Structural significance of modified nucleosides k²C and t⁶A present in the anticodon loop of tRNA^Ile

Abstract

The structural significance of the hypermodified nucleosides lysidine (k²C) and N⁶-(N-threonylcarbonyl) adenosine (t⁶A) present in the anticodon loop of E. coli tRNA^Ile has not been studied in detail theoretically at the atomic level. Hence, in the present paper, we investigated the conformational preferences of modified nucleosides k²C and t⁶A in the anticodon loop of tRNA^Ile using various quantum chemical methods. Multiple molecular dynamics simulation studies of the anticodon stem loop (ASL) of tRNA^Ile have also been made to see the solvation effect. The lysine moiety of lysidine orients back and forms a hydrogen bond with the 2′-hydroxyl group of the 34th ribose sugar whereas the t⁶A₍₃₇₎ side chain prefers a ‘distal’ conformation. This kind of conformation would be useful for recognition of the codons AUA instead of AUG. The t⁶A side chain prohibits canonical base pairing and maintains an anticodon open loop structure. The preferred conformations of k²C₍₃₄₎ and t⁶A₍₃₇₎ are stabilized by various intra as well as inter residual interactions within the anticodon loop which could be useful for maintaining a proper anticodon loop structure for smooth and in phase codon–anticodon interactions during the biosynthesis of proteins.

---

1 Introduction

Transfer RNA (tRNA) is an adapter molecule which plays a key role during the biosynthesis of proteins. Several modified nucleosides were discovered immediately after the discovery of tRNA.^1,2 Modified nucleosides are frequently found at the ‘wobble’ (34th) position and anticodon 3′-adjacent (37th) position in a range of tRNA in diverse species.^3–5 The hypermodified nucleosides present at the 3′-adjacent position avoid extended Watson–Crick base pairing during protein biosynthesis, whereas the 34th position tRNA modifications may limit or expand the scale of wobble base pairing.^6–8 The post-transcriptional modifications of the anticodon wobble position (34th) preserves codon–anticodon base pairing interactions by preventing codon misreading.⁹ Several wobble modifications are recognized to alter the codon specificity for precise decoding.^10–13

Lysidine is a lysine containing a cytidine derivative which occurs at the wobble (34th) position in the anticodon loop of E. coli tRNA^Ile_minor, Bacillus subtilus, Mycoplasama capricola tRNA^Ile, and potato mitochondrial tRNA^Ile.^14–17 The occurrence of lysidine (k²C) has also been reported at the 34th position of potato mitochondrial tRNA^Ile.¹⁸ Similarly, another modified nucleoside was also found at wobble position in tRNA^Ile of Haloferax volcani.¹⁹ The replacement of the oxygen atom in position 2 of cytidine with the ε-nitrogen atom of L-lysine results in the formation of lysidine. The modification of cytidine to lysidine involves the condensation of the lysine amino acid at the O(2) position of the cytidine. The lysidine modification prevents misrecognition of AUG as isoleucine and that of AUA as methionine.²⁰ The k²C modification also changes the aminoacylation identity of the tRNA from methionine to isoleucine.²¹ Agmatidine (agm²C), a polyamine-conjugated modified nucleoside has been identified at the 34^th position of archael tRNA^Ile which can also decode the AUA codon similarly to lysidine.²²

The hypermodified nucleoside N⁶-(N-theronlycarbonyl) adenosine (t⁶A) occurs at the 3′-adjacent (37th) position in the anticodon loop of tRNA.^23–25 The crystal structure of t⁶A was determined to understand the structural behavior of the modified nucleic acid base t⁶A.^26–28 The NMR study of the anticodon stem-loop of tRNA^lys showed that t⁶A at the 37th position plays a role in the removal of canonical hydrogen bonding within the anticodon loop.²⁹ The effects of t⁶A modification in codon discrimination have been studied crystallographically in ribosomes containing tRNA^lys and its cognate mRNA codon.³⁰ Transfer RNAs with an anticodon loop ending with uridine (U), the bulky hydrophilic modified base N⁶-(N-threonylcarbonyl) adenine (t⁶Ade), or 2-methylthio-N⁶-(N-threonylcarbonyl) adenine (ms²t⁶Ade) or N⁶-methyl-N⁶-(N-threonylcarbonyl) adenine (m⁶t⁶Ade) generally occur at the 3′-adjacent position of the anticodon. The tRNA containing 2-methylthio-N⁶-isopentenyl adenosine (ms²i⁶A)instead of t⁶A at the 3′-adjacent (37th) position to the anticodon loop have shown that the presence of uridine at the 36th position is responsible for the formation of t⁶A at the 3′-adjacent position in tRNA.⁸ The presence of the ureido HN–CO–NH linkage in the N(6) substituent is a common feature between molecules of hydrophilic modified adenine. Theoretically, it has been shown that orientation of the N(6)substituent in t⁶Ade and ms²t⁶Ade is ‘distal’ i.e., the N(6) substituent spreads away from the five-membered imidazole moiety of the adenine ring.^31,32 The conformational preferences of hydrophilic substituents t⁶A, m⁶t⁶A, i⁶A, ms²i⁶A, cis or trans isomers of zeatin (io⁶A) and its 2-methylthio derivatives^32–35 and the effects of protonation induced conformational transitions have also been studied.^36,37 Recently the conformational preferences of modified nucleoside ac⁴C in the anticodon loop and m²G, m²₂G in the hinge region of tRNA have been investigated by quantum chemical methods.^38–40

The role of modified bases in maintaining the conformational flexibility and stability of the anticodon loop structure of tRNA along with its function has been the topic of discussion in many studies.^41–45 Fluorescence detected circular dichroisum (FDCD) measurements and NMR techniques have been used to study the conformations of the anticodon loop of tRNA.^46,47 Molecular dynamics (MD) simulation studies of the anticodon loop structure have also been carried out to detect the conformational behavior of the anticodon loop in solution.^48–50 A crystallographic investigation of the anticodon loop of tRNA containing hypermodified nucleoside t⁶A has been carried out previously.²⁶

The conformational preferences of 34th and 37th positions containing modified bases in the anticodon loop of tRNA have not been studied in detail at the atomic level. Hence, the present molecular modelling studies have been undertaken to investigate the structural significance of the hypermodified nucleosides lysidine (k²C) and N⁶-threonylcarbonyl adenosine (t⁶A) in the anticodon loop structure of E. coli tRNA^Ile.

2. Nomenclature, convention and procedure

Fig. 1a describes the model anticodon loop segment (Me-p-C₍₃₂₎-p-U₍₃₃₎-p-k²C₍₃₄₎-p-A₍₃₅₎-p-U₍₃₆₎-p-t⁶A₍₃₇₎-p-A₍₃₈₎-p-Me) containing the hypermodified nucleoside lysidine (k²C) at the 34th ‘wobble’ position and N⁶-(N-threonylcarbonyl) adenosine (t⁶A) at the 3′-adjacent (37th) position of E. coli tRNA^Ile. The atom numbering and torsion angle definition for various acyclic chemical bonds in the hypermodified nucleoside lysidine, k²C (Fig. 1b) and N⁶-(N-threonylcarbonyl) adenosine, t⁶A (Fig. 1c) in the model anticodon loop segment of tRNA^Ile are shown in separate figures. In lysidine nucleoside, the torsion angle α₁[N(1)C(2)N(2)C(7)] describes C(7) rotation around the C(2)–N(2) bond and was measured in the right hand sense of rotation, with reference to the eclipsed orientation of N(1)C(2) and N(2)C(7) bonds. Likewise, the successive chemical bonds along the main extension of the substituent define the subsequent torsion angles β₁[C(2)N(2)C(7)C(8)], γ₁[N(2)C(7)C(8)C(9)], δ₁[C(7)C(8)C(9)C(10)], ψ₁[C(8)C(9)C(10)C(11)], ϕ₁[C(9)C(10)C(11)C(12)], ζ₁[C(10)C(11)N(11)H] and θ₁[C(10)C(11)C(12)O(12b)] similar to the lysine moiety. The torsion angle definition for the large side chain of t⁶A₍₃₇₎ are shown in (Fig. 1c). The torsion angle α₂[N(1)C(6)N(6)C(10)] denotes rotation around the C(6)–N(6) bond and was measured in the right hand sense of rotation, with reference to the eclipsed (0°) orientation of N(6)C(10) and C(6)N(1). Likewise, the remaining torsion angles such as β₂[C(6)N(6)C(10)N(11)], γ₂[N(6)C(10)N(11)C(12)], δ₂[C(10)N(11)C(12)C(13)], ε₂[N(11)C(12)C(13)O(13a)], θ₂[C(12)C(13)O(13a)H], ζ₂[N(11)C(12)C(14)O(14)], η₂[C(12)C(14)O(14)H] and ϕ₂[C(12)C(14)C(15)H] are as shown in Fig. 1c. The standard nomenclature system for the phosphate backbone along with the anticodon loop bases (Me-p-C₍₃₂₎-p-U₍₃₃₎-p-k²C₍₃₄₎-p-A₍₃₅₎-p-U₍₃₆₎-p-t⁶A₍₃₇₎-p-A₍₃₈₎-p-Me) was adopted from the IUPAC nomenclature system for polynucleotides.⁵¹ The starting structure for conformational energy calculations was taken from crystal structure data,⁵² where all the r bond distance, bond angle and torsion angle values for the anticodon loop bases along with the sugar-phosphate backbone were maintained throughout the conformational energy calculations. The glycosidic torsion angle values [χ₍₃₂₎, χ₍₃₃₎, χ₍₃₄₎, χ₍₃₅₎, χ₍₃₆₎, χ₍₃₇₎ and χ₍₃₈₎] were held ‘anti’ similar to the crystal structure⁵² and measured for O4′–C1′–N(9)–C(8) for purine nucleosides and O4′–C1′–N(1)–C(6) for pyrimidine nucleosides and their modified derivatives as for the crystal structure data.⁵² The ribose ring puckering was C3′-endo for all the seven nucleotides present in the anticodon loop of tRNA^Ile. The ribose-phosphate backbone was adopted from the yeast tRNA^Phe crystal structure.⁵²

Fig. 1 [a] Nomenclature and convention for model anticodon loop segment of tRNA^Ile. [b] Nomenclature and torsion angles definition of tautomeric lysidine (k²C) at the 34th position. [c] Nomenclature and torsion angle definition for t⁶A at the 37th position.

Conformational energy calculations of the lysine moiety of various forms of k²C and t⁶A side chain were started with the appropriate selection of bond lengths, bond angles and torsion angle values from earlier conformational studies.^53–55 The energy calculations were carried out in the gaseous state with the help of the quantum Chemical Perturbative Configuration Interaction with Localized Orbitals (PCILO) method.^56–58 Throughout the conformational search polarities of the chemical bonds were optimized for every molecular conformation. Energy correction terms up to the third order were included in the calculations of total ground state energy. This method has been found to be widely useful in conformational studies of various bioorganic molecules, including nucleic acid constituents, their derivatives⁵⁹ and peptide nucleic acids.⁶⁰ Recently, the PCILO results have also been confirmed at the ab initio level.^61,62 The variation of total energy with respect to torsion angles determining the lysine substituent and N⁶-threonylcarbonyl side chain orientation in the anticodon loop has been investigated. In the multidimensional conformational space the logical selection of grid points approach is used for searching the most stable structure and an alternative stable structure.⁶³ This procedure is well suited and quite fast for molecular conformations having several degrees of freedom.^34–37 In this way energetically preferred most stable and alternative stable conformations are obtained.

2.1 Automated full geometry optimization

Automated full geometry optimization calculations were carried out over the PCILO preferred conformations by using a semi-empirical quantum chemical RM1 (ref. 64) method (Fig. 2). At the time of full geometry optimization, the ribose-phosphate backbone of the anticodon loop of tRNA^Ile was frozen while bases (32 to 38) with glycosidic torsion angles were allowed to rotate freely. All these calculations were performed with the commercially available PC Spartan Pro (version 6.1.1.0, Wavefunction Inc.) software.

Fig. 2 [a] Full geometry optimized conformation by the RM1 method over the PCILO most stable structure of the model anticodon loop segment of tRNA^Ile. [b] RM1 optimized geometry of tautomeric lysidine (k²C) at the 34th position. [c] RM1 optimized geometry of t⁶A at the 37th position.

2.2 Molecular dynamic simulation

MD simulations were performed over the anticodon stemloop (ASL) model of tRNA (Fig. 3) containing modified nucleosides k²C at the 34th and t⁶A at the 37th positions. For lysidine and N⁶-(N-threonylcarbonyl) adenosine, RM1 preferred conformations were used (Fig. 2b). A model anticodon and stem loop of tRNA^Ile was surrounded by 17 Na⁺ ions and 5305 SPC/E water molecules filling a 57.63 × 63.857 × 58.910 ų rectilinear box.⁶⁵ Simulations were performed under the periodic boundary conditions by employing the Particle Mesh Ewald^66,67 method for calculation of long range interactions. MD trajectories were propagated at 2.0 fs time steps utilizing a shake algorithm⁶⁸ for all the hydrogen atoms with a non-bonded cut off of 9.0 Å. The non-bonded pair list was updated every 10 steps. The trajectories were calculated by maintaining a constant temperature (300 K) and constant pressure (1 atm) at a 2 fs time step according to the Berendsen coupling algorithm.⁶⁹

Fig. 3 Model anticodon stem and loop (ASL) in [a] clover leaf model and [b] three dimensional structure of tRNA^Ile considered for MD simulation.

An equilibration protocol similar to the earlier MD simulation study of nucleic acid was applied.^70,71 The equilibration protocol consisted of 5000 steps of steepest descent minimization followed by 50 ps of MD at 300 K applied to the relaxation of the initial strain present between the water molecules and the model anticodon stem and loop segment. Next the anticodon stem and loop segment was fixed while water molecules and Na⁺ counter ions were allowed to relax at 100 K (10 ps), 200 K (10 ps), and finally at 300 K for 930 ps, thus the equilibration protocol was completed at 1000 ps. The equilibrated system was further subjected to 5000 steps of steepest descend minimization to remove bad contacts between the water molecules and the model anticodon stem loop segment of tRNA^Ile. In further steps of MD simulation, no positional constraints were applied to the system and the temperature was progressively increased to 300 K in steps of 50 K with 1 ps at each step. Finally, the system was subjected to a production MD run (10 ns) at a temperature of 300 K and a constant pressure (1 atm) with a fully solvated and neutralized system. The ffbsc0 force field parameters available in Amber10 software were used in this MD simulation study.⁷² Parameters for the modified nucleoside were taken from ‘Modified Parameters Database servers’.⁷³ The PTRAJ module of Amber Tools 10 was used for analysis of average and snapshot structures.⁷⁴ MD simulation was performed with the Amber 10 simulation suit in the HP ProLiant-ML150G6 server.

Additional MD simulations of the anticodon stem loop segment of tRNA^Ile have been carried out to understand the effect of different temperature ranges and salt concentrations on lysidine conformations. The above mentioned simulation protocols were used for ASLs of tRNA^Ile simulation studies at various temperatures such as 270 K and 370 K. A MD simulation study of ASL with variation in salt concentrations was also carried out. Salt concentrations of 150 mM (9 Na⁺ and Cl⁻ each) and 1 M (62 Na⁺ and Cl⁻ each) were added to the neutralized solvated ASL system. All the MD simulations were performed for 10 ns of production run. The obtained simulation results were analyzed using the PTRAJ module⁷⁴ and a comparative account is discussed below.

3. Result and discussion

3.1 Orientation of the tautomer form of lysidine (k²C) at the 34th position in the anticodon loop of tRNA^Ile

The lysidine tautomer conformation contains a –NHR group at position 2, position 3 has –NH as the hydrogen bond donor group while the hydrogen bond acceptor group = NH is present at the 4th position of cytosine nucleoside. This form of lysidine can recognize the AUA codon instead of AUG as described in an earlier conformational study.⁵⁴ Hence, by considering the role of the tautomer form of lysidine in proper codon recognition for tRNA^Ile, the other three forms of lysidine viz. zwitterionic, non-zwitterionic and neutral forms are not discussed in the present paper. The PCILO preferred conformation of the tautomer form of lysidine, k²C₍₃₄₎ describing torsion angles are α₁ = 180°, β₁ = 180°, γ₁ = 30°, δ₁ = 180°, ψ₁ = 150°, ϕ₁ = 0°, ζ₁ = 0°, θ₁ = 120°. The lysine moiety of the tautomer form of lysidine in the anticodon loop of tRNA^Ile folds back and forms a hydrogen bond with the 2′-hydroxyl group of the 34th ribose ring in a similar manner to that reported in a previous study.⁵⁴ The glycosidic torsion angles [χ₍₃₂₎ = 31°, χ₍₃₃₎ = 33°, χ₍₃₄₎ = 5°, χ₍₃₅₎ = 26°, χ₍₃₆₎ = 25°, χ₍₃₇₎ = 13°, χ₍₃₈₎ = 10°] maintain ‘anti’ conformation similar to the crystal structure values.⁵²

The full geometry optimization over the PCILO stable structure of the model anticodon loop segment of tRNA^Ile containing the tautomer form of lysidine (k²C) at the 34th position and t⁶A at the 37th position has been performed using a semi-empirical RM1 method. The results of geometry optimization are shown in (Fig. 2a–c and Tables 1 and 2). The RM1 optimized torsion angles of the lysine moiety and the glycosidic torsion angles are listed in Table 1. The glycosidic torsion angles χ₍₃₂₎, χ₍₃₃₎ and χ₍₃₈₎ of the anticodon loop show small differences while χ₍₃₄₎, χ₍₃₅₎, χ₍₃₆₎ and χ₍₃₇₎ vary by 34–46° as compared to the crystal structure values.⁵² The optimization results revealed that all the glycosyl torsion angles prefer an ‘anti’ conformation similar to the crystal structure values.⁵² The minor changes found in the glycosidic torsion angles support the U-turn feature (O1P₍₃₆₎–HN(3)₍₃₃₎) interaction in the anticodon loop segment of tRNA^Ile (Table 2). Hydrogen bonding between O2′₍₃₃₎–HC(6)₍₃₄₎ (Table 2) takes part in maintaining base stacking interactions along with a U-turn feature within the anticodon loop.

Table 1 Optimized torsion angle values of tautomeric lysidine (k²C) and t⁶A in the model anticodon loop segment of tRNA^Ile obtained by using the RM1 method over the most stable structure (Fig. 2b and c)

Modified nucleoside Name	Torsion angles
Lysidine (k^2C)	α1 = 177°, β1 = 179°, γ1 = 71°, δ1 = 208°, ψ1 = 147°, ϕ1 = 300°, ζ1 = 320°, θ1 = 163°, χ(32) = 18°, χ(33) = 43°, χ(34) = 43°, χ(35) = 70°, χ(36) = 59°, χ(37) = 59°, χ(38) = 12°.
t^6A	α2 = 2°, β2 = 357°, γ2 = 177°, δ2 = 32°, ε2 = 55°, θ2 = 187°, ζ2 = 49°, ϕ2 = 184°, η2 = 53°, χ(32) = 18°, χ(33) = 43°, χ(34) = 43°, χ(35) = 70°, χ(36) = 59°, χ(37) = 59°, χ(38) = 12°.

---

Table 2 Geometrical parameters for hydrogen bonding from optimized conformations of tautomeric lysidine (k²C) and t⁶A in the model anticodon loop segment of tRNA^Ile

Method used atom 1–2–3	RM1		Figure reference
	r12	<123
O(12a)(34)–HO2′(34)	1.753	157.33	2b
O(12a)(34)–HN(2)(34)	2.475	147.72	2b
N(7)(35)–HO2′(33)	1.706	162.46	2a
O(2)(33)–HC3′(35)	2.674	169.46	2a
O1P(36)–HN(3)(33)	1.696	165.94	2a
N(1)(37)–HN(11)(37)	1.740	131.76	2c
O(13b)(37)–HO(14)(37)	2.549	119.15	2c
O(14)(37)–HN(6)(35)	2.504	107.13	2a
O(10)(37)–HO(14)(37)	2.567	117.55	2c
O2′(33)–HC(6)(34)	2.381	109.47	2a

---

The orientation of the lysine moiety obtained by the PCILO method and full geometry optimization results of the RM1 method (Fig. 2a) show that the lysine substituent may provide structural support to recognize AUA codons. Only the lysidine (k²C) tautomer form has the proper hydrogen bond donor–acceptor groups at the N(3) and N(4) sites respectively to form Watson–Crick hydrogen bonding with ‘A’ as compared to zwitterions, non-zwitterion and neutral forms of lysidine as shown in an earlier model.⁵⁴ Hence, the lysine substituent conformation of lysidine (k²C) also supports base stacking interactions in the anticodon loop of tRNA^Ile.

3.2 Orientation of N⁶-(N-threonylcarbonyl) adenine (t⁶A) in the model anticodon loop of tRNA^Ile

PCILO preferred conformation of the N⁶-(N-threonylcarbonyl) side chain (t⁶A) within the anticodon loop of tRNA^Ile describing torsion angle values of α₂ = 0°, β₂ = 0°, γ₂ = 180°, δ₂ = 0°, ε₂ = 90°, θ₂ = 180°, ζ₂ = 60°, ϕ₂ = 180° and η₂ = 60°. The results of full geometry optimization by the semi-empirical RM1 method performed over PCILO for the most stable structure of the model anticodon loop containing k²C at the 34th position and t⁶A at the 37th position are given in Tables 1and 2. In the presence of the lysidine tautomer form at 34th position, the hypermodified nucleoside N⁶-(N-threonylcarbonyl) adenine (t⁶A) preserves its initial geometry ‘distal’ conformation and the results are shown in Table 1. The RM1 optimization results (Fig. 2c) maintained starting geometry by retaining similar torsion angle values (Table 1) and hydrogen bonding interactions (Table 2).

The N⁶-(N-threonylcarbonyl) side chain prefers ‘distal’ conformation in the anticodon loop region i.e. the N⁶ substituent spreads away from the five-membered imidazole moiety of the adenine ring, but N(6)H point towards the atom N(7) of adenine ring as predicted crystallographically^26–28 and theoretically by the PCILO method.⁵⁵ This kind of orientation of the N⁶-(N-threonylcarbonyl) side chain helps to maintain a proper reading frame by preventing extended Watson–Crick base pairing interactions. The preferred conformation is stabilized by hydrogen bonding interactions between N(1)₍₃₇₎–HN(11)₍₃₇₎ and O(13b)₍₃₇₎–HO(14)₍₃₇₎ (Table 2). The ureido group (NH–CO–NH) of the t⁶A side chain is involved in base stacking interactions (Fig. 2c). The hydrogen atom of N(11)₍₃₇₎ of the ureido group of the t⁶A side chain interacts with N(1)₍₃₇₎ of the adenine ring in t⁶A. In the present study the bulky threonyl moiety of t⁶A prohibits itself from being incorporated into the helical stack in the anticodon loop whereas the ureido group of t⁶A has been found to be involved in base stacking interactions with adjacent bases as suggested by Murphy et al.³⁰

The intra-residual interactions observed in the preferred conformation of t⁶A (Fig. 2a and c) would be helpful to avoid canonical hydrogen bonding interactions between U₍₃₃₎–A₍₃₇₎ within the anticodon loop of tRNA^Ile as explained by an earlier experimental study.²⁹ The N⁶-threonylcarbonyl side chain of t⁶A does not interact with any other bases in the anticodon loop structure (Fig. 2a and c). Thus, these intra-residual interactions show that these modified bases are helpful for maintaining base stacking interactions within the loop as well as for proper recognition of the codon during protein biosynthesis. The uriedo and threonyl moiety of t⁶A maintains proper base stacking interactions within the anticodon loop structure of tRNA^Ile. Hence, the two bulky hypermodified nucleosides k²C and t⁶A fit well in the anticodon loop structure of tRNA^Ile and may help to maintain a proper anticodon loop structure.

Hence, molecular interactions of k²C and t⁶A observed in this study might be useful for maintaining the proper three-dimensional fold of the tRNA^Ile anticodon loop. Similarly, there are many studies in which quantum chemical methods such as RM1, PM3, PCILO, HF SCF and DFT along with MD simulations have been used to investigate structural significance of isolated modified nucleosides present at the 34th ‘wobble’ and 3′-adjacent (37th) positions in the anticodon loop^{31–34,38,54,55,62} as well as the hinge region of tRNA.^39,40 It has been reported that quantum chemical calculations can provide a more important description of chemical structures and can achieve similar accuracy for both base pairing and stacking interactions.⁷⁵

3.3. Molecular dynamic simulation study

MD simulation was also performed to study the effect of hydration on the conformations of lysidine (k²C) and N⁶-(N-threonylcarbonyl) adenine (t⁶A) within the anticodon loop segment of tRNA^Ile by using Amber10 software.⁷² An earlier report suggests that the accuracy of MD simulation depends upon the availability of a reasonable starting structure.⁷⁵ Hence, the RM1 preferred anticodon loop structure of tRNA^Ile containing k²C and t⁶A (Fig. 2) was used as a starting geometry for the 10 ns MD simulation. The results of simulated structures are depicted in (Fig. 4–7). MD simulation results were found to be in agreement with the conformations of k²C and t⁶A as obtained in the first part of this paper (Fig. 2and Tables 1 and 2).

Fig. 4 Analysis of torsion angles of lysine substituent occuring at the 34th position of tRNA^Ile during MD simulation.

3.3.1 Orientation of lysine substituent of k²C during MD simulation. During MD simulation the torsion angle α₁ of the lysine side chain maintains an initial value i.e., α₁ = 180° (Fig. 4a and 7a) which is similar to the RM1 most stable conformation (Fig. 2a) as well as to that previously reported.⁵⁴ The torsion angle β₁ value remains at ±180° until the end of the simulation except from 1.0 ns to 1.3 ns and 1.9 ns to 2.3 ns where it fluctuates between −60° and −150° as shown in (Fig. 4b). At the end of the simulation the torsion angle β₁ is 90° (Fig. 4b). In a similar manner to β₁, the torsion angle γ₁ = ±60°, ±180° (Fig. 4c). These fluctuations of the β₁ and γ₁ torsion angles determine the orientation of the lysine side chain with respect to the pyrimidine ring. During simulation it has been observed that the lysine substituent stays away from the ASL and below the pyrimidine ring when torsion angles β₁ and γ₁ are between −60° and −150° and 60° respectively (Fig. 4b and c). This kind of orientation may expose the carboxyl group of the lysine side chain to solvent molecules (Fig. 7b). However, torsion angle β₁ and γ₁ direct the lysine side chain towards ASL and above the pyrimidine ring of lysidine , when the values are also ±180° and 60° respectively (Fig. 4b and c). Such a type of coordination between the torsion angles β₁ and γ₁ may allow interactions of the amino and carboxyl groups of the lysine substituent of lysidine with the 35th adenine residue (Fig. 4b and c and 7c and d).

Torsion angle δ₁ maintains a value of ±180° throughout the simulation except from 4 ns to 4.1 ns and 4.9 ns to 5.1 ns, where it has a value of around 90°(Fig. 4d). The ψ₁ torsion angle is ±180° throughout the simulation period except for a small time span of 3.0 ns to 3.2 ns and 5.0 ns to 5.2 ns where it has a value of around 60° to 90° (Fig. 4e). Torsion angle ϕ₁ is within the range of ϕ₁ = 60°, ±180° (Fig. 4f) which determines the orientation of the –COO⁻ and –NH₃⁺ groups of the lysine substituent. These results suggest that free rotations around the torsion angle ϕ₁ are possible which would allow the –COO⁻ and –NH₃⁺ groups of the lysine substituent to interact with the ribose ring of the 34th and 35th nucleosides within the anticodon loop as well as with codons if present at the 3′ side of the anticodon loop of tRNA during the protein biosynthesis process. Torsion angle θ₁ fluctuates around ±90° to ±150° and determines the orientation of oxygen atoms O(12a) and O(12b) of the lysine side chain (Fig. 4g). Free movement of the –NH₃⁺ group of the lysine moiety could be possible by fluctuations around the torsion angle ζ₁ within the range of ±60°, ±180° (Fig. 4h). Such fluctuations might be useful during codon–anticodon interactions. Thus, the orientation of the lysine substituent could play an important role in recognition of the codon AUA instead of AUG.

3.3.2 Orientation of the N⁶-threonylcarbonyl moiety of t⁶A after MD simulation. The threonylcarbonyl side chain of t⁶A has a ‘distal’ orientation as observed in the RM1 preferred structure obtained from the conformational study of the anticodon loop of tRNA^Ile (Fig. 2c) and a crystal structure study.⁴⁹ Torsion angle α₂ maintains an initial value of ±30° (Fig. 5a) similar to the RM1 preferred structure (Fig. 2c) in this study as well as to that previously reported.⁵² The hydrogen bonding interaction between N(1)₍₃₇₎–HN(11)₍₃₇₎ (Fig. 6b) of the threonylcarbonyl (t⁶A₍₃₇₎) side chain helps to maintain a ‘distal’ configuration as found in Fig. 2c. Torsion angles β₂ and γ₂ maintain similar values similar to the RM1 preferred conformation (Fig. 2c) throughout the simulation period (Fig. 5b and c). Torsion angle δ₂ maintains a value of 60° up to 7 ns, and then fluctuates between −60° and −150° until the end of the simulation period (Fig. 5d). Torsion angle ε₂, fluctuates mostly in the range 60° to 150° during the MD simulation, except for the initial 1 ns time period where it varies between ±180° (Fig. 5e). The fluctuations of the ε₂ torsion angle determine the orientation of the –COOH group in the t⁶A side chain. Torsion angle θ₂ remains at 180° throughout the simulation, which determines the orientation of the –OH group, as can be seen in Fig. 5f. The torsion angle ζ₂ fluctuates between ±60° throughout the simulation time (Fig. 5g) while the torsion angle η₂ determines the orientations of the methyl group hydrogen, which varies periodically between ±60° and ±180° (Fig. 5h).

Fig. 5 Various torsion angles of t⁶A at the 37th position during the MD simulation period.

3.3.3 Hydrogen bonding interactions during MD simulation. Interaction between HN(3) of the 33rd residue with the oxygen atom of the ribose phosphate backbone of the 36th residue, popularly known as the U-turn feature,⁴³ is well maintained during the 10 ns simulation run (Fig. 6a) except for some minor fluctuations occurring from 1.9 ns to 2.1 ns, 2.5 ns to 2.7 ns, 3.9 ns to 4.1 ns, and 5.0 ns to 5.4 ns. These minor fluctuations could be because of limitations of the available force field which may not be sufficient to obtain meaningful results.⁷⁵

Fig. 6 Analysis of hydrogen bonding during the MD simulation period.

Hydrogen bonding interactions of t⁶A. Hydrogen bonding between N(1)₍₃₇₎–HN(11)₍₃₇₎ (Fig. 6b) was stable throughout the simulation period which is essential to maintain the ‘distal’ conformation of t⁶A. The interaction between O(13b)₍₃₇₎–HO(14)₍₃₇₎ (Fig. 2c) observed in the RM1 preferred structure did not occur during the 10 ns simulation period because atom O(13b)₍₃₇₎is in close proximity with HN(6) of the 35th adenine residue (Fig. 6c and 7d). However, in the RM1 optimization instead of O(13b)₍₃₇₎ the atom O(14)₍₃₇₎ interacts with HN(6) of the 35th adenine within the anticodon loop (Fig. 2c). These interactions could play an important role in maintaining the proper three-dimensional structure of tRNA.

Fig. 7 Average structures of four bases, 34–37 (the phosphate backbone is not shown for clarity). [a] 0.95–1.3 ns , [b] 1.4–2.7 ns, [c) 2.8–3.6 ns, [d] 3.6–6.1 ns and [e] 6.2–7.1 ns time trajectories.

Hydrogen bonding interactions of k²C during MD simulation. The hydrogen bonding interaction between HN(2) of the lysine substituent with O2′ of the hydroxyl group of the 34th ribose ring (Fig. 6h) maintained the RM1 preferred conformation (Fig. 2b)as well as that previously reported.⁵⁴ Similarly, interaction between O(12a)₍₃₄₎–HO2′₍₃₄₎ was observed between 2.8 and 7.7 ns of the simulation period (Fig. 6g). These interactions could be responsible for maintaining the ‘trans’ orientation of the lysine substituent during the simulation period. Interaction between the O(12b) and HN(11) of the lysine side chain was observed throughout the simulation (Fig. 6e) as found in the RM1 preferred structure (Fig. 2b). The –NH₃⁺ group of the lysine side chain interacts with O2′ of the 35th ribose ring during 3.0 ns to 7.8 ns of the simulation period (Fig. 6f). All the hydrogen bonding interactions observed during the MD simulation period are depicted in Table 3.

Table 3 Time period of various hydrogen bonding interactions during 10 ns of MD simulation study

Atom involved in hydrogen bonding	Time period interval (ns)	Figure reference
O2′ (34)–HN(2)(34)	0.0 to 8.5	6h
O(12a)(34)–HO2′(34)	2.8 to 7.7	6g
O(12b)(34)–HN(11)(34)	0.0 to 1.0, 5.5–6.2, 7.0 to 7.5, 8.5 to 10.0	6e
N(1)(37)–HN(11)(37)	0.0 to 10.0	6b
N(7)(35)–HO(13b)(37)	3.6 to 6.0, 8.3 to 8.5	6d
O(13a)(37)–HN(6)(35)	0.9 to 1.1, 2.0 to 2.5, 4.0 to 6.0, 8.0 to 8. 5	6c
O2′(35)–HN(11)(34)	3.5 to 4.0, 6.7 to 6.8, 7.5 to 7.6	6f
O1P(36)–HN(3)(33)	0.0 to 8.0	6a

---

Average structures were analyzed to understand the dynamic behavior of the modified nucleosides present at the 34th and 37th positions of the anticodon loop of tRNA^Ile (Fig. 7a–e). In average structures taken for the period of 0.95 ns to 1.3 ns (Fig. 7a) the lysine side chain bends away from ASL and below the plane of the pyrimidine ring. In this average structure (Fig. 7a) the threonylcarbonyl moiety maintains a ‘distal’ orientation similar to the RM1 preferred conformation (Fig. 2b).

The lysine substituent has an extended conformation as can be seen in the average structure selected for the period of 1.4 ns to 2.7 ns (Fig. 7b). In the average structure taken for the period 2.8 ns to 3.6 ns, the lysine moiety turns towards ASL above the plane of the pyrimidine ring supported by hydrogen bonding interactions between O2'₍₃₅₎–HN(11)₍₃₄₎, O2'₍₃₄₎–HN(2)₍₃₄₎, and O(12a)₍₃₄₎–HO2'₍₃₄₎ (Fig. 7c). The lysine side chain has a similar orientation to that found during 2.8 ns to 3.6 ns (Fig. 7c) for the period 3.6 ns to 6.1 ns, while the threonylcorbonyl moiety folds below the purine ring and turns towards the 35th adenine residue by forming hydrogen bonding interactions between O(13b)₍₃₇₎–HN(6)₍₃₅₎ and N(7)₍₃₅₎–HO(13a)₍₃₇₎ (Fig. 6c and d and 7d). The average structure extracted for the period 6.2 ns to 7.1 ns has a ‘trans’ orientation of the lysine substituent, which is stabilized by the hydrogen bonding interaction between O(12a)₍₃₄₎–HO2'₍₃₄₎ (Fig. 6g and 7e) as predicted by the RM1 method (Fig. 2b).

3.3.4 Molecular dynamics simulations of the anticodon stem loop of tRNA^Ile at various temperatures. MD simulation trajectories at various temperatures were analyzed and showed significant similarities with the above described results (Fig. 8 and 9). The sugar phosphate backbone structures and U-turn features of ASL were well maintained during simulations performed at 300 K, whereas minor deviations were observed in the backbone structure during simulation performed at 270 K as compared to 300 K MD simulation (Fig. 8 and 9). The RMS deviations for 270 K and 300 K temperatures showed significant stability by contributing various inter-residual hydrogen bonding interactions (Fig. 9b). The selected average structures showed close resemblance with the above described simulation results in respect to lysidine conformations and U-turn features. The lysidine side chain interacts with the O2′ of the 34th ribose sugar which might interact with the AUA codon during the translation process. MD simulation performed at 370 K resulted in a disturbed stem loop structure. This might be due to the instability of ASL at higher temperatures as represented by the higher RMSD value (Fig. 9b).

Fig. 8 MD simulated structures of ASL at different temperature ranges i.e. 270 K (tan color), 300 K (cyan color), 370 K (orchid color) containing k²C at the 34th position.

Fig. 9 Shows [a] stability of U turn feature hydrogen bonding during MD simulation and [b] RMS deviation of ASL sugar phosphate backbone during MD simulation at various temperatures.

3.3.5 Molecular dynamics simulations of the anticodon stem loop of tRNA^Ile at various salt concentrations. MD simulations of ASL with different salt concentrations were analyzed for RMS deviation of the sugar phosphate backbone as well as the characteristic U turn feature (Fig. 10 and 11). The sugar phosphate backbone structures were well preserved during simulation studies performed with a 150 mM salt concentration and shows comparable average structures with MD calculations carried out without salt, while excess salt concentration destabilizes the ASL structure (Fig. 10 and 11). The specific U turn hydrogen bonding is well maintained during MD simulations at 0 M and 150 mM salt concentrations, but an excess salt concentration (1 M) destabilizes the U turn feature as well as the anticodon loop structure (Fig. 10 and 11). MD simulation performed at an increased salt concentration resulted in a disordered ASL structure (Fig. 10). This could be due to some interactions of ions with ASL which might cause instability in the U turn hydrogen bonding resulting in a destabilized structure as represented by the higher RMSD value (Fig. 11b). However lysidine conformations without salt were comparable with conformations generated during MD simulation with a 150 mM salt concentration. Hence, the increased temperature as well as salt concentration destabilizes the ASL loop structure by disturbing the proper base pairing interactions within the stem loop structure.

Fig. 10 MD simulated structures of ASL at different salt concentrations i.e. 0 M (tan color), 150 mM (cyan color), 1 M (orchid color) containing k²C at the 34th position.

Fig. 11 [a] Stability of U turn feature hydrogen bonding during MD simulation and [b] RMS deviation of ASL sugar phosphate backbone during MD simulation at different salt concentrations.

4 Conclusions

The orientation of the lysine substituent of hypermodified nucleoside lysidine folds back and interacts with the 2′-OH group of the 34th ribose ring in the model anticodon loop segment in a similar manner to that observed in the isolated molecule.⁵⁴ The interaction of the amino group of the lysine side chain with the hydroxyl group of the 35th residue could be useful to enhance base stacking interactions in order to maintain the anticodon loop structure. Hence, this kind of conformation of the lysine moiety might be essential for recognition of the AUA codon instead of AUG during protein biosynthesis.

The N⁶-threonylcarbonyl side chain (t⁶A) prefers a ‘distal’ conformation within the anticodon loop as predicted in the earlier studies.^55,27 The threonylcarbonyl side chain interacts with N(6) of the 35th adenine residue in the model anticodon loop. Hence, it may help to prevent canonical C32–A37 base pairing which results in maintaining the proper anticodon loop structure. Other inter-residual interactions observed in the anticodon loop of tRNA^Ile support base stacking interactions. Separate MD simulations carried out at different temperature ranges and salt concentrations also revealed the significance of k²C and t⁶A to maintain the proper tRNA^Ile anticodon loop structure. Thus, the molecular interactions of k²C and t⁶A noticed in the context of other anticodon nucleosides could play an important role to maintain proper anticodon loop structure which would be helpful to have smooth and in phase protein biosynthesis process. However, it would be good to study the effects of such modified nucleosides by considering whole tRNA^Ile along with mRNA and ribosomes.

Acknowledgements

This work was supported by the Department of Science and Technology, Government of India, New Delhi through the DST Fast Track project sanctioned by KDS, and is gratefully acknowledged. SBS is thankful for the financial support provided by the DST-PURSE program sanctioned by the Department of Biochemistry, Shivaji University, Kolhapur.

References

1. D. B. Dunn, Biochim. Biophys. Acta, 1959, 34, 286–288.
2. J. D. Smith and D. B. Dunn, Biochem. J., 1959, 72, 294–301.
3. R. W. Adamiak and P. Gornicki, Prog. Nucleic Acid Res. Mol. Biol., 32, 27–74.
4. Y. Motorin, G. Bec, R. Tewari and H. Grosjean, RNA, 1997, 3, 721–733.
5. A. Morin, S. Auxilien, B. Senger, R. Tewari and H. Grosjean, RNA, 1998, 4, 24–37.
6. P. A. Limbach, P. F. Crain and J. A. McClowskey, Nucleic Acids Res., 1994, 22, 2183–2196.
7. B. C. Persson, Mol. Microbiol., 1993, 8, 1011–1016.
8. P. F. Agris, Prog. Nucleic Acid Res. Mol. Biol., 53, 79–129.
9. G. Bjork. Biosynthesis and Function of Modified Nucleosides, in tRNA: Structure, Biosynthesis, and Function, ed. D. Soll and U. L. RajBhandary, American Society for Microbiology, Washington, DC, 1995, pp. 165–205.
10. P. R. Schimmel, Similarities in the Structural Organization of Complexes of tRNAs with Aminoacyl-tRNA Synthetases and the Mechanism of Recognition, in Transfer RNA: Biological Aspects, ed. D. Soll, J. N. Abelson and P. R. Schimmel, Cold Spring Harbor Lab. Press, Plainview, NY, 1979, pp. 297–310.
11. H. Ozeki; H. Inokuchi; F. Yamao; M. Kodaira; H. Sakano; T. Ikemura and Y. Shimura, Genetics of Nonsense Suppressor tRNAs in Escherichia coli, in Transfer RNA: Biological Aspects, ed. D. Soll, J. N. Abelson and P. R. Schimmel, Cold Spring Harbor Lab. Press, Plainview, NY, 1980, pp. 341–362.
12. S. Yokoyama, Z. Yamaizumi, S. Nishimura and T. Miyazawa, Nucleic Acids Res., 1979, 6, 2611–2626.
13. A. P. Gerber and W. Keller, Science, 1999, 286, 1146–1149.
14. F. Harada and S. Nishimura, Biochemistry, 1974, 13, 300–307.
15. T. Muramatsu, S. Yokoyama, N. Horie, A. Matsuda, T. Ueda, Z. Yamaizumi, Y. Kuchino, S. Nishimura and T. A. Miyazawa, J. Biol. Chem., 1988, 263, 9261–9267.
16. J. Matsugi, K. Murao and H. J. Ishikura, J. Biochem., 1996, 119, 811–816.
17. Y. Kuchino, S. Watanabe, F. Harada and S. Nishimura, Biochemistry, 1980, 19, 2085–2089.
18. F. Weber, A. Dietrich, J. H. Weil and L. Marechal-Drouard, Nucleic Acids Res., 1990, 18, 5027–5030.
19. R. Gupta, J. Biol. Chem., 1984, 259, 9461–9471.
20. T. Muramatsu, K. Nishikawa, F. Nemoto, Y. Kuchino, S. Nishimura, T. Miyazawa and S. Yokoyama, Nature, 1988, 336, 179–181.
21. A. Soma, Y. Ikeuchi, S. Kanemasa, K. Kobayashi, N. Ogasawara, T. Ote, J. Kato, K. Watanabe, Y. Sekine and T. Suzuki, Mol. Cell, 2003, 12, 689–698.
22. Y. Ikeuchi, S. Kimura, T. Numata, D. Nakamura, T. Yokogawa, T. Ogata, T. Wada, T. Suzuki and T. Suzuki, Nat. Chem. Biol., 2010, 6, 277–282.
23. G. B. Chheda, R. H. Hall, J. Mozejko, D. I. Magrath, M. P. Schweizer, L. Stasiuk and P. R. Taylor, Biochemistry, 1969, 8, 3278–3282.
24. H. Ishikura, Y. Yamada, K. Murao, M. Saneyoshi and S. Nishimura, Biochem. Biophys. Res. Commun., 1969, 37, 990–995.
25. R. Brambilla, H. Rogg and M. Staehelin, Nature, 1976, 263, 167–169.
26. R. Parthasarathy, M. Soriano-Garcia and G. B. Chheda, Nature, 1976, 260, 807–808.
27. R. Parthasarathy, J. M. Ohrt and G. B. Chheda, Biochemistry, 1977, 16, 4999–5008.
28. D. A. Adamiak, T. L. Blundell, I. J. Tickle and Z. Kosturkiewicz, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1975, 31, 1242–1246.
29. J. W. Stuart, Z. Gdaniec, R. Guenther, M. Marszalek, E. Sochacka, A. Malkiewicz and P. F. Agaris, Biochemistry, 2000, 39, 13396–13404.
30. F. V. Murphy IV, V. Ramakrishnan, A. Malkiewicz and P. F. Agaris, Nat. Struct. Mol. Biol., 2004, 11, 1186–1191.
31. R. Tewari, J. Biomol. Struct. Dyn., 1990, 8, 675–686.
32. R. Tewari, Int. J. Quantum Chem., 1988, 34, 133–142.
33. K. D. Sonawane, U. B. Sonavane and R. Tewari, J. Biomol. Struct. Dyn., 2002, 19, 637–648.
34. U. B. Sonavane, K. D. Sonawane, A. Morin, H. Grosjean and R. Tewari, Int. J. Quantum Chem., 1999, 75, 223–229.
35. K. D. Sonawane, U. B. Sonavane and R. Tewari, Int. J. Quantum Chem., 2000, 78, 398–405.
36. R. Tewari, Int. J. Quantum Chem., 1994, 51, 105–112.
37. R. Tewari, Chem. Phys. Lett., 1995, 238, 365–370.
38. B. V. Kumbhar, A. D. Kamble and K. D. Sonawane, Cell Biochem. Biophys., 2013, 66, 797–816.
39. R. S. Bavi, A. D. Kamble, N. M. Kumbhar, B. V. Kumbhar and K. D. Sonawane, Cell Biochem. Biophys., 2011, 61, 507–521.
40. R. S. Bavi, S. B. Sambhare and K. D. Sonawane, Computational and Structural Biotechnology Journal, 2013, 5(6).
41. F. A. Vendeix, A. Dziergowska, E. M. Gustilo, W. D. Graham, B. Sproat, A. Malkiewicz and P. F. Agris, Biochemistry, 2008, 47, 6117–6129.
42. F. Claesens and R. Rigler, Eur. Biophys. J., 1986, 13, 331–342.
43. M. Sundaram, P. C. Durant and D. R. Davis, Biochemistry, 2000, 39, 12575–12584.
44. S. Kurata, A. Weixlbaumer, T. Ohtsuki, T. Shimazaki, T. Wada, Y. Kirino, K. Takai, K. Watanabe, V. Ramakrishnan and T. Suzuki, J. Biol. Chem., 2008, 283, 18801–18811.
45. J. W. Stuart, K. M. Koshlap, R. Guenther and P. F. Agris, J. Mol. Biol., 2003, 334, 901–918.
46. D. H. Turner, I. Tinoco and M. F. Maestre, Biochemistry, 1975, 14, 3794–3799.
47. K. Beardsley, T. Tao and C. R. Cantor, Biochemistry, 1970, 9, 3524–3532.
48. C. S. Tung, Biophys. J., 1997, 72, 876–885.
49. N. E. McCrate, M. E. Varner, K. I. Kim and M. C. Nagan, Nucleic Acids Res., 2006, 34, 5361–5368.
50. P. Auffinger and E. Westhof, Biophys. J., 1996, 71, 940–954.
51. A. Cornish-Bowden, Pure Appl. Chem., 1983, 55, 1273–1280.
52. H. Shi and P. B. Moore, RNA, 2000, 6, 1091–1105.
53. U. B. Sonavane, K. D. Sonawane and R. Tewari, J. Biomol. Struct. Dyn., 2002, 20, 473–485.
54. K. D. Sonawane and R. Tewari, Nucleosides, Nucleotides Nucleic Acids, 2008, 27, 1158–1174.
55. R. Tewari, Indian J. Biochem. Biophys., 1987, 24, 170–176.
56. S. Diner, J. P. Malrieu and P. Claverie, Theor. Chim. Acta, 1969, 13, 1–17.
57. S. Diner, J. P. Malrieu, F. Jordan and M. Gilbert, Theor. Chim. Acta, 1969, 15, 100–110.
58. J. P. Malrieu, in Semiempirical Methods of Electronic Structure Calculations Part A Techniques, ed. G. A. Segal, Plenum New York, 1977, pp. 69–103.
59. B. Pullman and A. Pullman, Adv. Protein Chem., 28, 347–526.
60. S. Sharma, U. B. Sonavane and R. Joshi, Int. J. Quantum Chem., 2009, 109, 890–896.
61. F. S. Nandel and A. Saini, J. Biophys. Chem., 2011, 2, 37–48.
62. N. M. Kumbhar and K. D. Sonawane, J. Mol. Graphics Modell., 2011, 29, 935–946.
63. R. Tewari, Int. J. Quantum Chem., 1987, 31, 611–624.
64. G. B. Rocha, R. O. Freire, A. M. Simas and J. P. Stewart, J. Comput. Chem., 2006, 27, 1101–1111.
65. H. J. C. Berendsen, G. R. Griegera and T. P. Straatsma, J. Phys. Chem., 1987, 91, 6269–6271.
66. T. Darden, D. York and L. Pedersen, J. Chem. Phys., 1993, 98, 10089–10092.
67. D. M. York, T. A. Darden and L. G. Pedersen, J. Chem. Phys., 1993, 99, 8345–8348.
68. J. P. Ryckaert, G. Ciccotti and H. J. C. Berendsen, J. Comput. Phys., 1977, 23, 327–336.
69. H. J. C. Berendsen, J. P. M. Postma, W. F. Van Gunsteren and A. DiNola, J. Chem. Phys., 1984, 81, 3684–3690.
70. P. S. Auffinger, S. Loise-May and E. Westhof, J. Am. Chem. Soc., 1995, 117, 6720–6726.
71. P. S. Auffinger, S. Loise-May and E. Westhof, Faraday Discuss., 1996, 103, 151–174.
72. A. Perez, I. Marchan, D. Svozil, J. Sponer, T. E. Cheatham, C. A. Laughton and M. Orozco, Biophys. J., 2007, 92, 3817–3829.
73. R. Aduri, B. T. Psciuk, P. Saro, H. Taniga, H. B. Schlegel and J. Santa Lucia Jr, J. Chem. Theory Comput., 2007, 3, 1464–1475.
74. D. A. Case, T. A. Darden, T. E. Cheathem III, C. L. Simmerling, J. Wang, R. E. Duke, R. Luo, M. Crowley, R. C. Walker, W. Zhang, K. M. Merz, B. Wang, S. Hayik, A. Roitberg, G. Seabra, I. Kolossvary, K. F. Wong, F. Paesani, J. Vanicek, X. Wu, S. R. Brozell, T. Steinbrecher, H. Gohlke, L. Yang, C. Tan, J. Mongan, V. Hornak, G. Cui, D. H. Mathews, M. G. Seetin, C. Sagui, V. Babin and P. A. Kollman, AMBER 10, University of California, San Francisco, 2008.
75. M. A. Ditzler, M. Otyepka, J. Sponer and N. G. Walter, Acc. Chem. Res., 2010, 43, 40–47.

---

Footnote

† Present address: Institute of Bioinformatics and Biotechnology, University of Pune, Ganeshkhind, Pune, Maharashtra, India.

---