Wild-Type RNA Microhelix^Ala and 3:70 Variants:  Molecular Dynamics Analysis of Local Helical Structure and Tightly Bound Water

Abstract

Molecular dynamics simulations of RNA microhelix Ala indicate that G:U and other 3:70 purine: pyrimidine wobble pairs induce local deviations from A-form geometry in their respective microhelices; the helix is underwound at the base-pair step above and overwound at the base-pair step below, in each case by about 7-9°compared to canonical A-form RNA. On the basis of analysis of average water densities and residence lifetimes, the wild-type microhelix strongly binds a water molecule in the minor groove of the 3:70 base pair, consistent with crystallographic analyses of an RNA duplex derived from the acceptor stem of Escherichia coli tRNA Ala . Other wobble pairs show water binding at this position but to a lesser degree; the strength of water binding correlates directly with the measured aminoacylation activities of the microhelices as substrates for E. coli alanyl-tRNA synthetase (G:U > 2AA:IsoC > G:dU > I:U). Watson-Crick base pairs at the 3:70 position show no tendency toward specific hydration. This tightly bound minor-groove water in the microhelices with 3:70 wobble pairs evidently does not function to stabilize a particular local helical structure, but it may play a role as a specific recognition element or serve as an indicator of interaction specificity between the microhelix and a hydrogen-bonding residue of the aminoacyl-tRNA synthetase.

Efficient aminoacylation of the tRNAAla acceptor stem: Dependence on the 2:71 base pair

“…Meaningful interpretation of MD simulations requires that each trajectory's stability be determined+ To this end, various parameters were examined to assess convergence+ The root-mean-square deviation (RMSD) from the average structure was calculated for the wild-type microhelix Ala simulation and each of the 2:71 variant simulations over a 2-ns production run time (Fig+ 5)+ Analyses of helical parameters were also examined, including local twist, base-pair inclination, and x-displacement from the global helical axis+ Appreciable fluctuations from a constant value were not observed in the RMSDs or in the helical parameters (data not shown)+ Thus, all trajectories in this study were established as reliable for interpretation+ All helices exhibited A-form geometries with the exception of local twist deviations around the G3:U70 wobble base pair+ Such geometries are consistent with prior NMR, X-ray, and MD simulation studies on the wild-type microhelix and related analogs (Ramos & Varani, 1997;Vogtherr et al+, 1998;Mueller et al+, 1999;Nagan et al+, 1999)+ Analysis of hydration indicated a water molecule to be tightly bound at the 3:70 position in every case examined here, consistent with prior experiments and simulations (Mueller et al+, 1999;Nagan et al+, 1999) for wild-type microhelix Ala + Along with structural properties, we also examined the relationship between base-pair substitution and RNA electrostatic properties+ The electrostatic potential (ESP) computed by solving the linearized PB equation is expressed as the free energy of interaction between the RNA and a unit positive charge+ The calculation is performed considering two different dielectric regions (solute and solvent) defined by the solvent-accessible surface+ This type of calculation can provide insights into how a cation, a positively charged side chain, or a protein with a positive electrostatic potential may interact with the RNA+ Such sites will also interact strongly with hydrogen-bond donating groups, including water+ PB calculations are routinely used to describe proteinsmall molecule (Ullmann et al+, 2000) and protein-DNA interactions (Misra et al+, 1994), and recently, Chin and coworkers have used PB methods to describe specific RNA electrostatics (Chin et al+, 2000)+ Van der Waals components can be further added to the electrostatic component in such a way that steric effects, derived from the actual size of atoms involved in the interaction, are also taken into account+ The sum of these two terms defines the molecular interaction potential (MIP), as originally described in the context of quantum mechanical systems by Orozco and Luque and later extended to classical systems (Orozco & Luque, 1993)+ MIP isopotential contours at Ϫ5+5 kcal/mol for various microhelices are shown in Figure 6+ All microhelices are shown from the major groove side+ The isopotential contour around the wild-type microhelix Ala indicates that there is a large region of negative interaction potential throughout the major groove, beginning at the 5:68 base pair and extending up the helix roughly to the 2:71 base pair+ Notwithstanding structural similarities between all microhelices studied, a striking difference in interaction potentials was observed with respect to...…”

“…The negative potential in the major groove of RNA implies that positively charged ions or polar groups acting as hydrogen bond donors (to include water) may be expected to specifically bind within this groove+ Divalent cations have been found bound to the major groove in many RNAs (Ott et al+, 1993;Cate & Doudna, 1996;Kieft & Tinoco, 1997;Correll et al+, 1997;Le et al+, 1998;Suga et al+, 1998)+ In our simulations we have also found that monovalent sodium ions tend to bind to the major groove of wild-type microhelix Ala (nonspecifically) near the tetraloop and at the top of the acceptor stem, as both regions are solvent-or ion-accessible (data not shown)+ Whether ion interactions or direct major-groove interactions with a positive surface on AlaRS are individually or jointly critical for aminoacylation activity cannot be determined in the absence of more structural information about the enzyme-substrate complex+ A separate and interesting possibility is that the change in MIP may influence the degree to which water is bound at the G3:U70 wobble position+ It is well established, both from crystallography (Mueller et al+, 1999) and MD simulations (Nagan et al+, 1999), that in the wild-type microhelix, a water molecule is very strongly bound to this position+ The release of this water, assuming it is replaced by an enzymatic contact having a similar enthalpy of interaction, may provide an entropic driving force for enzyme-substrate binding+ To the extent that the enthalpy of interaction decreases, it may be that looser water binding reduces the entropy available to the system from release of bound water to the solvent+ CONCLUSIONS In summary, 2:71 base-pair functional groups significantly affect the electrostatics of the alanine acceptor stem+ Sequence-specific effects upon ion binding Young et al+, 1997;Auffinger & Westhof, 2000) and molecular electrostatics (Chin et al+, 2000) have been examined by others+ For example, Chin and coworkers recently noted that PB calculations of surface potentials varied with RNA sequence and that these differences might predict drug or cation binding (Chin et al+, 2000)+ Our study provides new insights into the potential similar importance of electrostatics for protein recognition of an RNA surface+ In particular, we show that there is a statistically significant correlation between the electrostatic potential in the major groove of the tRNA acceptor stem and aminoacylation with alanine+ To our knowledge, this study provides the first example demonstrating the importance of such effects for specific protein-RNA recognition+…”

“…Initial structures for molecular dynamics simulations were derived from a high-resolution NMR structure of E. coli microhelix Ala (Fig+ 1) determined by Ramos and Varani (1997)+ Simulations of the wild-type G2:C71-containing microhelix Ala and seven variants (C:G, A:U, U:A, 2AA:U, U:2AA, 2AP:U, and U:2AP) were carried out+ The microhelices were surrounded by sodium counterions and 9 Å of TIP3P water (Jorgensen et al+, 1983)+ Particle mesh Ewald (PME; York et al+, 1993) was used to account for long-range electrostatics+ Simulations were carried out for 2+0 ns of production run time per microhelix using the SANDER module of AMBER 5+0 (Case et al+, 1997) with a 2+0-fs time step+ The SHAKE algorithm (Ryckaert et al+, 1977) was applied to all hydrogen atoms and the nonbonded interactions cutoff was set to 9+0 Å+ The pair list was updated every 25 steps+ Constant pressure (1 bar) and temperature (300 K) were maintained according to the Berendsen (Berendsen et al+, 1984) coupling algorithm+ The detailed equilibration protocol of Cheatham was employed )+ Simulations were equilibrated for an additional 500 ps beyond the initial equilibration period in the case of the C:G and all pur:pyr 2:71 base-pair variants+ The U:2AA and U:2AP variants were equilibrated for 1300 ps and 800 ps, respectively, beyond the initial equilibration period+ Simulations employed the Cornell et al+ (1995) force field as implemented in AMBER 5+0 (Case et al+, 1997) including additional nonstandard nucleotide parameters as needed+ Charges for 2AA were previously determined (Nagan et al+, 1999)+ Charges for the nonstandard base 2AP (charges and other parameters are available from the authors) were derived by fitting Hartree-Fock (Hehre et al+, 1986) electrostatic potentials (ESP charges) using the 6-31G* (Ditchfield et al+, 1971;Hehre et al+, 1972;Hariharan & Pople, 1972) basis set and normalizing to the existing force field (Cornell et al+, 1995)+ All electronic structure calculations were carried out using Gaussian 98, Revision A+6 (Frisch et al+, 1998)+…”

Molecular Dynamics Simulations of RNA Systems

“…[19,20], (2) the differences between RNA and DNA structures of similar sequence [19][20][21] or (3) the effects associated with the introduction of natural [22,23] and non-natural modified nucleotides [24,25]. In order to generate larger systems, model built fragments can also be assembled with motifs derived from X-ray or NMR structures [26].…”

RNA solvation: A molecular dynamics simulation perspective

“…OR(n ϩ 1) hydrogen-bond pattern observed in RNA. 20,21,29,30 Other methods rely on the time spent by a water molecule in a 5 Å sphere around a given solute atom 31 or use coordination correlation function approaches. 14,32,33 Worth mentioning is that none of the above methods can give statistically exact values for the residence times of the most strongly bound water molecules, since the length of a MD simulation should be many times longer than the longest calculated residence times.…”