Search tips
Search criteria 


Logo of narLink to Publisher's site
Nucleic Acids Res. 2010 August; 38(15): 5105–5118.
Published online 2010 April 12. doi:  10.1093/nar/gkq210
PMCID: PMC2926596

Dynamics on multiple timescales in the RNA-directed RNA polymerase from the cystovirus ϕ6


The de novo initiating RNA-directed RNA polymerase (RdRP), P2, forms the central machinery in the infection cycle of the bacteriophage ϕ6 by performing the dual tasks of replication and transcription of the double-stranded RNA genome in the host cell. By measurement and quantitative analysis of multiple-quantum spin-relaxation data for the δ1 positions of Ile residues that are distributed over the 3D-fold of P2, we find that the enzyme is dynamic both on the fast (ps–ns) and slow (µs–ms) timescales. The characteristics of several motional modes including those that coincide with the catalytic timescale (500–800/s) are altered in the presence of substrate analogs and single-stranded RNA templates. These studies reveal the plasticity of this finely tuned molecular machine and represent a first step towards linking structural information available from a host of crystal structures to catalytic mechanisms and timescales obtained from the measurements of kinetics for homologous systems in solution.


RNA directed RNA polymerases (RdRPs) utilize RNA templates to catalyze the formation of a phosphodiester bond between nucleoside triphosphate (NTP) substrates. RdRPs are key in sustaining the growth, propagation and infection cycle of RNA viruses. The high-resolution structures of a variety of viral RdRPs reveal a conserved overall domain architecture shaped like a cupped hand containing ‘fingers’, ‘thumb’ and ‘palm’ domains (Supplementary Figure S1a) (1). RdRPs contain six conserved sequence motifs also found in DNA polymerases (A–E in the palm domain and F in the fingers domain) (2,3). These motifs that play a central role in catalyzing the nucleotidyl transfer reaction are responsible for binding metal ions, NTPs and RNA.

P2 (4), the 75-kDa RdRP from the cystovirus ϕ6 inititates the de novo synthesis of both minus-strand RNA (replication) (5) and plus-strand RNA (semi-conservative transcription) (6) and lies at the heart of the four protein polymerase complex central to the cystoviral life cycle (7,8). Multiple high-resolution crystal structures exist for P2 in the apo-form, and in complex with NTPs (and NTP-analogs), single-stranded DNA (ssDNA) and single-stranded RNA (ssRNA) templates (9–11). While these structures of P2 and indeed of several other RdRPs (1) provide atomic snapshots along the catalytic pathway, extensive flexibility both at the domain level and at the level of local motifs seems essential to reconcile biochemical data with structural information (12). For example, in P2 and in the homologous hepatitis C virus polymerse (HCVpol), the central catalytic channel is enclosed by additional loops (‘fingertips’) that connect the thumb and fingers, limiting the possibility of large-scale opening and closing motions as in DNA polymerases (13). Thus, in the absence of domain motion, the narrow template channel (Supplementary Figure S1b) is capable of accommodating ssRNA but not RNA duplexes. However, these motions, though not large in amplitude, likely exist, since both enzymes are capable of processing circular templates (14). Reorientation of the fingers and thumb would then be necessary in order to accommodate the wider template. Indeed structures of HCVpol have been solved in both the ‘open’ and ‘closed’ forms (15).

On the other hand, evidence of motion on a more local scale has also been found in RdRPs. Studies on the poliovirus polymerase (3Dpol) and also on other classes of nucleic-acid polymerases (HIV-1 reverse transcriptase, RB69 DNA-dependent DNA polymerase and T7 DNA-dependent RNA polymerase) have suggested that a step that leads to significant rate enhancement during the nucleotide addition cycle involves protonation of the pyrophosphate product facilitated by conformational changes in motif D (16). Other studies have hinted towards the influence of motions involving motif A on catalysis (17). An allosteric surface site that binds NTPs has also been located in HCVpol (18). This site, that is a target for a host of nucleotide-mimetic antivirals, is dynamically coupled to the catalytic site that is more than 30-Å away (19). Thus, dynamics both local and long-range, seem to be ‘the missing link between structure and function’ (12) in these ubiquitous enzymes.

Measurement, quantitative analyses and interpretation of nuclear spin relaxation data in solution provides the most promising route towards characterization of dynamics on multiple timescales (20) to complement the large amount of structural and biochemical data available on RdRPs. These studies have proven to be quite successful in revealing functional dynamics in other proteins. For example, in the case of adenylate kinase, the lid opening motions were noted to be rate limiting in the catalytic cycle of the enzyme (21). Similar conclusions were drawn for cyclophilin A where the measured exchange rate (kex) obtained from relaxation dispersion experiments, was found to be close to the sum of the forward and reverse rates of substrate cis–trans isomerization (22).

While the RdRPs are large (>55 kDa) and generally difficult to solubilize in high enough quantities under conditions suitable for NMR studies, recent advances in experimental methodology for both the backbone (23) and sidechains (24) coupled with the availability of cryogenic probe technology (25), provide great promise for the accurate measurement and interpretation of spin-relaxation data for these enzymes in solution. In our attempts to elucidate the role of dynamics along the catalytic cycle of RdRPs, in atomic detail, we have commenced NMR studies of ϕ6 P2. While an analysis of the dynamics of all backbone and sidechain nuclei remains our goal, as a first step, we present here an analysis of the spin-relaxation of multiple quantum (MQ) coherences involving 13C and 1H nuclei at the δ1 position of the Ile residues in P2 (26–28). We find that the enzyme is dynamic on multiple timescales, a behavior not immediately obvious from crystallographic analyses alone (9). These various motional modes that are active in apo P2, are perturbed by the presence of Mg2+ ions, substrate NTPs and short ssRNA, and provide insight into the dynamic nature of the initiation complex. The timescale of the lower frequency motional modes that involve several conserved functional motifs coincides with that relevant for catalysis as determined from kinetic analyses of representative RdRPs (29).


Overexpression and Purification of ϕ6 P2

The ϕ6 P2 expression plasmid was constructed by PCR amplification of the pLM687 (30) plasmid that encodes of the cystoviral l segment with recombinant Pfu DNA polymerase (Stratagene) by using the following oligonucleotides—5′-GGTAAGCGCCATATGCCGAGGAGA-3′ and 5′-TACGAATTCCGGCATGATTACCTAGGCATTACA-3′ as upstream and downstream primers, respectively. The P2 gene fragment was digested with NdeI and EcoRI, and ligated to a digested pET28a( + ) vector (Stratagene). N-terminal His6-tagged full-length P2 was transformed into Escherichia coli BL21-DE3 (Stratagene) and grown overnight in 20 ml of unlabeled M9 medium prepared in 100% H2O and supplemented with kanamycin (50 mg/l). Following overnight growth, the bacteria were adapted for growth in media containing progressively higher concentrations of D2O from 10 to 100% at 37°C. The final overnight culture (20 ml grown in M9 medium prepared in 100% D2O) was used to inoculate 1 l of M9 medium in 100% D2O containing 3 g/l of U-[12C,2H]-glucose (CIL) and 1 g/l of 15NH4Cl (CIL) as the principal sources of carbon and nitrogen respectively. When the optical density at 600 nm reached ~0.6, 50 mg of 2-keto-3-d2-4-13C-butyrate (CIL) was added into the growth medium. After 1 h of growth, the culture was placed on ice for 30 min and induced with 1 mM IPTG. The growth was continued for 20 h at 16°C. The cells were harvested by centrifugation at 5000 rpm and resuspended in lysis buffer containing 20 mM Tris–HCl (pH 7.2), 300 mM NaCl, 5 mM Imidazole and two tablets of Complete Mini EDTA-free protease inhibitor (Roche). The suspension was lysed by sonication and the lysate was centrifuged at 16 000 rpm for 30 min at 4°C. The soluble lysate was then loaded onto a nickel affinity column (Ni–NTA agarose, Qiagen), and washed extensively with buffer containing 20 mM Tris–HCl (pH 7.2), 300 mM NaCl, and 20 mM imidazole. The bound recombinant P2 was eluted with buffer containing 20 mM Tris–HCl (pH 7.2), 300 mM NaCl, 500 mM imidazole and then dialyzed against the NMR buffer that contained 20 mM bis-tris, 400 mM NaCl and 4 mM DTT at pH 6.5. The uniformly-15N,2H-Ile-δ1-13C,1H-labeled protein was concentrated by using Amicon Ultra-4 or Amicon Ultra-15 centrifugal concentrators (Millipore) for NMR analyses.

Assignment of Ile δ1 resonances

All NMR experiments were performed using a Varian Inova 600 MHz spectrometer equipped with a triple-resonance {1H,13C}-cryogenic probe capable of applying pulsed field gradients along the z-axis. Resonance assignment of the 25 Ile δ1 resonances was achieved using an Ile-to-Leu mutation strategy described below (Supplementary Figure S2). Twelve Ile-to-Leu triple mutants were designed specifically avoiding simultaneous mutation of Ile residues that were sequentially or spatially proximal in the same set. In addition 13 single Ile-to-Leu mutants were also created. Mutations were introduced into the cDNA corresponding to wild-type ϕ6 P2 using the QuikChange Site-directed Mutagenesis and QuikChange Multisite-directed Mutagenesis kits (Stratagene) using primers designed utilizing the QuikChange Primer Design Program (Stratagene). The mutations were confirmed by sequencing. The mutated plasmids were transformed into E. coli BL21-(DE3)-RIPL competent cells (Stratagene) and each uniformly-15N,2H-Ile-δ1-13C,1H-labeled mutant P2 was overexpressed and purified as described previously. The resonances corresponding to mutated Ile δ1 positions vanished in an 13C,1H HMQC spectrum. Of 25 Ile-δ1 resonances, 15 were assigned using the strategy employing triple-mutants (see the legend to Supplementary Figure S2 for more details). However, some of these mutations resulted in protein misfolding and could not be used for assignments. Consequently, assignments for 8 of the remaining 10 Ile δ1 resonances were obtained using Ile-to-Leu point mutants. Resonances corresponding to Ile211 and Ile283 could not be unambiguously identified.

Chemical shift changes

The concentrations of Mg2+ (states 2–6, Table 1), of GMPPNP (Sigma-Aldrich) (states 3, 4 and 5) and of GMPCPP (Jena Biosciences) (states 3A and 6), were held constant at 5, 4 and 4 mM, respectively. The concentrations of the ssRNA were maintained in 3.5–5-fold excess of the concentration of P2 (250–350 µM).

Table 1.
Complexes of ϕ6 P2 utilized in this study

Scaled chemical shift changes (Δδij in Hz) between states i and j (i, j = 1, 2, 3, 3A, 4, 5, 6; Table 1) were calculated using the following equation

equation image

where δi/j,C/H are the 13C/1H chemical shifts in states i/j in 13C,1H HMQC spectra and ν0 is the spectrometer 1H frequency in MHz (600 MHz in the present case).

NMR relaxation measurements

All relaxation experiments were performed at 25°C and 600 MHz using sweep-widths of 9000 and 1600 Hz in the direct (1H, 512 complex points) and indirect (13C, 32 complex points) dimensions respectively. The 13C carrier was positioned at 13.15 ppm at the center of the Ile 13Cδ1 resonances. P2 concentration varied from 250 to 350 μM for the cross-correlated relaxation measurements and was around 250 µM for the relaxation dispersion measurements. All fits to theoretical curves both for the relaxation analyses [Equations (2) and (4)] and for the fluorescence measurements [Equation (6)] were performed using in-house programs that utilize the ODRPACK subroutines (31). Errors in the fitted parameters were obtained from the inverse covariance matrices of the fits and included both the random as well as the model selection errors.

Measurement of intra-methyl 1H–1H cross-correlation rates

In order to measure the intra-methyl 1H–1H dipole-dipole cross-correlated relaxation rate (η)(32), a set of two experiments were recorded. The signal intensity (IA) in the first experiment (A) during a variable relaxation delay (T) depends only on spin-dynamics in the I = 3/2 manifold of the Ile δ1 methyl H3 spin-system. In the second experiment (B), the signal intensity (IB) is affected by spin-dynamics in the I = 3/2 and I = 1/2 manifolds. The cross-correlation rate (η) is related to the intensity ratio (IA/IB) in the following way (32):

equation image

where δ depends on the 1H–1H cross-relaxation with external protons. Seven values of T = 1, 2, 4, 6, 8, 10 and 14 ms were used with 256 and 64 transients per t1 point for experiments A and B, respectively. η and δ values were obtained by fitting the experimentally determined IA/IB ratio as a function of T to Equation (2). The order-parameters (An external file that holds a picture, illustration, etc.
Object name is gkq210i1.jpg) were calculated from the following equation (32) using the rotational correlation time (τC) (assuming that the rotational diffusion was isotropic) estimated from hydrodynamic modeling:

equation image

where rHH = 1.813 Å, θHH = 90° and all other symbols have their usual meaning.

Multiple–quantum relaxation dispersion

13C,1H multiple quantum relaxation dispersion curves were obtained for states 1–6 (Table 1) at 600 MHz using the pulse sequences developed by Korzhnev et al. (33). Fifteen values of the RF–field strength (ν1 = 1/2τCPMG, where τCPMG is the spacing between successive π pulses) = 31.25, 62.5, 93.75, 125.0, 156.25, 187.5, 250.0, 312.5, 375.0, 437.5, 500.0, 625.0, 750.0, 875.0 and 1000 Hz were used with a constant relaxation delay of 32 ms. 128 transients were collected per t1 point. Data at each field were acquired in an interleaved fashion with all ν1 values collected prior to t1 incrementation in order to minimize possible errors resulting from sample heating and/or precipitation and to increase the accuracy of the fitted exchange parameters. All relaxation dispersion data were analyzed assuming exchange between two chemically distinct sites A and B with an exchange rate kex = kA→B + kB→A, where kA→B and kB→A are the rates for the forward and reverse processes, respectively. In the limit of fast exchange, when kex >> (ΔωωC)1/2 (where ΔωH=|ωH(A) – ωH(B)| and ΔωC=|ωC(A) – ωC(B)| are the differences between the chemical shifts of the two states for the 1H and 13C nuclei, respectively) and when ΔωH → 0, the effective multiple–quantum relaxation rate (ΓMQ,eff) for a given RF–field, ν1 is given by (33,34)

equation image

ξ = 2pA(1 – pAωHΔωC, pA is the fractional occupancy of state A (the fractional occupancy of state B, pB = 1 – pA) and ΓMQ(∞) is the multiple–quantum relaxation rate at infinite field strength i.e. the value of ΓMQ,eff in the absence of chemical exchange. The kex values were obtained as one of the parameters in a three–parameter [kex, ΓMQ(∞) and ξ] fit of the experimental dispersion curves to Equation (4).

A point to be made here is that for practical considerations, using Equation (4) should not result in substantial errors in kex values provided that the product, ΔωωC, is small and the dynamic timescale lies in the fast exchange limit. However, errors in ΓMQ(∞) and ξ, using this simplified approach, are larger and correlated with each other. Therefore we do not attempt to interpret ΓMQ(∞) and ξ values obtained in the present study.

Fluorescence measurements

All fluorescence anisotropy experiments were performed using single-stranded RNA oligos (5′-UUUCC-3′ and 5′-UUUUC-3′) labeled at the 5′-end with 6-carboxyfluorescein. Oligos were purchased PAGE purified from Dharmacon RNAi Technologies. Experiments were performed using a Panvera Beacon 2000 fluorescence anisotropy system. Fluorescence intensity data were acquired by passing the emission following an excitation at 490 nm through a 520 nm cutoff filter. A 10 s integration time was used for four combinations of the horizontal (H) and vertical (V) polarization filters. The measured intensities using parallel (IVV) and perpendicular (IVH) polarizer settings were converted into fluorescence anisotropy using the following equation:

equation image

Anisotropy values (A) were measured in the presence of increasing amounts of P2. Values of the dissociation constant (Kd) were obtained by fitting the dependence of A on P2 concentration using the following equation:

equation image

where A(C) is the anisotropy when the concentration of P2 is C and A is the limiting anisotropy value. The concentration of RNA (CRNA) was kept constant at 5 nM. All experiments were performed at pH 6.5 in the NMR buffer. Absence of non-specific binding to the fluorescein moiety was confirmed by measuring A values of fluorescein alone (300 nM) and in the presence 40 and 400 µM P2.

In order to measure the affinity of the GTP-analogs GMPPNP and GMPCPP towards P2, tryptophan (ϕ6 P2 contains 12 Trp residues) fluorescence spectra were obtained using a Spex Fluorolog-3 (Horiba Jobin Yvon) spectrometer with excitation and emission slits set to 1.6 nm. Fluorescence emission spectra (300–450 nm) were acquired in 0.5-nm steps and signal-averaged over three scans. Emission intensities were obtained by averaging the values at the maximum ± 4 nm. The excitation wavelength was set to 282 nm for each addition of either GMPCPP or GMPPNP to the protein sample in NMR buffer. Apparent Kd values were obtained by using fits to an analog of Equation (6) with the anisotropy values replaced by the corresponding emission intensities, the CRNA by the concentration of P2 (held constant at 10 µM) and C by CGMPNP/GMPCPP representing the variable concentration (0–2.5 mM) of the GTP analogs.

Hydrodynamic modeling

The diffusion tensor for full-length P2 was estimated by utilizing the bead model as implemented in the HYDRONMR program (35). The effective radius of the atomic elements was set to 3.2 Å, the temperature to 298.15 K and the viscosity to 9.11 cP. The minibead solvent shell radius was calculated automatically. The atomic coordinates of P2 were obtained from structures with the following PDB IDs: 1HHS, 1HHT, 1HI0, 1HI8, 1UVI, 1UVJ, 1UVL, 1UVM and 1UVN.


The principal focus of this work is to ascertain the functionally relevant motions in P2 for multiple states along its enzymatic pathway. The motions that are capable of influencing catalysis in RdRPs are typically slow, occurring on a millisecond-to–microsecond (µs–ms) timescale (29). Given this timescale, these motions are likely to involve relatively large structural segments and are thus expected to be non–local. This is in contrast to more local, fast dynamics on a picosecond–to–nanosecond (ps–ns) timescale involving only a small group of atoms. While characterizing dynamics in atomic detail, on both the ps–ns and µs–ms timescale, remains our goal, in the present work we use the δ1 position of the 25 Ile residues in P2 as probes of motional changes along its catalytic cycle. Given the near uniform distribution of the Ile residues over the structural domains (Figure 1a) and the key functional motifs (Figure 1b) of P2, this approach should provide valuable insight not only into the large amplitude µs-ms timescale motions in P2 that likely affect catalysis, but also into the local (ps–ns) flexibility of the catalytic motifs.

Figure 1.
(a) Distribution of Ile residues over the P2 tertiary structure. The fingers (1–30, 104–276, 333–397), thumb (37–91, 518–600), palm (277–332, 398–517) and C-terminal domains (601–664) are ...

Chemical shift assignment

13C,1H HMQC spectra of uniformly-deuterated,Ile-δ1-[13C,1H]-labeled P2 showed excellent dispersion and all 25 expected Ile δ1 13C, 1H resonances were seen/resolved (Figure 1c). Resonance assignments were carried out using a set of Ile-to-Leu single and triple mutants as described in the ‘Materials and Methods’ section. The methodology employing triple-mutants has been described in detail in the Supplementary Data (see the legend to Supplementary Figure S2). In all, 23 of 25 Ile resonances could be unambiguously assigned using this strategy. Resonances corresponding to Ile211 and Ile283 (indicated by asterisk in Figure 1c) could only be tentatively assigned and relaxation data for these residues were not interpreted further.

Choice of binary and ternary complexes of P2

We analyzed the spin relaxation rates of the Ile δ1 positions for P2 in the apo state, in complex with Mg2+, with GTP analogs—GMPPNP and GMPCPP, and two different 5-nt ssRNA constructs. A total of seven distinct states (states 1, 2, 3, 3A, 4, 5 and 6, Table 1) including binary and ternary complexes, were studied. The RNA construct, 5′-UUUCC-3′, chosen for states 4 and 6, corresponds to the final 5 nt from the 3′-end of the minus-strands (called s and m) of the small and medium segments of the three-segmented (S, M and L for small, medium and large) double-stranded RNA (dsRNA) genome of cystoviruses. Biochemical data for P2 reveals that transcription is far more efficient for the s and m strands compared with that of l (that contains a 5′-UUUAC-3′ sequence at the 3′-end) suggesting a role for the 3′-end sequence (more specifically of the penultimate nucleotide from the 3′-end) on the ability of ssRNA to be processed by P2 (6). The second 5-nt ssRNA construct chosen for the present study, differed in sequence from the first, only at the penultimate position from the 3′-end (U instead C–5′-UUUUC-3′), while introducing a minimal size differential. Choice of this construct provides an additional advantage that is described at length in the following paragraph.

The choice of the ternary complexes of P2 with substrate (Mg2+/GTP analogs) and ssRNA warrants some discussion (Figure 2). For RdRPs that initiate RNA synthesis de novo i.e. without the use of a primer, the initiation NTP acts as a 1-nt primer and base-pairs with the extreme 3′-nt (T1) of the template forming the first nucleobase (D1) from the 5′-end of the daughter strand. The second substrate NTP (D2) base pairs with the penultimate template nucleotide from the 3′-end (T2). Formation of a stable initiation complex requires both T1–D1 and T2–D2 base-pairing to occur. The first phosphodiester bond is thus formed between D1 and D2 with the cleavage of the latter between the α and β−phosphates releasing pyrophosphate (shown schematically in Figure 2a, also see Supplementary Figure S3). The process continues with each incoming substrate nucleotide (CTP, UTP, ATP or GTP) complementary to the corresponding template nucleobase (36). Thus, among the ternary complexes studied here, state 4 (Table 1, Figure 2b) represents a dead-end or stalled complex where the first phosphodiester bond between D1 and D2 can form since GMPPNP is a viable substrate for P2 (Supplementary Figure S3) given the fact that it can be cleaved between the α and β phosphates (37). However, additional polymerization beyond the one-nucleotide addition cycle and formation of a 2-nt daughter chain (5′-PPP-G-P-G-3′) becomes extremely inefficient, as the substrate nucleotide (A from ATP or an ATP-analog) complementary to T3 (U) is unavailable. Examples of stalling through nucleotide deprivation in ϕ6 P2 has been described in the literature (38) and characterized structurally (10). In state 5 (Table 1) a complex capable of forming a stable initiation platform cannot form since the complementary nucleotide D2 (A from ATP or analog) for T2 (U) is unavailable (Figure 2b). State 6 (Table 1), on the other hand, represents a stable initiation complex, stabilized by proper base-pairing of D1 with T1 and D2 with T2 (D1,2 = GMPCPP and T1,2 = C). However the phosphodiester linkage between D1 and D2 cannot form since in GMPCPP the bond between the α and β phosphates cannot be cleaved (Figure 2b and Supplementary Figure S3).

Figure 2.
(a) Formation of a stable initiation complex results from the proper base pairing of the template bases T1 (C) and T2 (C) (numbered 3′→5′) with the substrate (GTP analog) bases D1 (G) and D2 (G). Nucleophilic attack by the 3′-OH ...

Fast dynamics

Prior to probing the catalytically relevant µs-ms timescale, we investigated the local rigidity of the Ile δ1 positions on the fast ps–ns timescale. Thus, beginning with the apo enzyme (state 1), the difference between the relaxation rates of the fast and slow relaxing coherences corresponding to the I = 3/2 manifold of the methyl H3 spin-system at each Ile δ1 positions, was measured. The extracted difference relates to the intra-methyl 1H–1H dipole–dipole cross-correlation rate (η). This rate being independent of chemical exchange to first-order (27,28) is therefore uncontaminated by the effects of slow conformational dynamics and represents an accurate measure of local dynamics on the fast, ps–ns timescale through its relationship with the order parameter, An external file that holds a picture, illustration, etc.
Object name is gkq210i2.jpg [Equation (3)]. For residues that are highly disordered on the ps–ns timescale, An external file that holds a picture, illustration, etc.
Object name is gkq210i3.jpg values are close to 0, and residues that are ordered on this timescale display An external file that holds a picture, illustration, etc.
Object name is gkq210i4.jpg values that are close to 1.

η values for 21 of 25 Ile [1H,13C]-δ1 resonances were analyzed in detail for states 1 through 4. Signals corresponding to Ile617 and Ile641 from the C-terminal domain were weak and excluded from the analysis as were those belonging to Ile211 and Ile283. Representative fits to Equation (2) (see ‘Materials and Methods’ section) and the η values for the 21 Ile resonances for the four states analyzed are shown in Figure 3.

Figure 3.
(a) Representative fits to Equation (2) for Ile340 and Ile500 are shown for P2 in the apo state (state 1). The scaling factor used in the y-axis is the ratio (4) of the number of scans used in experiments A and B. (b) η values for the Ile residues ...

Prior to the availability of backbone 15N, 1H assignments and the analysis of backbone 15N relaxation rates, we relied on hydrodynamic calculations (35) to estimate the nature of the rotational diffusion tensor (see ‘Materials and Methods’ section) of P2. This was required to justify the use of Equation (3) that assumes isotropic rotational diffusion with a single correlation time (τC) and also to utilize the τC value to convert the experimental η to the more readily interpretable An external file that holds a picture, illustration, etc.
Object name is gkq210i5.jpg using Equation (3). These calculations predicted a nearly isotropic diffusion tensor [2Dzz/(Dxx+Dyy) = 1.07 ± 0.03] with a corresponding τC of 40.15 ± 0.76 ns. Thus using a τC value of 40 ns we estimated An external file that holds a picture, illustration, etc.
Object name is gkq210i6.jpg values for states 1–4 using Equation (3).

Apo P2 (Figure 4) displayed an average An external file that holds a picture, illustration, etc.
Object name is gkq210i9.jpg value of 0.63 ± 0.27 (10% trimmed — 0.60 ± 0.22) with the highest degree of order seen for the palm domain residues located in and around the catalytic site including those that form a part of motif C (445–461; Ile446, 0.96 ± 0.14 and Ile449, 0.82 ± 0.15) that bears the conserved catalytic DD motif (Asp453, Asp454). On the other hand, the most dynamic Ile residues were located on the edge of the template tunnel at the top the fingers domain—Ile12 (0.18 ± 0.01), Ile22 (0.26 ± 0.01), and Ile164 (0.37 ± 0.06); Ile37 (0.46 ± 0.07) on the thumb domain; and Ile96 (0.28 ± 0.02) located on the fingertips. This mobility at the entrance to the template pore could be the result of fast fluctuations at the top of the fingers and the thumb domains. The other parts of the fingers and thumb were relatively well ordered on the fast timescale. The other residues that were disordered on the fast timescale included Ile488 (0.28 ± 0.02, motif D) and Ile500 (0.20 ± 0.02, motif E) near the catalytic site at the entrance of the substrate channel.

Figure 4.
An external file that holds a picture, illustration, etc.
Object name is gkq210i7.jpg values for the Ile Cδ1 positions in apo-P2 (state 1). The Ile residues are indicated by spheres colored according to their respective An external file that holds a picture, illustration, etc.
Object name is gkq210i8.jpg values. Three views of P2 are shown and key residues labeled. Residues for which data could not be analyzed ...

An analysis of the An external file that holds a picture, illustration, etc.
Object name is gkq210i10.jpg values in states 2–4 revealed small but significant changes in the dynamic nature of several residues compared with the apo state. While the overall patterns of fast dynamics were generally similar, some very large variations were seen especially around the catalytic site for the motif C residues Ile446 and Ile449. Ile449, located near the top of the β-hairpin bearing the catalytic Asp residues displayed increased disorder in the complex with substrate GMPPNP/Mg2+ (state 3, An external file that holds a picture, illustration, etc.
Object name is gkq210i11.jpg = 0.42 ± 0.06 compared with 0.82 ± 0.15 for state 1). This motion is indicative of local conformational diversity of the top of the hairpin and likely originates from the continuous remodeling of the catalytic cavity to facilitate stabilizing local interactions with a variety of templates and substrates. Ile446, located at the base of the hairpin became highly ordered in state 4 (An external file that holds a picture, illustration, etc.
Object name is gkq210i12.jpg = 1.34 ± 0.27). This extreme rigidity near the catalytic site possibly reflects a distinct state for the stalled enzyme. Notably, Ile449 (An external file that holds a picture, illustration, etc.
Object name is gkq210i13.jpg = 0.77 ± 0.33) is also ordered in state 4. As will be discussed at length below, both the analysis of chemical shift perturbations as well as the dynamic characteristics on the µs-ms timescale also seemed to indicate that state 4 was quite different from all other states considered.

While the errors in the experimental η values (and consequently the An external file that holds a picture, illustration, etc.
Object name is gkq210i14.jpg values) were relatively large, the overall rigidity of the stalled complex (state 4, An external file that holds a picture, illustration, etc.
Object name is gkq210i15.jpg = 0.64 ± 0.20, 10% trimmed—0.57 ± 0.18) while comparable with that of the apo enzyme (state 1) was higher than both state 2 (Mg2+, An external file that holds a picture, illustration, etc.
Object name is gkq210i16.jpg = 0.54 ± 0.25, 10% trimmed—0.51 ± 0.20) and state 3 (GMPPNP/Mg2+, An external file that holds a picture, illustration, etc.
Object name is gkq210i17.jpg = 0.57 ± 0.27, 10% trimmed—0.52 ± 0.21). The increased fluctuations on the fast timescale in the Mg2+/GTP complexes could be an indication of conformational fine-tuning of the enzyme to facilitate the processing of a variety of ssRNA templates and substrate NTPs.

Thus, considering the An external file that holds a picture, illustration, etc.
Object name is gkq210i18.jpg values of all four states (states 1–4) together, the highest degree of flexibility on the ps–ns timescale was seen for the Ile residues in close proximity to the entry portals for the substrate NTPs and the template ssRNA. This was in contrast to the high rigidity seen for the two Ile residues located on the catalytic motif C, except for the increased disorder at the top of this motif in the Mg2+/GMPPNP-loaded state (state 3). Altogether, the fast dynamics seems connected to loading of both substrates and templates. This increased dynamics likely helps modulate local interactions allowing a variety of substrates and templates (with differences in local structure, e.g. variation in possible hydrogen-bond donors or acceptors) to be processed efficiently.

Binding affinity towards ssRNA and GTP-analogs

The ternary complexes (P2/Mg2+/GMPPNP/ssRNA) studied here involved two different ssRNA constructs. Prior to analyzing the dynamics on the µs–ms timescale it was necessary to confirm that perceived motional differences (as inferred from differences in kex values) between the complexes were not related to exchange between substrate/template-bound and free forms of P2 i.e. to simple on/off processes. To estimate the exchange contributions from these processes, knowledge of the affinities of the two GTP analogs (GMPPNP and GMPCPP) and the two different RNA constructs, used in this study, towards P2, was required.

In order to measure the binding affinity of the two 5-nt ssRNA constructs towards P2 we performed fluorescence anisotropy measurements using 5′-fluorescein tagged RNA. Fluorescein, in its singly protonated form (pKa = 6.4) has a reduced molar absorptivity (ε = 29000 M−1 cm−1) compared with the corresponding dianionic form (ε = 76900 M−1cm−1) (39) that is the dominant species above pH = 7. However, we chose to perform these experiments using the NMR buffer (pH = 6.5) for consistency with the relaxation studies for states 4, 5 and 6. In spite of the reduced fluorescence, changes in fluorescence anisotropy could be monitored with a high degree of accuracy and the resultant binding isotherms could be fit to Equation (6) to obtain apparent Kd values (Figure 5). These studies revealed that P2 has a higher affinity for 5′-UUUUC-3′ (Kd = 6.6 ± 0.4 µM) than for 5′-UUUCC-3′ (Kd = 50.0 ± 1.7 µM). No significant changes in binding affinity were noted in the presence of GMPPNP. Based on the binding affinities of the ssRNA templates and their concentration (~mM) in the NMR experiments one can estimate a kex = kon(CRNA+Kd) ~105 s−1 for exchange between the RNA-bound and free states of P2 assuming a diffusion limited on-rate. This value is about two orders of magnitude higher than the largest kex values (~103 s−1) described below. Thus these simple on/off processes are too fast to contribute to the kex values measured from the relaxation dispersion experiments.

Figure 5.
Binding affinity of the 5-nt ssRNA constructs towards P2. Data for 5′-UUUCC-3′ (black) and 5′-UUUUC-3′ (red) are shown. Solid lines represent theoretical curves obtained from fits to Equation (6) and circles represent the ...

The binding affinities of the two GTP analogs towards P2, as estimated from tryptophan fluorescence quenching (data not shown) were quite similar (GMPPNP, Kd = 317 ± 31 µM, GMPCPP, Kd = 271 ± 31 µM) and comparable to that seen in other RdRPs (40). Here again, the expected kex values for simple on-off processes were expected to be 50- to a 100-fold higher than the highest kex values obtained from the relaxation dispersion experiments.

Slow dynamics

In order to investigate dynamics on the µs-ms timescale in P2 we performed an analysis of the dependence of the Ile-Cδ1 13C,1H multiple-quantum (MQ) relaxation rates (ΓMQ,eff) on the strength (ν1) of an applied radiofrequency (RF) field (MQ relaxation dispersion) (33). These rates can be measured with high sensitivity in large proteins where traditional single-quantum (SQ) relaxation dispersion experiments are likely to fail (33). When the exchange rate (kex) is much greater than ν1, the ΓMQ,eff values are essentially field-independent. However, when kex and ν1 are of comparable magnitude, ΓMQ,eff becomes dependent on ν1, reaching a field-independent limiting value [the inherent relaxation rate, ΓMQ(∞)] in the limit ν1 > >kex. The dependence of ΓMQ,eff on ν1 is extremely complex and cannot, in general, be treated using the Carver-Richards formalism used for SQ relaxation (33). However, such an exact treatment requires high-quality data at multiple static magnetic fields. In the fast exchange regime, however, when the chemical shift differences between states A and B (see ‘Materials and Methods’ section) for the two nuclei participating in the MQ coherence, is small, a simple functional form may be used to extract kex values [Equation (4)] (34). We use this approach in the present case especially since the chemical shift changes seen (assuming that these changes are reflective of the difference between the two states) are sufficiently small so that the fast exchange condition kex >> (ΔωHΔωC)1/2 (where ΔωH and ΔωC are the chemical shift differences for 1H and 13C nuclei, respectively) was maintained in all of the cases analyzed below. Additionally, in this i.e. the fast exchange limit, MQ relaxation dispersion is uniquely sensitive to slow dynamics providing added incentive in choosing MQ relaxation over corresponding SQ relaxation measurements (41). Under the experimental conditions used here, kex values in the 100–2500 s−1 range can be determined with an acceptable degree of accuracy.

For apo-P2 (state 1) statistically significant kex values could be obtained for eight Ile residues by fitting the MQ dispersion data to Equation (4). These values that varied from 449 to 1690 s−1 (Table 2), could be broadly classified into two distinct groups—group I that had higher exchange rates (kex) implying motion on a faster timescale (mean kex = 1510 ± 168 s−1) and group II consisting of residues that were mobile on a slower timescale (mean kex = 600 ± 87 s−1).

Table 2.
Exchange rates (kex in s−1) determined from Ile [13Cδ1, 1Hδ1] multiple quantum relaxation dispersion

Group I consisted of two Ile residues that line one side of the entrance to the template tunnel (Ile37, Ile96) and one (Ile641) on the C-terminal domain (Figure 6a). kex values for the C-terminal Ile617 could not be extracted accurately due to extremely low S/N ratio for the corresponding resonance. However, visual inspection of the field-dependence of ΓMQ,eff suggested a kex value characteristic of a group I residue. In RdRPs, such as P2, that initiate RNA synthesis de novo, the absence of a primer requires additional interactions of the template ssRNA with the RdRP to stabilize the initiation complex. These interactions in P2 occur with a ‘priming loop’ located on the C-terminal domain and containing a conserved aromatic residue (Y630 in ϕ6 P2) (9). In addition, the accepted model for the exit of the newly synthesized daughter strands from the catalytic cavity involves motion of the C-terminal domain (9). It thus stands to reason that the conformational dynamics involving the C-terminal domain would be synchronized with those that allow the entry of ssRNA into the template tunnel. We therefore hypothesize that these faster motions are designed to facilitate entry of template and exit of the daughter chains in a coordinated fashion in line with the currently accepted model for RNA entry and egress (9). Extensive studies of the effects of mutations of key residues at the rim of the template tunnel and the C-terminal domain are available (42,43). However interpreting the effects of these mutations in terms of functional motions in a simple way is difficult. For example, the rim mutant K30A has reduced transcriptional activity but this may partially be explained by its reduced affinity for RNA. Replacing the priming loop 630YSW632 (on the C-terminal domain) by a smaller and more dynamic GSG sequence leads to enhanced transcription, though this increase may also be attributed to back-priming as opposed to de novo RNA synthesis (42,43). A similar effect is obtained using the mutant E634Q which is expected to lead to a more dynamic C-terminal domain due to disruption in a salt-bridge with K144. Since this mutant has smaller (~20%) back-priming activity compared to the GSG mutant, enhanced mobility of the C-terminal domain does seem to play a role in transcriptional efficiency (43).

Figure 6.
(a) Results of 13C, 1H multiple-quantum relaxation dispersion experiments. Ile δ1 positions that have detectable kex values are displayed on the tertiary structure of P2. Data for state 3 is shown as a representative example. Group I residues ...

Group II consisted of Ile164, Ile258, Ile488, Ile500 and Ile563. Of these residues, Ile488 (motif D), Ile500 (motif E) were located in and around the catalytic motifs on the palm and Ile563 on the base of the thumb (Figure 6a). Ile164 is located in the so-called plough region (Figure 6a). It has been shown that plough mutants enhance transcriptional activity though they do not have any significant effect on template binding (43). This is consistent with the nature of motions displayed by Ile164 with the kex being smaller than those of the group I residues. Additionally, given that the motion of Ile164 occurs on a similar timescale as the catalytic site residues suggests some degree of coupling of the plough region to the active site and therefore to the transcriptional activity. Ile258 is located on the diametrically opposite end of the fingers (Figure 6a) suggesting that the group II motions represent dynamic coupling of the outer edges of the fingers to the active site. Given that this motional timescale was slower than that seen for group I residues (putatively involved in RNA entry and egress), they could be expected to be rate limiting, if indeed they are catalytically relevant. A recent kinetic study on the poliovirus RdRP 3Dpol (29), suggested several steps involving conformational changes that are likely to play a role in catalysis. These included: (i) a conformational transition in the RdRP/NTP/RNA ternary complex to an activated state primed for catalysis (expected kex ~800 s−1), (ii) a transition from the activated complex to an activated product complex with pyrophosphate, the central reaction in the nucleotide addition cycle (expected kex ~895 s−1), and (iii) a deactivation of the product complex (kex ~500 s−1). The timescales associated with these three steps are consistent with the second, slower timescale seen in the present study (group II). The small differences in the predicted kex values between these three steps would make it difficult to pinpoint the exact conformational transition/s corresponding to the slower timescale. It is most likely that all of these transitions contribute to the distribution in kex values for the group II residues. However, the conjecture that these motions could be rate limiting in catalysis is reinforced by the detection of kex values in the 500–800 s−1 i. e. group II range for residues on several catalytic motifs, most notably in the Motif C residue Ile449 in the P2/Mg2+/GMPPNP complex (state 3) and in the stable initiation complex (state 6) discussed below.

For state 2 (P2 + Mg2+), as expected, no major changes in dynamic timescales are seen within group I (1516 ± 168 s−1) or group 2 (379 ± 157 s−1) residues. However, Ile500 that was a group II residue for state 1 now showed dynamics characteristic of group I (Table 2), accompanied by the largest Δδ1,2 [Equation (1)] between states 1 and 2 (19 Hz). This is not unexpected since Ile500 is within 4 Å of the divalent metal-binding sites. Comparatively large Δδ1,2 values were also seen for Ile164 (13 Hz), Ile449 (11 Hz) and Ile488 (14 Hz). All of these residues, with the exception of Ile164, were located at the catalytic center reinforcing the notion of long-range coupling of the plough region (where Ile164 is located) with the catalytic site.

For state 3, a similar timescale separation was seen for group I (1134 ± 203 s−1) and group II (427 ± 164 s−1) residues. However, Ile164 now displayed a higher kex value that was closer in magnitude to that for Ile500 (Table 2). It is possible that these residues comprised an additional intermediate timescale. Ile164 and Ile500 also displayed comparatively large Δδ2,3 values of 12 and 14 Hz, respectively. Here again, the correlated variation in kex and Δδ values suggested that conformational changes of the catalytic core were coupled to that of the plough region.

In state 4 (Figure 2b), the dead-end complex, all conformational dynamics involving Ile96 (Figure 6b), Ile164, Ile258, Ile449 and Ile563 were completely suppressed. Motion could only be detected for Ile488 and Ile500, with the former displaying kex values characteristic of an intermediate timescale (similar to Ile164 and Ile500 in state 3 above) and the latter similar to that seen for group I residues. Clearly, the dynamic nature of the dead-end complex was different from all other states both in the distribution as well as the magnitudes of the kex values. It has been suggested that P2 in this dead-end or stalled state can be reinitiated by supplementation with the missing NTPs (38). However, this reinitiated complex has a greatly reduced polymerization rate than that prior to stalling. Interestingly, the polymerization rate in the reinitiated complex had no measurable temperature dependence (in the 16−30°C range) (38) which suggests that the conformational sub-states likely giving rise to the RF-field dependent ΓMQ,eff were no longer thermally accessible for this state. This implied that the enzyme was transformed into a distinct conformation that was still capable of catalysis, but no longer efficient in doing so. The distinct conformation adopted by the stalled complex was also evident from Δδ3,4values that were, on average, far larger than corresponding values for the other ternary complexes, namely states 5 and 6 (Δδ3,5 and Δδ3A,6 discussed below, also see Supplementary Figure S4). Another observation unique to state 4 was that the Ile37 resonance showed extensive line broadening seemingly due to an unresolved doublet (Supplementary Figure S5). Based on ΔΓMQ,eff = ΓMQ,eff (31.25 Hz) – ΓMQ,eff (1000 Hz) = 15 s−1 (for the major component of the doublet), slow conformational dynamics persisted in this state for Ile37, though the dispersion curve could no longer be reliably fit assuming fast-exchange i. e. using Equation (4).

The distribution of kex values for the unstable initiation complex (state 5, see Figure 2b) was as expected, essentially similar to state 3 in their overall distribution—group I (Ile37, Ile96 and Ile500, kex = 1411 ± 149 s−1) and group II (Ile164, Ile258, Ile488 and Ile563, kex = 453±91 s−1). Ile164 (20 Hz) also displayed a relatively large Δδ3,5 value.

Analysis of the motion of states 1–5, however, did not allow the dissection of the motions that get altered simply by the formation of an initiation complex prior to start of the nucleotide addition cycle i.e. catalysis. In order to do so, data for two additional states (3A and 6), involving the GTP analog GMPCPP, that is non-cleavable between the α and β−phosphates, were analyzed.

While the data for the GMPCPP-bound states was of lower quality than the GMPPNP-bound states, some interesting motional characteristics emerged. For state 3A (Mg2+/GMPCPP, Table 1), while motional modes and kex values similar to state 3 were expected, substantial differences were seen (Table 2). This indicated that the nature of perturbations induced by the GTP-analogs GMPPNP and GMPCPP were different, a scenario confirmed by inspection of the Δδ3,3A values (Supplementary Figure S6). Motions involving Ile164, Ile258 and Ile563 were completely suppressed, adding credence to the hypothesis that the motions of these residues were coupled to the conformational fluctuations around the catalytic cavity at the substrate NTP binding sites. Given the fact that the affinities of P2 towards the two GTP analogs were about the same, as mentioned above, it is likely that these differences arose from subtle variations in the orientation of the triphosphate moiety at the active site possibly resulting in altered coupling of the active site to remote regions of the protein. This is discussed at length in the following section.

However, the motional characteristics of Ile37, Ile96 and Ile488 were similar to state 3 (Table 2), though small differences in kex values for Ile500 with this state was noted. While the kex value for Ile96 was similar to state 3, the ΔΓMQ,eff = ΓMQ,eff (31.25 Hz) – ΓMQ,eff (1000 Hz) = 3.8 s−1 (compared with 13.2 s−1 for state 3, Figure 6b) most likely due to alteration in ΔωH or ΔωC values or both rather than a simple population shift (with respect to state 3) given that Δδ3,3A value for this residue was quite small.

Finally for state 6 (Figure 2b), the ternary complex of Mg2+/GMPCPP/5′-UUUCC-3′, capable of forming a stable initiation complex was studied. In this state, field-dependent ΓMQ,eff could only be detected for Ile37 (characteristic of a group I residue), Ile449 (group II), Ile488 (group II) and Ile500, while no such effects were seen for Ile96, Ile164, Ile258 and Ile563. Thus, the only clear difference between the motional characteristics of the GMPCPP/Mg2+ (state 3A) and the stable initiation complex i.e. the GMPCPP/Mg2+/ssRNA (state 6) was the suppression of slow dynamics involving Ile96 (template entry) in the latter. This is not unexpected since the ssRNA is now likely locked in a stable complex. Unfortunately the weak signals for the C-terminal Ile617 and Ile641 residues did not allow a quantitative analysis of these resonances. Thus the nature of the motions assigned to facilitate the egress of the daughter strands could not be analyzed. However Ile37 (template entry) was still shown to display a kex value expected of a group I residue.


Based on the characterization of the seven states discussed above, the following observations can be made—the slow dynamics in P2 could be generally characterized by two distinct timescales—a faster timescale (group I, kex ~1200–1500 s−1) and a second, slower timescale (group II, kex ~500–800 s−1).

The faster timescale (group I) involved residues in close proximity to the template tunnel (Ile37 and Ile96) and on the C-terminal domain. These motions most likely assist in RNA translocation, facilitating the passage of the template through the tunnel and exit of the newly synthesized daughter strands by coordinated movements of the C-terminal domain. The nature of conformational dynamics in the C-terminal domain was also borne out by the large chemical shift changes in all the complexes with ssRNA—for Ile641, Δδ3,4 = 59 Hz, Δδ3,5 = 41 Hz and Δδ3A,6 = 38 Hz. Given the timescale of these motions, they are clearly not expected to be rate limiting. This is consistent with data obtained for both DNA (44) and RNA polymerases (29) where the translocation is not rate limiting in the process of polymerization. It has been suggested that the translocation step is coupled to pyrophosphate release from the enzyme. Rates of pyrophosphate release in have been found to be as slow as 1200 s−1 in T7 DNA polymerase (44) and as fast as 8100 s−1 in 3Dpol (29). The kex values of group I residues were within this range of timescales. Formation of a stable initiation complex lead to suppression of the motion of the fingertip Ile96 (Figure 6b) while that of the Ile37, that is closest to the ssRNA template, remained. This suggests that some motion around the template tunnel is required to thread the ssRNA through the RdRP even after the formation of an initiation complex. Ile96 was also no longer mobile in the stalled complex and motion involving Ile37 occurred on a slower exchange regime.

The second, slower motional mode (group II) that coincides with the catalytic timescale, involved several residues located in the catalytic motifs (motif C—Ile449, motif D—Ile488 and motif E—Ile500) and in the outer edge of the template tunnel (Ile164, Ile258 and Ile563). The distribution of kex values within these groups of residues suggests that there may be multiple motional modes involved in producing them. This fact is supported by the observation of selective damping of motions at the outer edge of the template tunnel upon binding GMPCPP, a behavior distinct from GMPPNP binding. This also suggests that minor perturbations at the active site can cause activation or deactivation of a subset of these modes involving distant regions of P2. It has been proposed that one of the conformational changes that occur on the catalytic timescale likely involves the reordering of the triphosphate moiety and concurrent changes in the enzyme (29). Given the differences induced by GMPPNP and GMPCPP both in static chemical shifts (in Ile145, Ile164, Ile449 and Ile500) and dynamics changes (in Ile164, Ile258 and Ile563) it is possible that this reorientation contributes in part to one component of the slower timescale. Indeed the largest differences in the binding mode of these two GTP analogs most likely occur due to the orientation of the triphosphate moiety. An analysis of GTP-bound structures of P2 (e.g. 1H10) reveals that the O atom (CH2 in GMPCPP) between the α and β-phosphates participates in a hydrogen bond with the Arg270 sidechain, additional hydrogen-bonds are also seen for the β- and γ-phosphates (with basic residues—Arg268, Arg270 and His328) defining their orientation. The conserved motif F (268–272) which has evolved to co-ordinate the triphosphate moiety does not contain any Ile residues, so is invisible to the analysis presented here. Another key observation that emerged from these studies was evidence of long-range coupling of the plough region (as reflected by Ile164) with the catalytic core region reinforcing observations from functional analyses (43).

Upon formation of a dead-end complex (state 4), motions of Ile449 on motif C are completely suppressed while motion of the motif D residue, Ile488 occurs on a distinct, faster timescale. This dead-end complex also seems structurally distinct from all other RNA-bound states. Interestingly, Ile488 displayed kex ~600 s−1 in all states we analyzed, including the dead-end complex, where it had a slightly higher kex value compared with the other states. An analysis of the available structures of a range of RdRPs indicate that the conformation of motif D is amongst the most variable (12). It has been suggested that this motif D, which bears a basic residue functioning as a general acid, undergoes a conformational change taking it into an optimal position to protonate the product pyrophosphate. Protonation of the pyrophosphate product leads to a large enhancement in the rate of nucleotidyl transfer (12,16). There are several Lys residues (476, 479 and 487) on motif D in ϕ6 P2, which could serve this purpose.

The case of the motif E residue Ile500 was rather curious. It displayed the largest variation in kex values (from 690 to 1811 s−1) over the several states, being slowest (and closest to the catalytic i.e. group II timescale) for states 1, 3, 3A and 6, and fastest (and closest to a group I timescale) for states 2, 4 and 5. The exact reason for this variation is unclear at the present time. Notably, Ile500 also showed significant chemical shift variations across the various states.

Another intriguing feature that was immediately evident was that several residues that were dynamic on the slow, µs–ms timescale (with significant kex values) across the states analyzed, were also dynamic on the fast, ps–ns timescale (with low η and consequently low An external file that holds a picture, illustration, etc.
Object name is gkq210i19.jpg values). These included Ile37, Ile96 near the template entry portal and Ile488, Ile500 at the substrate entry portal near the catalytic site. It is likely that motion on multiple timescales at the portal regions implies slow sampling of conformational states of the enzyme as well as local conformational variations allowing the processing of a wide range of substrate NTPs and template RNAs, accommodating differences both in overall size and shape as well local structural details.

Clearly, RdRPs in general, and P2 in particular, function as intricate molecular machines with many moving parts working in unison. This study that used 25 probes of molecular mobility represents a first step towards understanding these motions in a more quantitative way. We have only attempted to interpret the kex values in the present approach. Analysis of the chemical shifts and populations of the two states should also be highly informative (45–48), as would the extension of these studies to other methyl-bearing sidechains including Val, Leu, Met residues and finally to the backbone 15N nuclei. A drawback of our approach and indeed that of others is that we interpret field (ν1)-independent ΓMQ,eff values as lack of µs–ms timescale motion. RF-field-independent ΓMQ,eff may also occur if the differences in chemical shifts between the ground and excited states for 1H or 13C nuclei are vanishingly small i.e. ΔωH or ΔωC are 0. Thus, it is often useful to combine MQ dispersion measurements with corresponding SQ experiments. This, while easier to carry out in smaller, well-behaved proteins, is a non-trivial task in P2. In spite of these shortcomings, to the best of our knowledge this is the very first attempt to quantify dynamics along the catalytic cycle of this very important class of RNA processing enzymes. Additionally, the only complete kinetic analysis for the nucleotide addition cycle has been performed for the poliovirus RdRP (3Dpol) (29) that uses a primer to drive the nucleotide addition cycle. This is different from RdRPs like P2 that initiate RNA synthesis de novo. Thus, while the major dynamic modes contributing to the enzymatic activity may be similar, differences in the spatial as well as temporal features are likely present between the two sub-classes. It would therefore be important to use kinetic assays in de novo initiating RdRPs in conjunction with NMR analyses or conversely to extend our NMR experiments to RdRPs such as 3Dpol for which substantial kinetic data exist.


Supplementary Data are available at NAR Online.


National Science Foundation MCB0843141 (RG); National Institues of Health 5G12 RR03060 (partial support of the core facilities at CCNY); National Institutes of Health P41 GM66354 (partial support of NMR facilities at NYSBC). Funding for open access charge: NSF MCB0843141.

Conflict of interest statement. None declared.

Supplementary Material

[Supplementary Data]


The authors are grateful to Dr. Paul Gottlieb (CCNY) for his generous gift of the pLM687 plasmid. The authors thank Dr Lewis Kay (U of Toronto) and Dr Dmitry Korzhnev (U of Toronto) for the NMR pulse sequences for relaxation measurements and Dr Aneel Aggarwal (Mt. Sinai School of Medicine) for use of the Panvera Beacon 2000 spectrofluorimeter. The authors also thank Dr Andrea Piserchio (CCNY/NYSBC), Dr Leonard Mindich (PHRI), Dr Stewart Shuman (Sloan-Kettering Institute) and Dr Fabien Ferrage (CNRS, Ecole Normale Supérieure, Paris) for critical comments on this manuscript.


1. Ferrer-Orta C, Arias A, Escarmis C, Verdaguer N. A comparison of viral RNA-dependent RNA polymerases. Curr. Opin. Struct. Biol. 2006;16:27–34. [PubMed]
2. Bruenn JA. A structural and primary sequence comparison of the viral RNA-dependent RNA polymerases. Nucleic Acids Res. 2003;31:1821–1829. [PMC free article] [PubMed]
3. Bruenn JA. A closely related group of RNA-dependent RNA polymerases from double-stranded RNA viruses. Nucleic Acids Res. 1993;21:5667–5669. [PMC free article] [PubMed]
4. Makeyev EV, Grimes JM. RNA-dependent RNA polymerases of dsRNA bacteriophages. Virus Res. 2004;101:45–55. [PubMed]
5. Makeyev EV, Bamford DH. Replicase activity of purified recombinant protein P2 of double-stranded RNA bacteriophage ϕ6. EMBO J. 2000;19:124–133. [PubMed]
6. Makeyev EV, Bamford DH. The polymerase subunit of a dsRNA virus plays a central role in the regulation of viral RNA metabolism. EMBO J. 2000;19:6275–6284. [PubMed]
7. Mindich L. Packaging, replication and recombination of the segmented genome of bacteriophage ϕ6 and its relatives. Virus Res. 2004;101:83–92. [PubMed]
8. Poranen MM, Tuma R. Self-assembly of double-stranded RNA bacteriophages. Virus Res. 2004;101:93–100. [PubMed]
9. Butcher SJ, Grimes JM, Makeyev EV, Bamford DH, Stuart DI. A mechanism for initiating RNA-dependent RNA polymerization. Nature. 2001;410:235–240. [PubMed]
10. Salgado PS, Makeyev EV, Butcher SJ, Bamford DH, Stuart DI, Grimes JM. The structural basis for RNA specificity and Ca2+ inhibition of an RNA-dependent RNA polymerase. Structure. 2004;12:307–316. [PubMed]
11. Poranen MM, Salgado PS, Koivunen MR, Wright S, Bamford DH, Stuart DI, Grimes JM. Structural explanation for the role of Mn2+ in the activity of ϕ6 RNA-dependent RNA polymerase. Nucleic Acids Res. 2008;36:6633–6644. [PMC free article] [PubMed]
12. Cameron CE, Moustafa IM, Arnold JJ. Dynamics: the missing link between structure and function of the viral RNA-dependent RNA polymerase? Curr. Opin. Struct. Biol. 2009;19:1–7. [PMC free article] [PubMed]
13. Steitz TA. DNA polymerases: structural diversity and common mechanisms. J. Biol. Chem. 1999;274:17395–17398. [PubMed]
14. Ranjith-Kumar CT, Kao CC. Recombinant viral RdRps can initiate RNA synthesis from circular templates. RNA. 2006;12:303–312. [PubMed]
15. Biswal BK, Cherney MM, Wang M, Chan L, Yannopoulos CG, Bilimoria D, Nicolas O, Bedard J, James MN. Crystal structures of the RNA-dependent RNA polymerase genotype 2a of hepatitis C virus reveal two conformations and suggest mechanisms of inhibition by non-nucleoside inhibitors. J. Biol. Chem. 2005;280:18202–18210. [PubMed]
16. Castro C, Smidansky ED, Arnold JJ, Maksimchuk KR, Moustafa I, Uchida A, Gotte M, Konigsberg W, Cameron CE. Nucleic acid polymerases use a general acid for nucleotidyl transfer. Nature Struct. Mol. Biol. 2009;16:212–218. [PMC free article] [PubMed]
17. Arnold JJ, Vignuzzi M, Stone JK, Andino R, Cameron CE. Remote site control of an active site fidelity checkpoint in a viral RNA-dependent RNA polymerase. J. Biol. Chem. 2005;280:25706–25716. [PMC free article] [PubMed]
18. Bressanelli S, Tomei L, Rey FA, De Francesco R. Structural analysis of the hepatitis C virus RNA polymerase in complex with ribonucleotides. J. Virol. 2002;76:3482–3492. [PMC free article] [PubMed]
19. Di Marco S, Volpari C, Tomei L, Altamura S, Harper S, Narjes F, Koch U, Rowley M, De Francesco R, Migliaccio G, et al. Interdomain communication in hepatitis C virus polymerase abolished by small molecule inhibitors bound to a novel allosteric site. J. Biol. Chem. 2005;280:29765–29770. [PubMed]
20. Palmer AG., 3rd NMR probes of molecular dynamics: overview and comparison with other techniques. Annu. Rev. Biophys. Biomol. Struct. 2001;30:129–155. [PubMed]
21. Wolf-Watz M, Thai V, Henzler-Wildman K, Hadjipavlou G, Eisenmesser EZ, Kern D. Linkage between dynamics and catalysis in a thermophilic-mesophilic enzyme pair. Nature Struct. Mol. Biol. 2004;11:945–949. [PubMed]
22. Eisenmesser EZ, Millet O, Labeikovsky W, Korzhnev DM, Wolf-Watz M, Bosco DA, Skalicky JJ, Kay LE, Kern D. Intrinsic dynamics of an enzyme underlies catalysis. Nature. 2005;438:117–121. [PubMed]
23. Tzakos AG, Grace CR, Lukavsky PJ, Riek R. NMR techniques for very large proteins and RNAs in solution. Annu. Rev. Biophys. Biomol. Struct. 2006;35:319–342. [PubMed]
24. Tugarinov V, Kay LE. Methyl groups as probes of structure and dynamics in NMR studies of high-molecular-weight proteins. ChemBioChem. 2005;6:1567–1577. [PubMed]
25. Goger M, McDonnell JM, Cowburn D. Optimization of protein assignment with reduced cycle multidimensional NMR experiments. Spectroscopy. 2003;17:161–167.
26. Korzhnev DM, Kloiber K, Kay LE. Multiple-quantum relaxation dispersion NMR spectroscopy probing millisecond time-scale dynamics in proteins: theory and application. J. Am. Chem. Soc. 2004;126:7320–7329. [PubMed]
27. Tugarinov V, Kay LE. Relaxation rates of degenerate 1H transitions in methyl groups of proteins as reporters of side-chain dynamics. J. Am. Chem. Soc. 2006;128:7299–7308. [PubMed]
28. Tugarinov V, Sprangers R, Kay LE. Probing side-chain dynamics in the proteasome by relaxation violated coherence transfer NMR spectroscopy. J. Am. Chem. Soc. 2007;129:1743–1750. [PubMed]
29. Arnold JJ, Cameron CE. Poliovirus RNA-dependent RNA polymerase (3Dpol): pre-steady-state kinetic analysis of ribonucleotide incorporation in the presence of Mg2+ Biochemistry. 2004;43:5126–5137. [PMC free article] [PubMed]
30. Mindich L, Qiao X, Onodera S, Gottlieb P, Frilander M. RNA structural requirements for stability and minus-strand synthesis in the dsRNA bacteriophage ϕ6. Virology. 1994;202:258–263. [PubMed]
31. Boggs PT, Donaldson JR, Byrd RH, Schnabel RB. ODRPACK software for weighted orthogonal distance regression. ACM Trans. Math. Software. 1989;15:348–364.
32. Tugarinov V, Ollerenshaw JE, Kay LE. Probing side-chain dynamics in high molecular weight proteins by deuterium NMR spin relaxation: an application to an 82-kDa enzyme. J. Am. Chem. Soc. 2005;127:8214–8225. [PubMed]
33. Korzhnev DM, Kloiber K, Kanelis V, Tugarinov V, Kay LE. Probing slow dynamics in high molecular weight proteins by methyl-TROSY NMR spectroscopy: application to a 723-residue enzyme. J. Am. Chem. Soc. 2004;126:3964–3973. [PubMed]
34. Dittmer J, Bodenhausen G. Evidence for slow motion in proteins by multiple refocusing of heteronuclear nitrogen/proton multiple quantum coherences in NMR. J. Am. Chem. Soc. 2004;126:1314–1315. [PubMed]
35. Garcia de la Torre J, Huertas ML, Carrasco B. HYDRONMR: prediction of NMR relaxation of globular proteins from atomic-level structures and hydrodynamic calculations. J. Magn. Reson. 2000;147:138–146. [PubMed]
36. van Dijk AA, Makeyev EV, Bamford DH. Initiation of viral RNA-dependent RNA polymerization. J. Gen. Virol. 2004;85:1077–1093. [PubMed]
37. Kainov DE, Lisal J, Bamford DH, Tuma R. Packaging motor from double-stranded RNA bacteriophage ϕ12 acts as an obligatory passive conduit during transcription. Nucleic Acids Res. 2004;32:3515–3521. [PMC free article] [PubMed]
38. Vilfan ID, Candelli A, Hage S, Aalto AP, Poranen MM, Bamford DH, Dekker NH. Reinitiated viral RNA-dependent RNA polymerase resumes replication at a reduced rate. Nucleic Acids Res. 2008;36:7059–7067. [PMC free article] [PubMed]
39. Nygren J, Svanvik N, Kubista M. The interactions between the fluorescent dye thiazole orange and DNA. Biopolymers. 1998;46:39–51. [PubMed]
40. Arias A, Arnold JJ, Sierra M, Smidansky ED, Domingo E, Cameron CE. Determinants of RNA-dependent RNA polymerase (in)fidelity revealed by kinetic analysis of the polymerase encoded by a foot-and-mouth disease virus mutant with reduced sensitivity to ribavirin. J. Virol. 2008;82:12346–12355. [PMC free article] [PubMed]
41. Wang C, Palmer AG., 3rd Differential multiple quantum relaxation caused by chemical exchange outside the fast exchange limit. J. Biomol. NMR. 2002;24:263–268. [PubMed]
42. Laurila MR, Salgado PS, Stuart DI, Grimes JM, Bamford DH. Back-priming mode of ϕ6 RNA-dependent RNA polymerase. J. Gen. Virol. 2005;86:521–526. [PubMed]
43. Sarin LP, Poranen MM, Lehti NM, Ravantti JJ, Koivunen MR, Aalto AP, van Dijk AA, Stuart DI, Grimes JM, Bamford DH. Insights into the pre-initiation events of bacteriophage ϕ6 RNA-dependent RNA polymerase: towards the assembly of a productive binary complex. Nucleic Acids Res. 2009;37:1182–1192. [PMC free article] [PubMed]
44. Patel SS, Wong I, Johnson KA. Pre-steady-state kinetic analysis of processive DNA replication including complete characterization of an exonuclease-deficient mutant. Biochemistry. 1991;30:511–525. [PubMed]
45. Hansen DF, Vallurupalli P, Kay LE. Measurement of methyl group motional parameters of invisible, excited protein states by NMR spectroscopy. J. Am. Chem. Soc. 2009;131:12745–12754. [PubMed]
46. Vallurupalli P, Hansen DF, Kay LE. Structures of invisible, excited protein states by relaxation dispersion NMR spectroscopy. Proc. Natl Acad. Sci. USA. 2008;105:11766–11771. [PubMed]
47. Vallurupalli P, Hansen DF, Kay LE. Probing structure in invisible protein states with anisotropic NMR chemical shifts. J. Am. Chem. Soc. 2008;130:2734–2735. [PubMed]
48. Hansen DF, Vallurupalli P, Lundstrom P, Neudecker P, Kay LE. Probing chemical shifts of invisible states of proteins with relaxation dispersion NMR spectroscopy: how well can we do? J. Am. Chem. Soc. 2008;130:2667–2675. [PubMed]

Articles from Nucleic Acids Research are provided here courtesy of Oxford University Press