An inserted Gly residue fine tunes dynamics between mesophilic and thermophilic ribonucleases H

Abstract

Dynamic processes are inherent properties of proteins and are crucial for a wide range of biological functions. To address how changes in protein sequence and structure affect dynamic processes, a quantitative comparison of microsecond-to-microsecond time scale conformational changes, measured by solution NMR spectroscopy, within homologous mesophilic and thermophilic ribonuclease H (RNase H) enzymes is presented. Kinetic transitions between the observed major state (high population) and alternate (low population) conformational state(s) of the substrate-binding handle region in RNase H from the mesophile Escherichia coli (ecRNH) and thermophile Thermus thermophilus (ttRNH) occur with similar kinetic exchange rate constants, but the difference in stability between exchanging conformers is smaller in ttRNH compared to ecRNH. The altered thermodynamic equilibrium between kinetically exchanging conformers in the thermophile is recapitulated in ecRNH by the insertion of a Gly residue within a putative hinge between α-helices B and C. This Gly insertion is conserved among thermophilic RNases H, and allows the formation of additional intrahelical hydrogen bonds. A Gly residue inserted between α-helices B and C appears to relieve unfavorable interactions in the transition state and alternate conformer(s) and represents an important adaptation to adjust conformational changes within RNase H for activity at high temperatures.

Protein function relies on a combination of time-average static structure and time-dependent conformational fluctuations. Intrinsic dynamic properties influence many biological processes, including folding, ligand binding, catalysis, and allosterism (Wand 2001; Akke 2002; Benkovic and Hammes-Schiffer 2003; Kern and Zuiderweg 2003; Palmer 2004; Tousignant and Pelletier 2004; Kay 2005). Dynamic processes have been shown to be (partially) rate-limiting for catalysis, and differences in the time scale of motion are directly correlated with differences in the catalytic efficiency of homologous enzymes (Williams and McDermott 1995; Kohen and Klinman 2000; Codreanu et al. 2002; Cole and Loria 2002; Wolf-Watz et al. 2004). Several recent studies have highlighted the importance of individual amino acid residues in modulating dynamic processes within proteins (Eisenmesser et al. 2005; Kovrigin and Loria 2006). However, detailed mechanisms by which local interactions within the protein structure control the kinetic and thermodynamic properties of dynamic processes remain largely unknown.

Ribonucleases H (RNases H, EC 3.1.26.4) from the mesophile Escherichia coli (ecRNH) and thermophile Thermus thermophilus (ttRNH) share significant amino acid sequence and structural similarity, but display differences in thermal stability and catalytic activity. Typical of other mesophilic and thermophilic proteins (Jaenicke and Böhm 1998; Kumar et al. 2001; Vieille and Zeikus 2001), ttRNH is more resistant to thermal (and chemical) denaturation (Kanaya and Itaya 1992; Hollien and Marqusee 1999a,b) and is significantly less catalytically active than ecRNH at mesophilic temperatures (Kanaya and Itaya 1992). One source for these differences may stem from altered dynamic properties between these enzymes: Several regions display localized differences in dynamics on the picosecond-to-nanosecond (ps–ns) and microsecond-to-millisecond (μs–ms) time scales (Butterwick et al. 2004).

Chemical exchange line broadening is observed for nuclear spins in a highly basic loop between α-helices C (α_C) and D, termed the handle (see Fig. 1), in both ecRNH and ttRNH (Mandel et al. 1995, 1996; Kroenke et al. 1998; Yamasaki et al. 1998; Butterwick et al. 2004), suggesting that μs–ms time scale conformational changes in this region are an integral property of both enzymes. Residues in α_C and the handle are important for RNA:DNA hybrid recognition (Kanaya et al. 1991), although they are not essential for catalytic activity (Keck and Marqusee 1995). Nuclear magnetic resonance (NMR) spectroscopy and circular dichroism studies suggest that the handle changes conformation with respect to the catalytic domain upon complex formation (Oda et al. 1993; Iwai et al. 1996). The structural details of these transitions remain elusive; however, based on chemical shift changes upon RNA:DNA hybrid titration and models of the substrate-bound complex, residues between α_B and α_C have been suggested as a possible hinge in ecRNH (Oda et al. 1993). This region contains a single amino acid insertion in ttRNH relative to ecRNH: Gly85 is in a left-handed helical conformation (backbone dihedral angles ϕ = 70° and ψ = −9°), and allows an additional hydrogen bond between the Gly85 carboxyl and Gly89 amide, forming the N-terminal cap of α_C (Ishikawa et al. 1993d).

Figure 1.

Sequence and structural alignments of RNases H. (A) ClustalW (Thompson et al. 1994) multiple sequence alignment of RNase H from the mesophilic bacterium E. coli (37°C) and several thermophilic bacteria: Chlorobium tepidum (47°C), Moorella thermoacetica (58°C), T. thermophilus (70°C), and Thermoanaerobacter tengcongensis (75°C). The optimal growth temperature for each organism is indicated in parentheses. Conserved active site and substrate-binding residues are highlighted in blue; the inserted Gly is shown in red. α-Helical (α_A–α_E) and β-strand (β₁–β₅) secondary structural elements and the handle in WT ecRNH are indicated above the alignment. (B) Structural superposition of WT ecRNH (blue, PDB 2RN2), WT ttRNH (orange, PDB 1RIL), and iG80b ecRNH (gray, PDB 1GOA) using backbone atoms along core helices α_A and α_D. Structural elements from (A) and the N and C termini are indicated. Ribbon diagrams for this and subsequent figures were drawn using MOLMOL (Koradi et al. 1996).

Using NMR spectroscopy, a quantitative comparison of the μs–ms time scale dynamics of ecRNH and ttRNH is presented. Extending previous results (Butterwick et al. 2004), backbone amide ¹⁵N R _1ρ relaxation dispersion experiments show that conformational changes within the handle of both enzymes occur with similar kinetic exchange rate constants at 300 K. Analysis of the relaxation dispersion suggests an altered thermodynamic equilibrium among exchanging conformers exists between ecRNH and ttRNH. Reciprocal Gly insertion/deletion mutations within the putative α_B/α_C hinge are used to probe the role of this residue in modulating conformational changes within RNases H. Insertion of Gly80b into ecRNH between Gln80 and Trp81 (iG80b) or deletion of Gly85 from ttRNH (dG85) significantly alters μs–ms time scale dynamic processes compared to the wild-type (WT) enzymes. ¹⁵N R _1ρ dispersion experiments suggest that the thermodynamic distribution observed for WT ttRNH is largely recapitulated in the iG80b ecRNH mutant. Thus, the Gly insertion represents an important contribution in the adaptation of conformational changes within RNase H for activity at high temperatures.

Results

Gly insertion is conserved among thermophilic RNases H

Sequence alignments for mesophilic and thermophilic RNases H highlight several amino acid substitutions that may confer thermal stability or alter catalytic activity. More than 100 bacterial RNase H sequences with a conserved handle are found in the Swiss-Prot database, including four from thermophilic organisms (Fig. 1A). In addition to several amino acid substitutions known to promote increased thermal stability in ecRNH, for example H62P and V74L (Kimura et al. 1992b; Ishikawa et al. 1993b), the Gly insertion between α_B and α_C is conserved in thermophilic RNases H. Furthermore, this insertion is largely absent from mesophilic sequences: Only RNase H enzymes from mesophiles Lactobacillus plantarum and Deinococcus radiodurans contain this insertion.

Structural and functional consequences of Gly insertion

Although the Gly insertion is conserved in thermophilic RNases H, insertion of Gly80b into ecRNH provides only a moderate increase in thermal stability of 0.1 kJ/mol (Kimura et al. 1992b). However, this insertion leads to a drastic decrease in catalytic efficiency, increasing the binding affinity by 50-fold and decreasing the catalytic rate constant (k _cat) by 10-fold at 30°C (Ishikawa et al. 1993c), such that the activity of iG80b ecRNH is similar to the activity of ttRNH (Kanaya and Itaya 1992). Structurally, Gly80b is nicely accommodated between α_B and α_C, and adopts a similar left-handed helical conformation (ϕ = 52°, ψ = 46°) as Gly85 in ttRNH (Fig. 1B). In addition to the expected N-terminal cap hydrogen bond between Gly80b and Asn84, another hydrogen bond is present between the Ile78 carboxyl and Gly80b amide forming the C-terminal cap of α_B (Ishikawa et al. 1993c). The additional Gly also shifts the Gly77 carboxyl–Ile82 amide hydrogen bond to Gly77–Trp81. Given the minimal structural perturbations and marginal stabilization induced by the Gly80b insertion, the drastic decrease in activity may be due to modification of dynamic processes involving the substrate-binding handle.

Backbone amide ¹H and ¹⁵N resonances for the iG80b ecRNH and dG85 ttRNH mutant enzymes were assigned by comparison to WT spectra complemented with an HMQC-NOESY-HSQC spectrum recorded at 300 K. Chemical shift differences between WT and mutant spectra are small outside the site of mutation, suggesting that insertion or removal of the Gly residue does not significantly alter the protein structure. In addition, NOESY cross-peaks are observed between backbone amide protons separated by one or two residues throughout both α_B and α_C, confirming the helical nature of these regions (data not shown). The largest ¹⁵N chemical shift differences between WT and iG80b ecRNH are for residues Thr79 and Ile82: δ_iG80b−δ_WT = 4.94 ppm and −3.77 ppm, respectively. These shift changes are the correct sign and magnitude as expected for gain of a hydrogen bond involving the Ile78 carboxyl and switch of the hydrogen bond involving the Ile82 amide (Xu and Case 2002).

Interestingly, the equivalent residues also exhibit similar ¹⁵N chemical shift differences between WT and dG85 ttRNH, but with the opposite sign: δ_dG85−δ_WT = −4.78 ppm for Thr83 and 3.89 ppm for Leu87. Chemical shift changes for residues in α_B, α_C and the handle between WT and mutant ecRNH and ttRNH are negatively correlated overall (Fig. 2A). Not only are the changes in chemical shift similar in magnitude for these proteins, but the iG80b ecRNH chemical shifts resemble WT ttRNH, and dG85 ttRNH shifts resemble WT ecRNH (Fig. 2B). Thus, the hydrogen bonding pattern and backbone geometry is similar between WT ttRNH and iG80b ecRNH and between WT ecRNH and dG85 ttRNH. These results are in slight disagreement with the crystal structure of WT ttRNH (Fig. 2C) (Ishikawa et al. 1993d): Based on chemical shift changes, hydrogen bonds would be expected between carboxyls of Phe82 and Gly77 and amides of Gly85 and Leu87. These interactions appear to be present in solution and slight packing rearrangements may prevent these hydrogen bonds from being recognized in the crystalline state.

Figure 2.

Structural consequences of the Gly insertion. (A) Chemical shift changes between WT and iG80b ecRNH (Δδ_ec) and WT and dG85 ttRNH (Δδ_tt) are compared as Δδ = δ_mutant−δ_WT for backbone amide (left) ¹H and (right) ¹⁵N nuclei. Residues in α_B, α_C and the handle (70–100 in ecRNH) are indicated by open circles; the correlation coefficient for these residues are −0.63 for ¹H and −0.84 for ¹⁵N. (B) Overlay of ¹H–¹⁵N correlation spectra for iG80b ecRNH and dG85 ttRNH with (left) WT ttRNH and (right) WT ecRNH highlighting the positions of Thr79 (ecRNH) and Thr83 (ttRNH). (C) Main-chain hydrogen bonds in the α_B/α_C hinge and handle are shown in green for WT ecRNH (blue, PDB 2RN2), iG80b ecRNH (gray, PDB 1GOA), and WT ttRNH (orange, PDB 1RIL). Additional hydrogen bonds seen in the crystal structures of iG80b ecRNH and WT ttRNH are shown in magenta.

Chemical exchange is more pronounced in the presence of the inserted Gly

Dynamic processes on the μs–ms time scale were identified by the excess contribution to backbone amide ¹⁵N transverse relaxation (R ₂) from chemical exchange line broadening (R _ex). A TROSY-based Hahn-echo experiment (Wang et al. 2003) was used to measure R _ex for WT ecRNH and mutant RNases H at a static magnetic field of 14.1 T and a temperature (T) of 300 K (Fig. 3); data for WT ttRNH have been reported previously (Butterwick et al. 2004). In WT ecRNH, two residues show significant R _ex (R _ex ≥ 2.5 sec⁻¹): Lys60 in the α_A/β₄ loop and Trp90 in the handle (Mandel et al. 1995, 1996; Kroenke et al. 1998; Yamasaki et al. 1998). WT ttRNH is characterized by an increase in both the number of residues influenced by chemical exchange and the magnitude of R _ex. Specifically, Lys80, Thr83, Gly85, and Trp86 surrounding the α_B/α_C hinge, Trp95 and Arg96 within the handle, and His124 and Val 126 in β₅ have significant values of R _ex at 300 K (Butterwick et al. 2004).

Figure 3.

Chemical exchange line broadening in WT and mutant RNases H. Backbone amide ¹⁵N R _ex at 300 K, 14.1 T for WT ecRNH, WT ttRNH, iG80b ecRNH, and dG85 ttRNH are shown. Residues with significant chemical exchange line broadening (R _ex ≥ 2.5 sec⁻¹) are indicated by open circles. Vertical dashed lines indicate the sites of mutation. Secondary structural elements are diagrammed at the top of the figure. Data for WT ttRNH are from Butterwick et al. (2004).

Insertion of Gly80b into ecRNH or deletion of Gly85 from ttRNH within the α_B/α_C hinge significantly alters chemical exchange line broadening. iG80b ecRNH shows an increase in both the number of residues influenced by chemical exchange line broadening and magnitude of R _ex for residues within the handle and α_B/α_C hinge. In particular, Ile82, Arg88, Lys95, Lys99, and Asn100 show significant R _ex not seen in the WT enzyme, while Trp90 shows a significant increase in R _ex (over 10-fold). Relative to changes in the handle, chemical exchange for Lys60 is only moderately perturbed (decreased by 0.8-fold) in the iG80b ecRNH mutant. Conversely, dG85 ttRNH shows a significant decrease in R _ex for residues in both the handle and β₅. Specifically, residues near the Gly85 deletion no longer show significant R _ex, and Trp95 in the handle and His124 and Val126 in β₅ show decreased R _ex, relative to the WT enzyme. Importantly, changes in R _ex for both mutants are not localized solely to the site of mutation, but are distributed throughout the handle and other regions of the protein, suggesting that μs–ms time scale motions are highly coupled events that affect much of the enzyme.

Temperature dependence of chemical exchange

The temperature dependence of R _ex provides information on the time scale and energetics of conformational transitions (Mandel et al. 1996; Evenäs et al. 1999; Butterwick et al. 2004). R _ex measurements were repeated at temperatures between 285 K and 310 K for WT ecRNH and mutant RNases H; data for WT ttRNH have been reported previously (Butterwick et al. 2004). Values of R _ex increase with decreasing temperature in both WT enzymes, indicating that chemical exchange is intermediate to fast on the chemical shift time scale (Fig. 4A; Mandel et al. 1996^; Butterwick et al. 2004). In addition, although R _ex is not a kinetic rate constant per se, the temperature dependence of R _ex displays Arrhenius behavior over the temperature ranges studied, where the thermophilic enzyme is characterized by a larger “apparent” activation energy barrier (Ê _a) (Fig. 4B; Butterwick et al. 2004). At T ≤ 290 K, Ile82 and Asn100 have significant R _ex in WT ecRNH, but accurate Ê _a could not be determined for these residues (see Materials and Methods).

Figure 4.

Temperature dependence of chemical exchange line broadening. (A) R _ex measured at multiple temperatures and 14.1 T and (B) Arrhenius plots are shown for (left) Trp90 in WT and iG80b ecRNH and (right) Trp95 in WT and dG85 ttRNH. Open and closed symbols represent data for WT and mutant RNases H, respectively. Apparent activation energies, Ê _a, derived from the slope of the lines in B are listed in Table 1. Square symbols represent data that deviate from a linear Arrhenius relationship, and are not included for determining Ê _a. Data for WT ttRNH are from Butterwick et al. (2004).

In a similar manner to the WT enzymes, R _ex in both mutants increases with decreasing temperature (Fig. 4A). However, despite the differences in R _ex between WT and mutant enzymes, Ê _a remain strikingly similar (Fig. 4B; Table 1) This is not true for all residues: His124 and Val126 have depressed Ê _a in dG85 ttRNH relative to WT ttRNH. Amide nuclei in these residues are 14–16 Å away from the site of mutation, and backbone nuclei between the site of mutation and these residues are not preferentially broadened. Thus, indirect effects, such as time-dependent aromatic ring current shifts arising from conformational dynamics within the handle, are unlikely to contribute to the broadening of ¹⁵N nuclei in the loop, particularly given the low magnetogyric ratio of ¹⁵N. Thus, these results suggest coupling of motional processes between the handle and β₅ near the active site. Removal of Gly85 from ttRNH likely disrupts this connection, but leaves motion within the handle relatively unperturbed. Although the relatively large uncertainty in Ê _a for dG85 ttRNH renders quantitative comparisons difficult, the similarity between Trp95 in WT and dG85 ttRNH suggests that interactions within the handle determine the kinetic and thermodynamic properties of motion. In contrast to the WT proteins, R _ex for both mutants deviate from Arrhenius behavior at low temperatures; these data are not included in the determination of Ê _a.

Table 1.

Apparent activation energies for WT and mutant RNases H

Relaxation dispersion of backbone amide nuclei

To further characterize the exchange kinetics within WT and mutant RNases H, backbone amide ¹⁵N R ₂(ω _e) relaxation dispersion was measured at a temperature of 300 K for WT ecRNH, WT ttRNH, and iG80b ecRNH using R _1ρ spin relaxation techniques. The magnitude of exchange broadening in the dG85 ttRNH mutant is too small for accurate relaxation dispersion measurements (see Materials and Methods). All dispersion data are well described by a two-state exchange model

characterized by a kinetic exchange rate constant k _ex = k _AB + k _BA, where k _AB and k _BA are the forward and reverse rate constants between exchanging conformers A and B. On- and off-resonance R _1ρ experiments (Akke and Palmer 1996; Mulder et al. 1998; Massi et al. 2004) were combined to span effective fields of ω _e ≈ 10²–10⁴ sec⁻¹ at static magnetic fields of 11.7 and 18.8 T. For all exchanging residues in WT ecRNH and ttRNH, R _ex(ω _e) is barely suppressed at the highest ω _e obtained, such that k _ex >> 10⁴ sec⁻¹ for residues in both enzymes (Fig. 5; Table 2). In this regime, the individual forward and reverse rate constants and the thermodynamic and structural characteristics of the exchange process cannot be separated and only k _ex and ϕ _ex = p _A p _BΔω ² can be determined, where p _A and p _B are the fractional populations of the conformers A and B, respectively (p _A ≥ p _B), and Δω is the chemical shift difference between exchanging conformers (Palmer 2004).

Figure 5.

R _1ρ relaxation dispersion for backbone amide ¹⁵N nuclei. R ₂(ω _e) dispersion results are shown for residues Trp90 and Lys60 in WT and iG80b ecRNH, and Trp95 and Val126 in WT ttRNH at 300 K. Closed and open symbols represent data collected at 11.7 and 18.8 T, respectively.

Table 2.

Relaxation dispersion parameters for WT and mutant RNases H at 300 K

In WT ecRNH, R _ex(ω _e) could be quantified for Lys60 in the α_D/β₄ loop, and Trp90 and Asn100 in the handle. Residues in the handle were individually fit with similar k _ex ≈ 40,000–45,000 sec⁻¹; subsequent global analysis of these residues yielded k _ex = (42 ± 8) × 10³ sec⁻¹. In agreement with its reduced Ê _a, Lys60 exhibited much faster kinetics with k _ex = (100 ± 23) × 10³ sec⁻¹. Surprisingly, residues in WT ttRNH exhibited very similar exchange kinetics to residues in WT ecRNH. Individual fits of residues Lys80, Phe82, Thr83, Gly85, Trp86, Gly89, and Trp90 near the α_B/α_C hinge and Trp95, Arg96 and Arg106 in the handle region were characterized by k _ex ≈ 30,000–50,000 sec⁻¹. Interestingly, residues His124 and Val126 in β₅ were also individually fit with k _ex ≈ 40,000 sec⁻¹, suggesting a common motional process affects residues in the handle and near the active site. In a similar manner for WT ecRNH, these residues could be globally described by k _ex = (40 ± 4) × 10³ sec⁻¹. Thus, the larger values of R _ex for residues in WT ttRNH are due to approximately threefold larger ϕ _ex values and either the population distribution or chemical shift differences between the exchanging conformers are altered between the two enzymes.

In a similar manner to the WT enzymes, chemical exchange for all residues in iG80b ecRNH are fast on the chemical shift time scale (Fig. 5; Table 2). Residues in α_C and the handle region were generally individually fit with slower kinetics than WT ecRNH with k _ex ≈ 15,000–40,000 sec⁻¹; subsequent global analysis of Trp81, Ile82, Arg88, Trp90, Thr92, Lys95, Lys99, and Asn100 yielded k _ex = (27 ± 2) × 10³ sec⁻¹. The reduced k _ex does not completely account for the increased R _ex relative to WT ecRNH, and residues in iG80b ecRNH are fit with approximately fourfold larger ϕ _ex. Relaxation dispersion for Lys60, however, yields k _ex = (101 ± 30) × 10³ sec⁻¹, which is identical within experimental uncertainties to the value obtained for WT ecRNH. Thus, reduced R _ex for Lys60 in iG80b ecRNH compared to WT is due to an ∼0.8-fold reduction in ϕ _ex.

Interpreting chemical exchange effects

In the fast exchange regime, only k _ex and ϕ _ex can be determined from relaxation dispersion measurements; consequently, the underlying energetics and chemical shift changes of the conformational transition cannot be determined. However, ratios of k _ex and ϕ _ex provide information on the relative changes in the free energy difference between exchanging conformers (ΔΔG) and between transition states (ΔΔG ^‡) for two proteins. Assuming that k _AB and k _BA are described by transition state theory, k _AB = k _oexp[−ΔG ^‡/RT] and k _BA = k _oexp[−(ΔG ^‡−ΔG)/RT], and that site populations follow a Boltzmann distribution, p _B/p _A = exp[−ΔG/RT], then

where the subscripts 1 and 2 represent the two proteins of interest, k _o is the preexponential factor for k _AB and k _BA, R is the universal gas constant, and ΔG and ΔG ^‡ are the energy differences of the minor conformer and transition state relative to the major conformer, respectively, and ΔΔG = ΔG ₁ − ΔG ₂ and ΔΔG ^‡ = ΔG ₁ ^‡ − ΔG ₂ ^‡. The second equalities in Equations 2 and 3 arise if p _A − p _B ≈ 1 for either protein.

Given the small chemical shift changes between WT and iG80b ecRNH outside the site of mutation, the exchanging conformer structures probably are similar for both proteins, such that Δω _WT ≈ Δω _iG80b. In this case, using R _1ρ results for Trp90 (δ_N,iG80b − δ_N,WT = 0.41 ppm), ΔΔG = ΔG _iG80b − ΔG _WT = −3.6 ± 0.1 kJ/mol and ΔΔG ^‡ = ΔG ^‡ _iG80b − ΔG ^‡ _WT = −2.5 ± 0.5 kJ/mol at 300 K. Thus, insertion of Gly80b reduces the stability difference between exchanging conformers, increasing p _B relative to WT ecRNH (Fig. 6). The decreased stability difference between conformers is slightly larger than the decrease in activation energy, leading to slower kinetics in the mutant enzyme. Virtually identical results are obtained with data from Asn100.

Figure 6.

Summary of conformational changes in RNases H. Energy diagram comparing the relative stabilities and activation energies for exchanging conformers in WT ecRNH (blue), WT ttRNH (orange), and iG80b ecRNH (gray) at 300 K. Diagram is drawn to scale assuming p _B = 1% for WT ecRNH and k _o = k _B T/h, where k _B and h are Boltzmann's and Planck's constants, respectively.

The significant amino acid conservation and structural similarity between WT ecRNH and ttRNH also suggest that Δω probably are similar between these enzymes. Using R _1ρ results for Trp90 (ecRNH) and Trp95 (ttRNH) (δ_N,ttRNH − δ_N,ecRNH = 1.03 ppm), yields ΔΔG = ΔG _WT,Tt − ΔG _WT,Ec = −2.7 ± 0.1 kJ/mol and ΔΔG ^‡ = ΔG ^‡ _WT,Tt − ΔG ^‡ _WT,Ec = −2.6 ± 0.6 kJ/mol at 300 K. Similar exchange kinetics are observed between WT enzymes because ΔΔG ≈ ΔΔG ^‡; however, the minor conformer is stabilized relative to the major conformer in the thermophilic enzyme leading to a larger p _B (Fig. 6). Most surprisingly, the increase in transition state and minor conformer stabilities seen in WT ttRNH is similar to that conferred by the Gly80b insertion into ecRNH.

Discussion

Conformational changes are ubiquitous in biology. While structural studies have identified numerous cases in which structural changes occur during a biological process, the roles of specific residues and local interactions in controlling thermodynamics and kinetics of structural transitions remain largely uncharacterized. In the present work, NMR spectroscopy was used to characterize μs–ms time scale motional processes in homologous mesophilic and thermophilic RNase H enzymes. Comparison of the conformational dynamic properties of WT and Gly insertion/deletion mutants was used to probe the role of local interactions surrounding a putative hinge in modulating dynamics within the substrate-binding handle.

A Gly insertion between α_B and α_C is conserved in thermophilic RNases H that utilize a handle region for RNA:DNA hybrid recognition, suggesting that this residue is important for biological function at high temperatures. This insertion is a focal point for additional intrahelical hydrogen bonds that marginally stabilize the protein. Chemical shifts surrounding this region are similar between WT ttRNH and iG80b ecRNH and between WT ecRNH and dG85 ttRNH, suggesting that interactions within the iG80b ecRNH mutant are more “thermophilic-like” while those in dG85 ttRNH are more “mesophilic-like.” The similarity between WT ttRNH and iG80b ecRNH is further reflected in their reduced catalytic efficiencies relative to WT ecRNH (Kanaya and Itaya 1992; Ishikawa et al. 1993c). dG85 ttRNH, however, does not show significantly increased catalytic activity at low temperatures (to resemble WT ecRNH), possibly due to the uncoupling of conformational changes between the handle and β₅ near the active site (J.A. Butterwick and A.G. Palmer III, unpubl.). Thus, while the crystallographic structures of WT ttRNH and iG80b ecRNH display slight differences, both proteins likely incorporate the same additional hydrogen bonds in solution, compared to WT ecRNH and dG85 ttRNH.

Currently, the structural details of alternative conformations sampled in solution and leading to chemical exchange line broadening are unknown. Differences in R _ex between WT and mutant RNases H are spread throughout residues in α_C and the handle, suggesting that conformational changes within these regions are highly coupled. Large R _ex are not observed for most residues, indicating that Δω are small and the local structure in α_C and the handle is maintained. Because chemical exchange line broadening is not observed for residues Thr79, Gly80b and Asn84 in iG80b ecRNH, the additional hydrogen bonds most likely are maintained during conformational changes about this region. Residues Trp90 and Asn100 may serve as hinges for motions of the handle: Trp90 is the first and Asn100 is the last residue in the handle, both Trp90 and Asn100 show significant increase in R _ex in the iG80b ecRNH mutant, and both Trp90 and Asn100 have left-handed helical backbone dihedral angles in the observed major conformation.

The hypothesis that chemical exchange reflects movement of the entire handle as a whole is supported by the dynamic properties of a thermostabilized ecRNH quintuple G23A/H62P/V74L/K95G/D134H mutant (Yamasaki et al. 1998). Mutation of the left-handed Lys95 to Gly is well accommodated within the handle of ecRNH (Ishikawa et al. 1993a) and significantly stabilizes the protein (Kimura et al. 1992a). While no global changes in dynamics were observed in the quintuple mutant compared to the WT protein, several residues surrounding the sites of mutation experienced a change in dynamics across multiple time scales. Interestingly, an increase in R _ex was observed for residues Ile82, Trp90, Lys91, and Asn100 in the quintuple mutant (Yamasaki et al. 1998), although to a lesser extent than seen in iG80b ecRNH. Lys95 is 5–14 Å away from these residues, yet removal of its bulky side chain likely alters the thermodynamics or kinetics of structural transitions across the handle.

The remarkable similarity between the free energy profiles of iG80b ecRNH and WT ttRNH (Fig. 6) suggests that the Gly insertion plays a major role in modulating conformational dynamics of the handle region of RNase H. In iG80b ecRNH, both the transition and minor states are stabilized by a similar extent relative to WT ecRNH. Thus, the Gly insertion does not drastically alter the kinetic rate constants, but rather adjusts the thermodynamic equilibrium between conformers. Lacking a C^β atom, Gly occupies approximately threefold more Ramachandran space compared to other amino acids. This structural plasticity appears to alleviate unfavorable interactions in the transition state and minor conformer(s) in RNase H.

For RNases H, the kinetics of the chemical exchange process are likely too fast to be rate limiting for enzymatic activity. Previous characterization of the WT proteins using a nonanucleotide hybrid substrate at 30°C yielded k _cat ≈ 1.5 and 0.3 sec⁻¹ for WT ecRNH and WT ttRNH, respectively (Kanaya et al. 1990; Kanaya and Itaya 1992). Given the current results, the lower limit for WT ecRNH is p _B ≈ 1%, corresponding to k _AB ≈ 400 sec⁻¹ and k _BA ≈ 41,600 sec⁻¹, which are much faster than k _cat. Thus, the altered thermodynamic distribution rather than changes in the kinetic rate constants afforded by the Gly insertion likely represents an important adaptation to fine tune conformational changes within RNase H for activity at high temperatures. The Gly80b insertion into ecRNH does not significantly stabilize the protein; thus, the mutant ecRNH enzyme is unfolded and inactive at the optimal temperature for activity in the wild-type thermophilic ttRNH enzyme (∼343 K) (Kanaya and Itaya 1992). Additional amino acid substitutions, such as H62P and V74L (Kimura et al. 1992b; Ishikawa et al. 1993b), may be required to confer increased thermal stability in order for the effects of the Gly insertion to be seen at higher temperatures.

Insertion of Gly80b into ecRNH has few effects on the ground-state structure outside of the α_B/α_C hinge region and insignificantly affects overall thermodynamic stability, but renders the structure of the α_B/α_C hinge region and enzymatic activity more similar to the hinge structure and enzymatic activity observed for WT ttRNH (Ishikawa et al. 1993c). The crystallographically observed structures of apo WT ecRNH, WT ttRNH, and iG80b ecRNH are consistent with the dominant conformations observed in solution by NMR spectroscopy, but these structures appear incapable of productive interactions between the handle and substrate (Iwai et al. 1996). The present results demonstrate that the potential energy surface controlling μs–ms time scale transitions to alternate higher energy conformations also is more similar in iG80b ecRNH to WT ttRNH than WT ecRNH. These observations imply that enzymatic activity of RNases H depends upon the structures and energetics of the alternative conformations that are sampled at equilibrium in solution.

Materials and methods

Sample preparation

Plasmids containing the WT RNase H coding sequences have been described previously (Mandel et al. 1995; Hollien and Marqusee 1999b). For improved solubility, ttRNH contains four Cys mutations (Hollien and Marqusee 1999b): C17A, C45S, C67A, and C154S. Gly insertion (GGT codon) and deletion mutations were created using the Quik-Change protocol (Stratagene). WT and mutant RNases H were expressed in E. coli BL21(DE3) (Stratagene) using M9 minimal media (Sambrook et al. 1989) prepared with 99% (v/v) deuterium oxide and [98% ¹⁵N]-ammonium chloride (Cambridge Isotope Laboratories, Inc.) and purified as previously described (Kroenke et al. 1998; Hollien and Marqusee 1999b). ²H incorporation at nonexchangeable proton sites was estimated to be >85% for all samples by mass spectrometry. Final protein concentrations were 1.0 mM for WT and iG80b ecRNH, and 0.3 mM for WT and dG85 ttRNH in 100 mM d ₃-sodium acetate at pH 5.5, 10% (v/v) deuterium oxide, 0.02% (w/v) sodium azide, and 3 mM sodium 2,2-dimethyl-2-silapentane-sulfonate (DSS); ecRNH samples also contained 1 mM d ₁₀-dithiothreitol.

NMR spectroscopy

All NMR data were collected at Columbia University (11.7 and 14.1 T) or the New York Structural Biology Center (16.4 and 18.8 T). Data at static magnetic fields of 11.7 and 18.8 T were acquired on Bruker DRX500 and AV800, respectively, equipped with triple-resonance probes and three-axis gradients. Data at static magnetic fields of 14.1 and 16.4 T were acquired on Bruker DRX600 and AV700, respectively, equipped with triple-resonance cryoprobes with Z-axis gradients. For all experiments, the temperature was calibrated using a 100% (v/v) methanol standard (Cavanagh et al. 1996). Data processing and analysis were performed with NMRPipe (Delaglio et al. 1995) and Sparky (T.D. Goddard and D.G. Kneller, Sparky 3, University of California, San Francisco) software, along with in-house written programs.

Resonance assignments for mutant RNases H

Backbone ¹H and ¹⁵N resonance assignments for iG80b ecRNH and dG85 ttRNH were determined by comparison with WT spectra supplemented by HMQC-NOESY-HSQC spectra (Cavanagh et al. 1996) acquired at a static magnetic field of 16.4 T and a temperature of 300 K. The spectra were recorded using with (t ₁ × t ₂ × t ₃) 48 × 32 × 768 complex points and 2.857 × 2.857 × 10 kHz spectral widths. Assignments, referenced to internal DSS, have been deposited to the BioMagResBank (http://www.bmrb.wisc.edu) under accession codes 7277 (dG85 ttRNH) and 7278 (iG80b ecRNH).

Analysis of chemical exchange line broadening

For residues subject to chemical exchange line broadening, R ₂ = R ₂° + R _ex, where R ₂° is the intrinsic transverse relaxation rate constant which is sensitive to motion on the ps–ns time scale and the protein size. A TROSY-based Hahn-echo sequence (Wang et al. 2003) was used to measure R ₂° and R _ex at a static magnetic field of 14.1 T and temperatures between 285 K and 310 K. This method uses the transverse ¹⁵N chemical shift anisotropy/¹H–¹⁵N dipolar interference rate constant (η _xy) to determine R ₂° = κη _xy, where κ is the trimmed mean R ₂/η _xy ratio for nonexchanging residues. Values of κ were between 1.31 and 1.47 at all temperatures for all proteins. Analysis of the data with a weighted mean κ = 1.40 does not significantly change the results. Relaxation data were collected in an interleaved manner with (t ₁ × t ₂) 150 × 1024 complex points and 2.5 × 10 kHz spectral widths. A minimum of four experiments were acquired at each temperature for each protein. The average and standard deviation were taken as the relaxation rate constant and uncertainty, respectively.

Apparent activation energies, Ê _a, were determined from the temperature dependence of chemical exchange assuming an Arrhenius relationship for R _ex:

where ΔH ^‡ is the transition state enthalpy for the forward reaction and ΔH is the difference in enthalpy between the two exchanging states (Mandel et al. 1996; Evenäs et al. 1999; Butterwick et al. 2004). The second equality assumes that exchange is fast (k _ex >> Δω). For each residue, Ê _a was determined from a linear fit of ln(R _ex) versus 1/T using R _ex data acquired at a minimum of three temperatures. At low temperatures, deviations from linearity may become apparent if the exchange kinetics slow to the intermediate regime (k _ex ≈ Δω); these data are not included in the determination of Ê _a.

R_1ρ relaxation dispersion analysis

Chemical exchange line broadening is suppressed in the presence of an effective field, ω _e. In an R _1ρ experiment, magnetization is spin-locked parallel to ω _e by the application of a continuous radiofrequency (rf) field with frequency ω _rf and amplitude ω ₁. The magnitude of the effective field is ω _e ² = ω ₁ ² + ΔΩ², where ΔΩ = Ω_o − ω _rf is the observed resonance offset from the applied field. The relaxation rate constant, R _1ρ, for magnetization aligned parallel to ω _e is given by

where R ₁ is the longitudinal relaxation rate constant, θ = arctan(ω ₁/ΔΩ) is the tilt angle of the effective field, and

is the effective transverse relaxation constant where R _ex(ω _e) is the residual R _ex.

R ₁ and on-resonance R _1ρ relaxation rate constants were measured at a temperature of 300 K and static magnetic fields of 11.7 and 18.8 T using relaxation time delays between 0 and 1.2 sec and 0 and 100 msec, respectively (Farrow et al. 1994; Massi et al. 2004). Seven to nine unique time points were sampled with three duplicates for estimation of the error in signal intensity. The relaxation rate constants were determined from the best single exponential fit to the data with the error estimated using a jackknife algorithm. Off-resonance R _1ρ relaxation rate constants were measured from pairs of interleaved spectra with relaxation delays of 2 msec and 60–200 msec assuming a single exponential decay (Akke and Palmer 1996; Mulder et al. 1998); errors were estimated based on repeat experiments. For on-resonance experiments, magnetization was aligned using a free precession period of 1/ω ₁ (Yamazaki et al. 1994; Massi et al. 2004), while off-resonance experiments utilized 10 msec tanh/tan adiabatic pulses each with a frequency sweep of 15 kHz (Mulder et al. 1998). For all R _1ρ experiments, ω ₁ was calibrated using the residual scalar coupling in the presence and absence of the applied field (Palmer et al. 2001).

Analysis of relaxation dispersion data followed a similar protocol previously described (Grey et al. 2006). Initially, R ₁ and R _1ρ relaxation data at both static magnetic fields along with R ₂° and free precession R _ex estimated from TROSY Hahn-echo experiments acquired at 11.7 and 18.8 T were analyzed on a per-residue basis using Equations 5 and 6. F-statistical testing (at α = 0.05 confidence level) was used to select the appropriate functional form for R _ex(ω _e): either chemical exchange is not present, with R _ex(ω _e) = 0; or exchange is fast on the chemical shift time scale (k _ex >> ω _e), with

Additional R _ex(ω _e) models that include exchange outside the fast time scale regime (Trott and Palmer 2002), or a three-site exchange mechanism (Grey et al. 2003) did not significantly improve the fitted parameters. Optimization of R ₁ and R ₂° rate constants and ϕ _ex and k _ex exchange parameters were accomplished with routines written in Mathematica (v4.1, Wolfram Research, Inc.) utilizing the Levenberg-Marquart algorithm. Off-resonance R _1ρ data were included only if 20° ≤ θ ≤ 65° while on-resonance R _1ρ data required ΔΩ ≤ |0.4ω ₁/2π|. Residues characterized by similar k _ex values by inspection were subsequently analyzed globally with a single k _ex, maintaining individual ϕ _ex.