Equilibrium Unfolding of Bombyx mori Glycyl-tRNA Synthetase* h S

Unfolding of Bombyx mori glycyl-tRNA synthetase was examined by multiple spectroscopic techniques. Tryptophan fluorescence of wild type enzyme and an N-termi-nally truncated form (N55) increased at low concentrations of urea or guanidine-HCl followed by a reduction in intensity at intermediate denaturant concentrations; a transition at higher denaturant was detected as de-creased fluorescence intensity and a red-shifted emission. Solute quenching of fluorescence indicated that tryptophans become progressively solvent-exposed during unfolding. Wild type enzyme had stronger negative CD bands between 220 and 230 nm than the mutant, indicative of greater a -helical content. Urea or guani-dine-HCl caused a reduction in ellipticity at 222 nm at low denaturant concentration with the wild type enzyme, a transition that is absent in the mutant; both enzymes exhibited a cooperative transition at higher denaturant concentrations. Both enzymes dissociate to monomers in 1.5 M urea. Unfolding of wild type enzyme is described by a multistate unfolding and a parallel two state unfolding; the two-state component is absent in the mutant. Changes in spectral properties associated with unfolding were largely reversible after dilution to low denaturant. Unfolding of glycyl-tRNA synthetase is complex with a native state, a native-like monomer, partially unfolded states, and the unfolded state. Aminoacyl-tRNA

Unfolding of Bombyx mori glycyl-tRNA synthetase was examined by multiple spectroscopic techniques. Tryptophan fluorescence of wild type enzyme and an N-terminally truncated form (N55) increased at low concentrations of urea or guanidine-HCl followed by a reduction in intensity at intermediate denaturant concentrations; a transition at higher denaturant was detected as decreased fluorescence intensity and a red-shifted emission. Solute quenching of fluorescence indicated that tryptophans become progressively solvent-exposed during unfolding. Wild type enzyme had stronger negative CD bands between 220 and 230 nm than the mutant, indicative of greater ␣-helical content. Urea or guanidine-HCl caused a reduction in ellipticity at 222 nm at low denaturant concentration with the wild type enzyme, a transition that is absent in the mutant; both enzymes exhibited a cooperative transition at higher denaturant concentrations. Both enzymes dissociate to monomers in 1.5 M urea. Unfolding of wild type enzyme is described by a multistate unfolding and a parallel two state unfolding; the two-state component is absent in the mutant. Changes in spectral properties associated with unfolding were largely reversible after dilution to low denaturant. Unfolding of glycyl-tRNA synthetase is complex with a native state, a native-like monomer, partially unfolded states, and the unfolded state.
Aminoacyl-tRNA synthetases are a structurally diverse group of enzymes divided into two classes based on the topography of their adenylate binding sites and their modes of tRNA binding (1)(2)(3). The active sites of class I enzymes have a classic Rossman fold formed from two ␤-␣-␤ elements resulting in a structure of four parallel beta strands. The active sites of class II enzymes consist of a unique fold formed from an antiparallel seven-stranded sheet element first identified in seryl-tRNA synthetase (4). A structure similar to the active site of class II enzymes exists in the biotin synthetase/repressor protein (5) and the asparagine synthetase (6), both of which have an adenylate intermediate on their reaction pathways. The two aminoacyl-tRNA synthetase classes have different modes of nucleotide binding and approach tRNA from opposite sides of the polynucleotide. The differences in primary structure and modes of ligand binding suggest that the two classes arose independently (7)(8)(9).
Glycyl-tRNA synthetases are unusual in that there are two distinct enzyme types that are not closely related. They have developed modes of tRNA recognition that differ with respect to the discriminator base and other base pairs in the stem (10,11). The type exemplified by the Escherichia coli enzyme is found in Gram-negative and Gram-positive bacteria, whereas a second enzyme is found in other eubacteria such as Mycobacteria sp. and Mycoplasma sp., organisms classified as thermotoga (e.g. Thermus thermophilus), the archaea (e.g. Methanococcus jannashii), and in all eukaryotic organisms. Examination of the crystal structure of T. thermophilus glycyl-tRNA synthetase revealed an atypical motif 1 (12) that was not identified in earlier alignments of glycyl-tRNA synthetase (13,14). Although the T. thermophilus enzyme aligns with glycyl-tRNA synthetases from eukaryotic organisms, the eukaryotic enzymes have elements that are absent in the T. thermophilus enzyme. One such element in glycyl-tRNA synthetase from Bombyx mori, Homo sapiens, Caenorhabditis elegans, and Arabadopsis thalinia is a 50-to 60-residue N-terminal structure present in a number of other eukaryotic aminoacyl-tRNA synthetases, including the glutamyl-prolyl-, histidyl-, tryptophanyl-, and methionyl-tRNA synthetases. Although the physiological function of this structure is unclear, because it is apparently not required for amino acid activation or tRNA binding, some studies suggest it may have a role in binding to polynucleotides (15,16) or in mediating interaction with other proteins (17,18). One study of human histidyl-tRNA synthetase (19) indicated that removal of this element from the N terminus resulted in a drastic reduction in enzymatic activity, although the removal of the corresponding structure in B. mori glycyl-tRNA synthetase does not have the same effect (15). Studies by Raben et al. (19) indicated that the structure has a high content of ␣-helix and may form a coiled-coil, similar to the N-terminal domain of the E. coli (4,20) and T. thermophilus (21,22) seryl-tRNA synthetases.
We have examined the unfolding of the B. mori glycyl-tRNA synthetase to determine how the N-terminal structure interacts with other elements of the protein and to determine the unfolding pathway for this dimeric enzyme. Our studies indicate that the enzyme dissociates into relatively native monomers prior to effects on spectroscopic signals sensitive to changes in conformation; the monomers unfold through a multistate process. The first 55 residues of the protein have a high content of ␣-helix and unfold independently of the rest of the structure, suggesting that it constitutes a domain that is separate from the core catalytic domains.
Apart from our interest in aminoacyl-tRNA synthetase struc-ture and mechanism, the present studies are of general interest as an example of a complex unfolding pathway. Most protein folding studies have focused on simple two-state systems, whereas many functionally interesting proteins are likely to show more complex behavior. The techniques and analytical methods for coping with the added complexity need to evolve, and the present studies are a step in this direction.

EXPERIMENTAL PROCEDURES
Materials-Wild type B. mori glycyl-tRNA synthetase and a mutant lacking the first 55 N-terminal residues (N55) were expressed in E. coli BL21DE3pLysS transformed with plasmids encoding these proteins (pNADA and pNADAN55, described previously (15)). The proteins were purified by successive chromatography on Q-Sepharose, hydroxylapatite, and Sephacryl S200 (15). Stock solutions of the purified proteins at 15-25 mg ml Ϫ1 in 50 mM potassium phosphate (pH 7.5), 20% (v/v) glycerol, 0.5 mM EDTA, 0.5 mM dithiothreitol were stored as small aliquots at Ϫ80°C until use. Protein concentrations were determined from absorbance at 280 nm using extinction coefficients (based on the absorbance of a 1% solution of the protein) of 9.1 for the wild type enzyme and 10 for the N55 deletion mutant (Ref. 23 and the present study). Guanidine-HCl was obtained from Heico (Delaware Watergap, PA) and ultrapure urea was from Life Technologies, Inc. Concentrations of urea and guanidine-HCl solutions were calculated from a refractive index (24). 1-Anilinonapthalene-8-sulfonic acid (ANS) 1 was obtained from the Sigma Chemical Co.
Analytical Ultracentrifugation-Ultracentrifugation experiments were performed in a Beckman Optima XLA analytical ultracentrifuge employing absorbance optics and using an An60Ti rotor. Temperature calibration was performed as described previously (25); experiments were performed at 20°C. Sedimentation velocity studies were performed at 40,000 rpm in a charcoal-filled Epon double-sector centerpiece. Density and viscosity of the solvents were estimated using the program SEDNTERP. The partial specific volume was calculated by the method of Cohn and Edsall (26 -28). Velocity data were collected at 280 nm at 0.002-cm intervals with one average in a continuous scan mode. Sedimentation data was analyzed using DCDTϩ (29) and SVEDBERG (version 6.37) as detailed previously (25). Samples were prepared for sedimentation studies by chromatography on a column of Sephadex G-50 medium equilibrated in 50 mM potassium phosphate (pH 7.2), 0.1 mM EDTA containing 0.05 mM DTT and the indicated concentrations of urea. Samples were prepared 4 h prior to the experiment; protein concentrations were adjusted to 0.2 mg ml Ϫ1 .
Size Exclusion Chromatography-Stokes radii of wild type and N55 glycyl-tRNA synthetases were determined on a Superdex 200 column in 50 mM potassium phosphate (pH 7.2), 0.1 mM EDTA, and 0.05 mM DTT and different concentrations of urea. Standards (thyroglobulin (8.0 nm), immunoglobulin G (5.0 nm), ovalbumin (3.5 nm), and myoglobin (2.0 nm)) were run in the absence of urea. Samples were prepared 4 h before application to the column, and chromatography was performed at room temperature (23 Ϯ 0.5°C).
Unfolding Experiments-Samples for spectroscopic measurements were diluted at least 100-fold into solutions containing 50 mM potassium phosphate (pH 7.2), 0.1 mM EDTA, 0.05 mM DTT and the indicated concentration of guanidine-HCl or urea. Corresponding blanks were prepared using enzyme storage buffer. All spectral measurements were taken at 20°C. Sample solutions were prepared at least 4 h prior to taking measurements. Preliminary experiments revealed a time-dependent change in fluorescence intensity after preparation of the sample. Most of the fluorescence change occurred in the first hour and occurred more rapidly in samples containing urea or guanidine-HCl; there was no detectable difference in measurements taken at 4 and 16 h. Measurements of fluorescence were taken in an ISS, Inc. GREG 200 spectrofluorometer using a 10-by 2-mm quartz cell. Excitation was at 290 nm and emission was measured from 300 to 400 nm at 1-nm intervals. In some experiments, measurements were taken at a fixed wavelength as indicated in the figure legend. Circular dichroism was measured using a Jasco J500 spectropolarimeter at 20°C. Measurements in the far ultraviolet (185-240 nm) were taken in a 0.1-or 1-mm cell on solutions of 0.05-0.2 mg ml Ϫ1 and in the near ultraviolet (240 -300 nm) in a 1-mm cell on solutions at 0.2-0.4 mg ml Ϫ1 . Concentrations in specific experiments are indicated in the figure legends. Samples were prepared for CD at least 4 h before taking the measurement; no significant differences were detected in spectra taken between 4 and 16 h. Spectra were recorded by taking measurements at 0.2-nm intervals at a scan rate of 1 nm/min at a 1-nm band pass and a time constant of 64. Appropriate blank spectra were recorded on the buffer components and subtracted from spectra obtained on protein solutions. In experiments examining unfolding as a function of denaturant, data were collected at a fixed wavelength and averaged over 2 min. Fluorescence spectra in guanidine-HCl or urea were analyzed by singular value decomposition (30,31) using the program MATLAB 5.3 (The Math-Works Inc.). Singular value decomposition converts the data matrix into the product of a U matrix (basis spectra), an S matrix (singular values), and a V matrix (amplitude vectors). These three matrices allow for the determination of the number of spectrally significant species over a range of guanidine-HCl or urea. Singular values of Ͻ0.017 and autocorrelation values of Ͻ0.8 are taken as indicative of a random nonsignificant component. Fitting of models of unfolding to data was performed by the nonlinear least squares algorithm of Marquart (32) using the program FitAll (MTR Software). Intensity averaged emission wavelength was calculated, where I i is the intensity measured at i .
Refolding Experiments-Protein solutions were mixed with concentrated urea to a final concentration of 7 M and then diluted to different concentrations of urea for refolding; other buffer components were 50 mM potassium phosphate (pH 7.2), 0.1 mM EDTA. and 0.05 mM DTT. Spectral measurements were made after 20 h.
Models of Unfolding-Spectroscopic data obtained in unfolding experiments were analyzed taking into consideration four schemes for unfolding. A simple two-state model (33) with the native protein (N) and the unfolded species (U) is described by, where S obs is the observed spectroscopic signal, S N and S U are the signals for the native and unfolded species, respectively, ⌬G is the free energy of unfolding in the absence of denaturant, m is the partial derivative of ⌬G with respect to denaturant, and d is the concentration of denaturant. S obs and d are the dependent and independent variables, S N is treated as a constant, and S U , ⌬G, and m are treated as fitting parameters.
A more complicated three-state model (Reaction II, described in Ref.

34), which includes an intermediate species is given by,
where S obs , S N , S U , and d have the same definitions as for the two-state model. The terms ⌬G 1 , ⌬G 2 , m 1 , and m 2 are the free energies and m values for the N 7 I and I 7 U transitions. ⌬G 1 , ⌬G 2 , m 1 , m 2 , S I , and S U are treated as fitting parameters, and S N is treated as a constant based on the value of the observed signal in the absence of denaturant; however, when S N was treated as a fitting parameter, similar results were obtained. A linear term (m U [d]) was included as a fitting parameter to account for the post-transition baseline (34 -36). This model was used to fit CD data obtained for the unfolding of the N55 mutant. A variant of the three-state model was derived, which includes a third parallel transition to account for an unfolding event that occurs independently of the other transitions (Reactions III and IV below). This model was considered to account for the difference observed between wild type glycyl-tRNA synthetase and the N55 mutant when CD was monitored during unfolding, where the terms ⌬G 1 , ⌬G 2 , m 1 , m 2 , and S I are defined in the same manner as for the three-state model; S N is the signal of the native species in the three state unfolding in Reaction III. The terms ⌬G p m p , S Np , and S Up are used in the second term in Eq. 4 that describes the independent (or parallel) unfolding event given in Reaction IV; we have interpreted signals S Np and S Up as arising from the native and unfolded N-terminal element. Linear terms (m N [d] and m U [d]) were included in the equation as fitting parameters to account for the dependence of the post-transition baseline on denaturant concentration (34 -36). However, owing to the complexity of the models, including these terms had little effect on the result. An expression based on a four state model of unfolding was derived to analyze the fluorescence emission employing the same approach used for the three-state model. The thermodynamic parameters describing the N to I a and I a to I b transitions are indicated by a and b subscripts to avoid the implication that they correspond to transitions described by ⌬G 1 and ⌬G 2 in Eqs. 3 and 4. The term ⌬G 2 is the same in all three expressions to indicate that it corresponds to the same unfolding event.
Fluorescence Quenching Experiments-Quenching of intrinsic tryptophan fluorescence by iodide or acrylamide (37)(38)(39) was examined in 50 mM potassium phosphate buffer, 0.1 mM EDTA, 0.05 mM DTT, and the indicated concentrations of urea or guanidine-HCl. Excitation was at 290 nm, and emission was measured from 300 to 400 nm or by measuring emission at a fixed wavelength. Fluorescence in acrylamide quenching experiments was corrected for the inner filter effect due to the absorbance of acrylamide at 290 nm using the relationship F ϭ F obs 10e 0.5⑀cl , where ⑀ is the molar extinction coefficient, c is molar concentration, and l is the path length (1 cm). In experiments employing KI as a quencher, ionic strength was kept constant by the addition of KCl. Estimates of the dynamic and static quenching constants for acrylamide were obtained by fitting the hyperbolic form of the Stern-Volmer equation to the data, which includes an exponential term to account for static quenching (38), where K SV is the Stern-Volmer constant for collisional quenching and is the static quenching constant. Data are presented using the inverse form of the Stern-Volmer equation as F 0 /F versus [Quencher]. Because iodide quenching did not have a significant static component, the exponential term was not included in fits of the equation to these data. Fluorescence Lifetime Measurements-Fluorescence lifetime data were collected using an I.S.S. K2 multifrequency cross-correction phase and modulation fluorometer with a xenon arc lamp. A scattering solution of glycogen (0.8 mg/ml aqueous solution) was used as a reference. Excitation was at 290 nm, and emission was measured on emitted light that passed through a 305-nm cutoff filter to eliminate scattered radiation. Lifetimes were determined at 15 frequencies in the 1-to 200-MHz range. Data were collected until the S.D. from each measurement of phase and modulation was at most 0.2 and 0.004, respectively.
ANS Binding-ANS (1-anilinonapthalen-8-sulfonic acid) binding was detected by collecting fluorescence spectra in the presence of 10 M dye for the wild type glycyl-tRNA synthetase and the N55 mutant in the presence of varying concentrations of guanidine-HCl or urea. Excitation was at 380 nm, and emission was measured from 400 to 600 nm. Stock solutions of ANS were prepared in methanol and diluted into the samples such that the methanol concentration was less than 0.05% (v/v). Concentration of the dye was determined using an extinction coefficient of 8 ϫ 10 3 M Ϫ1 cm Ϫ1 at 372 nm. The presence of urea or guanidine-HCl in the absence of protein had no significant effect on the fluorescence of the dye.

RESULTS
To examine the unfolding of glycyl-tRNA synthetase and the interaction of the N-terminal structure with other elements of the protein, we examined the urea and guanidine-HCl unfolding process by multiple spectroscopic techniques (40). We also examined the oligomeric state of the protein under different solvent conditions. Examination of Protein Unfolding by Intrinsic Tryptophan Fluorescence-B. mori glycyl-tRNA synthetase has four tryptophans that are conserved in all eukaryotic glycine enzymes. Insofar as tryptophanyl residues are sensitive to protein conformation and local environment (41) they are potential probes of conformational change. We examined fluorescence emission spectra for the wild type enzyme and the N55 deletion mutant under native conditions and in the presence of different concentrations of urea (panel A) or guanidine-HCl (panel B) as shown in Fig. 1 for the wild type enzyme; results with the N55 mutant were similar (not shown). With both chaotropic agents there was an increase in fluorescence intensity at lower concentrations of denaturant followed by a red shift in the emission wavelength and a reduction in fluorescence intensity at higher concentrations of denaturant. Both mutant and wild  Fig. 2. The addition of KCl to 1 M in the absence of urea or guanidine-HCl did not induce an increase in fluorescence intensity as was observed at 1 M guanidine-HCl, indicating that the effect at this concentration of guanidine-HCl is not related to ionic strength or the presence of chloride. The similarity of the results obtained with urea and guanidine-HCl also suggests that the spectral changes reflect general features of the unfolding process and not properties of the agents used to induce unfolding.
The weighted basis spectra (expressed as the product of the U and S matrices) were obtained by singular value decomposition of the emission spectra from guanidine-HCl or urea induced denaturation (shown in Fig. 3) for the wild type enzyme; similar results were obtained for the N55 mutant. Singular values and values for the autocorrelation function for the first 10 basis spectra are summarized in Table I. The analysis indicates that two singular values make the largest contribution to the spectra with a third component making a negligible contribution; similarity of the basis spectra suggests that the variables are the same for the two denaturants for the mutant and wild type enzymes. Based on the magnitude of the singular values and the values of the autocorrelation function (values greater than 0.8 indicating significance), only the first four or five basis spectra make a contribution to the spectra. The existence of multiple species in the unfolding pathway is suggested by singular value decomposition of the intensity data. Although the basis spectra are not necessarily those of the intermediates, they suggest that there are species with spectral properties intermediate between the native and denatured forms of the enzyme. The red shift and the intensity changes associated with unfolding indicate that the tryptophans are buried and at least partially shielded from solvent (37, 41), a conclusion also supported by iodide and acrylamide quenching studies (presented later). The increase in fluorescence intensity suggests quenching of fluorescence of one or more of the tryptophanyl residues by proximity to another group is relieved by a conformational change induced by low concentrations of denaturant. No concentration dependence was observed in unfolding experiments, where fluorescence was observable over a range of 0.05-0.2 mg ml Ϫ1 ; the propensity of the protein to adsorb to glass and plastic below 0.05 mg ml Ϫ1 precluded examination of lower protein concentrations.
The red shift in the emission spectrum, shown as the intensity-averaged emission wavelength in Fig. 4, was used to monitor urea-induced (Fig. 4, A and B) and guanidine-HCl-induced (C and D) unfolding. Multiple transitions reflected in fluorescence changes are inconsistent with a simple two-state model (Reaction I and Eq. 2), suggesting a more complex scheme is required to describe the unfolding process. Although the change in emission wavelength in Fig. 4 can be fit using a three-state model, fits to the intensity changes shown in Fig. 2 and the emission wavelength in Fig. 4 showed a nonrandom distribution of residual values (Fig. 4, B and D). A better fit was obtained with a model with two intermediate species (Reaction V and Eq. 5); values for the F-statistic for different experiments ranged from 4 to 10 placing the fit to the four-state model well above the 95% interval. The values for ⌬G and m for the transition occurring at high denaturant concentration are given in Table I for the wild type enzyme. Examination of the Effect of Unfolding on Tryptophan Fluorescence Lifetimes-The fluorescent properties of tryptophans in glycyl-tRNA synthetase during guanidine-HCl-induced denaturation was examined by fluorescence lifetime measurements (see Fig. 1s in the Supplemental Material). Two classes of lifetimes were detected with one class accounting for 70 -80% of the fluorescence. Consistent with the steady-state measurements, there was an increase in the lifetime of the larger class of fluorophors with increasing denaturant concentration followed by a decrease as the protein unfolds at high concentrations; this result is seen in both the wild type enzyme and N55 mutant.
Examination of Tryptophan Exposure by Fluorescence Quenching-To corroborate the results of the unfolding studies indicating that the tryptophans in the native proteins are at least partially shielded from the solvent, we examined iodide and acrylamide quenching of tryptophan fluorescence. Both wild type and mutant gave linear Stern-Volmer plots with iodide, whose slopes increased with increasing concentrations of urea or guanidine-HCl (shown in Fig. 2s, A and B, in the Supplemental Material). Quenching by acrylamide gave a similar pattern of increasing quenching upon addition of urea or guanidine-HCl (see Fig. 2s, C and D, in the Supplemental Material) but exhibited both static and dynamic components as indicated by the upward curvature in the Stern-Volmer plots. The quenching studies, summarized in Table II, indicate that the tryptophans in native glycyl-tRNA synthetase are inaccessible to a polar quencher (KI) and only slightly more accessible to a more hydrophobic quencher (acrylamide), consistent with most of the four tryptophans being shielded from solvent in the protein interior. That both the wild type and N55 mutant exhibited similar behavior qualitatively and quantitatively is consistent with the steady-state fluorescence results indicating that the tryptophans are buried. Perturbation of the structure of both enzymes with a relatively low concentration of denaturant resulted in the same extent of exposure of tryptophanyl residues to solvent. As expected for unfolded proteins, both wild type and mutant showed the same susceptibility to iodide and acrylamide quenching when treated with high concentrations of urea (6.4 M) or guanidine-HCl (5.8 M).
Examination of Unfolding by Circular Dichroism-We employed circular dichroism in the analysis of the unfolding of glycyl-tRNA synthetase as a spectroscopic probe that is sensitive to protein secondary structure. The CD spectra of the wild type and N55 deletion mutant in Fig. 5 revealed greater negative ellipticity for the wild type enzyme in the region of 210 -230 nm (Fig. 5A), a region with bands characteristic of ␣-helix. The difference spectrum derived from the wild type and mutant proteins has features characteristic of ␣-helix (not shown).
There was a small, though significant, difference between wild type and mutant in the CD spectra recorded in the near-UV range (panel B). The far-UV CD spectra of the native and the unfolded/refolded wild type and N55 mutant proteins were also examined as shown in Fig. 5 (C and D, respectively). The native (C, closed symbols) and unfolded/refolded (C, open symbols) wild type enzyme were similar, indicating that unfolding of a significant fraction of secondary structure is reversible. The CD spectrum of the unfolded/refolded N55 mutant (D, open symbols) is also similar to the native form of the protein (D, closed symbols). We examined the change in ellipticity at 222 nm at different concentrations of urea for the wild type and N55 mutant proteins as shown in Fig. 6A. The wild type enzyme showed significantly higher ellipticity than did the mutant, which declined with a transition with a midpoint at ϳ2.7-3 M urea; this transition was absent or at least far less pronounced in the mutant. At higher concentrations of urea, the behavior of the mutant and wild type proteins were similar, both showing a relatively noncooperative decrease in ellipticity between 3 and 5.5 M urea followed by a cooperative transition centered at 6 -6.1 M urea. The latter transition occurs at the same concentration of urea as the transition seen when fluorescence was examined and appears to correspond to the global unfolding of the protein. When guanidine-HCl was employed to induce unfolding (Fig. 6B), a similar result was obtained; a transition at low concentrations of denaturant was present in the wild type enzyme but absent in the mutant and both exhibited a cooperative transition at 2.7 M guanidine-HCl. Despite the difference between the wild type and mutant proteins below 3 M urea in unfolding experiments, both could be fit to a three-state unfolding model; the result may be a reflection of the small amplitude of the spectral change in the mutant and the transitions in the wild type occurring coincidentally over a similar range of denaturant. Although application of the F-statistic to the fit of four-state models to the data for N55 indicated that these complex models could not be justified, the wild type enzyme gave a better fit to the four-state model (F ϭ 5.8, 95% confidence interval) described by Eq. 5. A more appropriate model (Reactions III and IV and Eq. 4) for the wild type enzyme includes a three-state unfolding process that describes features of both the wild type and mutant and an independent, parallel unfolding process that accounts for the transition seen in the  wild type enzyme at low denaturant concentration, and it was this model that was used for analysis of the wild type enzyme.
Although the fit to the four-state scheme described by Eq. 5 was better than to the three-state model (F ϭ 11, with p greater than 95%), it was almost the same as the fit obtained to Eq. 4. Tables III and IV summarize the thermodynamic parameters (m and ⌬G values) derived from urea-and guanidine-HClinduced unfolding experiments. Discrepancies for free energies between urea and thermal unfolding and guanidine-HCl-induced unfolding noted previously (24) have been attributed to the stabilizing effect of guanidine-HCl on some structures (34,42). Makhatadze (43) has attributed some of the effects of guanidine-HCl on protein stability to the nature of the anion and suggests that guanidine-HCl is unsuitable for determining thermodynamic parameters of protein unfolding using the commonly employed linear extrapolation method. The work of Smith and Scholtz (44)  respect to the use of guanidine-HCl, the unfolding of glycyl-tRNA synthetase with the two denaturants was quite similar. The similarity in the values of ⌬G 2 where fluorescence or CD were observable suggests that they reflect the same unfolding process.
Examination of ANS Binding-ANS binding was used as a probe of the extent of exposure of hydrophobic regions of the protein during unfolding. Glycyl-tRNA synthetase bound ANS in the absence of denaturant (Fig. 3s in the Supplemental Material). The fluorescence of the dye in the presence of the protein increased with increasing urea (Fig. 3A) or guanidine-HCl (Fig. 3B) and then declined at higher concentrations. At 4 -6 M guanidine-HCl or 6 -8 M urea the fluorescence of the dye in the presence of protein was the same as that of the free dye. Essentially the same results were obtained with the N55 mutant (data not shown). With both urea and guanidine-HCl, ANS fluorescence increased significantly at denaturant concentrations (1 M urea or 0.4 M guanidine-HCl) that caused little change in either intrinsic tryptophan fluorescence or CD at 222 nm. The results are consistent with increased exposure of hydrophobic residues as the protein unfolds with increasing denaturant concentration.
Examination of the Oligomeric State of Glycyl-tRNA Synthetase during Unfolding-Protein denaturation through a dimeric intermediate confers a concentration dependence on the unfolding process and is described by different expressions than for proteins, which denature through monomeric intermediates (45,46). Because glycyl-tRNA synthetase is a dimer, such an intermediate must be considered. To determine the extent to which dissociation of the dimer contributes to changes in the spectroscopic signals, we examined the sedimentation constant at concentrations of urea (shown in Fig. 4s, A, of the Supplemental Material) associated with changes in both the CD and fluorescence spectra. In the absence of denaturant the wild type protein had a weighted average value of S 20,w of 6.91, consistent with previous determinations of this parameter (15,47) and consistent with a dimer. At 1.5 M urea the weighted average S 20,w was 5.75. The latter value is consistent with other data indicating that the protein is a dimer of subunits of M r 76,919 (15,47) and that it dissociates reversibly into monomers (23). The sedimentation data collected in the presence of 1.5 M urea is accounted for by a monomeric species with an S 20,w of 4.57 constituting 76% of the material with the remaining material reflecting an early stage of aggregation. The presence of the larger material skews the weighted average for S 20,w to 5.75. At 4 M urea the protein is aggregated and polydisperse. The transition from dimer to monomer occurs at a concentration of urea where there is no significant change in the fluorescence emission spectrum or the CD spectrum of the protein. The result indicates that dissociation of the dimer into monomers occurs prior to conformational changes reflected in fluorescence and CD spectra and that the monomers are probably similar to the native structure. In experiments not shown we examined the effect of guanidine-HCl on the oligomeric state of the wild type and N55 mutant enzymes by cross-linking the proteins with succinimidylsuberate under conditions that we had previously established for cross-linking the enzymes (15). In the absence of guanidine-HCl we observed a species with a mobility expected for the dimer. At low concentrations of guanidine-HCl (up to 1 M) there was a reduction in cross-linking. At higher concentrations of denaturant (up to 2.2 M), the cross-linked species just entered the gel migrating at a position expected for cross-linked aggregates. Above 2.5 M the protein migrated at a position expected for the monomer. We also examined the wild type and N55 mutant enzymes by size exclusion chromatography (shown in Fig. 4s, B, of the Supplemental Material). The native proteins eluted from Superose 12 at a retention time corresponding to the dimer (R s 5.5 nm for wild type, 5.3 for N55), whereas protein in the presence of 1.5 M urea eluted at a retention time characteristic of the monomer (R s 4.9 nm for wild type, 4.7 nm for N55); at 4 M urea the protein eluted at a retention time indicating a size greater than thyroglobulin (Ͼ8 nm) and suggesting that it was aggregated. These results indicate that the wild type and mutant enzymes dissociate into monomers at low concentrations of denaturant, aggregate at intermediate concentrations, and then dissociate to unfolded monomers at high denaturant concentration. Because the dissociation of dimers to monomers is not reflected in spectral changes at low denaturant and there is no detectable concentration dependence to the unfolding, the data can be fit to a model which need only account for unfolding of the monomer.

DISCUSSION
Unfolding of Glycyl-tRNA Synthetase-Examination of the denaturation of glycyl-tRNA synthetase by urea or guanidine-HCl using spectroscopic techniques sensitive to protein conformation reveals a complex unfolding pathway expected of a multidomain, dimeric enzyme (48). Depending on the spectroscopic signal examined, unfolding of the wild type enzyme can be described by a three-state (for CD) or a four-state (fluorescence) model. Although a three-state model can be satisfactorily fitted to the CD data obtained for the wild type enzyme, analysis of the deletion mutant lacking the first 55 residues suggests that an additional parallel unfolding process is required to describe the unfolding of the N terminus. The results with the two proteins suggest that the N-terminal element constitutes an independent folding entity. This may also be the case for E. coli seryl-tRNA synthetase (49) where the unfolding of the N-terminal coiled-coil has been examined, although this element was examined while separated from the rest of the protein. Based on the difference in the CD spectra of the wild type and N55 mutant, the N-terminal element of glycyl-tRNA synthetase appears to be largely ␣-helix, and conforms to the 7-amino acid repeat (a through g with hydrophobic residues at a and d and charged residues usually at positions e and g) (50) expected for a typical coiled-coil. Although we do not have a structure of the N-terminal element of glycyl-tRNA synthetase, an NMR structure reported recently (51) for a homologous element from hamster glutamyl-prolyl-tRNA synthetase demonstrates that this structure is an antiparallel coiled-coil; the structure binds to polynucleotides as originally suggested from studies of glycyl-tRNA synthetase (15). The structure appears to fold independently and its removal from glycyl-tRNA synthetase does not alter the environment of tryptophanyl residues as detected by examination of unfolding by fluorescence and by susceptibility of the tryptophanyl residues to quenching by iodide or acrylamide. The independent folding properties of this structure are in accord with its being present in the eu- karyotic glycyl-tRNA synthetases and absent in the prokaryotic enzyme and the fact that the N-terminal element is not required for amino acid activation or tRNA aminoacylation. This structure is present in a number of aminoacyl-tRNA synthetases of both classes, in the N terminus (tryptophanyl-, histidyl-, and glycyl-), in the C terminus (methionyl-), and as multiple copies between two tRNA synthetase domains (glutamyl-prolyl-).
Although transitions seen at lower concentrations of denaturant detected by CD are absent in the mutant, the transition at higher denaturant concentration associated with global unfolding of the protein occurs at the same concentration of denaturant in both mutant and wild type and has the same amplitude. A major transition at high denaturant detected by fluorescence corresponds to the transition detected by CD. Caution must be exercised in interpreting the results of experiments where the observable is fluorescence, because, as pointed out by Eftink (41), there is no simple relationship describing fluorescence or max for a mixture of states; the state with the highest quantum yield will tend to dominate the emission, skewing the measured parameter toward that state. However, singular value decomposition analysis (Table I and Fig. 1) of glycyl-tRNA synthetase unfolding suggests that the fluorescence properties of intermediates in the unfolding must be similar to the native state. That the value for ⌬G b is poorly determined by the data is in accord with this difficulty in the analysis of fluorescence data, the complexity of the model required to account for the unfolding of glycyl-tRNA synthetase, and the small amplitude of the spectral change. However, the close correspondence between the transitions occurring at high denaturant concentration observed by fluorescence and CD suggests that they reflect the same unfolding process. The minimal model describing the unfolding of glycyl-tRNA synthetase requires the native state, the unfolded state, at least two intermediate species, and an independent unfolding process with its own native and unfolded states.
The dissociation of the dimeric enzyme into subunits is not accompanied by changes detectable by CD and fluorescence, suggesting that a native-like monomer is formed. This is consistent with earlier work (23) indicating that at low protein concentrations, in the absence of chaotropic agents, the enzyme dissociates into inactive monomers. Similar results were obtained in studies of the homologous glycyl-tRNA synthetase from S. cerevisiae (52), suggesting that this enzyme dissociates reversibly to inactive monomers. Although the dissociation to monomers must be considered as part of the overall unfolding pathway, we have not taken it into consideration in the analysis of our data, because there is no spectral manifestation of it and no protein concentration dependence was seen in the unfolding transitions. The absence of concentration dependence can be explained by the dissociation greatly favoring the monomer conformation even at relatively low denaturant concentrations. Presumably, concentration dependence would be detected if sufficiently high protein concentrations could be examined. Although analysis based solely on the spectroscopic data will underestimate the total free energy of unfolding, the  contribution of dimer dissociation to the process appears to be small, because it occurs at relatively low concentrations of denaturant. The unfolding of dimeric proteins involves free energy changes that are typically higher than monomeric proteins and roughly proportional to the number of residues (42); the relationship between size and free energy of unfolding appears to reflect the surface area available for interaction between subunits. It would appear that in the case of glycyl-tRNA synthetase the dimer-dimer contacts do not make a major contribution to the overall stability of the protein.
In other systems such as the dimeric N-terminal domain of tyrosyl-tRNA synthetase, the dimer is converted to a monomeric intermediate (53). Less commonly, as in the cases of luciferase (46), organophosphorus hydrolase (54), and 3-isopropylmalate dehydrogenase (55), a native dimer is converted to a dimeric intermediate, which is then converted to an unfolded monomer. The Arc repressor (56,57) unfolds directly to monomers, whereas the remarkably stable tetrameric lac repressor (58) forms an unfolded tetramer, which dissociates to monomers at higher denaturant concentration. One aspect of the unfolding we have not considered in our analysis is the formation of the oligomeric species with the wild type and mutant enzymes at intermediate concentrations of denaturant. Guanidine-HCl and urea can promote aggregation in unfolding studies due to the association of an intermediate (59) or the denatured state (60). Partially folded intermediates of staphylococcal nuclease aggregate to form more structured species (61,62); partially folded conformations appear to be intermediates in the formation of most aggregates (59) and can be mistaken for kinetic folding intermediates when they form transiently (63). Rigorous analysis of such a system would presumably require fitting data to a polynomial whose order would depend on the number of monomers in the aggregate; an additional vexing problem is the existence of multiple oligomeric states. The unfolding process usually exhibits dependence on the concentration of the monomer as observed with staphylococcal nuclease (61,62). Although analysis of the effect of oligomerization is a tractable problem when describing the unfolding of a dimer or a monomer that goes through a dimeric intermediate, analysis of a higher order polynomial where n is indeterminate poses a significant obstacle. The absence of concentration dependence in the unfolding experiments where fluorescence was measured warrants some comment. One would expect both the dissociation of a dimer and the formation of oligomeric species to show some concentration dependence such that any expression that accounts for the unfolding would require a mass action term(s). Concentration dependence might be detectable at lower concentrations than were examined, but the propensity of the protein to adsorb to surfaces at low concentrations precluded our observing it. Although there appears to have been no explicit treatment of aggregating systems for denaturants such as guanidine and urea, pressure-induced unfolding of higher ordered structures (64) such as erythrocruorin (65) and the capsid protein of Brome mosaic virus (66) exhibit a lack of concentration dependence so that such structures exhibit behavior characteristic of macroscopic bodies.