Yeast hexokinase isoenzyme ScHxk2: stability of a ... - Oxford Journals

Protein Engineering, Design & Selection vol. 24 no. 1 –2 pp. 79–87, 2011 Published online November 12, 2010 doi:10.1093/protein/gzq098

Yeast hexokinase isoenzyme ScHxk2: stability of a two-domain protein with discontinuous domains Hauke Lilie 1, Dorit Ba¨r 2, Karina Kettner 2, Ulrich Weininger 3, Jochen Balbach 3, Manfred Naumann 4, Eva-Christina Mu¨ller 5, Albrecht Otto 5, Klaus Gast 6, Ralph Golbik 1,7 and Thomas Kriegel 2 1

Institute of Biochemistry and Biotechnology, Martin Luther University Halle-Wittenberg, Kurt-Mothes-Strasse 3, Halle D-06120, Germany, 2Carl Gustav Carus Medical Faculty, Institute of Physiological Chemistry, Dresden University of Technology, Fetscherstrasse 74, Dresden D-01307, Germany, 3 Institute of Physics, Biophysics Group and ‘Mitteldeutsches Zentrum fu¨r Struktur und Dynamik von Proteinen’, Martin Luther University HalleWittenberg, Betty-Heimann-Strasse 7, Halle D-06120, Germany, 4Medical Faculty, Institute of Biochemistry, University of Leipzig, Johannisallee 30, Leipzig D-04103, Germany, 5Max Delbru¨ck Center for Molecular Medicine, Neuroproteomics, Robert-Ro¨ssle-Strasse 10, Berlin D-13125, Germany and 6 Institute of Physical Biochemistry, University of Potsdam, Karl-LiebknechtStrasse 24, Potsdam D-14476, Germany 7

To whom correspondence should be addressed. E-mail: [email protected] Received October 8, 2010; revised October 8, 2010; accepted October 12, 2010 Edited by Henning Tidow

The hexokinase isoenzyme 2 of Saccharomyces cerevisiae (ScHxk2) represents an archetype of a two-domain protein with the active site located in a cleft between the two domains. Binding of the substrate glucose results in a rigid body movement of the two domains leading to a cleft closure of the active site. Both domains of this enzyme are composed of discontinuous peptide sequences. This structural feature is reflected in the stability and folding of the ScHxk2 protein. Structural transitions induced by urea treatment resulted in the population of a thermodynamically stable folding intermediate, which, however, does not correspond to a molecule with one domain folded and the other unfolded. As demonstrated by different spectroscopic techniques, both domains are structurally affected by the partial denaturation. The intermediate possesses only 40% of the native secondary structural content and a substantial increase in the Stokes radius as judged by circular dichroism and dynamic light scattering analyses. One-dimensional 1H NMR data prove that all tryptophan residues are in a non-native environment in the intermediate, indicating substantial changes in the tertiary structure. Still, the intermediate possesses quite a high stability for a transition intermediate of about DG 5 222 kJ mol21. Keywords: dynamic light scattering/NMR/ScHxk2/stability/ transition intermediate

Introduction Glucose kinases represent a subclass of phosphotransferases comprising hexokinases and glucokinases. The well-

established physiological role of these enzymes is the immobilization of glucose inside the cell and the initiation of its glycolytic degradation by phosphorylation. In addition, hexokinases are involved in glucose sensing in yeast, plants and mammals (Moore et al., 2003). The latter feature was exploited to design glucose-detecting biosensors in fermentation processes (Hussain et al., 2005). In humans, the alteration of hexose kinase activity may be associated with diseases such as cardiomyopathy (Barrie et al., 1979), hemolytic anemia (Peters et al., 2001) and maturity onset diabetes in the young (Vionnet et al., 1992; Gupta et al., 1997). Owing to their biological multifunctionality, similarity with human homologous enzymes and convenient availability, yeast hexokinases have become model enzymes in basic and biomedical research. The yeast Saccharomyces cerevisiae harbors two hexokinases (ScHxk1 and ScHxk2) and one glucokinase (ScGlk1) (Colowick, 1973; Lobo and Maitra, 1977; Walsh et al., 1991). ScHxk2 predominates in the presence of glucose; in contrast, the corresponding gene ScHXK2 is repressed and the genes ScHXK1 and ScGLK1 are derepressed during growth on non-fermentable carbon sources (Rodriguez et al., 2001). ScHxk2 represents a multifunctional enzyme, which in addition to its catalytic function in glucose metabolism plays a prominent role in glucose repression (Moreno and Herrero, 2002; Moreno et al., 2005) and is presumed to be involved in the signal transduction indicating glucose limitation (Heidrich et al., 1997; Randez-Gil et al., 1998; Golbik et al., 2001; Ahuatzi et al., 2007). The underlying molecular basis is the interaction of ScHxk2 with protein kinases (Kriegel et al., 1994) and phosphoprotein phosphatases (Heidrich et al., 1997; Alms et al., 1999; Sanz et al., 2000; Pelaez et al., 2009). Phosphorylation of ScHxk2 at Ser15 as occurring in vivo during glucose limitation causes the dissociation of the homodimeric enzyme and the export of its subunits from the nucleus by the exportin Xpo1. Interestingly, the formation of monomeric hexokinase is accompanied by an increase in the affinity of the enzyme for its glycolytic substrates (Golbik et al., 2001). In contrast, the signal transduction indicating glucose abundance involves the import of ScHxk2 into the nucleus by a so far unknown mechanism. The latter event is likely to be related to the unphosphorylated and presumably dimeric enzyme (Vojtek and Fraenkel, 1990; Kriegel et al., 1994; Pelaez et al., 2009). On a structural level, the monomer of ScHxk2 is composed of two domains, a small a/b domain and a large mainly a-helical domain, forming a deep cleft where the sugar substrate is bound. The binding event induces a rigid body movement of the domains leading to a closure of the active site cleft (Kuser et al., 2008). Interestingly, both domains of ScHxk2 are composed of discontinuous peptide sequences. Comprehensive analyses of the stability and folding of twodomain proteins with one or both domains consisting of continuous peptide sequences indicated that very often these domains behave as independent folding units (Jaenicke,

# The Author 2010. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]

79

H.Lilie et al.

1999). In the case of dihydrofolate reductase (DHFR), a circular permutation study suggested that always the domain with the continuous peptide sequence is most stable. Thus, depending on the artificially positioned N- and C-termini of the circular permuted variants, the highest domain stability switches between the adenosine-binding domain and the loop domain of DHFR (Svensson et al., 2006). In this work presented, we analyzed the stability and folding of yeast hexokinase isoenzyme ScHxk2 consisting of two domains with a discontinuous peptide sequence. Thermodynamic and spectroscopic analyses of urea-induced structural transitions indicated a thermodynamically stable folding intermediate, which is enzymatically inactive. Remarkably, both structural domains are partially denatured in this central intermediate, even though tryptophan fluorescence, dynamic light scattering (DLS) and one-dimensional 1 H NMR indicated a still compact but non-native structure.

Materials and methods

Protein expression, purification and activity measurements Plasmid-encoded ScHxk2 was expressed in and isolated from the triple kinase mutant strain DFY632 (hxk1::LEU2 hxk2::LEU2 glk1::LEU2 lys1 – 1 leu2-1 ura3-52) of S. cerevisiae (Vojtek and Fraenkel, 1990) grown overnight and re-cultivated prior to harvest for 90 min in yeast nitrogen base medium (Sherman, 1991) containing 0.2% casamino acids and 2% glucose at 308C. The chromatographic procedure described previously (Kriegel et al., 1994) was modified by substituting repeated separation on HiLoad 26/10 Q Sepharose HP (GE Healthcare) for DEAE-cellulose and hydroxylapatite chromatography using buffers A (20 mM Tris/HCl, 1 mM EDTA, pH 7.4) and B (buffer A containing 1 M NaCl, linear gradient 0– 500 mM) as the solvents. The purification procedure was completed by a hydrophobic interaction chromatography step on a Resource ISO 1 ml column (GE Healthcare) using buffers C (50 mM potassium phosphate, 1 mM EDTA, pH 7.4) and D (buffer C containing 50% ammonium sulfate, linear gradient 50– 12.5% ammonium sulfate) as the solvents. Protein solutions were desalted on PD-10 columns when necessary. Dithiothreitol (1 mM) and phenylmethylsulfonyl fluroide (0.5 mM) were present during all purification steps. The isolated enzyme was stored as ammonium sulfate suspension at 90% saturation of the salt at 2 – 48C. Prior to the stability studies, the enzyme was desalted and finally purified on a HiLoadTM 16/ 60 column (GE Healthcare) equilibrated with 50 mM potassium phosphate buffer, pH 7.6. The molecular mass of the isolated ScHxk2 protein lacking the N-terminal methionine was determined by electrospray ionization mass spectrometry using a Q-TOF1 mass spectrometer (Waters GmbH, Germany) to 53 810 Da with an accuracy of 0.01%. Protein concentrations were determined from the absorbance at 280 nm using a molecular absorption coefficient of 43 240 l mol21 cm21 as determined by the method of Gill and von Hippel (Gill and von Hippel, 1989). The enzymatic activity of ScHxk2 was determined by monitoring the glucose-dependent ATP consumption. The rate of NADPH formation was determined in a coupled assay using glucose 6-phosphate dehydrogenase as an auxiliary enzyme (Golbik et al., 2001). 80

Equilibrium folding studies Chemical unfolding of ScHxk2 was monitored by fluorescence and far-UV circular dichroism (CD). The intrinsic fluorescence after excitation at 280 nm was recorded from 295 to 500 nm at different urea concentrations on a Fluoromax-2 or Fluoromax-3 device at 208C. The bandpass for both the excitation and the emission monochromators were set to 5 nm, the integration time was 0.1 or 0.2 s nm21, respectively, and the response time automatically adjusted by the device. Measurements were performed using microcuvettes (10/4 mm) and all spectra were corrected for buffer baseline. The relative fluorescence intensity at 320 nm was used for monitoring denaturant-induced unfolding transitions. The final protein concentration was set to 15 mg ml21 in 20 or 50 mM potassium phosphate buffer, pH 7.6. Far-UV CD spectra of ScHxk2 were recorded on a Jasco J-810 spectropolarimeter at 208C. In these measurements, the final protein concentration was 150 mg ml21 using an optical path length of d ¼ 0.1 cm, or 850 mg ml21 and d ¼ 0.1 mm in 50 mM potassium phosphate buffer, pH 7.6. Spectra were acquired at a scan speed of 20 nm min21, a slit width of 1 nm and a response time of 2 s. The CD spectra obtained at the two different protein concentrations were identical even though the enzyme exists predominantly as a dimer at the higher concentration in the absence of any denaturant (Golbik et al., 2001). The urea-induced unfolding of ScHxk2 revealed a biphasic transition. A two-state model was fitted employing the program Sigma Plot to the second transition, in which the central intermediate (I) and the unfolded (U) states are present. The values for DG 0 (free energy of unfolding in the absence of denaturant) were determined by the linear extrapolation method using Equations [(1) – (3)]. DGA!B ¼ DG0A!B m ½D

ð1Þ

DGA!B ¼ RT ln KA=B

ð2Þ

KA=B ¼

XA X X XB

ð3Þ

DGA – B is the free energy of unfolding (between two states) at a given denaturant concentration [D], m the slope of the plot, XA the value of the specified optical signal (fluorescence F320 at 320 nm or ellipticity Q225 at 225 nm) of the initial folding state of the protein, XB refers to the final folding state of the protein and X is related to the protein at a given denaturant concentration in the respective transition. KA/B represents the equilibrium ratio of the respective protein species at any specified denaturant concentration, T is the absolute temperature and R the gas constant. Evaluation of the equilibrium requires extrapolation of the pre- and post-unfolding baselines into the transition region. Considering the intercepts (nA, nB) and the slopes (mA, mB) of the pre- and post-unfolding regime, Equation (4) was used for fitting a transition in the denaturation curve

Transition intermediate of yeast hexokinase 2

of ScHxk2. ðnA þ mA ½DÞ þ ðnB þ mB ½DÞ expðm ð½D50% ½DÞ=RTÞ X¼ 1 expðm ð½D50% ½DÞ=RTÞ

ð4Þ

Folding kinetics Folding reactions were initiated by jumping from native to unfolding conditions or vice versa at different final denaturant concentrations. Fluorescence measurements were performed as described for equilibrium folding studies. Far-UV CD measurements were performed at 220 nm and at 208C using a p * – 180 CDF Circular Dichroism Spectrometer (Leatherhead, UK). Slits were set to 2 nm, the optical path length was 2 mm. ScHxk2 was dissolved in 50 mM potassium phosphate buffer, pH 7.6 at a final protein concentration of 0.13 mg ml21 (2.5 mM).

NMR measurements One-dimensional 1H NMR spectra consisting of 1024 scans of ScHxk2 were recorded at various urea concentrations at a Bruker Advance II spectrometer using 1 mM protein samples in 20 mM sodium phosphate, pH 7.6, containing 7%/93% 2 H2O/1H2O, respectively, at 258C. Spectra were processed and analyzed by FELIX. In order to minimize disturbances from signal overlap, the right flank of W_N2 (broad peak) was not integrated beyond 10.5 ppm, and W_N3 (sharp peak) was only integrated in a narrow range around the maximum, to minimize contributions from W_N2 and W_N4.

domains, which, in contrast to many two-domain proteins as molecular models in protein folding studies so far, are each composed of a discontinuous peptide sequence. The large domain consisting of 12 a-helices and one 5-stranded b-sheet comprises the amino acids (aa) 1 – 76 and 212– 455, whereas the small a/b domain is formed by residues 77– 211 and 456 – 486 (Fig. 1). The two-domain structure of ScHxk2 is reflected in its stability. Reversible urea-induced unfolding measured by fluorescence revealed a two-step transition with a central intermediate populated at 3 – 4 M urea (Fig. 2). The midpoint of the first transition is at 1.6 M urea, the second midpoint at 4.6 M urea. A two-step transition was observed by far-UV CD as well (Fig. 2). The corresponding spectrum of this populated central intermediate showed only 40% of the signal of the native protein, indicating denaturation of major structural elements of ScHxk2. The second transition leading to a complete unfolding of the enzyme was identical in both the fluorescence and CD analysis. The thermodynamic stability of the respective central intermediate with respect to the unfolded state was calculated to DG 0 ¼ 221.9 + 2.3 kJ mol21. The cooperativity of the transition was characterized by an m-value of 13.2 + 0.9 kJ mol21 M21. In contrast, at low urea

Dynamic light scattering DLS was measured at a scattering angle of 908 and a wavelength of 532 nm using a laboratory-built apparatus, equipped with a diode-pumped, continuous-wave laser Millennia IIs (Spectra-Physics, USA) and a high quantum yield avalanche photodiode. Translational diffusion coefficients D obtained from the measured autocorrelation functions using the program CONTIN (Provencher, 1982) were converted into Stokes radii via the Stokes–Einstein equation RS ¼ kB T/ (6ph0D), where kB is Boltzmann’s constant, T the temperature in Kelvin, and h0 the solvent viscosity. Solvent viscosities and densities were measured using an Ubbelohde type viscometer PVS 1/1 (Lauda, Lauda-Ko¨nigshofen, Germany) and a digital density meter DMA 58 (Anton Paar, Graz, Austria), respectively. Refractive indices of the solvent mixtures were measured with an Abbe type refractometer. The solvents and protein solutions were filtered through 100 nm pore-size Anotop filters (Whatman, Maidstone, UK) directly into ultra-micro-fluorescence cells with a fill volume of 70 ml (Hellma, Mu¨llheim, Germany). Results

Equilibrium denaturation of ScHxk2 ScHxk2, a 54 kDa protein, exhibits a monomer/dimer equilibrium, which is affected by the protein concentration, substrates, cofactors and denaturants (Golbik et al., 2001). Under the conditions used throughout this work, ScHxk2 is monomeric unless otherwise stated. The enzyme consists of two

Fig. 1 Two-domain structure of ScHxk2 (pdb: 1IG8). The ribbon diagram shows the large mainly a-helical domain (upper right) and the small a/b domain (down left). The distribution of tyrosine (green) and tryptophan residues (red) throughout the protein is indicated in a ball-and-stick representation. The N- and C-termini of the ScHxk2 protein are labeled. The very C-terminal helix belonging to the small a/b domain is shown in black. In the crystal structure, the aa sequence starts with Asp18. For structural presentation, the program PyMol was used. In the schematic representation, the assignment of the aa sequence to the structural domains is shown. The numbering represents the aa sequence. White and gray boxes correspond to the large and small domains, respectively.

81

H.Lilie et al.

concentrations, the two applied spectroscopic techniques revealed different structural transitions, indicating an additional intermediate characterized by a high content of secondary structure and substantial loss of tertiary contacts, but being less stable than the populated central intermediate. Furthermore, the urea-induced inactivation of ScHxk2 occurred even slightly before structural changes became detectable by fluorescence analysis (Fig. 2). Consistent with the stability of ScHxk2 on chemical denaturation observed by activity measurements, the presence of the substrate glucose slightly destabilized the protein. The first unfolding transition of the enzyme in the presence of 5 mM glucose is identical to that of the corresponding activity measurements (Fig. 2). This result clearly demonstrates that the conformational changes of ScHxk2 induced by substrate binding leads to an overall destabilization of the protein. In contrast to the transition of the native to the intermediate state, the second transition from the intermediate to the fully denatured protein was not affected by glucose. Hence, the central intermediate does not bind the sugar substrate. These data revealed a complex denaturation behavior of the protein during the first transition. Taken together, the overall folding/unfolding equilibria clearly indicated a regime where the native state (N) unfolds to a stable, enzymatically inactive central intermediate (I) via a complex structural process as illustrated in Scheme 1. The central intermediate is quantitatively populated at 3.5– 4 M urea. At higher concentrations of denaturant, it is in equilibrium with the fully unfolded state (U).

Fig. 2 Unfolding transitions of ScHxk2. The protein was incubated overnight at 208C in 50 mM potassium phosphate, pH 7.6 containing various concentrations of urea. The structural transitions were analyzed by monitoring the enzyme activity (final protein concentration 2.73 mg ml21, gray circles), fluorescence (final protein concentration 15 mg ml21, excitation at 280 nm, emission at 320 nm) in the presence of 5 mM glucose (gray squares) and in the absence of glucose (black triangles), far-UV CD at 225 nm [final protein concentration 150 mg ml21 (black circles) and the integral over the spectrum between 210 and 250 nm (open circles)]. The second transition (3.5– 8 M urea) was analyzed according to a two-state model yielding DG 0 ¼ 221.9 + 2.3 kJ mol21 for both fluorescence and CD.

Scheme 1 Complex folding/unfolding of ScHxk2 via intermediates.

82

Folding and unfolding kinetics of ScHxk2 The complex equilibrium transition of denaturation of ScHxk2 is reflected in the kinetics of folding and unfolding. Figure 3A and B exemplarily shows the different folding and unfolding reactions as measured by stopped-flow CD. The

Fig. 3 Folding kinetics of ScHxk2 monitored by far-UV CD and intrinsic fluorescence. The enzyme was incubated at 208C in 50 mM potassium phosphate, pH 7.6 in the absence and presence of denaturant. Folding reactions were initiated by jumping to different denaturant concentrations as indicated in the figures. (A and B) The changes in the far-UV CD signal during unfolding or refolding were recorded at 220 nm and at an optical path length of 2 mm by a stopped-flow device. The final protein concentration was 0.13 mg ml21. N, I and U denote for the native, central intermediate and unfolded state. (C) The Chevron plot was derived from determining the rate constants of the respective folding phases at different denaturant concentrations by monitoring progress curves of the intrinsic fluorescence changes at excitation of 280 nm, at emission of 320 nm and at a protein concentration of 15 mg ml21. Refolding (open circles) was started from either 7 M urea (unfolded state) or 3.5 M urea (central intermediate). Unfolding was monitored starting from either 0 M urea (native state, closed circles) or 3.5 M urea (central intermediate, closed triangles).


full amplitude of folding/unfolding was obtained by diluting either the denatured enzyme (6 M urea) into buffer containing no denaturant or the native protein into 6.6 M urea, respectively. Several phases could be observed in folding and unfolding measurements. The fastest reaction(s) of folding/ unfolding could not be resolved by the stopped-flow technique (dead time 1 ms). There are two remaining phases present both in folding and in unfolding. Surprisingly, the folding/unfolding of the thermodynamically populated central intermediate did not follow a kinetic two-state reaction when jumping between the denatured (6.0 M urea) and the intermediate state (3 M urea), but rather showed a more complex behavior. At least, two phases could be observed, one of them being too fast for the time resolution in stoppedflow CD measurements. The folding/unfolding amplitude of the central intermediate was about 50% of the overall folding/unfolding reaction amplitude and corresponded to the respective amplitude of the equilibrium measurements (Fig. 2). These results clearly indicated that the thermodynamically observed central intermediate is kinetically populated in the folding/unfolding reaction as well. The dependence of the observed rate constants on the urea concentration was also determined in folding/unfolding reactions using fluorescence measurements. The respective Chevron plot is shown in Fig. 3C. The two V-shaped curves display minima at 1.6 and 2.7 M urea, which correspond to the first equilibrium transition between the native and the central intermediate state. The third reaction shown in Fig. 3C represents the folding/unfolding of the central intermediate and is related to the equilibrium transition between the central intermediate and the unfolded state of ScHxk2.

Spectroscopic characterization of the ScHxk2 intermediate The amplitudes of the transition measured by far-UV CD (Figs 2 and 4A) suggested that large conformational changes occurred in the central intermediate compared with the native structure. In order to attribute these conformational rearrangements to the denaturation of both or a special single domain of ScHxk2, the enzyme was analyzed by fluorescence and one-dimensional 1H NMR spectroscopy. ScHxk2 contains 4 Trp and 16 Tyr residues, which are asymmetrically distributed within the protein (Fig. 1). The large domain contains 1 Trp and 13 Tyr, while 2 Trp and 3 Tyr residues are present in the small a/b domain. As shown in Fig. 1, another Trp residue is located in the interface between the two domains. The fluorescence spectra of native, denatured and central intermediate state of ScHxk2 revealed a remarkable relationship. A large drop in fluorescence intensity was detectable in going from the native to the central intermediate state, whereas the maximum emission wavelength shifted only slightly (Fig. 4B). The latter observation suggested only a small change in solvent exposure or the microenvironments of the tryptophans in this step. In the second transition from the central intermediate to the unfolded state, however, the signal amplitude of the intrinsic fluorescence did not change further, but a large shift of the maximum emission wavelength occurred, indicating a solvent exposure of all tryptophans (Fig. 4B). A more detailed structural analysis could be obtained using one-dimensional 1H NMR (Fig. 5). In the native state of ScHxk2, the signals of all four tryptophan residues could be resolved in the range of 10.25 – 10.95 ppm, whereas in the

Fig. 4 Spectroscopic properties of ScHxk2 at various urea concentrations. ScHxk2 was incubated at 208C in 50 mM potassium phosphate, pH 7.6 in the absence of denaturant (filled circles) and in the presence of 3.5 M (open circles) and 6 M urea (dotted). (A) Far-UV CD spectra were recorded at a protein concentration of 850 mg ml21 using a cuvette of 0.1 mm optical pathlength. (B) Fluorescence spectra were measured at excitation of 280 nm and a protein concentration of 15 mg ml21.

fully denatured state, a single tryptophan signal only was monitored at 10.1 ppm. It is possible to analyze, even without an assignment of the four tryptophans to the respective signals of the native protein in a one-dimensional 1H NMR spectrum, whether the environments of all or only several of these residues are affected during the formation of the central intermediate in ScHxk2. The signal intensities of the tryptophan residues in dependence of urea from the NMR measurements are shown in Fig. 5B. At the high protein concentration used in this experiment, ScHxk2 is dimeric in the absence of urea; in the presence of at least 1 M denaturant, the protein remains in the monomeric state. Thus, changes in the tryptophan signals until 1 M urea might be associated with either structural changes or just with the dissociation of the subunits. At concentrations 1 M urea, any changes in the signals correspond to conformational changes of the monomeric enzyme. The single signal of the tryptophans in the fully denatured state and three tryptophan signals obtained under native conditions could be evaluated quantitatively. The fourth tryptophan signal of the native protein (the shoulder at 10.25 ppm) was too small to be analyzed in this manner. This signal showed the same shift on urea treatment of the protein as monitored for the other tryptophan signals. In all four cases, the signals decreased substantially in the transition region in going from the native to the central intermediate state (from 1 to 3 M urea) and displayed a plateau at around 3 M urea. These results indicate 83

H.Lilie et al.

Fig. 5 Low-field region of 600 MHz one-dimensional 1H NMR spectra of ScHxk2. The spectra were recorded at a protein concentration of 1 mM in 20 mM sodium phosphate, pH 7.6 at 258C. The buffer contained 7% 2H2O. (A) Spectra of ScHxk2 at 0 –6 M urea (front to back) are depicted. The signals of the four tryptophans (native conditions) and the single tryptophan signal under fully denatured conditions are marked by W_N1-4 (10.95 ppm, 10.55 ppm, 10.4 ppm and 10.25 ppm) and W_U (10.1 ppm). (B) Quantitative analysis of the urea-dependent tryptophan signals. The intensities of the respective tryptophan signals were plotted against the urea concentration.

the tryptophans of ScHxk2 to be no longer native-like in the central intermediate structure. Thus, the structure of both domains of ScHxk2 is changed in the equilibrium reaction between the two folded states. Furthermore, the NMR data prove the tryptophan residues not to be located in a fully denatured environment in the central intermediate state. The respective signal at 10.1 ppm exhibited only a small increase in the concentration range of 0 – 3 M urea, but evolved significantly in the second structural transition from the central intermediate to the unfolded state (4 – 6 M urea). Similar results were obtained in monitoring high-field shifted methyl groups by one-dimensional 1H NMR (data not shown). Thus, in the urea-induced unfolding reaction of ScHxk2 via the central intermediate state all tryptophan residues of the protein display a similar switch from a native to a non-native environment in the first transition, meanwhile a fully random-coil environment was achieved in the second transition only. 84

Hydrodynamic properties of the ScHxk2 intermediate The decrease in the CD amplitude of the central intermediate by 60% compared with the native protein of ScHxk2 indicates large conformational differences between these two states. Furthermore, the NMR data suggest the whole protein being part in these structural rearrangements. Large conformational changes of a protein are usually accompanied by changes of the molecular volume. The latter parameter can be estimated from the Stokes radius RS of a molecule measured by DLS. The RS value of fully denatured ScHxk2 was determined to 6.3 nm (Fig. 6). In contrast, native and monomeric ScHxk2, when analysed at 1 M urea, exhibited a Stokes radius of 3.7 nm. Measurements below 1 M urea were not considered due to the monomer/dimer equilibrium of the enzyme manifesting at the protein concentration of 0.84 mg/ml employed in these experiments. ScHxk2 equilibrates between monomers and dimers at low denaturant concentrations. Since both


Fig. 6 Urea-dependent changes of the Stokes radius RS of ScHxk2 at 208C. DLS of ScHxk2 was measured at a protein concentration of 0.84 mg ml21 in 50 mM potassium phosphate, pH 7.6 and at varying urea concentrations. All measurements were performed twice.

activity and spectroscopic data clearly demonstrated ScHxk2 to be fully structured at 1 M urea (cf. Fig. 2), the RS value measured at that urea concentration can be taken as a reliable parameter of the monomeric protein in the absence of denaturant. The urea-dependent transition of the Stokes radius of ScHxk2 showed a small plateau region around 3 M urea and an RS value of 4.8–4.9 nm (Fig. 6). This reflects the radius of the quantitatively populated central intermediate under the conditions used. Assuming the molecular volume scales approximately with the third power of RS, the Stokes radii of ScHxk2 in its native, intermediate and fully denatured state (3.7 nm for N, 4.8–4.9 nm for I, 6.3 nm for U) clearly indicate that almost half of the volume change during denaturation of ScHxk2 occurs in the structural transition from the native to the central intermediate state. The results fit well to the spectroscopic data indicating large structural changes in its formation. However, despite these substantial conformational rearrangements, this intermediate still possesses a stable, though non-native structure with a DG 0 ¼ 221.9 + 2.3 kJ mol21 to the unfolded state (cf. Fig. 2). Discussion The present paper describes the stability as well as some structural characteristics of a central folding intermediate of hexokinase isoenzyme 2 from S. cerevisiae (ScHxk2). In contrast to most two-domain proteins utilized so far as molecular models in protein folding, monomeric ScHxk2 consists of two domains that are both formed from discontinuous peptide sequences (Kuser et al., 2008). The large a-helical domain consists of residues 1 – 76 and 212 – 455, whereas the small a/b domain comprises aa 77 –211 and 456 – 486 (cf. Fig. 1). In the urea-induced equilibrium transition analyzed in this study, one stable central intermediate accumulated quantitatively (Fig. 2). Earlier differential scanning calorimetry (DSC) studies have shown that thermally induced unfolding of ScHxk occurs as a two-step process (Catanzano et al., 1997; Kumar et al., 2004). The respective two unfolding reactions have been attributed to the two domains of the protein (Takahashi et al., 1981; Kumar et al., 2004). However, the transitions could not be specifically correlated to the individual domains. Furthermore, since the transitions were highly overlapping, the structural central intermediate

was only partially populated during heating. Only at pH .9 both transitions could be well separated. On the basis of intrinsic tryptophan fluorescence, it has been speculated that the small domain was unfolded whereas the large domain remained folded in the respective intermediate (Kumar et al., 2004). This interpretation is clearly limited by the fact that such fluorescence data are difficult to interpret, since the quantitative contribution of each single tryptophan to the overall signal is unknown. In order to address the problem of a structural interpretation of the central folding intermediate of ScHxk2 experimentally, the thermodynamically stable intermediate identified in the present work was subjected to CD, onedimensional 1H NMR and hydrodynamic analyses in addition to fluorescence measurements. These studies indicated that 60% of the secondary structure of ScHxk2 is lost in this populated species and the microenvironment of all tryptophans distributed throughout the whole protein exhibits nonnative, but not random-coil characteristics. The dramatic structural changes measured are accompanied by a substantial increase in the Stokes radius. The entirety of the novel data indicates a very unusual structure of the central transition intermediate of the two-domain protein ScHxk2 with both domains being partially unfolded and the remaining structure revealing a high thermodynamic stability. The folding of two-domain proteins has been analyzed in great detail. Remarkably, all two-domain proteins used so far as model proteins in folding studies consist of at least one domain comprising a continuous peptide sequence or even two sequential domains connected by a peptide linker. Examples for the latter case are the antibody light chain, g-crystallin, phosphoglycerate kinase and the gene-3-protein (G3P) of fd phage (Tsunenaga et al., 1987; Rudolph et al., 1990; Parker et al., 1996; Holliger et al., 1999). In all these cases, the domains fold independently of each other even though domain – domain interactions in the native state may stabilize one of the two domains significantly as described for g-crystallin (Palme et al., 1997). Equilibrium denaturation of g-crystallin is characterized by a pronounced two-step transition, where an intermediate is populated with the N-terminal domain structured and the C-terminal domain unfolded (Rudolph et al., 1990). Even variants of g-crystallin with an engineered original linker between the two domains still exhibit two-step equilibrium transitions (Mayr et al., 1994; Jaenicke 1999). Similarly, intermediates with one domain folded and the other domain unfolded have been observed in antibody light chains (Tsunenaga et al., 1987) and phosphoglycerate kinase (Parker et al., 1996). Lysozyme, DHFR and G3P are well-studied examples for two-domain proteins with one continuous and one discontinuous domain. Chemical denaturation of these proteins occurs cooperatively without accumulation of a thermodynamically stable intermediate (Touchette et al., 1986; Ibarra-Molero and Sanchez-Ruiz, 1997). In the case of G3P, both domains are formed prior to their slow docking representing the ratelimiting step, which is limited by prolyl isomerization (Martin and Schmid, 2003). For DHFR, a variety of circular permutations have been analyzed (Svensson et al., 2006). Several of them revealed two-step equilibrium transitions upon urea-induced or thermally induced unfolding. In the respective intermediates, always the continuous domain was folded and the discontinuous domain unfolded, even though, 85

H.Lilie et al.

depending on the circular permutation, in one variant, the continuous domain was the adenosine-binding domain, whereas in another variant, the continuous domain was the originally discontinuous loop domain (Svensson et al., 2006). These results indicate that the discontinuous domain of DHFR is always disfavored in stability. Still, all these variants can fold to the native state and display enzymatic activity. In contrast to the proteins described above, ScHxk2 consists of two domains composed of discontinuous peptide sequences. As shown in this study, this protein populates a thermodynamically stable central intermediate at 3– 4 M urea. This intermediate does not consist of a single folded and an unfolded domain. Instead, both domains are structurally affected in the transition between the native and central intermediate state. The very C-terminal fragment of the a/b domain (aa 458– 486; 3.1 kDa) contains a single a-helix (black in Fig. 1), which is packed against the concavely shaped b-sheet of the same domain. On the other side, this helix faces a short helix of the large domain, which is located at the concave side of a b-sheet as well. Thus, the contact site of the two ScHxk2 domains is largely represented by two perpendicularly aligned helices. Sliding of these helices against each other permits the rigid body movement of the two domains, finally leading to a closure of the cleft between them on sugar substrate binding (Kuser et al., 2000). The a-helix propensity of ScHxk2 was predicted by the program APSSP2 (ExPASy Proteomics Server) and revealed very high values at aa 211– 220 and aa 458 – 486. On the basis of these structural data related to the enzyme mechanism of ScHxk2, it is obvious that a structural communication exists between the two domains. It seems conceivable that partial unfolding of one domain is transmitted to the other domain via structural changes in the domain – domain contact site by strongly destabilizing the contact helix. In this manner, both domains are affected simultaneously in the central intermediate. Interestingly, statistical data analyses of hexokinase showed a co-evolved set of aa comprising the active site and the hinge region between the two domains (Kuser et al., 2008). Thus, there is a network of interactions connecting the active site with the interaction site of the two domains, which is functionally important in the induced fit mechanism of enzyme catalysis (Kuser et al., 2008). These conserved residues might also be important in the signaling of conformational changes from one domain to the other during folding/unfolding reactions. There are some examples where helices only form in the presence of native protein contacts. In the case of the interaction of RNase S and its S-peptide, the latter is unfolded in its isolated state, but adopts an a-helical conformation on binding to RNase S. On the other hand, RNase S folds oneself, but undergoes an induced fit on binding the S-peptide (Goldberg and Baldwin, 1998a,b). The situation is slightly different in nicked staphylococcal nuclease, where the fragments 6 – 48 and 49– 149 are largely unstructured, but can associate to form a partially active enzyme (Taniuchi and Anfinsen, 1969). The nuclease fragment 6 – 48 contains part of a b-sheet, whereas the large fragment 49– 149 consists of two additional b-strands and three helices (Chen et al., 2000). Remarkably, two of these helices as well as two b-turns are stabilized by local interactions even in the denatured state (Alexandrescu et al., 1994). In the case of 86

ScHxk2, the non-covalent contacts between the two domains are mediated almost exclusively by two helices, which are aligned perpendicularly (cf. Fig. 1). Opposite to their interaction sites, both helices pack against b-sheets of the respective domains. It seems reasonable to assume that the short interacting helix (aa 211– 220) of the large domain as well as the helix of the a/b domain (aa 458– 486) might only be stably structured when interacting with each other and in turn stabilize the adjacent b-sheets of both domains. In conclusion, a transition intermediate of ScHxk2 has been characterized on a structural, kinetic and thermodynamic level. CD and DLS data indicate a substantial loss of secondary structure and an increase in the Stokes radius of the central intermediate compared with the native structure. Fluorescence spectra prove significant changes in the tertiary structure of this intermediate in comparison with the native state. 1H NMR showed that these structural changes are not restricted to one of the two domains of ScHxk2. Instead, both domains are partially unfolded in a central transition intermediate. Although the large conformational changes affect the whole enzyme, the stability of the intermediate structure still resembles that of folded proteins. In order to refine a structural picture of the ScHxk2 central intermediate, it appears promising to analyze variants containing mutations that disrupt the contact sites of the domains and to design single engineered domains of the enzyme. Acknowledgement We thank Renate Nitsch for technical assistance.

Funding This work was supported by the GRK1026 of the Deutsche Forschungsgemeinschaft. References Ahuatzi,D., Riera,A., Pelaez,R., Herrero,P. and Moreno,F. (2007) J. Biol. Chem., 282, 4485–4493. Alexandrescu,A.T., Abeygunawardana,C. and Shortle,D. (1994) Biochemistry, 33, 1063–1072. Alms,G.R., Sanz,P., Carlson,M. and Haystead,T.A.J. (1999) EMBO J., 18, 4157– 4168. Barrie,S.E., Saad,E.A., Ubatuba,S., Dasilvalacaz,P. and Harris,P. (1979) Res. Commun. Chem. Path., 23, 375–381. Catanzano,F., Gambuti,A., Graziano,G. and Barone,G. (1997) J. Biochem., 121, 568 –577. Chen,J.M., Lu,Z.Q., Sakon,J. and Stites,W.E. (2000) J. Mol. Biol., 303, 125–130. Colowick,S.P. (1973) In Boyer,P.D. (ed.), The Enzymes. Academic Press, New York, pp. 1– 48. Gill,S.C. and von Hippel,P.H. (1989) Anal. Biochem., 182, 319– 326. Golbik,R., Naumann,M., Otto,A., Mu¨ller,E.C., Behlke,J., Reuter,R., Hu¨bner,G. and Kriegel,T.M. (2001) Biochemistry, 40, 1083–1090. Goldberg,J.M. and Baldwin,R.L. (1998a) Biochemistry, 37, 2546– 2555. Goldberg,J.M. and Baldwin,R.L. (1998b) Biochemistry, 37, 2556– 2563. Gupta,B.L., Nehal,M. and Baquer,N.Z. (1997) Ind. J. Exp. Biol., 35, 792–795. Heidrich,K., Otto,A., Behlke,J., Rush,J., Wenzel,K.W. and Kriegel,T. (1997) Biochemistry, 36, 1960–1964. Holliger,P., Riechmann,L. and Williams,R.L. (1999) J. Mol. Biol., 288, 649–657. Hussain,F., Birch,D.J.S. and Pickup,J.C. (2005) Anal. Biochem., 339, 137–143. Ibarra-Molero,B. and Sanchez-Ruiz,J.M. (1997) Biochemistry, 36, 9616– 9624. Jaenicke,R. (1999) Prog. Biophys. Mol. Biol., 71, 155– 241.

Transition intermediate of yeast hexokinase 2 Kriegel,T.M., Rush,J., Vojtek,A.B., Clifton,D. and Fraenkel,D.G. (1994) Biochemistry, 33, 148–152. Kumar,D.P., Tiwari,A. and Bhat,R. (2004) J. Biol. Chem., 279, 32093– 32099. Kuser,P.R., Krauchenco,S., Antunes,O.A.C. and Polikarpov,I. (2000) J. Biol. Chem., 275, 20814–20821. Kuser,P., Cupri,F., Bleicher,L. and Polikarpov,I. (2008) Proteins, 72, 731–740. Lobo,Z. and Maitra,P.K. (1977) Arch. Biochem. Biophys., 182, 639–645. Martin,A. and Schmid,F.X. (2003) J. Mol. Biol., 329, 599–610. Mayr,E.M., Jaenicke,R. and Glockshuber,R. (1994) J. Mol. Biol., 235, 84– 88. Moore,B., Zhou,L., Rolland,F., Hall,Q., Cheng,W.H., Liu,Y.X., Hwang,I., Jones,T. and Sheen,J. (2003) Science, 300, 332– 336. Moreno,F. and Herrero,P. (2002) FEMS Microbiol. Rev., 26, 83– 90. Moreno,F., Ahuatzi,D., Riera,A., Palomino,C.A. and Herrero,P. (2005) Biochem. Soc. Trans., 33, 265–268. Palme,S., Slingsby,C. and Jaenicke,R. (1997) Protein Sci., 6, 1529–1536. Parker,M.J., Spencer,J., Jackson,G.S., Burston,S.G., Hosszu,L.L.P., Craven,C.J., Waltho,J.P. and Clarke,A.R. (1996) Biochemistry, 35, 15740– 15752. Pelaez,R., Herrero,P. and Moreno,F. (2009) J. Biol. Chem., 284, 20548– 20555. Peters,L.L., Lane,P.W., Andersen,S.G., Gwynn,B., Barker,J.E. and Beutler,E. (2001) Blood Cell Mol. Dis., 27, 850–860. Provencher,S.W. (1982) Comput. Phys. Commun., 27, 229–242. Randez-Gil,F., Sanz,P., Entian,K.D. and Prieto,J.A. (1998) Mol. Cell. Biol., 18, 2940– 2948. Rodriguez,A., de la Cera,T., Herrero,P. and Moreno,F. (2001) Biochem. J., 355, 625– 631. Rudolph,R., Siebendritt,R., Nesslauer,G., Sharma,A.K. and Jaenicke,R. (1990) Proc. Natl Acad. Sci. USA, 87, 4625– 4629. Sanz,P., Alms,G.R., Haystead,T.A.J. and Carlson,M. (2000) Mol. Cell. Biol., 20, 1321– 1328. Sherman,F. (1991) Methods Enzymol., 194, 3 –21. Svensson,A.K.E., Zitzewitz,J.A., Matthews,C.R. and Smith,V.F. (2006) Protein Eng. Des. Sel., 19, 175–185. Takahashi,K., Casey,J.L. and Sturtevant,J.M. (1981) Biochemistry, 20, 4693–4697. Taniuchi,H. and Anfinsen,C.B. (1969) J. Biol. Chem., 244, 3864– 3875. Touchette,N.A., Perry,K.M. and Matthews,C.R. (1986) Biochemistry, 25, 5445–5452. Tsunenaga,M., Goto,Y., Kawata,Y. and Hamaguchi,K. (1987) Biochemistry, 26, 6044– 6051. Vionnet,N., Stoffel,M., Takeda,J., et al. (1992) Nature, 356, 721– 722. Vojtek,A.B. and Fraenkel,D.G. (1990) Eur. J. Biochem., 190, 371 –375. Walsh,R.B., Clifton,D., Horak,J. and Fraenkel,D.G. (1991) Genetics, 128, 521–527.

87