Substrate-Induced Change in the Quaternary Structure of Type 2 Isopentenyl Diphosphate Isomerase from Sulfolobus shibatae

Abstract

Type 2 isopentenyl diphosphate isomerase catalyzes the interconversion between two active units for isoprenoid biosynthesis, i.e., isopentenyl diphosphate and dimethylallyl diphosphate, in almost all archaea and in some bacteria, including human pathogens. The enzyme is a good target for discovery of antibiotics because it is essential for the organisms that use only the mevalonate pathway to produce the active isoprene units and because humans possess a nonhomologous isozyme, type 1 isopentenyl diphosphate isomerase. However, type 2 enzymes were reportedly inhibited by mechanism-based drugs for the type 1 enzyme due to their surprisingly similar reaction mechanisms. Thus, a different approach is now required to develop new inhibitors specific to the type 2 enzyme. X-ray crystallography and gel filtration chromatography revealed that the enzyme from a thermoacidophilic archaeon, Sulfolobus shibatae, is in the octameric state at a high concentration. Interestingly, a part of the regions that are involved in the substrate binding in the previously reported tetrameric structures is integral to the formation of the tetramer-tetramer interface in the substrate-free octameric structure. Site-directed mutagenesis at such regions resulted in stabilization of the tetramer. Small-angle X-ray scattering, tryptophan fluorescence, and dynamic light scattering analyses showed that substrate binding causes the dissociation of an octamer into tetramers. This property, i.e., incompatibility between octamer formation and substrate binding, might provide clues to develop new specific inhibitors of the archaeal enzyme.

INTRODUCTION

Isopentenyl diphosphate isomerase (IDI) catalyzes the isomerization of active units for isoprenoid biosynthesis, i.e., isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) (17). Because DMAPP is the first primer substrate for prenyl elongation reactions, the enzyme plays an important role in control of isoprenoid metabolism. Especially for the organisms that utilize only the mevalonate pathway for the biosynthesis of the active isoprene units, IDI is essential for growth. Type 2 IDI is found in almost all archaea and in some bacteria, including human pathogens such as methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus, while its nonhomologous isozyme, type 1 IDI, exists in eukaryotes and in other bacteria. Thus, type 2 IDI is regarded as a good target for the development of antibiotics effective against certain pathogenic bacteria. Because type 2 IDI is a flavin mononucleotide (FMN)- and NADPH-dependent flavoenzyme, it has been expected to catalyze the same reaction with a different mechanism from that used by type 1 IDI, which is a nonflavoenzyme without any homology with type 2 IDI (15). However, recent studies have shown that the reaction mechanism of type 2 IDI is surprisingly similar to that of type 1 IDI (22, 24, 25, 29, 30, 33). Both types of enzymes catalyze the isomerization between IPP and DMAPP via a protonation-deprotonation mechanism, probably through the same carbocation intermediate. Indeed, the mechanism-based inhibitors developed for type 1 IDI can also inhibit the activity of type 2 IDI (14, 24). Thus, a different approach is now required to develop new drugs that can inhibit only type 2 IDI.

The crystal structures of a few type 2 IDIs have been reported so far. The structures of type 2 IDIs from Bacillus subtilis (27) and Thermus thermophilus (7) have been reported as homooctamers, but type 2 IDI from thermophilic archaeon Sulfolobus shibatae has a homotetrameric configuration (22, 33). Interestingly, a boundary sedimentation experiment for B. subtilis type 2 IDI showed that the octamer dissociates into a smaller quaternary structure in the presence of FMN, NADPH, and a substrate (18). In addition, type 2 IDIs from the thermophilic archaeon Methanothermobacter thermautotrophicus (1) and cyanobacterium Synechosystis sp. strain PCC 6803 (2) are reported to exist in an equilibrium between tetrameric and octameric states in a concentration-dependent manner, with K_ds (dissociation constants) of 17 and 0.2 μM, respectively. These results suggest that these enzymes are either in a tetrameric or octameric state or in an equilibrium and that their quaternary configurations depend on the concentrations, the presence of the cofactors and/or the substrates, and their origins. However, the relationship between the quaternary structure and the function of the enzyme was unclear.

In the present study, we found that S. shibatae type 2 IDI also forms an octamer under certain conditions for crystallization. Interestingly, in its octameric form, some of the amino acid residues that have been shown to be involved in substrate binding move away from the active site and interact with the other subunit at the tetramer-tetramer interface. The formation of an octamer in the solution was confirmed by gel filtration column chromatography. Double mutations introduced at the tetramer-tetramer interface, which are aimed at the disturbance of intersubunit interactions without affecting the formation of the substrate-binding site, successfully stabilized the tetrameric state. Moreover, dissociation of an octamer into tetramers by the addition of the substrate was confirmed by small-angle X-ray scattering (SAXS), tryptophan fluorescence, and dynamic light scattering (DLS) analyses. From DLS analysis, we determined that the apparent IPP concentration that caused the dissociation was between the K_d for IPP, which was obtained from tryptophan fluorescence analysis, and the K_m for IPP. These data clearly show that the substrate binding affects the octamer-tetramer equilibrium and suggest the possibility of the reverse relationship. Thus, the octamer formation can be an important property that controls enzyme activity and, therefore, might help us generate new agents that will specifically inhibit the enzyme.

MATERIALS AND METHODS

Enzyme purification and gel filtration chromatography.

S. shibatae type 2 IDI was recombinantly expressed in Escherichia coli BL21(DE3) containing the plasmid pET-idi, as described elsewhere (34). The enzyme with a polyhistidine tag at its N terminus was purified from the disrupted E. coli cells by heat treatment at 55°C for 30 min and by affinity chromatography using a HisTrap column (GE Healthcare). The tag was excised from the enzyme by treatment with restriction-grade thrombin (Novagen), and the excised tag was removed by another round of affinity chromatography. The enzyme was dialyzed against a buffer that contained 10 mM Tris-HCl (pH 7.7), 1 mM EDTA, and 10 mM 2-mercaptoethanol and was then purified by ion-exchange chromatography using a Mono Q 5/50 column (GE Healthcare) run with the same buffer and a linear gradient of 0 to 1 M NaCl. Purified enzyme was used for the analyses after the buffer was exchanged to buffer I, which contained 10 mM Tris-HCl (pH 7.7), 1 mM EDTA, 10 mM 2-mercaptoethanol, and 0.15 M NaCl, or buffer II, which contained 100 mM 3-(N-morpholino)propanesulfonic acid-NaOH ( pH 7.0) and 0.1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate, either by dialysis or via Amicon Ultra 10K centrifugal filters (Millipore).

For crystallization, the enzyme was further purified by gel filtration chromatography using HiLoad 16/60 Superdex 200 prep grade (GE Healthcare) eluted with buffer I. The column was also used to determine the molecular weight of the wild-type and mutant type 2 IDIs. Ferritin, catalase, and aldorase were used as molecular weight markers, while thyroglobulin was used to determine the void volume.

Crystallization, X-ray data collection, structure solution, and refinement.

The purified S. shibatae type 2 IDI was crystallized at 20°C using the hanging-drop vapor diffusion method with a reservoir solution containing 0.1 M Tris-HCl (pH 8.0), 0.2 M sodium citrate, and 30% (vol/vol) polyethylene glycol 400 (PEG 400). The crystals were soaked in a reservoir solution containing 32% (vol/vol) PEG 400 and 10 mM NADH for 1 h and placed directly into a nitrogen stream at 95 K. Data for a 1.7-Å resolution were collected using a Beamline BL-5A at the Photon Factory (KEK, Tsukuba, Japan). X-ray diffraction data of the crystal were processed and scaled using an HKL2000 (23). Data collection statistics are summarized in Table 1. The new, substrate-free structure in a reduced form at a resolution of 1.7 Å was refined using a Refmac program (21) from the CCP4 suite and the atomic coordinate of the previously reported structure at a resolution of 2.3 Å (PDB code 2ZRV) (33). Manual fitting of the model was carried out using a Coot program (9). The Ramachandran plot was generated by RAMPAGE (19). The refinement statistics are summarized in Table 1. The figures for the protein model were drawn using a PyMOL program (Schrödinger, LLC).

Table 1.

Data collection and refinement statistics

Crystal type	Type 2 IDI, octameric forma
Data collection and processing statistics
Beamline	PF-BL5A
Space group	P43212
Unit cell dimension (Å)
a = b (Å)	100.84
c (Å)	336.65
a = b = g (°)	90.000
Wavelength (Å)	1.0000
Resolution (Å)	50.00–1.70 (1.73–1.70)
Measured	1,538,069
I/σI	25.5 (5.5)
Redundancy	8.5 (3.7)
Completeness (%)	94.7 (77.1)
Rmergeb (%)	14.3 (31.4)
Refinement statistics
Resolution	35.15-1.70
Protein atoms	11,200
Ligand atoms	124
Water molecule	724
Rwork/Rfree(%)	18.5/21.7
Root mean square deviations
Bond lengths (Å)	0.033
Bond angles (°)	2.628
Ramachandran plot
Favored regions (%)	97.2
Allowed regions (%)	2.8
Outlier regions (%)	0

^aNumbers in parentheses are for the highest shell.

^bR_merge = 100Σ|I− <I>|/Σ I, where I is the observed intensity and <I> is the average intensity of multiple observations of symmetry-related reflections.

Mutagenesis.

The E13R mutation was introduced into pET-idi using a QuikChange mutagenesis kit (Stratagene) and the oligonucleotide primer 5′-GTGGAGCACGTGCGCATTGCCGCTTTTG-3′ along with its complementary primer. Then the R235E mutation was additionally introduced using a 5′-GATAAGGGATATCGAAAGAGGGAACTG-3′ primer along with its complementary primer.

Enzyme assay and kinetics.

An IDI assay was performed as described elsewhere (11). Kinetic parameters were calculated by fitting the Michaels-Menten equation to the substrate concentration versus initial velocity plot using Kaleidagraph (SYNERGY software).

SAXS measurements.

SAXS measurements were carried out using the optics and detector system for SAXS (superadvanced X-ray emission spectrometer [SAXES] system) (31, 32) installed on a Beamline BL-10C with a 2.5-GeV storage ring at the Photon Factory, KEK,Tsukuba, Japan. The circulating electron current in the storage ring was 454 mA. A wavelength (λ) of 1.488 Å was used, and the specimen-to-detector distance was about 90 cm. Raw data from Q values of 1.7 × 10⁻² to 2.00 × 10⁻¹ Å ⁻¹ (the Bragg spacing equivalent to dB = 370 to 31.4 Å; Q = 4πsin θ/λ, where 2θ is the scattering angle) were measured using a 2-dimensional imaging plate (R-AXIS VII; Rigaku, Tokyo, Japan). A detailed description of SAXS measurements is provided elsewhere (13). The net scattering intensities were obtained by subtraction from the blank buffer. The radius of gyration, Rg, and forward scattered intensity normalized with respect to the protein concentration, J(0)/C, were determined using Guinier's approximation (10), log J(Q) versus Q × Q. The weight-average molecular weight, M_w, was determined by referring to J(0)/C of bovine liver catalase. To distinguish differences in the quaternary structure (12, 13), Kratky plots (16), J(Q)× Q × Q versus Q, were constructed. For measurement, 40 μl of a 2.8-mg/ml enzyme solution in either buffer I or II was used. To the solution, 0.5 mM NADH, 2.5 mM MgCl₂, 100 μM IPP, or both 2.5 mM MgCl₂ and 100 μM IPP were added to elucidate their effects on the quaternary structures of the wild-type and E13R/R235E mutant enzymes.

Tryptophan fluorescence analysis.

To 500 μl of 2.5-mg/ml or 0.002-mg/ml enzyme solution in buffer II, 1 mM IPP was gradually added for titration. Fluorescence from tryptophan residues in the enzyme was measured using an F-4500 fluorometer (Shimadzu). The excitation and emission wavelengths were 295 and 350 nm, respectively. The fluorescence intensity (F) and the concentration of IPP ([IPP]) were corrected for the increase in the volume of the solution. The data were plotted ([IPP] versus F) and fitted with the following equation to calculate K_d for IPP using Kaleidagraph: F = F_i + ΔF[IPP]/([IPP] + K_d), where F_i is the initial fluorescence intensity. For fitting, K_d and ΔF, the maximal change in fluorescence intensity expected when [IPP] ≫ K_d, were set as variable parameters.

DLS analysis.

Two hundred microliters of 2.5-mg/ml solution of the wild-type enzyme was used for DLS analysis on a Zetasizer Nano (Malvern Instruments). The solution was titrated with 1 mM IPP, and the radius sizes of particles (r) were measured. By fitting the following equation to the plot of IPP concentration ([IPP]) versus r, the K_app value, which represents the [IPP] that caused the half-maximal change in the r value, was calculated. This equation is based on a simple one-to-one equilibrium between the octameric and tetrameric states, because the stoichiometric ratio of the states (1:2) is offset by the difference in the intensity of scattering between them. The equation is as follows: r = r_o + Δr[IPP]/([IPP] + K_app), where r_o represents the average of r values in the absence of IPP, and Δr is the maximal change in r expected when [IPP] ≫ K_app. Whereas [IPP] was corrected for the increase of the volume of the solution, r was not adjusted because it was hardly changed when water was used for titration instead of the IPP solution.

RESULTS

Crystal structure of octameric S. shibatae type 2 IDI.

We tried to obtain the crystal structure of S. shibatae type 2 IDI in complex with NADH, which is considered to be a physiological reducing agent for the flavoenzyme as well as for NADPH (34), by using an enzyme that was carefully purified for crystallization and by soaking the crystals in the buffer containing a high concentration of NADH. As a result, we succeeded in solving the new structure of S. shibatae type 2 IDI at a resolution of 1.7 Å, which is the highest level yet obtained (Table 1, PDB code 3VKJ). Although the new structure contained no electron density of NADH, a remarkable difference between the new and the previously reported structures (22, 33) was found in the N-terminal region (Fig. 1A and B). The new structure contained the N-terminal region interacting with the neighboring tetramer unit related by the 2-fold axis in a crystallographic symmetry (Fig. 1A). In contrast, no model could have been built for the N-terminal region (6 residues) of the previous NADH-reduced structure (PDB code 2ZRV; resolution, 2.3 Å) due to the lack of electron density, which had led us to decide that the structure is a tetramer. A root mean square deviation between the two structures without the N-terminal region (8 residues) is 0.247 Ų, which means that the two structures are almost the same except the N-terminal region. The new and the previous NADH-reduced structures were obtained under the same conditions. Therefore, revealing the electron density in the N-terminal region of the new structure is thought to be a result of improvement of the crystal quality and the resolution. Moreover, the orientation and conformation of the N-terminal region in the new structure distinctly differ from those of the previously reported substrate-complex structures (22, 33), in which the N-terminal region forms an α-helix to construct a part of the substrate-binding site. A superimposed structure of the type 2 IDI-FMN-IPP ternary complex in the reduced state (PDB code 3B05) showed that the N-terminal α-helix in the new structure partially unwinds to interact with the α-helix (α11′) and loop regions between α11′ and α12′ in the neighboring tetramer unit (Fig. 1B). As a result of the movement of the unwound N-terminal region, Ile4, Val5, Arg7, Glu10, Glu13, and Phe17 on α1, Arg235 and Arg236 on α11, and Gly237 and Asn238 on the loop region between α11 and α12 all interact with the next tetramer unit through hydrogen binding and hydrophobic interactions (Fig. 1C). Besides these direct interactions of the residues, Arg7, Glu13, Arg235, H₂O-1, H₂O-2, Arg7′, and the main chain oxygen atom of Arg235′ create a hydrogen bonding network that forms interactions between the 2 tetramer units via the water molecules (Fig. 1C). The interface surface area between the 2 tetramer units of the new structure is about 670 Ų per monomer, which is about 2.8 times greater than that of the type 2 IDI-FMN-IPP complex structure (about 240 Ų per monomer). These interactions of the new structure between the 2 tetramer units, which have been regarded as the fundamental quaternary structure of S. shibatae type 2 IDI, imply the formation of an octamer complex.

Fig 1.

Crystal structure of type 2 IDI from S. shibatae obtained by soaking the crystals in NADH solution. (A) Octameric structure represented by changing the color of each monomeric subunit. The N-terminal residue (Ile4) in each subunit is shown by a sphere. FMN is shown in the stick model. (B) Structural change is indicated by superimposing a subunit in the tetrameric, reduced type 2 IDI-FMN-IPP complex structure (pink, PDB code 3B05) onto that in the octameric structure (blue). A subunit from another tetramer, which interacts with the blue subunit, is shown in gray. The polypeptide chains are shown in the ribbon model, while FMN and IPP are shown in sticks and Mg²⁺ is in a sphere. (C) Stereoscopic view of the tetramer-tetramer interface, from basically the same angle as for panels A and B. A subunit (green) interacts with another (white) at the interface. The expected hydrogen bonds important for subunit interaction are shown in the dotted lines. The residues mutated in this study, i.e., Glu13 and Arg235, are shown by numbers in red.

Because octamer formation was reported for type 2 IDIs from other organisms (1, 2, 7, 18, 26, 27), we considered that it might not be an artifact observed only under the conditions for crystallization. Thus, we used gel filtration column chromatography to examine the quaternary structure of the enzyme (Fig. 2). In our previous study (34), an enzyme from S. shibatae with a fused polyhistidine tag on its N terminus was eluted from a gel filtration column at the retention time of a homotetramer. In the present study, however, the enzyme from which the tag had been excised was used. Approximately 1 ml of a solution of 5.4 mg/ml (130 μM monomer) purified enzyme, which tightly bound FMN, was loaded onto a 16- by 600-mm column. The elution buffer (buffer I; see Materials and Methods) did not contain NADH, substrates, or an Mg²⁺ ion, all of which are required for enzyme activity. As shown in Fig. 2A, the enzyme was eluted from the column as a single peak at the retention time of a 319-kDa protein (see Fig. 2D), suggesting that an S. shibatae type 2 IDI, which is ∼41 kDa in monomer form, exists mostly in the octameric state. When 1 ml of 0.6 mg/ml (15 μM monomer) of the enzyme solution was loaded, 2 small, comparable peaks were observed (Fig. 2B). One of the peaks was eluted almost simultaneously with the octamer, while the other was eluted at the retention time of a 174-kDa protein (Fig. 2D). This result suggests that S. shibatae type 2 IDI is in an equilibrium between the octameric and tetrameric states, as reported with a few other type 2 IDIs. Because the heights of the two peaks were almost identical, the K_d for the octamer-tetramer equilibrium of the S. shibatae type 2 IDI was estimated—if dilution in the column was ignored—to be ∼3.75 μM. This parameter lies between those of the homologues from M. thermautotrophicus (1) and Synechocystis (2) which had been determined by sedimentation equilibrium experiments to be 17 and 0.2 μM, respectively.

Fig 2.

Gel filtration chromatography of the wild type and E13R/R235E mutant. The elution profiles of 5.4 (A) and 0.6 (B) mg/ml of the wild-type enzyme and 5.7 mg/ml of the E13R/R235E mutant (C) were used for molecular weight calculation. The standard curve (D) was constructed with aldorase, catalase, and ferritin. The open and closed circles and diamonds represent the correspondence with the elution peaks.

Mutagenesis aimed at stabilizing the tetrameric form.

To inhibit the formation of an octamer by disrupting the interaction between two tetramers, we performed site-directed mutagenesis at the regions involved in the interaction. Glu13 at the N-terminal α-helix 1 and Arg235 at α-helix 11 were selected as the mutated positions to form the E13R/R235E double mutant because Glu13 interacts with Arg7′ from the other tetramer via water (Fig. 1C). Glu13 also interacted with Arg235 of the same subunit, as observed in the crystal structures of the tetrameric, substrate-complexed type 2 IDI (33). Thus, the mutual substitution was expected to diminish octamer formation by disturbing the interactions between two tetrameric subunits, without significantly affecting the substrate-binding function. The mutant was shown to be, at least partially, heat stable because it tolerated heat treatment at 55°C for 30 min in the purification process and retained enzyme activity after the treatment. According to our kinetic studies, the E13R/R235E mutant and the wild-type enzyme showed similar V_max values (108 ± 6 and 139 ± 21 nmol/mg/min [means and standard deviations], respectively) while the K_m of E13R/R235E (7.14 ± 1.47 μM) was smaller than that of the wild type (46.3 ± 14.8 μM). In the gel filtration column chromatography of a 1-ml solution of the E13R/R235E mutant of 5.7 mg/ml (140 μM monomer), the enzyme was eluted as a single peak at the retention time of a 182-kDa protein (Fig. 2C and D). This result indicated that the mutant was in the tetrameric state even at a concentration where the wild-type enzyme dominantly formed an octamer.

Effect of substrate binding on the quaternary structure.

Dissociation of an octamer to a smaller quaternary structure under the conditions that enable the isomerase reaction, i.e., in the presence of FMN, NADPH, IPP, and Mg²⁺ ion, has been reported with type 2 IDI from B. subtilis (18). Besides, the fact that the amino acid residues, such as Arg7, that are critical for substrate binding in S. shibatae type 2 IDI (33) are involved in octamer formation implies that the substrate binding and the octamer formation are incompatible. Therefore, we performed SAXS analysis of both the wild-type S. shibatae type 2 IDI and the E13R/R235E mutant, both of which tightly bind FMN, to research the effect of binding of the substrate and/or cofactors on their quaternary structures. The radius of gyration (Rg) values of the octamer and tetramer were calculated to be 44.0 Å and 37.8 Å, respectively, based on the crystal structures via the CRYSOL program (28). The difference in the Rg value between an octamer and a tetramer was only 6.2 Å, so it was difficult to use a Guinier plot (10), which enables one to evaluate Rg, to determine the quaternary state of the enzyme, which is likely in an equilibrium between octamer and tetramer. Moreover, it also was difficult to calculate the molecular weight change by the forward scattering intensity normalized to the protein concentration, J(0)/C, which can also be given from a Guinier plot. Thus, we used a Kratky plot, which is sensitive to changes in quaternary structure, to compare the difference between the octamer and the tetramer (16). As shown in Fig. 3, the curves obtained for all conditions of the wild type and E13R/R235E mutant contained two peaks, which is typical for oligomeric proteins (11, 12). The Q values of the first peak and the curves between first peak and second peak reflect the quaternary structure. The Kratky plot of a 2.8-mg/ml (68 μM monomer) solution of the wild-type enzyme (dark blue) fitted well with the theoretical curve calculated from the octameric crystal structure described above (green) (Fig. 3A). Although each addition of 0.5 mM NADH (magenta), 2.5 mM MgCl₂ (yellow), and 100 μM IPP (light blue) had no significant effect, the addition of both MgCl₂ and IPP (red) largely changed the shape of the Kratky plot into that suggesting the formation of much larger complexes. On the other hand, as shown in Fig. 3B, the Kratky plot of the E13R/R235E mutant in the same concentration (dark blue) fitted well with the theoretical curve of a tetramer (black), even after the addition of NADH (magenta), MgCl₂ (yellow), or IPP (light blue). However, it was curious that the addition of both MgCl₂ and IPP (red) changed the curve into one similar to the theoretical curve of an octamer (green). Because Mg²⁺ is involved in the binding of a diphosphate moiety of the substrate to S. shibatae type 2 IDI (33), the drastic changes in the quaternary structures observed with both the wild-type and mutant enzymes were possibly induced by the tight binding of the substrate, which would cause structural changes with each subunit.

Fig 3.

Kratky plots from SAXS analyses of the wild type (A and C) and E13R/R235E mutant (B and D). The solution of 2.8-mg/ml enzyme in buffer I (A and B) or II (C and D) was used for the measurement. Data were obtained in the absence of the substrate and cofactors (dark blue) and in the presence of only 0.5 mM NADH (magenta), only 2.5 mM MgCl₂ (yellow), only 100 μM IPP (light blue), or both 2.5 mM MgCl₂ and 100 μM IPP (red). The theoretical curves calculated from the crystal structures of octamer and tetramer are shown in green and black, respectively.

Because the change in the shape of the Kratky plot observed by the addition of both MgCl₂ and IPP to the wild-type enzyme was considered to arise from an aggregation of the enzyme, we performed similar experiments after exchanging the buffer (buffer I) with one that contained a detergent (buffer II) to avoid aggregation (Fig. 3C). In the absence of the substrate and cofactors, the Kratky plot of the wild-type enzyme drew a curve (dark blue) that fell between those theoretically expected from the octameric structure (green) and from the previously reported tetrameric crystal structure (black). The addition of only IPP (light blue) or both MgCl₂ and IPP (red) changed the shape into one more similar to the theoretical curve of a tetramer than to that of an octamer, while NADH (magenta) and MgCl₂ (yellow) had no effect. As for the E13R/R235E mutant, the Kratky plots in all the conditions fitted well with the theoretical curve of a tetramer, even after the addition of both MgCl₂ and IPP (Fig. 3D). From this result, the substrate and metal ion-induced structural change of the mutant enzyme observed in the absence of detergent is considered to be based mainly on a hydrophobic interaction that is easily disturbed by detergent. In contrast, the octameric structure of the wild-type enzyme was relatively stable in the presence of detergent, which could be explained by the electrostatic interactions observed in the tetramer-tetramer interface of the octameric crystal structure. Thus, the octamer-like plot observed for the E13R/R235E mutant in a detergent-free solution (buffer I) in the presence of both MgCl₂ and IPP was probably derived from an unknown structure that is different from a wild-type octamer.

Substrate concentration that affects the quaternary structure.

Next, using tryptophan fluorescence analysis, we examined the structural change of the wild-type and mutant enzymes caused by the addition of only IPP, which was enough to change the shape of the Kratky plot of a wild-type enzyme from octamer-like to tetramer-like. The solution of the enzymes at 0.002 mg/ml (49 nM monomer) or 2.5 mg/ml (61 μM monomer) in buffer II was titrated with IPP, and change in the intensity of tryptophan fluorescence was plotted (Fig. 4). Fitting of the theoretical curve, which depicted a simple equilibrium between two states with different fluorescence intensities, to the plot for a 0.002-mg/ml solution of the wild-type enzyme and the E13R/R235E mutant gave comparable K_d values for IPP: 3.04 and 4.51 μM, respectively (Fig. 4A and B). However, the tryptophan fluorescence from a 2.5-mg/ml solution of the wild-type enzyme violently fluctuated and did not allow fitting (Fig. 4C), while the plot for 2.5-mg/ml E13R/R235E was similar to those of 0.002-mg/ml enzymes and gave a K_d of 2.65 μM (Fig. 4D). The fluctuation of tryptophan fluorescence observed only with the wild-type enzyme at a high concentration was probably derived, at least in part, from the octamer-tetramer dissociation caused by the addition of IPP at the tested range of concentrations. On the other hand, the wild-type enzyme at a low concentration and the E13R/R235E mutant at both high and low concentrations seemed to be in a similar structural state, i.e., tetramers, and the changes in their fluorescence intensities supposedly arose from intrasubunit structural changes caused by the substrate binding.

Fig 4.

Tryptophan fluorescence analyses of the wild type (WT) and E13R/R235E mutant. The 0.002-mg/ml solutions of the wild type (A) and the mutant (B) and the 2.5-mg/ml solutions of the wild type (C) and the mutant (D) in buffer II were titrated with IPP. Through the titration, change in the intensity of tryptophan fluorescence (excitation at 295 nm, emission at 350 nm) was recorded. Curve fitting to the theoretical equation was performed as described in Materials and Methods.

Therefore, we tried to determine the IPP concentration that brought the change in quaternary structure of wild-type S. shibatae type 2 IDI by using DLS. The DLS analysis of a 2.8-mg/ml (68 μM monomer) solution of the wild-type enzyme in buffer II yielded a large peak of particle size distribution, as shown in Fig. 5A. The solution was titrated with IPP, and change in the radius size of the particle (r), which was represented by the peak position, was plotted (Fig. 5B). It should be noted that, other than the peak discussed above, peaks with larger r values (>100 nm) emerged at higher IPP concentrations, suggesting that part of the enzyme was aggregated (data not shown). However, the aggregation was limited to a small portion (less than 15%) when the IPP concentration was below 17.4 μM, and thus, it was unlikely to significantly affect the octamer-tetramer equilibrium. The r value decreased from ∼7 to ∼6 nm with an increase in IPP concentration from 0 to 17.4 μM, suggesting that the dissociation of an octamer to tetramers proceeded and that their average distribution was represented as a single peak. The molecular mass corresponding to the r value in the absence of IPP was ∼330 kDa, which is close to that of an octamer, although the enzymes were considered to be in the mixture of the octameric and tetrameric states, as shown by the SAXS analyses. However, the molecular mass of the enzyme with an r value of 6 nm was ∼220 kDa, which is somewhat larger than that of a tetramer. This inconsistency might have arisen from insufficient dissociation or from the nonspherical structure of the tetramer. Fitting of the theoretical curve of a simple one-to-one equilibrium to the plot gave a K_app value of 8.40 ± 4.68 μM for IPP, which represents an IPP concentration that causes a half-maximal change in the r value. This was larger than the K_d value for IPP obtained from tryptophan fluorescence analysis and smaller than the K_m for IPP, although the K_m value was obtained under different buffer conditions. This fact suggested that the change in the r value, probably caused by dissociation of (a part of) the enzymes from the octameric to the tetrameric state, was induced by substrate binding at the active site.

Fig 5.

DLS analysis of the wild-type enzyme. (A) Typical particle size distribution, i.e., radius size (r) versus intensity, in a 2.8-mg/ml enzyme solution in the absence of IPP. (B) The radius sizes corresponding to the peak of the octamer/tetramer were plotted through titration with IPP. Curve fitting to the theoretical equation was performed as described in Materials and Methods.

DISCUSSION

As shown by gel filtration chromatography, octamer-tetramer transformation of S. shibatae type 2 IDI occurs at around 0.6 mg/ml. It should be noted that the concentration was overrated without considering dilution in a column. Judging from the elution profiles in Fig. 2, the enzyme solution with an initial volume of 1 ml seemed to be diluted several times through gel filtration chromatography. Because total cellular protein concentration in prokaryotic cells has been determined to be 200 to 300 mg/ml (8, 20), the concentration of each protein is estimated at ∼0.1 mg/ml if all proteins encoded in a prokaryotic genome are equally expressed. Thus, S. shibatae type 2 IDI is likely to form, at least partially, an octamer when it is expressed at a level above average in the cells of the archaeon.

Interestingly, the results from X-ray crystallography, SAXS analysis, tryptophan fluorescence analysis, and DLS analysis suggest the incompatibility between the substrate binding and octamer formation of S. shibatae type 2 IDI. The binding of the substrate molecules to all (or most) of the 8 active sites in an octamer causes the structural change of each subunit, e.g., winding of the N-terminal α-helix, and then leads to dissociation into tetramers, which can catalyze isomerization of the substrate. The change in the quaternary structure of the enzyme is reminiscent of that of a well-studied allosteric enzyme, phosphofructokinase-2 (PFK-2) from E. coli (3–5). The E. coli enzyme takes inactive homotetrameric configuration in the presence of a homotrophic inhibitor, MgATP, which binds not only to the active sites but also to the effector sites, but turns into active homodimers when fructose-6-phosphate binds to the active sites. Similarly with PFK-2, the enzyme activity of S. shibatae type 2 IDI can be inhibited by octamer formation. The substrate-complex tetrameric structures of the archaeal enzyme (22, 33) suggest that unwinding of the N-terminal α-helix observed in the octameric structure precludes substrate binding because the helix is involved in active-site formation. If so, octamer dissociation proceeds when the production of IPP via the mevalonate pathway is enhanced and upregulates isoprenoid biosynthesis initiated from the isomerization of IPP. We tried to evaluate the inhibitory effects of octamer formation on activity but failed because of the difficulty of a quantitative assay at high enzyme and low substrate concentrations (data not shown). However, if an allosteric regulation mechanism like that of E. coli PFK-2 also exists in S. shibatae type 2 IDI, it might be possible to inhibit the enzyme with reagents that can stabilize the octameric form, as MgATP does in PFK-2. Or, small molecules that can mimic the subunit interactions used for octamer formation, such as a peptide with the N-terminal sequence of S. shibatae type 2 IDI, might also have an inhibitory effect.

Bacterial type 2 IDIs have also been reported to take homooctameric configurations (27), but dissociation of an octamer by the addition of the substrate and cofactors has been shown only with the enzyme from B. subtilis (18). It is noteworthy that the geometric placement of the subunits is somewhat different between the octameric structures of type 2 IDIs from B. subtilis (27) and S. shibatae. Namely, the tetramer-tetramer interfaces used for octamer formation in the enzymes are not identical. In the substrate-free crystal structure of the B. subtilis enzyme, α-helices corresponding with α3 and α16′ in the S. shibatae enzyme reportedly contributed to octamer formation, while the region corresponding to α1 in the S. shibatae enzyme was unseen, probably because the region was disordered in the absence of the substrate. Moreover, the quaternary structures of the enzymes from pathogenic bacteria such as S. aureus and Enterococcus in the presence of the substrate remain unclear.