Crystal structures of 3-isopropylmalate dehydrogenases with ...

Download PDF
2 downloads 0 Views 182KB Size Report
Comment
Crystal structures of 3-isopropylmalate dehydrogenases with mutations at the .... and the reservoir solution was 0.1 M sodium acetate buffer an R factor of 27.4%.
Protein Engineering vol.13 no.4 pp.253–258, 2000

Crystal structures of 3-isopropylmalate dehydrogenases with mutations at the C-terminus: crystallographic analyses of structure–stability relationships

Zeily Nurachman1, Satoshi Akanuma2, Takao Sato1, Tairo Oshima3 and Nobuo Tanaka1,4 1Department

of Life Science, Faculty of Bioscience and Biotechnology, Tokyo Institute of Technology, Nagatsuta 4259, Yokohama 226-8501, 2Institute of Physical and Chemical Research (RIKEN), Wako 351-0198 and 3Department of Molecular Biology, Tokyo University of Pharmacy and Life Science, Horinouchi, Hachioji 192-0392, Japan 4To

whom correspondence should be addressed. E-mail: [email protected]

Thermal stability of the Thermus thermophilus isopropylmalate dehydrogenase enzyme was substantially lost upon the deletion of three residues from the C-terminus. However, the stability was partly recovered by the addition of two, four and seven amino acid residues (called HD177, HD708 and HD711, respectively) to the C-terminal region of the truncated enzyme. Three structures of these mutant enzymes were determined by an X-ray diffraction method. All protein crystals belong to space group P21 and their structures were solved by a standard molecular replacement method where the original dimer structure of the A172L mutant was used as a search model. Thermal stability of these mutant enzymes is discussed based on the 3D structure with special attention to the width of the active-site groove and the minor groove, distortion of β-sheet pillar structure and size of cavity in the domain–domain interface around the C-terminus. Our previous studies revealed that the thermal stability of isopropylmalate dehydrogenase increases when the active-site cleft is closed (the closed form). In the present study it is shown that the active-site cleft can be regulated by open–close movement of the minor groove located at the opposite side to the activesite groove on the same subunit, through a papercliplike motion. Keywords: closed conformation/3-isopropylmalate dehydrogenase/minor groove/modified C-terminus/paperclip motion

Introduction Thermal tolerance of proteins in a thermophilic organism depends on the folding process of their polypeptide chains to a particular conformation. There are many subtle factors that influence protein stability, e.g. tighter hydrophobic packing (Yutani et al., 1987), increased hydrophobicity (Fersht and Serrano, 1993), increased hydrogen bonds and salt bridges (Shortle, 1992) and disulfide bonds (Matsumura et al., 1989). 3-Isopropylmalate dehydrogenase (IPMDH, EC 1.1.1.85), encoded by a leuB gene of an extreme thermophile, Thermus thermophilus, is a thermostable enzyme catalyzing the oxidative decarboxylation of (2R,3S)-3-isopropylmalate (IPM) to 2oxoisocaproate in leucine biosynthesis. The catalytic properties and thermal stability of the enzyme have been studied (Yamada et al., 1990) and its three-dimensional structure has also been determined at high resolution (Imada et al., 1991). The subunit © Oxford University Press

structure of IPMDH is composed of two domains, domain 1 and domain 2. The first domain includes both termini and consists of residues 1–99 and 252–345, and the remaining amino acid residues from 100 to 251 form the second domain. Numerous efforts to improve the thermal stability of IPMDH from T.thermophilus particularly emphasized the role of Ala residue at position 172 in the conformational stability (Kotsuka et al., 1996, Akanuma et al., 1997). The thermostability of IPMDH A172L mutant (A172L hereafter), in which Ala was replaced with Leu at position 172, is 3°C higher than that of the wild-type enzyme. The thermostability enhancement of A172L is caused by the fact that the larger side chain of Leu in the minor groove gives extra interactions at the interface between domains in the active-site cleft (Qu et al., 1997). In a sense, the enzyme is more stabilized when the active-site cleft is closed (closed form). In a further study of A172L, the last five amino acid residues at the C-terminus (Leu–Arg–His–Leu–Ala) were replaced with a dipeptide Gly–Ile. It results in a temperature-sensitive IPMDH mutant, called CD071. The melting temperature of CD071 dropped sharply from that of A172L by 14°C (Akanuma et al., 1996). Furthermore, insertion of tandemly duplicated sequences (Table I) to the C-terminus of CD071 produces novel mutants that partly restore the thermal stability. The catalytic activities of these mutants are not significantly different from that of the wild-type IPMDH or A172L. The IPMDH mutants (except CD071) have been purified and crystallized isomorphously with A172L. The crystals belong to the monoclinic space group P21. In this paper we describe the 3D structures of these mutant enzymes and try to explain the restoration of the thermal stability of IMPDH mutants from their structural viewpoints. Materials and methods Expression and purification of the IPMDH mutant The mutants of T.thermophilus IPMDH were constructed and overexpressed in Escherichia coli as described previously (Akanuma et al., 1996). Purification of the IPMDHs was performed according to Kotsuka et al. (1996) with 10 min heat treatment at 72, 70 and 70°C for HD711, HD708 and HD177, respectively. The purified enzymes were analyzed with SDS–PAGE.

Table I. Amino acid sequences of IPMDH A172L and its mutants in the Cterminal region

A172L CD071 HD177 HD708 HD711

Amino acid sequence

Thermostability

∆Tm (°C)

341 341 341 341 341

Thermostable Thermosensitive Partly restored Partly restored Partly restored

0 –14 –11 –8 –6

LRHLA 345 GI 342 GMGI 344 TATVGI 346 EAFTATVGI 349

253

Z.Nurachman et al.

Table II. Summary of data collection statistics HD711 Wavelength (Å) X-ray source

1.00 BL6A (Photon Factory) Detector Weissenberg camera Temperature (K) 293 Resolution limit (Å) 36.1–2.2 No. of unique 19524 (616) reflections Completeness 85.5% to 2.8 Å 57.6% to 2.2 Å (14.5%) a 0.071 (0.340) Rmerge

Table III. Summary of refinement statistics

HD708

HD177

1.54 Rigaku RU300

1.00 BL18B (Photon Factory) Weissenberg camera 100 53.5–2.7 14939 (685)

R-AXIS IV 100 54.2–1.8 39276 (332) 83.0% to 2.1 Å 64.5% to 1.8 Å (18.1%) 0.047 (0.200)

86.7% to 2.7 Å (79.0%) 0.051 (0.203)

⫽ Σ|II – ⬍II⬎|/ΣII. Values in parentheses are for the outer shell of 2.30–2.2 0, 1.86–1.78 and 2.75–2.70 Å for HD711, HD708 and HD177, respectively. aR merge

Crystallization and data collection The enzyme was dissolved in 20 mM potassium phosphate buffer (pH 6.5) containing 0.5 mM EDTA. Crystals of IPMDH HD711 were grown at 25°C by the hanging drop vapor diffusion method. The drops contained 0.05 M sodium acetate buffer (pH 4.8), 10% (v/v) PEG400 and 5–7.5 mg/ml protein and the reservoir solution was 0.1 M sodium acetate buffer (pH 4.8) containing 20% (v/v) PEG400. Rod-like crystals with the largest size of 0.3⫻0.5⫻0.3 mm were observed after overnight incubation. Crystals of IPMDH HD708 and HD177 were grown under the conditions similar to those for HD711 except for a twofold higher concentration of PEG400. Data sets for HD711 and HD177 were collected at room temperature and 100 K, respectively, with use of the Weissenberg camera for macromolecular crystallography installed at the BL6A and BL18B of Photon Factory (Tsukuba, Japan). The intensity data were indexed, integrated, merged and scaled using the DENZO packages (Otwinowski and Minor, 1997). The crystal HD711 belongs to the monoclinic space group P21, with cell dimensions a ⫽ 55.4 Å, b ⫽ 87.6 Å, c ⫽ 70.9 Å and β ⫽ 100.5°. HD177 crystallizes also in the monoclinic space group P21 with unit cell dimensions a ⫽ 69.9 Å, b ⫽ 86.0 Å, c ⫽ 54.4 Å and β ⫽ 100.3°. Assuming a dimer per asymmetric unit, the Vm values of HD711 and HD177 are 2.30 and 2.20 Å3/Da, respectively (Matthews, 1968). Detailed data collection statistics are listed in Table II. The intensity data of HD708 were collected from crystals that had been flash-cooled to 100 K with a Rigaku R-AXIS IV. The reservoir solution was used as a cryoprotectant prior to flash-cooling. The X-ray generator producing Cu Kα radiation was operated at a power of 50 kV, 80 mA. The data were indexed by the data processing package PROCESS (Rigaku). The crystal belongs to the monoclinic space group P21, with cell dimensions of a ⫽ 55.6 Å, b ⫽ 84.3 Å, c ⫽ 72.3 Å and β ⫽ 103.0°. There are two monomers per asymmetric unit (calculated Matthews’ coefficient 2.24 Å3/Da). Structure determination and refinement HD711. HD711 was solved by the molecular replacement method using the program X-PLOR (Bru¨ nger, 1992). The dimer A172L structure (PDB, accession code 1OSJ) was utilized as a search model. A rotation search using data between 10 and 4 Å resulted in the unique solution that the molecule corresponds to the search model, which is rotated 254

Resolution range (Å) No. of reflections used R/Rfree factor (%) (Fo ⬎ 2σ)a No. of protein atoms No. of water molecules R.m.s.d. from ideality: Bond lengths (Å) Bond angles (°) Dihedral angles (°) Improper torsions (°) Ramachandran plotb: Most favored regions (%) Additional allowed regions (%) Generously allowed regions (%) Disallowed regions (%)

HD711

HD708

HD177

6.8–2.8 13270 17.2/26.4 5170 69

8.0–2.2 27758 24.9/30.8 5188 143

6.5–2.7 13010 20.1/30.2 5160 73

0.007 1.4 29.3 0.69

0.010 1.5 28.9 0.86

0.007 1.3 29.9 0.71

89.0 10.8 0.2 0.0

88.7 11.2 0.2 0.0

86.0 13.6 0.4 0.0

aR factor ⫽ Σ||F | – k|F obs calc||/Σ|Fobs|. bThe plot was generated with the program

PROCHECK (Laskowski et al.,

1993).

by 10.2° about the axis of the direction cosine of (–0.3, –0.9, 0.2). A translation search of the oriented molecule followed by rigid body refinements also yielded a unique solution with an R factor of 27.4%. After the molecule was located in the unit cell, manual-rebuilding structure was performed with the program TURBO-FRODO (Jones, 1985). Again, rigid body and positional refinements using data between 6.88 and 2.8 Å were carried out with the program X-PLOR 98.1. Solvent positions were automatically assigned by WATERPEAK standard protocol with the hydrogen bond distance set between 2.5 and 4 Å and they were also checked with 2Fo – Fc map using the program TURBO-FRODO. Further molecular dynamics stimulated-annealing refinements were achieved with 10% of data left out randomly for Rfree factor calculation. In the last refinements, non-crystallographic symmetry (NCS) restraint was applied to the domain backbones to obtain a good stereochemical quality and a bulk solvent correction was also made. Residues 346–349 in subunit A and residues 345–349 in subunit B were not included in the structural model because of their uninterpretable electron densities. The final model has an R factor of 17.7% and an Rfree factor of 26.7%. Detailed refinements statistics are given in Table III. HD708 and HD177. In procedures similar to those for HD711, HD708 and HD177 were solved by molecular replacement using the same search model. The molecules were oriented in the crystals through rotation of search model by 9.8° about the axis of (–0.4, –0.7, 0.6) for HD708 and 172.5° about the axis of (–0.6, 0.1, –0.8) for HD177. A translation search followed by rigid body refinements produced a model which has R factors of 34.4 and 37.3% for HD708 and HD177, respectively. The molecular structures in the C-terminal region were modified using the TURBO-FRODO program. Further refinements were performed using data between 8 and 2.2 Å (HD708) and 6.5 and 2.7 Å (HD177). Water molecules were modeled automatically with hydrogen bond distances between 2.5 and 3.5 Å. The final R factor and Rfree factor (in parentheses) for HD708 and HD711 are 24.9% (30.8%) and 20.1% (30.2%), respectively (Table III).

IPMDH with mutations at the C-terminus

Table V. Root mean square differences (in Å) of pillar structure atoms of mutant IPMDHs

Subunit A Subunit B Overall molecule

Fig. 1. Schematic diagram of IPMDH in (a) monomer and (b) dimer forms. Circles represent α-helices and arrows or boxes denote β-sheets. The armlike region is not displayed in the dimer form. Solid and dotted lines symbolize positions at the upper and bottom side of the paper face, respectively.

Table IV. Orientation between subunits or domains (in degrees) forming dimer for A172L and its mutants

Subunit Domain 1 Domain 2

A172La

HD711

HD708

HD177

177.6 (0.8) 176.4 (0.4) 179.7 (0.3)

178.4 (0.7) (172.2 (0.1) 179.9 (0.1)

175.7 (0.6) 173.2 (0.1) 178.9 (0.1)

178.9 (0.9) 177.3 (0.1) 179.6 (0.1)

aData

were calculated from 1OSJ coordinates (Protein Data Bank). Values in parentheses are the r.m.s. deviations.

Results Overall structure IPMDH is a homodimer (subunits A and B hereafter) and each subunit consists of two domains; domain 1 contains five β-strands (A–E) and seven helices (a–d and i–k) and domain 2 contains seven β-strands (F–L) and four helices (e–h) (Figure 1a) (Imada et al., 1991). Ten β-strands from A to J are important pillars for the IPMDH structure while the surrounding α-helices and loops cover the pillar from the solvent. The active-site groove is composed of helix h, strand F, strand E and helix d (Figure 1b). The minor groove between helices e and j has no specific function. Table IV summarizes the rotation angles of HD711, HD708 and HD177. The proper arrangement of the subunits may be essential for rigidity of IPMDH conformation. The dimer fitting angles of HD711 and

HD711

HD708

HD177

1.0 1.0 1.0

2.4 2.4 1.7

2.1 2.0 1.8

HD177 are close to the twofold symmetry angle but that of HD708 deviates slightly from the symmetry angles. Changes in pillar structure To detect the conformational changes in the mutants, 64 Cαatoms in β-strands forming the pillar structure (Imada et al., 1991) (residues 2–7, 35–40, 66–69, 100–110, 126–133, 179– 184, 210–215, 232–236, 258–263 and 266–271) of each mutant were superimposed on those of A172L and the root mean square (r.m.s.) differences between the two coordinate sets were calculated for all mutants. The pillars of the mutants were fitted to the A172L pillar by rotation of 0.3° about the axis of (0.4, –0.7, 0.6) for HD711, 2.9° about the axis of (1.0, –0.5, –0.4) for HD708 and 83.8° about the axis of (–0.2, –1.0, 0.2) for HD177, then by shifting oriented pillars in the direction of the orthogonal vectors of (–28.1, 4.1, 0.5), (–26.5, 5.0, 1.4) and (–21.1, 82.2, 79.4) in units of ångstroms for HD711, HD708 and HD177, respectively. The r.m.s. differences of overall molecule and subunits are listed in Table V. In addition, the r.m.s. deviations of the pillar for HD711, HD708 and HD177 are 0.61, 0.70 and 0.60 Å, respectively. Compared with the pillar of A172L (the r.m.s. deviation is 0.54 Å), those of mutants may shift owing to the mutation around the C-terminus. HD708 is the enzyme deviating most in structure from A172L. Dimer interface interactions The dimer interface involves twofold symmetry interactions of four helices, g, h (in subunit A), g⬘ and h⬘ (in subunit B), two sticking out arm-like peptide chains hooking both subunits and two loops (Figure 2). The dimer interface interactions were evaluated by measuring the cavity volumes with radius probe of 1.4 Å. The largest cavity in the dimer interface was found in the space between the sticking out arm-like peptide chains and the mouth of a four-helix-bundle structure. The cavity volumes for A172L, HD711, HD708 and HD177 are 49.6, 54.0, 42.4 and 53.7 Å3, respectively. Moreover, a potent hydrophobic well buried at the center of the four-helix bundle structure is also measured. The solvent accessible volumes of the hydrophobic wells of HD711 and HD177 are both 0.7 Å3 (twice as large as that of A172L). In contrast, the volume in the corresponding location of HD708 is 0.2 Å3 (half that of A172L), suggesting that the mutation around the C-terminus of HD708 may induce the strained dimer interface. Active-site groove The active-site groove is sandwiched between helix d and h and is in a closed conformation for all mutants used in this study. The width of the substrate binding cleft is defined by the distance between Cα atoms of Glu87 and Asp241 and that of the coenzyme binding cleft can be measured by estimating the distance between Cα atoms of Gly255 and Ile279. As shown in Table VI, the average widths of the substrate binding 255

Z.Nurachman et al.

Fig. 2. Cα atom diagram of IPMDH dimer.

Table VI. Width of substrate and coenzyme binding clefts in (Å) for A172L and its mutants A172L

HD711

Subunit A 16.5a (13.9)b 16.3 (13.5) Subunit B 16.7 (13.3) 16.8 (13.5) Average molecule 16.6 (13.6) 16.6 (13.5) aValues bValues

HD708

HD177

16.4 (13.2) 15.3 (13.4) 15.9 (13.3)

16.2 (13.2) 16.7 (13.2) 16.5 (13.2)

are for substrate binding. are for coenzyme binding.

clefts of HD711 and HD177 are similar to that of A172L. These results are consistent with the result of functional analyses that the kinetic constants of these IPMDHs are similar to each other (Akanuma et al., 1996). In contrast, the average width of the substrate-binding cleft of HD708 is closer than the others. The average widths of the coenzyme binding clefts of these IPMDHs remain unchanged (in a range 13.2–13.6 Å). Conformational changes around the C-terminal region The minor groove in IPMDH is located at the opposite side from the active-site groove. The function of this groove is not known. It may play a role in ‘open and close’ movement of the active-site enzyme. The active site is opened upon binding of the substrate (Kadono et al., 1995). In the wild-type enzyme or A172L, the structure around the C-terminus was stabilized by a hydrophobic pocket surrounded by two residues on helix a (Val22, Leu26), one on helix j (Leu307) and three on helix k (Val340, Leu341, Leu344) and ionic interactions among His343, Arg344 (helix k), Lys317 and Glu321 (helix j) (Figure 3a). However, the corresponding hydrophobic interactions in HD711 seem to be weakened due to mutation of Leu341→Glu and the ionic interactions in this region may also be decreased owing to mutations of Arg342→Ala and His343→Phe (Figure 3b). Moreover, the extended four hydrophobic residues at the C-terminus may also weaken the structure. This part is difficult to observe in an electron density map, presumably, because of the highly disordered structure. Although the conformational change takes place in the C-terminal region of HD711, the mouth of the minor groove is similar in size to that of A172L (Table VII). In the case of HD708 and HD177, which possesses a shorter C-terminus sequence than that of HD711, the hydrophobic 256

Fig. 3. Residual arrangement of IPMDHs (a) A172L, (b) HD711, (c) HD708 and (d) HD177 in the C-terminal region. Dashed lines indicate the width of the minor groove containing a positively charged loop (1) and a hydrophobic loop (2). This figure was drawn using MOLSCRIPT (Kraulis, 1991).

Table VII. Width of the mouth of minor groove in (Å) for A172L and its mutants

Subunit A Subunit B Average molecule aValues

A172L

HD711

HD708

HD177

10.5a

10.5 9.5 10.0

8.2 9.0 8.6

9.9 8.8 9.4

9.5 10.0

are based on the distance between Cα atoms of Lys175 and Val305.

pocket may be strengthened and as a result the minor groove is closed (Table VII). In HD708, mutations of Arg342→Ala and His343→Thr break ionic interactions in the C-terminal region (Figure 3c). Moreover, the mutations of Leu341→Thr, Leu344→Val and the addition of Ile346 may also increase the hydrophobic forces. The average of the accessible surface area, which is calculated with a radius probe of 1.4 Å and a Z spacing factor of 0.05, for the whole residue at positions of 341 and 344 is 45.32 and 20.18 Å2, respectively. The areas in the corresponding location of A172L are larger, 48.65 and 25.96 Å2, respectively. The hydrophobic forces from the C-terminal sequences of HD708 seem the strongest among those of the mutants. In HD177, the ionic interactions in the C-terminal region are lost owing to mutations of Arg342→Met and His343→Gly (Figure 3d). In addition, mutation of Leu344→Ile may increase the hydrophobic force that pushes the minor groove closer (Table VII). To detect conformational changes due to the mutations in the C-terminal region, the cavity volumes of the region surrounded by strand B, helices a, i and j in the first domain, are also measured. In the case of A172L, the total accessible

IPMDH with mutations at the C-terminus

volume of the cavity is 1.6 Å3. The volumes of HD708 and HD177 are smaller than the volume of A172L, indicating that the strengthened hydrophobic pocket shifts not only the minor groove but also the pillar structure. In contrast, the volume in the corresponding location of HD711 is larger than that of A172L. Discussion Although a number of successful examples of protein stabilization have been reported, the mechanism of the protein stabilization has a far from comprehensive understanding. The thermostable IPMDH from T.thermophilus, for instance, is homologous to its counterpart from E.coli except for higher thermal-denaturation temperature and lower activity at room temperature (Za´ vodszky et al., 1998). At least three major intramolecular interactions can be pointed out as an essential interaction for the folding of T.thermophilus IPMDH. First, the IPMDH pillar structure is built with a series of β-strands from A to J (Figure 1b). The active-site groove and the minor groove are sandwiched between the first and second domains with a hinge pivot consisting of strands E and F. Second, the subunit interface interactions are composed of four-helix bundle structures with a hydrophobic well in the center, hydrogen bonds between two symmetric arm-like stretches from each subunit and interactions between two symmetric loops from each subunit (residues P117–L118, Figure 2). Finally, helices and loops prevent the pillar from making contact with the solvent. In the present study, mutant IPMDHs with different C-terminal sequences and different denaturation temperatures were analyzed using X-ray crystallographic techniques in connection with the change in thermal stability. Although the experimental temperatures were either room temperature or 100 K, the IPMDH structure did not change as much as reported by Nagata et al. (1996). In fact, the catalytic activity and the major internal interactions of each domain of HD711, HD708 and HD177 are similar to those of A172L, hence the change in thermostability can be attributed to a result of ‘mechanical movement’ alterations of the active-site groove and/or the minor groove. IPMDH is a paperclip-like enzyme and the mouth of the active-site groove is closed when the minor groove is expanded. The inner space of the minor groove is mainly built by hydrophobic interactions except the space at its mouth segment. This segment is composed of two loops: a loop with hydrophobic residues (residues 301AFGLV305) in the first domain and a positively charged loop (residues 174RLRRL178) in the second domain. The antagonistic movements of the mouth of the minor groove may regulate the active-site groove, which is located at the opposite side of the enzyme molecule. Although it is not easy to define a simple criterion for the open–close movement of the active site because of the small angle between the substrate binding cleft and the hinge pivot, it is possible to define the movement of the minor groove by measuring the distance between two Cα atoms at the mouth segment. With respect to the C-terminal region of the wild-type IPMDH or A172L, the width of the minor groove may depend on the inflexibility of helices a, j and k (Figure 3a). The top side of helix j, which is close to the hydrophobic loop of the minor groove, is stabilized by hydrophobic interactions among the side chains of Leu307 (helix j), Leu26 and Leu32 (helix a). The middle part of helix j is also supported by hydrophobic

interactions among the side chains of Ala314 (helix j), Val22, L26 (helix a) and L344 (helix k), among the side chains of Ala318 (helix j), Val340 and Phe336 (helix k). The bottom side of helix j, which is exposed to the solvent, is stabilized by ionic interactions (Rhode and Martin, 1999) among the side chains of Lys317, Glu321 (helix j), Arg342 and His343 (helix k). In addition, hydrophobic interactions between the side chains of Leu341 (helix k) and Ala25 (helix a) make helices a, j and k more compact structures. HD711 The cleavage of ionic interactions at the bottom of helix j due to mutation of Arg342→Ala and His343→Phe brings helix k close to helix j. On the other hand, the side chain of Glu341 located at the hydrophobic pocket tends to push helix k away (Figure 3b). Although destabilization of the structure occurs particularly in the C-terminal region of helix k, the width of the minor groove remains constant. As shown in Table VII, the width of the minor groove of HD711 is similar to that of A172L. In accordance with this movement, the width of the active-site cleft is also similar to that of the untruncated enzyme (Table VI). The results may explain the partial restoration of the thermal stability of HD711 and also provide the reason why this enzyme is the most thermostable among the mutants. However, the slight constraints in the pillar structure (Table V) and the expansion of the cavity at the interface between the domains are induced by the mutations and destabilize the structure of HD711, giving rise to a lower denaturation temperature than that of A172L. HD708 Unlike HD711, the mutations in HD708 improve the strength of the hydrophobic pocket in the C-terminal region (Figure 3c). The lack of ionic interactions at the bottom of helix j and the exposed side chain of the hydrophobic residue (Ala342) to the solvent may support the strengthened hydrophobic pocket, by shifting helix k close to helix j. In addition, the side chains of Thr341 and Val344 located in the hydrophobic pocket drive helix k close to helix a. This makes the cavity in the first domain smaller and the minor groove closer (Table VII). The closing of the mouth of the minor groove may explain the thermal instability of HD708 compared with that of HD711 or A172L. At the same time, the increase in the hydrophobic forces in the C-terminal region pushes the second domain away from the first domain. This gives rise to distortion of the interface angle between the subunits (Table IV) and reduces the cavity in the dimer interface. The distorted dimer interface makes the active-site groove closer (Table VI), suggesting the reason why HD708 restores thermal stability higher than that of HD177. HD177 As the case in HD708, the mutations in HD177 also improve the hydrophobic interactions in the C-terminal region but the improvement is not as great as that in HD708. The loss of ionic interactions at the bottom of helix j makes helix k closer (Figure 3d). The side chain of Ile344 undergoes hydrophobic interactions with the side chains of Val22, Val36 (helix a) and Ala 314 (helix j). However, the residue of Gly341 seems not to contribute to the enhancement of the hydrophobic interactions because of its too small size. The mutations in HD177 only make the minor groove closer (Table VII) and compress the cavity in the first domain. The closing of the mouth of the minor groove and the expansion of the cavity in the dimer 257

Z.Nurachman et al.

interface may explain the thermal instability of HD177 and may also be the reason why HD177 has the lowest thermal stability among the mutants. Hence it is obvious that the molecular mechanisms for the restored thermostability of the three mutants are different from one another. Although different mechanisms restored the thermal stability of three IPMDH mutants in this study, the results suggest that maintenance of the pillar structure and the interactions at the dimer interface is important for keeping the enzyme structure rigid. In addition, it is predictable from the present results that the thermostability of IPMDH can be engineered by point mutations (e.g. replacing Val305 with a larger hydrophobic residue), by inserting ions or a specific ligand in the mouth of the minor groove to expand the width of the groove and eventually close the active-site groove located at the opposite side of the subunit through a papercliplike movement of the β-strand pillar structure. The mechanical movement mechanisms that are deduced from the present study are useful for the protein thermostability arrangement without any drastic change in the catalytic properties by way of the adjusted minor groove. The atomic coordinates and structure factors have been deposited at the Protein Data Bank with the accession codes 1DPZ, 1DR0 and 1DR8 for HD711, HD708 and HD177, respectively. Acknowledgements We thank Dr H.Moriyama and Dr N.Igarashi for invaluable assistance and discussions during the data collection at the Photon Factory. This work was supported in part by the ACT-JST Program, Japan Science and Technology Corporation and Grants-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan (Nos 11157209 and 11160203). Data collections with the Weissenberg camera were carried out at the Photon Factory, Institute of Materials Structure Science, High Energy Accelerator Research Organization, with the approval of the Photon Factory Advisory Committee, Japan (Proposal No. 98G-140).

References Akanuma,S., Yamagishi,A., Tanaka,N. and Oshima,T. (1996) J. Bacteriol., 178, 6300–6304. Akanuma,S., Qu,C., Yamagishi,A., Tanaka,N. and Oshima,T. (1997) FEBS Lett., 410, 141–144. Bru¨ nger,A.T. (1992) A System for X-ray Crystallography and NMR, X-PLOR, Version 3.1. Yale University Press, New Haven, CT. Fersht,A.R. and Serrano,L. (1993) Curr. Opin. Struct. Biol., 3, 75–83. Imada,K., Sato,M., Tanaka,N., Katsube,K., Matsuura,Y. and Oshima,T. (1991) J. Mol. Biol., 222, 735–738. Jones,T.A. (1985) Methods Enzymol., 115, 157–171. Kadono,S., Sakurai,M., Moriyama,H., Sato,M., Hayashi,Y., Oshima,T., Tanaka,N. (1995) J. Biochem., 118, 745–752. Kotsuka,K., Akanuma,S., Tomuro,M., Yamagishi,A. and Oshima,T. (1996) J. Bacteriol., 178, 723–727. Kraulis,P.J. (1991) J. Appl. Crystallogr., 24, 946–950. Laskowski,R.A., MacArthur,M.W., Moss,D.S. and Thornton,J.M. (1993) J. Appl. Crystallogr., 26, 283–291. Matsumura,M., Becktel,W.J., Levitt,M. and Matthews,B.W. (1989) Nature, 334, 406–410 Matthews,B.W. (1968) J. Mol. Biol., 33, 491–497. Nagata,C.,Moriyama,H., Tanaka,N., Nakasako,M., Yamamoto,M., Ueki,T. and Oshima,T. (1996) Acta Crystallogr., D52, 623–630. Otwinowski,Z. and Minor,W. (1997) Methods Enzymol., 276, 307–326. Qu,C., Akanuma,S., Moriyama,H., Tanaka,N. and Oshima,T. (1997) Protein Engng, 10, 45–52. Rhode,D.J. and Martin,B.L. (1999) Biochem. Biophys. Res. Commun., 258, 179–183. Shortle,D. (1992) Q. Rev. Biophys., 25, 205–250.

258

Yamada,T., Akutsu,N., Miyazaki,K., Kakinuma,K., Yoshida,M. and Oshima,T. (1990) J. Biochem., 108, 449–456. Yutani,K., Ogasahara,K., Tsujita,T. and Sugino,Y. (1987) Proc. Natl Acad. Sci. USA, 84, 4441–4444. ´ . and Petsko,G.A. (1998) Proc. Natl Acad. Za´ vodszky,P., Kardos,J., Svingor,A Sci. USA, 95, 7406–7411. Received January 11, 2000; revised February 9, 2000; accepted February 14, 2000