Active site regions may therefore ... a single amino acid residue by using site-directed mutagene- sis as part of ..... This would make the region of the mobile loop.
Eur. J. Biochem. 224, 249-255 (1994) 0 FEBS 1994
A single amino acid mutation enhances the thermal stability of Escherichia coli malate dehydrogenase Christopher R. GOWARD', Julie MILLER', David J. NICHOLLS', Laurence I. IRONS', Michael D. SCAWEN', Ronan O'BRIEN' and Babur Z. CHOWDHRY'
'
Division of Biotechnology, Centre for Applied Microbiology and Research, Porton Down, England
* University of Greenwich, School of Biological and Chemical Sciences, England (Received April 5/June 20, 1994)
-
EJB 94 0480/3
The stability of wild-type Escherichia coli malate dehydrogenase was compared with a mutant form of the enzyme with the amino acid residue at position 102 changed from arginine to glutamine. The mutation occurs on the underside of a mobile loop which closes over the active-site cleft on formation of the enzyme/cofactor/substrate ternary complex. The mutant enzyme is kinetically compromised while the wild-type enzyme is highly specific for oxaloacetate. The mutant enzyme was shown to be more resistant to irreversible thermal denaturation by thermal inactivation experiments and high-sensitivity differential scanning calorimetry than the wild-type enzyme. In contrast, resistance of both enzymes to reversible unfolding in guanidinium chloride was similar. Circular dichroic spectropolarimetry shows the secondary structures of the enzymes are similar but there is a demonstrable difference in tertiary structure. From the position of the mutation, it is conjectured that the substitution on a mobile surface loop results in partial closure of the loop and greater resistance to thermal inactivation of the mutant enzyme. However, molecular modelling combined with circular dichroic spectropolarimetry indicate that the mutation may have a more widespread effect on the structure than simply partial closure of the mobile surface loop as the environment of distant tyrosine residues is altered. Resistance of the wild-type enzyme to thermal inactivation can be increased by cofactor addition, which may have the effect of partial closure of the mobile surface loop, but has little effect on the mutant enzyme.
Protein engineering can be used to explore and modify the molecular basis of protein stability (Nosoh and Sekiguchi, 1990). Deliberate engineering to produce enhanced stability of enzymes has obvious commercial significance. The aim, in most cases, is to increase stability without loss of catalytic efficiency. Few general principles have become apparent for quantitative prediction of amino acid substitutions to enhance stability from extensive studies of interactions that determine the thermodynamic stability of proteins (Alber et al., 1987) although some guidelines have emerged (Pace, 1990; Geisow, 1991). It may prove very difficult to Correspondence to C. R. Goward, Division of Biotechnology, Centre for Applied Microbiology and Research, Porton Down, Salisbury, England SP4 OJG Fax: +44 980 610898. Abbreviations. E, fully folded enzyme; E*, unfolded enzyme; E+, irreversibly denaturated enzyme; HSDSC, high-sensitivity differential scanning microcalorimetry ; GdmC1, guanidinium chloride ; [Q102R]lactate dehydrogenase, mutant form of Bacillus stearothermohilus L-lactate dehydrogenase with the amino acid residue at position 102 changed from glutamine to arginine; [R102Q]malate dehydrogenase, mutant form of Escherichia coli L-malate dehydrogenase with the amino acid at position 102 changed from arginine to glutamine; tl,*, half life of an enzyme at one temperature; t,, maxima of heat-capacity functions ; AG[H,O], free energy change for unfolding of an enzyme in the absence of denaturant at 20°C; AH,,,, calorimetric enthalpy. Enzymes. L-Malate dehydrogenase (EC 1.1.1.37); L-lactate dehydrogenase (EC 1.1.1.27).
achieve enhanced stability if large changes in conformation are required in the catalytic mechanism (Geisow, 1991). It is known that the folded conformation of a protein may be stabilised by substrates, cofactors or inhibitors (Klibanov, 1983 ; Pace, 1990). An active site may be situated in a more flexible region than the molecule as a whole. Active site regions may therefore be more sensitive to denaturants than is the whole molecule (Tsou, 1986) and are often considered the weakest part of an enzyme structure. It is therefore difficult to alter active site configuration to enhance stability and, simultaneously, to preserve catalytic proficiency. This is a significant problem which has barely been addressed. L-Lactate dehydrogenase and L-malate dehydrogenase convert 2-hydroxy acids to the corresponding 2-keto acids with NAD'NADH as cofactor. They have extensive structural similarities and so form the 2-hydroxy acid dehydrogenase family (Birktoft et al., 1982). Escherichia coli malate dehydrogenase is a dimer of identical subunits and catalyses the reversible reduction of oxaloacetate to L-malate. There are high-resolution crystal structures of E. coli malate dehydrogenase (Hall et al., 1992; Hall and Banaszak, 1993), and of malate dehydrogenases from pig cytoplasm (Birktoft et al., 1989), pig mitochrondria (Roderick and Banaszak, 1986) and Thermus9avus (Kelly et al., 1993). E. coli malate dehydrogenase is most similar to the mitochondria1 malate dehydrogenase with 59% sequence identity and a similar tertiary structure (Kelly et al., 1993). Despite a relatively low amino acid sequence identity, there is considerable structural iden-
250 tity between the subunit tertiary structures of malate dehydrogenase (Hall et al., 1992) and X-ray structures for lactate dehydrogenase from dogfish (Abad-Zapatero et al., 1987) and Bacillus stearothermophilus (Wigley et al., 1992). E. coli malate dehydrogenase is a suitable enzyme for stability studies as there is much known about its structure and it is easy to work with. E. coli malate dehydrogenase was specifically altered at a single amino acid residue by using site-directed mutagenesis as part of an investigation to study factors controlling molecular recognition of keto acid substrates (Nicholls et al., 1992). A conserved arginine residue at amino acid position 102 (lactate dehydrogenase numbering: Eventoff et al., 1977) on a mobile surface loop over the active-site cleft was replaced with a gIutamine residue to produce a mutant form of the enzyme ([R102Q]malate dehydrogenase). The mutation resulted in loss of the high degree of specificity for the natural substrate oxaloacetate. The difference in relative binding energy for oxaloacetate between the wild-type and mutant enzyme was approximately 29 kJ . mol-' and this was explained by the large hydration potential o f arginine and formation of a salt bridge with a carboxylate group o f oxaloacetate (Nicholls et al., 1992). Initial studies indicated that the mutant enzyme was more stable to heating than was the wildtype enzyme. The aim of this study was to characterise and attempt to explain this apparent increase in stability.
EXPERIMENTAL PROCEDURES Bacterial strains and vectors E. coZi TG2[K12, (laq-pro),supE, thi, hsdDS/F', traD36, proA'B', ladq, lacZ M15 recA-] was the host strain for recombinant plasmids and phage. The bacteria were grown in 2XYT broth [2% (masshol.) tryptone, 2% (masshol.) yeast extract, 1% (masshol.) NaCl] at 37°C. Ampicillin (100 yg . ml-') was used where necessary to select for clones containing recombinant plasmids. Wild-type E. coli malate dehydrogenase was expressed from the recombinant plasmid pDN41. The [R102Q]malate dehydrogenase was expressed using the same system. Site-directed mutagenesis and subcloning strategy Standard recombinant DNA techniques were used as described by Sambrook et al. (1989). Nucleotide sequences were determined by using the Sequenase version 2.0 kit with deoxy-7-deazaguanosine according to the manufacturer's instructions (Cambridge Bioscience, Cambridge, England). Oligonucleotide site-directed mutations were introduced into the mdh gene by using the single primer-extension method (Winter et al., 1982). The mutagenic oligonucleotide primer (ACCCGGTTTCTGCGCTACGCC) was synthesised on an Applied Biosystems model 380A DNA synthesiser. The mutated gene was ligated as a 2.2-kb SphI-EcoRI fragment into pMTL23 (Chambers et al., 1988). Overexpression of the recombinant mutant enzyme was directed by the the E. coli gene promoter contained in the insert DNA. The entire nucleotide sequence of the coding region was determined to check that only the desired mutation was incorporated following mutagenesis. Purification of wild-type and [R102Q]malate dehydrogenase Wild-type and [R102Q]malate dehydrogenase were purified using triazine dye affinity chromatography on Procion
Red HE-3B Sepharose 4B (Nicholls et al., 1989). The purity of the enzyme preparations was assessed by SDSPAGE
Enzyme and protein assays Enzyme specific activities were determined at 30°C in 50mM Tris/HCl, pH7.5 containing 0.14mM NADH and either 0.3 mM oxaloacetate or 0.3 mM pyruvate as the substrate. The initial reaction rates were measured by following the change in A,,,. Steady-state kinetic constants were determined with 0.14 mM NADH, 0.3 mM oxaloacetate or 0.3 mM pyruvate as the fixed concentrations of cofactor and substrate. Variable substrate and cofactor concentrations were in the range 0.005-50 mM. Kinetic parameters were calculated by using a non-linear regression data analysis program (Enzfitter, Elsevier-Biosoft). Protein concentrations were determined by using the Biuret method (Gornall et al., 1949) with bovine serum albumin as the standard.
Thermal inactivation Irreversible enzyme inactivation was measured with a protein concentration of 1.5 yM subunits in 20 mM potassium phosphate, pH 7.2. Samples were incubated at 60'C in sealed glass tubes for various time periods and'were removed from the source of heat and placed on ice for 20 min. Residual enzyme activity was determined using oxaloacetate as the substrate for wild-type enzyme and pyruvate as the substrate for the [R102Q]malate dehydrogenase. The enzyme samples were incubated in the presence and absence of 1 mM N.4DH. A sample kept on ice for 20 min was used as the zero-time control.
High-sensitivity differential scanning microcalorimetry Thermal denaturation of wild-type and [R102Q]malate dehydrogenase was studied by high-sensitivity differential scanning microcalorimetry (HSDSC) at pH 7.2-3.6. Calorimetric measurement was carried out by using a Microcal MC-2 instrument (Microcal Ltd., Amherst, MA) and the DA2 dedicated software package (provided by Microcal Ltd.) for data accumulation and analysis. The reference cell was filled with appropriate buffer and all experiments were performed under approximately 100 kPa nitrogen to prevent solvent loss by evaporation. Samples were equilibrated in the high-sensitivity differential scanning calorimeter for a minimum of 45 min prior to each scan. The scan rates used were 10 K . h-' and 60 K . h-' for the wild-type and 60 K . h-' for [R102Q]malate dehydrogenase. Data analysis was based on a subunit molecular mass of 33000 Da for E. coli malate dehydrogenase. The values oft, are defined as Celsius temperature maxima of the heat-capacity functions.
Circular dichroic spectroscopy CD spectra were obtained at room temperature (1820 "C) with a Jasco J-600 spectropolarimeter calibrated with 0.6% (masshol.) ammonium (+)-camphor-l O-sulphonate. The spectra were recorded in the near ultraviolet (320250 nm) and far ultraviolet (250-190 nm) in cylindrical cells of pathlength 0.5 cm and 0.02 cm, respectively. Each spectrum was the mean of four accumulated scans obtained with a time constant of 2 s and a scan speed of 10 nm . minUnspecified solvent dichroic absorbances were subtracted from the spectra by computer manipulation. The final spectra
25 1
5r
Table 1. Steady-state kinetic data. Values of the K, and k,,, for wild-type malate dehydrogenase with pyruvate were not measureable (n.m.). The kJK, value for wild-type enzyme with pyruvate was calculated from the gradient of the saturation curve which was assumed to be linear at low substrate concentrations. Data was taken from a previous study (Nicholls et al., 1992).
Enzyme
Binding of
Kinetic parameter
pM Wild-type malate dehydrogenase [R102Q]malate dehydrogenase
oxaloacetate NADH pyruvate oxaloacetate NADH pyruvate
0.04 0.05
n.m. 3 0.03 25
s-1
M-1
. s-1
930 23.25 X lo6 750 15.00X lo6 n.m. 0.14 0.8 270 0.13 4300 3.3 130
5 c 1
\!
0
were converted to mean residue ellipticities by using a mean residue mass of 103.8. Denaturation of the enzyme was followed in guanidinium chloride (GdmC1). BDH Aristar GdmCl was dissolved in 20 mM potassium phosphate, pH 7.2. Enzyme solutions contained 6 yM subunit. The concentration of GdmCl solutions was determined by measurement of the refractive index.
I
I
I
I
5
10
15
20
Time (rnin) Fig. 1. Time course for thermal inactivation. Enzyme (1.5 pM subunits) was heated at 60°C in 20 mM potassium phosphate, pH 7.2. Irreversible thermal inactivation was measured for wild-type malate dehydrogenase (0, 3.0 min), wild-type malate dehydrogenase with 1 mM NADH (V,tllz 4.9 rnin), [R102Q]malate dehydrogenase (m, t1,28.0 rnin), [R102Q]malate dehydrogenase with I mM NADH (A,tliz8.4 min).
RESULTS Enzyme purification
I
I
I
55
60
65
Wild-type and [R102Q]malate dehydrogenase were purified to specific activities of 1900 U . mg-' and 5 U . mg-I, respectively. SDSPAGE revealed a single protein band of 34 000 Da for both enzymes. h
Steady-state kinetics
The apparent K, for NADH is unaffected by the mutation whereas the value of K, for oxaloacetate is increased 75-fold and k,,, is considerably reduced for both oxaloacetate and NADH (Table 1). Pyruvate is a poor substrate for the wildtype enzyme while the [R102Q]malate dehydrogenase shows a substantial improvement with this substrate as k,,,/K, is increased approximately 950-fold (Table 1).
50
3 r/)
Y0 $
0 50
Temperature ( OC )
Fig. 2. Thermal denaturation monitored by HSDSC. Scan-rate normalised HSDSC (Microcal MC-2, Microcal Ltd, Amherst, MA) and The half life (t,J at 60°C for irreversible thermal inacti- recordings for thermal denaturation of wild-type (-) vation of [R102Q]malate dehydrogenase (tlf28.0 min) is [R102Q]malate dehydrogenase (. . . .) in potassium phosphate, pH 7.2, with a scan rate of 60 K . h-' and 44 pM subunits (wildgreater than the half-life for the wild-type enzyme (tll2 type malate dehydrogenase) and 58 pM subunits ([R102Q]malate 3.0 min) (Fig. 1). The presence of 1 mM NADH in the incu- dehydrogenase). The values of t, are 59.0"C for the wild-type enbation buffer has little effect on stability of the [R102Q]ma- zyme and 60.6"C for the [RlOZQlmalate dehydrogenase. Cp, heat late dehydrogenase ( t l R 8.4 min, a 4% increase). However, capacity at constant pressure. the presence of NADH increased stability of the wild-type enzyme (tlR 4.9 min, a 60% increase) (Fig. 1). Similar experiments carried out in the presence of higher concentra- nase and wild-type enzyme, respectively, with corresponding tions, i.e. up to 10 mM NADH, produced curves identical to calorimetric enthalpies (AH,,,) of 440 kJ . mol-' and 430 kJ those obtained for 1 mM NADH. . mol-' (Fig. 2). The increase in t, caused by the mutation was small and was shown to be independent of pH at pH 73.6 (Fig. 3). There was an exothermic component at temperHigh-sensitivity differential scanning microcalorimetry atures above the main transition, probably associated with The HSDSC-transitions are characterised by a single formation of a precipitate which was observed when the sampeak with a t, at 60.6"C and 59.0"C for malate dehydroge- ple was removed from the cell. There was no discernible
Thermal inactivation
A
-
I
I
I
I
260
210
280
290
60
h
2
3
40
4
N
G2
20
9 c s o -20 250
300
Wavelength (nm)
3
I
I
I
I
I
4
5
6
7
8
B
I
I
I
I
I
I
1
PH Fig. 3. Change in apparent denaturation temperature with pH. Values oft,,, of the wild-type ( 0 )and [R102Q]malate dehydrogenase (V)were recorded with a scan rate of 6 0 K . h-' and a protein concentration of 45 pM subunits at various pH values.
transition on reheating the sample and thus the transitions were calorimetrically irreversible. The t, values were independent of protein concentration, at least up to 109 pM subunits (data not shown), and aggregation was not considered the primary reason for the irreversibility. Denaturation of the wild-type enzyme was, however, shown to be dependent on the scan rate; there was a decrease in t,,, to 56.9"C when the protein was heated at 10 K . h-I. Kinetic control of the denaturation process was further demonstrated by thermoinactivation studies carried out at 60"C, a temperature close to the t,,, of both proteins, and at the same pH. This demonstrated that the t,,Z values of inactivation were 4.7 min and 10.0 min for wild-type and [R102Q]malate dehydrogenase, respectively, at 60°C as measured by HSDSC. This close correlation with thermal inactivation data of 3.0min and 8.0min, respectively, for tllzat 60"C, as shown in the previous section, suggests that the HSDSC transition represents the same process which results in thermal inactivation. The time spent in the calorimetric transition region (at 60 K . h-I) was approximately 8.0 min. These features preclude the use of equilibrium thermodynamic models for interpretation of the calorimetric data. Calorimetry is however a useful technique for measuring the thermal stability of proteins, especially when the differences are quite small as in the present case. Both proteins were heated in the presence of 50 mM NADH at pH 7.2 and at 60 K . h-'. This resulted in exothermic transitions and t, values close to those observed in the absence of NADH for the mutant enzyme (t", 62.7"C). For the wild-type enzyme, t,,, was increased to a value similar to that of the mutant (tm 60.3"C). Although not amenable to detailed analysis, this increase demonstrates that the process under these conditions is likely to be kinetically limited and may infer that the rate constants associated with irreversible inactivation have been altered in the mutant enzyme.
Circular dichroic spectroscopy The secondary structures appear to be similar, as shown by far-ultraviolet spectra but there is a clear difference in
I
I
I
I
1
190
200
210
220
230
240
250
Wavelength (nm)
Fig.4. CD spectra for the wild-type and [RlOZQlmalate dehydrogenase in the near-ultraviolet and far-ultraviolet regions. CD and [RIO2Q]mdlate dehydrogenase spectra of wild-type (-) (. . . .) were determined in 20 mM potassium phosphate, pH 7.1, with a Jasco 5-600 spectropolarimeter. The protein solutions contained 130 pM subunits for (A) near-ultraviolet spectra and 13 pM subunits for (B) far-ultraviolet spectr. Cylindrical cells of pathlength 0.5 cm and 0.02 cm were used for near-ultraviolet and far-ultraviolet recordings, respectively. The time constant was 2 s, the mean scan speed was 10 nm . min-l and each spectrum was the mean of four scans.
tertiary structure, as shown by the near-ultraviolet spectra (Fig. 4). E. coli malate dehydrogenase contains nine phenylalanine and four tyrosine residues but no tryptophan residues. The near-ultraviolet spectrum of the enzyme at pH 7.2 is positive and showed Cotton effects due to a contribution from phenylalanine at 270-250 nm and tyrosine at 290-270 nm. Contribution to the spectrum from phenylalanine becomes more visible with the mutant due to a probable change in contribution from tyrosine. The change in structure of the protein with increasing concentration of GdmCl was monitored by measuring the change in mean residue ellipticity at 222 nm (Fig. 5). These results showed that there was a progressive loss of structure. A concentration of 0.84M GdmCl was required to unfold the wild-type enzyme by 50% and 0.80M GdmCl was required for similar unfolding of the mutant enzyme. Complete loss of structure occurred in approximately 1.5 M GdmCl for both enzymes. When refolding was initiated by replacing 2.0 GdmCl with 20 mM potassium phosphate, pH 7.2, by rapid buffer-exchange chromatography on Sephadex G-25, the CD
253 mobile loop, which amounts to 1.4 nm for Thermus malate dehydrogenase (Kelly et al., 1993). As a result, the environment in the active-site cleft becomes more hydrophobic and the guanidinium group of the arginine residue at amino acid position 102, located on the mobile loop, becomes buried. X-ray structural evidence shows that in the malate dehydrogenasekofactor binary complex the loop is partially closed (Kelly et al., 1993). The mutation R102-+Qhas an affect on the kinetics of the reaction; in the case of [R102Q]malate dehydrogenase, k,,, is reduced approximately 1200-fold for oxaloacetate (Table 1; Nicholls et al., 1992). If the loop were forced to assume a position of partial closure, as found in the enzymekofactor binary complex (Kelly et al., 1993), the effect would be to limit access to the active-site cleft and hence the turnover of substrate and cofactor. The denaturation of an enzyme may be considered simply by the following scheme :
-2 s m
L.i
m
x
0 0.0
0.5
1.o
4
1.5
2.0
GdrnCl (rnol l-’) Fig. 5. Unfolding of wild-type and [RlOZQlmalate dehydrogenase in GdmC1. Malate dehydrogenase (6 pM subunits) was incubated for 24 h at 20°C in 20 mM potassium phosphate, pH 7.2, containing various concentrations of GdmC1. CD spectra were recorded in the far ultraviolet with cylindrical cells of pathlength 0.02 cm, a time constant of 2 s and a scan speed of 10 nm . min-’. The mean residue ellipticity at 222 nm was calculated from a mean of four scans. Values of dG[H,O] were calculated for wild-type (0,dG[H,O] 14.3 ld . mol-’) and [R102Q]malate dehydrogenase dC[H20] 13.1 k J . mol-I).
spectra for both enzymes showed the refolding to be complete (data not shown). The difference in conformational stability between the wild-type and [R102Q]malate dehydrogenases was estimated from the free energy change for unfolding of the enzymes in the absence of denaturant at 20°C (AG[H,O]). The values of AG at each concentration of GdmCl in the transition region (Fig. 5 ) were transformed according to the method described by Pace (1990), using the following relationship : AG = -RT In [(Y”-y,d(Yots-Yd)I, where R is the gas constant, T is the absolute temperature, yn is the mean residue ellipticity for the native conformation, yobsis the observed mean residue ellipticity at a particular GdmCl concentration and yd is the mean residue ellipticity for the denatured conformation. The value of AG[H,O] was estimated, assuming that the linear dependence of AG on GdmCl concentration continues to zero concentration, by least-squares analysis to fit data to the following equation: AG
=
AG[H,O]-m[GdmCl],
where m is a measure of the dependence of AG on denaturant concentration. These AG[H,O] values amount to 14.3 kJ . mol-’ for wild-type and 13.1 kJ . mol-‘ for the [R102Q]malate dehydrogenase.
DISCUSSION The enzyme reaction for malate dehydrogenase follows a compulsory order in which binding of cofactor is followed by binding of substrate to form an active ternary complex (Fig. 6). This is accompanied by movement of the 13-residue
E kz
k3
E*
Er,
where k , -k3 are rate constants, E is fully-folded enzyme, E* is unfolded enzyme and E’ is irreversibly denatured enzyme in the final state arrived at from E*. E has a lower free energy than E*. Anything that stabilises E andor decreases destabilizing forces of E* will increase the free-energy change for unfolding and therefore stabilize the protein (Nosoh and Sekiguchi, 1990). Irreversible inactivation dependends on the conformation of E* (Nosoh and Sekiguchi, 1990). Thermal stability is rate controlled and not thermodynamically controlled. For the irreversible inactivation, k, becomes greater than k2 so that most of the molecules are converted to Er rather than returning to E from E*. The concentration of E* becomes relatively low and the process is thus effectively irreversible and may be considered E+E+ and kinetically controlled by k, as was shown for thermolysin (Sanchez-Ruiz et al., 1988). The mutation appears to confer thermal stability on the enzyme and the difference in tl,, and t, values between wildtype and [R102Q]malate dehydrogenase suggests that the mutation has structural consequences. The gain in stability is perhaps surprising, as mutagenesis of other structurally wellcharacterised proteins has shown that residues in mobile surface loops are frequently of little importance to thermal stability of folded proteins (Alber et al., 1987; Reidhaar-Olson and Sauer, 1988). The role of surface loops in protein stability is unclear, as mobile surface loops have been shown to be weak points for thermal denaturation in Arthrobacter xylose isomerase (Siddiqui et al., 1993). It has recently been shown that mutations to surface loops in Bacillus subtilis neutral protease lead to an increase in thermal stability (Hardy et al., 1994). The presence of cofactor was shown to increase stability of the wild-type enzyme but has no effect upon the mutant. A conformational rearrangement occurs in the wild-type Thermus malate dehydrogenase upon addition of cofactor which involves partial closure of the mobile loop (Kelly et al., 1993). This would make the region of the mobile loop less susceptible to thermal effects by collisional energy transfer from the bulk solvent. Consequently the loop would be less susceptible to thermal vibration and a decrease in the k, rate. The mutant enzyme is intrinsically more thermostable such that addition of cofactor has no further effect. This suggests that the loop conformation in the mutant apoenzyme may be similar to that in the wild-type enzyme/cofactor binary complex.
254 Conformational changes in the mobile loop region are thought to be part of the catalytic mechanism in malate dehydrogenase. By analogy with lactate dehydrogenase, the loop moves towards the active site in the presence of bound cofactor, and makes specific interactions with the substrate (Wigley et al., 1992; Clarke et al., 1986; Grau et al., 1981). The arginine residue at amino acid position 102 on the mobile loop closes down onto an a-helix (a2G) (Wilks et al., 1990) and bulk water is excluded from the active-site cleft. The protein conformational change was shown to be the rate-limiting step for B. stearothermophilus lactate dehydrogenase with its natural substrate (Waldman et al., 1988). The glutamine residue at position 102 is located on the underside of the mobile loop and ends up in close proximity to the substrate following the conformational change (Dunn et al., 1991). In a reverse situation to that described for E. coli malate dehydrogenase, B. stearothermophilus lactate dehydrogenase was specifically altered to produce a mutant form of the enzyme with the amino acid at position 102 changed from glutamine to arginine ([Q102R]lactate dehydrogenase) which showed preferential malate dehydrogenase activity rather than lactate dehydrogenase activity (Wilks et al., 1988). However, the [Q102R]lactate dehydrogenase showed no significant change in stability; the t , was 74.7"C compared with 74.5"C for the wild-type enzyme, at a scan rate of 60 K . h-' (Kallwass et al., 1992). The effect of this reverse but complementary mutation is thus rather different in terms of irreversible thermal denaturation compared with [R102Q]malate dehydrogenase. Although [RI 02Qlmalate dehydrogenase has increased resistance to thermal inactivation compared with the wildtype enzyme, which involves only the rates k, and kz, the mutant enzyme has similar resistance to denaturation in GdmC1. The denaturation curves are almost superimposable (Fig. 5). This implies that factors which influence irreversible thermal inactivation are not necessarily the same as those which influence reversible equilibrium denaturation by GdmC1. The position of the mobile loop may inhibit the rate step for thermal inactivation in the mutant enzyme, however the global effect on structure, indicated by the near-ultraviolet spectra (Fig. 4A), may allow easier access of the denaturant to sensitive parts of the molecule. The secondary structure of the enzyme is not altered by the mutation, as shown by the far-ultraviolet CD spectra (Fig. 4A). However, this single mutation did result in a marked difference in tertiary structure as shown by the nearultraviolet spectra (Fig. 4B). The chiral nature of environments near aromatic residues of proteins often produce Cotton effects in near-ultraviolet CD spectra which are sensitive indicators of structural change. The observed change in the CD spectrum is probably due to a change in contribution from one or more tyrosine residues. The four tyrosine residues of E. coli malate dehydrogenase are all 2-3 nm from the arginine residue at amino acid position 102 (Fig. 6). This implies that any conformational arrangement due to substitution of glutamine for arginine has an effect distant from the position of the mutation, and thus the conformational change may not be considered as simply localised to the mobile loop region. A concerted effect between loop movement and more widespread changes in structure is therefore suggested. This contrasts with data obtained from molecular modelling which suggest that there is little movement on formation of the ternary complex other than loop movement (Hall and Banaszak, 1993). The situation is therefore not as simple as a partial
Fig.6. E. coli malate dehydrogenase. A molecular diagram of E. coli malate dehydrogenase was produced on an Evans and Sutherland PS390 graphics system interfaced with a Digital Equipment Corporation Microvax I1 running the molecular modelling package SYBL version 5.3 (Tripos Associates, St Louis, MO). The threedimensional coordinates were obtained from the Brookhaven Protein Database (Hall et al., 1992). The mobile surface loop is shown as a ribbon. The side chain of arginine at amino acid position 102 projects below the surface on the underside of the loop (lactate dehydrogenase numbering: Eventoff et al., 1977). The position of L-malate and NAD' are shown in the active-site cleft. The position cQ side chains of the tyrosine residues Y33, Y138, Y253 and Y260 (6coli malate dehydrogenase numbering: Hall et al., 1992) are shosn. The distances of the a-carbon of the tyrosine residue from the a-carbon of arginine at amino acid position 102 is 2.18 (Y33), 2.97 (Y138), 2.14 (Y253) and 2.87 nm (Y260). None of these tyrosine residues are on the subunit interface.
closure of the loop as occurs with the Thermus malate dehydrogenasekofactor binary complex (Kelly et al., 1993 j. Increased resistance to thermal inactivation does not necessarily imply an intrinsically more stable protein and thus a more desirable industrial catalyst. The mutation of E. coli malate dehydrogenase has resulted in greater stability to thermal activation but a similar resistance to denaturation by GdmC1. The enzyme is also a far less efficient catalyst. This study demonstrates the need to examine stability from several approaches when assessing a potentially more stable industrial catalyst and emphasises that, even if the protein has a greater resistance to thermal inactivation, the catalytic properties may be less desirable. We thank Prof Tony Atkinson for helpful comments in the preparation of the manuscript and Dr Melanie Duffield for help with molecular modelling. This work was supported in part by a grant from the SERC DTI LINK scheme, England.
REFERENCES Abad-Zappatero, C., Griffith, J. P., Sussman, J. L. & Rossmann, M. G. (1987) Refined crystal structure of dogfish M, apo-lactate dehydrogenase, J. Mol. Biol. 198, 445-467. Alber, T., Dao-pin, S., Nye, J. A., Muchmore, D. C. & Matthews, B. W. (1987) Temperature-sensitive mutations of bacteriophage T4 lysozyme occur at sites with low mobility and low solvent accessibility in the folded protein, Biochemistry 26, 3753- 3758. Birktoft, J. J., Fernly, R. T., Bradshaw, R. A. & Banaszak, L. J. (1982) Amino acid sequence homology among the 2-hydroxyacid dehydrogenases. Mitochondria1 and cytoplasmic malate dehyrogenases form a homologous system with lactate dehydrogenase, Proc. Natl Acad. Sci. USA 79, 6166-6170.
255 Birktoft, J. J., Rhodes, G. & Banaszak, L. J. (1989) Refines crystal structure of cytoplasmic malate dehydrogeuase at 2.5-A resolution, Biochemistry 28, 6065 -6089. Chambers, S. P., Prior, S. E., Barstow, D. A. & Minton, N. P. (1988) The PMTL nic-. cloning vectors. I. Improved pUC polylinker regions to facilitate the use of sonicated DNA for nucleotide sequencing, Gene (Amst.) 68, 139-149. Clarke, A. R., Wigley, D. B., Chia, W. N., Barstow, D., Atkinson, T. & Holbrook, J. J. (1986) Site-directed mutagenesis reveals role of mobile arginine residue in lactate dehydrogenase catalysis, Nature 324, 699-702. Dunn, C. R., Wilks, H. M., Halsall, D. J., Atkinson, T., Clarke, A. R., Muirhead, H. & Holbrook, J. J. (1991) Design and synthesis of new enzymes based on the lactate dehydrogenase framework, Philos. Trans. R. SOC.Lond. B. Biol. Sci. 332, 177-184. Eventoff, W., Rossman, M. G., Taylor, S. S., Tort, H.-J., Meyer, H., Keil, W. & Kiltz, H. H. (1977) Structural adaptations of lactate dehydrogenase isozymes, Proc. Natl Acad. Sci. USA 74, 26772681. Geisow, M. J. (1991) Stabilizing protein products: coming in from the cold, Trends Biotechnol. 9, 149-150. Gornall, A. G., Baradawill, C. J. & David, M. M. (1949) Determination of serum proteins by means of Biuret reagent, J. Biol. Chem. 177, 751 -766. Grau, U. M., Trammer, W. E. & Rossman, M. G. (1981) Structure of the active ternary complex of pig heart lactate dehydrogenase with S-lac-NAD at 2.7 A resolution, J. Mol. Biol. 151, 289-307. Hall, M. D., Levitt, D. G. & Banaszak, L. J. (1992) Crystal structure of Escherichia coli malate dehydrogenase. A complex of the apoenzyme and citrate at 1.87 A resolution, J. Mol. Biol. 226, 867-882. Hall, M. D. & Banaszak, L. J. (1993) Crystal structure of a ternary complex of Fscherickia coli malate dehydrogenase citrate and NAD at 1.9 A resolution, J. Mol. Biol. 232, 213-222. Hardy, F., Vriend, G., Van der Vinne, B., Frigerio, F., Grandi, G., Venema, G. & Eijsink, G. H. (1994) The effect of engineering surface loops on the thermal stability of Bacillus subtilis neutral protease, Protein Eng. 7 , 425 -430. Kelly, C. A., Nishiyama, M., Ohnishi, Y., Beppu, T. & Birktoft, J. J. (1993) Determinants of protein thermostability observed in the 1.9-A crystal structure of malate dehydrogenase from the thermophilic bacterium Thermusflavus, Biochemistry 32, 391 3 3922. Kallwass, H. K. W., Surewicz, W. K., Parris, W., Macfarlane, E. L. A., Luyten, M. A,, Kay, C. M., Gold, M. & Jones, J. B. (1992) Single amino acid substitutions further increase the stability of a thermophilic L-lactate dehydrogenase, Protein Eng. 5, 769 -774. Klibanov, A. M. (1983) Stabilization of enzymes against thermal inactivation, Adv. Appl. Microbiol. 29, 1-28. Nicholls, D. J., Minton, N. P., Atkinson, T. & Sundaram, T. K. (1989) Cloning sequencing and over-expression of Escherichia coli malate dehydrogenase, Appl. Microbiol. Biotech. 31, 376382.
Nicholls, D. J., Miller, J., Scawen, M. D., Clarke, A. R., Holbrook, J. J., Atkinson, T. & Goward, C. R. (1992) The importance of arginine 102 for the substrate specificity of Escherichia coli malate dehydrogenase, Biochem. Biophys. Res. Commun. 189, 1057 - 1062. Nosoh, Y. & Sekiguchi, T. (1990) Protein engineering for thermostability, Trends Biotechnol. 8, 16- 19. Pace, C. N. (1990) Measuring and increasing protein stability, Trends Biotechnol. 8, 93-98. Reidhaar-Olson, J. F. & Sauer, R. T. (1988) Combinatorial cassette mutagenesis as a probe of the informational content of protein sequences, Science 241, 53-57. Roderick, S. L. & Banaszak, L. J. (1986) The three-dimensional structure of porcine heart mitochondria1 malate dehydrogenase, J. Biol. Chem. 261, 9461-9464. Sambrook, T., Fritsch, E. F. & Maniatis, T. (1989) Molecular cloning: a laboratory manual, 2nd edn, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Sgnchez-Ruiz, J. M., L6pez-Lacomba, J. L., Cortijo, M. & Mateo, P. L. (1988) Differential scanning calorimetry of the irreversible thermal denaturation of thermolysin, Biochemistry 27, 16481652. Siddiqui, K. S., Rangarajan, M., Hartley, B. S., Kitmitto, A,, Panico, M., Blench, I. P. & Moms, H. R. (1993) Arthrobacter D-xylose isomerase. Partial proteolysis with thermolysin, Biochem. J. 289, 201 -208. Tsou, C.-L. (1986) Location of the active sites of some enzymes in limited and flexible molecular regions, Trends Biochem. Sci. 11, 427-429. Waldman, A. D. B., Hart, K. W., Clarke, A. R., Wigley, D. B., Barstow, D. A., Atkinson, T., Chia, W. N. & Holbrook, J. J. (1988) The use of a genetically engineered tryptophan to identify the movement of a domain of B. stearothemophilus lactate dehydrogenase with the process which limits the steady-state turnover of the enzyme, Biochem. Biophys. Res. Commun. 150, 752-759. Wigley, D. B., Gamblin, S. J., Turkenburg, J. P., Dodson, E. J., Piontek, K., Muirhead, H. & Holbrook, J. J. (1992) Structure of an allosteric lacjate dehydrogenase from Bacillus stearothermophilus at 2.5 A resolution, J. Mol. Biol. 223, 317-335. Wilks, H. M., Hart, K. W., Feeney, R., Dunn, C. R., Muirhead, H., Chia, W. N., Barstow, D. A., Atkinson, T., Clarke, A. R. & Holbrook, J. J. (1988) A specific, highly active malate dehydrogenase by redesign of a lactate dehydrogenase framework, Science 242, 1541-1544. Wilks, H. M., Halsall, D. J., Atkinson, T., Chia, W. N., Clarke, A. R. & Holbrook, J. J. (1990) Designs for a broad substrate specificity keto acid dehydrogenase, Biochemistry 29, 8587-8591. Winter, G., Fersht, A. R., Wilkinson, A. J., Zoller, M. & Smith, M. (1982) Redesigning enzyme structure by site-directed mutagenesis. Tyrosyl tRNA and ATP binding, Nature 299, 756-758.
