Purification, catalytic properties and thermostability of 3-isopropylmalate dehydrogenase from Escherichia coli

BIOC'HIMICA ET BIOPHYSICA AC']'A

ELSEVIER

Biochimica et BiophysicaActa 1337 (1997) 105-112

Purification, catalytic properties and thermostability of 3-isopropylmalate dehydrogenase from Escherichia coli Gerlind Wallon ~, Kazutaka Yamamoto b, Hiromi Kirino b, Akihiko Yamagishi b, Susan T. Lovett a, Gregory A. Petsko a, Tairo Oshima b,, Rosenstiel ~,tedical Sciences Research Center, Brandeis University, Waltham, MA 02254-9110, USA b Department of Molecular B,:ology, Tokyo Unicersi~" of Pharmacy and Life Science, 1432 Horinouchi, Hachioji, Tokyo, 192-03, Japan

Received 27 August 1996; accepted 4 September 1996

Abstract 3-isopropylmalate dehyOrogenase (IPMDH) from Escherichia coli was overexpressed, purified and crystallized. The enzyme was characterized and compared to its thermophilic counterpart from Thermus thermophilus strain HB8. As in the thermophile enzyme, the activity of E. coli IPMDH was dependent on the divalent cations, Mg 2÷ or Mn 2÷, with Mn 2+ being the preferred cation. Activity was also strongly influenced by KCI: 0.3 M were necessary for the optimal activity. At 40°C the K m of E. coli IPMDH was 105 p.M for IPM and 321 jzM for NAD, the kca t w a s 69 s -~. The half denaturation temperature was 64°C, which was 20°C lower than that of the thermophile enzyme. Keywords: 3-Isopropylmalatedehydrogenase; Carboxylating-dehydrogenase;Thermostability; (Escherichia coli)

1. Introduction The comparison of the properties of proteins from mesophilic microorganisms with their more stable counterparts from thermophiles provides a logical approach to the investigation of the structural basis of protein stability. It has been observed that the free energy change of stabilization, AGstab, from thermophilic proteins does not appear to exceed significantly the average values measured for mesophilic proteins. Jaenicke and Zavodszky [1] have argued that this additional free energy, AG.~tab can be provided by a

Abbreviations: IPMDH, 3-isopropylmalatedehydrogen&se(EC 1.1.1.85); KP,, potassium phosphate buffer consisting of 20 mM K2HPOa/KH2PO4 (pH 7.6), I mM EDTA. ° Corresponding author. Fax: +81 426 767145; E-mail: [email protected]

few extra hydrogen bonds, salt bridges or hydrophobic interactions. Comparison of the amino-acid sequences of a mesophilic/thermophilic protein pair of bacterial origin generally shows only about 4 0 - 7 0 % sequence identity, implying that most of the differences in sequence must be neutral or have been fixed during evolution for reasons other than to stabilize the protein. Petsko and Ringe [2] and Vihinen [3] have argued that thermostable proteins pay for their increased thermostability with a decrease in flexibility: their activities at mesophilic temperatures are generally much lower than those of their mesophilic counterparts at that temperature. This leads to the assumption that flexibility and thermostability are interrelated. In the best case a single mutation in the sequence could transform one protein into the other. We have approached this problem by studying the properties and structures of the products of the leuB

0167-4838/97/$17.00 Copyright © 1997 Elsevier Science B.V. All rights reserved. PH S0167-4838(96)00157-4

1(~

G. Walhm et al. / Biochimica et Bi,q~hysica Acta 1337 (1997) 105-112

gene, 3-isopropylmalate dehydrogenase (IPMDH, EC 1.1.1.85), from a thermophilic (Thermus thermophilus) and a mesophilic ( Escherichia coli) organism. The enzyme catalyzes the third step in the biosynthesis of the amino-acid leucine in microorganisms and plants. The reaction involves a dehydrogenation and subsequent decarboxylation of threo-O3-isopropylmalate to 2-oxoisocaproate with the concomitant reduction of NAD +. This dimeric enzyme belongs to a new family of dehydrogenases which comprises only two members: isocitrate dehydrogenase (ICDH) and isopropylmalate dehydrogenase (IPMDH). These enzymes do not display the so-called 'Rossman-fold', found in many nucleotide binding enzymes. The sequence of IPMDH has been determined from a variety of species including mesophilic (Sac-

charomyces cererisiae, Saccharomyces pombe, Salmonella ~phimurium, Bacillus subtilis, Bacillus coagulans and Escherichia coli), moderately thermophilic (Bacillus stearothermophilus) and extreme thermophilic microorganisms (Thermus thermophilus, Thermus aquaticus, Bacillus caldotenax). The T. thermophilus and the E. coli enzymes have an amino-acid sequence identity of 50%, in fact, the sequence identity between any pair is at least 50%. The 3-dimensional structure of the thermophilic enzyme from T. thermophilus has been solved [4] and can be used as a model to determine the structures of IPMDH from other organisms. The structure of IPMDH can be characterized as a 'parallel a / ~ doubly wound /3-sheet'. It is clearly divided into two domains, which are connected through a 10-stranded central fl-sheet with a cleft between the domains. One domain contains an armlike protrusion that makes contact with the same domain of the other subunit, forming a 4-stranded intersubunit fl-sheet. No direct experimental evidence for the location of the active site has been obtained so far, but a tentative determination has been made from its location in the homologous enzyme (ICDH) [5]. The substrate interacts with ten amino acids in the active site, only two of which are not conserved between the homologues, which presumably are involved in the binding of the isopropyl group. Nine of these residues are conserved among the IPMDHs from different organisms. Another important feature is the presence of a divalent metal ion,

Mn 2+, which coordinates one of the substrate carboxyl groups and the substrate hydroxyl group. In the present study we describe the purification and characterization of the E. coli enzyme, the gene of which has been cloned and sequenced by Kirino et al. [6], and a comparison with the thermophilic enzyme. This work is necessary preliminary to the comparison of the three dimensional structures of both enzymes and for further studies on the structural basis of thermophilicity.

2. Materials and methods

2.1. Materials DNA modification enzymes were products of New England Biolabs; threo-o,e-3-isopropylmalic acid was purchased from Wako; NAD was obtained from Sigma. Substrate analogs were kindly provided by Dr. Kakinuma (Department of Chemistry, Tokyo Institute of Technology, Tokyo, Japan). Butyl- and DEAE-Toyopearl 650S were obtained from Toyo Soda. DEAE-Sepharose was obtained from Pharmacia.

2.2. Bacterial strains and plasmids IPMDH was overexpressed in E. coli BL21 (F', ompT, hsdS, with a A prophage carrying the TTRNA polymerase gene under Plac control) harboring the plasmid pBluescript K S - (Stratagene), which carries the leuB gene under the control of the T7 RNA polymerase promoter (pWallyl). The wild-type T. thermophilus IPMDH was purified and expressed as described earlier [7]. For complementation assays, the growth rate of JA221 with a low c o p y n u m b e r v e c t o r p C L I 9 2 1 / 1 9 2 0 [8] harboring the E. coli or the T. thermophilus leuB gene (pEc and pTt) was compared on minimal medium without leucine.

2.3. Enzyme purification E. coli cells harboring the recombinant plasmid were grown in 4 1 of 2 X YT medium. Overexpression was induced by addition of IPTG to a final concentration of 1 mM. The cells were spun down

G. Wallon et a l . / Biochimica et Biophysica Acta 1337 (1997) 105-112

and washed in 75 ml of 20 mM potassium phosphate and 1 mM EDTA, pH 7.6 (KP i buffer) and stored at - 7 0 ° C until further use. The cells were thawed and sonicated for 10 min. on ice. The crude extract was subjected to ultracentrifugation at 90 000 X g for 20 min. The following chromatographic steps were then performed on a FPLC (Pharmacia) at room temperature: the supernatant was loaded on a self packed DEAE-Toyopearl (Toyo Soda) column. The proteins were eluted with a linear gradient of 0 to 2.0 M KCI in KP:buffer. The active fractions were then pooled and applied to a Butyl-Toyopearl 650S column, after addition of ( N H 4 ) 2 S O 4 to 25% ( w / v ) . The bound protein was eluted in a linear gradient of 25% to 0% (NH4)2SO 4 ( w / v ) in KPi-buffer. The active fractions were pooled and applied to a HcA34 Ultrogel-gelfiltration-column (LKB Bromma) for desalting and size fractionation. The active fractions were pooled and applied to a DEAE-Sepharose column (Pharmacia). The buffer system used for the first DEAE-column was modified by the addition of 10% glycerol. The final active fractions were pooled and stored frozen at -70°C.

2.4. Enzyme activity assay The standard assay mixture consisted of 20 mM potassium phosphate buffer (pH 7.6); containing 0.3 M KCI, 0.2 mM MnC12, 0.8 mM NAD ÷ and 0.4 mM IPM (threo-D,L-3-isopropylmalic acid), and appropriately diluted enzyme preparation in 20 mM potassium phosphate buffer (pH 7.6) in a final volume of 600 /xl. The enzyme activity was determined by measuring the initial rate in the absorbance at 340 nm using a temperature controlled Gilford Response spectrophotometer or a Hitachi 2000 Spectrophotometer. One unit of activity was defined as the amount of enzyme that reduces one /xmol of NAD + per minute, using a molar absorption coefficient for NADH at 340 nm of 6.22 × 103 M - J cm-~. The specific activity is defined as units per mg protein.

107

albumin as standard. SDS-PAGE was performed in slab gels [10], and protein bands were detected by staining with Coomassie brilliant blue R. N-terminal protein sequence analysis was performed by the Edman-method using a gas phase sequencer (Model 470, Applied Biosystems). The heat stability of the enzyme was determined by measuring the remaining activity after heat treatment. The enzyme was incubated in 20 mM potassium phosphate buffer (pH 7.6) at temperatures from 35°C to 78°C for 10 min. Aliquots were removed from each sample to determine the remaining activity under standard assay conditions at 40°C (E. coli enzyme) or 60°C (T. thermophilus enzyme). The temperature dependence of the enzymatic reaction was determined by pre-incubating the assay mixture at the specified temperatures (15-80°C) until thermal equilibrium was reached. Enzyme was then added to initiate the reaction. The pH dependence of the reaction was determined by adjusting the pH of the assay mixture with wide-range-buffer [11]. The buffer contains boric acid, citric acid, KH2PO 4 and diethyl barbituric acid, each at 0.0286 M; the final concentration in the assay mixture was 20 mM. The assay mixture was then incubated at 40°C and the reaction was started as above.

2.6. Circular dichroism A thermal melting profile was obtained on a Jobin-Yvon spectropolarimeter at a wavelength of 220 nm. The temperature of the enzyme solution was controlled by a Haake Temperature Programmer PG20. The sample temperature was monitored with a thermoelement that penetrated into the water stream close to the quartz-cell, the heating rate was l°C/min. A CD spectrum from 200 to 250 nm was recorded on a Jasco J-40A spectropolarimeter at a constant temperature of 25°C.

3. Results

2.5. Analytical procedures 3.1. Purification The protein concentration was determined by the Bradford method [9] using the standard reagent purchased from Bio-Rad or Pierce and bovine serum

The enzyme fractions eluted from the DEAE anion-exchange column were homogeneous as judged

108

G. Wallon et al. / Biochimica et Biophysica Acta 1337 (1997) 105-112

Table 1 Purification of isopropylmalate dehydrogenase from E. coli 100

Step

Total activity Total Relative activity Purification (dA. min - i ) protein (mg) (dA- min- i m g - i ) index

Extract DEAE Butyl Sizing DEAE

131 600 56800 43719 43600 40000

4120 1394 200 96 77

32 41 219 454 520

1.0 1.3 6.8 14.2 16.3

g*

80

.>

60 i

4o or"

Note: purification was started from 4 I culture of E. coli cells harboring the plasmid pWallil.

2O

,

I

,

1.0

20

Concentration [mM]

by SDS-PAGE (not shown). The purification procedure is summarized in Table 1. The N-terminal amino-acid sequence was determined as Asp-LysAsn-Tyr-His-Ile-Ala-Val-Leu-Pro. The sequence is identical to the amino-acid sequence predicted from the nucleotide sequence of the leuB gene with the exception of the first methionine coded by the initiation codon. The CD spectrum of the native enzyme was recorded (Fig. 1). The spectrum has a minimum at 220 nm.

E E

% -10

-15

-20 20O

I

I

210

220

I

I

2.30 240

250

Wavelength [nm]

Fig. 1. Circular dichroism spectrum of ,5". coli isopropylmalate dehydrogenase. The protein concentration was 0.22 m g / m l in 20 mM KP i buffer (pH 7.6). The spectrum was recorded at 25°C in a quartz cuvette of 0.1 cm path length.

Fig. 2. Dependence of the activity on the two divalent cations Mn 2- ( ~ ) and Mg 2- ( 0 ) . The activity was investigated in 50 mM Hepes-buffer at pH 7.6.

3.2. Effects of the di~,alent cations Mn 2 + and Mg 2 +, the monoualent cation K +, pH and temperature on the catalytic activity The presence of a divalent cation is essential for the enzyme activity - - i.e., chelation with EDTA leads to complete loss of activity. The metal ion is believed to bind in the active site and chelate the non-leaving carboxyl-group and hydroxyl-group of the substrate where it may stabilize the negative charge developing on the hydroxyl group in the transition state for dehydrogenation [12]. To test whether this binding site is specific for Mn 2+ or Mg 2+ we dialyzed the enzyme against 5 mM EDTA overnight to chelate the metal ion and assayed its activity in the standard assay system with varying MgCI 2 or MnC12 concentrations using 50 mM Hepes-buffer (pH 7.6). From Fig. 2, it is obvious that Mn 2+ is the preferred divalent cation. The activity was twice as high with Mn 2+ than with Mg 2+ when saturation was reached at 0.2 mM cation concentration. This corresponds to the values found for the T. thermophilus dehydrogenase. Yamada et al. [7] found that monovalent cations, specifically K--ion, enhance the activity of T. thermophilus IPMDH. The effect of K+-ion on the E. coli enzyme was investigated. The results in Fig. 3 demonstrate that the optimum concentration for K + is slightly higher than that of the thermophile en-

G. Wallon et al. / Biochimica et Biophysica Acta 1337 (1997) 105-112

109

120

100'

100

Z

•

~

"

0

8O 60 63°C

83oc

40 "~

40

20

2O

0

i

35 I

0

o

1

45

55

I

1 KCI[M]

Fig. 3. Effect of KCI on the activity. The dependence of the activity on KCI was investigqted under standard conditions. E. coli I P M D H ( O ) , T. thermophilus I P M D H (C)).

zyme. Concentrations higher or lower than 0.3 M K ÷ lead to a pronounced decrease in its activity. The effect of pH on the enzyme activity was studied. Since a wide-range-buffer was used for the pH dependence, it is assumed that the enzyme activity was not greatly affected by the buffer species. A bell-shaped curve with a maximum activity at pH 7.6 was obtained. In another experimenl the thermal stability of the mesophilic IPMDH was measured by incubating the enzymes for 10 min at temperatures between 35°C and 85°C and recording the remaining activity under the standard assay conditions. The temperature of half inactivation for wild-type E. coli IPMDH was 64°C and 83°C for the 7"~ thermophilus enzyme (Fig. 4). The thermal denaturation profile as determined by CD showed irreversible denaturation, with midpoints at 62.5°C and 78.6°C for the E. coli and the T. thermophilus enzymes, respectively (data not shown). We compared the activities of both the thermophilic and mesophilic wild-type enzymes in a temperature range from 20°C to 80°C. Fig. 5 indicates that the mesophilic enzyme is the more active species at all temperatures. The optimum temperature of the E. coli enzyme was about 70°C and that of T. thermophilus enzyme was higher than 80°C. This temperature dependence curve was used to construct an Arrhenius plot and to calculate activation energy parameters for both enzymes. The results are tabu-

65

75

85

95

Temperature [Oc]

Fig. 4. Remaining activity after heat treatment. The enzymes were incubated for 10 min. at the respective temperatures. The remaining activity was then measured under standard conditions. E. coli I P M D H ( O ) , T. thermophilus I P M D H ( O ) .

160 140 120

100 w

80 60 40 20

0 20

30

40 50 60 Temperature [°C]

70

80

Fig. 5. The temperature dependence of the activity of the E. coli T. thermophilus ( ~ ) IPMDH. The catalytic activity (initial rate) was measured with the respective standard buffers at the temperatures stated. ( 0 ) and the

Table 2 Energy of activation ( E a) and thermodynamic activation parameters (G,H,S) at 40°C Parameter

E. coli I P M D H T. thermophilus IPMDH

k~., ( s i) E~ (kJ m o l - i ) . AG " (lO m o l - ~) AH " (kJ tool- t) AS " ( J m o l - l d e g - I )

69 27 68 24 _140

8 43 71 38 -104

" Experimental energy of activation above 30°C.

110

G. Walhm et al. / Biochimica et Biophysica Acta 1337(1997) 105-112

Table 3 Enzymatic parameters of E. coli and 7~ thermophilus IPMDH at two different temperatures Species T Km (gM) kc,,, (°C) IPM NAD (s- r) 40 60 40

E. coli T. thermophilus T. thermophilus

105 44 316

321 139 157

69 19 8

k ~t/ K,,,

(s I M l ) 6.56 + 105 4.28 × 105 2.66 × 10 4

a kcat is expressed as reaction per dimer.

lated in Table 2. The lower energy of activation for the mesophile enzyme as compared to the thermophile one correlates with its higher specific activity, but the contribution of the enthalpy and entropy terms to ZIG* differ in both enzymes. 3.3. Kinetic analogs

constants

and

analysis

o f substrate

Kinetic constants were determined at the respective standard conditions. The values for the wild-type enzymes are tabulated in Table 3. There is an obvious difference between the thermophilic and the mesophilic enzyme at their respective optimum temperatures. This difference is manifested mainly in a larger k~ t for the mesophile enzyme (see also Fig. 5). The difference in catalytic efficiency kcat/Km is not as great, mainly due to the lower K m of the thermophile enzyme. Upon lowering the temperature for the thermophile enzyme to 40°C (a non-optimal value), K m increases by a factor of 5 - 8 , whereas the kcat does not decrease significantly. This leads to a decrease in the catalytic efficiency by an order of magnitude as compared to the results at 60°C.

The substrate and three of its analogs containing different substituents at C-3 position were tested for their ability to react with the E. coli enzyme. The compounds are listed in Table 4 according to the size of their substituents. These results indicate that ethylmalate and IPM are equally good substrates for the enzyme. It could be concluded that the additional hydrophobic interactions that are possible with a propyl-group do not contribute significantly to the binding of the substrate. Isobutylmalate, with its longer alkyl-chain, was not a substrate for the E. coli enzyme. This result suggests that the part of the active site that binds the hydrophobic moiety of the substrate is fairly shallow. Tert-butylmalate was a bad substrate. Its side-chain may nevertheless be accommodated because of its more compact conformation. 3.4. Complementation

Both the wild-type mesophile and thermophile enzymes complemented a chromosomal leuB deficiency (leuB6) equally well if they are expressed from an uninduced high copy number vector (e.g.

Table 4 Enzymatic parameters of E. coli IPMDH for substrate analogs Substrate Km kcal kcat/ K m (/xM) (s- ~) (s-' M-') 7.47 × 105 Ethylmalate 90 70 6.56 × 105 lsopropylmalate 105 69 Isobutylmalate na " na " na b 2.6 × 104 tert-Butylmalate 214 6 Data taken from Ref. [13]. b na: No activity.

Relative value of E. coli 1.1 1.0 na b

0.04

kcat/Km T. thermophilus ~

1.2 1.0 0.6 0.08

G. Wallon et al. / Biochimica et Biophysica Acta 1337 (1997) 105-112

pBluescript KS-, Stratagene) on minimal medium without leucine. However, if they are placed in a low copy number vector, there was a marked difference in growth rate. E. coli JA221 harboring E. coli leuB gene grew within 24 h at 37°C to well visible size, whereas the same strain harboring T. thermophilus leuB gene did not grow at all during that time span.

4. Discussion The CD spectrum of E. coli isopropylmalate dehydrogenase (Fig. 1) closely resembles that of T. thermophilus [7], suggesting that the secondary structures of E. coli enzyme is similar to those of the thermophile enzyme. Amino-acid sequences of these two enzymes are highly homologous to each other [6]. Based on these observations, we can speculate that the overall structures of these enzymes are similar to each other. The mesophilic enzyme from E. coli shared many properties with its therrnophilic counterpart from T. thermophilus. It showed a similar dependence on Mg 2+ and Mn 2+, with Mn 2÷ being the preferred divalent cation and it required K ÷ ions for optimum activity. Both enzymes had their maximum activity at a pH of approx. 7.5. A comparison of the substrate specificity of the two enzymes displayed opposed behavior towards isobutylmalate. Isobutyhnalate was not a substrate for E. coli IPMDH but was a reasonable one for its thermophilic counterpart (ratio k c a t / K m IPM/isobutyl = 1.8 for T. thermophilus IPMDH). Tertbutylmalate was a bad substrate for both the mesophilic and thermophilic IPMDH (ratio kcat/K m IPM/tert-butyl = 25 for E. Coli IPMDH and 12 for T. thermophilus IPMDH, [13]). These results demonstrate that the active sites of the two enzymes are slightly different in their sensitivity towards substrate analogs: the thermophilic enzyme can more easily accommodate a more elongated alkyl chain (i.e., isobutylmalate) than the mesophile enzyme can. A comparison of the amino-acid residues involved in substrate binding in the proposed active site shows that 9 out of l0 side chains are identical, most of which are involved in binding the malic acid part of the substrate. The only difference occurs in the alkyl-binding site, where Thr-88 in the thermophile

111

enzyme has been changed to Arg-92 in the mesophile enzyme. This change from a short to a long amino acid may lead to the difference in substrate preference that we have found. As mentioned in the introduction it has been suggested that thermophilic enzymes have a rigid structure which provides greater stability, and allows them to function at high temperatures but which would make them too inflexible and stiff (i.e., less likely to function well) at low temperatures. This is true for T. thermophilus and E. coli IPMDH as well, as has been demonstrated by comparing their activities over a range of temperatures (see Table 3 and Fig. 5). The k c a t / g m for both enzymes at their respective standard temperatures (40°C for E. coli and 60°C for T. thermophilus) was very similar. Upon lowering the temperature for the thermophile enzyme to 40°C the k c a t / K m decreased one order of magnitude. The kca t of the mesophile enzyme at 40°C was 3-times as high as that of the thermophile enzyme at 60°C. It is difficult to interpret these data in terms of active-site properties (e.g., flexibility), since kcat may be influenced by product release if that is the rate-limiting step. This interpretation would of course go well with our hypothesis: inflexible loops in the binding site (active sites) would decrease the kcat in the thermophile. We have established that the mesophile and the thermophile enzymes are very similar in their biochemical behavior - - i.e. pH, preference of Mn 2÷ as the divalent cation, requirement of K + for activity - but quite different in their stability and activity versus temperature and the specificity for some of the substrate analogs that we tested. The mesophile enzyme has been crystallized (G. Wallon, unpublished resuits). Structural comparisons are under way to try to understand these differences on a molecular basis.

References [I] Jaenicke, R. and Zavodszky, P. (1990) FEBS Left. 268, 344-349. [2] Petsko, G.A. and Ringe0 D. (1984) Annu. Rev. Biophys. Bioeng. 13, 331-371. [3] Vihinen, M. (1987) Protein Eng. 1,477-480. [4] lmada, K., Sato, M., Tanaka, N., Katsube, Y., Matsuura, Y. and Oshima, T. (1991) J. Mol. Biol. 222, 725-738.

112

G. Wallon et al. / Biochimica et Biophysica Acta 1337 (1997) 105-112

[5] Stoddard, B.L., Dean, A.M. and Koshland, D.E. (1993) Biochemistry 32. 9310- 9316. [6] Kirino, H., Aoki, M., Aoshima, M., Hayashi, Y., Ohba, M., Yamagishi, A., Wakagi, T. and Oshima, T. (1994) Eur. J. Biochem. 220, 275-281. [7] Yamada, T., Akutsu. N., Miyazaki, K., Kakinuma, K., Yoshida, M. and Oshima, T. (1990) J. Biochem. 108, 449456. [8] Lerner, C.G. and Inouye, M. (1990) Nucleic Acids Res. 18, 4631.

[9] Bradford, M.M. (1976) Anal. Biochem. 72, 248-254. [10] Laemmli, U.K. (1970) Nature 227, 680-685. [11] Britton, H. T. S. and Robinson, R. A. (1931) J. Chem. Soc. 458-473. [I 2] Hurley, J.H., Dean, A.M., Koshland, D.E. and Stroud, R.M. ( 1991 ) Biochemistry 30, 8671-8678. [13] Miyazaki, K., Kakinuma, K., Terasawa, H. and Oshima, T. (1993) FEBS Lett. 332, 35-36.