Physicochemical characterisation of the two active site mutants Trp52→Phe and Asp55→Val of glucoamylase from Aspergillus niger

Biochimica et Biophysica Acta 1601 (2002) 163 – 171 www.bba-direct.com

Physicochemical characterisation of the two active site mutants Trp52!Phe and Asp55!Val of glucoamylase from Aspergillus niger Trine Christensen a,1, Torben P. Frandsen b, Niels C. Kaarsholm c, Birte Svensson b, Bent W. Sigurskjold a,* a

Department of Biochemistry, August Krogh Institute, University of Copenhagen, Universitetsparken 13, DK-2100 Copenhagen Ø, Denmark b Department of Chemistry, Carlsberg Laboratory, Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark c Novo Nordisk A/S, Novo Alle´, DK-2880 Bagsværd, Denmark Received 21 March 2002; received in revised form 2 September 2002; accepted 25 September 2002

Abstract Glucoamylase 1 (GA1) from Aspergillus niger is a multidomain starch hydrolysing enzyme that consists of a catalytic domain and a starch-binding domain connected by an O-glycosylated linker. The fungus also produces a truncated form without the starch-binding domain (GA2). The active site mutant Trp52 ! Phe of both forms and the Asp55 ! Val mutant of the GA1 form have been prepared and physicochemically characterised and compared to recombinant wild-type enzymes. The characterisation included substrate hydrolysis, inhibitor binding, denaturant stability, and thermal stability, and the consequences for the active site of glucoamylase are discussed. The circular dichroic (CD) spectra of the mutants were very similar to the wild-type enzymes, indicating that they have similar tertiary structures. The D55V GA1 mutant showed slower kinetics of hydrolysis of maltose and maltoheptaose with DDGzi22 kJ mol 1, whereas the binding of the strong inhibitor acarbose was greatly diminished by DDGji52 kJ mol 1. Both W52F mutant forms have almost the same stability as the wild-type enzyme, whereas the D55V GA1 mutant showed slight destabilisation both towards denaturant and heat (DSC). The difference between the CD unfolding curves recorded by near- and far-UV indicated that D55V GA1 unfolds through a molten globule intermediate. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Glucoamylase; Enzyme mechanism; Protein unfolding; Circular dichroism spectroscopy; Differential scanning calorimetry; Isothermal titration calorimetry

1. Introduction Glucoamylase (1,4-a- D -glucan glucohydrolase, EC 3.2.1.3) is an inverting exo-acting hydrolase that degrades starch and related oligo- and polysaccharides of glucose from the non-reducing end, releasing h-D-glucose [1,2]. Glucoamylase occurs naturally in two forms, glucoamylase 1 (GA13) and glucoamylase 2 (GA2). GA1 consists of a catalytic domain (Ala1 – Thr440; glycoside hydrolase family

Abbreviations: CBM20, carbohydrate-binding module family 20; CD, circular dichroism spectroscopy; DSC, differential scanning calorimetry; GA1, glucoamylase 1; GA2, glucoamylase 2; GH15, glycoside hydrolase family 15; ITC, isothermal titration calorimetry; rec-GA, recombinant wildtype glucoamylase; SBD, starch-binding domain * Corresponding author. Tel.: +45-3532-1748; fax: +45-3532-1567. E-mail address: [email protected] (B.W. Sigurskjold). 1 Present address: Department of Chemistry, Duke University, A020 Levine Science Research Center, Box 90317, Durham, NC 27708, USA.

15 (GH15)) [3] and a starch-binding domain (SBD) (Cys509 – Arg616; carbohydrate-binding module family 20 (CBM20)) [3] connected by a highly O-glycosylated linker region [4]. GA2 lacks the SBD, but the primary structure of GA2 is identical to that of Ala1 – Pro512 of GA1 [2,4]. The three-dimensional structure of the catalytic domain including part of the linker region of a glucoamylase from Aspergillus awamori var. X100 has been solved by X-ray crystallography [5] and the structure of the SBD has been solved by NMR [6,7]. Trp52 and Asp55 are both located in the active site of glucoamylase and are involved in activity, but their individual roles in the catalytic process are different [8– 10]. Fig. 1 shows the side chains of selected amino acid residues and the inhibitor acarbose bound in the active site of glucoamylase from A. awamori var. X100 [11]. Trp52 is located close to the active site and a hydrogen bond between the side chains of Trp52 and Glu179 is observed in the crystal structure [5,12,13]. Glu179 has been identified as the general acid

1570-9639/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 1 5 7 0 - 9 6 3 9 ( 0 2 ) 0 0 4 6 3 - 6

164

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

and of the two active site mutants, Trp52 ! Phe and Asp55 ! Val. Heat-induced unfolding was monitored by CD at two different wavelengths, 222 and 280 nm. In this way, separate information is obtained on the disruption of secondary and tertiary structures of glucoamylase.

2. Materials and methods 2.1. Materials

Fig. 1. A section of the active site of glucoamylase showing side chains of selected amino acid residues (thin) and acarbose (thick) [11]. The rings of acarbose are labelled A – D from the nonreducing end. A is the valienamine in the bottom of the active site. Asp55 is situated in the bottom of the active site and Trp52 is placed close to the catalytic site between rings A and B of the inhibitor.

catalyst [14,15] and one of the roles of Trp52 is therefore probably to keep the right orientation of Glu179 during the catalytic process. Also, hydrophobic interactions between the tryptophan side chain and the inhibitor acarbose in the active site are observed [16] and inhibitor binding affects hydrogen exchange from residue Trp52 [17]. After Trp52 is mutated to Phe, substrates are bound more tightly, while the rate of catalysis is very low [9]. Asp55 is situated at the bottom of the active site, and in uncomplexed crystal the side chain of Asp55 adopts two different conformations, reflecting a high degree of flexibility [5,12]. In the complexes with the active site inhibitors 1-deoxynojirimycin and acarbose, however, only a single conformation of the side chain of Asp55 was found, which was involved in formation of two hydrogen bonds to the ligand [11,13,16]. Mutations of Asp55 to either Asn, Gly, or Glu resulted in enzyme variants able to bind substrates, but which are essentially without hydrolytic activity [8,10]. In ligand complexes, Asp55, located at subsite  1, interacts with the two key polar groups of maltose 6V-OH and 4V-OH [18 –21]. Unfolding of glucoamylase has been studied in detail [22 –24]. The catalytic domain unfolds irreversibly at all pH values studied, whereas the SBD unfolds reversibly at pH 5.5 –7.5. Heat-induced unfolding followed by DSC of a form of the enzyme purified from a commercial preparation of glucoamylase shows that the catalytic domain unfolds irreversibly according to a one-step model, whereas the SBD unfolds reversibly [22,24,25]. Here, two different methods, differential scanning calorimetry (DSC) and circular dichroism (CD) spectroscopy, are used to study the thermal unfolding process of wild-type glucoamylases

Maltose, maltoheptaose, and the glucose oxidase kit were from Sigma (St. Louis, MO). Guanidine hydrochloride was from Fluka (Buchs, Switzerland). Acarbose was a generous gift from Bayer AG (Wuppertal, Germany). Water was drawn from a Milli-Q system (Millipore, Bedford, MA). Other chemicals used were of analytical grade. The wildtype glucoamylases were produced as culture supernatants from a recombinant wild-type strain (rec-GA1 and recGA2). All wild-type glucoamylase culture supernatants were obtained from Novo Nordisk A/S (Bagsværd, Denmark). The wild-type and mutant GA1 and GA2 forms were separated and purified as essentially described previously [2,26 – 28]. D55V was constructed using the following primer 5V-ACC TGG ACT CGA GTC TCA GGC CTC GTC CTC AAG-3V and transformation and expression in Aspergillus niger was performed essentially as described [27]. W52F was constructed as previously described [9]. 2.2. Enzyme kinetics Initial rates of hydrolysis were determined for maltose and maltoheptaose at 45 jC in 0.05 M sodium acetate at pH 4.5 using up to 12 substrate concentrations ranging from eight times under Km to eight times above Km. Released glucose was measured by the glucose oxidase method with monitoring in microtiter plates [27]. The kinetic measurements were performed using the GA1 forms of mutant and wild-type glucoamylases at enzyme concentrations of 2– 7 AM for Asp55 ! Val and 0.00117– 0.007 AM for wild-type GA. The kinetic parameters kcat and Km were obtained by non-linear regression analysis of the values of the initial rates, v, as a function of [S] to the Michaelis– Menten equation using the program ENZFITTER. Changes in activation energy for substrate hydrolysis were calculated from DDGz =  RTln{(kcat/Km)mutant/(kcat/ Km)wild-type} [29]. 2.3. Guanidine hydrochloride denaturation Unfolding of GA (0.24 AM) was performed essentially as described [30] by guanidine hydrochloride from 0 to 8 M in 50 mM sodium acetate, pH 4.5, and followed by fluorescence changes using a Perkin – Elmer (Shelton, CT) LS50B luminescence spectrometer. Excitation and emission wavelengths were 280 and 320 nm, respectively, with a slit width

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

165

Table 1 Kinetic parameters for the hydrolysis of maltose and maltoheptaose and increase in transition-state binding energy, DDGz, between wild-type and Asp55 ! Val glucoamylase GA1 from A. nigera Maltose

Wild-type Asp55 ! Val

Maltoheptaose

kcat (s 1)

Km (mM)

kcat/Km (s 1 mM 1)

DDGzb (kJ mol 1)

kcat (s 1)

Km (mM)

kcat/Km (s 1 mM 1)

DDGz (J mol 1)

10.7 F 0.6 (2.1 F 0.1)  10 3

1.21 F 0.14 0.95 F 0.21

8.84 F 1.14 (2.2 F 0.1)  10 3

– 22.0 F 0.2

59.7 F 1.6 0.013 F 0.004

0.12 F 0.01 0.12 F 0.04

498 F 44 0.11 F 0.05

– 22.3 F 0.3

a

Determined at 45 jC in 0.05 M sodium acetate, pH 4.5. Determined as DDGz =  RTln{(kcat/Km)mutant/(kcat/Km)wild-type} [29]. Uncertainties on kcat and Km are standard errors from the regression analysis and uncertainties on kcat/Km and DDGz have been calculated from error propagation. b

of 5 nm. From the measured fluorescence intensity, the fraction of unfolded protein, fu, was calculated from fu=( FN  Fobs)/( FN  FU), where FN and FU are fluorescence intensities with the folded and unfolded protein, respectively, and Fobs is the fluorescence observed after 1 h incubation at room temperature.

MicroCal. This instrument has been described in detail by Wiseman et al. [31]. The reference cell was filled with water and the instrument was calibrated with electrical pulses. The concentration of the glucoamylase mutant was 70 AM and the ligand concentration in the syringe was 580 AM. Twenty injections were carried out at 27 jC and the data were analysed as described previously [32].

2.4. Differential scanning calorimetry 2.6. CD spectroscopy DSC experiments were performed using an MCS differential scanning calorimeter from MicroCal, Inc. (Northampton, MA). A scan rate of 60 jC h 1 was used in all experiments. Matching buffer –buffer scans were recorded corresponding to each protein –buffer scan and the reversibility was tested each time. All experiments were carried out in 50 mM sodium phosphate buffer, pH 7.5. Protein concentrations of the wild-type rec-GA1 and rec-GA2 were 38.9 and 37.3 AM, respectively. Concentrations of the mutant glucoamylases were 24.5 and 20.1 AM for Trp52 ! Phe GA1 and GA2, respectively, and 19.1 AM for Asp55 ! Val GA1. DSC profiles corresponding to irreversible and reversible transitions were fitted to an irreversible one-step model and a reversible non-two-state model, respectively. Data analysis of the irreversible unfolding process of glucoamylase has been described previously [22,24]. 2.5. Isothermal titration calorimetry (ITC) Ligand binding thermodynamics measurement was performed by using an OMEGA titration calorimeter from

The CD spectroscopy experiments were performed in a Jasco Corp. (Tokyo, Japan) J-715 spectropolarimeter supplied with a JTC-340 temperature control program. The CD spectra of glucoamylases were obtained in the far-UV and near-UV regions using the wavelength ranges 185 – 260 and 250 – 350 nm, respectively. The light paths of the cuvettes were 0.01 and 0.1 cm for the far- and near-UV experiments, respectively. In the unfolding experiments, the proteins were heated with a scan rate of 60 jC h 1 (as in the DSC experiments) using an external water bath, and the temperature was computer controlled. Thermal unfolding were followed in the farUV region at 222 nm and in the near-UV region at 280 nm. Protein concentrations of wild-type and mutant glucoamylases were 12.6– 57.9 AM for measurements in the near-UV region and 0.49 – 11.9 AM in the far-UV region. In the calculations, the molar concentration of peptide bonds was used in the far-UV region and molar protein concentrations were used in the near-UV region. Noise reduction of the obtained spectra was done according to instructions using the J700 for Windows software

Table 2 Association constants and thermodynamic parameters for the binding of acarbose to the active site of wild-type and Asp55 ! Val glucoamylase GA1a Wild-typeb Asp55 ! Val a

Ka (M 1)

DHj (kJ mol 1)

 TDSj (kJ mol 1)

DGj (kJ mol 1)

(9.4 F 6.0)  1011 (9.8 F 0.3)  102

 32.8 F 1.4  125.0 F 2.0

 36.1 F 2.1 108.0 F 2.0

 68.9 F 1.6  17.0 F 0.1

Determined at 27 jC in 0.05 M sodium acetate, pH 4.5. Values were measured in a displacement experiment with 1-deoxynojirimycin where no binding of the inhibiting ligand was detected. However, it is possible that weak binding occurred below the detectable limit. Therefore, the uncertainties may be larger than indicated. b From Ref. [33]. Uncertainties on K and DH are standard errors from the regression analysis and uncertainties on DSj and DGj have been calculated from error propagation.

166

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

catalytic transition state, is reduced nine orders of magnitude in the mutant corresponding to a change in the binding free energy of more than 40 kJ mol 1. This is approximately twice the value of DDGz obtained for loss of transition-state stabilisation for hydrolysis of maltose and maltoheptaose (Table 1). 3.2. Guanidine hydrochloride denaturation Fig. 2 shows the denaturation of wild-type and Asp55 ! Val GA1 induced by guanidine hydrochloride. The mutant is only slightly less stable towards denaturant than the wild-type enzyme. Similar destabilisations were previously observed for the mutants Trp317 ! Phe and Asp309 ! Glu [30]. Fig. 2. Guanidine hydrochloride induced denaturation of wild-type (o) and Asp55 ! Val mutant ( ) GA1 glucoamylase.

.

supplied by Jasco. The traces obtained by thermal unfolding were normalised and the degree of unfolding is shown between 0 and 1.

3. Results 3.1. Enzyme kinetics and inhibitor binding The kinetic parameters for the hydrolysis of the substrates maltose and maltoheptaose at 45 jC are summarised in Table 1. Asp55 ! Val GA1 showed extremely low kcat values on maltose and maltoheptaose, which were reduced over 5000-fold compared to the wild-type enzyme. The Km values, in contrast, were essentially unchanged for Asp55 ! Val compared to the wild-type enzyme. The second-order rate constant, kcat/Km, is thus also highly reduced, resulting in changes in the activation energy, DDGz, of around 22 kJ mol 1 for Asp55 ! Val compared to wildtype GA (Table 1). The thermodynamic parameters of acarbose binding to wild-type and Asp55 ! Val GA1 mutant enzymes were determined by ITC and are shown in Table 2. The binding of this pseudotetrasaccharide, which might resemble the

3.3. Differential scanning calorimetry All recorded calorimetric traces obtained by thermal unfolding of the wild-type and the active site mutant glucoamylases, Trp52 ! Phe and Asp55 ! Val, show endothermic peaks. All GA2 forms, both the wild-type enzyme and the mutant, unfold completely irreversibly, whereas a small peak corresponding to an SBD appears in the rescans of all GA1 forms. To verify if the unfolding process of the catalytic domains of rec-GA1, rec-GA2, and the mutants follow a one-step model, experiments using different scan rates ought to be carried out, but due to lack of protein, these experiments have not been performed. The GA1 and the GA2 forms of the wild-type glucoamylases have identical primary structures and until now, no significant differences have been observed in the three-dimensional structures between the mutant and the wild-type glucoamylases [34]. Consequently, it is plausible and reasonable to assume that the catalytic domains of both the wild-type and the mutant glucoamylases unfold according to the irreversible one-step model. The one-step irreversible unfolding model can be outlined as k

F ! D;

ð1Þ

DH

where F is the folded protein, D is the irreversibly denatured protein, k is the rate constant of the process, and DH is the

Table 3 Activation energy, unity temperature, and enthalpy for the thermal unfolding process of the catalytic domain of glucoamylase GA1 and GA2 forms of recombinant wild-type and active site mutant glucoamylases Cell content

Peak 1 Ea (kJ mol 1)

rec-GA1 rec-GA2 Trp52 ! Phe GA1 Trp52 ! Phe GA2 Asp55 ! Val GA1a a

Peak 2 Tu (jC)

DHj (kJ mol 1)

Ea (kJ mol 1)

311 F 27 273 F 24

63.1 F 0.7

514 F 44

Peak 3 Tu (jC)

56.1 F 0.6

DHj (kJ mol 1)

203 F 30

Ea (kJ mol 1)

Tu (jC)

DHj (kJ mol 1)

275 F 4 292 F 36 264 F 2 284 F 6 363 F 10

66.9 F 0.1 66.7 F 0.1 62.1 F 0.1 64.9 F 0.1 71.4 F 0.1

2108 F 27 1574 F 16 2332 F 12 1487 F 30 1839 F 40

The first peak corresponds partly to unfolding of the SBD. Uncertainties are standard errors from the regression analysis.

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

167

molar heat accompanying the unfolding. The temperature dependence of the excess heat capacity is then given by [35]:   Z k 1 T hCp;exc i ¼ DH exp  kdT : ð2Þ v v T0 Here v is the scan rate, T is the absolute temperature, and T0 is a reference temperature at which the protein is completely folded. The rate constant is furthermore determined by the Arrhenius equation:    Ea 1 1 k ¼ exp   ; ð3Þ R T TU where Ea is the activation energy, TU is the unity temperature at which the rate constant is 1 min 1, and R is the universal gas constant. The obtained calorimetric traces are fitted to Eq. (2) combined with Eq. (3) and the results are summarised in Table 3. The catalytic domains of rec-GA1 and rec-GA2 unfold in a single transition (Fig. 3 and Table 3). Trp52 ! Phe GA1 and Trp52 ! Phe GA2 are slightly destabilised compared to the respective wild-type forms and the denaturation temperatures are 57.8 and 61.1 jC, respectively (Fig. 4A,B). The corresponding denaturation temperatures for the rec-GA1

Fig. 4. Calorimetric traces of unfolding of the two glucoamylase active site mutants (first scan). (A) Trp52 ! Phe GA1, (B) Trp52 ! Phe GA2, and (C) Asp55 ! Val GA1. The solid lines are the experimentally obtained DSC profiles, the dashed curves are the discernible peaks, and the dotted lines are the resulting curve obtained from fitting the experimentally obtained curves to Eqs. (2) and (3).

Fig. 3. Calorimetric traces of unfolding the recombinant wild-type glucoamylase forms (first scan). (A) rec-GA1 and (B) rec-GA2. The solid lines are the experimentally obtained DSC profiles and the dotted lines are the resulting curve obtained from fitting the experimentally obtained curves to Eqs. (2) and (3).

and rec-GA2 are 62.7 and 63.1 jC, respectively (Fig. 3). Data analysis of the recorded calorimetric trace obtained from the unfolding process of Trp52 ! Phe GA1 reveals one transition and the obtained parameters are shown in Table 3. The major difference between Trp52 ! Phe GA1 and recGA1 is a decrease in the unity temperature of the mutant glucoamylase of almost 5 jC. The phenylalanine residue introduces a higher degree of flexibility in the active site, which may result in a slightly less stable molecule. The DSC profile of the Trp52 ! Phe GA2 mutant shows a transition that is resolved into two peaks (Table 3, Fig.

168

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

Table 4 Denaturation temperature, calorimetric and van’t Hoff enthalpies for the reversible unfolding process of the starch-binding domain of wild-type and mutant glucoamylases Cell content

Td (jC)

DHj (kJ mol 1)

DHvH (kJ mol 1)

rec-GA1 Trp52 ! Phe GA1 Asp55 ! Val GA1

58.9 F 0.1 59.3 F 0.1 57.4 F 0.1

281 F 2 255 F 5 207 F 2

435 F 3 460 F 12 498 F 5

Uncertainties are standard errors from the regression analysis.

4B), whereas the transition of the rec-GA2 only shows one peak (Fig. 3B). Again, Trp52 ! Phe GA2 has a more flexible active site, and when the SBD is absent, a small part of the catalytic domain unfolds before the rest of the molecule [22,24]. The calorimetric trace obtained from the thermal unfolding process of Asp55 ! Val GA1 shows two discernible peaks in the first scan, a minor peak at lower temperatures and a major peak at higher temperatures (Fig. 4C). The DSC profile of the rescan shows a transition corresponding to an SBD located at the same position as the minor peak of the first scan (Table 4). However, the enthalpy of the minor peak obtained from fitting to Eq. (2) in the first scan of Asp55 ! Val GA1 (Table 3) is more than twice as large as the enthalpy of unfolding the SBD analysed according to the reversible non-two-state model (Table 4). This means that the enthalpy obtained from unfolding the minor peak in the first scan of Asp55 ! Val GA1 probably corresponds to the unfolding processes of both the SBD and of part of the catalytic domain. The major peak obtained by unfolding Asp55 ! Val GA1 presumably corresponds to the catalytic domain, and this domain unfolds at elevated temperatures of about 10 jC compared to the wild-type enzyme. The aspartic acid residue is placed at the bottom of the active site and somehow the valine in position 55 stabilises the catalytic domain. Both the activation energy and the unity temperature are increased for the unfolding process of the mutant compared to the wild-type enzyme. Hence, more energy has to be applied to the protein to reach the transition state of unfolding. All GA1 forms studied here show a reversible transition in the rescans, corresponding to thermal unfolding of the SBD. Therefore, the obtained DSC profiles were fitted to a reversible non-two-state model and the calculated thermodynamic parameters are shown in Table 4. The denaturation temperatures of the SBD from recGA1, Trp52 ! Phe GA1, and Asp55 ! Val GA1 are very similar. The van’t Hoff enthalpies are larger than the calorimetric enthalpies in every case. Differences in the calorimetric and van’t Hoff enthalpies means that the unfolding process does not follow a simple two-state mechanism, and when the van’t Hoff enthalpy is the larger, oligomerisation or other intermolecular interactions occur during unfolding [36,37]. These processes are not dependent on protein concentration in the range that can be measured

by DSC and they are probably reversible since five successive scans show only minor losses in enthalpy of unfolding [37]. 3.4. CD spectroscopy When comparing the spectra of the GA1 and the GA2 forms obtained in the far-UV region (180 – 260 nm), no large differences are seen in spite of the fact that the SBD of GA1 contains eight h-strands [6,7]. Consequently, the h-sheet structure does not contribute significantly to the CD spectra.

Fig. 5. CD spectra obtained in the far-UV region (185 – 260 nm) of glucoamylase variants. (A) rec-GA1 and rec-GA2. (B) rec-GA1 and recGA2, and Trp52 ! Phe GA1 and GA2. (C) rec-GA1 and rec-GA2, and Asp55 ! Val GA1.

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

Fig. 6. CD spectra obtained in the near-UV area (250 – 350 nm) of glucoamylase variants.

The spectrum of Trp52 ! Phe GA1 corresponds closely to that of the wild-type enzyme, whereas that of Trp52 ! Phe GA2 shows minor changes (Fig. 4B). Trp52 is located just before the start of the second a-helix [5,12], and since the GA2 molecules seem to be slightly more flexible than the

169

GA1 molecules, the mutation may result in minor local changes in the secondary structure. The spectra of Asp55 ! Val GA1 are slightly different from those of the wild-type glucoamylase (Fig. 5C). Asp55 is placed in the second a-helix and, as mentioned previously, the side chain of Asp55 adopts different conformations in the crystal structure of the free glucoamylase [5,12]. The mutation probably distorts the local arrangement of secondary structure in this region. Spectra obtained in the near-UV region (250 – 350 nm) of the wild-type and the mutant glucoamylases are shown in Fig. 6. The near-UV spectra of GA1 and GA2 forms differ because the two forms of the enzymes contain different numbers of tryptophan residues, namely 19 and 15, respectively. The near-UV spectra of Trp52 ! Phe GA1 and GA2 differ slightly from those of the wild-type enzymes, presumably due to the lack of one tryptophan residue in the mutant form, and it is therefore difficult to make the comparison. However, the Trp52 ! Phe mutant forms probably have secondary and tertiary structures that correspond closely to those of the wild-type enzyme. From molecular modelling experiments where Trp52 has been replaced with a phenylalanine residue, no major differences were seen in the structure compared to that of the wild-type enzyme [9].

Fig. 7. Thermal unfolding of glucoamylase variants monitored by CD spectroscopy in the far-UV (F-UV) region at 222 nm and in the near-UV (N-UV) region at 280 nm. (A) rec-GA1 and rec-GA2. (B) Trp52 ! Phe GA1 compared to rec-GA1. (C) Trp52 ! Phe GA2 compared to rec-GA2, (D) Asp55 ! Val GA1 compared to rec-GA1.

170

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

The environment of the tryptophan residues in the structure of Asp55 ! Val GA1 corresponds closely to that of the wildtype enzyme. However, the overall structure of the mutant glucoamylase is probably closely related to the wild-type enzymes. The thermal unfolding experiments of the wild-type and the mutant glucoamylases were measured by CD at 222 and 280 nm to follow the disruption of the secondary and tertiary structure elements, respectively (Fig. 7). The unfolding process of rec-GA2 corresponds closely to that of recGA1. Thermal unfolding recorded by CD shows that Trp52 ! Phe GA1 unfolds as rec-GA1 where the secondary and tertiary structures unfold simultaneously (Fig. 7B). However, this mutant is slightly destabilised, which is also seen in unfolding monitored by DSC. The secondary and tertiary structures of Trp52 ! Phe GA2 unfold concurrently, but at lower temperatures compared to those of rec-GA2 (Fig. 7C). The Asp55 ! Val mutant unfolds differently from the wild-type glucoamylases. The secondary structure of the GA1 form of the mutant unfolds at higher temperatures compared to the secondary structure of rec-GA1. However, the tertiary structure of the Asp55 ! Val GA1 mutant unfolds at the same temperatures as the tertiary structure of the wild-type enzyme. Consequently, the Asp55 ! Val GA1 mutant unfolds with a molten globule intermediate (Fig. 7D).

4. Discussion From CD spectra obtained in the far-UV region (180 – 260 nm) and in the near-UV region (250 – 350 nm), it is observed that the glucoamylases investigated in this study have secondary and tertiary structures that are similar to each other. Therefore, none of the mutated forms result in structural changes detectable by CD. The so-called catalytic domain A proteolytically cleaved form of the Asp55 ! Val mutant containing residues Ala1 –Val470 was produced by subtilisin proteolysis [28] and crystallised. No major changes between this structure and the wild-type structure were seen (B. Stoffer, personal communication). Therefore, the starting point in this investigation is that the threedimensional structures of the wild-type and mutant glucoamylases are essentially identical and changes in the thermal stability originate from the mutated amino acid residue. The thermal unfolding process of the GA1 and GA2 wild-type and mutant forms of glucoamylase follows different pathways. Unfolding of rec-GA1 and rec-GA2 are the exceptions and these two glucoamylases unfold similarly, irrespective of whether the process is recorded by DSC or by CD (Figs. 3 and 7A,B). From the crystal structure, it is known that the first part of the linker region (residues 420– 470) wraps around the catalytic core in a random coil-like structure [5]. The recombinant wild-type glucoamylases have a higher extent of glycosylation in the linker region

compared to the commercial wild-type glucoamylases [38]. Therefore, the higher extent of glycosylations stabilises the catalytic domain, resulting in an unfolding process of recGA2 similar to that of rec-GA1. As demonstrated using double-headed heterobidentate synthetic oligosaccharide inhibitors targeted to the two domains [39,40], the catalytic domain and the SBD are presumably close enough in space for domain– domain interactions, which keeps the linker region wrapped around the catalytic domain. This conformation stabilises the GA1 forms and prevents parts of the catalytic domain from unfolding differently from the rest of the molecule. However, the same stabilisation is observed in rec-GA2 and, therefore, the more glycosylated linker region has a similar stabilising effect on the catalytic domain as has the SBD. Whether the stabilisation of rec-GA2 arises from interactions between the more glycosylated linker region and the catalytic domain or from other interactions is not known. It is perhaps also possible that intermolecular interactions (oligomerisation) could be involved in the stabilisation achieved for rec-GA2. The thermal unfolding of the Trp52 ! Phe mutants shows that the secondary and tertiary structures of both the GA1 and GA2 forms unfold simultaneously, but at slightly lower temperatures compared to the wild-type enzymes (Fig. 7B,C). Hardly any differences are seen in the activation energies compared to the wild-type enzymes (Table 3). This confirms that the structures of the mutant glucoamylases are very similar to those of the wild-type enzymes. However, the destabilising effect, which is seen in the unfolding process of the mutant form, arises from an increase in activation entropy, and therefore the mutation results in a slightly more flexible structure, maybe due to the loss of the Trp52 –Glu179 hydrogen bond by the Phe substitution (Fig. 1). The unfolding process of the Trp52 ! Phe GA2 form recorded by DSC results in a calorimetric trace that is resolved into two peaks, whereas the rec-GA2 only shows one transition. It is probably more difficult for the Trp52 ! Phe GA2 form to keep the native structure during heating than the Trp52 ! Phe GA1 form, because of the lack of the stabilising effects of the linker region and the SBD. Trp52 plays a significant role in catalysis, where it is essential in assisting the hydrolysis process and it is also important for substrate binding [9]. The Trp52 ! Phe mutant also binds 1-deoxynojirimycin and acarbose, but with slightly reduced association constants [41]. The phenylalanine in position 52 is able to retain some of the wildtype properties, whereas mutations to non-hydrophobic amino acid residues without the capability of forming hydrophobic interactions might result in more significant changes in thermostability. The Asp55 ! Val GA1 unfolds thermally very differently from any other glucoamylase investigated until now. Unfolding of this mutant form monitored by DSC shows two discernible peaks, a major and a minor one (Fig. 4C). Unfolding of the catalytic domain, which corresponds to the major peak in the recorded DSC profile, occurs at temper-

T. Christensen et al. / Biochimica et Biophysica Acta 1601 (2002) 163–171

atures about 10 jC higher than the wild-type. Asp55 ! Val GA1 unfolds with a molten globule intermediate, since CD reveals that the tertiary structure of this protein unfolds at lower temperatures than the secondary structure. Compared to the wild-type enzyme, the tertiary structure of this mutant unfolds at similar temperatures, whereas the secondary structure of the mutant is stabilised and unfolds at elevated temperatures. The temperature where the unfolding process of the secondary structure occurs corresponds to those of the major peak in the DSC experiment. Consequently, most of the enthalpy monitored by DSC originates from unfolding of the secondary structure of the Asp55 ! Val GA1 mutant, while the disruption of the tertiary structure is largely entropy-driven [42,43]. The Asp55 ! Val mutant has a very low rate of catalysis, but it appears as if it binds the substrates maltose and maltotetraose almost identically to the wild-type enzyme. Moreover, ITC experiments show that the Asp55 ! Val mutant binds acarbose very weakly, with an association constant decreased by nine orders of magnitude (Table 2) compared to the wild-type enzyme [41]. Therefore, Asp55 is very important for substrate catalysis and binding of acarbose.

Acknowledgements This work was supported by grants from the Danish Research Councils’ Committee on Biotechnology (Grant no. 9502014) and the Danish Natural Science Research Council. We thank Dr. Jan Lehmbeck for construction of the mutants Trp52 ! Phe and Asp55 ! Val, Ms. Anne-Marie Kolstrup for performing the CD experiments, Dr. Carolyn R. Berland for performing the ITC experiment, Dr. Jørgen Sauer for making Fig. 1, and Ms. Sidsel Ehlers for technical assistance.

References [1] K. Hiromi, K. Takahashi, Z. Hamauzu, S. Ono, J. Biochem. 59 (1966) 469 – 475. [2] B. Svensson, T.G. Pedersen, I. Svendsen, T. Sakai, M. Ottesen, Carlsberg Res. Commun. 47 (1982) 55 – 69. [3] P.M. Coutinho, B. Henrissat, Carbohydrate-active enzymes server, at URL: http://www.afmb.cnrs-mrs.fr/CAZY/, 1999. [4] B. Svensson, K. Larsen, I. Svendsen, E. Boel, Carlsberg Res. Commun. 48 (1983) 529 – 544. [5] A. Aleshin, A. Golubev, L.M. Firsov, R.B. Honzatko, J. Biol. Chem. 267 (1992) 19291 – 19298. [6] K. Sorimachi, A.J. Jacks, M.-F. Le Gal-Coe¨ffet, G. Williamson, D.B. Archer, M.P. Williamson, J. Mol. Biol. 259 (1996) 970 – 987. [7] K. Sorimachi, M.-F. Le Gal-Coe¨ffet, G. Williamson, D.B. Archer, M.P. Williamson, Structure 5 (1997) 647 – 661. [8] Y. Yamashita, N. Hanada, M. Itoh-Andoh, T. Takehara, FEBS Lett. 243 (1989) 343 – 346. [9] T. Christensen, B. Stoffer, B. Svensson, U. Christensen, Eur. J. Biochem. 250 (1997) 638 – 645.

171

[10] M.R. Sierks, B. Svensson, Biochemistry 32 (1993) 1113 – 1117. [11] A.E. Aleshin, L.M. Firsov, R.B. Honzatko, J. Biol. Chem. 269 (1994) 15631 – 15639. [12] A.E. Aleshin, C. Hoffman, L.M. Firsov, R.B. Honzatko, J. Mol. Biol. 238 (1994) 575 – 591. [13] E.M.S. Harris, A.E. Aleshin, L.M. Firsov, R.B. Honzatko, Biochemistry 32 (1993) 1618 – 1626. [14] M.R. Sierks, C. Ford, P.J. Reilly, B. Svensson, Protein Eng. 3 (1990) 193 – 198. [15] B. Svensson, A.J. Clarke, I. Svendsen, H. Møller, Eur. J. Biochem. 188 (1990) 29 – 38. [16] A.E. Aleshin, B. Stoffer, L.M. Firsov, B. Svensson, R.B. Honzatko, Biochemistry 35 (1996) 8319 – 8328. [17] L.M. Firsov, K.N. Neustroev, A.E. Aleshin, C.M. Metzler, D.E. Metzler, R.D. Scott, B. Stoffer, T. Christensen, B. Svensson, Eur. J. Biochem. 223 (1994) 293 – 302. [18] M.R. Sierks, K. Bock, K.S. Refn, B. Svensson, Biochemistry 31 (1992) 8972 – 8977. [19] T.P. Frandsen, B.B. Stoffer, M.M. Palcic, S. Hof, B. Svensson, J. Mol. Biol. 263 (1996) 79 – 89. [20] M.R. Sierks, B. Svensson, Biochemistry 39 (2000) 8585 – 8592. [21] J. Sauer, B.W. Sigurskjold, U. Christensen, T.P. Frandsen, E. Mirgorodskaya, M. Harrison, P. Roepstorff, B. Svensson, Biochim. Biophys. Acta 1543 (2000) 275 – 293. [22] T. Christensen, B. Svensson, B.W. Sigurskjold, Biochemistry 38 (1999) 6300 – 6310. [23] S.M.A. Khan, P.J. Reilly, C. Ford, Starch/Sta¨rke 52 (2000) 385 – 397. [24] J. Sauer, T. Christensen, T.P. Frandsen, E. Mirgorodskaya, K.A. McGuire, H. Driguez, P. Roepstorff, B.W. Sigurskjold, B. Svensson, Biochemistry 40 (2001) 9336 – 9346. [25] G. Williamson, N.J. Belshaw, T.R. Noel, S.G. Ring, M.P. Williamson, Eur. J. Biochem. 207 (1992) 661 – 670. [26] A.J. Clarke, B. Svensson, Carlsberg Res. Commun. 49 (1984) 111 – 122. [27] T.P. Frandsen, C. Dupont, J. Lehmbeck, B. Stoffer, M.R. Sierks, R.B. Honzatko, B. Svensson, Biochemistry 33 (1994) 13808 – 13816. [28] B. Stoffer, T.P. Frandsen, P.K. Busk, P. Schneider, I. Svendsen, B. Svensson, Biochem. J. 292 (1993) 197 – 202. [29] A.J. Wilkinson, A.R. Fersht, D.M. Blow, G. Winter, Biochemistry 22 (1983) 3581 – 3586. [30] T.P. Frandsen, T. Christensen, B. Stoffer, J. Lehmbeck, C. Dupont, R.B. Honzatko, B. Svensson, Biochemistry 34 (1995) 10162 – 10169. [31] T. Wiseman, S. Williston, J.F. Brandts, L.-N. Lin, Anal. Biochem. 179 (1989) 131 – 137. [32] D.R. Bundle, B.W. Sigurskjold, Methods Enzymol. 247 (1994) 288 – 305. [33] B.W. Sigurskjold, C.R. Berland, B. Svensson, Biochemistry 33 (1994) 10191 – 10199. [34] B. Svensson, T.P. Frandsen, I. Matsui, N. Juge, H.-P. Fierobe, B. Stoffer, K.W. Rodenburg, in: S.B. Petersen, B. Svensson, S. Pedersen (Eds.), Carbohydrate Bioengineering, Elsevier, Amsterdam, 1995, pp. 125 – 145. [35] J.M. Sa´nchez-Ruiz, Biophys. J. 61 (1992) 921 – 935. [36] E. Freire, Methods Enzymol. 259 (1995) 144 – 168. [37] T. Christensen, PhD Thesis, University of Copenhagen, 1999. [38] H.-P. Fierobe, E. Mirgorodskaya, T.P. Frandsen, P. Roepstorff, B. Svensson, Protein Expr. Purif. 9 (1997) 159 – 170. [39] B.W. Sigurskjold, T. Christensen, N. Payre, S. Cottaz, H. Driguez, B. Svensson, Biochemistry 37 (1998) 10446 – 10452. [40] N. Payre, S. Cottaz, C. Boisset, R. Borsali, B. Svensson, B. Henrissat, H. Driguez, Angew. Chem., Int. Ed. 38 (1999) 974 – 977. [41] C.R. Berland, B.W. Sigurskjold, B. Stoffer, T.P. Frandsen, B. Svensson, Biochemistry 34 (1995) 10153 – 10161. [42] O.B. Ptitsyn, V.N. Uversky, FEBS Lett. 341 (1994) 15 – 18. [43] O.B. Ptitsyn, Adv. Protein Chem. 47 (1995) 83 – 229.