Use of protein engineering to explore subunit interactions in an ...

Protein Engineering vol.11 no.7 pp.569–575, 1998

Use of protein engineering to explore subunit interactions in an allosteric enzyme: construction of inter-subunit hybrids in Clostridium symbiosum glutamate dehydrogenase

Suren Aghajanian and Paul C.Engel1 Department of Biochemistry, University College Dublin, Belfield, Dublin 4, Republic of Ireland 1To

whom correspondence should be addressed

Hybrids of different forms of clostridial glutamate dehydrogenase (GDH) have been constructed in order to probe the basis of allosteric interaction in this hexameric enzyme. It was shown that the C320S mutant, which is fully active and shows allosteric behaviour similar to that of the wildtype enzyme, can also be renatured after unfolding in urea. Mixtures of unfolded wild-type and C320S subunits gave rise to hybrids upon refolding. A purely random reassembly would lead to a simple binomial distribution. However there was a slightly better overall recovery of wild-type subunits and there appears to be a tendency for rapidly formed structured wild-type subunits in a mixture to nucleate further refolding in a way that biases the final distribution against the formation of C320S hexamers. Only the wild-type subunits in such hybrid mixtures are able to react with Ellman’s reagent, 5,59-dithiobis-(2-nitrobenzoate) (DTNB). Accordingly, after modification of hybrid hexamers with DTNB only the mutant subunits can bind NAD1. This permits fractionation on an NAD1– agarose affinity column. The elution pattern in itself indicates cooperativity since DTNB modification of just one subunit in a 1:5 wild-type/C320S hybrid largely abolished binding to the column. Kinetic studies were carried out on a fractionated preparation in which hexamers containing only one C320S subunit and five wild-type subunits were the predominant active species. Measurements of activity were made both before and after treatment with an excess of β-mercaptoethanol to remove the blocking thionitrobenzoate moieties. Before β-mercaptoethanol treatment this sample, with only one active subunit per hexamer, gave strictly hyperbolic (Michaelis–Menten) kinetics with NAD1 at pH 7.0, whereas after β-mercaptoethanol (all six subunits now active) the markedly kinked Eadie–Hofstee plot characteristic of wild-type enzyme was obtained. On the other hand the sigmoid response to glutamate at high pH persisted (Hill coefficient 5 3.6) even without β-mercaptoethanol, reflecting the fact that the inactive subunits can still bind glutamate. β-Mercaptoethanol treatment restored full positive cooperativity (Hill coefficient 5 5.2). These results prove beyond doubt that the non-classical kinetic behaviour of clostridial GDH is a direct result of interaction between NAD1 binding sites on the six (normally) identical subunits of a hexamer. Keywords: glutamate dehydrogenase/refolding/hybrids/chemical modification/cooperativity Introduction Glutamate dehydrogenases (GDHs, L-glutamate NAD(P)1 oxidoreductase, EC 1.4.1.2-4) are very widely distributed in © Oxford University Press

Nature, but the sustained interest they have attracted from enzymologists since the purification and crystallization of the first member of the family in 1952 (Olson and Anfinsen, 1952), is not merely a reflection of the metabolic importance of the reaction catalysed by GDH. Complex regulatory behaviour and striking departures from classical Michaelis–Menten kinetics have been described for GDHs from many different sources (Goldin and Frieden, 1971; Smith et al., 1975; Hudson and Daniel, 1993). Such behaviour in an enzyme is usually attributed to interaction between remote sites in the protein, mediated through conformational change (Monod et al., 1963, 1965; Koshland et al., 1966). In some cases, notably those of haemoglobin, aspartate transcarbamoylase and phosphofructokinase, X-ray crystallography has provided a detailed insight into the likely structural basis of such an interaction (Perutz, 1970; Kantrowitz and Lipscomb, 1988; Lipscomb, 1991; Schirmer and Evans, 1990; Evans, 1991). In the case of GDH, however, even though there is now a high-resolution structure available for the clostridial enzyme (Baker et al., 1992), there is still no direct proof of subunit interaction nor clear insight into the possible structural basis of any such interaction. All GDHs studied in detail so far have an oligomeric, usually hexameric, structure (Smith et al., 1975; Eisenberg et al., 1976; Gore, 1981; Hudson and Daniel, 1993), and clearly this potentially offers the possibility of interaction between subunits. In the case of the GDHs of vertebrates, however, there have been suggestions of several nucleotide binding sites per monomer (Fisher, 1970; Jallon and Iwatsubo, 1971; Krause et al., 1974; Pal and Colman, 1979; Hudson and Daniel, 1993) and there is also, therefore, the potential for allosteric interaction within a single subunit to account for at least part of the observed kinetic complexity. It must also be borne in mind that kinetic explanations of non-classical behaviour have been advanced that do not invoke subunit interaction (Ferdinand, 1966; Fisher, 1972). There is thus a continuing need for some direct experimental test for the basis of kinetic complexity in GDH. Bacterial GDHs lack the heterotropic responses to GTP and ADP found, for example, in GDH from vertebrates (Goldin and Frieden, 1971; Smith et al., 1975; Hudson and Daniel, 1993), but nevertheless do display non-classical kinetic behaviour. The NAD1-dependent GDH from Clostridium symbiosum, for example, has revealed complex behaviour in response to variation of NAD1 concentration (Syed et al., 1991). More recently strong positive cooperativity at high pH in the dependence of the enzyme reaction on glutamate concentration has been found (Wang and Engel, 1995). The construction of inter-subunit hybrids offers a method for distinguishing the intrinsic capabilities of a single subunit without drastically altering its environment and also for systematically investigating subunit interaction. Such an approach has been used to great effect, for example, with native and chemically modified subunits of aspartate transcarbamoylase (Lahue and Schachman, 1986; Yang and Schachman, 1987). 569

S.Aghajanian and P.C.Engel

For the Clostridium symbiosum GDH a number of mutant species of enzyme with substitution of residues involved in substrate binding (Wang and Engel, 1995; Wang et al., 1995), catalysis (Dean et al., 1995) or subunit interactions (Pasquo et al., 1996) have been constructed in our laboratory by sitedirected mutagenesis and characterized. This allows hybrid constructs to be envisaged in which subunits are disabled in various precise ways. In the presence of urea, hexameric molecules of clostridial GDH dissociate into denatured monomers (Aghajanian and Engel, 1994; Aghajanian et al., 1995) which refold and reassociate into the native hexameric structure upon dilution into 0.1 M potassium phosphate buffer, pH 7, containing 2 mM NAD1 (Aghajanian and Engel, 1994, 1997). These results have made possible the construction of hybrid hexamers which appear to be stable once formed, in the sense that there is no detectable subunit interchange (Aghajanian et al., 1995). Another important requirement for investigation of hybrids is accurate determination of subunit composition. Wild-type clostridial GDH contains just two cysteine residues in the entire polypeptide chain (Teller et al., 1992) and only one of these, Cys320, is accessible to thiol-specific reagents such as Ellman’s reagent, 5,59-dithiobis-(2-nitrobenzoate) (DTNB), or p-hydroxymercuribenzoate (p-HMB) (Syed et al., 1994). Modification with DTNB completely inactivates the enzyme. Although this can be prevented by high concentrations of NAD1 or NADH (Syed et al., 1994; Basso and Engel, 1994), the reactive residue Cys320 is essential neither for catalysis nor for binding of coenzymes or substrates, since the C320S mutant retains full catalytic activity with similar kinetic characteristics to those of the wild-type enzyme (Wang and Engel, 1994). Inactivation of wild-type clostridial GDH by DTNB appears to be a consequence of steric blocking of coenzyme binding by the bound thionitrobenzoate (TNB–) moiety (Wang and Engel, 1994). This mutant has made it possible first of all to count the wild-type subunits in a hybrid hexamer and also secondly to attempt a separation of hybrid mixtures by affinity chromatography. This paper reports details of the construction in vitro of inter-subunit hybrids involving wild-type and C320S mutant subunits of Clostridium symbiosum GDH. It also presents an investigation of the cooperative characteristics in a 5:1 (wildtype:C320S) hybrid in which, after DTNB modification, only one subunit per hexamer can bind the coenzyme. Some of this work has been briefly presented elsewhere (Aghajanian et al., 1996). Materials and methods Materials Grade II NAD1 (free acid), grade II NADH (disodium salt) and 2-oxoglutarate (disodium salt) were obtained from Boehringer Mannheim (UK). L-Glutamate (monosodium salt), Tris (Tris[hydroxymethyl]aminomethane) and NAD1–agarose resins (NAD1 immobilized to cross-linked 4% beaded agarose through the (i) N-6 and (ii) C-8 atoms of the adenine ring, and (iii) hydroxyl group of ribose), were purchased from Sigma Chemical Co. Sephadex DEAE A-50, Sephadex G-25 (fine), Thiopropyl Sepharose 6B and Activated Thiol Sepharose 4B were from Pharmacia Biotech Ltd. All other chemicals were analytical reagent grade and supplied by BDH Chemicals Ltd (UK) or Fisons Scientific Ltd (UK). 570

Enzyme preparation Escherichia coli strains TG1 and Q100 carrying respectively wild-type clostridial GDH (Teller et al., 1992) and C320S mutant (Wang and Engel, 1994) genes were kindly provided by Drs J.K.Teller and X.G.Wang. Wild-type and C320S transformants of E.coli were incubated at 37°C overnight in LB medium supplemented with 100 µg/ml ampicillin and 0.5 mM isopropyl β-thiogalactopyranoside (IPTG). Bacteria were harvested, suspended (1:10 w/v) in 0.1 M Tris–HCl buffer, pH 7.0, containing 1 mM EDTA, sonicated (Branson Sonicator 7532B) in short bursts with cooling in a salt-ice bath and centrifuged at 27 000 g for 30 min. Both enzymes were purified similarly by a single-step affinity-chromatography (Syed et al., 1991) modified as described elsewhere (Aghajanian et al., 1995). Specific activities of wild-type and C320S enzymes were 25 and 28 U/mg protein, respectively. The enzymes, stored in 60% saturated (NH4)2SO4 at 4°C, were desalted on a Sephadex G-25 (fine) column before use. Determination of protein concentration The concentrations of wild-type clostridial GDH and the C320S mutant were determined spectrophotometrically at 280 nm by using an absorption coefficient of 1.05 litre·g–1·cm–1 (Teller et al., 1992). In the case of DTNB-modified enzymes the concentrations were determined by the standard Bradford method (Bradford, 1976). Enzyme assays GDH activity was measured spectrophotometrically at 25°C (Uvikon 941 Plus, Kontron Instruments) by recording the change in A340 due to the production of coenzymes (NAD1 or NADH). Concentrations of substrates were 1 mM NAD1 and 40 mM L-glutamate (forward reaction) or 0.1 mM NADH, 10 mM 2-oxoglutarate and 50 mM NH4Cl (reverse reaction) in 0.1 M potassium phosphate buffer, pH 7.0, containing 1 mM EDTA. Denaturation and renaturation of clostridial GDH Denaturation of clostridial GDH by 4 M urea was carried out in 0.1 M potassium phosphate buffer, pH 7.0, containing 1 mM EDTA and 1 mM β-mercaptoethanol at 20°C (Aghajanian et al., 1995). Enzyme activity after 60 min incubation was less than 0.2%. Denatured enzyme was reactivated by 50-fold dilution into the same buffer containing 2 mM NAD1 and incubation at room temperature over 2 days (Aghajanian and Engel, 1997). DTNB modification of enzyme samples and recovery of activity after modification Enzyme samples (wild-type, mutant or hybrid mixture) were modified with 2 mM DTNB in 0.1 M potassium phosphate buffer, pH 7.0. After 30 min incubation at 20°C, conditions previously shown to be sufficient for complete inactivation of wild-type GDH (Basso and Engel, 1994; Syed et al., 1994), modified samples were separated from the low molecular weight compounds by gel-filtration on a Sephadex G-25 (fine) column equilibrated with 50 mM potassium phosphate buffer, pH 7.0, containing 0.5 mM EDTA. To recover the catalytic activity of wild-type subunits after DTNB-modification, the samples were diluted 5-fold into 0.1 M potassium phosphate buffer, pH 7.0, containing 5 mM β-mercaptoethanol and were incubated for 5–10 min at 20°C. This procedure releases the TNB– residues from the enzyme with quantitative recovery of catalytic activity (Aghajanian et al., 1995).

Construction and use of hybrid hexamers in GDH

Table I. Calculated theoretical values of hybrid hexamers with defined composition formed from randomly distributed subunits of wild-type (() and mutant (s) enzymes depending on mixing ratiosa Hybridsb

n

Percentage of hybrid hexamers with n wild-type subunits for the samples obtained by mixing wild-type and mutant enzymes in the ratios: 0:6

1:5

2:4

3:3

4:2

5:1

6:0

0.

sss sss

100

33.49

8.78

1.56

0.14

0.002

0

1.

sss ss( sss s(( sss ((( ss( (((

0

40.19

26.34

9.38

1.65

0.064

0

0

20.09

32.92

23.44

8.23

0.804

0

0

5.36

21.95

31.25

21.95

5.36

0

0

0.804

8.23

23.44

32.92

20.09

0

0

0.064

1.65

9.38

26.34

40.19

0

0

0.002

0.14

1.56

8.78

33.49

100

0

1/6

1/3

1/2

2/3

5/6

1

2. 3. 4.

s(( ((( ((( ((( p(i) 5

5. 6.

an is the number of wild-type subunits and p(i) is their probability of incorporation in a hybrid hexamer for the mixing ratio i:(6 – i) of two enzymes. Details of the calculation are given in the Materials and methods. bThe symbols illustrating the hybrids indicate only the subunit stoichiometry and not the precise arrangement within a hexamer.

Calculation of probability of hybrid formation with defined composition of subunits Probabilities (Table I) of formation of hybrid hexamers with n incorporated wild-type subunits, depending on the mixing ratio of wild-type (i parts) and mutant (6 – i parts) enzymes, and based on a standard binomial distribution, were calculated according to Equation (1) (Kempthorne and Folks, 1971):

{

Pn(i) 5 Pn–1(i) 3 P0(i) 5

qN(i)

p(i) q(i)

3

N – (n – 1) n

3 100;

; Nùn.0 (1) n50

where: N 5 6 for the hexamer; p(i) 5 1 – q(i). p(i) and q(i) are the probabilities of incorporation of wild-type and mutant subunits respectively with the mixing ratios of enzymes i:(6 – i). Results and discussion Construction of inter-subunit hybrids by refolding of mixed wild-type and C320S mutant species of clostridial GDH Comparative studies of urea-induced denaturation and subsequent renaturation of wild-type enzyme and the C320S mutant revealed no major differences in behaviour, although the rate of reactivation for the C320S mutant was slightly slower. Thus in a typical refolding experiment under identical conditions half-lives of reactivation were 29.7 and 40.2 min for the wild-type and C320S enzymes, respectively (Aghajanian and Engel, 1997). On the simplest assumption, if the reactivation processes are generally similar for both enzymes, the recombination of subunits should be a random process following a binomial distribution. Estimated theoretical yields of hybrids of defined composition formed from randomly distrib-

uted wild-type and mutant subunits, depending on the mixing ratios of the two enzymes, were calculated according to Equation (1) (see Materials and methods) and are given in Table I. The calculated values are independent of the route of association of monomers into hexamer. The Table provides useful guidance in defining the ratios in which wild-type and mutant enzymes must be mixed in order to obtain reasonable amounts of a hybrid with any particular defined subunit composition. Except for mixing ratios exceeding 5:1, not more than 40% of any one hybrid ratio is predicted for any of the refolding mixtures. It must be emphasised however that in the best case these predictions can only be approximate, and they rely on two assumptions which may not be entirely valid, namely that the different subunit types refold with similar speed and efficiency and that their reassociation is entirely random. For the construction of inter-subunit hybrids, the following procedure was used: wild-type clostridial GDH and the C320S mutant (1.5 mg/ml each), after individual denaturation in urea over 60 min, were mixed in different volume ratios (wildtype/C320S 5 6:0, 5:1, 4:2, 3:3 etc.) and renatured as described in Materials and methods. The reactivated samples were concentrated and separated from NAD1 by ion-exchange chromatography on DEAE Sephadex A-50. Samples were eluted with 0.5 M NaCl in 0.1 M potassium phosphate buffer, pH 7.0. Determination of partial concentrations of wild-type and C320S mutant enzymes in refolded samples Refolded samples were modified with 2 mM DTNB (see Materials and methods) and subjected to gel-filtration to remove free thionitrobenzoate (TNB–) and excess DTNB. The normalized absorption spectra of DTNB-modified samples are shown in Figure 1. As expected, the samples containing wildtype subunits showed absorption spectra characteristic for bound TNB–, with a peak at 336 nm. The intensity of the peak increased in proportion to the fraction of wild-type enzyme in the initial mixture. To estimate the actual partial concentrations 571

S.Aghajanian and P.C.Engel

two species in the final refolded material. The results show a small departure from the simple agreement expected if both the two species were refolded with equal efficiency. Thus in all the refolded mixtures there was a slight tendency for overrepresentation of wild-type subunits. For example an equal mixture (3:3) gave back a refolded preparation containing 53% wild-type subunits and 47% C320S. This presumably reflects the small differences in refolding efficiencies mentioned above. The spectrophotometric method presented allows us, however, to define the precise subunit ratio in the refolded samples.

Fig. 1. Normalized absorption spectra of refolded samples of clostridial GDH with different wild-type and C320S mixing ratios after DTNBmodification. Wild-type and C320S mutant enzymes (1.5 mg/ml) after ureadenaturation were mixed in different volume ratios (wild-type/C320S ratios are indicated). Samples were refolded, purified and modified with DTNB as described in the text. Absorption spectra were normalized to A280 5 1.

of wild-type and mutant subunits in refolded samples from the absorption spectra shown in Figure 1 the following procedure was used. Since bound TNB– would contribute to the total UV absorption of modified wild-type enzyme, the protein concentrations of DTNB-modified samples of pure wild-type (sample 6:0) and pure C320S (sample 0:6) with the same optical density at 280 nm were determined by the Bradford method (Bradford, 1976). The concentration of sample 6:0 was 10 6 0.5% less than the concentration of sample 0:6. This result allowed us to estimate the absorption coefficients at 280 and 336 nm for the wild-type enzyme fully modified with DTNB. Assuming a value for the pure C320S enzyme at 280 nm (ε C320S 280 ) the same as for unmodified wild-type enzyme (1.05 litre·g–1·cm–1) (Teller et al., 1992) the value for the ) must be 10% DTNB-modified wild-type enzyme (ε mod-WT 280 greater i.e. 1.155 litre·g–1·cm–1. Thus concentrations of samples 6:0 and 0:6 with one optical unit absorption at 280 nm are 0.86 and 0.95 mg/ml, respectively. From these data and the absorption spectrum of modified wild-type enzyme (sample 6:0, Figure 1) with the relative absorption peak at 336 nm ∆A336 5 0.138, the absorption coefficient of modified wildtype enzyme at 336 nm (ε mod-WT ) can be calculated as 0.16 336 litre·g–1·cm–1 (0.138/0.86). Now the concentrations of wildtype (xi) and C320S (y6–i) subunits incorporated into reactivated samples obtained by the mixing of i parts of wild-type and 6 – i parts of C320S mutant can be calculated from the absorption spectra of modified samples (Figure 1) by using absorption coefficients obtained above: ; xi 5 ∆A336(i) / ε mod-WT 336 y6–i 5 [∆A280(i) – xi 3

ε mod-WT ] 280

(2) /

ε C320S 280 ;

(3)

where ∆A336(i) and ∆A280(i) are the relative absorptions at 336 and 280 nm of the samples i:(6 – i) obtained by the mixing of i parts of wild-type and 6 – i parts of C320S mutant enzymes. Table II shows the application of this approach to calculating the subunit composition of refolded samples. In each case the two components of the refolding mixture were present in a known initial ratio. The spectrophotometric calculations allowed estimation of the representation of the 572

Affinity chromatography of refolded samples An attempt was made to separate different hybrid components of refolded samples by affinity chromatography first of all by using the reactive thiol group present in wild-type subunits but missing in C320S. Wild-type enzyme was loaded onto activated Thiol Sepharose 4B or Thiopropyl Sepharose 6B affinity columns (both equilibrated with 50 mM potassium phosphate buffer, pH 7.0), in the hope that the enzyme might bind to the resins by a thiol-disulfide interchange reaction (Brocklehurst et al., 1974). However when equilibration buffer was applied after 60 min all loaded protein was washed off the columns. This result could be explained by the location of Cys320 deep in the active site cleft of enzyme (Baker et al., 1992) and thus perhaps inaccessible to the reactive groups of the affinity resins. As an alternative, affinity chromatography on NAD1–agarose was explored. It has been shown elsewhere (Aghajanian et al., 1995) that DTNB-modified wild-type clostridial GDH, in contrast to unmodified enzyme, does not bind to NAD1– agarose in consequence of the blocking of the coenzyme binding site. Three NAD1 affinity resins were examined, in which NAD1 was attached to agarose respectively through (i) N6 and (ii) C8 positions of the adenine ring and (iii) through the hydroxyl groups of ribose. The first two resins showed similar affinity for the native clostridial enzyme, but binding to the third was very poor. Bound enzyme was eluted from the first two resins with linear gradients of either NAD1(0–2 mM) or NaCl (0–1 M). There were no remarkable differences in protein elution profiles on using a gradient of NaCl rather than NAD1(data not shown). On the other hand, use of NAD1 impeded spectrophotometric detection of proteins because of its high absorbancy. Thus in all further experiments for the separation of constructed hybrids the NAD1–agarose affinity column with an attachment through the N6 position of NAD1 was used with elution by NaCl. Refolded samples (200 µl; previously modified with DTNB) with protein concentrations 1 mg/ml each in 50 mM potassium phosphate buffer, pH 7.0, were loaded separately on an NAD1– agarose column (6 ml) equilibrated with the same buffer. The elution profiles of samples with varying ratios of the two components are shown in Figure 2. Only the sample 0:6 (pure C320S mutant enzyme with 100% activity after DTNB treatment) was bound on the NAD1–agarose and eluted from the column after applying a salt gradient. All other samples eluted from the column without binding tightly, although the shapes of elution profiles, and the size and position of protein peaks depended on the nature of the loaded samples. Since the peak corresponding to C320S hexamer is missing from most of the elution profiles it can clearly be concluded that the refolded samples were not simple mixtures of pure wildtype and pure C320S mutant hexamers but rather inter-subunit hybrids between the two subunit species. The fact that the

Construction and use of hybrid hexamers in GDH

Table II. Estimated partial concentrations of wild-type and C320S mutant enzymes and percentage of incorporation of wild-type enzyme into the reactivated samples obtained by mixing of denatured enzymes in different ratiosa Refolded samples with the ratios of wild-type:C320S 0:6 1:5 2:4 3:3 4:2 5:1 6:0

A336

0.004 0.030 0.057 0.081 0.103 0.124 0.142

Partial concentration of enzymes (mg/ml)

Percentage of incorporated wild-type subunits

Wild-type

C320S

Estimated

Expected

0 0.16 0.33 0.48 0.62 0.75 0.86

0.95 0.77 0.59 0.42 0.27 0.13 0

0 17 36 53 70 85 100

0 16.7 33.3 50 66.7 83.3 100

aThe

values were estimated from the normalized absorption spectra of samples (Figure 1) with the use of Equations (2) and (3). The absorption coefficients used were 1.155 litre·g–1·cm–1 at 280 nm and 0.16 litre·g–1·cm–1 at 336 nm for the wild-type enzyme and 1.05 litre·g–1·cm–1 at 280 nm for the C320S mutant (see the text for details).

Fig. 2. Affinity chromatography of refolded samples of clostridial GDH with different wild-type and C320S mixing ratios on an NAD1–agarose column after DTNB-modification. 200 µl of DTNB-modified refolded samples with protein concentrations 1 mg/ml each in 50 mM potassium phosphate buffer, pH 7.0, were separately loaded on an NAD1–agarose column. In each case elution of bound enzyme was carried out with a linear gradient of NaCl (0–1 M) in the same buffer. The samples with different wild-type/C320S mixing ratios are indicated.

sample obtained by mixing wild-type and mutant enzymes in 1:5 ratio (with approximately 85% of residual activity after DTNB modification) eluted from the affinity column virtually without binding to it suggests a cooperative interaction between subunits in hybrid hexamers since inactivation of an average of one subunit in a hexamer almost abolished the affinity of the other five subunits for the immobilized coenzyme. This interesting effect unfortunately limits the resolution of the column in separating different hybrids. For all the reactivated samples, 1 ml fractions were collected and activities were determined before and after treatment with β-mercaptoethanol (Figure 3). Incubation of DTNB-modified samples for 5–10 min with 5 mM β-mercaptoethanol removed practically all TNB– residues from the wild-type enzyme with almost 100% regeneration of catalytic activity (Aghajanian et al., 1995). Relative activities for each fraction were calculated as a ratio of activities before and after treatment with βmercaptoethanol. This ratio indicates relative content of wildtype and mutant enzymes in each fraction. The results in Figure 3 (open squares) show an increase of relative activity

Fig. 3. Results of activity analysis of fractions showed in Figure 2. Activities of fractions (1 ml each) were measured before (s) and after (d) treatment with 5 mM β-mercaptoethanol. Relative activities (u) were expressed as a ratios of activities of fractions before and after treatment. Treatment of fractions with β-mercaptoethanol was carried out as described in Materials and methods.

with increasing fraction number, indicating differing extents of interaction of NAD1–agarose with distinct hybrids present within each refolded mixture. Thus, for example in the case of the 1:5 sample the relative activity climbs from approximately 0.5 in fraction 4 to 0.73 in fraction 7 (reasonably close to the theoretical value of 0.69 for pure 2:4 hybrid) and then to a fairly constant value of about 0.83 over the next 10 fractions (very close to the predicted value of 0.85 for pure 573

S.Aghajanian and P.C.Engel

1:5 hybrid). Finally in a separate small peak comprising fractions 23–25 a ratio close to 1.0 is seen, as expected for pure 0:6 i.e. C320S hexamer. Thus, in spite of the cooperative effect mentioned above, it is still possible to obtain fractions in which a single hybrid species predominates. Surprisingly small amounts of pure hexameric C320S were found in samples obtained by refolding mixtures of wild-type and mutant enzymes in the ratios of 1:5 or 2:4 (Figures 2 and 3). These were not more than 5% of the total protein, in comparison with the theoretical values of 33.5 and 8.8% calculated for random refolding and reassembly (Table I). This suggests that formation of hybrids between wild-type clostridial GDH and the C320S mutant is not entirely random and cannot be accurately described by the binomial distribution. Taking into consideration two other observations discussed above: (a) incorporation of more wild-type than C320S mutant subunits into reactivated samples (Table II); (b) observed minor differences in refolding of the separate pure wild-type and C320S enzymes, we can suggest that, after dilution of denatured wildtype and mutant enzymes into the reactivation mixture, rapidly formed structured wild-type subunits behave as nuclei promoting formation of correctly folded mutant subunits and their further association into the active hexameric structure. Investigation of cooperative effects in clostridial GDH Cooperative interaction of the subunits affecting enzyme– coenzyme affinity during chromatography on an NAD1– agarose column has already been shown above for the sample 1:5 (Figures 2 and 3). The fact that inactivation of only one subunit in the hexamer can alter the coenzyme affinity of the other five subunits suggests that all six subunits of the hexamer are simultaneously involved in cooperativity, although a possibility of stronger interaction within dimers or trimers is not ruled out. In response to variation of NAD1 concentration at pH 7 wild-type clostridial GDH displays a remarkable departure from classical Michaelis–Menten kinetics (Syed et al., 1991) with a strikingly abrupt transition in the slope of the Lineweaver–Burk plot which has been attributed to allosteric interaction. Moreover, at pH 9, where the enzyme appears inactive under standard assay conditions (1 mM NAD1, 40 mM L-glutamate) (Syed and Engel, 1990), it shows a steep sigmoidal rise of activity in the range of L-glutamate concentrations higher than 100 mM, with a Hill coefficient equal to 5.4 (Wang and Engel, 1995). Analogous experiments with the C320S mutant showed cooperative behaviour similar to that of the wild-type enzyme (data not shown). These properties of clostridial GDH were examined for the hybrid containing only one active and five inactive subunits. For these experiments, fraction 6 (in Figure 3, shown by the arrow) of the DTNB-modified 5:1 sample was used. The predicted relative activities with and without DTNB treatment for pure 6:0, 5:1 and 4:2 wild-type:C320S hybrid hexamers are 0, 0.18 and 0.36. The measured figures for relative activity before and after β-mercaptoethanol treatment of fractions 4–9 in the elution of the 5:1 refolding mixture (Figure 3) are 0.13, 0.11, 0.15, 0.19, 0.20 and 0.31 respectively. The ratio for fraction 9 suggests a predominance of the 4:2 hybrid and the low ratios in fractions 4 and 5 indicate that these are mainly mixtures of 6:0 and 5:1 hybrids. The ratios for fractions 6–8 suggest that in these fractions 5:1 hybrid is the predominant species. Fraction 6 probably contains a slightly higher contamination with pure 6:0, but the choice of this fraction for the experiments was deliberately made on the basis that, compared 574

Fig. 4. Eadie–Hofstee plots of activity with varied NAD1 concentration for fraction 6 of refolded sample 5:1. Activities were measured before (s) and after (d) treatment with β-mercaptoethanol. Enzyme concentrations in assay solution were 2.3 µg/ml (s) and 0.35 µg/ml (d).

Fig. 5. (A) Activity of fraction 6 of refolded sample 5:1 as a function of L-glutamate concentration at pH 8.8. (B) The Hill plots for the same data. Activities were measured in 0.1 M Tris–HCl buffer, pH 8.8, with 1 mM NAD1, before (s) and after (d) treatment of enzyme with βmercaptoethanol. Enzyme concentrations in assay solution were 2.3 µg/ml (s) and 0.35 µg/ml (d).

with subsequent fractions, it contained relatively fewer hybrids with more than one active C320S subunit in the hexamer. Obviously, the presence of pure wild-type hexamers in a fraction would not affect the results of such experiments because they are fully inactive owing to DTNB modification. A few experimental data obtained using fraction 5 were qualitatively practically identical to those obtained for fraction 6 (data not shown). As a control, the same fraction 6, after treatment with 5 mM β-mercaptoethanol and following

Construction and use of hybrid hexamers in GDH

separation from low molecular weight components on Sephadex G-25 (fine), was used. Eadie–Hofstee plots of the kinetics with NAD1 for fraction 6 before and after treatment with β-mercaptoethanol are shown in Figure 4. In contrast with fully active enzyme, the sample with only one active subunit per hexamer shows classical Michaelis–Menten kinetics with NAD1, suggesting that ‘nonlinear’ kinetic behaviour of the fully active enzyme is the result of the interaction of active subunits in the hexamer of clostridial GDH. Activities of the same fraction 6 before and after βmercaptoethanol treatment were measured at pH 8.8 as a function of L-glutamate concentration (Figure 5A). The sample with only one active subunit per hexamer, like the fully active enzyme, showed a sigmoidal rise of activity, but with a Hill coefficient of 3.6, as compared with 5.2 for the enzyme with six active subunits (Figure 5B). In the hybrid molecules examined in this study, only one subunit (the single C320S subunit) was able to bind NAD1 after modification with DTNB, and accordingly no interaction in NAD1 binding was evident in the kinetics. Evidently there is no contribution from any unsuspected second binding site within a single subunit. On the other hand there is no reason to believe that the DTNB modification significantly compromises glutamate binding. Thus all six subunits in the hybrid can presumably still bind glutamate. Accordingly, even though catalytic activity is expressed in only one subunit, that activity reflects the cooperativity of glutamate binding shown by active and inactive subunits alike. It appears that the hybridization procedure described here has the potential to dissect the allosteric interactions in clostridial GDH. The availability of subunit types disabled in other ways than by blockage of coenzyme binding may allow precise definition of the basis of subunit interaction.

Jallon,J.-M. and Iwatsubo,M. (1971) Biochem. Biophys. Res. Comm., 45, 964–971. Kantrowitz,E.R. and Lipscomb,W.N. (1988) Science, 241, 669–674. Kempthorne,O. and Folks,L. (1971) Probability, Statistics, and Data Analysis. Iowa State University Press, Ames, Iowa, USA. Koshland,D.E.,Jr, Nemethy,G. and Filmer,D. (1966) Biochemistry, 5, 365–385. Krause,J., Bu¨hner,M. and Sund,H. (1974) Eur. J. Biochem., 41, 593–602. Lahue,R.S. and Schachman,H.K. (1986) J. Biol. Chem., 261, 3079–3084. Lipscomb,W.N. (1991) Chemtracts, 2, 1–15. Monod,J., Changeux,J.P. and Jacob,F. (1963) J. Mol. Biol., 6, 306–329. Monod,J., Wyman,J. and Changeux,J.P. (1965) J. Mol. Biol., 12, 88–118. Olson,J.A. and Anfinsen,C.B. (1952) J. Biol. Chem., 197, 67–79. Pal,P.K. and Colman,R.F. (1979) Biochemistry, 18, 838–845. Pasquo,A., Britton,L.K., Stillman,T.J., Rice,D.W., Co¨lfen,H., Harding,S.E., Scandurra,R. and Engel,P.C. (1996) Biochim. Biophys. Acta, 1297, 149–158. Perutz,M.F. (1970) Nature, 228, 726–739. Schirmer,T. and Evans,P.R. (1990) Nature, 343, 140–145. Smith,E.L., Austin,B.M., Blumenthal,K.M. and Nyc,J.F. (1975) In Boyer,P.D. (ed.), The Enzymes. Academic Press, New York, USA, Vol.11, pp. 293–367. Syed,S.E.-H. and Engel,P.C. (1990) Biochem. J., 271, 351–355. Syed,S.E.-H., Engel,P.C. and Parker,D.M. (1991) Biochim. Biophys. Acta, 1115, 123–130. Syed,S.E.-H., Hornby,D.P., Brown,P.E., Fitton,J.E. and Engel,P.C. (1994) Biochem. J., 298, 107–113. Teller,J.K., Smith,R.J., McPherson,M.J., Engel,P.C. and Guest,J.R. (1992) Eur. J. Biochem., 206, 151–159. Wang,X.G. and Engel,P.C. (1994) Protein Engng, 7, 1013–1016. Wang,X.G. and Engel,P.C. (1995) Biochemistry, 34, 11417–11422. Wang,X.G., Britton,K.L., Baker,P.J., Martin,S., Rice,D.W. and Engel,P.C. (1995) Protein Engng, 8, 147–152. Yang,Y.R. and Schachman,H.K. (1987) Anal. Biochem., 163, 188–195. Received August 28, 1997; revised February 20, 1998; accepted March 2, 1998

Acknowledgements This work was supported by the former Agriculture and Food Research Council (UK) through a Linked Institution Grant and latterly by the Biotechnology and Biological Sciences Research Council (UK). The work was initiated at the Krebs Institute for Biomolecular Research, University of Sheffield (UK) and completed in Dublin.

References Aghajanian,S.A. and Engel,P.C. (1994) Biochem. Soc. Trans., 22, 318S. Aghajanian,S. and Engel,P.C. (1997) Biochem. J., 326, 649–655. Aghajanian,S.A., Martin,S.R. and Engel,P.C. (1995) Biochem. J., 311, 905–910. Aghajanian,S.A., Wang,X.G. and Engel,P.C. (1996) Biochem. Soc. Trans., 24, 136S. Baker,P.J., Britton,K.L., Engel,P.C., Farrants,G.W., Lilley,K.S., Rice,D.W. and Stillman,T.J. (1992) Proteins Struct. Funct. Genet., 12, 75–86. Basso,L.A. and Engel,P.C. (1994) Biochim. Biophys. Acta, 1209, 222–226. Bradford,M.M. (1976) Anal. Biochem., 72, 248–254. Brocklehurst,K., Carlsson,J., Kierstan,M.P.J. and Crook,E.M. (1974) Meth. Enzymol., 34, 531–544. Dean,J.L.E., Wang,X.G., Teller,J.K., Waugh,M.L., Britton,L.K., Baker,P.J., Stillman,T.J., Martin,S.R., Rice,D.W. and Engel,P.C. (1994) Biochem. J., 301, 13–16. Eisenberg,H., Josephs,R. and Reisler,E. (1976) Adv. Protein Chem., 30, 101–181. Evans,P.R. (1991) Curr. Opin. Struc. Biol., 1, 773–779. Ferdinand,W. (1966) Biochem. J., 98, 278–283. Fisher,H.F. (1970) Adv. Enzymol., 39, 369–417. Fisher,J.R. (1972) Arch. Biochem. Biophys., 152, 638–645. Goldin,B.R. and Frieden,C. (1971) Curr. Top. Cell. Regul., 4, 77–117. Gore,M.G. (1981) Int. J. Biochem., 13, 879–886. Hudson,R.C. and Daniel,R.M. (1993) Comp. Biochem. Physiol., 106B, 767– 792.

575