Gorga, F. R. & Lienhard, G. E. (1981) Biochemistry 20,5108-5113. Gorga, F. R. ... Morris, H. R., Allard, W. J., Lienhard, G. E. & Lodish, H. F.. (1985) Science 229 ...
Eur. J. Biochem. 22I, 513-522 (1994) 0 FEBS 1994
The kinetics and thermodynamics of the binding of cytochalasin B to sugar transporters Adrian R. WALMSLEY', Allan G. LOWE' and Peter J. F. HENDERSON'
'
Krebs Institute for Biomolecular Research, Department of Molecular Biology and Biotechnology, University of Sheffield, England Department of Biochemistry and Molecular Biology, The Medical School, University of Manchester, England Department of Biochemistry and Molecular Biology, University of Leeds, England
(Received September 7November 24, 1993)
-
EJB 93 1358/6
The kinetics of the binding of cytochalasin B to the proton-linked L-arabinose (AraE) and Dgalactose (GalP) symporters from Escherichia coli and to the human erythrocyte glucose transporter (GLUTl) have been investigated by exploiting the changes in protein fluorescence that occur upon binding the ligand. Steady-state measurements yielded l ( d values of 1.1, 1.9 and 0.14 pM for the AraE, GalP and GLUT1 proteins, respectively. The association and dissociation rate constants for the binding of cytochalasin B have been determined by stopped-flow spectroscopy. In each case, the apparent Kd was calculated from the corresponding rate constants, yielding values of 1.5, 0.4 and 1.6 pM for AraE, GalP and GLUTl, respectively. The differences between these apparent Kd values and those measured by fluorescence titration is interpreted in terms of the following three step mechanism where CB represents cytochalasin B : Scheme 1
CB The transporter is proposed to alternate between two different conformational forms (TI and T,), with cytochalasin B binding only to the T, conformation, to induce a further conformational transition of the transporter to the T, form. The values for the overall dissociation constants show that the TI conformation is favoured by AraE and GalP in the absence of ligands, but the T, conformation is favoured by GLUTl. Thus, the binding of cytochalasin B to GLUTl alters the equilibrium towards the T,(CB) conformational state, producing the observed tight binding, in contrast to the changes in the equilibrium observed with the binding of cytochalasin B to AraE and GalP. A thermodynamic analysis of these conformational transitions has been performed. The T, and T, conformations may represent transporter states in which the binding site is facing outwards and inwards, respectively.
A number of sugar transporters, from organisms as diverse as bacteria and man, have been cloned and sequenced. Several mammalian passive glucose transporters have been identified (Gould and Bell, 1990; Carmthers, 1990; Silverman, 1991). A transporter from the human hepatoma HepG2 cell line (Mueckler et al., 1985), and one from rat brain (Birnbaum et al., 1986), are very similar, if not identical in amino acid sequence, to the extensively studied human erythrocyte transporter (Baly and Horuk, 1988). Four other transporters, including one found in all tissues (Kayano Correspondence to A. R. Walrnsley, Krebs Institute for Biomolecular Research, Department of Molecular Biology and Biotechnology, University of Sheffield, P. 0. Box 594, Firth Court, Western Bank, Sheffield, England S10 2UH Fax: +44 742 728691. Abbreviation. GalP, the D-galactose-H' symporter of Escherichia coli; AraEi, the L-arabinose-H' symporter of E. coli; GLUTl , the human erythrocyte glucose transporter; CB, cytochalasin B ; T,, conformational form of transporter.
et al., 1988), one from liver (Thorens et al., 1988), one from small intestine (Kayano et al., 1990), and the insulinregulatable transporter from adipose tissue (James et al., 1989) have been sequenced and shown to be distinct, since they have 40- 65 % identity with the erythroid-type transporter. More recently, a microsomal transporter from the endoplasmic reticulum of liver has been sequenced and shown to be distinct from the plasma-membrane transporter, with which it has 68% identity (Waddell et al., 1992). All these mammalian glucose transporters have approximately 30% identity with the D-galactose, L-arabinose and D-xylose proton-linked transporters found in Escherichia coti, which can also catalyse the passive translocation of sugars (Maiden et al., 1987: Baldwin and Henderson, 1989: Henderson, 1990; Henderson and Maiden, 1990: Henderson et al., 1992). Taking into account conservative substitutions, the different mammalian and bacterial transporters have a similar degree of similarity with one another. Indeed, it has now become apparent that both the mammalian and bacterial transporters
514 belong to a superfamily, which includes sugar transporters from mammals, plants, yeasts, fungi, algae, protozoa, cyanobacteria and eubacteria, and also several bacterial antibiotic transporters (Griffith et al., 1992). A consequence of the similarity of the amino acid sequences of the various sugar transporters, is that they are all thought to have a common membrane topology, comprising 12 membrane-spanning a-helices, with helices 6 and 7 connected by a cytoplasmic hydrophilic domain composed of 60-70 amino acids (Griffith et al., 1992). A common structure is further indicated by the sensitivity of the erythrocyte glucose transporter and the bacterial D-galactose and L-arabinose proton symporters to cytochalasin B (Baldwin et al., 1980; Baldwin and Lienhard, 1981; Shanahan, 1982; Gorga and Lienhard, 1981, 1982; Walmsley, 1988; Carruthers and Helgerson, 1991 ; Helgerson and Carruthers, 1987, 1991 ; Carruthers, 1990; Henderson and Maiden, 1990; Cairns et al., 1991; King et al., 1991; Burant and Bell, 1992). Considering the structural similarities between these different sugar transporters, it would seem likely that they also share a common mechanism of operation, with differences attributable to variations in the specificities for sugar substrates and cations. In this study, we have investigated the kinetics and thermodynamics of the binding of cytochalasin B (CB) to three members of this superfamily of homologous sugar transporters : the glucose transporter (GLUT1) from erythrocytes, and the D-galactose (GalP) and L-arabinose (AraE) proton-linked symporters from E. coli. The aim of this study was to probe both the mechanism of cytochalasin B binding, about which little is known, and to determine the basis of the differences in the cytochalasin-B-binding affinities of these transport proteins. It was possible to address these aims by using a stopped-flow spectroscopic technique to monitor the changes in protein fluorescence upon the binding of cytochalasin B by the substantially over-expressed or purified proteins (Walmsley et al., 1993). We demonstrate that in each case the kinetics of the binding of cytochalasin B is consistent with at least a three-step mechanism, in which the ligand-free transporter alternates between two different conformations, to only one of which the cytochalasin B binds, inducing a further conformational change in the transporters. As such, the data presented in this study support the alternating conformation model (Walmsley, 1988). The difference in affinities of GLUTl and GalP (Cairns et al., 1991) for cytochalasin B is attributable to differences in the equilibrium constants for the two isomerisation steps, rather than any difference in the initial interaction. The thermodynamic parameters for the binding of cytochalasin B were determined by varying the temperature. The energetics of the interaction and the nature of the resultant conformational changes are discussed.
MATERIALS AND METHODS Growth of the AraE and GalP over-producing E. coli strains The E. coli strain AR120 (pMM27) was used for the over-expression of the AraE protein (Maiden, 1986; Maiden et al., 1988). The E. coli strain JMllOO (pPER) was used for the over-expression of the GalP protein (Roberts, 1992). E. coli strains AR120 (pMM27) and JMllOO (pPER3) were grown as described by Walmsley et al. (1993) and Roberts (1992), respectively.
Preparation of inner membranes containing over-expressed protein E. coli cells were disrupted by explosive decompression in a French press at 137.5 MPa, and inner membranes were prepared essentially as described by Osborn et al. (1972). This procedure yields predominantly inside-out vesicles (Futai, 1978). They were suspended in 5OmM sodium phosphate, 150 mM NaCl, 1 mM EDTA, pH 7.4.
Quantification of over-expressed protein A sample of each membrane preparation (30 pg protein) was subjected to SDSRAGE. After staining with Coomassie blue the gels were scanned with a Molecular Dynamics 100A computing densitometer and the proportion of transport protein was measured in each track.
Preparation of erythrocyte glucose transporters The preparation of alkali-washed membranes (pH 12) and the subsequent purification and reconstitution of GLUTl were performed as described by Baldwin et al. (1982).
Protein assays The concentration of protein in the AraE and GalP membrane preparations was assayed by the method of Schaffner and Weissman (1973). The concentration of GLUTl was determined by the method of Bradford (1976).
Fluorescence studies Fluorescence titrations were performed in a Jasco FP777 spectrofluorimeter, operated at the indicated temperatures. The protein was excited at 280nm and the fluorescence quenching was monitored at 333 nm, using excitation and emission band widths of 3 nm and 10 nm, respectively. This excitation wavelength was chosen since it provides the maximal signal change for all the transporters. Titrations were performed by making additions (pl) of concentrated cytochalasin B (in ethanol) to 0.5 ml vesicles in a 1-cm-path-length cuvette. These additions led to less than a 10% dilution of the samples. The protein and antibiotic concentrations were corrected for these small dilution effects. Rapid reactions were followed using an Applied-Photophysics (London, UK) spectrofluorimeter, operated at 20°C as described by Walmsley et al. (1993). The kinetics of the binding of cytochalasin B to the AraE and GalP symporters from E. coli, and the kinetics of binding of cytochalasin B to GLUTl were determined by mixing vesicular preparations of the respective transport proteins with cytochalasin B in the stopped-flow device, and monitoring the resultant changes in protein fluorescence. In the case of AraE and GalP, vesicles were prepared from strains of E. coli that over-produce these proteins. The level of membrane enrichment of AraE and GalP was estimated by densitometry of Coomassie-blue-stained SDS/PAGE gels to be 24% and 52% of the total membrane protein, respectively. Such high levels of AraE and GalP in these membrane preparations have enabled us to monitor directly changes in the tryptophan fluorescence of the protein upon binding cytochalasin B. In the case of GLUTl, we have largely been restricted to using purified, reconstituted, transporters (Baldwin et al., 1982). Only small fluorescence changes were observed when alka-
515 line-washed (pH 12) membranes, prepared from red-cell ghosts, were used for stopped-flow fluorescence studies ; GLUTl only constitutes approximately 5 % of the erythrocyte membrane protein in such preparations (Allard and Lienhard, 1985). Consequently, all the studies were performed with purified transporters, unless stated otherwise. In order to establish that changes in protein fluorescence, resulting from cytochalasin B binding to the vesicles, were not simply due to changes in light scattering, the absorbance at 340 nm and 600 nm was monitored for vesicles mixed with up to 50 pM cytochalasin B. No significant changes in the absorbance have been noted. Similarly, the effect of sugars upon the light-scattering properties of the vesicles were investigated. For sugar (mixing chamber) concentrations below 150mM there was little change in either the absorbance at 340 nm or 600 nm and only a slight increase in the protein fluorescence.
Assay conditions Unless stated otherwise, fluorescence measurements were performed in 5 0 m M potassium phosphate, 1OOmM NaCl and 1 mM EDTA, pH 7.4, at the indicated temperature. Routinely, the membranes were used at protein concentrations of 50, 25 and 8 pg/ml of AraE, GalP and GLUT1, respectively, in this buffer. Five or more concentrations of cytochalasin B were used in determining the association rate constants.
t
A
l-----7
LL
I
I
.
2
2
Ti me (s)
Fig. 1. The binding of cytochalasin B to subcellular vesicles. Stopped-flow records are presented showing the quenching of protein fluorescence (excitation = 280 nm, emission > 335 nm) upon the high-affinity binding of ( 5 pM) cytochalasin B to subcellular vesicles. The vesicles were as follows: (A) inside-out vesicles prepared from a strain of E. coli, AR120 (pMM27), that over-expresses the AraE protein (100 pg/ml membrane protein); (B) inside-out vesicles prepared from a strain of E. coli, JMllOO (pPER3), that overexpresses the GalP protein (50 kg/ml membrane protein); (C) leaky vesicles, prepared from erythrocyte membrane lipids, in which the purified glucose transporter (GLUTl) was reconstituted (50 pg/ml membrane protein); (D) alkaline-washed erythrocyte membranes (1 mg/ml membrane protein). Each division on the ordinate axes represents a fluorescence change of 1.25%.
Data analysis Data were routinely analysed using the non-linear leastsquares regression program supplied with the Applied Photophysics SE17MV stopped-flow equipment.
Determination of thermodynamic parameters The activation energies for the measured rate constants were determined by fitting the values obtained for these constants at various temperatures directly to the Arrhenius equation by non-linear regression. The enthalpies for specific mechanistic steps were determined either by linear-regression analysis of a van't Hoff plot of the values obtained for the equilibrium (or dissociation) constants, or from the difference in the activation energies for the forward and reverse rate constants for each step. Gibbs-free-energy changes were calculated from the measured equilibrium constants at 20 "C. The entropy change was calculated from standard thermodynamic relationships :
AGO= -RTlnK AGO= AH"-TAS". Thermodynamic parameters for the formation of the conformational transition states of the proteins were calculated by applying the following relationships derived from transition-state theory:
E,
=
AHf
+- R T ,
-
kT
-
h
AG"
=
p(ASf/R)
AHf-TAS'
. p-- ( A H f / R n ,
where E, is the activation energy for a process which occurs with rate constant k , k, is Boltzmann's constant, h is Planck's constant, R is the gas constant, T is the absolute temperature
and A G f , A H + , and A S f are the standard Gibbs free energy, enthalpy and entropy changes, respectively, associated with formation of the transition state.
RESULTS The changes in the equilibrium fluorescence of the transporters induced by cytochalasin B at 20°C The binding of cytochalasin B to AraE, GalP, pH 12 membranes and purified, reconstituted GLUTl, were all characterized by a quench in the protein fluorescence (excitation wavelength = 280 nm, emission wavelength = 333 nm), which amounted to 7-9, 14-18, 3-5 and 7-9%, respectively, for the binding of 50 pM cytochalasin B to each protein, respectively. When the proteins were excited at 297 nm, to preferentially excite tryptophan residues, the fluorescence quench was reduced to 3-5, 12-16, 1-2 and 3-5%, for AraE, GalP, pH 12 membranes and purified GLUT1, respectively. These values represent the ranges for at least five individual measurements for each protein. In each case, the maximum fluorescence emission- wavelength was .at 331 335 nm, and the binding of cytochalasin B did not cause any significant shift in the wavelength of the maximum fluorescence.
The resolution of the time course for the binding of cytochalasin B to the transporters at 20°C The time courses of the binding of cytochalasin B to AraE, GalP, pH 12 membranes and purified, reconstituted GLUTl were resolved by stopped-flow fluorescence spectroscopy. Typical data for the time course of the binding of 5 pM cytochalasin B to each of the transport proteins are given in Fig. 1. All the traces were biphasic and better fits of
indicated a lower fluorescence quenching than measurements performed in the fluorirneter (i.e, the time-resolved and equilibrium measurements yielded values of 7.6% and 14% quenching in protein fluorescence, respectively, for the binding of 5 pM cytochalasin B (excitation wavelength = 280nm). However, D-glucose and D-galactose failed to reverse completely the quenching due to the binding of cytochalasin B at equilibrium (in the fluonmeter), suggesting that this discrepancy may be due, at least partially, to nonspecific binding. When the binding of cytochalasin B was followed by specifically monitoring the changes in tryptophan fluorescence (excitation wavelength = 297 nm, emis2 4 6 8 sion > 335 nm) essentially the same results were obtained, Time (s) except that the signal amplitude was reduced for each of the transport proteins. That the observed fluorescence changes were attibutable to the specific binding of cytochalasin B to the respective transport proteins was established from measurements of the inhibition of the fluorescence change by different sugar isomers. The quenching of AraE fluorescence by cytochalasin B was inhibited by L-arabinose but not D-arabinose, a nontransported isomer of the natural substrate L-arabinose Fig. 2. Analysis of the time course of the binding of cytochalasin (Walmsley et al., 1993). Similarly, D-glucose and D-galactose B to GalP. Stopped-flow record over an extended time base of 10 s, were inhibitors of the quenching of GalP and GLUTl fluowhich shows the quenching of protein fluorescence upon the binding rescence by cytochalasin B, but neither L-glucose nor L-gaof cytochalasin B (5 pM) to GalP (50 pg/ml membrane protein). lactose was an inhibitor (Walmsley, A. R. and Henderson, P. Superimposed upon the record are the best fits obtained by non- J. F., unpublished results). Moreover, when vesicles, which linear regression analyses of the data in terms of (A) single-expo- were not enriched in either AraE or GalP, were mixed with nential and (B) double-exponential functions. For the double-expo- cytochalasin B, no significant change in protein fluorescence nential fit of the data, the rapid phase accounted for 75.4% of the was observed (Walmsley et al., 1993; Walmsley, A. R. and decrease in fluorescence and had a first-order rate constant of Henderson, P. J. F., unpublished results). 22.9 s-'; the slow phase accounted for 24.6% of the decrease in fluorescence and had a first-order rate constant of 2.8 s-'. The total quenching of fluorescence was 9.0%. The inset shows the first second of the stopped-flow record, and the fit for a single-exponential Comparisons of the kinetics of the binding function is super-imposed upon the data. The lower panels (A, B), of cytochalasin B to the sugar transporters at 20°C show the residual variance of the data about the best fit single exponential (A) and double exponential (B). The residual variance is For each of the proteins, the rate constant for the binding magnified sevenfold in B relative to A. Note the systematic devia- of cytochalasin B was determined over a range of cytochation of the residual variance about the best fit in A, indicating that lasin B concentrations. This was achieved by performing a the function provides an inadequate fit of the data. The residual series of experiments similar to those shown in Fig. 1, and variance in B shows only random scatter about the best fit. determining the rate of the early phase of the change in protein fluorescence by fitting to a double-exponential function. In each case, the rate constant for the early phase of binding the data were obtained for a double-exponential function, increased linearly with the cytochalasin B concentration rather than a single-exponential function. In the case of AraE (Fig. 3 ) and this behaviour can be interpreted in terms of a and GLUTl (pH 12 membranes and purified, reconstituted simple one-step binding process : transporters), the biphasic nature was only slight and the lCBl k2 second slow phase amounted to only 10-13% of the total T, CB 7 TZ (CB) Scheme 1. k-2 signal amplitude, which occurred with a rate constant of 2-4 s-'. In the case of GalP, the biphasic nature was more pronounced, with the second slow phase constituting approx- where T, is the transport protein and CB is cytochalasin B. The observed rate constant (kobs)is approximated by the imately 25% of the total fluorescence change (Fig. 2). This slow phase occurred with a rate constant of 3 s-' and was following function under pseudo-first-order conditions : independent of the cytochalasin B concentration, indicative kobs = h[CBl + k-2, (1) of a slow isomerisation of the transporter-cytochalasin B complex. Indeed, the slow phase for any of the transporters, where k, and k-, are the rate constants for association and cannot be attributed to a second population of transporters dissociation, respectively. The initial conformation of the that bind cytochalasin B, otherwise the rate of this phase transporter is termed T, in order to be consistent with the would have changed with the cytochalasin B concentration. schemes developed below for GalP and GLUTl (e.g. There was no change in the rate of this phase with changing Scheme 1). The dissociation rate constants for cytochalasin B were cytochalasin B concentrations for any of the transporters (data not shown). In each case this phase must be due to a measured directly by displacing cytochalasin B, bound to cytochalasin-B-induced isomensation of the transporter. each transporter, with sugar. Each transporter was previously Stopped-flow measurements of the change in protein fluores- equilibrated with 10 pM cytochalasin B and mixed with cence, arising from the binding of cytochalasin B to GalP, 200 mM sugar in the stopped-flow apparatus to displace the
+
9
517
1
ID
B 80
3
[cytochalasin
B]
6
9
1
Table 1. Kinetic parameters for the binding of cytochalasin B to AraE, GalP and GLUTl at 20°C. The association rate constants (k,) were calculated from the linear plots of the binding rate versus the cytochalasin B concentration given in Fig. 3. The dissociation rate constants (k-,) were determined by sugar displacement of cytochalasin B. The dissociation constants for cytochalasin B (K,) were calculated from the measured association and dissociation rate constants (Kz = k-2/kz). The overall dissociation constants for cytochalasin B (Kd)were determined by titration of the protein fluorescence. Transporter
k,
AraE GalP GLUT1
6.0 ( 2 0.23) 9.0 ( 2 0.38) 1S O ( 20.09) 1.lo ( 2 0.39) 6.1 ( 2 0.14) 2.5 (2 0.20) 0.41 (? 0.03) 1.90 ( 2 0.60) 2.4 ( 2 0.10) 3.9 ( 2 0.16) 1.63 ( 2 0.10) 0.14 (? 0.01)
k-2
K2
Kd
2
(pM)
Fig. 3. The concentration dependence of the rate of association of cytochalasin B to (A) AraE, (B) GalP, (C) reconstituted GLUTl and (D) alkaline-washed (pH 12) erythrocyte membranes. Linear-regression analyses of each data set yielded the following values for the apparent association and dissociation rate constants and Kd: (A) k,,, = 6.0 pM-' . s-', koff= 7.4 s - I , Kd = 1.2 pM; (B) k,, = 6.1 pM-' . s-', k,, was too small for accurate measurement; (C) k,, = 2.4 pM-' . s-', k,, = 4.6 s-', K, = 1.9 pM; (D) k,, = 1.7 pM-' . S-', k,, = 2.7 s-', K, = 1.6 pM.
bound cytochalasin B. This displacement was monitored as an increase in the protein fluorescence (Walmsley et al., 1993). For AraE, L-arabinose was used as the displacing sugar; for GalP and GLUT1, D-glucose was used. When vesicles were mixed with sugars, in the absence of cytochalasin B, there was no slow change in the protein fluorescence that could be resolved by stopped-flow fluorimetry, for any of the transporters. Hence, the displacement of bound cytochalasin B from the transporters, by sugars, was observed as an increase in protein fluorescence. It has previously been reported that the binding of D-glucose to purified, reconstituted, GLUTl induces quenching of the protein fluorescence (Gorga and Lienhard, 1982; Carruthers, 1986; Pawagi and Deber, 1990; Chin et al., 1992), but this quenching is too rapid for direct measurement by stopped-flow fluorimetry (Appleman and Lienhard, 1985, 1989) and the reaction is essentially complete within the dead-time of the instrument. In contrast, the binding of cytochalasin B to GLUTl induces a much larger quenching of the protein fluorescence than Dglucose (Gorga and Lienhard, 1982; Carruthers, 1986; Pawagi and Deber, 1990; Chin et al., 1992), and occurs at a sufficiently slow rate that it can be monitored by stoppedflow fluorimetry. Hence, the displacement of GLUTl -bound cytochalasin B by D-glucose is observed as an increase in the protein fluorescence that can be monitored by stopped-flow fluorimetry. However, the signal change for alkaline-washed membranes was too small to allow accurate determination of the cytochalasin B dissociation rate constant for GLUTl by this method, although it was possible with the purified, reconstituted, transporter. In each case, the increase in protein fluorescence resulting from the displacement of bound cytochalasin B by the sugar could be fitted to a single-exponential function. The rate constants for the dissociation of cytochalasin B from AraE, GalP and purified GLUTl are given in Table 1 (k-J, and are shown not to differ significantly from those obtained by extrapolation of the cytochalasin B concentration dependence of the apparent binding rate constant (Fig. 3 and Table
1). It might be expected that since the association process was biphasic that the dissociation processes should also be biphasic. A possible explanation for this difference is that the apparently monophasic dissociation of cytochalasin B from the transporters might reflect the fact that this occurs at a similar rate to the slow conformational change. When the rate constants differ by less than a factor of three it is difficult to distinguish between the two processes, especially when one of the phases is of small amplitude (Walmsley and Bagshaw, 1989). In the case of GalP, which has the largest amplitude for the slow phase, the dissociation and isomerisation rate constants have similar values up to approximately 20°C, but thereafter the difference becomes greater with increasing temperature. However, the amplitude of the slow phase decreases with increasing temperature (data not shown), so that the data for the dissociation of cytochalasin B at these higher temperatures could well appear monophasic. The concentration of cytochalasin B used in sugar displacement experiments was sufficient to achieve over 80% saturation of the transporters (Table 1). Consequently, most of the transporters will be sequestered in the inward-facing conformation. Under these conditions, attainment of equilibrium for the transporter-sugar complexes will be rate limited by the dissociation of cytochalasin B, since for each transporter the sugar has direct access to this face of the membrane. In the cases of AraE and GalP, the inside-out vesicles are reported to be sealed (Futai, 1978), so that the sugar can only displace cytochalasin B by binding initially to the inward-facing conformation that is exposed to the outside of the vesicles. In the case of GLUTl, the vesicles are reported to be right-side out but leaky (Appleman and Lienhard, 1985), allowing D-glucose to compete with cytochalasin B at both faces of the membrane. Under these circumstances, D-glucose could displace cytochalasin B by binding to the outward-facing or the inward-facing site. However, the rate constant measured for cytochalasin B displacement was slower than those previous obtained for-reorientation of the transporter (Appleman and Lienhard, 1985, 1989; Lowe and Walmsley, 1986), indicating that it is a true measure of the cytochalasin B dissociation rate constant, rather than including a component of sugar accessibility. The values for the association (k2) and dissociation (k-,) rate constants and the calculated dissociation constants (k2/ kz) are given in Table 1. AraE and GalP were characterised by similar association rate constants but cytochalasin B dis-
518 In the case of GalP, there is no evidence for a secondary conformational change in the cytochalasin B complexes [i.e. no T3(CB) state] since K2 is less than the Kd for this transporter. As a first approximation we can therefore assume that 11 K3 is zero, and K , can then be calculated as 3.63 for GalP. This is in fact a minimum value for K, since if 1 / K , has a finite value K, will have a higher value than 3.63. In any event, it is evident that at 20°C the outward-facing conformation predominates for GalP. In the case of AraE, K2 had a slightly higher value than Kd in the present study, but in a previous study this situation was reversed (K2 = 0.72 pM and Kd = 0.80-1.76 pM; Measurement of the overall K , for cytochalasin B Walmsley et al., 1993). This might suggest that the inwardby titration of the protein fluorescence at 20°C facing conformation of AraE predominates and that formaThe overall dissociation constants (Kd) for cytochalasin tion of T,(CB) is negligible. However, K, and Kq could have B were determined for AraE,GalP and purified, reconstituted similar values. The above model also provides a partial explanation for GLUT1 by titration of the protein fluorescence (excitation = 280 nm). The quenching of protein fluorescence was deter- the biphasic natures of the changes in fluorescence after mined for a range of cytochalasin B concentrations, from binding of cytochalasin B. In the case of GLUTl, the early approximately 0.2X Kd to 1OX Kd in each case. The data were phase probably involves binding to the 80% of transporters fitted to an equation describing the titration curve for a sec- initially present in the T, conformation (88% total quenching ond-order binding process : in fluorescence), while the later phase (12% total quenching) involves both additional binding to transporters undergoing the (relatively slow) transition from T, to TI, and any change FIFO= (dF/'T1) . {(Kd+ [TI [CB]) associated with reorientation of T,(CB) to T,(CB). Fo In the case of GalP, the initial rapid quenching of fluorescence is presumably associated with formation of T2(CB) where T and CB represent the transporter and cytochalasin from those carriers in the T, conformation (20% total for B, F, is the initial fluorescence, AF is the total fluorescence GalP). The late phase of fluorescence involves binding of change, and Kd is the dissociation constant. Such an analysis cytochalasin B to transporters which were initially in the T, yielded the K,, values given in Table 1, indicating the form and must undergo the slow TI to T, transition before following order of affinities for cytochalasin B : purified binding. The fact that the amplitude of the late quenching of fluorescence is less than could be predicted from the initial GLUT1 > AraE > GalP. The apparent discrepancies between the values of K2 and proportions of T, and T, presumably indicates that the T, to Kd for GLUTl, AraE and GalP (Table 1) can be explained T, transition itself involves an increase in protein fluorescence which partially masks the decrease caused by the bindby the following model : ing of cytochalasin B. The inward facing form of GLUTl kl k2 LCBI k3 has been reported to have a greater intrinsic fluorescence T2 (CB) T, (CB) , Scheme 1. than the outward facing form (Appleman and Lienhard, Ti 7 T, k-2 1989). in which T, and T, are the outward-facing and inward-facing It is notable that there is also a second slow phase of forms of the unloaded transporters, respectively, and T,(CB) fluorescence change when cytochalasin B binds to AraE, and T,(CB) are two different conformations of the transpor- suggesting that a proportion of the transporters are in the ter-cytochalasin B complex. The inward and outward orien- outward facing conformation. If this is the case, then a protations are defined with respect to the plasma-membrane of portion of the transporters must adopt the T,(CB) conformathe cell. For example, in the case of AraE and GalP, the tion. However, the amplitude of the late fluorescence change outward-facing conformation is that in which the binding site is less for AraE than GalP, suggesting that before cytochaof the transporter faces the outside of the cell or the inside lasin B binds a greater proportion of transporters are in the of the (inside-out) vesicles. For such a model the apparent T, conformation for GalP than for AraE. dissociation constant of the transporter -cytochalasin B complex is given by the following function:
sociated from GalP at a much slower rate than from AraE. The association rate constant for purified GLUT1 was approximately 2-3 fold slower than from AraE or GalP. Both the association and dissociation rate constants for cytochalasin B binding to alkaline-washed red cell membranes were slightly slower than for reconstituted transporters, indicating that purification with octylglucoside has only a modest effect upon the kinetics of the binding of cytochalasin B to GLUT1 (Fig. 3).
+
-
-
A
7
(3) with K,, K2 and K3 defined as K, = k - , / k , , K, = k-,/k2 and K 3 = k-,/k,. In the case of GLUTl, k, and k - , (and hence K , ) are known (Lowe and Walmsley, 1987; Appleman and Lienhard, 1989) and so K3 can be calculated as 0.074 by substituting K, (k, = 600 s-', k-, = 148 s-' and Kl = 0.247 at 20°C; Lowe and Walmsley, 1987) and the measured values of K,, and K2 into Eqn (3). Thus at equilibrium (20"C), the GLUTlcytochalasin B complex is present as 8% T,(CB) and 92% Tz(CB).
The temperature dependence of the cytochalasin B association rate constant The cytochalasin B association rate constants for AraE, GalP and GLUTl were determined at 3-42"C, so that the thermodynamics of the binding of cytochalasin B could be analysed. Fig. 4 is a set of Arrhenius plots showing the variation in the asssociation rate constant with temperature for each of the proteins; the thermodynamic parameters calculated from each plot are presented in Table 2. All the Arrhenius plots were linear, with no changes in slope that might be attributable to membrane-phase changes, suggesting that binding is a single-step process in each case.
:F\ I
519
Al
0 1
3N
2
1
0
1
'm
34
32
103 T
36
0
m
32
36
34
lo3 (K-1) T
(K-')
Fig. 4. Arrhenius plots of the temperature dependencies of the second-order association rate constants (Icon) for the binding of cytochalasin B to (A) AraE, (B) GalP and (C) purified GLUT1. The activation energies for k,, (k2 in Scheme 1) were calculated by a direct fit of the data to the Arrhenius equation by non-linear leastsquares regression. This analysis yielded activation energies of 36.7 ( 5 0.5) kJ . mol-', 58.8 ( 5 3.3) kJ . mol-' and 36.5 (t0.8) kJ . mol-' for AraE, GalP and GLUTI, respectively.
Fig.5. Arrhenius plots of the temperature dependencies of the dissociation rate constants (kom)for the (A) AraE, (B) GalP and (C) purified GLUTl cytochalasin B complexes. The dissociation rate constants were measured either by sugar displacement (H) or by extrapolation of the data for the concentration dependence of the observed rate of binding of cytochalasin B (0).Where both types of measurement were made, the data were combined for analysis. The activation energies for k,, (k2 in Scheme 1) were calculated by a direct fit of the data to the Arrhenius equation by non-linear leastsquares regression. This analysis yielded activation energies of 43.0 (f1.81) kJ . mol-', 59.3 ( 21.0) kJ . mol-' and 58.8 ( ? 1.5) kJ . rno1-I for AraE, GalP and GLUTl, respectively.
The temperature dependence of the rate constant for dissociation of transporter-cytochalasin B complexes For each transporter, the rate constant for cytochalasin B dissociation was determined by sugar displacement at a series of temperatures. The resulting data gave linear Arrhenius plots for all three transporters, and the activation energies were determined directly from fits of the data to the Arrhenius equation (Fig. 5). In the cases of AraE and GLUTl, the dissociation rate constants were fast enough to be measured independently by extrapolation of the cytochalasin-B-binding data, and this data is also presented in Fig. 5. The two procedures gave broadly comparable dissociation rate constants for the cytochalasin B complexes of GalP and AraE across the full temperature range studied.
The temperature dependence of the slow isomerisation associated with formation of the GalP-cytochalasin B complex (k,) When membrane vesicles containing GalP were mixed with 16 pM cytochalasin B the initial rapid change in fluorescence was followed by a slow change occurring over a period of 2.5-10 s, depending on the temperature of the experiment. The rate constant for the slower process was obtained, at a series of temperatures, by fitting the data to a double-exponential expression, and Fig. 6 illustrates the temperature dependence of the rate constant for the slow change
Table 2. Thermodynamic parameters for the interaction of cytochalasin B with the sugar transporters. Transition
Entropy ( A S ) for
Enthalpy ( A H ) for AraE
GalP
GLUT1
-
-17.4 -
-17.4
GalP
GLUTl
J . mol-I . K-'
k J . mol-' T, to T, T, to T,(CB) T,(CB) to T,(CB) Overall
AraE
-5.1 -0.5 -
-6.6
-46.0 -22.3 +16.5 -51.8
-
+55.6 -
+55.6
-
37.9
+ 120.5 + 87.0
- 145.0
+
+ -
34.8 64.5 45.7
520
t
0.2
103 (K-1)
B
-0.5
T
Fig. 6. Arrhenius plot of the temperature dependence of the rate constant for the slow isomerisation (kiJ of GalP (k, in Scheme 1). The activation energy for k, was calculated by a direct fit of the data to the Arrhenius equation by non-linear least-squares regression, yielding a value of 49.1 ( r 0 . 6 ) kJ . mol-I.
-3.51
,
33
,
,
34
,
I 35
Fig. 8. van't Hoff plots of the temperature dependencies of the dissociation constants (Kd values) for the complexes formed between cytochalasin B and (A) GalP and (B) purified GLUT1. The enthalpy changes associated with the binding of cytochalasin B were determined by linear regression. This analysis yielded enthalpies of -6.6kJ . rnol-' for GalP and of -51.8 kJ . mol-' for GLUTl. -2
34
52
T 103
36
(K")
Fig. 7. van%Hoff plot of the temperature dependence of 1/K, for AraE. This constant ( l / K 2 )provides an approximate measure of the overall Kd for the AraE-cytochalasin B complex. The enthalpy change associated with the binding of cytochalasin B was determined by linear regression, yielding a value of -17.4 kJ . rnol-'.
in fluorescence in the form of an Arrhenius plot. This process is slower, and has a much smaller activation energy than sugar-induced dissociation of cytochalasin B from GalP, and as discussed above is probably attributable to the slow isomerization of GalP from the TI to T2 conformation with subsequent binding of cytochalasin B to this initially 'hidden' pool of GalP.
The temperature dependence of the overall Kd for the dissociation of cytochalasin B from the transporters The overall dissociation constants (Kd values) for the AraE-cytochalasin B, GalP-cytochalasin B and GLUTl catochalasin B complexes were determined by titration of the protein fluorescence, in a fluorimeter, at four different temperatures (15-30°C). In the case of AraE, the Kd had little dependence upon the temperature in the studied range, with similar values to K2 (data not shown), whilst for GalP there was a slight increase in the Kd with increasing temperature (Fig. 8A, Table 2), indicating that the binding process is slightly exothermic. For purified GLUTl , there was more pronounced dependence of Kd on temperature indicating a more exothermic binding process (Fig. 8B, Table 2). Since for GLUT1, K , , K , and their temperature dependencies are known it is possible to calculate the enthalpy of isomerization of T,(CB) to T,(CB) from the following relationship: dH(overal1) = dH(1)
The temperature dependence of the dissociation constants for the T,(CB) complexes The dissociation constants (K2) for the cytochalasin B complexes of AraE, GalP and GLUTl were calculated from the measured association and dissociation rate constants obtained over a range of temperatures. The K2 value for AraE and GalP had little dependence on temperature, with K,(AraE) increasing from approximately 0.5 pM to 1.5 pM at 3-42°C (Fig. 7) and K,(GalP) remaining almost constant in this temperature range. Calculation of the enthalpies of binding from the activation energies associated with rate constants k, and k-, indicated that the binding of cytochalasin B is slightly exothermic for both AraE and GalP. Binding of cytochalasin B to GLUTl showed a similar temperature dependence to AraE and GalP (with K2 values of 1.5-3 pM at 3-42°C). However, calculation of the enthalpy of binding of cytochalasin B to GLUTl from the activation energies for k2 and k-, indicated that in this case the binding process is moderately exothermic (AH" = -22 kT . mol-').
+ dH(2j + d H ( 3 ) .
(4) This gives a value of +16.5 kJ . mol-' for dH(3). Similarly, dS(3) can be calculated as -45.7 J . K-' . mol-'. If the assumption is made that there is no isomerization of the GalP-cytochalasin B complexes [T,(CB) to T,(CB)], then dH(1) and dS(1) can be calculated in an analogous manner (Table 2).
DISCUSSION The kinetics and thermodynamics of the binding of cytochalasin B to the proton-linked arabinose and galactose symporters of E. coli and the human erythrocyte glucose transporter have been determined. The overall affinity of these sugar transporters for cytochalasin B (measured at equilibrium) decreases in the order GLUTl > AraE > GalP, with about an order of magnitude difference in the Kd values between GLUTl and GalP, but the order of affinities of the initial formed complexes with cytochalasin B (measured by stopped-flow fluorimetry) decreases in the reverse order.
521 This apparent paradox can be accounted for by the following model : CB . kl Q"(k2 . 4 T, p - l Tz T. Tz (CB) T, (CB)
7
CB
'-2
7
Scheme 1.
in which the equilibrium binding of cytochalasin B depends on isomerizations of the 'unloaded' (T,/T,) and 'loaded' [T,(CB)/T,(CB)] forms of the transporters in addition to the primary binding process. In the case of GalP, the high Kd for equilibrium binding relative to the dissociation constant for primary binding ( K J can be accounted for if the equilibrium between T, and T, is such that a substantial proportion of transporters are in the T, state and there is a negligible proportion of T,(CB). However, the relatively small Kd for GLUTl can be accounted for if the equilibrium between T,(CB) and T,(CB) is substantially in favour of the latter and the majority of unloaded GLUTl molecules are in the T, state. The near equivalence of K2 and Kd for AraE suggests that most of the transporters are in the T, state and formation of the T,(CB) state is negligible. The relationship between the model outlined above and the transport processes catalysed by AraE, GalP and GLUTl is not known with certainty, but it is likely that the T, and T, conformations represent those states of the alternating conformation model in which the sugar-binding site is respectively outward facing and inward facing. In the case of GLUTl, these two conformations of the transporter have been shown to be in dynamic equilibrium (model reviewed by Baldwin and Lienhard, 1981; Walmsley, 1988; Carruthers, 1990; Silverman, 1991) with the inward-facing conformation predominating at low temperatures (Lowe and Walmsley, 1986, 1987; Walmsley and Lowe, 1987; Walmsley, 1988; Appleman and Lienhard, 1989), and analysis of the competitive inhibitory effects of cytochalasin B and the externally-acting inhibitor phloretin (Deves and Krupka, 1978 ; Gorga and Lienhard, 1981) has demonstrated that cytochalasin B binds to the inward-facing (T,) form of GLUTl. The kinetic data reported in this study are fully consistent with this model. The fact that the T,(CB) to T,(CB) transition of GLUTl is accompanied by increases in entropy and enthalpy that are smaller than for the change from T, to T, is of considerable interest. One possible explanation is that the T,(CB) to T,(CB) transition is similar to, but less complete than, the T, to T, transition, with the transporter going part of the way to reorientating from the inward-facing to outward-facing conformation. This would agree with the hypothesis that GLUTl and cytochalasin B can form a transition-state complex which is analogous to, but much more stable than, the transition state formed between GLUTl and glucose during transport. The rate of interconversion of the T,(CB) and T,(CB) forms is more than two orders of magnitude slower than the rate of reorientation of the unloaded transporter at 20 "C (Lowe and Walmsley, 1986), which suggests the presence of a state in which the cytochalasin B is temporarily occluded from the aqueous medium. Moreover, the displacent of cytochalasin B from GLUTl by D-glucose appears monophasic. This might be attributable to the dissociation of cytochalasin B from T,(CB), and the isomerisation, between T,(CB) and T,(CB), which are processes occurring at similar rates. However, we may have failed to observe dissociation of the cytochalasin B from the small T, population, via a slow, ratelimiting, interconversion of T,(CB) to T,(CB), if this oc-
curred over a time-scale of several minutes. In support of the latter argument, Gorga and Lienhard (1981), measuring the rate of dissociation of rH1cytochalasin B from GLUTl by filtration at 3-4"C, have reported a rate constant that is an order of magnitude slower than that obtained in the present study. The rate constants measured for GLUTl in the present study would then only be for the dissociation of cytochalasin B from the T,(CB) conformation. King et al. (1991) have also concluded, on the basis of the ability of thermolysin to digest the human erythrocyte transporter, that cytochalasin B induces a different conformational state of the transporter to that induced by the binding of sugars to the internal or external facing site. They have proposed that cytochalasin B 'locks' the transporter in a conformation that prevents the transporter from adopting either an inward-facing or outward-facing conformation with sugars. In the cases of GalP and AraE, there is no independent evidence to determine whether cytochalasin B binds to the the inward-facing or outward-facing conformations. However, in view of the amino acid similarities between the three transporters, it seems reasonable to assume that cytochalasin B again binds to the inward-facing T, conformations of AraE and GalP. If this is the case, the analysis presented in the Results section indicates that for both AraE and GalP the thermal equilibrium between TI and T, lies more in favour of the outward-facing T, conformation than is the case for GLUTl. It is also notable that the entropy and enthalpy changes for the T, to T, transitions of AraE and GalP are smaller than those previously reported for GLUTl (Walmsley and Lowe, 1987; Walmsley, 1988; Appleman and Lienhard, 1989). This may indicate that for AraE and GalP the change from T, to T, involves a smaller increase in hydration than for GLUT1. Formation of the initial complexes [T,(CB)] of AraE and GalP with cytochalasin B are entropically driven, in that for both transporters the enthalpy term does not contribute much to the free energy of binding (Table 2). In contrast, formation of the corresponding GLUTl complex is associated with a moderate increase in entropy and decrease in enthalpy. These changes in entropy and enthalpy give some insight into the processes involved in cytochalasin B binding. In the cases of AraE and GalP, the increases in entropy could arise from displacement of water from the transporter-binding sites by cytochalasin B, and release of 'structured' water associated with the hydrophobic cytochalasin B in aqueous solution. Furthermore, the small change in enthalpy for binding of cytochalasin B to AraE and GalP could result from the enthalpies of (hydrogen) bond formation between cytochalasin B and the transporters being balanced by the breaking of similar numbers of (hydrogen) bonds between the transporters and water molecules displaced from the binding sites. In the case of GLUTl, the smaller increase in entropy and larger decrease in enthalpy for cytochalasin B binding may indicate that binding of cytochalasin B to GLUTl displaces less water than is the case for the two bacterial transporters. The kinetic model, developed in the present study, of the binding of cytochalasin B to sugar transporters that have similar amino acid sequences, and which are presumably structurally similar, will be of immense value as a basis for future studies. It should be possible to determine the effects of imposing a pH gradient or membrane potential upon the kinetics and thermodynamics of cytochalasin B binding to the proton-linked L-arabinose and D-galactose symporters, and to assess how this differs from binding to passive transporters, such as GLUT1. Moreover, the application of the fluores-
cence techniques developed in this study to mutants generated in AraE, GalP and GLUT1, will enable us to determine the energetics of the interaction of cytochalasin B with individual amino acid residues, and to assess which are involved in stabilising the various conformational states. This study was supported by research grants from the Wellcome Trust (to A. R. W., P. J. F. H. and A. G. L.) the Science and Engineering Research Council (to P. J. F. H.) and the Medical Research Council (to A. R. W. and P. J. F. H.) and by equipment grants from the University of Sheffield (to A. R. W.) and the Royal Society (to A. R. W.). The Krebs Institute is a designated centre for molecular recognition studies and is supported by the Science and Engineering Research Council. We are grateful to Dr V. A. Lucas, Dr N. G. Rutherford, and Mr R. Choolun for isolating and characterizing the inner-membrane preparations from E. coli.
REFERENCES Allard, W. J. & Lienhard, G. E. (1985) J. Biol. Chem. 260, 86688675. Appleman, J. R. & Lienhard, G. E. (1985) J. Biol. Chem. 260, 4575 -4578. Appleman, J. R. & Lienhard, G. E. (1989) Biochemistry 28, 82218227. Baldwin, J. M., Lienhard, G. E. & Baldwin, S. A. (1980) Biochim. Biophys. Acta 599, 699-714. Baldwin, S. A,, Baldwin, J. M. & Lienhard, G. E. (1 982) Biochemistry 21, 3836-3842. Baldwin, S. A. & Henderson, P. J. F. (1989)Annu. Rev. Physiol. 51, 459-71. Baldwin, S. A. & Lienhard, G. E. (1981) Trends Bzochem. Sci. 6, 208-21 1. Baly, D. L. & Horuk, R. (1988) Biochim. Biophys. Acta 947, 571 590. Birnbaum, M. J.. Haspel, H. C. & Rosen, 0. M. (1986) Proc. Nut1 Acad. Sci. USA 83, 5784-5788. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254. Burant, C. F. & Bell, G. I. (1992) Biochemistry 31, 10414-10420. Cairns, M. T., McDonald, T. P., Home, P., Henderson, P. J. F. & Baldwin, S. A. (1991) J. Biol. Chem. 266, 8176-8183. Carruthers, A. (1986) J. Biol. Chem. 261, 11028-11037. Carruthers, A. (1990) Physiol. Rev. 70, 1135-1176. Carruthers, A. & Helgerson, A. L. (1991) Biochemistry 30, 39073915. Chin, J. J., Jhun, B. H. & Jung, C. Y. (1992) Biochemistry 31, 19451951. Deves, R. & Krupka, R. M. (1978) Biochim. Biophys. Acta 510, 186-200. Futai, M. (1978) in Bacterial transport (Rosen, B. P., ed.) pp. 741, Marcel Dekker, New York. Gorga, F. R. & Lienhard, G. E. (1981) Biochemistry 20,5108-5113. Gorga, F. R. & Lienhard, G. E. (1982) Biochemistry 21, 19051908. Gould, G. W. & Bell, G. I. (1990) Trends Biochem. Sci. 15, 18-23. Griffith, J. K., Baker, M. E., Rouch, D. A,, Page, M. G. P., Skurray, R. A,, Paulsen, I. T., Chater, K. F., Baldwin, S. A. & Henderson, P. J. F. (1992) Curr. Opin. Cell Biol. 4 , 684-695.
Helgerson, A. L. & Carruthers, A. (1987) J. Biol. Chem. 262, 54645475. Helgerson, A. L. & Carruthers, A. (1991) Biochemistry 30, 39073915. Henderson, P. J. F. (1990) J. Bioenerg. Biomembr. 22, 525-569. Henderson, P. J. F. & Maiden, M. C. J. (1990) Philos. Trans. R. SOC. Lond. B Biol. Sci. 326, 391-410. Henderson, P. J. F., Baldwin, S. A., Cairns, M. T., Charalambous, B., Dent, H. C., Gunn, F., Liang, W. J., Lucas, V. A., Martin, G. E., McDonald, T. P., McKeown, B. J., Muiry, J. A. R., Petro, K. R., Roberts, P. E., Shatwell, K. P., Smith, G. & Tate, C. G. (1992) Int. Rev. Cytol. 137, 149-208. James, D. E., Strube, M. & Mueckler, M. (1989) Nature 338, 8387. Jung, E. K. Y., Chin, J. J. & Jung, C. Y.(1986) J. Biol. Chem. 261, 915-9160, Kayano, T., Fukumoto, H., Eddy, R. L., Fan, Y.-S., Byers, M. G., Showa, T. G. & Bell, G. I. (1988) J. Biol. Chem. 263, 1524515248. Kayano, T., Burant, C. F., Fukumoto, H., Gould, G. W., Fan, Y.-S., Eddy, R. L., Byers, M. G., Shows, T. B., Seino, S. & Bell, G. I. (1990) J. Biol. Chem. 265, 13276-13282. King, A. P. J., Tai, P.-K. K. & Carter-Su, C. (1991) Biochemistry 30, 11546-11553. Lowe, A. G. & Walmsley, A. R. (1986) Biochim. Biophys. Acta 857, 146- 154. Lowe, A. G. & Walmsley, A. R . (1987) Biochim. Biophys. Acta 903, 547-550. Macpherson, A. J. S., Jones-Mortimer, M. C. & Henderson, P. J. F. (1981) Biochem. J. 196, 269-283. Maiden, M. C. J., Davis, E. O., Baldwin, S. A., Moore, D. C. M. & Henderson, P. J. F. (1987) Nature 325, 641-643. Maiden, M. C. J., Jones-Mortimer, M. C. & Henderson, P. J. F. (1988) J. Biol. Chem. 263, 8003-8010. Mueckler, M., Caruso, C., Baldwin, S. A., Panico, M., Blench, I., Morris, H. R., Allard, W. J., Lienhard, G. E. & Lodish, H. F. (1985) Science 229, 941-945. Osbom, M. J., Gander, J. E., Parisi, E. & Carson, J. (1972) J. Biol. Chem. 247, 3962-3972. Pawagi, A. B. & Deber, C. M. (1990) Biochemistry 29,950-955. Roberts, P. E. (1 992) Ph. D. thesis, University of Cambridge. Schaffner, W. & Weissman, C. (1973) Anal. Biochem. 56,502-514. Shanahan, M. F. (1982) J. Biol. Chem. 257, 7290-7293. Silverman, M. (1991) Annu. Rev. Biochem. 60, 757-794. Thorens, B., Sarkar, H. K., Kaback, H. R. & Lodish, H. F. (1988) Cell 55, 281 -290. Waddell, I. A., Zomerschoe, A. G., Voice, M. W. & Burchell, A. (1992) Biochem. J. 286, 173- 177 Walmsley, A. R. (1988) Trends Biochem. Sci. 13, 226-231. Walmsley, A. R. & Bagshaw, C. R. (1989) Anal. Biochem. 176, 313-318. Walmsley, A. R. & Lowe, A. G. (1987) Biochim. Biophys. Acta 901, 229-238. Walmsley, A. R., Petro, K. R. & Henderson, P. J. F. (1993) Eur. J. Biochem. 215, 43-54.