306

Biochimica et Biophysica Acta, 1038 (1990) 306-314 Elsevier

BBAPRO 33623

Interactions of glycolytic enzymes with erythrocyte membranes S i m o n J. H a r r i s * a n d D o n a l d J. W i n z o r Department of Biochemistry, University of Queensland, St. Lucia (Australia) (Received 30 October 1989)

Key words: Membrane-protein interaction; Erythrocyte membrane; Band 3 protein; Glycolytic enzyme

Partition equilibrium experiments have been used to characterize the interactions of erythrocyte ghosts with four glycolytic enzymes, namely aldolase, glyceraldehyde-3-phosphate dehydrogenase, phosphofructokinase and lactate dehydrogenase, in 5 mM sodium phosphate buffer (pH 7.4). For each of these tetrameric enzymes a single intrinsic association constant sufficed to describe its interaction with erythrocyte matrix sites, the membrane capacity for the first three enzymes coinciding with the band 3 protein content. For lactate dehydrogenase the erythrocyte membrane capacity was twice as great. The membrane interactions of aldolase and glyceraidehyde-3-phosphate dehydrogenase were mutually inhibitory, as were those involving either of these enzymes and lactate dehydrogenase. Although the binding of phosphofructokinase to erythrocyte membranes was inhibited by aldolase, there was a transient concentration range of aldolase for which its interaction with matrix sites was enhanced by the presence of phosphofructokinase. In the presence of a moderate concentration of bovine serum albumin (15 m g / m l ) the binding of aldolase to erythrocyte ghosts was enhanced in accordance with the prediction of thermodynamic nonideality based on excluded volume. At higher concentrations of albumin, however, the measured association constant decreased due to very weak binding of the space-filling protein to either the enzyme or the erythrocyte membrane. The implications of these findings are discussed in relation to the likely subcellular distribution of glycolytic enzymes in the red blood cell.

Introduction A recent theoretical investigation [1] has given detailed consideration to the possibility that glycolysis within the red blood cell is facilitated by spatial localization of the glycolytic enzymes in a complex adsorbed to the erythrocyte membrane. According to the proposed model this adsorption of the multienzyme complex occurs via interactions with the intrinsic membrane protein that is variously termed band 3 protein or porin (the transmembrane anion transport protein). In that regard, the binding of glycolytic enzymes to erythrocyte ghosts has certainly been considered to involve band 3 protein [2-11], the highly acidic region of its cytoplasmic domain having been implicated in this process [8,9]. However, attractive though the concept of an adsorbed glycolytic complex may be from the viewpoint of metabolic control, there is very little quantitative information on the strengths of the enzyme-membrane

* Present address: E.A. Doisy Department of Biochemistry, St. Louis University School of Medicine, St. Louis, MO 63104, U.S.A. Correspondence: D.J. Winzor, Department of Biochemistry, University of Queensland, St. Lucia, Queensland 4067, Australia.

interactions and hence on the likely extents to which they might occur in the red blood cell. Multivalency of the glycolytic enzymes in their interactions with erythrocyte membranes has been a complication that has delayed definitive characterization of these binding phenomena. Thus, although the curvilinear form of Scatchard [12] plots for these systems has invariably been ascribed to heterogeneity of membrane sites [2-10], the consequent requirement that band 3 protein must be located in at least two different environments has been given scant attention. Subsequent reconsideration of results for the membrane interactions of aldolase [4,5] and glyceraldehyde-3-phosphate dehydrogenase [10] in terms of enzyme tetravalency [13] has signified their conformity with the concept that binding to band 3 protein is governed by a single intrinsic association constant. The initial aim of this investigation is to confirm the adequacy of a single intrinsic binding constant to descibe the interaction of a glycolytic enzyme with erythrocyte ghosts. Partition equilibrium studies have therefore been conducted: (i) to obtain more definitive sets of data for aldolase and glyceraldehyde-3-phosphate dehydrogenase; and (ii) to obtain information on the interactions of two other enzymes, viz., lactate dehydrogenase and phosphofructokinase, with erythrocyte membranes.

0167-4838/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

307 The physiological relevance of the interactions between glycolytic enzymes and erythrocyte membranes is still to be proven [14], since the studies mentioned above have been performed at low ionic strength (I--0.01). Indeed, the fact that the interactions of aldolase and glyceraldehyde-3-phosphate dehydrogenase with erythrocyte ghosts cannot be detected at ionic strengths greater than 0.08 [13] seemingly supports the contention that the adsorption phenomena observed at low ionic strengths are merely hypotonic artifacts [15,16]. However, as noted previously [17], the results of in vitro experiments refer to conditions approaching thermodynamic ideality; and thus do not reflect the effects of thermodynamic nonideality emanating from the highly concentrated macromolecular environment within the red blood cell. This investigation concludes with an exploratory study in which bovine serum albumin is added as a space-tilling solute to ascertain the extent to which enzyme adsorption might be enhanced by thermodynamic nonideality.

Binding studies

In studies of the interactions of glycolytic enzymes with erythrocyte membranes, reaction mixtures (1 ml) containing ghosts (0.29-0.35 mg protein/ml) and glycolytic enzymes, either singly or in mixtures with another enzyme, in 5 mM sodium phosphate buffer (pH 7.4) were incubated at 4 ° C for 2 h. After centrifugation at 12000 × g for 15 min to pellet the membranes the supernatants were assayed for glycolytic enzyme activity [28]. Division by the specific activity of the respective stock solution then yielded CA, the molar concentration of free enzyme in the mixture with known total enzyme concentration CA. The binding data were analyzed in accordance with the general counterpart of the Scatchard plot [29], viz., rl/(CA) 1/I = qk~,x - ]kAxrf((~A)(/- 1)/I

where ,1 = [ ( C A ) ' / / - - ( C A ) W ] / C ×

Materials and Methods

Preparation of erythrocyte ghosts

Unsealed human erythrocyte ghosts were prepared from outdated blood samples, kindly provided by the Red Cross Blood Transfusion Service in Brisbane, essentially by the method of Steck and Kant [18]: specific details of the method and also of the procedures used to characterize the permeability and sidedness of the membranes appear elsewhere [13]. For use in binding studies the ghosts were equilibrated with 5 mM sodium phosphate buffer (pH 7.4) by repeated cycles of centrifugation and resuspension of the pellet in buffer. Protein concentrations of the resulting stock suspensions of erythrocyte membranes were determined by the biuret procedure [19] calibrated with bovine serum albumin. Preparation of enzyme solutions

Aldolase, lactate dehydrogenase, glyceraldehyde-3phosphate dehydrogenase and phosphofructokinase from rabbit muscle were obtained as ammonium sulphate suspensions from either Boehringer-Mannheim or Sigma Chemical Co. Before use the enzymes were freed of ammonium sulphate by extensive dialysis at 4 ° C against 5 mM sodium phosphate buffer (pH 7.4). Concentrations of the dialyzed solutions were determined spectrophotometrically at 280 nm on the basis of the following absorption coefficients (Alcm) 1~ and molecular weights: 14.4 and 140000 for lactate dehydrogenase [20]; 9.1 [21] and 160000 [22] for aldolase; 10.0 [23] and 144000 [24,25] for glyceraldehyde3-phosphate dehydrogenase; 10.7 [26] and 380000 [27] for phosphofructokinase.

(la)

(lb)

and f, the valence of the glycolytic enzymes, has been taken as 4 [13,29]. The total matrix concentration, Cx, was based on the weight-concentration of membrane protein in the reaction mixtures, a value of 6 . 1 0 -13 g protein per ghost [30] and the existence of 1.43-106 band 3 protein molecules per ghost [4]. For interactions governed by a single intrinsic association constant, kAx, thermodynamic characterization was obtained from the linear plot of r4//(CA) 1/4 VS. r4(CA) 3/4 by virtue of its slope (--4kgx) and abscissa intercept (q/4), the latter providing the valence (q) of a matrix site in its interaction with enzyme. Results and Discussion

Quantitative characterization of glycolytic erythrocyte membrane interactions

enzyme-

Results from partition studies of the interactions of aldolase and glyceraldehyde-3-phosphate dehydrogenase with erythrocyte membrane suspensions in 5 mM sodium phosphate buffer (pH 7.4, I 0.011) are presented as general counterparts [29] of Scatchard plots in Figs. la and lb, respectively. The linear form of these plots signifies the adequacy of a single intrinsic association constant to describe the interaction of either enzyme with the erythrocyte ghost preparation; and thereby confirms the conclusion [13] that the source of the curvilinearity in traditional Scatchard [12] plots for these two systems is the erroneous value (unity) assigned to the valence ( f ) of the enzymes in their interactions with membrane. Also confirmed, is the correspondence between the number of matrix sites and the band 3 protein content of erytrhocyte membranes, the abscissa intercept of 0.25 signifying a value of unity for

308 10

)

~'~

i

i

a]

t

0 0 2

O. lr4C3/40"2A '

I

I

+'~ 1~

0.3 b

-

._× ~-'¢ 0

0

0.1

0.2

0.3

r4~A3 / 4

Fig. 1. Quantitative characterizationof the interactions of (a) aldolase and (b) glyceraldehyde-3-phosphatedehydrogenasewith erythrocyte ghosts in 5 mM phosphate buffer (pH 7.4), the results being presented as generalized Scatchard plots for a tetravalent enzyme (Eqn. 1 with f = 4) and a total concentrationof membrane sites (Cx) based on the band 3 protein content.

q, the valence of a matrix site, X: it should be recalled that in the determination of r4 (Eqn. lb), the value of Cx was expressed in terms of band 3 protein concentration. The intrinsic association constant, kAx, of (8.6 + 0.8)- 105 M-1 for the aldolase-membrane interaction is smaller than that of 3.5.106 M-1 deduced [13,29] from data [3] obtained in 5 mM phosphate (pH 7.0). On the other hand, the kAx of (1.6 + 0.2)- 10 6 M-1 for the glyceraldehyde-3-phosphate dehydrogenase-membrane interaction is larger than that of 1.0- 10 6 M-1 inferred [13] from results [10] obtained in 5 mM phosphate (pH 8.0). These intrinsic association constants add a quantitative dimension to qualitative observations [30,31] that glyceraldehyde-3-phosphate dehydrogenase binds more strongly to erythrocyte membranes than does aldolase. From the multivalent Scatchard plots for the erythrocyte membrane interactions of phosphofructokinase (Fig. 2a) and lactate dehydrogenase (Fig. 2b) it is evident that linear relationships are again obtained. That for phosphofructokinase is very similar to the corresponding plot for glyceraldehyde-3-phosphate dehydrogenase (Fig. lb), not only in regard to the magnitude of kAx, viz., (1.6 + 0.2)-106 M -], but also in regard to the abscissa intercept (0.25). The consequent conclusion that the number of membrane binding-sites coincides with the band 3 protein content is consistent with findings for aldolase and glyceraldehyde-3-phosphate dehydrogenase (Fig. 1), and with an earlier claim

[6] of similar stoichiometry for the phosphofructokinase-membrane interaction. The most notable feature of the multivalent Scatchard plot for the interaction of lactate dehydrogenase with erythrocyte membranes (Fig. 2b) is the magnitude of 0.5 for the abscissa intercept, which signifies a value of 2 for q, the matrix valence. Thus, although enzyme binding is characterized by a smaller intrinsic association constant (kAx = (2.5 + 0.2). 105 M - ] ) , the interaction of lactate dehydrogenase entails an additional site to the band 3 protein site already proposed as the site for membrane interactions of other glycolytic enzymes [8,9]. The nature of this second site has not been established, but its affinity for lactate dehydrogenase must be very similar to that exhibited by the band 3 protein site. Whereas any marked disparity between the magnitudes of intrinsic constants should lead either to a curvilinear multivalent Scatchard plot or at least to increased uncertainty in the linear relationship obtained by linear regression analysis, the relative error in kAx is comparable with that for any of the interactions with q --- 1.

Binding studies with mixtures of enzymes The fact that erythrocyte membranes exhibit the same capacity for aldolase, glyceraldehyde-3-phosphate dehydrogenase and phosphofructokinase does not necessarily indicate that those enzymes are binding to the same membrane sites. However, such a situation seems likely in view of the assertion [8,9] that the membrane interactions with glycolytic enzymes involve the N-

2 ¢,D I

o

I

I

a

1

x

~< 0

2,, 0

0

O. 1 _3/40.2

0.3

r4CA

'° ~

4

,-'~" o

_e

0

0.2

--3/4

0.4

0.6

r4CA

Fig. 2. The quantitative characterization of the interactions of (a) phosphofructokinase and (b) lactate dehydrogenasewith erythrocyte ghosts in 5 mM phosphate buffer (pH 7.4), the results being presented as generalized Scatchard plots for a tetravalent enzyme (Eqn. 1 with f = 4) and a total concentration of membrane sites (Cx) based on the band 3 protein content.

309 terminal region of the band 3 protein. The existence of competition between matrix sites clearly runs counter to the formation of a multienzyme complex attached to the membrane matrix; and is accordingly detrimental to the Kurganov concept [1] of such a complex effecting more efficient flux of glycolytic intermediates. Partition equilibrium studies have therefore been performed to determine the extent to which the binding of one enzyme to erythrocyte ghosts is affected by the presence of a fixed concentration of another. Competitive binding of two multivalent enzymes (A and B) to matrix sites (X) is described by the expression [32], [gG/(Ca)1/Z] Z = k ^ x q G + R [ g G / ( C A ) '/z ]

0.8

~

I

I

I

I

I

I

I

I

a

0.8

0.4

O

,~0.2

,

N

-'.0

-0.8 -0.6 -0.4 -0.2 1/4 (r4Ox/CA)[kAx(ex_&)_Zr4Cx -

1.0

I

I

o.8-~ ~0.4

I

I

0 A - / -G 1/4qj b

5 ~

0.6

-

U_

X( k^x(qG - g G ) - Z r / G / ( C^) 'If } (2a)

m

where R is the ratio of the two intrinsic association constants ( k B x / k A x ) and z = 1 + I k A x ( G , ) 1 / / ( G ) ( / - 1)//

0.2

0

0

I 0.5

I 1.0

I 1.5

I 2.0

2.5

Free Aldolase Conc. CA (laM) (2b)

Thus, provided that values may be ascribed to f and g, the respective valences of A and B in their interactions with membrane sites, Eqn. 2 affords a convenient linear transform of experimental results for the binding of enzyme A in the presence of total concentration C B of the second enzyme. The effect of glyceraldehyde-3-phosphate dehydrogenase (0.8/xM) on the binding of aldolase to erythrocyte membranes in 5 mM phosphate (pH 7.4) is summarized in Fig. 3. Although casual inspection of the results showed that the extent of aldolase binding was obviously diminished in the presence of glyceraldehyde3-phosphate dehydrogenase, analysis of the data in terms of competitive inhibition led to a result (Fig. 3a) which is incompatible with Eqn. 2 in that the negative slope implies a negative value for the ratio of association constants for the two enzymes. The physical description )f this mathematical inference is that the binding of glyceraldehyde-3-phosphate dehydrogenase to a membrane site is having a greater inhibitory effect than that attributable to its complete prevention of aldolase binding to that site. Such a situation could well arise if glyceraldehyde-3-phosphate dehydrogenase and aldolase were interacting in the aqueous phase to form a complex with no affinity for membrane, in which case the anomalous form of Fig. 3a would reflect an overestimation of the free aldolase concentration in the aqueous phase. However, frontal gel chromatography of the individual enzymes and mixtures thereof [33] on Sephadex G-200 failed to reveal any evidence of complex formation under the same experimental conditions

Fig. 3. The effect of glyceraldehyde-3-phosphatedehydrogenase(B) on the binding of aldolase (A) to erythrocyte ghosts in 5 mM phosphate buffer (pH 7.4). (a) Test of results obtained in the presence of 0.8/~M glyceraldehyde-3-phosphatedehydrogenasefor competitive binding of these two tetravalent enzymes, the results being plotted according to Eqn. 2 with f =g = 4. (b) Comparison of the same binding data (I) with the correspondingbinding curve (o) obtained in the absence of glyceraldehyde-3-phosphatedehydrogenase. (pH 7.4, I 0.011). Another possible explanation is that the binding of glyceraldehyde-3-phosphate dehydrogenase to a membrane site not only prevents the binding of aldolase to that site but also induces a structural perturbation in the membrane that impairs its affinity for aldolase, i.e., a perturbation that leads to a decrease in kAx- Irrespective to the mechanism by which glyceraldehyde-3-phosphate dehydrogenase affects the extent of aldolase binding, the important conclusion is that this enzyme does inhibit the binding of aldolase. This point is more clearly evident from Fig. 3b, which compares the two binding curves that are obtained using the classical binding function, r [12,34]. Since results for other enzyme mixtures were also only amenable to qualitative interpretation, the same format will be retained for the presentation of those comparisons. The inhibitory effect of aldolase (0.5 gM) on the binding of glycerladehyde-3-phosphate dehydrogenase is evident from Fig. 4a. Taken in conjunction with the results of the reverse experiment (Fig. 3b), this observation confirms the report [30] that the binding of these two enzymes is mutuall2~ inhibitory. Although it could be argued that such a situation is a logical necessity that requires no confirmation, the need to do so was dictated

310 1.5

I

I

e

1.0 Ii

0.5

0 Free

1.5

I

i

0.4

0.8

1.2

E n z y m e Conc. (pM) I I b

t-

,~1°-" ~ L~. r 0.5

h~ A

o

o

0.5

Free Enzyme

1.o

1.5

Conc. (isM)

Fig. 4. The effect of a second glycolyticenzyme on the interactions of glyceraldehyde-3-phosphate dehydrogenase and lactate dehydrogenase with erythrocyte ghosts in 5 mM phosphate buffer (pH 7.4). (a) Binding curves for glyceraldehyde-3-phosphatedehydrogenase in the absence (©) and presence (e) of 0.5/~M aldolase. (b) Corresponding curves for lactate dehydrogenase in the absence (©) and presence of 1.3 #M aldolase (e) or 0.7 /~M glyceraldehyde-3-phosphate dehydrogenase (A).

strued as evidence that at least lactate dehydrogenase and one other enzyme can bind simultaneously to the m e m b r a n e matrix. This pattern of evidence against the formation of a multienzyme complex of glycolytic enzymes attached to the erythrocyte membrane extends to the effect of aldolase (1.3 # M ) on the binding of phosphofructokinase (Fig. 5a). However, the partition study of the reverse situation, viz., the effect of phosphofructokinase (0.1 /~M) on the binding of aldolase, provides the first evidence in this investigation for the enhanced binding of an enzyme due to the presence of another (Fig. 5b): qualitatively similar results were obtained in the corresponding study of the binding of glyceraldehyde-3phosphate dehydrogenase in the presence of phosphofructokinase. On the basis of the intrinsic association constant of 1.6.106 M - l obtained from Fig. 2a, essentially all (97%) of the phosphofructokinase is adsorbed to the erythrocyte membranes (Cx = 1.2 /~M) in the absence of aldolase; and from the low concentration region of the binding curve in Fig. 5b this relatively major extent of phosphofructokinase attachment to the matrix is enhancing the binding of aldolase. However, with increasing aldolase concentration the binding curve in the presence of phosphofructokinase approaches that

1.0

,

I

'

y

L_

-° .i-,

o

o

I

o

a

o

0 e-

~_ 0.5 by a subsequent report [31] that the binding of aldolase is inhibited by glyceraldehyde-3-phosphate dehydrogenase, but that aldolase has no effect on the binding of glyceraldehyde-3-phosphate dehydrogenase. Since the erythrocyte membrane exhibits double the capacity for lactate dehydrogenase that it does for the other enzymes (Fig. 2), effects of aldolase (1.3 #M) and glyceraldehyde-3-phosphate dehydrogenase (0.7/xM) on this interaction were also determined (Fig. 4b). In keeping with the concept that the mutually inhibitory actions of these two enzymes signifies their interactions with the same or overlapping matrix sites, the inhibitory effect of aldolase on the binding of lactate dehydrogenase by erythrocyte ghosts is also exhibited by glyceraldehyde-3-phosphate dehydrogenase. Moreover, the extent of this inhibition suggests that the binding of either aldolase or glyceraldehyde-3-phosphate dehydrogenase to a membrane site effectively precludes the binding of lactate dehydrogenase to either of the two sites for that enzyme. Thus, inasmuch as band 3 protein also seems to be implicated in both erythrocyte membrane sites for lactate dehydrogenase, the additional membrane capacity for this enzyme cannot be con-

.E

0

, 0

I 0.05

i

I 0.10

Free E n z y m e Conc. (pM) 1,0

'

I

J

I

b c O °~ O e-

u. 0.5 C ,m m

0

,

0

I

1.0

i

I

2.0

Free E n z y m e Conc, (pM) Fig. 5. Interactions of aldolase and phosphofructokinase with erythrocyte ghosts (pH 7.4, I 0.011) in the presence of each other. (a) Binding of phosphofructokinase in the absence (o) and presence (e) of aldolase (1.3 ~M); and (b) binding of aldolase in the absence (o) and presence (e) of phosphofructokinase(0.1 ~M).

311 in its absence, a result consistent with the displacement of phosphofructokinase from matrix sites by aldolase (Fig. 5a). In the range of aldolase concentrations for which its interaction with erythrocyte membranes is insufficiently strong (kAx = 8.6.105 M -1) to displace phosphofructokinase (kBx = 1.6- 106 M-l), the additional aldolase associated with the membrane phase is presumably binding with high affinity to sites generated by the interaction of phosphofructokinase with erythrocyte membranes. Since the difference between the binding curves in this region signifies the creation of three-to-four sites per phosphofructokinase molecule bound to matrix, it seems logical to consider that these sites are on the phosphofructokinase rather than on the matrix. In the model of the multienzyme complex proposed by Kurganov [1], the other glyco!ytic enzymes bind to the self-association sites of phosphofructokinase. A relatively low concentration of phosphofructokinase (0.1 #M) was used in these partition studies to allow opportunity for its displacement by the more weakly bound aldolase; but it transpires that a value of 0.1/~M may well approximate the concentration of this enzyme in erythrocytes [35,36]. For aldolase a similar coupling of erythrocyte activity [35] with the specific activity of highly purified human erythrocyte enzyme [5] yields a total aldolase concentration of 0.7/~M. Whilst it is tempting to identify the stimulated binding of aldolase by erythrocyte ghosts for this combination of enzyme concentrations (Fig. 5b) as verifying the postulated existence of multiple enzyme attachment to a single matrix site, no such conclusion can be drawn because of the low ionic strength used to obtain the evidence [14]. On the positive side, this evidence of 'piggy-back' binding of enzymes, albeit under conditions irrelevant to the physiological environment, does illustrate the generation of binding sites for one enzyme by the immobilization of another on a subcellular structure - an essential prerequisite of the various models for enhanced glycolytic flux by a multienzyme complex attached to the membrane [1] or myofibrillar [37,38] matrix.

is zero or merely of insufficient magnitude for measurement by a technique in which binding governed by an association constant less than 104 M-1 is undetectable. Should the latter situation prevail, a negligible extent of enzyme binding under the thermodynamically ideal conditions of the partition studies could well assume significant proportions in the highly concentrated, thermodynamically nonideal, environment within the red blood cell [13,17,43]. From the viewpoint of physiological relevance the obvious protein to use as a space-filling solute is haemoglobin, but its intense coloration in concentrated solution is clearly an impediment to estimations of enzyme distribution between particulate and aqueous phases in partition experiments. In this exploratory study of thermodynamic nonideality effects on the interaction between aldolase and erythrocyte ghosts, serum albumin is substituted as the space-filling solute. To test that the inclusion of serum albumin does lead to enhanced binding of aldolase, partition experiments were again conducted at low ionic strength (pH 7.4, I 0.011), conditions under which the thermodynamic association constant (kAx) for the enzyme-membrane interaction had already been determined. Allowance for the effects of thermodynamic nonideality in Eqn. 1 merely requires the equilibrium constant so defined to be regarded as an apparent association constant ( k ~ ) , which then needs to be related to the thermodynamic constant (kAx) calculated by the application of Eqn. 1 to results obtained under conditions approaching thermodynamic ideality (Fig. la). A quantitative relationship between kAx and k~,~' is derived by expressing the various activity coefficients (y) as second virial coefficients (a) written in terms of covolume (U), charge-charge and anhydrous molar volume (M'f) contributions [44]. Specifically,

Need for enhancement of erythrocyte membrane-enzyme interactions by thermodynamic nonideality Since interactions of glycolytic enzymes with erythrocyte membranes involve the highly acidic amino-terminal region of band 3 protein [8,9], the effect of ionic strength on such interactions needs to be considered. Despite claims for the existence of significant amounts of membrane-bound enzymes at physiologically relevant ionic strength [7,39-42], the interactions of aldolase and glyceraldehyde-3-phosphate with erythrocyte ghosts become unquantifiable at ionic strengths greater than 0.08 [13]. The question at issue is whether the value of kAx at physiological ionic strength

where j encompasses all species, including i. The charge-charge term employs standard Debye-Huckel nomenclature in that Zi and a i denote the nett charge and radius, respectively, of i, ~ being the inverse screening length. For evaluation of covolume it is customary to assume spherical geometry for the various species, whereupon U,,j = 4qrN( a i + a j)3/3. Although consideration of the erythrocyte membrane in these geometrical terms is precluded on the grounds that molecular dimensions cannot be ascribed to a solid phase, it is permissible to consider 'the size of aldolase (A) to be miniscule in comparison with that of the membrane matrix; and hence that UA,x -----UAX,X. Provided that the

yi=exp[~,(ai,jCj)]

(3a) Z i Z j ( I + ra i + xaj)

ai'j=Ui'j-MjOJ+ 2I(l+xa,)(l+raj)

(3b)

312 approximation is then made that charge is conserved in all enzyme-matrix interactions, it follows [45] that k ~,P~ kAx( yAYx)/YAx =

(4)

4

k~P~ = kAx exp[aA,ACA+ aA,xC x + O/A,MCM]

(5)

where M refers to the inert space-filling solute (albumin) added to effect thermodynamic nonideality. In the present experimental context the free concentrations of enzyme (CA) and matrix sites (Cx) are sufficiently small in relation to the concentration of albumin (CM) for the exponent in Eqn. 5 to be dominated by the final term, whereupon ZAZM(1 + Kah+ ~aM) 1 k~PxP/kAx=exp UA.M--MMSM+ 21(l+~aA)(l+~aM) C M (6) =

As in earlier studies [17,44,45], the Stokes radius has been assumed to approximate the radius of a globular protein suitable for covolume calculation, a value of 1300 litre per mol for (UA, M - - M M O M ) being obtained from the following parameters: a A = 4.6 nm [22], a M = 3.5 nm, vra = 0.734 m l / g and MM = 66 500 [46]. On the basis that x = 3.27. I07~/I, an estimate of 3.43.10 6 is obtained for the inverse screening length, and hence the only parameters for which values are still required to complete the calculation of 0/A,M are the two nett charges. Z M has been taken as - 2 2 , a value obtained by direct estimation of the charge on albumin at pH 7.4 [47]; and Z A as - 2 , a value obtained by combining the electrophoretic mobility of - 0 . 8 • 10 -5 cm 2 • s -1 • V - I for aldolase under these conditions [48] with the Abramson expression [49]. On making all of these substitutions Eqn. 6 simplifies to ( k~,PxP/kAx) = exp(2600C M)

(7)

Results of partition equilibrium experiments in which bovine serum albumin (Sigma) was included in reaction mixtures are summarized in Fig. 6, where the broken line denotes the dependence of k~PP/kAx upon albumin concentration (CM) that is predicted by Eqn. 7. The first point to note is that the extent to which the apparent intrinsic association constant is increased by 0.23 mM albumin (15 m g / m l ) corresponds very closely to the enhancement predicted by considerations of thermodynamic nonideality. Secondly, the fact that kapp//., AX/~'AX then decreases with further increase in CM points to the invalidity of considering albumin's effect solely in terms of a space-filling role, i.e., to a breakdown in the assumption that albumin is inert. The consequent conclusion that albumin must interact weakly with either aldolase or with erythrocyte mem-

//

I

X

,,/

/

.sis

~" which, on substitution of Eqn. 2a for the activity coefficients, becomes

I

2-

T--"

o

l

I

0.2

0.4

-

~XX

0

I 0.6

Albumin Conc. C M (mM)

Fig. 6. The effect of bovine serum albumin on the interaction of aldolase with erythrocyte ghosts (pH 7.4, I 0.011), the results being plotted as the ratio of the measured intrinsic association constants in the presence(k~,P~)and absence(kAx) of this space-fillingsolute (M). The broken line is the theoretical relationship predicted (Eqn. 7) on the statistical mechanical basis of excludedvolume.

brane sites clearly runs counter to its use for comment on the extent to which thermodynamic nonideality effects may enhance the binding of aldolase to band 3 protein in the physiological environment. However, the results of Fig. 6 do serve an important role by emphasizing that the high concentrations of space-filling solute required to effect enhancement of interactions via thermodynamic nonideality place stringent demands on the degree to which the solute is inert. In that regard the selection of albumin as space-filling solute may well have been an appropriate model for the erythrocyte system, since haemoglobin is known to interact with band 3 protein [50]. Combination of that finding with the demonstration (Fig. 6) that weak chemical interaction can readily suppress any enhanced extent of interaction emanating from macromolecular crowding clearly casts doubt on the validity of the concept [1] that glycolysis within the red blood cell is facilitated by a multienzyme complex of glycolytic enzymes attached to band 3 protein of the erythrocyte membrane.

Concluding remarks This partition equilibrium study of interactions between erythrocyte ghosts and glycolytic enzymes has yielded several important findings. (i) It has provided evidence (Figs. 1 and 2) for the equivalence and independence of erythrocyte membrane sites in their interactions with four glycolytic enzymes. For aldolase, glyceraldehyde-3-phosphate dehydrogenase and phosphofructokinase the membrane capacity for enzyme coincides with the band 3 protein content, but for lactate dehydrogenase the membrane capacity is twice as great. (ii) The concept that facilitated glycolytic flux within the red blood cell is effected by a multienzyme complex attached to the erythrocyte membrane [1] is seemingly contradicted by demonstrations that the membrane interactions of aldolase and glyceraldehyde3-phosphate dehydrogenase are mutually inhibitory (Fig.

313 3), as are those involving either of these enzymes and lactate dehydrogenase (Fig. 4). (iii) Although the binding of phospofructokinase to erythrocyte membranes is inhibited by aldolase (Fig. 5a), there is a transient concentration range of aldolase for which its interaction with matrix sites is enhanced by the presence of phosphofructokinase (Fig. 5b). This finding is significant inasmuch as it illustrates the feasibility of generating binding sites for one enzyme by the adsorption of another to a subcellular matrix, a concept that is central to the formation of the envisaged multienzyme complex anchored to the membrane by the interaction of phosphofructokinase with band 3 protein [1]. (iv) The attempt to increase the strength of the aldolase-erythrocyte membrane interaction by thermodynamic nonideality as the result of macromolecular crowding was thwarted by very weak binding of serum albumin to either reactant that overrode the enhancement emanating from excluded volume effects (Fig. 6). In keeping with a recent appraisal of the situation [14], the findings of the present investigation are extremely difficult to reconcile with the concept [8,9] that the highly acidic cytoplasmic domain of band 3 protein provides an adsorption site for glycolytic enzymes in the red blood cell. Involvement of electrostatic attraction in the observed interactions between glycolytic enzymes and erythrocyte ghosts is certainly implicated by the inverse dependence of the strength of binding upon ionic strength [13]. However, the binding of glycolytic enzymes to erythrocyte ghosts becomes undetectable at the ionic strength of 0.15 that is usually associated with the physiological environment [14]; and accordingly, the persistence of these interactions in the red blood cell would require the operation of a large excluded volume effect to achieve sufficient enhancement of any thermodynamic association constant for membrane-enzyme adsorption [17]. Although haemoglobin is present at a sufficiently high concentration in the red blood cell to assume such a space-filling role, its interaction with band 3 protein [50] would almost certainly offset any enhancement of enzyme binding to this matrix site as the result of macromolecular crowding (Fig. 6). In contrast with these conclusions based on quantitative in vitro studies, there are also claims for the existence of significant amounts of membrane-bound glycolytic enzymes at physiological ionic strength [7,39-42]. Of those reports, the most recent [42] comprises a detailed morphological study of the red blood cell. Irrespective of whether the contents of the red blood cell were fixed with paraformaldehyde or immobilzied by freezing, glyceraldehyde-3-phosphate dehydrogenase appeared to be localized in the vicinity of the erythrocyte membranes [42]. One possible explanation of the disparity between those observations and the conclusions drawn from in vitro studies is that the localization of glyceraldehyde-3-phosphate dehydrogenase is an

artifact emanating from the fixing/freezing procedures used in the morphological study [51]. Alternatively, the erythrocyte m e m b r a n e may be sufficiently highly charged to effect a localized Donnan redistribution of ions, thereby rendering the ionic strength in its immediate vicinity much lower than the average value of 0.15 that is usually ascribed to the physiological environment [52].

Acknowledgements The financial support of this investigation by the Australian Research Council is gratefully acknowledged, as is the receipt (by S.J.H.) of a Commonwealth Postgraduate Research Award.

References 1 2 3 4 5 6

Kurganov, B.I. (1986) J. Theor. Biol. 119, 445-455. Kant, J.A. and Steck, T.L. (1973) J. Biol. Chem. 248, 8457-8464. Yu, J. and Steck, T.L. (1975) J. Biol. Chem. 250, 9176-9184. Strapazon, E. and Steck, T.L. (1976) Biochemistry 15, 1421-1424. Strapazon, E. and Steck, T.L. (1977) Biochemistry 16, 2966-2971. Higashi, T., Richards, C.S. and Uyeda, K. (1979) J. Biol. Chem. 254, 9542-9550. 7 Kliman, H.J. and Steck, T.L. (1980) J. Biol. Chem. 255, 6314-6321. 8 Prasanna Murthy, S.N., Liu, T., Kaul, R.K., Kohler, H. and Steck, T.L. (1981) J. Biol. Chem. 256, 11203-11208. 9 Tsai, I.-H., Prasanna Murthy, S.N. and Steck, T.L. (1982) J. Biol. Chem. 257, 1438-1442. 10 Beth, A.H., Balasubramanian, K., Wilder, R.T., Venkataramu, S.D., Robinson, B.H., Dalton, L.R., Pearson, D.E. and Park, J.H. (1981) Proc. Natl. Acad. Sci. USA 78, 4955-4959. 11 Jenkins, J.D., Kezdy, F.J. and Steck, T.L (1985) J. Biol. Chem. 260, 10426-10433. 12 Scatchard, G. (1949) Ann. NY Acad. Sci. 51, 660-666. 13 Kelley, G.E. and Winzor, D.J. (1984) Biochim. Biophys. Acta 778, 67-73. 14 Maretzki, D., Reimann, B. and Rapoport, S.M. (1989) Trends Biochem. Sci. 14, 93-96. 15 Maretzld, D., Groth, J., Tsamaloukas,A.G., Grundell, M., Kruger, S. and Rapoport, S. (1974) FEBS Lett. 39, 83-87. 16 Simpson, R.J., Brindle, K.M. and Campbell, I.D. (1983) Biochem. Soc. Trans. 11, 281-282. 17 Harris, S.J. and Winzor, D.J. (1985) Arch. Biochem. Biophys. 243, 598-604. 18 Steck, T.L. and Kant, J.A. (1974) Methods Enzymol. 31, 172-180. 19 Gornall, A.G., Bardawill, C.J. and David, M.M. (1949) J. Biol. Chem. 177, 751-766. 20 Jaenicke, R. and Knof, S. (1968) Eur. J. Biochem. 4, 157-163. 21 Baranowski, T. and Niederland, T.R. (1949) J. Biol. Chem. 180, 543-551. 22 Kawahara, K. and Tanford, C. (1966) Biochemistry5, 1578-1584. 23 Fox, J.B., Jr. and Dandliker, W.B. (1956) J. Biol. Chem. 221, 1005-1017. 24 Harrington, W.F. and Karr, G.M. (1965) J. Mol. Biol. 13, 885-793. 25 Harris, J.I. and Perham, R.N. (1968) Nature (Lond.) 219, 10251028. 26 Hesterberg, L.K. and Lee, J.C. (1982) Biochemistry 21, 216-222. 27 Parmeggiani, A., Luft, J.H., Love, D.S. and Krebs, E.G. (1966) J. Biol. Chem. 241, 4625-4637. 28 Scopes, R.K. (1977) Biochem. J. 161, 253-263. 29 Hogg, P.J. and Winzor, D.J. (1985) Biochim. Biophys. Acta 843, 159-163.

314 30 Solti, M. and Friedrich, P. (1976) Mol. Cell. Biochem. 10, 145-152. 31 Wilson, J.E., Reid, S. and Masters, C.J. (1982) Arch. Biochem. Biophys. 215, 610-620. 32 Harris, S.J. and Winzor, D.J. (1988) Anal. Biochem. 169, 319-337. 33 Masters, C.J. and Winzor, D.J. (1981) Arch. Biochem. Biophys. 209, 185-190. 34 Klotz, I.M. (1946) Arch. Biochem. 9, 109-117. 35 Beutler, E. (1971) Red Cell Metabolism: A Manual of Biochemical Methods, Grune & Stratton, New York. 36 Tarui, S., Kono, N. and Uyeda, K. (1972) J. Biol. Chem. 247, 1138-1145. 37 Clarke, F.M., Stephan, P., Morton, D.J. and Weidemann, J. (1985) in Regulation of Carbohydrate Metabolism (Beitner, R., ed.), Vol. II, pp. 1-33, CRC Press, Boca Raton, FL. 38 Kurgnaov, B.I., Sugrobova, N.P. and Mil'man, L.S. (1985) J. Theor. Biol. 116, 509-526. 39 Keokitichai, S. and Wrigglesworth, J.M. (1980) Biochem. J. 187, 837-841. 40 Yeltman, D.R. and Harris, B.G. (1980) Arch. Biochem. Biophys. 199, 186-196.

41 Solti, M., Barthe, F., Halasz, N., Toth, G., Sirokman, F. and Friedrich, P. (1981) J. Biol. Chem. 256, 9260-9265. 42 Rogalski, A.A., Steck, T.L. and Waseem, A. (1989) J. Biol. Chem. 264, 6438-6446. 43 Liu, S.-C. and Palek, J. (1984) J. Biol. Chem. 259, 11556-11562. 44 Wills, P.R., Nichol, L.W. and Siezen, R.J. (1980) Biophys. Chem. 1l, 71-82. 45 Ford, C.L., Winzor, D.J., Nichol, L.W. and Sculley, M.J. (1983) Biophys. Chem. 18, 1-9. 46 Baldwin, R.L. (1957) Biochem. J. 65, 503-512. 47 Ford, C.L. and Winzor, D.J. (1983) Biochim. Biophys. Acta 756, 49-55. 48 Velick, S.F. (1949) J. Phys. Colloid Chem. 53, 135-149. 49 Abramson, H.A., Moyer, L.S. and Gorin, M.H. (1942) Electrophoresis of Proteins, Reinhold, New York. 50 Low, P.S. (1986) Biochim. Biophys. Acta 864, 145-167. 51 Cooper, J.A., Loftus, D.J., Frieden, C., Bryan, J. and Elson, E.L. (1988) J. Cell Biol. 106, 1229-1240. 52 Comper, W.D. and Preston, B.N. (1974) Biochem. J. 143, 1-9.