ATP regulation of the human red cell sugar transporter.

Purified human red blood cell sugar transport protein intrinsic tryptophan fluorescence is quenched by D-glucose and 4,6-ethylidene glucose (sugars that bind to the transport), phloretin and cytochalasin B (transport inhibitors), and ATP. Cytochalasin B-induced quenching is a simple saturable phenomenon with Kd app of 0.15 microM and maximum capacity of 0.85 cytochalasin B binding sites per transporter. Sugar-induced quenching consists of two saturable components characterized by low and high Kd app binding parameters. These binding sites appear to correspond to influx and efflux transport sites, respectively, and coexist within the transporter molecule. ATP-induced quenching is also a simple saturable process with Kd app of 50 microM. Indirect estimates suggest that the ratio of ATP-binding sites per transporter is 0.87:1. ATP reduces the low Kd app and increases the high Kd app for sugar-induced fluorescence quenching. This effect is half-maximal at 45 microM ATP. ATP produces a 4-fold reduction in Km and 2.4-fold reduction in Vmax for cytochalasin B-inhibitable D-glucose efflux from inside-out red cell membrane vesicles (IOVs). This effect on transport is half-maximal at 45 microM ATP. AMP, ADP, alpha, beta-methyleneadenosine 5'-triphosphate, and beta, gamma-methyleneadenosine 5'-triphosphate at 1 mM are without effect on efflux of D-glucose from IOVs. ATP modulation of Km for D-glucose efflux from IOVs is immediate in onset and recovery. ATP inhibition of Vmax for D-glucose exit is complete within 5-15 min and is only partly reversed following 30-min incubation in ATP-free medium. These findings suggest that the human red cell sugar transport protein contains a nucleotide-binding site(s) through which ATP modifies the catalytic properties of the transporter.

Purified human red blood cell sugar transport protein intrinsic tryptophan fluorescence is quenched by D-glucose and 4,6-ethylidene glucose (sugars that bind to the transport), phloretin and cytochalasin B (transport inhibitors), and ATP. Cytochalasin B-induced quenching is a simple saturable phenomenon with Kd of 0.15 p~ and maximum capacity of 0.85 cytochalasin B binding sites per transporter. Sugar-induced quenching consists of two saturable components characterized by low and high Kdapp binding parameters.
These binding sites appear to correspond to influx and efflux transport sites, respectively, and coexist within the transporter molecule. ATP-induced quenching is also a simple saturable process with Kd,, of 50 pM.
Indirect estimates suggest that the ratio of ATP-binding sites per transporter is 0.87:l. ATP reduces the low Kd,, and increases the high Kd,, for sugar-induced fluorescence quenching. This effect is half-maximal at 45 p~ ATP. ATP produces a 4-fold reduction in K,,, and 2.4-fold reduction in V,,, for cytochalasin B-inhibitable D-glucose efflux from inside-out red cell membrane vesicles (IOVs). This effect on transport is half-maximal at 45 p~ ATP. AMP, ADP, a,@-methyleneadenosine 5'-triphosphate, and B,ymethyleneadenosine 5"triphosphate at 1 mM are without effect on efflux of D-glucose from IOVs. ATP modulation of K , for D-glucose efflux from IOVs is immediate in onset and recovery. ATP inhibition of V,, for D-glucose exit is complete within 5-15 min and is only partly reversed following 30-min incubation in ATP-free medium. These findings suggest that the human red cell sugar transport protein contains a nucleotide-binding site(s) through which ATP modifies the catalytic properties of the transporter.
Insulin stimulation of sugar transport in muscle and adipose seems not to result from increased intrinsic activity of plasmalemmal sugar transport proteins but rather from an insulin-induced "recruitment" of sugar transport protein from microsomal membranes to the plasma membrane (1-4). Anoxia, metabolic depletion, and contractile activity also markedly stimulate sugar transport in skeletal muscle (5,6). However, a number of lines of evidence suggest that insulinindependent acceleration of skeletal muscle sugar transport is mediated by recruitment-independent mechanisms.
Insulin-induced, recruitment-mediated sugar transport stimulation is abolished in metabolically depleted adipocytes (4). It seems unlikely, therefore, that acceleration of skeletal * This work was supported by National Science Foundation Grant DCB-8510876 and National Institutes of Health Grant RO1-AM 36081-01. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. muscle sugar transport by metabolic depletion could result from an energy-dependent recruitment-like mechanism. Moreover, whereas insulin stimulation of sugar transport is both rapid in onset and recovery (4), the rapid stimulation of sarcolemmal sugar transport by metabolic depletion and contractile activity recovers only very slowly upon restoration of resting sarcoplasmic ATP and Ca2+ levels (5, 6). Is sugar transport also subjected to a recruitment-independent form of regulation in which the direct covalent or allosteric modification of plasmalemmal sugar transport protein results in altered catalytic activity? Direct support for this hypothesis is unavailable. Indirect support has been obtained from studies of adipose and human erythrocyte sugar transport. In adipose, a variety of agents that act to increase cytosolic CAMP levels also inhibit sugar transport without altering the distribution of carrier between plasmalemmal and microsomal compartments (7). Human erythrocyte sugar transport is modulated by low molecular weight factors present in cytosol (8), possibly ATP (9). The erythrocyte-lacking the specialized intracellular organelles necessary for membrane cycling-is, of course, incapable of carrier recruitment. The red cell therefore provides the almost unique potential for the study of recruitment-independent sugar transport regulation under conditions where physiological effects on transport are not obscured by the possible occurrence of carrier redistribution between plasmalemmal and intracellular compartments.
This study demonstrates that micromolar concentrations of ATP drastically modify sugar transport in human red cell inside-out vesicles and substrate binding to purified human red cell sugar transport protein.

EXPERIMENTAL PROCEDURES
Materials-Outdated, whole human blood was obtained from the University of Massachusetts Medical Center Blood Bank.
Solutions-Tris medium consisted of 50 mM Tris-HC1, 0.2 mM EDTA adjusted to pH 7.0 using 1 M Tris base.
Preparation of Inside-out Vesicles (Z0Vs')"IOVs were formed and loaded with 60 mM D-glucose as described previously (8). Sialic acid accessibility assays were performed as described by Steck and Kant (10) and sialic acid released quantitated by the method of Warren (11). 95% of membrane sialic acid groups in IOVs are inaccessible to the enzyme sialidase.
Fluorescence Measurements-Fluorescence measurements were performed at 23 "C using a Farrand Spectrofluorimeter MK 2 with excitation at 295 nm and excitation and emission bandwidths of 2.5 and 10 nm, respectively. Band 4.5 samples (25 pg in 100 pl) were added to 2 ml of Tris medium (final transporter concentration, 0.22 PM). Sugars, nucleotides, and transport inhibitors (stock solutions adjusted to pH 7.0 using 1 M Tris base) were injected into the cuvette from above. The contents of the cuvette were constantly stirred using a Spectrocell Inc. cuvette stirrer. Fluorescence quenching data represent quenching of steady-state emission at 334 nm. In all instances, fluorescence changes upon addition of ligand were complete within 5-20 s and remained stable for at least 5-15 min. At greater excitation bandwidths, fluorescence decayed monoexponentially (t% = 4-8 min at 23 "C) due, presumably, to photolysis of tryptophan residues.
Steady-state quenching data are subject to the following systematic errors. 1) Dilution artifacts resulting from addition of volumes of stock ligand solution; 2) changes in emission spectral position/shape upon interaction of transporter with ligand; 3) changes in transporter absorption spectrum upon interaction with ligand; 4) variations in the refractive index of the transporter solution upon addition of ligand; 5) the inner filter effect (attenuation of the exciting light due to absorption at 295 nm by added ligand); 6) reabsorption of transporter fluorescence by ligand.
Dilution corrections were assessed using the ligands L-glucose, mannitol, and cytochalasin D and E. These agents do not interact significantly with the transport protein, and following correction for dilution effects at maximum ligand concentrations, emission was within 98-100% of control emissions. Sugars that interact with the transporter and cytochalasin B induce a shift in the emission spectrum peak of band 4.5 to shorter wavelengths by about 1-3 nm. The necessary correction for this shift was less than 1% in 180 mM Dglucose and 50 p~ cytochalasin B (assuming no quenching). Changes in the transporter's absorption spectrum upon addition of ligand were assessed by analysis of the absorption spectra of band 4.5 in the absence and presence of the highest concentrations of ligands added. The difference spectra of ligand/buffer/transporter-ligandbuffer were virtually identical (in position, shape, and absorbance) to the absorption spectrum of transporter in buffer alone. No correction for absorption spectral changes was considered necessary. Refractive index corrections range from zero to n2 (17). The correction for 180 mM D-glucose is ~2 % .
Absorption by sugars and cytochalasin B at 295 and 334 nm was barely distinguishable from absorption by buffer alone, hence the systematic errors caused by inner filter effects and fluorescence reabsorption were negligible in studies where the added ligands were sugars and/or cytochalasin B. AMP, ADP, ATP, AMP-CPP, and AMP-PCP at 1 mM absorb weakly at 295 nm (Am = 0.057, 0.057, 0.058, 0.063, and 0.066, respectively) and insignificantly at 334 nm. The nucleotides therefore produce apparent quenching due to attenuation of the exciting light at 295 nm but leave emission at 334 nm unaffected. This inner filter effect was corrected as suggested by Parker (18). The correction takes the form F,, -2.203 A(d2d l ) F 10-Adl -10-Ad2 where A is the total absorbance at the excitation wavelength and d l and d2 refer to the geometry of the observed volume. With ATP at 1 mM, this correction results in a 3.4% increase in transporter emission at 334 nm. The transport inhibitor, phloretin, absorbs very strongly both at 295 and 334 nm, hence both inner filter effects and fluorescence reabsorption corrections were necessary. Initially, fluorescence reabsorption by phloretin was corrected by first correcting for inner filter effects, then, after determination of the absorption coefficient bert's law to calculate the original intensity of emission. Although of phloretin at each experimental concentration, by applying Lamsound in principle, this series of corrections could be subject to error due to uncertain geometries. Eventually, a more empirical approach was adopted. The fluorescence of a standard (N-acetyltryptophanamide, NATA) was measured as a function of added ligand (19). After correction for the above systematic errors, the emission of NATA in the presence of all ligands was within 98-100% of control emission (zero ligand). This indicates that NATA does not interact with the ligands employed in this study. Band 4.5 quenching data were then corrected for artifactual quenching by nucleotides and phloretin by running parallel experiments with NATA. The results of corrections using NATA were not significantly different from those obtained by applying first principles.
Unless stated otherwise, all data presented in this study have been corrected for artifactual quenching of emission by ligand. Absorption photometer. measurernenta were made at 23 "C using a Beckman DU-8 Spectro-Quuntitation of Sugar Transport Protein Numbers-Sugar transporter numbers were quantitated by determining the number of cytochalasin B-binding sites present in each band 4.5 preparation. Equilibrium [3H]cytochalasin B binding assays were carried out as described previously (14). An alternative procedure is to monitor cytochalasin B-induced quenching of transport protein intrinsic tryptophan fluorescence (20, 21). Under conditions where a significant proportion of ligand is bound to the transporter, the dissociation constant, Kd, for cytochalasin B binding to transporter is given by where CL is the total cytochalasin B concentration, R is the fractional saturation of transporter by cytochalasin B, CE the total transporter concentration, and n the number of binding sites per transporter. In practice, R is determined by titrating the transporter with cytochalasin B until no further quenching of intrinsic fluorescence is observed. Kd and x intercept of nCE. Fig. 1 illustrates the results of a series of cytochalasin B-induced fluorescence quenching measurements on a single band 4.5 preparation. Both radiolabel and fluorometric analyses are in agreement indicating that the stoichiometry of cytochalasin B binding to band 4.5 protein is 0.851 (probably 1:1, see Table   I). This stoichiometry is based upon protein determinations using the Lowry procedure (see below) and may not be as precise as previous estimates using amino acid analysis for protein determinations (20).

R is given by
Transport Determinations-D-Glucose efflux from IoVs was monitored by turbidimetry as described previously (8, 22). K, and Vmax for sugar exit were obtained from efflux records by use of the integrated, Michaelis-Menten equation for zero-trans sugar efflux (23).

Plotting In (St/So)/(So -S t ) versus t / ( S o -S t ) (where t is time and
subscripts 0 and t refer to time zero and time t respectively) produces a straight line with slope ( Vm,/K,) and y intercept -(l/K,,, + 1/P) where P is the osmolality of membrane-impermeant (non-glucose) species in solution. Recent objections to the use of the integrated Michaelis-Menten equation due to a suggested differential transport of a-and p-anomers of D-glucose (24) have proven to be without experimental basis (22). Experiments were performed using the stopped-flow spectrophotometer described in our previous studies (8, 14, 22) with data acquisition, temperature control, and data analysis under on-line control by computer (Apple II+). Only results from transport analysis in which least squares analysis of the transformed transport data provided correlation coefficients greater than 0.95 are reported here. Fig. 4 illustrates two typical efflux runs and their corresponding predicted time courses from the calculated K, and tal and theoretical time courses illustrates the suitability of the Vmax parameters. The very close correspondence between experimenintegrated rate analysis of time course data.
Analytical Methods-Triton X-100 assays were performed according to the method of Garewal (25). Protein assays were as described by Lowry et a1 (26). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was carried out on 10% gels as described previously (16). In instances where Eadie-Scatchard plots of quenching data indicated multi-component quenching (e.g. see Figs. Id, 3a, and 3b), two components were resolved by the method of successive approximation (27).

RESULTS
Fluorescence Quenching Measurements-When excited at 295 nm, purified sugar transporter displays a fluorescence emission spectrum characterized by two major peaks-one at 295 nm (due to scattering of the excitation beam) and one centered at 334 nm resulting from the tryptophan fluorescence of the protein (21). Upon addition of the transport inhibitor cytochalasin B, or transport substrate D-glucose, the emission peak at 334 nm is shifted by 1-3 nm toward shorter wavelengths and the peak height is reduced by about 10% (21). Addition of ligand to transporter can result in "quenching" of E 3 3 4 for a number of reasons. These include "artifactual" quenching (resulting, for example, from absorption of the exciting and emitted light by ligand) or "true" quenching, resulting from interaction of transporter with ligand. Unless stated otherwise, the data presented here represent true quenching of E334 produced by ligand/transporter interaction. Procedures for correcting artifactual quenching are described under "Experimental Procedures." In the case of cytochalasin B, ligand-induced fluorescence quenching is a simple (one-component) saturable phenomenon and (assuming a significant fraction of the ligand is bound; see     (Table I). D-Glucose increases Kd for cytochalasin B-induced quenching with an Kjapp of 32.1 & 3.7 mM (Table 11). Phloretin, a competitive inhibitor of sugar entry in human red cells (26), also induces quenching of band 4.5 protein intrinsic fluorescence. This is a simple saturable phenomenon with Kdapp of 0.96 f 0.08 p M (Fig. Ib). D-Glucose increases & for phloretininduced quenching with an Kiapp of 2.1 & 0.3 mM (Table 11).
Fig. IC shows the concentration dependence of D-glucoseinduced band 4.5 fluorescence quenching. These are raw quenching data corrected only for dilution effects encountered during titration of transporter against stock (1 M) D-glucose solution. L-Glucose is without significant effect upon E334 (Fig. IC). Upon first inspection, the D-glucose data appear to be approximated by a section of a single rectangular hyperbola. However, Eadie-Scatchard analysis of these data indicates two components of quenching-one mediated by a low Kd (1.7 f 0.3 mM) site and the second via a high Kd (25.3 f 1.2 mM) site. These values correspond closely to the Ktapp values for D-glucose inhibition of phloretin-and cytochalasin B-induced band 4.5 fluorescence quenching, respectively. This suggests that band 4.5 contains both the cytochalasin B-and phloretin-binding sites detected in intact cell transport studies (28). Moreover, asymmetry in D-glucose binding to the transporter appears to be an intrinsic property of the system (24). In intact cells, phloretin binds to an exofacial site on the sugar transport protein and cytochalasin B to an intracellular site (29). As competitions between phloretin and D-glUCOSe and between cytochalasin B and D-glucose are mediated by the low and high Kd D-glucose-binding sites, respectively, these data suggest that the low Kd substrate-binding site is associated with the external (influx) orientation of native transporter and the high Kd site with the internal (efflux) orientation of native transporter. Two-component saturable quenching was also observed using 4,6-ethylidene glucose (see Fig. 3a), a non-transported D-glucose analogue that binds to the extracellular substrate-binding site of the transporter with high affinity and to the intracellular site with much lower affinity (28). Other sugars known to act as substrate for the red cell hexose transport system also induce two-component quenching (Table 111). Two components of sugar-induced transporter fluorescence quenching were not observed in a previous study. However, the lowest substrate concentration employed in this study (21) was in the order of 10 mM, at which concentration the low Kd Bpp binding sites would be 85-90% saturated with sugars and, therefore, undetectable. It should be emphasized that as the low Kdapp is detected under conditions where dilution artifacts are minimal and substrate concentration are lowest, those systematic errors that influence steady-state fluorescence measurements (see above) are less likely to influence the measurements than under conditions (high ligand concentrations and greatest dilution factors) where the high Kdapp is detected.
Addition of ATP to purified transporter results in significant fluorescence quenching of the protein (see Fig. 2a).
Quenching is complete within 20 s (data not shown). Two lines of evidence suggest that ATP-induced fluorescence quenching is the result of specific ligand binding and not external quenching by nucleotide. 1) AMP, ADP, and ATP share almost identical absorption spectra (see "Experimental Procedures") yet quenching produced by ATP is considerably greater than that produced by either AMP or ADP (Fig. 2).  data for the inner filter effect results in the loss of the linear component of quenching leaving a simple saturable curve for ATP data alone with Kdapp of 51 pM. These considerations indicate that ATP-induced quenching is neither artifactual nor mediated solely via interaction of the transporter with the ribose moiety of the nucleotide. 2) Tripolyphosphate pentasodium (10 pM-1 mM) is without significant effect on E334 (Fig. 2b). These data indicate that quenching is ATPspecific and is not mediated solely by interaction of positively charged transporter domains with polyanionic species. Addition of 6 mM MgClz is without effect on the kinetics of nucleotide-induced fluorescence quenching (Table IV). The stoichiometry of ATP interaction with band 4.5 can be assessed indirectly from fluorescence quenching data using the procedures employed for analysis of the stoichiometry of cytochalasin B binding to band 4.5 (see Fig. la). Fig. 2c shows such an analysis. Here, the linear component of quenching produced by ATP was substracted from the total quenching by substracting AMP-induced quenching from ATP-induced quenching data. Band 4.5 protein was present as 25 pg in 2.1 ml. Over the range 50 nM-90 p~ ATP, the extrapolated concentration of ATP binding sites was 0.188 f 0.013 p~.
Assuming an average molecular size of 55 kDa for band 4.5 protein, this represents an ATP to band 4.5 binding ratio of 0.87 +. 0.06:1, a result indistinguishable from the ratio of cytochalasin B-binding sites per band 4.5 protein (Fig. l a ) and assuming each cytochalasin B-binding site represents a functional transporter molecule, this result suggests that each transport protein contains a single ATP binding/interaction site. Direct binding studies using labeled ATP are necessary to confirm this point. AMP-CPP and AMP-PCP (metabolizable and non-metabolizable ATP analogues, respectively) both appear to share ATP's ability to induce quenching but with 5-fold lower affinity (Kd = 230-250 p M ) and &fold lower maximum quenching (data not shown). The magnitude of fluorescence changes is low (2-3% following correction) and the calculated quenching parameters are unlikely to be as reliable as those obtained using ATP. ATP modifies the ability of substrate to induce fluorescence quenching of the transporter. Fig. 3A shows that ATP (500 p~) reduces the low KdaPp for EGlc-induced fluorescence quenching from 2.16 f 0.21 to 0.38 f 0.04 mM (Fig. 3A). The high &pp is increased slightly from 46.2 f 1.8 to 55.6 f 1.4 mM. Similar results were obtained using D-glucose (Fig. 3B). These  The non-metabolizable and metabolizable ATP analogues (AMP-PCP and AMP-CPP) appear to share the ability of ATP to induce low Kd and increase high Kd for sugar-induced quenching, but must be present at concentrations greater than 2 mM (data not shown). ATP was without effect on the Kdapp for cytochalasin B-induced quenching of transporter fluorescence ( n = 2, data not shown).
Sugar Transport Determinations-Sugar efflux from IOVs is, in principle, equivalent to sugar influx in right-side-out vesicles. Cytochalasin B is a noncompetitive inhibitor of efflux from IOVs as the cytochalasin B-binding site is exposed to  sugar-free medium in efflux experiments (29,30). In this current study, 5 p~ cytochalasin B reduced the rate of Dglucose efflux from IOVs by more than 99% (see Fig. 4). The measured permeability coefficient (PC) for D-glucose efflux from IOVs in the presence of cytochalasin B is 7.8 f 2.8 X lo-' cm/s ( n = 23; 20 "C) whereas PC for unrestricted efflux (efflux in the absence of cytochalasin B; V,.,/K,,, (30)) is 1.35 f 0.12 x cm/s ( n = 19). These values are similar to those reported for extensively washed red-cell ghosts (8, 30,31,32). As reported previously (8), efflux from IOVs resembles both efflux and influx in red cell ghosts but not influx in intact red cells. In terms of Michaelis-Menten parameters, V,,, and K,,, for cytochalasin B-inhibitable D-glucose efflux from IOVs (and influx in ghosts) are some 4-5-fold greater than Vmax and K,,, for influx in intact cells. However, if efflux from IOVs is monitored into red cell lysate (obtained by lysis of cells), the rate of glucose exit is reduced and K,,, and Vmax are reduced to values resembling K , and Vmax for influx into intact cells (8). Fig. 4 shows that ATP mimics the ability of red cell lysate to reduce the rate of D-glucose exit from IOVs. Analysis of these time course data using the integrated Michaelis-Menten rate equation reveals that ATP reduces K , and Vmax for cytochalasin B-inhibitable D-glucose efflux from IOVs (Fig.  5 B ) . This effect of ATP is half-maximal at about 44 p M ATP.
Addition of 6 mM MgC12 is without effect on the kinetics of ATP modulation of transport (Table IV). AMP, ADP, AMP-CPP, and AMP-PCP are unable to mimic the ability of ATP to reduce K,,, and Vmax for efflux from IOVs (Fig. 5C). ATP was without effect on cytochalasin B-insensitive D-glucose efflux (leakage) from IOVs (Fig. 4). As with the effects of cell lysate on sugar efflux from IOVs (8), the effects of ATP (0.5 mM) on K,,, for D-glucose exit from IOVs are seen immediately upon mixing IOVs and ATP (Table V) and are reversed immediately upon assay of D-glUCOSe exit from ATP-treated IOVs into ATP-free medium ( Table V). Effects of ATP on V, , , for exit from IOVs are expressed more slowly and are not reversed within 30 min of incubation of ATP-treated IOVs in ATP-free medium (Table V). DISCUSSION This study demonstrates that ATP modifies the ability of human red cell membrane IOVs to transport D-glucose. This effect is not brought about by nucleotide-induced alterations in nonspecific transmembrane D-ghCOSe flux (leakage) for cytochalasin B, a potent inhibitor of the catalytic activity of the red cell sugar transport protein (29,30), reduces the permeability of IOVs to D-glucose 2000-fold, leaving a residual flux (a leakage of less than 0.1% of control flux) that is unaffected by ATP. Moreover, ATP modifies the ability of Dglucose to induce quenching of the intrinsic tryptophan fluorescence of purified sugar transport protein. In both instances, these effects are half-maximal at about 50 p~ ATP, and indirect analysis suggests a probable stoichiometry of ATP transport protein interaction of 1:l. These findings establish that the catalytic properties of the red cell sugar transporter are subject to modulation by ATP and suggest the potential  for a mechanism of sugar transport regulation in cells that is fundamentally different to the well-documented, insulin-dependent recruitment of intracellular hexose transporter to plasma membrane (1-4). As these effects are observed in the human red cell, such a mechanism must necessarily involve modification of the catalytic properties of transport proteins located within the plasma membrane.
Although it has been suggested that a higher molecular weight protein and not band 4.5 may be the native sugar transport protein (33), this current study shows that band 4.5 protein contains the transport substrate-binding sites (exposed at endo-and exofacial orientations of the transporter), the normally endofacial cytochalasin B-binding site, the normally exofacial phloretin-binding site, and an ATP-binding site. These sites are integral components of the red cell sugar transport system (28, see also above). In addition, it is known that reconstitution of band 4.5 into artificial membranes results in the reconstitution of sugar transport activity with kinetic properties (Michaelis constants and turnover numbers, exchange fluxes) that closely resemble the properties of native sugar transport (24). These findings further strengthen the view that band 4.5 protein contains the red cell sugar transporter (12).
Sugar transport in intact human red cells displays catalytic asymmetry (24,28,30). V,,, and K,,, for glucose influx are some 5-10-fold lower than the corresponding values for efflux (28). It has been suggested that the complexation of cell water and D-gluCOSe by hemoglobin could result in apparent sugar transport asymmetry in red cells (28). Transport asymmetry is reduced or even lost in red cell ghosts (8, [30][31][32]. The net effect is increased VmaX and K, for sugar entry and reduced K , for sugar exit (8). This is a reversible phenomenon. Relysis of ghosts followed by resealing in the presence of high concentrations of cell lysate restores transport asymmetry (8). These effects seem not to be brought about by the presence of hemoglobin in cell lysate but rather by the presence of low molecular size factors ( 4 0 kDa) (8),. This present study supports the view that ATP may be such a factor, because the nucleotide mimics the ability of cell lysate to reduce both K , and V,,, for D-glucose efflux from IOVs. Half-maximal inhibition of sugar efflux from IOVs by cell lysate was produced at a lysate dilution of approximately x45 (8). If we assume the red cell ATP content is 2.24 mmol/l of cell water (9), then half-maximal effects would be observed at 50 PM ATP. This value is very close to that ATP concentration producing halfmaximal inhibition of VmaX and K,,, for sugar exit from IOVs (44 PM), half-maximal quenching of band 4.5 protein tryptophan fluorescence (50 p~) , and half-maximal reduction in the low Kdapp for D-glucose-induced fluorescence quenching of band 4.5 (45 PM). In a previous study of the effects of ATP on sugar influx in human red cells, it was reported that ATP stimulated sugar entry (9). However, no effect of red cell lysis and resealing upon D-glucose entry was detected (9). The reasons for these discrepancies remain unclear. It is interesting that two ATP-sensitive D-glucose-binding sites are detected in fluorescence measurements made with band 4.5 protein. The Kdapp for binding to these sites are approximately 1 and 25 mM D-glucose, and these sites appear to correspond to the sugar influx and efflux orientations of native transporter, respectively. These sites (low and high Kdapp sites) are also detected in red cell ghost membranes stripped of peripheral membrane proteins (34). The question arises as to whether these sites exist simultaneously on the transport molecule or whether the occupation of one site by substrate precludes the occupation of the second site by substrate. This is an important question because to date, onesite sugar transport models have been uniformly unsuccessful in predicting the complex kinetic features of human red cell sugar transport (28). The one-site model (with respect to sugar-induced fluorescence quenching) may be described by the scheme shown below.
where X is the transporter which may only bind a single sugar at either an internal or external site at any point in time, and aq and pq refer to quenching produced by the XG, and XGi complexes, respectively. For a non-transported sugar (e.g. EGlc or maltose), sugar-induced quenching, q is given by where Qm is maximum theoretical quenching and S replaces Gi and Go (the protein/lipid particles are freely permeable to sugar). Substituting values for KA and Ks (1 and 25 mM) and varying a and parameters results in a series of parallel lines in Eadie-Scatchard plots with no indication of two-component quenching kinetics. The two-site model, where two sugars can bind simultaneously to the carrier, is shown below.
Here, the transporter X can bind sugars simultaneously at internal and external sites. aq, Pq, and yq refer to quenching produced by XG,, XGi, and XGiGo complexes, respectively. For a non-transported sugar, sugar-induced quenching, q, is given by the following.
Substituting values for K, and KB and varying a, P, y, and a parameters results in a series of plots in Eadie-Scatchard analyses, some with no indication (e.g. when a = ,6 = y = d = 1) and others with clear indications of multi-component quenching. Multi-component quenching is predicted both by a system of two identical binding sites (KA = KB) with strong negative cooperativity (a > 1) and by a system of two different binding sites with insignificant interaction (8 = 1; K A C KB).
With the former, the fluorescence quenching produced by formation of the binary and ternary complexes of X , Gi, and Go can be identical. With the latter system, quenching produced by formation of the ternary complex must be greater than that produced by formation of the binary complex of X and G. It is not possible, using the available experimental data, to distinguish between these possibilities. We may conclude, however, that the one-site model is inconsistent with the experimental data. This present study does not support the view that the transporter is of the one-site type. However, two previous studies have provided evidence in favor of the one-site model (29,35). The study of Krupka and Deves (29) modeled inhibitions of red cell sugar transport (mediated by both one-and two-site kinetic schemes) produced by phloretin and cytochalasin B. They demonstrated that transport was consistent with the one-site model. They also described conditions under which their conclusions would be invalidated. These conditions require that phloretin and cytochalasin B cannot bind to the transporter simultaneously and that binding of sugar to one of the two sites on the two-site carrier does not prevent inhibitor or sugar binding to the second. It has previously been shown that phloretin inhibits cytochalasin B binding to purified transporter (36), and this present study supports the view that binding of sugar to one binding site does not prevent binding of inhibitor or sugar to the second. The criteria necessary to reject their conclusion of one-site kinetics are, therefore, fulfilled. Gorga and Lienhard (35) adopted a different approach to this problem. Measuring cytochalasin B binding to red cell membranes stripped of peripheral proteins they found that ethylidene glucose (EGlc-a non-transported sugar that binds preferentially but not exclusively to the external or orientation of the transporter) competitively inhibited binding with an Kiapp of 26 mM. They concluded that EGlc can displace CCB bound to the internal site by binding to the external orientation of the one-site transporter. In this study, it is shown that two EGlc-binding sites are detected in band 4.5-one with an Kd of 2 mM and a second with Kd app of 30 mM. In red cell membranes stripped of peripheral proteins, these values are 1 and 26 mM (34). It is possible, therefore, that the observed EGlc-inhibition of CCB binding to the transporter was mediated via the high Kdapp binding site (K,,,, would be -30 mM) and EGlc binding to the high affinity, low Kdapp site was without effect on CCB binding. In support of this possibility, the predictions of a two-site model in which Gi and CCB compete for binding to an internal site but in which Go and CCB do not compete for binding have been derived. The basic scheme is shown below.  Gorga and Lienhard (35). Values were also calculated for inhibition of CCB binding by propylglucoside, a sugar that binds preferentially to the interior orientation of the transporter with an Kdapp of about 30 mM. These calculations serve to illustrate how the two-site model can also account for the available experimental data, provided we assume that external sugar does not inhibit CCB binding to the transporter. The Kdapp for EGlc binding to the external-facing orientation of the transporter (1.9 mM in this study) is considerably lower than the measured Kiapp for inhibition of D-glUCOSe transport by external EGlc. Extracellular EGlc inhibits erythrocyte D-glUCOSe exchange transport with an KIBpp of 11 mM at 16 "C (37) and 7 mM at 25 "C (38). However, the true Ki for inhibition is governed by the mechanism by which transport occurs (37). For example, with the mobile carrier (onesite model), the true K; for inhibition of exchange by external EGlc e Kiepp. With the two-site model (where a 10-fold asymmetry is assumed) Ki = Ki.,,/6 (37). These considerations illustrate how interpretations of transport data are governed by the presumed mechanism of transport. As the The major enigma surrounding red cell transport centers upon the observation of two Michaelis constants for sugar exit but only one for sugar entry (28). One-site transport models cannot account for these features of sugar transport. It has been suggested that the problem lies not with the models but rather with the experimental determination of sugar transport kinetics (24, 35). Specifically, it has been proposed that previous transport determinations are in error, possibly due to the differential transport of a-and /3-D-glucose by the human red cell (24, 35). A number of direct, experimental observations argue against this hypothesis. 1) The predicted multi-component washout of D-glucose from loaded red cells due to the differential transport of D-glUCOSe anomers is not observed experimentally (39). 2) a-and 8-D-glucose are transported identically by the intact human red cell (22). 3) Human red cell sugar transport complexity (two Michaelis constants for exit and one for entry) is lost in red cell ghosts (8) and may be restored upon reincorporation of red cell lysate (8) or ATP (34) into ghosts. 4) Previous sugar transport determinations made in a variety of laboratories using a variety of experimental techniques and procedures are in substantial agreement (28). It appears, therefore, that human red cell sugar transport complexity is a real phenomenon that may, in some way, be due to modulation of the intrinsic properties of the transporter by factors (possibly ATP) present in red cell cytosol. In addition, the inability of one-site transport models to account for human red cell sugar transport is sufficient to reject their use in kinetic descriptions of the transfer mechanism.

KL
A two-site transport model for human red cell sugar transport has been proposed (34) in which ATP interacts with the transporter (in as yet unspecified fashion). This model predicts the following properties of the transport system. 1) Asymmetry in Vmax and K , parameters for sugar entry and exit in fresh cells and ATP-loaded ghosts; 2) the presence of two Michaelis parameters for sugar exit and one for entry in fresh cells and ATP-loaded ghosts; 3) Transport symmetry in ATP-free ghosts (only one Michaelis parameter for exit and entry); 4) increased K , for equilibrium-exchange D-glucose transport in ATP depleted red cells. This model is consistent with both available transport data and ligand binding data. However, the transport system is not of the rapid equilibrium type. As the KdaPp parameters for sugar binding to internal and external orientations of the transporter are widely different (see above), yet K , and V,,, for influx and efflux in ghosts are identical, we must conclude that the Michaelis and velocity constants measured in transport determinations are the product of both substrate binding and translocation constants (e.g. see Ref. 40). Such a model requires steady-state analysis and is of the Hybrid Ping Pong Random type (40) with a central Random Bi Bi segment to account for exchange fluxes and two Ordered Uni Uni segments to account for the partially reactive zero-trans fluxes (transport in the absence of sugar at the opposite side of the membrane). In addition, steps must be included to account for the effects of ATP on transport. ATP acts to increase K , for exit, reduce K , and V,,, for (1.9 -I-0.3 mM). entry, and to reduce K , for transport under exchange conditions.
What is the molecular mechanism of ATP action of sugar transport? Although ATP modulates the properties of purified transporter with almost identical potency to its modulation of sugar transport in IOVs, it is not yet possible to ascertain whether ATP modification of transport reflects direct transporter/nucleotide interaction or whether intermediate species (e.g. membrane and cytoskeletal kinases) are also involved. Certainly, purified transporter is not phosphorylated by ATP in the absence of exogenous kinase but is capable of acting as substrate for protein kinase C (41). In addition, the transporter is phosphorylated in vivo in the absence of phorbol ester and exogenous protein kinase C (41). In view of the inability of AMP and ADP to mimic ATP inhibition of sugar transport in IOVs, the rapid onset and reversibility of ATPaction on K , for transport and the somewhat slower onset and delayed recovery of ATP action on Vmax for transport, it is interesting to speculate that phosphorylation of the transporter results in reduced catalytic activity (Vmax) and that simple allosteric interaction with ATP modulates K,,, for transport (see Randle and Smith (5)). In the absence of additional experimental evidence, the elucidation of the mechanism of this recruitment-independent form of sugar transport regulation must await further study. Although the red cell may never encounter physiological conditions in which intracellular ATP levels fall below 50 p~, these findings may be of physiological significance in muscle and avian erythrocytes where, as with the human red cell, ATP depletion results in insulin-independent sugar transport stimulation (5,6, 42).