a family of integral membrane proteins called glucose transporters (GLUTs) (40) mediates equilibrative sugar transport in mammalian cells. The glucose transport protein GLUT1 catalyzes sugar transport in cells of the reticuloendothelial system (7, 61) and presents an interesting experimental puzzle. The steady-state kinetics of GLUT1-mediated sugar transport in rabbit (70), rat (38, 62), and avian (7, 8) erythrocytes and in basal (insulin-starved) rat adipocytes (76) are consistent with classical models for carrier-mediated solute transport (5, 47). GLUT1-mediated sugar transport in human red cells, however, displays a kinetic complexity that has proven difficult to reconcile with models for carrier-mediated transport (4, 21, 32, 49, 56, 79).

Transport complexity is especially obvious in zero-trans exit and infinite-cis entry conditions (13). In the zero-trans exit condition, cells are loaded with various starting sugar concentrations and the initial rate of exit is measured (49, 56), or the complete time course of exit is analyzed by using an integrated Michaelis-Menten equation (4, 16, 42, 56). Initial rate measurements (49, 56) routinely provide estimates of apparent K_m [K_m(app)] for sugar exit that are two to three times lower than those obtained by analysis of the complete time course of sugar exit (4, 16, 42, 56). In the infinite-cis sugar uptake experiment, the external sugar level is saturating and the concentration of intracellular sugar that reduces net sugar uptake by one-half is measured. K_m(app) for infinite-cis entry is routinely 5- to 10-fold lower than values predicted by classical carrier models for sugar transport (13, 14, 21, 22).

Why is it that human red cell sugar transport displays kinetic complexity, whereas GLUT1-mediated sugar transport in rabbit, rat, and avian erythrocytes and in rat adipocytes is consistent with sugar transport models? Human, rat, and rabbit GLUT1 share 98.4% identity (3, 7, 61). Six common residues in rabbit and rat GLUT1 diverge from human GLUT1 sequence, but all six substitutions show positive scores on the point-accepted mutation similarity scale (1), suggesting that each substitution is unlikely to affect structure or function. If sequence divergence is not the cause of transport complexity, GLUT1 phenotypic variation must result from differences in cellular environment, GLUT1 expression levels, or artifacts of measurement.

The sugar transport capacity of human red cells is 220- to 10,000-fold greater than that of rat basal adipocytes (76), rat red blood cells (38), and avian erythrocytes (24). Naftalin and Holman (63) discussed several ways by which this could give rise to transport complexity. 1) Transport measurements in human red blood cells are technically challenging even at low temperatures owing to the very high GLUT1 density of human blood red cells and the high catalytic turnover of GLUT1 (49). Significant back flux of imported sugar during the course of a transport determination would lead to underestimation of net import. 2) If net cellular sugar import were composed of two steps (transport followed by intracellular diffusion/distribution), the diffusional step could become rate limiting if the transport step were sufficiently rapid. Evidence for nonuniform intracellular distribution of sugars has been obtained in both human and rat erythrocytes (4, 36, 38, 62, 64).

According to the diffusional barrier hypothesis, human red cell net sugar import is composed of rapid transport (owing to high cellular GLUT1 content) and slow intracellular diffusion and/or distribution. The overall result is one where net sugar import is rate limited by intracellular diffusion/distribution and not by transport. Measurements of sugar uptake in rat or rabbit erythrocytes, however, largely reflect low-capacity GLUT1-mediated transport and thus provide a more accurate description of the intrinsic properties of GLUT1.

This hypothesis infers that GLUT1-mediated sugar transport is inherently simple but that operational complexity is caused by factors extrinsic to the transport system. Transport complexity is lost in cytosol-depleted human red cell ghosts (16, 20, 37, 41) but is preserved if ATP is included in artificial cytosol during red cell ghost resealing (12, 37). Does this mean that human GLUT1 is uniquely ATP sensitive? Human GLUT1 is an ATP-binding protein (15, 22) and contains three sequence motifs that form the ATP-binding pocket of human adenylate kinase (27). One of these domains (GLUT1 residues 332–338) has been identified by peptide mapping, by microsequencing of proteolyzed, azidoATP-photolabeled human GLUT1 (43), and by scanning alanine mutagenesis (44) as a domain critically involved in nucleotide-GLUT1 interaction. This domain is unchanged in rabbit and rat GLUT1 sequences where there is no a priori reason to assume that it does not serve a similar function. It is not known, however, whether rabbit or rat red cell sugar transport is modulated by altered intracellular ATP. Parental (hamster) and exogenous (human) GLUT1-mediated sugar transport are insensitive to cellular ATP depletion in Chinese hamster ovary cells (22, 44), whereas ATP-sensitive, human GLUT1-mediated sugar transport is observed in transfected HEK cells (22, 44) and in clone 9 cells (73). This suggests that cellular environment influences GLUT1 phenotype.

These observations stimulate two questions. 1) Is human red cell sugar transport complexity an artifact of measurement? 2) If complexity is real, do human erythrocytes and clone 9 and HEK cells uniquely express cellular factors that partner in ATP modulation of sugar transport? We address the former question in the present study. Human red cell membrane GLUT1 content approaches 10% total protein by mass (36), resulting in extremely rapid sugar transport rates at all temperatures studied (8). Measurements of transport are, therefore, complicated by the necessity for rapid sampling procedures (8, 49) but may be simplified by use of the equilibrium exchange condition (77). In equilibrium exchange, intracellular sugar concentration = extracellular sugar concentration, no net transport occurs, and unidirectional sugar fluxes are measured by addition of radiotracer sugar either to the external or to the internal medium. In principle, radiotracer fluxes under these conditions are first order and thus monoexponential in a uniform population of cells. We have used this condition to investigate whether sugar transport in red blood cells reflects the properties of transmembrane sugar flux alone or whether steps subsequent to transmembrane flux also contribute to overall transport behavior.

MATERIALS AND METHODS

Materials.

³H-labeled 3-O-methylglucose (3MG), ¹⁴C-labeled 3MG, [³H]uridine (Urd), and [¹⁴C]Urd were purchased from Sigma. Rabbit antisera raised against a synthetic carboxy-terminal peptide of GLUT1 (C-Ab) were obtained from East Acres Biologicals. Sheep antisera raised against tetrameric GLUT1 (δ-Ab) were prepared as described previously (34). Fluorescein-conjugated goat anti-rabbit antibody was purchased from Molecular Probes. Fluorescein-conjugated goat anti-sheep antibody was purchased from Calbiochem. Human blood was purchased from Biological Specialties. Other reagents were purchased from Sigma.

Solutions.

Kaline consisted of (in mM) 150 KCl, 5 MgCl₂, 5 EGTA, and 5 HEPES (pH 7.4). Lysis buffer contained 10 mM Tris·HCl and 2 mM EDTA (pH 8.0). Stripping solution contained 2 mM EDTA and 15.4 mM NaOH (pH 12). Sugar-stop solution consisted of ice-cold kaline containing 20 μM CCB and 200 μM phloretin. Urd stop solution consisted of ice-cold kaline containing 50 μM S-nitrothioinosine.

Red blood cells.

Red blood cells were isolated by washing whole human blood in four or more volumes of ice-cold kaline and centrifuging at 10,000 g for 15 min at 4°C. Serum and buffy coat were removed by aspiration, and the wash, centrifugation, aspiration cycles were repeated until the buffy coat was no longer visible. Cells were resuspended in four volumes of sugar-free or sugar-containing kaline and incubated for 1 h at 37°C to deplete or load intracellular sugar.

Red cell ghosts.

Ghosts were hypotonically lysed by resuspending washed red blood cells in 10 volumes of ice-cold lysis buffer for 10 min. Membranes were harvested by centrifugation at 27,000 g for 20 min. Ghosts were repeatedly washed with lysis buffer and centrifuged until the membranes appeared light pink (∼3 cycles). Ghosts were then washed with 10 volumes of ice-cold kaline and collected by centrifugation at 27,000 g. Harvested membranes were resealed by incubation in four volumes of kaline ± 4 mM ATP (37°C ) for 1 h and collected by centrifugation at 27,000 g for 15 min at 4°C. Resealed ghosts were stored on ice until use.

Net 3MG uptake.

Sugar-depleted cells or ghosts were incubated in 20 volumes of ice-cold kaline containing 100 μM unlabeled 3MG and 0.5 μCi/ml labeled 3MG. Uptake was allowed to proceed for intervals as short at 6 s to intervals as long as 3 h. Uptake was arrested by addition of ice-cold stop buffer, and ghosts were centrifuged at 14,000 g for 1 min. The supernatant was removed by aspiration, and ghosts were washed with 20 volumes of sugar stop buffer and recentrifuged, and supernatant was aspirated. The ghost pellet was extracted with 500 μl of 3% perchloric acid and centrifuged, and samples of the clear supernatant were counted in duplicate. Zero time points were collected by the addition of sugar stop solution to ghosts followed by uptake medium. Samples were then immediately processed. Radioactivity associated with cells at zero time was subtracted from all nonzero time points. Equilibrium time points were collected with an overnight incubation. All time points were normalized to the equilibrium time point. All solutions and tubes used in the assay were preincubated on ice for 30 min before the start of the experiment. Triplicate samples were processed for each time point.

Initial rate, zero-trans 3MG uptake: time dependence.

Sugar-depleted cells or ghosts were allowed to rest in 20 volumes of ice-cold kaline for intervals as short as 30 s or as long as 3 h. Thirty seconds before the end of the rest interval, a small volume of 3MG and 0.5 μCi/ml tracer 3MG were added to the suspension to a final concentration of 3MG of 100 μM. Uptake was stopped by addition of sugar stop solution, and cells or ghosts were processed as above. A zero time point was obtained as described above. An equilibrium time point was obtained by allowing ghosts to uptake unlabeled and labeled 3MG overnight at 4°C. Uptake rates were expressed as micromoles per minute per liter cell water by subtracting the zero time point and normalizing to the equilibrium time point.

3MG equilibrium exchange, uptake, and exit.

Ghosts were resealed in the presence of 0.1, 2.5, 10, or 20 mM unlabeled 3MG and trace amounts (0.5 μCi/ml) of [¹⁴C]3MG and centrifuged, and the supernatant was aspirated. Ghosts were then incubated in 20 volumes of ice-cold kaline containing the same concentration of 3MG and 0.5 μCi/ml [³H]3MG. Exchange was allowed to proceed for time intervals as short as 6 s to intervals as long as 10 h. Exchange was then stopped by the addition of ice-cold sugar stop, and ghosts were treated as above with zero-trans uptake.

Urd equilibrium exchange, uptake, and exit.

Experiments were performed as with 3MG equilibrium exchange except that the transport substrate was Urd, and Urd stop was used to stop the reaction and to wash the ghosts.

GLUT1 immunofluorescence microscopy and fluorescence-activated cell sorting analysis.

For C-Ab immunofluorescence staining, 50 μl of nonfixed unsealed ghosts were washed once in ice-cold kaline and centrifuged at 4°C, and the supernatant was aspirated. A 1:500 dilution of C-Ab was added to the pellet, the suspension was incubated at 4°C for 2 h and centrifuged, and the supernatant was aspirated. Ghosts were washed with ice-cold kaline. Secondary antibody (1:500 dilution) was added to the pellet, and the suspension was incubated at 4°C for 1 h and centrifuged, and the supernatant was removed by aspiration. Ghosts were washed five times with ice-cold kaline and brought to a final volume of 500 μl. For δ-Ab staining, resealed ghosts were incubated with a 1:500 dilution of δ-Ab at 37°C for 1 h and then processed as above. For microscopy, 10 μl of the suspension were added to a polylysine-coated coverslip and adhered by low-speed centrifugation. Coverslips were washed five times with 10 ml of ice-cold kaline. Fluorescence microscopy was performed on an Olympus BX-51 fluorescence microscope. For fluorescence-activated cell sorting (FACS) analysis, 50 μl of the immunofluorescence-stained ghosts were mixed with ∼1 ml of FACS buffer in a FACS sample tube. FACS analysis was performed with a Becton Dickinson FACS scan sampled for 10,000 counts and analyzed with Cell Quest 3.3.

Ghost-accessible volumes.

Ghosts were prepared and resealed with and without intracellular ATP. After ghosts were resealed, 2.5 mM 3MG or Urd and tracer [³H]3MG or [³H]Urd and [¹⁴C]sucrose were added to the ghosts, and the radiotracers were allowed to equilibrate with cell water. The samples were centrifuged at 14,000 g for 1 min, and known volumes of supernatant and pellet were sampled and counted for both isotopes. Ghosts were counted on a hemocytometer. Tracer accessible volume (V_acc) was calculated as

where ¹⁴C and ³H refer to the respective tracers (dpm), V is the sample volume, and subscripts S and P refer to supernatant and pellet, respectively. In this manner, the extracellular volume of the pellet is subtracted by measuring the sucrose space. Alternatively, ghosts were resealed with 2.5 mM 3MG and 2.5 mM Urd plus [³H]3MG and [¹⁴C]Urd or [¹⁴C]3MG and [³H]Urd radiotracers. At 0 and 15 h after resealing was completed, ghosts were sedimented by centrifugation and the supernatant was sampled. The ghost pellets were washed twice with 20 μM CCB, 200 μM phloretin, and 50 μM S-nitrothioinosine in ice-cold kaline. Known volumes of supernatant and pellet were counted for 3MG and Urd, and pellet-to-supernatant ratios were computed.

Red blood cell and ghost electron microscopy.

Scanning electron microscopy (SEM) of red blood cells and ghosts was carried out as previously described (35). Red blood cells and ghost cells were fixed by immersion in 2.5% (vol/vol) glutaraldehyde in 0.5 M sodium phosphate buffer (pH 7.2) for 1 h at room temperature. Fixed samples were then washed three times in the same buffer. After the third wash, the cells were postfixed for 1 h in 1% osmium tetroxide (wt/vol) in the phosphate buffer, washed three times in buffer, and left overnight at 4°C. Samples were dehydrated through a graded series of ethanol to 100%, and then the bottoms of the tubes were excised, place in porous sample holders, and critical point dried in liquid CO₂. After the drying step, tubes were emptied onto aluminum SEM stubs coated with adhesive carbon tape. The edges were then painted with silver conductive paste, and the SEM stubs were sputter coated with Au/Pd (80/20). The specimens were then examined by an ETEC autoscan scanning electron microscope at 20-kV accelerating voltage.

Circular dichroism.

Freshly dissolved α-3MG or equilibrium solutions of 3MG (each 20 mM) were prepared in kaline, and the circular dichroism of the sugar was monitored. Data were collected at 4°C by a Jasco J-810 spectropolarimeter and a 0.1-cm pathlength cuvette.

Proton NMR.

Data were collected at 24°C on a 400-MHz Oxford NMR.

Curve-fitting procedures.

Where appropriate, data sets were analyzed by nonlinear regression using the software packages proFit (version 6.06; Quantum Soft, Uetikon am See, Switzerland) or Igor-Pro (version 5; Wavemetrics, Lake Oswego, OR). Numerical integration was performed by using Berkeley Madonna X (version 8.3.14).

RESULTS

Effect of ATP on 3MG equilibrium exchange transport.

The effect of intracellular ATP on equilibrium exchange 3MG transport at 4°C was monitored over the course of 5 h by measuring the simultaneous uptake of tracer [³H]3MG and exit of [¹⁴C]3MG (Fig. 1). Resealed ghosts lacking intracellular ATP show an exchange time course that follows a single, simple exponential rise or decay for uptake and exit, respectively. This is expected for a passive transport process characterized by a single rate-limiting step and occurring in a uniform population of cells. When ghosts are resealed with 4 mM intracellular ATP, however, the time course changes, becoming biphasic for both unidirectional uptake and exit. The early phase of transport is accelerated fourfold in the presence of ATP. After 10 min when one-half to two-thirds of unidirectional exchange is complete, the rate of transport declines by more than 20-fold to a rate approaching 10-fold slower than transport in the absence of ATP. Table 1 summarizes fit parameters for ghosts lacking and containing ATP. Equilibrium exchange time courses have been studied previously by this laboratory, but biphasic transport was observed only at the lowest 3MG concentration employed (0.1 mM) (23), where biphasic exchange may be related to GLUT1 sugar binding. That study did not monitor transport beyond 5 min, which explains why biphasic equilibrium exchange transport at higher 3MG concentrations was not observed.

One explanation for biphasic transport in the presence of ATP is that a significant subpopulation of ghosts spontaneously reseal before ATP is introduced so that the measurements sample a mixed population of ATP-containing (fast) and ATP-lacking (slow) ghosts. This simple hypothesis is refuted by the observation that the slow phase of 3MG transport in ATP-containing ghosts is nearly 10 times slower than transport in ATP-free ghosts.

Red cell ghosts present a relatively uniform population of cell sizes and GLUT1 content.

A mixed population of ghost cell sizes or GLUT1 content could account for biphasic 3MG exchange transport. As cell size decreases, the surface area-to-volume ratio increases, thereby increasing the rate of GLUT1-mediated equilibration of radiotracer sugar. Altered GLUT1 content (or activity) at the membrane surface also determines the rate constant for radiotracer equilibration.

GLUT1 content was assayed by immunofluorescence microscopy of fixed or living cells and by FACS. Figure 2, A–C, shows fluorescence micrographs of ghosts stained with antibodies directed to cytoplasmic (Fig. 2A) or extracellular (Fig. 2, B and C) GLUT1 epitopes. The use of cytoplasmic epitope-directed antibodies (C-Ab) necessitates cell fixation/permeabilization. FACS analysis of GLUT1 staining by exofacial antibodies (δ-Ab) suggests a uniform population of cells ± ATP (Fig. 2E). C-Ab staining of unsealed ghosts ± ATP produces similar results (data not shown). Cell size was quantified by FACS analysis of single cell light scattering. Ghosts show a single population of scattering intensities as detected at either low (data not shown) or high angles (Fig. 2D) and scattering is independent of cytoplasmic ATP concentration.

Effect of ATP on human equilibrative nucleoside transporter 1-mediated Urd exchange transport.

Previous models have rationalized sugar transport complexity by hypothesizing an unstirred layer beneath the red cell membrane (4, 62, 63). Although predicted deviations from simple Michaelis-Menten kinetics at low 3MG concentrations are observed for sugar exit in human and rat red blood cells (36, 38), direct evidence for this unstirred layer is not available.

Human equilibrative nucleoside transporter-1 (ENT1)-mediated Urd transport was examined to determine whether the putative unstirred layer is sugar transport specific. Figure 3 shows the time course of 2.5 mM Urd exchange in ghosts ± ATP. Although there are 100-fold fewer nucleoside transporters per red blood cell than there are glucose transporters, K_m for Urd transport is ∼100 times lower than K_m(app) for GLUT1-mediated 3MG transport (44). The apparent rate constant for transport of any species (k) is proportional to the ratio of V_max to K_m(app) (75); hence, sugar transport and Urd transport proceed at similar rates. Urd uptake and exit are monoexponential in ghosts regardless of cellular ATP content (Fig. 3). If an intracellular unstirred layer or diffusional barrier does exist, it must be selective for GLUT1 substrates. Monophasic Urd exchange transport in ghosts lacking and containing ATP also argues against multiple cell sizes in resealed ghosts.

Effect of intracellular ATP on accessible sugar and Urd volumes.

Table 2 summarizes two experiments in which the intracellular 3MG and Urd spaces of red cell ghosts were measured in the absence and presence of 4 mM intracellular ATP. 3MG and Urd spaces are identical and are unaffected by inclusion of 4 mM ATP during resealing. Differences between experiments may result from cell counting errors caused by low contrast of red cell ghosts in the hemocytometer. Because transport experiments span several hours, the time dependence of cell volumes was also investigated. The normalized internal-to-external ratios of 3MG and Urd concentrations ± ATP at 0 and 15 h are shown in Fig. 4. There is no change in the accessible volume over that time.

Effect of preincubation time on initial rates and time courses of transport.

We have found that ATP levels are reduced by ∼50% during the 5 h of an exchange time course experiment (data not shown). Although AMP and ADP do not directly affect transport, they serve as competitive inhibitors of ATP binding to GLUT1 (15). It is possible, therefore, that nucleotide-dependent changes in GLUT1 intrinsic activity occur during the time course of our measurements. To examine this possibility, both the time course of net 3MG uptake and the instantaneous rate of 3MG uptake were measured in ATP-free and ATP-containing ghosts before and after 5 h of mock (sugar-free) transport. Figure 5A shows that ATP modulation of zero-trans uptake persists after 5 h. No change in the instantaneous (initial) rate of zero-trans 3MG uptake is observed after 5 h (Fig. 5B). Biphasic exchange transport persists in ghosts that have previously exchanged 3MG for 15 h before they are assayed for radiotracer exchange (Fig. 5C). These data indicate that time-dependent changes in the intrinsic activity of GLUT1 do not account for biphasic sugar transport.

ATP modulation of red cell ghost morphology.

Red blood cells and ghosts containing or lacking ATP were examined by scanning electron microscopy. Figure 6 shows the morphological changes induced by ATP depletion. Red cell membranes become spherical and highly crenated after ghosting in the absence of ATP. Ghost resealing in the presence of MgATP reverses this effect.

DISCUSSION

Human erythrocyte sugar transport displays a kinetic complexity that is not explained by available models for carrier-mediated facilitated diffusion (4, 21, 32, 49, 57, 79). Rat, rabbit, and pigeon erythrocyte sugar transport are, by contrast, compatible with carrier models (38, 62, 70, 74). Several factors suggest that human red blood cell sugar transport complexity is directly attributable to the very high GLUT1 content of human erythrocytes, which, because the resulting sugar transport is extremely rapid, contributes to inaccuracies in sugar flux determinations (4, 14, 63). Studies with resealed erythrocyte ghosts further suggest that cytosolic factors contribute to transport complexity (11, 12, 16, 37).

Naftalin and Holman (63) proposed that an unstirred layer at the cytosolic surface of the membrane would increase K_m(app) for rapid sugar exit because, during the course of an experiment, the concentration of sugar beneath the membrane falls more precipitously than total intracellular sugar. Evidence for an unstirred layer comes from three sources. Net 3MG uptake in human red blood cells displays biphasic kinetics, suggesting rapid equilibration with an unstirred layer and slower equilibration with bulk cytosol (23). Steady-state sugar exit in human red blood cells deviates from Michaelis-Menten kinetics at low intracellular sugar concentration in a way that is characteristic of an intracellular unstirred layer (10, 28). Counterflow experiments in rat red blood cells suggest a nonuniform distribution of unlabeled and labeled sugars within the cell (38, 62).

The unstirred layer hypothesis was modified after the discovery of an ATP-dependent, high-affinity GLUT1 sugar binding site (36), which, it was hypothesized, might also explain biphasic sugar uptake. Sugar translocation and subsequent binding at an endofacial GLUT1 domain account for the fast phase of transport. The slower phase of transport was ascribed to slow release into cytosol, a process limited by a diffusional barrier or cage formed by endofacial GLUT1 domains (COOH terminus and cytoplasmic loop-6). Although delayed within the cage, the probability of sugar reassociation with the exit site and of translocation back to interstitium is high. The net effect is reduced net sugar import. This model predicts that the size of the fast transport component in any cell is proportional to GLUT1 concentration. Transport complexity, however, should be independent of cellular GLUT1 concentration because diffusionally restricted sugar release results from properties intrinsic to each GLUT1 molecule. The nominal “absence” of transport complexity in rat, rabbit, and avian red blood cells refutes this hypothesis.

To eliminate complexities inherent to zero-trans experiments, we measured transport under equilibrium exchange conditions where no net change in sugar concentration occurs. Several studies have examined GLUT1-mediated sugar equilibrium exchange in detail. Weiser et al. (77) measured exchange uptake and exit in both cold-stored and fresh human red blood cells. They concluded that exchange kinetics were monophasic but used early time points (<30 s) in their analysis and thus missed the second slower phase. Previous studies by this laboratory (23), which used times of ≤5 min, demonstrated biphasic sugar exchange transport in intact red blood cells but only at the lowest 3MG concentration tested (<1 mM). This led to the conclusion that biphasic transport was due to sugar binding in the cell. However, we now know that detection and analysis of the slower phase of transport at 4°C require measurements beyond 5 min; this prompted us to extend our analysis of equilibrium exchange transport to later time points.

When red cell ghosts are nominally ATP free, radiotracer 3MG equilibrium exchange follows simple monophasic kinetics. Exchange in ATP-loaded red cell ghosts shows biphasic exchange kinetics at all 3MG concentrations tested. Neither fast nor slow phases of exchange transport can be attributed to sugar complexation by GLUT1 because GLUT1 is at 20 μM whereas 3MG ranges from 0.1 to 20 mM. ATP-dependent deviation from monophasic exchange kinetics is not due to heterogeneity in ghost size, cellular GLUT1 content, or GLUT1 activity. Monophasic Urd exchange kinetics in the presence of intracellular ATP also argue against ATP-dependent ghost heterogeneity.

Urd exchange kinetics imply that biphasic 3MG exchange kinetics are either GLUT1 or sugar specific. If a sugar-specific compartment exists in ATP-loaded cells, then the 3MG space of ATP-loaded cells should be greater than the Urd space. ATP removal should be without effect on the 3MG space but will increase the Urd space. This was not observed. If a specific sugar-binding compartment explains the ATP-dependent slow phase of exchange transport, then the 3MG space of ATP-loaded cells should be greater than the Urd space, whereas ATP depletion will reduce the 3MG space of the cell to that of the Urd space, which remains unchanged. The observation that ATP has no affect on 3MG or Urd space refutes both hypotheses.

The possibility of a time-dependent change in the intrinsic kinetics of transport over the 5 h required for measurements is refuted by two observations: 1) instantaneous 3MG transport rates do not change with time and 2) biphasic exchange of radiolabeled 3MG is recapitulated after 15 h of exchange of unlabeled sugar. In summary, these results suggest two serial barriers to 3MG transport. The first barrier is translocation through GLUT1, whereas the nature of the second barrier is unknown. Two possibilities exist: the barrier is a physical barrier (e.g., permeability barrier or unstirred layer) or the barrier is a chemical barrier (e.g., sugar is reversibly converted to a second transported species).

The physical barrier hypothesis.

A physical barrier model for ATP-dependent, GLUT1-mediated biphasic 3MG equilibrium exchange is illustrated in Fig. 7A. In normal red blood cells, GLUT1 is anisotropically distributed in the plasma membrane. The cytosol, however, contains two diffusionally isolated compartments (e.g., see Ref. 28). Cytosol adjacent to GLUT1-enriched membrane rafts (compartment 1) is accessible to newly imported sugars. Cytosol adjacent to GLUT1-depleted membrane (compartment 2) is inaccessible to newly imported sugars. Thus sugar transport rapidly fills compartment 1, whereas equilibration of compartment 2 requires intracellular diffusion across the barrier separating the compartments. The Urd transporter ENT1, by contrast, is uniformly distributed in the membrane such that both 3MG compartments are accessible to imported Urd. With Mg-ATP depletion-induced echinocytosis (48), GLUT1 becomes uniformly distributed in the plasma membrane and/or the diffusion barrier between cytoplasmic compartments is lost, making the entire cellular space directly accessible to transported 3MG.

This model predicts 1) agents that promote spheroechinocyte formation will induce monophasic 3MG exchange kinetics, 2) GLUT1 membrane localization is different from ENT1 localization, 3) diffusion-mediated 3MG transport in discocytes should be monophasic, and 4) slowing transport by use of inhibitors will inhibit the fast component of transport but not the slow phase.

The chemical barrier hypothesis.

A chemical barrier describes sugar conversion to a second molecular species within the cell (Fig. 7B). A sugar could, for example, be metabolized, undergo anomerization, or become complexed to a binding partner. If metabolized, the transformed species must remain a substrate for bidirectional transport (by either GLUT1 or other transporters); otherwise, the cytoplasmic distribution space available to sugar plus its metabolite would exceed cell water volume. 3MG is, however, a nonmetabolizable sugar that is not transformed after transport either into or out of the cell (51, 52). We therefore reject the metabolism hypothesis. The binding partner hypothesis is refuted by the observation of equal Urd and 3MG equilibration spaces.

Differential transport of sugar anomers could explain biphasic sugar transport. d-Glucose and 3MG anomerize slowly in aqueous solution between α- and β-anomers (half-time = 10 min at 37°C and 250 min at 4°C; equilibrium ratio of α- to β-anomer of ∼33:66) (45). The anomeric hydroxyl group (located at C-1) is oriented below the pyranose ring in α-d-glucose and 3MG but is equatorial for the β-anomers. Could the orientation of this hydroxyl group affect transporter affinity and/or capacity for substrate? Epimers of d-glucose (glucose stereoisomers with different hydroxyl group orientations at specific ring carbons) show strikingly different affinities for the red cell sugar transporter. Studies with d-galactose (a C-4 epimer of d-glucose), d-allose (C-3), or d-mannose (C-2) indicate that opposite hydroxy configurations at C-4, C-3, or C-2 cause a 12-, infinite, or 3-fold, respectively, increase in apparent inhibition constant (relative to d-glucose) for inhibition of sorbose uptake by red blood cells (6). Where it has been studied (e.g., d-galactose), there is no accompanying change in V_max for sugar transport (31).

The equilibrium ratio of β-3MG to α-3MG at 4–20°C is 65:35, which is very close to the relative sizes of fast and slow 3MG compartments observed in ATP-loaded red cell ghosts (see Table 1). If biphasic sugar transport is explained by differential transport of α- and β-anomers, several predictions follow: 1) β-3MG is the preferred GLUT1 substrate in the presence of ATP (ratio of fast to slow transport rates of ∼19:1; see Table 1). In the absence of ATP, 3MG anomers are transported equally; therefore, 2) equilibrium compartment sizes in ATP-containing cells are proportional to equilibrium anomer distributions, 3) acceleration of mutarotation to a rate severalfold faster than the slow rate of transport eliminates the slow phase of transport, and 4) slowing transport by use of inhibitors will inhibit both the fast and slow phases of transport because both phases are carrier mediated.

Testing the barrier hypotheses.

The physical and chemical barrier hypotheses predict mutually exclusive outcomes for the partial inhibition test. The physical barrier hypothesis predicts that only the rapid phase of transport is inhibited by transport inhibitors because the slow phase is a GLUT1-independent event (diffusion). The chemical barrier hypothesis predicts inhibition of both phases because each is GLUT1 mediated. Our results show that both fast and slow phases are inhibited by cytochalasin B (Fig. 8) and by maltose (39). We therefore conclude the physical barrier hypothesis is untenable.

Refining the anomer (chemical barrier) hypothesis.

Figure 9A illustrates a simulated exchange uptake of α- and β-3MG. Figure 9 models experimental data to a mechanism in which α- and β-3MG compete for GLUT1-mediated uptake and where β-3MG is transported with greater capacity and affinity than α-3MG. Exchange transport is assumed to follow simple Michaelis-Menten kinetics (14, 38, 47), and uptake of radiotracer 3MG (Q) is given by

where where v and v are unidirectional uptake of radiolabeled α- and β-3MG, v and v are unidirectional exit of radiolabeled α- and β-3MG, V_β is V_max for exchange transport of β-3MG, K_β is K_m for β-3MG transport, V_α is V_max for exchange transport of α-3MG, K_α is K_m for α-3MG transport, S_i and S_o are the total unlabeled intra- and extracellular 3MG concentrations, Q_i and Q_o are the total radiolabeled intra- and extracellular 3MG concentrations, α is the fraction of S or Q that exists as α-3MG (0.35), and β is the fraction of S or Q that exists as β-3MG (0.65).

Total (radiolabeled α- plus β-3MG) exchange uptake data at 2.5 mM 3MG and 10 μM radiolabeled 3MG (4°C) were simulated by fourth-order Runge Kutta numerical integration (Fig. 9A). V_max and K_m(app) parameters for α- and β-3MG transport were varied by an iterative least-squares procedure until the deviation between observed and predicted uptake reached a minimum. The result suggests that β-3MG [V_max = 2.4 mmol·l⁻¹·min⁻¹ and K_m(app) = 4.7 mM] is a more efficient exchange substrate than is α-3MG [V_max = 0.8 mmol·l⁻¹·min⁻¹ and K_m(app) = 37.1 mM]. Figure 9 also shows the observed kinetics of α-3MG mutarotation at 4°C as measured by circular dichroism. 3MG mutarotation is significantly slower than exchange uptake of α-3MG, a prerequisite of the anomer (chemical-barrier) model for sugar transport. The complete 3MG exchange data set of Table 1 has been reanalyzed by this procedure (e.g., see Fig. 1B, dashed curve fits), and the results are summarized in Table 3. V_β, K_β, V_α, and K_α are independent of 3MG concentration and direction of radiotracer 3MG movement (uptake or exit). V_β is 2.5-fold greater than V_α, whereas K_α is 7.4-fold greater than K_β. Thus the major impact of reorientation of the anomeric hydroxyl group to the α-configuration is a sevenfold loss in affinity of sugar binding. In the absence of intracellular ATP, radiolabeled 3MG exchange transport is consistent with a single transport process and is described by

where where V and K are V_max and K_m(app), respectively, for 3MG exchange transport. Analysis of exchange transport in the absence of ATP (dashed curve fits of Fig. 1A, Table 3) indicates K_m(app) and V_max are 35 mM and 5.2 mmol·l⁻¹·min⁻¹, respectively, indicating that ATP depletion results in the loss of high-affinity β-3MG binding. ATP-GLUT1 interaction is thus a primary determinant of high-affinity β-3MG exchange transport.

Few previous studies have examined the transport of α- and β-d-glucose in red blood cells (17, 26, 58). Unlike liver, which contains significant mutarotase activity (10), the human red blood cell and red cell ghosts lack mutarotase (17). Preliminary in vitro measurements indicate that 3MG is not a substrate for porcine mutarotase and that mutarotase-acceleration of d-glucose anomerization is unaffected by 4 mM ATP (Leitch and Carruthers, unpublished observations). Our observations are, therefore, not explained by ATP-dependent inhibition of erythrocyte mutarotase. When freshly dissolved as pure anomers, α- and β-d-glucose are relatively equipotent in their inhibition of the initial rate of radiolabeled d-glucose uptake from an equilibrium mixture of d-[α-¹⁴C]glucose and d-[β-¹⁴C]glucose (17). This suggests that differences in rate constants for transport of α- and β-anomers reflect differential capacity for transport rather than differences in binding affinity, a conclusion that stands in contrast to our analysis of 3MG exchange transport. We therefore reexamined zero-trans [¹⁴C]3MG uptake in red cell ghosts containing 4 mM ATP. Proton NMR of freshly dissolved, unlabeled α-3MG and of an equilibrium solution of 3MG indicates that “α-3MG” is in fact 80% α-anomer and 20% β-anomer. The equilibrium solution contains a 35-to-65 ratio of α- to β-3MG. These sugar solutions were used as inhibitors of [³H]3MG uptake at 4°C. [³H]3MG is an equilibrium mixture of [α-³H]3MG and [β-³H]3MG. Uptake was measured at 8 nM [³H]3MG with increasing amounts of unlabeled 3MG (0–10 mM). Figure 9B shows that α-3MG and equilibrium 3MG inhibit unidirectional [³H]3MG uptake half-maximally at 0.6–1.2 mM (see x-intercepts of curve fits). This is consistent with our earlier study (17) comparing α- and β-d-glucose affinities and suggests only small differences in the affinity of GLUT1 for α- and β-3MG. However, the slope of the inhibition data employing equilibrium 3MG is significantly greater than the slope of the α-3MG inhibition data. At radiolabeled 3MG concentrations ≪ unlabeled 3MG concentrations, the ratio of control radiolabeled 3MG uptake to inhibited 3MG uptake (v^c/vⁱ) increases linearly with total concentration of unlabeled 3MG and is characterized by the slope

where α and β are the fractional amounts of α- and β-3MG present in the uptake medium and K_α and K_β are K_m(app) for uptake of α- and β-3MG, respectively. Using the computed slopes of Figure 9B and the known ratios of α- to β-anomers present in solution, we calculate that K_α and K_β for zero-trans net α- and β-3MG uptake are 5.29 ± 0.47 and 0.77 ± 0.13 mM, respectively (Table 3). These affinity constants are significantly lower than those for exchange transport. However, K_m(app) for zero-trans d-glucose and galactose zero-trans uptake are 14- and 11-fold lower, respectively, than K_m(app) for exchange uptake of the same sugars (18, 29, 30). This more detailed analysis strongly supports the hypothesis that GLUT1 shows greater affinity for β-3MG than α-3MG.

Physiological significance.

A preference for α-d-glucose in transport and/or metabolism has been reported for liver (10, 54, 59), pancreatic beta cells (53, 55, 60, 71), muscle (80), adipose (70), and yeast cells (25). Cardiac muscle hexokinase displays high affinity but low capacity for α-d-glucose and low affinity but high capacity for β-d-glucose (50). Brain, retina erythrocytes, and lens cells appear to prefer β-d-glucose or show no preference (17, 26, 52, 59, 65, 67, 68). β-d-Glucose is more potent than α-d-glucose in inhibiting vagally mediated secretion of gastric acid (77). These observations emphasize the tissue-specific physiology of anomer transport and utilization and further suggest that cells expressing GLUT1 or GLUT3 may prefer the β-anomer, whereas cells expressing GLUT2 or GLUT4 may prefer the α-anomer.

Significance to previous transport studies.

These findings may necessitate reevaluation of red cell glucose transport determinations. Transport measurements are normally obtained by initial rate analysis over intervals of 1 min or less. If initial rates of transport are derived largely from uptake of β-sugar, they are related to (and must be corrected for) the amount of β-sugar present in solution during the measurement. Thus K_m(app) and V_max parameters may be overestimated by 1.5-fold. Furthermore, computed rates and K_m(app) must be corrected for competitive inhibition by the omnipresent, lower affinity α-sugar. This result could also explain (see Refs. 2 and 78 for lucid expositions) how integrated rate equation analysis of time course data results in overestimation of K_m(app) for transport because all time courses are artificially extended owing to slower transport of α-sugar. This laboratory has previously interpreted the slow phase of sugar import as diffusionally limited release of translocated sugar into cytoplasm (8, 9). This interpretation is no longer tenable.

Conclusion.

ATP-containing human red blood cell ghosts exhibit biphasic equilibrium exchange 3MG transport at sugar concentrations ranging from 1 to 20 mM. This observation is inconsistent with previous models for ATP modulation of GLUT1 activity. We propose that ATP-GLUT1 interactions promote preferential transport of β-3MG and that extended time courses of sugar transport in human red blood cells reveal deviations from simple kinetics owing to low-affinity, low-capacity α-3MG transport and continuous β-3MG and α-3MG mutarotation. In the absence of ATP, both anomers may be transported equally or differences in the kinetics of β-3MG and α-3MG become too small to measure with existing methodologies. Differential transport of 3MG anomers (and thus transport complexity) is only observable in cells where rate constants for 3MG transport are greater than those for 3MG mutarotation.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-36081 and DK-44888.

FOOTNOTES

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.