Evidence that forskolin binds to the glucose transporter of human erythrocytes.

Binding of [4-3H]cytochalasin B and [12-3H]forskolin to human erythrocyte membranes was measured by a centrifugation method. Glucose-displaceable binding of cytochalasin B was saturable, with KD = 0.11 microM, and maximum binding approximately 550 pmol/mg of protein. Forskolin inhibited the glucose-displaceable binding of cytochalasin B in an apparently competitive manner, with K1 = 3 microM. Glucose-displaceable binding of [12-3H]forskolin was also saturable, with KD = 2.6 microM and maximum binding approximately equal to 400 pmol/mg of protein. The following compounds inhibited binding of [12-3H]forskolin and [4-3H]cytochalasin B equivalently, with relative potencies parallel to their reported affinities for the glucose transport system: cytochalasins A and D, dihydrocytochalasin B, L-rhamnose, L-glucose, D-galactose, D-mannose, D-glucose, 2-deoxy-D-glucose, 3-O-methyl-D-glucose, phloretin, and phlorizin. A water-soluble derivative of forskolin, 7-hemisuccinyl-7-desacetylforskolin, displaced equivalent amounts of [4-3H]cytochalasin B or [12-3H]forskolin. Rabbit erythrocyte membranes, which are deficient in glucose transporter, did not bind either [4-3H]cytochalasin B or [12-3H]forskolin in a glucose-displaceable manner. These results indicate that forskolin, in concentrations routinely employed for stimulation of adenylate cyclase, binds to the glucose transporter. Endogenous ligands with similar specificities could be important modulators of cellular metabolism.


It has recently been
shown that forskolin, a diterpene activator of adenylate cyclase (l), strongly inhibits facilitated diffusion of glucose in human erythrocytes (2) and blood platelets (3) and in rat or human adipocytes (4, 5 ) . Published data (2-5) have suggested that this effect of forskolin is not entirely mediated by cyclic AMP. Herein we present evidence that forskolin interacts directly with the hexose transporter of human erythrocyte membranes. These results have important implications for the coordinate regulation of metabolism in hormonally responsive tissues. Co. Sodium dodecyl sulfate was from Bio-Rad. Preparation of Erythrocyte Membranes-Human erythrocytes were obtained from freshly drawn blood of normal adult donors. Essentially hemoglobin-free membranes ("gbosts") were prepared by the method of Dodge et al. (6). The membranes were suspended in 5 mM sodium phosphate buffer, pH 8.0, at concentrations of 1.6-2.4 mg of protein/ ml and stored in liquid nitrogen.
Membrane Protein Determination-Membranes were solubilized in 0.5 N NaOH at 50 "C overnight and then assayed by the method of Lowry et al. (7) with crystalline bovine serum albumin as standard.
Binding Assays-Binding of [4-3H]cytochalasin B or [12-3H]forskolin to erythrocyte membranes was determined by a centrifugation method. Incubation mixtures contained tritiated ligand (0.03 pCi of [4-3H]cytochalasin B or 0.35 pCi of [12-3H]forskolin), nonradioactive competing ligand as appropriate, monosaccharide (500 mM D-glucose or 500 mM sorbitol), and erythrocyte membranes (0.15-0.2 mg of protein/ml) in 5 mM sodium phosphate buffer, pH 8.0, at a final volume of 1.0 ml. In some experiments, other monosaccharides replaced D-glucose. Stock solutions of forskolin and cytochalasin B were dissolved in methanol and ethanol, respectively. Appropriate vehicle controls were included in each experiment, so the final concentration of alcohol in all tubes was 2.1% (v/v). All tubes contained 10 p~ cytochalasin E, which has been shown to compete with [4-3H] cytochalasin B for binding to nontransporter-related sites, without affecting its binding to the glucose transporter (8)(9)(10). In preliminary experiments, we confirmed that cytochalasin E reduced nonspecific (glucose-nondisplaceable) binding of [12-3H]forskolin to the membranes, without affecting glucose-displaceable binding. Therefore, we routinely employed cytochalasin E in the incubations. All incubations were performed in triplicate.
Incubations were started by addition of 100 pl of membrane suspension to tubes containing all the other components, with mixing. In preliminary experiments we determined that binding of either ligand was stable with incubation times from 5 to 15 min at room temperature. The standard procedure was to incubate for 5 min at room temperature, followed by centrifugation for 3 min in a microcentrifuge. After centrifugation, 500 pl of supernatant was transferred to a scintillation vial, and the rest of the supernatant was carefully aspirated off the membrane pellet. Each pellet was then dissolved in 50 pl of sodium dodecyl sulfate (11 mg/ml) and transferred to another scintillation vial. The incubation tube was then rinsed with 50 pl of water and the rinse added to the vial containing the dissolved membranes. The vials were counted in a liquid scintillation counter, with Hydrofluor (National Diagnostics, Inc.) as the scintillation fluid.
Concentrations of bound and free ligands were determined from the counts in the vials containing membranes and supernatants, respectively. The counting efficiency of 3H in the vials containing Forskolin Binds to Erythrocyte Glucose Transporter supernatants was from 88 to 90% that in the vials with membranes. Also, from 3 to 9% free ligand was trapped in the pellets, as determined by parallel incubations with ['4C]sucrose, which does not enter human erythrocytes (11). All the data have been appropriately corrected for counting efficiency and for trapping, which were measured independently in each day's experiment. Glucose-displaceable binding of ligand was calculated as the difference between the amount bound in the presence of 500 mM sorbitol (total binding) and that bound in the presence of 500 mM D-glucose. Binding in the presence of 500 mM D-glucose, which amounted to 6% total counts for [4-3H]cytochalasin B and 4% counts for [12-3H]forskolin, was nonsaturable and considered not to be specific for the glucose transporter. The relative amount of nonspecific binding increased with concentration of ligand, varying from 10 to 50% total binding over the range of 20 nM to 2 p~ [4-3H]cytochalasin B, and from 13 to 50% total binding over the range of 1 nM to 30 p M [12-3H]forskolin. The data were analyzed by the "LIGAND" program of Munson and Rodbard (12), kindly provided by Drs. Peter Munson and Michael Beveridge of the National Institutes of Health. The Scatchard plot of glucose-displaceable binding of [ 12-"Hlforskolin to erythrocyte membranes is shown in Fig. 2. The data suggest the existence of nonsaturable binding even after correction for displacement by glucose. After correction for residual nonsaturable binding (12, 17), we obtained the following coefficients: KO = 2.6 p M , maximal binding = 400 pmol/mg of protein.

Glucose
As a check on the accuracy of the above methods for determination of nonspecific binding, we also analyzed the raw data without correction for displacement of 3H-ligand by D-glUCOSe. In this analysis, nonspecific binding was treated as a statistic to be estimated by the curve-fitting algorithms of the "LIGAND" program. This method yielded results essentially identical with those derived from the corrected data: for L L : %, -. ., . """"""_..""_"""""""""" asin B in the presence of 500 mM D-glucose, presumably reflecting binding to nontransporter-related sites (8), was only minimally (less than 12%) displaced by 50 F M forskolin (data not shown).
Cytochalasin B displaced forskolin from the membranes. Glucose-displaceable binding of between 0.4 and 3 PM [12-'Hlforskolin was half-maximally inhibited by 0.2-0.6 PM cytochalasin B (data not shown), concentrations which approximate the KD for binding of cytochalasin B. As shown in Table   I, other cytochalasins displaced [12-3H]forskolin and [4-3H] cytochalasin B equivalently, with an order of potency parallel to their published relative potencies as inhibitors of glucose transport (8).
We also investigated displacement of forskolin and cytochalasin B by a water-soluble derivative of forskolin, 7-0hemisuccinyl-7-desacetylforskolin. The results, in Table 11,   Table 111, the relative orders of potency and degrees of inhibition were the same for either labeled ligand L-rhamnose and L-glucose were ineffective; D-galactose and D-mannose were of intermediate potency; and Dglucose, 2-deoxy-~-ghcose, and 3-O-methyl-D-glucose were the most powerful inhibitors of binding. This order of potency closely parallels the reported affinities of these sugars for the hexose transport system (19). Phloretin inhibits hexose transport in human erythrocytes (19, 20) and has previously been shown to inhibit binding of cytochalasin B to partially purified hexose transporter (8,16).   log of phloretin with little or no transport-inhibitory activity in erythrocytes (19, 21), did not displace either ligand.

As shown in
In an attempt to elucidate the binding specificities of [12-3H]forskolin, we studied membranes from rabbit erythrocytes, which transport glucose slowly and in a nonstereospecific manner (19, 22, 23) and lack transport-related binding sites for cytochalasin B (22, 24), presumably because they are deficient in hexose transporter. Our results, in Fig. 4, A and B, show that binding of both cytochalasin B and forskolin by these membranes was linear and apparently nonsaturable, with no displacement by glucose. The total binding of either ligand to rabbit membranes was quantitatively the same as the nonspecific binding to human membranes. DISCUSSION Stereospecific facilitated diffusion of glucose into mammalian cells is inhibited by the fungal metabolite cytochalasin B (8,25), an effect which seems to be mediated by binding to the hexose transporter (8,(14)(15)(16)25). Recentlypublished work has indicated that forskolin, thought previously to be a specific activator of adenylate cyclase, inhibits glucose transport and competes with cytochalasin B for binding to membranes from human erythrocytes (2, 26) and rat adipocytes (5). Our data elucidate further the binding interactions between forskolin and cytochalasin B.
In this work, we have minimized binding of radiolabeled ligands to sites not associated with glucose transport (8,9,24) by correction for extraneous binding in the presence of 500 mM D-glucose, and by inclusion of nonradioactive cytochalasin E, which competes with cytochalasin B for nontrans- The Scatchard plots ( Figs. 1 and 2) suggest that the number of glucose transporter-related [12-3H]forskolin-binding sites (400 pmol/mg of protein) is approximately equal to that for cytochalasin B (550 pmol/mg of protein). Because the limit of solubility of forskolin, approximately 30 PM, is less than 5fold greater than the KD for binding, it was necessary to extrapolate from data obtained at less than saturating concentrations, a procedure which leads to inaccuracy (27). However, the nonextrapolated data also suggest equal numbers of sites for forskolin and cytochalasin B. For example, the number of glucose-displaceable forskolin-binding sites at the KI for competition with cytochalasin B was 260 pmol/mg of protein, compared with half-maximal binding of 275 pmol/ mg of protein for cytochalasin B. Also, the nonradioactive cytochalasins and 7-hemisuccinyl-7-desacetylforskolin displaced equivalent amounts of [4-3H] by a cascade mechanism involving cyclic AMP-dependent protein phosphorylation. Indeed, from currently available data, it seems that none of the components of the adenylate cyclase system is sufficiently abundant in erythrocytic membranes to have contributed measurably to the observed level of [12-3H]forskolin binding.' The KI for displacement of cytochalasin B by forskolin (3 PM) lies within the range of reported half-maximal concentrations (2-7.5 PM) for inhibition of hexose transport in human platelets (3) and erythrocytes (2) and for inhibition of glucose transport and oxidation in rat adipocytes (5,31). Also, the KD for binding of [12-3H]forskolin approximates that reported (1-2 PM) for binding of forskolin to low affinity sites on rat brain and liver membranes (32,33). In contrast, Joost and Steinfelder (5) recently reported a nearly 50-fold greater potency of forskolin for displacement of cytochalasin B from adipocyte membrane vesicles ( K D = 200 nM), with an equivalent half-maximal dose for inhibition of 3-0-methylglucose transport. The reasons for the discrepancy with our erythrocyte data are yet unknown. Further work with isolated components will be required in order to determine whether forskolin and cytochalasin B compete for the same binding site, and to clarify the relationships of the high and low affinity binding sites for forskolin (reviewed in Ref. 34) to the hexose transporter and to adenylate cyclase, in various tissues.
Because human erythrocytic plasma membranes are relatively rich in glucose transporter (24, 35, 36) and poor in adenylate cyclase catalytic activity,' they provide a useful system for the study of actions of forskolin unrelated to activation of adenylate cyclase. Photolabeling with cytochalasin B (25, 37 38), immunologic methods (35)(36)(37)(38)(39)(40), and, most The highest catalytic activity so far reported for unfractionated human erythrocytic membranes has been 6 pmol/min/mg of protein (28); assuming this is only 0.1% of maximal activity, with a turnover number of 1000. min" (29), this would represent 6 pmol of catalytic units/mg of protein. Similarly, quantitative ADP-ribosylation has yielded estimated levels of 1 and 4 pmol/mg of protein for the NS and Nr (stimulatory and inhibitory guanyl nucleotide-binding) regulatory subunits, respectively (30). These numbers are 2 orders of magnitude lower than the level of binding we observed.
Forskolin has hitherto been thought to act specifically on adenylate cyclase (reviewed in Ref. 1). The finding that it competes with cytochalasin B for binding to the glucose transporter indicates that interpretation of experiments in which forskolin is applied to intact cells may be complicated by direct actions of the drug on uptake of glucose. In order to clarify the roles of cyclic AMP and glucose transport on control of metabolism, it will be necessary to employ agents that selectively interact with one or the other of these two systems. Furthermore, we speculate that there may exist cellular analogs of forskolin which could reciprocally modulate cyclic AMP generation and glucose transport. In this regard, we note with great interest the recent reports that mediators of insulin action can be derived from lipid-soluble precursors in plasma membranes (42, 43) and that insulin (44, 45) and catecholamines (46) can rapidly alter the activity of glucose transporters in adipocyte plasma membranes. Intramembranal lipids would be attractive candidates for the putative endogenous ligands to cytochalasin B/forskolin binding site(s) on the hexose transporter.