Sulfonylureas Rapidly Cross Phospholipid Bilayer Membranes by a Free-Diffusion Mechanism

Egg phosphatidylcholine was purchased from Avanti Polar Lipids (Pelham, AL). Tolbutamide and glibenclamide were purchased from Sigma (St. Louis, MO). Pyranin (8-hydroxy-1,3,6-pyrenetrisulfonate), [2′,7′-bis (2-carboxyethyl)-5(6)-carboxyfluorescein] (BCECF), and its acetoxymethylester (BCECF-AM) were purchased from Molecular Probes (Eugene, OR). D₂O was purchased from Cambridge Isotope Labs (Boston, MA), and G-25 Sephadex gel was from Amersham Pharmacia Biotech (Uppsala, Sweden). Sodium 3-(trimethylsilyl)[2,2,3,3-²H₄] propionate (TSP) was from Wilmad (Buena, NJ). Fatty acid-free albumin was from Sigma. Sample preparation for ¹H nuclear magnetic resonance experiments. Tolbutamide was dissolved (3.5 mmol/l final concentration) in 10 mmol/l 2-morpholinoethanesulfonic acid (MES) buffer with 50 mmol/l KCl. An equivalent amount of NaOH was added to dissolve all tolbutamide, and D₂O was added for nuclear magnetic resonance (NMR) analysis. After each measurement, the sample was removed from the NMR tube to adjust the pH to the desired value with small (5-μl) aliquots of 0.1 mmol/l HCl or 0.1 mmol/l NaOH. The sample of 20 mol% tolbutamide (0.2 mol tolbutamide/mol phospholipid) incorporated in small unilamellar vesicles (SUVs) was prepared as follows: 45 mg egg phosphatidylcholine in chloroform was mixed with 3.0 mg tolbutamide. The chloroform was evaporated under nitrogen, and the sample was lyophilized for 1 h; 1.62 ml of 0.5% wt/vol of KCl and 0.18 ml D₂O were added to the lyophilized sample, and the sample was allowed to hydrate for 2 h at 4°C. The solution was then sonicated (Branson 350 sonifier) in an ice/water bath for 1 h using a microtip and a 30% duty cycle. The sample of 10 mol% glibenclamide (0.1 mol glibenclamide/mol phospholipid) incorporated into the SUV was prepared by adding a small aliquot of glibenclamide in DMSO to SUV prepared as above except without any sulfonylurea added. Samples were titrated from low to high pH by adding 1- to 2-μl aliquots of 1 mol/l KOH. The pH was measured in the NMR sample with a Beckman pH meter fitted with a glass micro-electrode.

NMR experiments were performed on a Bruker DMX 500 and a Varian 400-MHz spectrometer. One-dimensional ¹H NMR experiments were performed at 20°C with 500–700 μl of sample containing TSP as an internal chemical-shift standard, using a “zgpr” pulse sequence. The frequency of the water presaturation pulse (O₁) was calibrated for each experiment because its value is pH dependent. The downfield phenyl ring chemical shift of both tolbutamide and glibenclamide protons was measured and plotted against the pH of the solution. All dissociation constant (pK_a) determinations were done by plotting chemical shift values versus bulk pH (13) and curve fitting to a sigmoidal curve. The pH value corresponding to the midpoint of this curve is the apparent pK_a of the sample.

SUVs were prepared as above except with pyranin or BCECF present; large unilamellar vesicles (LUVs) were prepared by extrusion as described previously (14). Untrapped pH dye was removed by gel filtration (Sephadex G25), and the phospholipid concentration was determined as described previously (14,15). Clonal pancreatic β-cells (HIT) were cultured and prepared for experiments as before (16). Cells were incubated with 1 μmol/l BCECF-AM for 30 min. Untrapped BCECF and external albumin were removed by three wash steps with Krebs buffer without albumin (16).

The fluorometric assays were performed on a FluoroMax 2 fluorometer as described in detail before (14,15).

Sulfonylureas (Fig. 1) are weak acids, with pK_a values of 5.43 (tolbutamide) and 5.3 (glibenclamide) in aqueous solution (17). The pK_a of the sulfonamide group by itself is 10.1 in solution but is lowered to between 5 and 6.5 when the group is next to a carbonyl. Because ionization is difficult to monitor by ¹H NMR of the ionizing hydrogen but can be reflected in the chemical shifts of nearby groups in weak acids (13), we measured the downfield chemical shift of a phenyl-ring hydrogen of both tolbutamide and glibenclamide that are ortho to the sulfonamide group (Fig. 1).

The pK_a of tolbutamide was determined both in solution and incorporated in the phospholipid bilayer at a ratio of 20 mol% (0.2 mol tolbutamide/mol phospholipid). The determinations were made as comparable as possible by dissolving similar amounts of tolbutamide in solution and in the vesicle suspension. We found that the pK_a of tolbutamide in solution was 5.41 and was shifted up to 6.64 in the phospholipid bilayer (Fig. 2). This upward shift is similar to that found for fatty acids (FAs) and other weak lipophilic acids (18). Sulfonylurea derivatives are amphiphilic molecules and have limited solubility in pH 7.4 HEPES buffer (19) and, in the presence of SUVs, become incorporated into the phospholipid bilayer structure. Additional evidence for the incorporation of tolbutamide in the phospholipid bilayer comes from the differences in chemical-shift values of the phenyl-ring hydrogens at most pH values (Fig. 2) and line-width changes (not shown).

We were unable to obtain spectra for aqueous glibenclamide because of its very low solubility in water. However, glibenclamide could be incorporated into phospholipid vesicles, and we attempted to measure its pK_a in SUVs by the same method we used for tolbutamide. Beginning at a high pH (10.63) with a translucent solution and titrating to a lower pH, we visually observed increasing turbidity and progressive line-broadening of the signals for the phenyl ring next to the sulfonylurea moiety (not shown). The peaks were broadened beyond detection at pH 5.44, so that only a rough estimate of the pK_a (∼7.0) was possible.

Previously, we developed a simple fluorescent pH assay to monitor the diffusion of FAs and other weak lipophilic acids, such as bile acids, across a protein-free phospholipid bilayer (14,15). Because of the increased pK_a in the phospholipid interface, sulfonylureas also exist almost equally in non-ionized and ionized forms at neutral pH.

To test whether the same mechanism (flip-flop) is feasible for sulfonylureas, we added small aliquots (1 mmol/l dissolved in DMSO or ethanol) of glibenclamide or tolbutamide to a suspension of vesicles with trapped pyranin. The observed drop in fluorescence intensity (Fig. 3) is due to a drop in pH inside the vesicles. As with FAs, the decrease in internal pH (pH_in) began immediately upon the addition of the sulfonylurea and was complete within 1 s—the time resolution of the measurement (15). Addition of FA-free BSA to the external buffer of SUV with oleic acid, glibenclamide (Fig. 3), or tolbutamide (not shown) resulted in a rapid return to the initial fluorescence intensity. BSA reportedly has three binding sites for tolbutamide, with a K_d of 21 μmol/l (20), and the rise in pH reflects extraction of sulfonylurea from the SUVs (accompanied by flip-flop). The observed drop in fluorescence intensity for tolbutamide was much smaller than that for the same amount of added glibenclamide (Fig. 3), indicating decreased transport of tolbutamide, most likely due to the lower partitioning of tolbutamide into the membrane.

The fluorescence results for both sulfonylureas were reproduced in numerous additional experiments and were independent of whether the trapped pH dye was pyranin or BCECF and whether the model membranes were SUV or LUV. The change in fluorescence was consistently nonlinear and progressively smaller with increasing concentration of added sulfonylurea, as illustrated for glibenclamide in Fig. 4. This pattern is due to the build-up of a pH-gradient across the phospholipid bilayer, as previously demonstrated for FAs in protein-free phospholipid vesicles, and provides further evidence for the flip-flop mechanism (14).

To assess the physiological relevance of the flip-flop of sulfonylurea observed in model membranes, we monitored the intracellular pH of insulin-secreting (HIT) cells. When glibenclamide was added to a suspension of 2 ml of cells preloaded with BCECF, the pH_in dropped ∼0.1 pH units in a time course of 3 min (Fig. 5). On addition of albumin, the pH_in partially recovered. As with FAs (16), the pH_in changes after exposure of HIT cells to glibenclamide were slower than those observed in SUVs (Fig. 2).

In recent years, many studies have proposed a role for transport proteins in the delivery of FAs to cells, and the importance of free diffusion for the transport of FAs across the phospholipid bilayer has been a topic of debate (21–25). Sulfonylureas are good candidates for the investigation of the free-diffusion model, because these compounds are known to be hydrophobic and are also weak acids. These derivatives are the kinds of molecules for which the Overton rule was formulated more than 100 years ago (26,27). According to this rule, the entry of a molecule into a cell depends on the solubility of that molecule in the phospholipid membrane surrounding the cell. Provided that the molecule can assume an uncharged form in the lipid bilayer, it also can translocate through the bilayer. The major findings of our study are that 1) upon binding to phospholipid bilayers, the apparent pK_a values for tolbutamide and glibenclamide increase to neutral pH values; 2) the uncharged form of the sulfonylurea rapidly diffuses across a protein-free phospholipid bilayer; and 3) glibenclamide diffuses through the plasma membrane of insulin-secreting HIT cells.

To study molecular interactions between sulfonylureas and phospholipid bilayers and the transport of sulfonylureas through bilayers, we used ¹H NMR spectroscopy to demonstrate sulfonylurea binding to vesicles and to measure the pK_a of the ionizable sulfonamide group. Direct interactions of the sulfonylurea with phospholipids were shown by changes in chemical shifts and line-widths of selected ¹H NMR peaks of the sulfonylurea in the presence of phospholipid bilayers relative to those of the sulfonylurea in aqueous buffer. The pK_a of tolbutamide increased in accord with predictions from previous studies of weak amphiphilic acids such as FAs and bile acids (27–29). Although the pK_a could only be estimated for glibenclamide, the estimate was good enough to permit us to conclude that the interfacial pK_a values for the two sulfonylureas are not significantly different. The titration of the sulfonylurea in vesicles indicated that the sulfonamide group is readily accessible to the H⁺/OH⁻ added to the external buffer and therefore positioned at the aqueous surface of the phospholipid bilayer. Exposure of the titratable group (Fig. 1) would most likely be accomplished by immersion of the more hydrophobic groups into the bilayer; such a bent conformation also would enhance diffusion of the molecule through the bilayer by reducing its cross-sectional area. A study of 53 compounds that have a clinically established ability or lack of ability to cross the blood-brain barrier by diffusion (30) revealed three main requirements for diffusion. These are 1) partitioning into the lipid bilayer, 2) a pK_a between 4 and 10 if the molecule is charged, and 3) a small cross-sectional area (<80 Å) in the interface. The NMR results show directly that both glibenclamide and tolbutamide fulfill the first and third requirements and show indirectly that the second requirement is also fulfilled.

To study the transmembrane movement of sulfonylureas, we performed fluorescence measurements of pH_in of entrapped pH dye after addition of the sulfonylurea to suspensions of SUV or LUV. Both sulfonylureas caused a dose-dependent decrease in pH_in and a progressively smaller decrease with increasing equivalent doses. These results are in accord both with theoretical predictions for the flip-flop mechanism and experimental results for FAs and other weak lipophilic acids (14). The pH changes observed with tolbutamide were much smaller than those observed with glibenclamide (Fig. 3). Because the interfacial pK_a values were similar, this is likely a result of the lower partitioning into the lipid phase of the more water-soluble tolbutamide. Figure 3 provides a graphic illustration that lower partitioning results in decreased diffusion through the membrane and that higher doses of tolbutamide would be required to achieve the same cellular concentrations as those for glibenclamide.

Our fluorescence assay also showed that glibenclamide diffused through the lipid bilayer plasma membrane of HIT cells, although not as rapidly as in model membranes. The bi-directional movement of sulfonylureas was demonstrated by the reversal of pH_in by the incomplete recovery reflecting weaker binding of glibenclamide to albumin compared with the binding of long-chain FAs. Similarly slow kinetics of pH_in changes such as those obtained for glibenclamide diffusion across the plasma membrane of HIT cells have been observed for the entry of FAs into HIT cells (16). Entry of FAs into adipocytes is significantly faster (31,32). In interpreting the kinetics of the pH_in changes, one must keep in mind that the pH assay measures the arrival of protons in the cytosol after adsorption of glibenclamide to the outer leaflet of the plasma membrane and its translocation through the membrane. Passive diffusion through the plasma membrane of the HIT cell might be slower than diffusion through protein-free model membranes because of the diffusion through the unstirred water layer (glycocalyx) adjacent to the plasma membrane or slower diffusion through the plasma membrane. This is not a unique characteristic of the sulfonylurea: the same considerations apply to the relatively slow pH changes after the addition of FAs to HIT cells (16).

We have shown that both tolbutamide and glibenclamide can cross the phospholipid bilayer rapidly by a free-diffusion mechanism (Fig. 6). Although our assay does not provide information about the rate of dissociation of sulfonylureas from the bilayer, it is likely that these molecules dissociate rapidly (or at least at a time scale significant for cell physiology), as in the case of FAs (33), and become available for binding sites in cytosol after they have diffused through the plasma membrane. These findings are especially interesting in that we have shown previously that these compounds can directly stimulate exocytosis of insulin (5–8). The discovery of a transport mechanism for sulfonylurea compounds across the phospholipid bilayer offers an explanation at the molecular level for how the sulfonylurea reaches intracellular targets. Furthermore, knowledge that sulfonylureas effectively permeate the plasma membrane and understanding of an underlying molecular mechanism may be of fundamental clinical importance in identifying new pharmacological principles and developing drugs for the treatment of type 2 diabetes.

Financial support was obtained from National Institutes of Health Grants DK 58508, HL 26335, HL 67188, and DK46200; the Swedish Research Council; the Swedish Diabetes Association; the Novo Nordisk Foundation; Bert von Kantzow’s Foundation; and Funds of Karolinska Institutet.

We thank Dr. Gordon Yaney for providing the HIT cells and Prof. Klaus Beyer for help with the NMR measurements.