[3H]Bumetanide Binding to Membranes Isolated from Dog Kidney Outer Medulla RELATIONSHIP TO THE Na,K,Cl CO-TRANSPORT SYSTEM*

We have synthesized the radiolabeled “loopy’ diuret-In order to study the molecular properties of the Na+,K+,Cl-ics [3H]bumetanide and [3H]benzmetanide (3-benzyl-co-transport process, it will be necessary to identify and purify amino-4-phenoxy- 5 -sulfamoylbenzoic acid) and have the transport protein; for this purpose, a specific marker is tested their potential as reversible labels of the Na,K,Cl needed with which the transporter can be followed through co-transport system. These compounds bind with high steps of membrane isolation, and protein solubilization and affinity (& 5 30 nM, under optimal conditions) to purification. Here we report the synthesis of [3H]bumetanide membranes isolated from dog kidney; we found -2 and show that it binds with high affinity to membranes Pmol/mg of sites in Crude membranes from the Outer isolated from dog kidney outer medulla; the ion requirements medulla, and ( O s 5 pmol/mg in a preparation for binding and the pharmacological specificity are convincing from kidney cortex. On Sucrose gradient centrifuga-evidence that the Na,K,Cl cotransport system is involved. A tion, a peak of [3Hlbumetanide binding activity (30 preliminary report has been presented (7). pmol/mg) is obtained at 37% (w/v) sucrose, distinct from the basolateral membranes in outer medulla and from brush borders in proximal tubule; our hypothesis is that this peak contains luminal membranes from the

The coupled transmembrane movement of N~+ , K+, and column (Whatman Partisil ODS 111, 10 r m ) with 30 MeOH, 20 "is mediated in many by a membrane fractions were pooled, evaporated, and taken up in 1 ml 10% dimethyl CH,CN, 0.2 CH,COOH, 50 H20 as solvent (2 ml/min). The peak transport system that is inhibited by furosemide,' bumetansulfoxide (yield, 0.7 m~i ; for benzmetanide, 0.5 m~i from ide, and other loop diuretics (1-4). Recently, it has been shown reaction). From the absorption spectrum (using = 9200 and that furosemide-sensitive active C1 uptake in the TALH in = 40001, the specific activity of the product was found to be 1.25 Ci/ mammalian kidney is Na+ and K+ dependent (5, 6) and mmol. Purity of the radioactive material, which co-migrated with the probably involves a N~, K , c~ co-transport system localized in authentic compound, was greater than 90% as determined by scintilthe lumina' membranes Of the In view Of the with 80 CHCl,, 2.5 MeOH, 10 CH3COOH, 1 cyclohexane; or Whatman lation counting of samples from a TLC plate (Silica Gel 60 F run extraordinarily high rate of transport across the luminal mem- KC.18 with 60 EtOH, H,O, 2 CH,COOH; visualization by bran% approximately 2 nmol/cm*.s (51, this tissue should be fluorescence in 254 nm light). We have determined the ether/H20 a rich source of the transport system. partition coefficient of [,H]bumetanide to be 8. 8, 8.7, 8.0, 3.9, 0.88, 0.25,0.04,0.02, and 0.02 at pH 2,3,4,5,6, 7,8,9, and 10, respectively, indicating that the relevant pK is near 5.
Sucrose gradient centrifugation was performed as previously described (9).

I"H]Burnetanide-binding
Experiments-Binding of (3H]bumetanide and ['Hlbenzmetanide to kidney membranes was determined by filtration on cellulose ester filters; other techniques, including nucleopore filtration without a rinse, and pelleting in an Airfuge (no wash) gave unacceptably high backgrounds, while filtration on Sephadex G-50 (1.5-mI bed volume) was satisfactory, but more time consuming. The basic binding experiment was performed as follows. Kidney membranes (0.2-20 mg/ml depending on the preparation used) in 10 p1 of 0.25 M sucrose, 30 mM histidine, pH 7.2, were added to tubes containing 15 pl of ["]burnetanide or ['Hlbenzmetanide (typically 0.2 p M final), appropriate ions (typically 30 mM Na' , 30 mM K' , 30 mM C1-, 15 mM S04'-), with or without unlabeled bumetanide or its analogs (typically 2 7 p~ bumetanide, final). After 20 min a t 22 "C, the sample was diluted into 5 ml of ice-cold 10 mM Tris-CI (pH 7.0), filtered on a 0.45-pm cellulose ester filter (Gelman GN-6 or Millipore HAWP), and rinsed with an additional 5 ml of cold buffer; elapsed time from dilution to the end of filtration was 15-20 s. Binding of ["Hlbumetanide to the cellulose ester filter alone was 0.3% of the amount filtered (about 20% of the nonspecific binding in Fig. 1).
In equilibrium binding experiments analyzed by Scatchard plot (Fig. 1) samples of each binding medium were counted, to determine total bumetanide (T, cpm, and thus concentration on the abscissa in  ( B l , Bz, cprn). Free bumetanide ( F , cpm) was calculated as the total ( T , cprn) minus the amount bound in the absence of cold bumetanide ( B l , cprn). Bound bumetanide ( E , cprn) was calculated as the difference between the amounts bound in the absence ( B l , cpm) and presence ( H z , cpm) of 10 p~ unlabeled bumetanide; it was plotted on the ordinate ( B , prnollrng) after division by the specific activity of ['HI bumetanide and the protein concentration.
Assays-(Na,K)-ATPase was assayed, after detergent treatment to disrupt membrane vesicles, as previously described (12). Alkaline phosphatase activity was determined by the hydrolysis of p-nitrophenyl phosphate (Sigma Bulletin No. 104). Protein was assayed by dye binding (13) or by the method of Lowry et al. (23).

Equilibrium Binding of ['HI Bumetanide to Kidney Mem-
branes-When we incubated plasma membranes from outer medulla of dog kidney with M ["Hlbumetanide in a medium containing Na, K, and C1, we found that a fraction of the drug remained bound to the membranes when they were filtered on a cellulose ester filter and washed briefly a t 0 "C; as shown in Fig. 1A, a component of the binding showed saturation behavior with increasing concentrations of ["HI bumetanide. As will be shown below, the 20-min incubation was sufficient to attain at least 90% of the equilibrium binding level, at concentrations above 0.1 KM [3H]bumetanide.' From the Scatchard analysis'' in Fig. 1B, the apparent number of binding sites was found to be 4.7 pmol/mg, and the affinity of the sites for ["Hlbumetanide, under these ionic conditions, was 45 nM. ['HIBenzmetanide also exhibited saturable binding and in a given membrane preparation the amount of specific binding was the same (within &20%) as that with ["Hlbumetanide, consistent with the hypothesis that the two compounds bind to the same membrane sites. The affinity for ['HHIbenzmetanide was somewhat higher ( K d = 22 nM under the same conditions), as expected from its greater inhibitory potency and diuretic efficacy in other preparations (15).
Partial Purification of Membranes that Bind r3H] Bumetanide-We examined different membrane fractions in a conventional differential centrifugation procedure (see "Experimental Procedures") and found that while the "plasma membrane" fraction (48,000 x g pellet) always had the highest specific binding activity, 1-5 pmol of ["H]bumetanide/mg of protein compared to 5 1 pmol/mg in the 7,500 X g pellet and 500 X g pellet, the total amount of binding was usually greater in the low speed pellets, which contained most of the protein. It appears that different size fragments of the same membranes are responsible for the wide dispersion in differential centrifugation, since when the three fractions were further purified ' At the lowest concentrations used in Fig. 1, equilibrium is attained more slowly than with 0.15 p M ['Hlbumetanide, the concentration used in Fig. 3A. Theoretical analysis of the ligand-binding site interaction indicates that for the rate constants derived from Fig. 3, and the four lowest concentrations in Fig. 1, the degree of achievement of equilibrium would be 85, 91, 95, and 99%. The analysis also indicates that this degree of disequilibrium may introduce 5 1 0 % systematic error in the direction of overestimating the number of binding sites, and 520% in the direction of underestimating the binding affinity ( 5 . Forbush, unpublished calculation).
,' Klotz (14) has pointed out that, by fitting a straight line to data on a Scatchard plot, it is arbitrarily assumed that the data reflect a single population of binding sites, having simple kinetic behavior. Although the data of Fig. 1 and our other data (not shown) are consistent with a single population of sites in either a Scatchard plot or in the plot advocated by Klotz, the data are not of sufficient quality to rule out the possibility of other populations of binding sites having lower affinities for bumetanide. This is due to the increasing scatter in the data at high [3H]bumetanide concentrations because of the increasing fraction of nonspecific binding. by centrifugation on a 32-42% (w/v) sucrose gradient, the same profile of ["Hlbumetanide binding was obtained for each. An example of this profile is shown in Fig. 2, for the 7,500 x g pellet. Membranes binding ['Hlbumetanide were found in a peak centered at 37% sucrose and were distinctly (but not completely) separated from the basolateral membrane (Na,K)-ATPase peak at 40% sucrose. Most of the protein consisted of mitochondria which were pelleted through the gradient. Although there is no well defined marker for the luminal membrane of the TALH, alkaline phosphatase activity, which is a luminal membrane marker in the proximal tubule (16), co-sedimented with the ['Hlbumetanide-binding membranes. This suggests that alkaline phosphatase may also be found in the luminal membrane in the TALH in dog kidney, although the possibility of co-sedimentation of two populations of membranes cannot be ruled out (but proximal tubule luminal membranes sediment at >40% sucrose; see below).
To further study factors affecting ["Hlbumetanide binding, we have utilized the population of membranes sedimenting at 37% sucrose in all other experiments reported in this paper. These membranes have usually been obtained as the population at the 34/39% interface on a sucrose step gradient, and they have been stored at -20 "C and thawed only once. The specific binding of these preparations was 3-15 pmol of sites/ mg of protein.
We have found considerable binding activity (0.5-1.5 pmol/ mg) in each of six membrane preparations purified from dog renal cortex by several steps of divalent cation precipitation4; these membranes are predominantly brush borders from the proximal tubule (10, 11). Significantly, when this material was further purified on a sucrose density gradient (32-44%, w/v), membranes that bound [3H]bumetanide sedimented at about 37% sucrose (as had those from outer medulla), while the brush-border membranes were found at the bottom of the gradient in fractions containing 75% of the protein and alkaline phosphatase activity (at 40-44%, w/v, sucrose; not shown). The occurrence of the Na,K,CI co-transport system in the renal cortex as well as medulla is consistent with previous observations of furosemide-sensitive transport in cortical TALH in rabbit kidney ( 5 , 6). While the particular experiments characterizing ['Hlbumetanide binding in this When membranes were prepared from renal cortex by the method used for renal medulla, we found 0.2-0.4 pmol/mg of ['Hlbumetanide binding in three preparations; however, in two other preparations, specific binding was not detectable above the background.   Fig. 3, for the same incubation conditions used in Fig. 1. T o determine the [3H]bumetanide association rate (Fig. 3A), 20-p1 samples of 0.3 p~ ['Hlbumetanide were mixed with equal volumes of 2.6 mg/ml membrane protein, and after an appropriate time, 8-pl aliquots were diluted into 5 ml of cold stop medium and filtered (see "Experimental Procedures"). As shown in Fig. 3A, the data describing the association of [3H] bumetanide with the binding site are adequately fit by a theoretical curve for the second order association of binding site and ligand, with a rate constant of Kl = 2.5 X lo4 mol-, s-', using the dissociation rate constant from Fig. 3B (K, = 1.23 X lo-" s-'). The equilibrium binding constant K = 49 nM that is obtained from the ratio of K-, and kl is in excellent agreement with the value obtained by analysis of the data shown in Fig. 1.
The dissociation rate was determined (Fig. 3B) after preincubation of 3.9 mg/ml membrane protein with 0.2 p~ [3H] bumetanide for 20 min, by dilution of 40-pl aliquots into 4.6 ml of the "binding medium" containing 100 p~ unlabeled bumetanide; after appropriate times, 200-p1 aliquots were further diluted into 5 ml of cold stop medium and filtered. In other experiments, not shown, it was found that at this dilution the rate of dissociation of ['Hlbumetanide was the same with or without the addition of unlabeled bumetanide. At 0 "C, dissociation of [3H]bumetanide is very slow (kl = 8 X s-'; upper curue, Fig. 3B) confirming that dissociation of label from this population of sites is negligible during the 10-20-s dilution-filtration/wash procedure. ['HIBumetanide dissociation at 22 "C is fit by an exponential decay curve with a rate constant of k-, = 1.23 X lo-' s-l (Fig. 3B, middle curue).
Although it is readily seen that a multiexponential curve would yield a more accurate fit, we have not investigated possible reasons for this complexity. The rate of dissociation was found to be unaffected by the ionic composition of the medium, in that the rate in the absence of Na, K, and C1 was identical with the rate in the presence of 100 mM Na, 100 mM K, and 200 mM C1 (not shown; the same as the rate of Fig. 3).
Thus, the ionic effects on the equilibrium binding levels of 13H]bumetanide, reported below, must be due to changes in the association rate constant, K1, rather than in the dissociation rate constant.
Competition by Other Loop Diuretics-To find if saturable ['Hlbumetanide binding is due to binding to the inhibitory site on the Na,K,CI co-transporter, we have examined the ability of related compounds to compete with [3H]bumetanide for the binding sites in a displacement assay; the results obtained with five compounds are presented in Fig. 4. The authenticity of the radioactive compounds ['Hlbumetanide and ['HJbenzmetanide is confirmed by the finding that the affinity for the unlabeled drugs is the same as that previously determined for the radiolabeled compounds under the same conditions (not shown). On comparing the concentrations5 at which these analogs prevent half of the ['Hlbumetanide binding in kidney membranes (K1J with the concentrations at which they inhibit Na+,K+,Cl-co-transport in isolated kidney tubules (18) and in avian red cells (15), an excellent correla-' Assuming competitive binding to a single population of sites, the affinities of the binding site for the unlabeled drugs ( K d values) are related to the K, values determined in a displacement binding assay (Fig. 4) by a proportionality factor that is dependent on the concentration of the labeled compound in relation to the affinity for the labeled compound (cf. Ref. 18); this factor is ~0 . 6 7 for the concentration of ['HJbumetanide used in the experiment of Fig. 4. tion is seen (top, Fig. 4). Note, for instance, that benzmetanide is the most potent compound in each case, whereas 4'-methoxy benzmetanide is ineffective as an inhibitor (and as a diuretic; Ref. 5) and is 300-fold less effective than benzmetanide in competition with [3H]bumetanide for binding to kidney membranes. We have also obtained an excellent correlation between the affinity of these drugs for membranes from the dogfish shark rectal gland and their corresponding inhibitory potencies in that organ, although the affinities are about 10fold lower than in kidney (7).6 We feel that these data are very strong evidence that we are measuring ['Hlbumetanide binding to the Na,K,Cl co-transport system in epithelial membranes.
Ion Requirements for Bumetanide Binding-Since earlier studies had indicated that both Na' and K' decrease the Iso for bumetanide inhibition of co-transport in avian red cells (15,19), and our preliminary data showed that [3H]bumetanide did not bind to kidney membranes in the absence of salts, we examined the appropriate ligands of the Na,K,Cl cotransport system as to their effect on saturable [3H]bumetanide binding. Fig. 5 shows the results of an experiment in which either Na', K', or C1-in the incubation medium was replaced with choline or SO4'-. It is seen that all three ions are required for high affinity [3H]bumetanide binding, since with any of the ions omitted, only a low level of saturable binding is obtained. We have found that optimal binding is obtained with 10 mM Na', 30 mM K' , 4 mM C1-(18 mM so4'-) in 12 mM histidine, pH 7.2 (Fig. 4, and from other experiments with lower total salt concentrations). Under these conditions, the Kd values for ['Hlbumetanide and [3H]benzmetanide are 30 and 515 nM, respectively (by Scatchard analysis, not shown).
The KT,, for stimulation of binding by Na' is -2 mM, and by K' " I mM (Fig. 5); these correlate well with the K , values of the transport system in the TALH of rabbit kidney: 3.6 and <3 mM, respectively (1, 6). Although there is no direct evidence that the sites stimulating binding are the same as the transport sites, we could speculate that diuretics bind to the conformation of the transport protein in which Na' and of [3H]bumetanide. Kidney membranes (~2 mg/ml) were incubated with [3H]bumetanide (0.23 PM) in the pres-c7co ence of various salts, and [3H]bumetan-z .E ide binding was determined after 20 min. Unless shown otherwise, the ion concentrations were 128 mM Na' , 64 mM K' , 128 mM c1-, 32 mM sod2-. In each experiment, the concentration of one ion K' are bound to their transport sites. This is further supported by the recent observation that bumetanide inhibition of transport in avian red cells is promoted by extracellular Na and K (15,19) with the same apparent affinities as their K, values for transport (19). We also noted that in the absence of Na+, t3H]bumetanide binding increased from less than 15% of maximal at 10 mM K' to 50% at 64 mM KC1 (Fig. 5). This suggests that K' may bind with low affinity to the Na site.
Chloride has a biphasic effect on binding: it is required a t low concentrations (KBh = 1 mM) for ['Hlbumetanide binding, but a t higher concentrations it inhibits binding. The latter finding is in agreement with that of Haas and McManus (19,20) that extracellular C1-increases the Iso for bumetanide inhibition of Na+,K+,Cl-co-transport in avian red cells. In other experiments, it was found that NO3-supports [3H] bumetanide binding in the same manner as C1-(not shown); when substituted by SO4-, NO3-stimulated binding at low concentrations (half-maximal a t 3 mM, maximal at 20-30 mM) and inhibited a t concentrations greater than 30 mM (45% inhibition at 120 mM).

DISCUSSION
We have shown that ['Hlbumetanide binds with high affinity to a population of membranes from dog kidney outer medulla. [3H]Bumetanide binding requires the presence of all three transported ions, Na+, K' , and C1-, and binding is competed for by various other "loop" diuretics at concentrations similar to those needed for inhibition of transport. We feel that this is very strong evidence that the [3H]bumetanidebinding site is the inhibitory site on the Na,K,Cl co-transport protein.
Our finding that cations are required for ['Hlbumetanide binding is in agreement with earlier reports that increasing concentrations of extracellular Na and K promote bumetanide inhibition of co-transport in avian red cells (15,19). Recently, Haas (21) has extended the red cell studies in showing that the apparent affinities with which the cations promote bumetanide inhibition are similar to those with which they are required for transport; this is the same observation we have made here with regard to bumetanide binding to kidney membranes. Since the loop diuretics including furosemide and bumetanide are negatively charged at neutral pH (pK = 5; see "Experimental Procedures"), it has been proposed that the compounds may inhibit transport by binding to the anion transport site; this hypothesis is supported by our result that C1-decreases bumetanide binding at high concentrations as well as the finding by Haas and McManus (19,20) that extracellular C1 increases the 150 for bumetanide inhibition of cotransport. Since in Fig. 5 it is seen that C1-is required at low concentrations to give optimal [3H]bumetanide binding, we suggest that the diuretic-binding site may only involve one of the two anion-binding sites, and that the stable inhibited conformation is that of the fully loaded carrier (Na' + K' + C1-+ bumetanide). However, this is clearly speculative in view of the possibility that allosteric interactions could yield similar behavior in models with separate sites and in view of the possibility that there are modifying sites for the various ions.
Since it is known that NO3does not substitute for C1-in the co-transport process (1-4), it would be expected that it would not affect L3H]bumetanide binding. Instead we found that it behaved exactly as did C1-, with somewhat lower affinities. Thus, it appears that either 1) the anion sites involved in stimulation and inhibition of [3H]bumetanide binding are not transport sites, or 2) NO3can bind to the transport sites but cannot be transported. The latter possibility could be tested in future flux experiments by looking for competitive inhibition of Na,K,C12 co-transport by NO,-.
The amount of specific [3H]bumetanide binding per weight of protein is rather low even in the best preparations from outer medulla, considering the very high rate of transport across the cells of the TALH (5). However, it is clear from the overlap with (Na,K)-ATPase activity (cf. Fig. 2) that the sucrose gradient peak is not comprised of a homogeneous population of membranes and, thus, that the specific binding activity of pure luminal membranes would be somewhat higher than attained so far. Furthermore, it is possible that the transport system is hormonally modulated (41, and is "turned off' in the membrane preparation, or that binding sites are inaccessible on inside-out vesicles; it is also possible that proteolytic degradation takes place on membrane purification. These possibilities will be investigated in future studies. Although we have no information as to the "tightness" and sidedness of membrane vesicles that most probably comprise our membrane preparation, it seems unlikely that these are important considerations with regard to the results reported here. Given the high lipid solubility of [3H]bumetanide (ether/ water partition coefficient 0.2; see "Experimental Procedures"), the lipid bilayer will not be a significant barrier to ['HIBumetanide Binding to Kidney Membranes diffusion of bumetanide into vesicles, and we can estimate that a membrane vesicle (51 ~L M in diameter) will equilibrate within milliseconds. It might be expected that entry of the ions required for bumetanide binding would be rate limiting, in vesicles of one orientation or the other. However, given the very high co-transport rate reported by Greger (5) for the intact cell membrane, we have calculated that the Na, K, and C1 would equilibrate within seconds at 37 "C (flux = 2 nmol/ cm'/s; diameter = 0.5 Fm, 40 mM salts: t,,, = 0.5 s). Thus, even a t 22 "C, we anticipate that ion concentrations will equilibrate rapidly on the time scale of the binding experiments (cf. Fig.  3).
T h e radiolabeled compounds ["Hlbumetanide and ['HI benzmetanide should be of great use in the identification, purification, and further characterization of the membrane protein responsible for Na,K,Cl co-transport. Because the binding assay is very simple and involves few assumptions, and because ion-promoted ['Hlbumetanide binding is very specific, this method appears preferable to the possible alternative assay for the co-transport system, detection of diureticsensitive co-transport after reconstitution of membrane protein into phospholipid vesicles. The high affinity of the binding site for the diuretic molecule (bumetanide Kd = 30 nM, under optimal conditions) and the number of modifications that can be made to the drug molecule with partial retention of potency (22) argue the feasibility of photoaffinity labeling as an approach to identification of the transport protein.
In addition, because of the high binding affinity, it is also hoped that these compounds will be of use in localization and characterization of the Na,K,Cl co-transporting system in tissues in which the density of transport sites is much lower than in the renal medulla. Finally, since we have shown the binding of ["Hlbumetanide to be affected by the concentrations of Na+, K' , and C1-, most probably through interactions of the ions at the binding site, we expect that further characterization of bumetanide binding will assist in gaining an understanding of the co-transport process.