Diuresis and reduced urinary osmolality in rats produced by small-molecule UT-A-selective urea transport inhibitors

Abstract

Urea transport (UT) proteins of the UT-A class are expressed in epithelial cells in kidney tubules, where they are required for the formation of a concentrated urine by countercurrent multiplication. Here, using a recently developed high-throughput assay to identify UT-A inhibitors, a screen of 50,000 synthetic small molecules identified UT-A inhibitors of aryl-thiazole, γ-sultambenzosulfonamide, aminocarbonitrile butene, and 4-isoxazolamide chemical classes. Structure-activity analysis identified compounds that inhibited UT-A selectively by a noncompetitive mechanism with IC₅₀ down to ∼1 μM. Molecular modeling identified putative inhibitor binding sites on rat UT-A. To test compound efficacy in rats, formulations and administration procedures were established to give therapeutic inhibitor concentrations in blood and urine. We found that intravenous administration of an indole thiazole or a γ-sultambenzosulfonamide at 20 mg/kg increased urine output by 3–5-fold and reduced urine osmolality by ∼2-fold compared to vehicle control rats, even under conditions of maximum antidiuresis produced by 1-deamino-8-d-arginine vasopressin (DDAVP). The diuresis was reversible and showed urea > salt excretion. The results provide proof of concept for the diuretic action of UT-A-selective inhibitors. UT-A inhibitors are first in their class salt-sparing diuretics with potential clinical indications in volume-overload edemas and high-vasopressin-associated hyponatremias.—Esteva-Font, C., Cil, O., Phuan, P.-W., Su, T., Lee, S., Anderson, M. O., Verkman, A. S. Diuresis and reduced urinary osmolality in rats produced by small-molecule UT-A-selective urea transport inhibitors.

The formation of a concentrated urine by the kidney involves a countercurrent multiplication mechanism in which a hypertonic medullary interstitium is generated by the coordinated actions of salt, water, and urea transporters (UTs; refs. 1–5). UT proteins of the UT-A class, encoded by the Slc14a2 gene, are expressed in tubule epithelial cells, and UT-B, encoded by the Slc14a1 gene, is expressed in vasa recta microvascular endothelial cells (6–15). Phenotype analysis of various UT-A-knockout (16–20) and UT-B-knockout (21) mice, as well as mathematical modeling (22), indicate that UT-A1 in inner medullary collecting duct is the UT isoform whose inhibition is predicted to have the greatest diuretic action. The potential utility of UT inhibitors as novel, salt-sparing diuretics has been discussed in several recent reviews (23–26).

Until recently, the only UT inhibitors were chemical analogs of urea with millimolar potency (27). We previously developed a high-throughput screen to identify UT-B inhibitors based on an erythrocyte lysis assay (28). UT-B inhibitors with low nanomolar IC₅₀ were identified and optimized, but produced only a mild reduction in maximum urinary concentrating function (29, 30). Other studies reported diuretic effects in rats injected with high doses of a triazolothienopyrimidine (31) or a urea analog (dimethylthiourea; ref. 32), though the inhibition selectivity and pharmacology of these compounds were not determined. More recently, recognizing the greater predicted diuretic efficacy of UT-A vs. UT-B inhibition, we developed a cell-based fluorescence screen to identify UT-A inhibitors, which produced compounds with high UT-A selectivity and others that inhibited both UT-A and UT-B (33). However, the unfavorable metabolic stability and solubility of the original UT-A inhibitors precluded their testing in animal models.

The purpose of this study was to identify small-molecule UT-A-selective inhibitors with suitable pharmacological properties to test their diuretic efficacy in rats. A high-throughput screen of 50,000 synthetic small molecules, and optimization by structure-activity analysis, yielded selective, noncompetitive inhibitors of UT-A urea transport that produced a strong diuretic response in rats, providing proof of concept for UT-A as a target for development of salt-sparing diuretics with a novel mechanism of action.

MATERIALS AND METHODS

Cell culture

Madin-Darby canine kidney (MDCK) cells stably expressing rat UT-A1 (34), yellow fluorescent protein (YFP)-H148Q/V163S (35), and human aquaporin-1 (AQP1), as described previously (33), were grown in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS), penicillin G (100 U/ml), streptomycin (100 μg/ml), zeocin (500 μg/ml), geneticin (600 μg/ml), and hygromycin (500 μg/ml) at 37°C, 5% CO_2. A cell clone with bright YFP fluorescence and relatively low AQP1 expression was used for the screen to maximize dynamic range. All reagents were purchased from Invitrogen (Grand Island, NY, USA).

Rats

Rats (Wistar males, 250–300 g) were purchased from Charles River Laboratories (Wilmington, MA, USA). Rats were used for pharmacokinetic and diuretic studies. Procedures described below were approved by the Committee on Animal Research at the University of California, San Francisco.

High-throughput screening

Primary screening was done using a collection of 50,000 chemically diverse, drug-like compounds from ChemDiv (San Diego, CA, USA). For analysis of structure-activity relationships (SARs), >300 commercially available analogs (ChemDiv and Asinex, Winston-Salem, NC, USA) were tested. Cells were plated on black 96 well Costar microplates with clear plastic bottoms (Corning, Corning, NY, USA) at 15,000 cells/well and cultured for 24 h at 37°C before assay. Microplates containing cultured cells were washed twice with PBS, and 150 μl of test compound (25 μM final) in PBS was added. Eighty wells contained test compounds, and the first and last columns of each plate contained negative [dimethyl sulfoxide (DMSO)] and positive (0.35 mM phloretin) controls. The assays were done on a plate reader (Tecan Trading AG, Männedorf, Switzerland) equipped with a custom YFP filter set. Each assay consisted of a continuous 15 s read (5 Hz) in which 50 μl of 3.2 M urea in PBS was injected at 1 s (at 130 μl/s) to give a 800 mM final urea gradient.

Synthesis procedures

All purchased materials and reagents were used without further purification. Flash chromatography was done with silica gel columns. ¹H and ¹³C NMR spectra were obtained in deuterated chloroform (CDCl₃), dimethyl sulfoxide (DMSO-d₆), or methanol (CD₃OD) using a 300 MHz Varian spectrometer (Varian Medical Systems Inc., Palo Alto, CA, USA) referenced to tetramethylsilane. Chemical shifts are expressed in hertz. Splitting patterns are designated as singlet (s), doublet (d), triplet (t), and multiplet (m). Mass spectrometry was done using a Waters liquid chromatography/mass spectrometry (LC/MS) instrument [Waters Micromass ZQ with high-performance LC (HPLC) Waters 2695; Waters Corp., Milford, MA, USA). LC was done on an Xterra MS C18 column (2.1×100 mm, 3.5 μm; Waters) with 0.2 ml/min water/acetonitrile (containing 0.1% formic acid), 16 min linear gradient, 5–95% acetonitrile. UV absorbance was detected at 254 nm.

N-(4-(2-(1H-indol-3-yl)thiazol-4-yl)phenyl)acetamide [UT-A inhibitor (UTA_inh)-E02]

A mixture of N-4-(bromoacetylphenyl) acetamide (compound 1; 61 mg) and indole-3-thiocarboxamide (compound 2; 50 mg) in 1 ml ethanol (EtOH) was refluxed for 2.5 h. Volatiles were then removed in vacuo, diluted with ethyl acetate (EtOAc) and washed with water and dried over anhydrous sodium sulfate (Na₂SO₄). The drying agent was removed by filtration, and the solvent was evaporated under reduced pressure to yield a beige-colored solid, which was purified using flash chromatography to yield UTA_inh-E02 (35 mg, 45%) as a white solid. Molecular formula: C₁₉H₁₅N₃OS; MS (ES+) (m/z) [M+1]⁺ 334; ¹H NMR (300 MHz, CD₃OD): δ 8.25-8.29 (2H, m), 7.98 (H, s), 7.96 (2H, s), 7.64-7.67 (2H, m), 7.50 (H, s), 7.46-7.49 (H, m), 7.23-7.27 (2H, m), 2.16 (3H, s); ¹³C NMR (75 MHz, CD₃OD): δ 170.2, 164.0, 154.4, 138.2, 136.9, 130.6, 126.5, 125.3, 124.6, 122.3, 120.5, 120.0, 119.8, 111.5, 111.1, 108.9, 22.7.

3-Chloro-N-(2-chloro-5-nitrophenyl)propane-1-sulfonamide (compound 4)

To a solution of 2-chloro-5-nitroaniline (compound 3; 600 mg, 3.48 mmol) in dichloromethane (7 ml) was added 3-chloropropanesulfonyl chloride (0.47 ml, 3.82 mmol) and pyridine (2.81 ml, 34.8 mmol). The reaction mixture was stirred at room temperature for 4 h, then diluted with EtOAc and washed with water. The organic layer was separated and dried over Na₂SO₄. The drying agent was filtered, and the solvent was evaporated under reduced pressure. The product was recrystallized with EtOH/acetonitrile to afford yellowish solid compound 4 (750 mg, 69%). Molecular formula: C₉H₁₀C_l2N₂O₄S; MS (ES+) (m/z) [M+1]⁺ 313; ¹H NMR (300 MHz, CD₃OD): δ 8.25-8.29 (2H, m), 7.98 (H, s), 7.96 (2H, s), 7.64-7.67 (2H, m), 7.50 (H, s), 7.23-7.27 (2H, m), 2.16 (3H, s).

2-(2-Chloro-5-nitrophenyl)-1,2-thiazolidine 1,1-dioxide (compound 5)

To a solution of compound 4 (750 mg, 2.39 mmol) in acetonitrile (20 ml) was added potassium carbonate (826 mg, 5.99 mmol). The reaction mixture was stirred overnight, then diluted with water and extracted with EtOAc. The organic layer was washed with brine and dried over Na₂SO₄. The drying agent was filtered, and solvent was evaporated under reduced pressure. The product was recrystallized with EtOH/acetonitrile to afford yellowish solid compound 5 (430 mg, 66%). Molecular formula: C₉H₉ClN₂O₄S; MS (ES+) (m/z) [M+1]⁺ 277; ¹H-NMR (300 MHz, CD₃OD): δ 8.48 (H, dd, J=0.3, 2.7 Hz), 8.25 (H, dd, J=2.7, 8.8 Hz), 7.82 (H, d, J=8.8 Hz), 3.85 (2H, t, J=6.7 Hz), 3.33 (2H, t, J=7.0 Hz), 2.58-2.63 (2H, m).

4-Chloro-3-(1,1-dioxide-1,2-thiazolidin-2-yl)aniline (compound 6)

To a solution of compound 5 (60 mg, 0.18 mmol) in EtOH (2 ml) was added Pd/C on activated charcoal; the solution was bubbled with hydrogen gas overnight. The product was filtered on a Celite pad, and the solvent was evaporated under reduced pressure, giving compound 6 (50 mg, 93%) as a white solid. Molecular formula: C₉H₁₁ClN₂O₂S MS (ES+) (m/z) [M+1]⁺ 247; ¹H NMR (300 MHz, CD₃OD): δ 7.43 (H, d, J=8.0 Hz), 7.29 (H, br), 7.05 (H, d, J=7.7 Hz), 3.95 (2H, t, J=6.7 Hz), 3.61 (2H, t, J=7.0 Hz), 2.55 (2H, m).

N-[4-Chloro-3-(1,1-dioxido-1,2-thiazolidin-2-yl)phenyl]-2-methoxy-5-methylbenzene-sulfonamide (UTA_inh-F11)

To a solution of aniline compound 6 (40 mg, 0.16 mmol) in pyridine (1 ml) was added 6-methoxy-m-toluenesulfonyl chloride (36 mg, 0.16 mmol) and stirred at 60°C for 2 h. The reaction mixture was diluted with water, and the product was extracted with EtOAc. The organic layer was separated and dried over Na₂SO₄. The drying agent was filtered, and the solvent was evaporated under reduced pressure. The product was recrystallized with EtOH/acetonitrile to afford UTA_inh-F11 (49 mg, 70%) as a white solid. Molecular formula: C₁₇H₁₉ClN₂O₅S₂; MS (ES+) (m/z) [M+1]⁺ 431; ¹H NMR (300 MHz, DMSO-d₆): δ 10.29 (H, s), 7.60-7.61 (2H, m), 7.35-7.41 (3H, m), 7.03–7.10 (2H, m), 3.81 (3H, s), 3.56 (2H, t, J=6.7 Hz), 3.38 (2H, t, J=Hz), 2.39 (2H, t, J=7.6 Hz), 2.25 (3H, s); ¹³C NMR (75 MHz, DMSO-d₆): δ154.7, 138.3, 136.0, 135.2, 131.1, 131.0, 129.7, 127.6, 125.8, 120.7, 120.3, 113.2, 56.6, 49.0, 47.1, 20.2, 19.4.

In vitro functional studies

Reversibility of UT-A1 inhibition was tested by preincubating MDCK cells expressing rat UT-A1, YFP-H148Q/V163S, and AQP1 with inhibitors for 15 min (at 1.5 μM) and then washing the cells with PBS prior to assay. The urea concentration dependence of UT-A1 inhibition was studied from inhibitor concentration-response data using different urea gradients at 0 initial intracellular urea concentration, or after preincubation of cells with different concentrations of urea for 10 min prior to a 1600 mM urea gradient. The kinetics of UT-A1 inhibition was measured by adding inhibitors (3 μM) at different time points (0–10 min) prior to assay.

UT-B inhibition measurements

As described previously (28), whole rat blood was diluted to a hematocrit of ∼1.5% in PBS containing 1.25 M acetamide and 5 mM glucose. Erythrocyte suspensions (100 μl) were added to 96-well round-bottom microplates, and test compounds (1 μl) were added. After 15 min incubation, 20 μl of the erythrocyte suspension was added rapidly to each well. Following vigorous mixing, erythrocyte lysis was quantified by absorbance at 710 nm. Nonlysed controls (isosmolar buffer) and lysed controls (0.7 mM phloretin) were added in each plate as negative and positive controls, respectively. Percentage erythrocyte lysis was computed as 100% × (A_neg − A_test)/(A_neg − A_pos), where A is absorbance at 710 nm.

Transepithelial transport measurements

UT-A1-expressing MDCK cells were grown on 12-mm-diameter collagen-coated Transwell inserts (0.4 μm pore size; Costar; Corning Inc., Corning, NY, USA) to form tight monolayers (resistance >500 Ω cm²), as described previously(28). Urea flux in the basolateral-to-apical direction was measured in response to a 15 mM urea gradient in which PBS containing forskolin (10 μM), with or without UT-A1 inhibitor and 0.7 mM phloretin, was added to both the apical-facing (0.2 ml) and basal-facing (2 ml) surfaces. After 30 min, the basal-facing solution was replaced by PBS containing test compound + 15 mM urea. Apical fluid samples (5 μl) were collected at specified times for enzymatic assay of urea (Quantichrome Urea Assay Kit; BioAssay Systems, Hayward, CA, USA). To measure inhibitor permeability across MDCK cell layers, compounds (20 μM) were added on the basal-facing surface, and fluid samples (20 μl) from the apical-facing surface were collected at different time points and analyzed by LC/MS.

Homology modeling and docking computations

A homology model of rat UT-A1 was generated using the SWISS MODEL online utility (http://swissmodel.expasy.org) (36, 37) in automated mode, using the rat UT-A1 sequence (accession code: NP_062220.2). The model was generated using coordinates from the X-ray crystal structure of bovine UT-B bound to selenourea [Protein Data Bank (PDB): 4EZC; solved to 2.5 Å; ref. 38] as a homology template. Two structural models were generated, comprising residues 105–449 (65.2% identity with bovine UT-B) and 568–909 of UT-A1 (67.5% sequence identity with bovine UT-B). Due to the slightly improved sequence identity, the latter model was used for docking computations. Because the UT-B template structure was solved as a homotrimer, the homology model was also generated in this format, and thus a single structure of the UT-A1 model was isolated for docking computations. The homology model for UT-A1 was prepared for docking using the FRED-RECEPTOR 2.2.5 utility (OpenEye Scientific, Santa Fe, NM, USA; http://www.eyesopen.com), with cytoplasmic and extracellular domains defined with 10 ų boxes. An analogous homology model of rat UT-B was prepared in a similar fashion using SWISS MODEL, with the sequence of full rat UT-B protein (accession code: P97689). Structures of small-molecule inhibitors were drawn in ChemDraw (Cambridge Software, Cambridge, MA, USA), converted to SMILES strings, transformed to 3-dimensional conformations, and minimized using Pipeline Pilot (Accelrys, San Diego, CA, USA). The single conformations were passed through Molcharge 1.5.0 (OpenEye Scientific) to apply MMFF charges (39) and through Omega 2.4.6 (OpenEye Scientific) to generate multiconformational libraries (40). The inhibitor conformational libraries were docked using FRED 2.2.5 (OpenEye Scientific; ref. 41), which was configured to use consensus scoring, using the scoring functions ChemGauss3, ChemScore, OEChemScore, ScreenScore, ShapeGauss, PLP, and ZapBind. Docking of the inhibitors was carried out free of pharmacophore restraint. The final protein-inhibitor complexes were visualized using PyMol (Schrödinger, San Diego, CA, USA).

In vitro metabolic stability

As described previously (42), compounds (each 5 μM) were incubated for specified times at 37°C with rat liver microsomes (1 mg protein/ml; Sigma-Aldrich, St. Louis, MO, USA) in potassium phosphate buffer (100 mM) containing 1 mM NADPH. The mixture was then chilled on ice, and 0.5 ml of ice-cold EtOAc was added. Samples were centrifuged for 15 min at 3000 rpm, the supernatant evaporated to dryness, and the residue was dissolved in 100 μl mobile phase (acetonitrile:water, 3:1, containing 0.1% formic acid) for LC/MS. Reverse-phase HPLC separations were carried out using a Xterra MS C18 column (2.1×100 mm, 3.5 μm; Waters) equipped with a solvent delivery system (Waters 2695). The solvent system consisted of a linear gradient from 5–95% acetonitrile containing 0.1% formic acid, run over 16 min (0.2 ml/min flow rate).

Cell toxicity and off-target effects

MDCK cells on black 96-well Costar microplates with clear plastic bottoms were cultured for 24 h at 37°C and incubated with test compounds (0–25 μM) for 24 h. Cell viability was assayed using AlamarBlue (Pierce, Rockford, IL, USA) in which the reagent was added to each well and incubated for 30 min, and absorbance was measured at 590 nm. For short-circuit current measurement, snapwell inserts containing MDCK cells were mounted on Ussing chambers. Test compounds (20 μM) were incubated with MDCK cells at both apical and basolateral sides for 10 min at 37°C prior to addition of 20 μM forskolin and then 100 μM ATP. The apical and basolateral chambers contained identical solutions: 120 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, 5 mM HEPES, 25 mM NaHCO₃, and 10 mM glucose. Solutions were bubbled with 5% CO₂/95% O₂ and maintained at 37°C. Hemichambers were connected to a DVC-1000 voltage clamp (World Precision Instruments Inc., Sarasota, FL, USA) via Ag/AgCl electrodes and 1 M KCl agar bridges for current recording.

Pharmacokinetics

UTA_inh-E02 or UTA_inh-F11 was administered at 20 mg/kg intravenously by tail vein to rats. UTA_inh-E02 was formulated as a 5 mg/ml saline solution containing 20% dimethylacetamide and 40% 2-hydroxypropyl γ-cyclodextrin; UTA_inh-F11 was formulated as a 5 mg/ml saline solution containing 20% dimethylacetamide and 0.6 mg/ml NaOH. Blood samples were collected from the tail vein into K3EDTA minitubes (Greiner, Kremsmunster, Austria) at specified time points. Plasma samples were separated by centrifugation and kept frozen at −20°C until analysis. Urine samples were also collected. Calibration standards were prepared in urine and plasma from control (nontreated) rats to which known amounts of UTA_inh-E02 or UTA_inh-F11 were added. Sixty microliters of sample was mixed with an equal amount of acetonitrile containing 0.1% formic acid. The mixture was centrifuged for 20 min at 13,200 rpm, and 90 μl of supernatant was taken for LC/MS. The solvent system consisted of a linear gradient from 5 to 95% acetonitrile containing 0.1% formic acid over 16 min (0.2 ml/min flow). Mass spectra was acquired on a mass spectrometer (Waters 2695 and Micromass ZQ) using electrospray (+) ionization, mass ranging from 100 to 1500 Da, cone voltage 40 V.

Diuretic studies in rats

Compound effects on urine output, osmolality, and urea concentration were investigated in two conditions. Rats were placed in metabolic cages with water and food available ad libitum, and spontaneously voided urine was collected for 3 h. Rats were then injected intravenously with 20 mg/kg UTA_inh-E02 or UTA_inh-F11 in formulations listed above (or vehicle alone), and returned to the metabolic cages without access to water, but free access to food. Spontaneously voided urine was collected every 3 h for a 6 h period. In some experiments, rats were administered 1-deamino-8-d-arginine vasopressin (DDAVP; 4 μg/kg, i.p.) just before compound (or vehicle) injection and 3 h later. Rats did not have access to water, but had free access to food in DDAVP experiments. Urine osmolality was measured in water-diluted urine samples by freezing-point osmometry (Micro-osmometer; Precision Systems, Natick, MA, USA). Urea concentration was determined by a colorimetric enzymatic assay as described above.

RESULTS

UT-A1 inhibitors identified in a small-molecule screen

To identify UT-A inhibitors suitable for testing in rats, a collection of 50,000 synthetic small molecules was screened. As diagrammed in Fig. 1A (top panel), the fluorescence cell-based screen utilized MDCK cells stably expressing UT-A1, water channel AQP1, and the cytoplasmic chloride-sensing fluorescent protein YFP-H148Q/V163S. As described and validated previously (33), extracellular addition of urea results in osmotic water efflux and cell shrinking, which increases intracellular chloride concentration and reduces YFP fluorescence. Urea entry results in increasing YFP fluorescence as cells re-swell. Inhibitors of UT-A1 urea transport produce a lesser reduction in fluorescence and slowed recovery. Figure 1A (bottom panel) shows concentration-dependence data for two active compounds identified in the primary screen, with positive (nonselective UT inhibitor phloretin) and negative (vehicle) controls.

Figure 1.

Identification of small-molecule UT-A1 inhibitors by high-throughput screening. A) Top panel: assay method showing rapid exposure of MDCK cells stably expressing UT-A1, AQP1, and YFP-H148Q/V163S to a 800 mM inward urea gradient. A decrease in cell volume (reduced fluorescence) due to AQP1-mediated water efflux is followed by cell swelling (increased fluorescence) due to urea and water influx. Bottom panel: positive (phloretin, red curve) and negative (DMSO, green curve) controls with concentration-dependence data for two active compounds. B) Summary of screening results. C) Chemical structures of UT-A inhibitors identified in the screen shown with prior reported UT-A inhibitors described in ref. 33.

The primary screen produced 13 compounds giving >98% UT-A1 inhibition at 25 μM, with compounds of 4 distinct chemical scaffolds having IC₅₀ < 10 μM (Fig. 1B). Figure 1C shows chemical structures of 4 classes of verified UT-A1 inhibitors identified in the screen: an aryl-thiazole (UTA_inh-E01), γ-sultambenzosulfonamide (UTA_inh-F01), aminocarbonitrile butene (UTA_inh-G01), and 4-isoxazolamide (UTA_inh-H01). These four chemical scaffolds are different from previously reported urea transport inhibitors, whose structures are shown for comparison.

Structure-activity analysis of UT-A1 inhibitors

Of the inhibitor classes shown in Fig. 1C, two chemical classes, E and F (nomenclature UTA_inh-Exx and UTA_inh-Fxx), were found to have suitable metabolic stability and pharmacokinetics in rats (see below). Figure 2A shows concentration-inhibition data of representative class E and F compounds. A small set of 50 commercially available class E analogs were tested, with the best compound being UTA_inh-E02, with IC₅₀ ∼ 1 μM. Preliminary SAR analysis (Fig. 2B) showed that an acetamide group on the aniline ring and a thiazole was best, while larger alkyl amides reduced UT-A1 inhibition. SAR analysis of 200 commercially available class F analogs gave 20 compounds that produced > 9% UT-A1 inhibition at 25 μM. The most potent compounds, UTA_inh-F01, -F09, -F11, and -F17, gave similar values of IC₅₀ ∼ 1–2.5 μM. Several class F compounds also inhibited the other major UT isoform, UT-B (see below). Figure 3 summarizes inhibition data for active class E and F compounds. The key structural determinants of class F compounds for UT-A1 inhibition activity, as deduced from the compounds listed in Fig. 3 and many inactive analogs, are summarized in Fig. 2C. Class F compounds are 1,3-diaminobenzenes with a benzosulfonamide group and a 5-member γ-sultam cyclic ring attached on the 1- and 3-amino positions, respectively (UTA_inh-F01 to -F13). Analogs based on the 1,4-diaminobenzene ring were inactive, as was a 5-oxo group on the γ-sultam ring. Of note, analogs with the sultam endocyclically attached to 1,3-diaminobenzene are active (UTA_inh-F14 to -F20). Further substitutions on the 1,3-diaminobenzene ring affected activity: 4-chloro and 6-methoxy (UTA_inh-F01, -F09, and -F11) were tolerated, while a 6-methyl group reduced activity (UTA_inh-F05 vs. -F08). Disubstituted benzosulfonamide gave good activity, with the 2-methoxy-5-methyl substituent giving the best UT-A1 inhibition (UTA_inh-F01 and -F17).

Figure 2.

Structure-activity analysis of UT-A inhibitors. A) Concentration-inhibition data for UT-A1 inhibition by indicated compounds (means±se, n=3). Fit to single-site inhibition model shown. B) Structural determinants of class E compounds for UT-A1 inhibition. C) Structural determinants of class F compounds. D) Synthesis of UTA_inh-E02 and UTA_inh-F11. a) EtOH, Et₃N, reflux, 45%. b) 3-Chloropropyl sulfonyl chloride, pyridine, CH₂Cl₂, 69%. c) K₂CO₃, CH₃CN, 66%. d) Pd/C, H₂, EtOH, 93%. e) ArSO₂Cl, pyridine, 60°C, 70%. Arabic numbers indicate compounds 1–6; see Materials and Methods.

Figure 3.

Structure-activity analysis of class E and F inhibitors.

For further analysis, UTA_inh-E02 and UTA_inh-F11 were resynthesized, purified to >99%, and characterized. Condensation of commercially available N-4-(bromoacetylphenyl)acetamide (compound 1) and indole-3-thiocarboxamide (compound 2) gave UTA_inh-E02 in moderate yield (Fig. 2D, top diagram). Synthesis of 3-γ-sultam-1-benzosulfonamidebenzene UTA_inh-F11 was achieved in 4 steps (Fig. 2D, bottom diagram). Commercial 2-chloro-5-nitroaniline (compound 3) was sulfonated with 3-chloropropylsulfonyl chloride to give sultam precursor, compound 4, which was cyclized under basic conditions using K₂CO₃ to yield nitro-γ-sultam, compound 5. Palladium-catalyzed reduction under atmospheric hydrogen efficiently converted nitro-γ-sultam (compound 5) to amino-γ-sultam (compound 6). Finally, condensation of compound 6 with 2-methoxy-5-methylbenzosulfonyl chloride under basic conditions gave UTA_inh-F11 in good yield.

In vitro characterization of UT-A inhibition

Inhibitor reversibility was studied by incubation of the UT-A1-expressing cells with compounds for 15 min, then washing, followed by assay of UT-A1 inhibition. UT-A1 urea transport after inhibitor washout was the same as before inhibitor addition for all compounds (Fig. 4A), indicating full reversibility. Inhibitor competition with urea, as might occur by steric hindrance if the inhibitor and urea occupied physically overlapping sites on the UT-A1 protein, was studied from the urea concentration-dependence of apparent inhibitor IC₅₀. Studies were done with 0 urea inside cells initially and different urea concentrations outside, and with different urea concentrations inside cells initially and the same inward urea gradient. Figure 4B shows similar IC₅₀ values with different urea concentrations under each set of conditions, providing evidence for a noncompetitive inhibition mechanism, which is desirable as urea concentration is very high in renal inner medulla.

Figure 4.

In vitro characterization of UT-A inhibitors. A) Reversibility studied by incubation with inhibitors (at 1.5 μM) with the transfected MDCK cells for 15 min, washing for 15 min, and then assay for UT-A1 inhibition. B) Left panels: urea competition studied by assay of UT-A1 inhibition using different urea concentrations (curves). Right panels: apparent IC₅₀ as a function of extracellular urea concentration, [urea]_e, at 0 initial intracellular urea concentration (third panel), and as a function of intracellular urea concentration, [urea]_i, for fixed 1600 mM urea gradient (right panel). C) Left panels: inhibition kinetics studied by assay of UT-A1 urea transport at different times after addition of 3 μM inhibitor. Top right panel: summary of kinetic data (means± se, n=3). Bottom right panel: inhibitor permeability across MDCK cell monolayers on porous filters (means±se, n=3). D) Top panel: inhibitor selectivity studied by measurement of UT-B inhibition by an erythrocyte lysis assay in rat erythrocytes. Bottom panels: UT-B concentration-inhibition data. E) Transepithelial urea transport in UT-A1-expressing MDCK cells. Cells were treated with 10 μM forskolin alone, forskolin + phloretin (0.7 mM), or forskolin plus inhibitor at 3 and 20 μM (means±se, n=3).

To further evaluate the potential site of inhibitor action, the kinetics of UT-A1 inhibition was measured by assay of UT-A1 inhibition at different times after compound addition. UTA_inh-E02 inhibition occurred over minutes, suggesting an intracellular site of action, whereas inhibition by F-class compounds was more rapid and would be consistent with an extracellular site of action if their transport into cells is slow compared with their inhibition kinetics (Fig. 4C, top left and right panels). To assess inhibitor membrane permeability, we measured transport of UTA_inh-E02 and UTA_inh-F11 across MDCK monolayers grown on porous filters by LC/MS. Transport of UTA_inh-F11 was much faster than that of UTA_inh-E02 (Fig. 4C, bottom right panel), with computed transepithelial permeability coefficients of 7 × 10⁻⁷ and 1.1 × 10⁻⁵ cm/s, respectively. From surface-to-volume considerations, we compute that the equilibration times of UTA_inh-E02 and UTA_inh-F11 are 250 and 14 s, respectively, in MDCK cells, which are consistent with intracellular sites of action of both inhibitors.

Inhibitor selectivity for rat UT-A1 vs. UT-B was investigated using an erythrocyte lysis assay of rat UT-B urea transport, which involved measurement of hypotonic lysis (by near-infrared absorbance) of acetamide-loaded erythrocytes following rapid dilution into acetamide-free buffer. Concentration-inhibition curves in Fig. 4D (top panel) show high selectivity of UTA_inh-E01 and UTA_inh-E02 for UT-A1 inhibition; data in Fig. 4D (bottom panels) shows moderate UT-A1 vs. UT-B selectivity for class F compounds.

Last, to verify UT-A1 inhibition using an independent cell-based assay that does not rely on water transport or fluorescent dyes, urea transport was measured in UT-A1-transfected MDCK cells cultured on porous filters. Transepithelial urea transport from the basolateral to the apical solution was measured by enzymatic assay of urea in the apical solution following urea addition to the basal-facing solution. Urea permeability in this cell model was increased by forskolin, and reduced by high concentrations (20 μM) of UTA_inh-E02 or UTA_inh-F11 to that of phloretin-treated cells (Fig. 4E); inhibitor concentrations of 3 μM, near their IC₅₀ values found in plate reader assays, produced ∼50% inhibition.

Computational modeling was done to identify putative binding sites and bound conformations of UTA_inh-E02 and UTA_inh-F11. Structures for rat UT-A1 and UT-B were generated by homology modeling based on the high-resolution X-ray crystal structure of bovine UT-B solved at 2.5 Å (PDB: 4EZD; ref. 38). There is at present no structural information on UT-A proteins, nor are there high-resolution structure data for rat UT-B. Several structural features were observed in the bovine UT-B X-ray crystal structure, including low-energy binding sites for urea, identified as S_o and S_i, adjacent to hydrogen bond acceptors, Gln²²⁷ and Gln⁶³, respectively, as well as a narrow constriction region identified as S_m, consisting of Thr³³⁴ and Thr¹⁷² (39). Notably, our homology models of rat UT-A and UT-B contained homologous residues corresponding to these sites, and were positioned similarly in the central pore region. In the model of rat UT-A, the S₀ and S_i sites were adjacent to Gln⁷⁶³ and Gln⁵⁹⁹, respectively, while the S_m constriction site consists of Thr⁸⁷⁰ and Thr⁷⁰⁸.

Ligand and receptor preparation, as well as docking simulations, were performed with the OpenEye Scientific suite of utilities, including the OMEGA (43) and FRED 2.2.5 (41) software. UTA_inh-E02 and UTA_inh-F11 were docked into the extracellular and cytoplasmic domains of rat UT-A1 and UT-B homology models. Active inhibitors docked into the extracellular domain, as well as a selection of inactive inhibitor analogs docked into either domain, showed low-energy binding poses, and appeared to bind nonspecifically. Docked conformations of UTA_inh-E02 and UTA_inh-F11 are presented in Fig. 5. The lowest energy docked pose of aryl-thiazole UTA_inh-E02 into rat UT-A1 orients the indole ring into the pore in the vicinity of Gln⁵⁹⁹ at position S_i, with the central thiazole and attached aromatic ring surrounded by several hydrophobic residues lining the outer region of the pore, including Leu⁶⁵², Leu⁸⁹⁵, Glu⁵⁷², and Phe⁸³², as well as Ser⁷⁰⁰ (Fig. 5A). The less selective γ-sultambenzosulfonamide UTA_inh-F11 is shown in its lowest energy docked pose into rat UT-A1 (Fig. 5B) and rat UT-B (Fig. 5C). This class of inhibitors docked best with the substituted aryl sulfonamide motif orienting into the pore, with the 5-atom cyclic sultam motif interacting with the pocket of hydrophobic residues (listed above) that line the outer pore. UTA1_inh-F11 docked in a similar manner into UT-B, with the substituted aryl sulfonamide oriented into the pore, and the cyclic sultam group positioned in the outer hydrophobic pocket.

Figure 5.

Computational modeling of UT-A-selective and nonselective inhibitors. Putative inhibitor binding sites on rat UT-A1 and rat UT-B based on functional measurements, homology modeling, and computational docking. Zoomed-in and zoomed-out representations of UTA_inh-E02 (A) and UTA_inh-F11 (B) bound to the rat UT-A1 cytoplasmic domain, and UTA_inh-F11 (C) bound to the rat UT-B cytoplasmic domain.

Pharmacological properties of UT-A inhibitors in rats

In vitro metabolic stability was measured by LC/MS after incubation with rat hepatic microsomes and NADPH. Figure 6A shows the kinetics of disappearance of the original (nonmetabolized) inhibitors UTA_inh-E02 and UTA_inh-F11. At 60 min, ∼50% metabolism was found, which represents reasonable in vitro metabolic stability, much better than that for several other inhibitor classes where the t_1/2 for compound disappearance was 15 min or less (data not shown).

Figure 6.

UT-A inhibitor pharmacology. A) In vitro metabolic stability measured in rat hepatic microsomes. Kinetics of disappearance of indicated compounds (at 5 μM) following incubation with hepatic microsomes and NADPH, showing original LC/MS traces and time course data (means±se, n=3). B) Left panel: kinetics of plasma UTA_inh-E02 concentration following bolus intravenous administration of 5 mg UTA_inh-E02 in saline containing 20% dimethylacetamide and 40% γ-hydroxypropyl cyclodextrin. Inset: original LC/MS traces. Second panel: UTA_inh-E02 concentration in urine collected at 0–3 and 3–6 h. Right panels: similar analysis done for UTA_inh-F11 following bolus intravenous administration of 5 mg in saline containing 20% dimethylacetamide and 0.6 mg/ml NaOH. C) Left panel: cytotoxicity measured by Alamar blue assay in MDCK cells treated for 24 hs with the indicated [compounds] (means±SE, n=3). Middle panel: short-circuit current recordings. Cells were incubated at 37°C for 10 min with 20 μM UTA_inh-E02 or 20 μM UTA_inh-F11 (or DMSO, vehicle) before addition of 20 μM forskolin and 100 μM ATP. Right panel: summary of changes in short-circuit current (ΔI_sc) produced by forskolin alone and forskolin + ATP (means±se, n=3 cultures each).

Several vehicles and administration routes were tested in order to give predicted therapeutic inhibitor concentrations in blood and urine. Selection of vehicle for administration of UTA_inh-E02 was challenging because of its limited solubility, as various combinations of vehicles gave low (≪1 μM) plasma and urine concentrations after intraperitoneal, subcutaneous, or oral administration of up to 100 mg/kg. We found that intravenous administration of 20 mg/kg of UTA_inh-E02 (5 mg/ml in saline, 20% dimethylacetamide, 40% 2-hydroxypropyl γ-cyclodextrin) to rats yielded ∼6 μM plasma and ∼3 μM urine levels initially (Fig. 6B, left panels), with plasma elimination t_1/2 ∼ 4.5 h. UTA_inh-F11 dissolved well in 20% dimethylacetamide and 0.6 mg/ml NaOH in saline. Because of the alkaline pH of the administered compounds, the vehicle control for UTA_inh-F11 was 20% dimethylacetamide and 1.2 mg/ml NaHCO₃ in saline to give identical pH. Intravenous administration of 20 mg/kg UTA_inh-F11 to rats yielded higher initial plasma levels than UTA_inh-E02, but lower t_1/2 ∼ 1 h (Fig. 6B, right panels). Urine levels of UTA_inh-F11 were > 3 μM over 6 h.

Compound toxicity was evaluated by compound incubation for 24 h with MDCK cell cultures followed by Alamar blue assay. Figure 6C (left panel) shows little toxicity of UTA_inh-E02 and UTA_inh-F11 at up to 25 μM, near their solubility limits. To assess potential ion channel off-target effects, short-circuit current was measured in MDCK cells in response to cAMP (forskolin) and calcium (ATP) agonists (Fig. 6C, right panels), which depends on the actions of multiple anion (CFTR, calcium-activated chloride channels) and cation (K⁺, Na⁺) channels. Neither inhibitor at 20 μM altered short-circuit current.

Diuretic action of UT-A inhibitors in rats

Diuretic efficacy was tested in rats under two different conditions. Inhibitor effects on maximum urinary concentrating ability was studied in rats administered the V2-selective agonist DDAVP at 4 μg/kg every 3 h. Inhibitors were administered intravenously at 20 mg/kg, as done in the pharmacokinetics experiments above. Figure 7A shows similar urine volume and osmolality of vehicle and treatment groups prior to inhibitor administration (−3 to 0 h). Each inhibitor produced a marked increase in volume and reduction in osmolality in urine collected over 0–3 h. Little difference was seen at 3–6 h, which is consistent with pharmacokinetics data, showing reversible inhibitor action. Figure 7B shows that the diuresis produced by UTA_inh-E02 was relatively urea selective, as shown by the increased ratio of urea vs. nonurea solutes in urine after inhibitor treatment.

Figure 7.

Diuretic action of UT-A inhibitors in rats. A) Maximal urinary concentration was produced by DDAVP (4 μg/kg, i.p., every 3 h) and dehydration. Inhibitors (5 mg) administered by intravenous injection at time 0. Urine volume (left panel) and osmolality (right panel) for indicated 3 h collections. Bars represent means ± se, 4 rats/group. B) Excretion of urea vs. nonurea solutes (expressed as ratio) shown at baseline (−3 to 0 h collection) and after inhibitor treatment (0 to 6 h collection). Bars represent means ± se. C) Study done as in A, but with control/hydrated rats. Bars represent means ± se, 3 rats/group. *P < 0.05, **P < 0.01.

In a separate set of studies, UTA_inh-E02 was tested in control hydrated rats. Figure 7C shows significantly increased urine volume and reduced urine osmolality at 0–3 h after UTA_inh-E02 administration. Urine volume and osmolality returned to near baseline in the 3–6 h collection.

DISCUSSION

A small-molecule screen identified compounds that selectively inhibited UT-A urea transport with low micromolar potency by a noncompetitive inhibition mechanism. The screen was done against isoform UT-A1, the rate-limiting urea transporter expressed on the luminal membrane of epithelial cells in kidney inner medullary collecting duct (9). UT-A1 inhibitors probably also inhibit isoform UT-A3, which is expressed at the basolateral membrane in the same cells and is highly homologous to UT-A1, as UT-A1 consists of one UT-A3 and one UT-A2 molecule in tandem. Also, urinary concentrating ability was not impaired in UT-A1/A3-double-knockout mice after transgenic replacement of UT-A1 (20). The other renal UT-A isoform, UT-A2, is expressed in the thin descending limb of Henle's loop (9), but appears to play a minimal role, as urinary concentrating function is unimpaired in UT-A2-knockout mice (19). The relatively mild urinary concentrating defect in UT-B-knockout mice (21) and in humans with loss of function mutations in UT-B (44, 45) suggests a much less important role for UT-B compared to UT-A isoforms in urinary concentrating function, which is supported here by the marked diuresis in rats produced by a UT-A1 selective inhibitor. On theoretical grounds, inhibition of urea transport in inner medullary collecting duct would prevent uptake of urea from luminal fluid into the medullary interstitium and hence reduce interstitial osmolality. Inhibition of UT-B, which impairs urea uptake from the interstitium into the renal microvasculature, is predicted to have minimal further effect when little urea is delivered to the interstitium with UT-A1 inhibition.

To the best of our knowledge, there are no previously reported biological activities for the γ-sultambenzosulfonamide compound class identified here. Though sulfonamides are commonly found in drugs and bioactive compounds, such as in antidiabetic sulfonylureas, there are no reported biological activities of a γ-sultam and benzosulfoaminde with 1,3-diaminobenzene linkage, as required for UT-A1 inhibition. SAR analysis of the γ-sultambenzosulfonamide analogs showed that both the geometry and polarity of substituents are important for inhibition. The 1,4-diaminobenzene linkage resulted in loss of activity, and the activity of 1,3-diaminobenzene analogs depended on the phenylsulfonamide substituents, with greater activity of ortho-methoxy and methyl substituents and reduced activity with para-methoxy, halide, and methyl substituents. For class E inhibitors, UTA_inh-E02 has been reported as a PIM-1 kinase inhibitor with micromolar potency (46), and the more general aryl-thiazole scaffold is reported to have antimicrobial (47) and antitumor (48) activities. The two chemical classes discovered in this study have drug-like properties, including the presence of multiple hydrogen bonding acceptors, favorable molecule weight, aLogP, and topological polar surface areas. The average molecular weights are 333 and 421 for class E and F compounds, respectively; average aLogP values are 4.6 and 2.0, and average topological polar surface areas are 86 and 95 Ų. These values are within the Lipinski (49) and Veber (50) criteria for orally bioavailable drugs. Facile synthesis of these two compound classes, as shown in this study, will allow rapid preparation of targeted chemical analogs to improve inhibition activities and physicochemical properties.

In vitro functional studies indicated a noncompetitive mechanism for UTA_inh-E02 and UTA_inh-F11 inhibition of urea transport. A cytoplasmic binding mode deduced by molecular docking was supported experimentally by the noncompetitive inhibition mechanism and inhibitor membrane permeabilities. Notwithstanding the limitations of homology modeling of membrane transport proteins and of molecular docking computations, UTA_inh-E02 and UTA1_inh-F11 appear to be oriented in the pore region, with additional structural features positioned to interact with an outer hydrophobic pocket. Further analysis of binding sites will require crystal structure information and/or mutagenesis studies.

In summary, four chemical classes of UT-A1 inhibitors were identified, two of which were tested in rats and found to produce a marked diuresis. Functional studies and homology/docking computations suggested putative inhibitor binding sites on the intracellular-facing surface of the UT-A1 protein. The rat data provide proof of concept for application of UT-A1 inhibitors as salt-sparing diuretics, or “urearetics,” with a novel mechanism of action. UT-blocking diuretics may have utility alone, or in combination with conventional salt transport-blocking diuretics, in edema due to fluid overload (congestive heart failure, nephrotic syndrome, and cirrhosis) and in hyponatremia due to chronic elevation in vasopressin (syndrome of inappropriate antidiuretic hormone secretion). Because of their unique mechanism of action on renal countercurrent multiplication, UT-A1 inhibitors may be effective in states of refractory edema where conventional diuretics such as furosemide and thiazides have limited efficacy. Following optimization of their pharmacological properties, testing of UT-A1 inhibitors in clinically relevant animal models of edema and hyponatremia will be important to define their potential clinical indications.