Epithelial plasma membrane transporters as drug targets

Abstract

Small molecule discovery and drug development are increasingly being pursued in academic settings, expanding beyond their traditional confinement to the pharmaceutical industry. The initial steps in drug discovery typically include identification and validation of a target, screening of chemical libraries to identify modulators of target activity, and subsequent prioritization and optimization of lead compounds using in vitro systems and animal models, with emphasis on compound potency, selectivity and pharmacological properties. This review focuses on early-stage discovery of small molecules that target plasma membrane transporters on epithelial cells, including absorptive and secretory epithelia in kidney, gastrointestinal tract, lung and eye. Of the estimated 500 distinct epithelial plasma membrane transporters, fewer than a dozen are the targets of approved drugs, most of which have been in clinical use for decades. We discuss the logistics and challenges associated with small molecule discovery in an academic setting. Specific epithelial cell targets are considered, including chloride channels, solute-coupled transporters, urea transporters and aquaporins, with therapeutic implications spanning constipation and secretory diarrheas, cystic fibrosis, dry eye disease, edema, hypertension and kidney stones. We conclude by identifying unmet needs and outlining opportunities to enable next-generation pharmacological modulation of epithelial transport processes.

Graphical Abstract

Clinical Highlights Summary

Although relatively few approved drugs directly target epithelial transport proteins, those that do have had substantial clinical impact. Established examples include diuretics for hypertension and edema, probenecid for gout, and proton–potassium ATPase inhibitors for acid-related disorders. Pharmacological modulators of CFTR have transformed outcomes in cystic fibrosis, demonstrating the therapeutic power of transporter-directed strategies. SGLT2 inhibitors not only improve glycemic control but also slow progression of chronic kidney disease, and bile acid transport inhibitors are approved for constipation and cholestatic disorders. Emerging agents targeting epithelial ion channels, solute carriers, aquaporins, and urea transporters are in development for a broad spectrum of indications, including secretory diarrheas, constipation, dry eye disease, nephrolithiasis, polycystic kidney disease, inflammatory lung disease, refractory edema, chronic kidney disease, acid–base disorders, heart failure, and pancreatitis. These advances underscore the expanding therapeutic landscape of epithelial transporter pharmacology and its potential to address major unmet clinical needs.

Epithelial transporters as targets for drug discovery

Of the ~20,000 distinct protein-encoding human genes, approximately 6000 encode membrane proteins, and of those, ~500 encode epithelial plasma membrane transporters. Despite their central roles in epithelial fluid, electrolyte, and solute homeostasis, only a small fraction of these transporters are targeted by approved therapeutics, and most such drugs have been in clinical use for decades (1). Recent approvals targeting previously untapped epithelial transport mechanisms including ivacaftor (in 2012), a potentiator of the CFTR chloride channel (ABCC7); canagliflozin (in 2013), an inhibitor of the sodium–glucose cotransporter SGLT2 (SLC5A2) (2); and elobixibat (in 2021), which targets the ileal bile acid transporter (SLC10A2) (3). CFTR as a drug target is discussed in detail herein. Following the introduction of canagliflozin, several additional SGLT2 inhibitors were approved, with indications extending beyond glycemic control to cardiovascular risk reduction and preservation of renal function in type 2 diabetes. The dual SGLT1/2 inhibitor sotagliflozin, approved in 2023, further expands this strategy by targeting intestinal glucose absorption. Inhibitors of hepatic and ileal bile acid transporters are now approved for chronic constipation and cholestatic disorders, including progressive familial intrahepatic cholestasis and Alagille syndrome–associated pruritus. These advances highlight the therapeutic potential of epithelial transporter targeting and the striking underrepresentation of this target class in modern drug development.

Diuretics targeting renal tubule epithelial transporters have been used for decades (4). Hydrochlorothiazide, approved in 1959, targets the sodium-chloride cotransporter NCC (SLC12A3) in the distal convoluted tubule; the loop diuretic furosemide, approved in 1966, targets the sodium-potassium-chloride cotransporter NKCC2 (SLC12A1) in the thick ascending limb of Henle; and the potassium-sparing diuretic amiloride, approved in 1981, targets the epithelial sodium channel ENaC (SCNN1A/B/G) in late distal tubules and collecting duct. Other older drugs targeting epithelial plasma membrane transporters include probenecid, which was approved in 1951 for treatment of gout and hyperuricemia, and omeprazole in 1989 for treatment of acid-related gastric disorders. Probenecid targets the uric acid transporter URAT1 (SLC22A12) as well as organic acid transporters (OAT1/3, SLC22A6/SLC22A8). Omeprazole targets the gastric proton-potassium pump (ATP4A/B). Digoxin, which was approved in 1954, inhibits the sodium-potassium pump (ATP1A1) to increase cardiac contractility and treat cardiac rhythm disturbances. In epithelial cells, inhibition of pump function by digoxin is an undesired, on-target effect. Though not targeting epithelial cell transport directly, many drugs can alter the absorption or clearance of epithelia-targeted drugs, such as cyclosporin and ritonavir, which block P-glycoprotein (ABCB1), and drugs which block organic acid transporters (OATs), including cimetidine, metformin, and some antivirals, statins, non-steroidal anti-inflammatory drugs, antibiotics, and anti-cancer drugs (5).

In the sections to follow, we examine the scientific and logistical challenges inherent to small molecule discovery targeting epithelial plasma membrane transporters, with emphasis on efforts originating in academic laboratories. Using examples from preclinical studies of epithelial ion, urea, and water transport, we highlight key determinants of successful discovery programs, including assay design, high-throughput screening strategies, compound prioritization, and in vivo validation. We also opine on unmet needs and opportunities in epithelial transport drug development. This review focuses on small molecule modulators rather than biologics, as antibody- and peptide-based agents have thus far demonstrated limited utility as direct regulators of epithelial transporter function. Given its primary audience of academic investigators, the emphasis is placed on preclinical discovery, with only limited consideration of clinical development and post-approval issues.

The drug discovery process from an academic perspective

Fig. 1 diagrams a representative workflow for small molecule discovery, spanning target identification through early lead optimization (6). Central to this process is rigorous target selection and validation, which can utilize several convergent lines of evidence including phenotype data on humans and experimental animal models with target mutations, pharmacological data for existing modulators, target expression profiling, and mechanistic data on target biology. Although target-based screens are generally most straightforward, phenotype-based screens may be appropriate if target identity is uncertain and suitable cell models and phenotype screening assays are available (7).

Figure 1.

Typical workflow for a small molecule discovery project. Following target validation, active compounds identified in a screen are verified, characterized, prioritized and optimized. Lead candidates emerging may serve as research tools and potentially as drug development candidates.

Development of a high-throughput screening assay, capable of automated screening of tens to hundreds of thousands of candidate molecules, is a pivotal component of the discovery process and often demands considerable technical innovation and optimization (8). Widely used optical readouts in multi-well plate format include integrated fluorescence (intensity, energy transfer, polarization, lifetime), luminescence or absorption. Where pathway complexity or subcellular phenotypes are central to the biology, high-content imaging platforms permit multiparameter quantification of cellular morphology, protein localization, and signaling dynamics. Beyond optics, specialized instruments support automated electrophysiological measurements of ion channel activity, including patch-clamp, as well as flow cytometers, mass spectrometers, surface plasmon resonance instruments, and liquid handling systems capable of carrying out complex, multi-step assays.

A quantitative and crucial index of assay quality is the Z’-factor, which provides a measure of the intrinsic variability of assay readouts normalized by the separation between positive and negative controls (9): Z’ = 1 – 3 (SD_pos-SD_neg) / (A_pos-A_neg), where SD_pos and SD_neg are standard deviations for positive and negative controls, respectively, and A_pos and A_neg are corresponding assay signals. A Z’ factor of 1 represents a perfect assay, Z’ > 0.5 is generally acceptable, and Z’ < 0.5 may be problematic and associated with many false positives. Importantly, an acceptable Z′-factor is a necessary but not sufficient condition for assay quality. Systematic confounders, including compound autofluorescence, aggregation, nonspecific cytotoxicity, interference with reporter systems, or limited readout specificity, can compromise biological interpretability despite favorable statistical metrics. Accordingly, orthogonal secondary assays and counter-screens are often needed to distinguish true target engagement from assay artifacts.

Screening libraries used in contemporary discovery programs typically comprise 50,000 or more chemically diverse, drug-like small molecules selected to maximize coverage of biologically relevant chemical space while satisfying physicochemical constraints consistent with oral bioavailability and synthetic tractability. Many commercial sources are available for purchase of chemically diverse compound collections. Smaller sets of compounds for screening may be appropriate if compound selection can be biased for the target of interest, for example using high-resolution structure data, computational chemistry (virtual screening), pharmacophore modeling of known target modulators and, increasingly, machine learning–assisted prioritization of chemical matter (10–12). These strategies can substantially increase hit rates and reduce screening burden, particularly for targets with defined binding pockets.

An alternative strategy leverages focused collections of approved drugs, investigational compounds, or bioactive molecules, enabling repurposing efforts that may accelerate clinical translation by capitalizing on established safety and pharmacokinetic data (13). This approach is particularly attractive in diseases with urgent unmet need or well-defined mechanistic hypotheses. Natural products remain a historically important source of pharmacophores, reflecting evolutionary selection for protein interaction. However, their structural complexity and challenges in chemical optimization can complicate development pipelines (14). The one example in the epithelial transport space is Crofelemer, a natural product from the Amazon tree Croton lechleri that was approved in 2012 for treatment of HIV-related diarrhea (15). Crofelemer is reported to have multiple antidiarrheal actions including weak inhibition of intestinal chloride channels (16). The approved SGLT2 inhibitors, though not natural products, are chemically similar to the natural product phlorizin, a dihydrochalcone flavonoid that inhibits glucose transport.

Following primary screening, putative active compounds, or ‘hits’, undergo confirmation in secondary, orthogonal assays designed to validate target engagement and exclude assay artifacts. For example, candidate ion channel modulators identified in fluorescence-based screens are typically verified using direct electrophysiological measurements, such as patch-clamp, to establish effects on channel conductance and gating. Early structure-activity relationship (SAR) studies, often enabled by commercially available analogs of hits, permit rapid prioritization of chemotypes. Given the availability of millions of synthetically accessible small molecules, iterative SAR exploration can efficiently define preliminary pharmacophores and eliminate liabilities at an early stage.

Promising scaffolds are then characterized for mechanism of action and efficacy in cell and animal models, as well as pharmacological properties including pharmacokinetics, oral bioavailability and toxicity. Medicinal chemistry optimization is then generally required to improve on compound properties such as target affinity, aqueous solubility or metabolic stability, with the goal of balancing potency and pharmacological properties. Lead candidates emerging from early studies may be useful as research tools, such as selective inhibitors to study target biology in cell and animal models. Pharmacological studies with a selective inhibitor can provide useful information to complement studies in transgenic knockout animal models, which can be confounded by chronic compensatory effects and altered regulation of off-target genes. If a lead molecule is a potential development candidate, then additional steps are taken, including protection of intellectual property by patent filings, expanded collection of efficacy and pharmacology data, and testing in disease-relevant animal models. A robust data set can greatly derisk a program when seeking funding for preclinical development, with the goal of filing an investigation new drug (IND) application to initiate human clinical trials.

Although numbers vary widely depending on therapeutic area and target biology, the typical cost to advance a single development candidate to IND filing is 5–15 million US dollars, much of which is spent on lead optimization and animal studies (17,18). The typical duration from initial target validation to IND filing is 3–6 years, with the most time-consuming step being hit identification to selection of a lead candidate. Another consideration is the success rate in moving from a lead candidate to IND filing and ultimate FDA approval. More than half of molecules at the lead candidate stage fail to be advanced because of issues with pharmacological properties, and of those that do advance to preclinical testing only ~30% advance to IND filing. Of the development candidates that enter clinical trials, less than 10% receive FDA approval, most often because of unanticipated toxicity or lack of efficacy in meeting pre-determined clinical endpoints. Taking into account the cost of failure, the estimated cost to produce one new approved drug is estimated to be at least 500 million US dollars, with some estimates considerably higher.

Where does academia fit in the small molecule / drug discovery paradigm? There are now an estimated 100–200 drug discovery centers at academic institutions worldwide, many with resources to develop and execute small molecule screens, characterize compounds in cell and small animal models, and optimize lead candidates by medicinal chemistry. At the practical level, screening capacity in academia spans a broad spectrum. Focused campaigns, such as repurposing screens of approved or investigational drug libraries, can be performed with relatively modest infrastructure and, in some cases, semi-manual workflows. Medium-throughput efforts involving tens of thousands of compounds typically require automated liquid-handling platforms and multi-mode plate readers, but do not necessarily demand the industrial scale of compound and assay management systems. The limiting factor in academic discovery programs is rarely compound availability or instrumentation alone. Successful translation requires multidisciplinary expertise, including assay development, pharmacology, small-animal testing, medicinal chemistry, and, importantly, biological insight into the target under investigation. Academia is uniquely positioned to contribute at the earliest stages of discovery, where mechanistic understanding and physiological context are paramount.

Advancement of a development candidate requires substantial financial investment, which may derive from government mechanisms, disease-focused private foundations, or, more commonly at later stages, commercial sources such as venture capital financing to launch a biotechnology company or out-licensing to an established pharmaceutical partner. Table 1 lists the major considerations in evaluating the commercial viability of a small molecule lead candidate. Return-on-investment and low risk-adjusted reward are primary concerns of potential investors. Though not absolute requirements, a lead candidate should have activity against a target with low nanomolar potency, excellent selectivity and drug-like properties. Its chemical structure should be novel and hence patentable, readily synthesized by a scalable route, water soluble and chemically stable. Favorable pharmacological properties are very important, including slow pharmacokinetics, oral bioavailability, accumulation in target tissues, efficient elimination, and slow metabolism to non-toxic, rapidly eliminated metabolites. Of central important is demonstration of efficacy in experimental animal models, with minimal toxicity. Evaluation of preclinical toxicity generally also includes in vitro screens such as Eurofins panels, genotoxicity, hERG cardiac channel testing and CYP inhibition. Beyond these scientific criteria, potential funders also scrutinize the strength and durability of intellectual property, competitive landscape and unmet clinical need, regulatory strategy, clinical trial design, and market size.

Table 1.

Considerations for commercial development of drug candidate

Compound properties

In vitro potency, target selectivity, drug-like properties

Chemistry

Novel chemical matter, synthesis, solubility, chemical stability

Pharmacology

Pharmacokinetics, oral bioavailability, metabolism, elimination, formulation

Small animal studies

Efficacy, on- and off-target effects

Toxicity

Toxicity screens, genotoxicity, hERG testing, CYP inhibition

Intellectual property

Patent protection - composition of matter, method of use, formulation

Clinical trials design

Size, duration, cost, approvable endpoint, biomarkers

Market analysis

Patient population, drug pricing, competitive landscape, payer acceptance

In the sections that follow, we describe several small molecule discovery programs pioneered by our laboratory that exemplify the major steps in the discovery process including assay development, compound discovery, validation, prioritization and optimization. Medicinal chemistry and small animal testing are the crucial, albeit generally the most laborious, steps in a project. To preserve focus on the discovery paradigm and its interface with epithelial physiology, we do not attempt comprehensive treatment of target genetics, regulatory mechanisms, high-resolution structural biology, or detailed drug–target binding mechanisms. Nor do we provide an exhaustive account of synthetic chemistry strategies. Rather, the intent is to highlight the translational logic linking physiological insight to chemical probe development and, ultimately, therapeutic advancement. Notably, while mechanistic understanding at the molecular level is desirable and can accelerate rational optimization, regulatory approval of safe and effective therapeutics has, in some instances, preceded elucidation of target identity or inhibition mechanism.

Assay development for chloride channels and solute carrier transporters

Initially motivated by the defect in epithelial chloride transport in cystic fibrosis (CF) and later by the discovery of CFTR as the chloride channel that is mutated in CF, we developed a series of intracellular chloride sensors for physiology studies and drug screening (19). Chloride plays a major role in cellular physiology in epithelial fluid absorption and secretion, cell volume and pH regulation, muscle contraction and neural signal transduction. In addition to CFTR, other classes of chloride channels include calcium-activated chloride channels (CaCCs), voltage-gated chloride channels (ClCs) and ligand (GABA, glycine) gated chloride channels. There are >2000 distinct CFTR mutations that cause CF, and mutations in other chloride channels cause various inherited diseases including Startle disease, myotonia, renal salt wasting, Dent’s disease, Bartter’s syndrome, osteopetrosis and epilepsy (20).

Functional assays of chloride transport generally involve measurement of kinetic responses to imposed anion gradients or electrical potential differences. Older radioisotope assays utilized ³⁶Cl⁻ or ¹²⁵I⁻. Electrophysiological methods such as short-circuit current and patch-clamp are useful for definitive characterization of ion channel modulators, though are generally not required for chloride channel drug discovery because most anion channels exhibit slow gating kinetics and relative voltage insensitivity. Intracellular chloride sensors are especially useful for automated drug screening because of their technical simplicity and rapid, large and selective signal responses. Indirect assays can also be used for chloride transport screens such as membrane potential-sensing dyes, cell volume sensors, and, for chloride-bicarbonate or proton coupled transporters, pH sensors.

We developed a series of chemical sensors of chloride and iodide, a halide that can often be substituted for chloride in screening applications (Fig. 2A). The first chloride indicators, the quinoliniums SPQ and MQAE (21,22), are blue-fluorescent dyes whose fluorescence is reversibly and instantaneously quenched by halides by a collisional mechanism (23). Variants of these dyes were synthesized for specialized applications, including a cell permeable dye, diH-MEQ, which is chemically modified in cytoplasm to a membrane-impermeable and fluorescent chloride sensor (24), and a dual-wavelength sensor, XPQ, which consists of a chloride-sensitive and insensitive moieties (25). Quninolinium-based dyes have been widely used in physiology studies; however, they are of limited utility for drug screening because of their dim blue fluorescence, rapid photobleaching, leakage out of cells and imperfect chloride selectivity, as quinolinium fluorescence is also quenched by cytoplasmic anionic macromolecules whose concentration is sensitive to cell volume and pH. A different chemical class of sensors, exemplified by the compound LZQ, are brightly green-fluorescent, highly sensitive and selective for iodide (though insensitive to chloride), and slowly leak out of cells (Fig. 2B) (26). Fig. 2C shows measurements of CFTR activity in cell cultures, using SPQ and LZQ sensors, in which nitrate replacement by iodide increases indicator fluorescence following CFTR activation by a cAMP agonist. In addition to the chemical sensors, a series of sensor-macromolecule conjugates were synthesized, based on the acridinium chromophore, for measurement of chloride concentration in intracellular organelles and extracellular spaces (27).

Figure 2.

Chemical halide sensors. A. Chemical structures of halide sensors. SPQ and MQAE are blue-fluorescent sensors with low-moderate cell permeability. diH-MEQ is a cell-permeable compound that is oxidized in the cytosol to a fluorescent halide sensor. LZQ is a green-fluorescent sensor that is sensitive to iodide but not chloride. Bis-XPQ is a ratioable, dual-wavelength halide sensor. B. Fluorescence excitation and emission spectra of LZQ, showing fluorescence quenching by iodide. C. Time course of SPQ and LZQ fluorescence measured in FRT cells stably expressing human wildtype CFTR and subject to an iodide-nitrate exchange protocol. Cells were loaded with indicator and initially bathed in an iodide-containing solution, followed by replacement of iodide by nitrate, then addition of cAMP agonists followed by thiocyanate. LZQ fluorescence is quenched by iodide and thiocyanate but not by nitrate. CFTR activation by cAMP agonists facilitates iodide efflux, which increases indicator fluorescence. Figure adapted from ref. 26, used with permission.

For screening applications, the more recently developed genetically encoded sensors, based on the green fluorescent protein (GFP), are generally superior to chemical sensors because they do not require cell loading and washing, and remain entrapped in specified cellular compartments. We reported the first applications of the original GFP to study pH-dependent processes in cells (28) and of yellow fluorescent protein (YFP) mutants to measure chloride transport (29). GFP and YFP are brightly fluorescent, relatively resistant to photobleaching and can be stably transduced into cells. YFP-H148H fluorescence is sensitive to chloride by a mechanism involving a chloride-induced shift in pKa such that at constant pH YFP-H148H fluorescence decreases with increasing chloride concentration (Fig. 3A). YFP-H148Q mutagenesis produced YFP variants with different anion sensitivities (30); the YFP-H148Q/I152L mutant has been particularly useful because of its high sensitivity to iodide vs. chloride (Fig. 3B). Fig. 3C shows a CFTR assay in YFP-H148Q/I152L transfected cells in which CFTR activation by forskolin increases the rate of fluorescence drop following chloride-iodide exchange. Several newer YFP variants have been reported with different halide sensitivities and optical properties, including ratioable sensors (31,32). It should be noted that the intrinsic pH sensitivity of YFP fluorescence can be a concern in assays in which cytoplasmic pH may change.

Figure 3.

Genetically encoded yellow fluorescent protein (YFP) halide sensor. A. Fluorescence of YFP-H148Q as a function of pH in solutions containing indicated concentrations of chloride.

B. Fluorescence of YFP-H148Q/I152L in solutions containing increasing concentrations of chloride, nitrate or iodide at pH 7.4. Inset. Stopped-flow study showing rapid kinetics of YFP fluorescence change in response chloride addition. C. (left) FRT cells expressing human wildtype CFTR and YFP-H148Q/I152L, showing cytoplasmic staining. (right) Cell fluorescence in response to replacement of chloride by iodide followed by addition of cAMP agonist forskolin and then replacement of iodide by chloride. Figure adapted from refs. 29 and 30, used with permission.

A generic anion screening approach is diagrammed in Fig. 4. Cells expressing an anion transporter of interest, together with cytoplasmic YFP, are cultured on a multi-well plate. Targets may include anion channels, exchangers or cotransporters, as diagrammed. Following incubation of cells with a drug candidate, and if required a transport activator, the assay is initiated by extracellular addition of halide in which halide influx is followed from the kinetics of YFP fluorescence. Transport activators or inhibitors are identified by altered kinetics of fluorescence following halide addition.

Figure 4.

Generic protocol to assay anion transporter activity using cytoplasmic halide-sensing fluorescent sensors. A. Assay of halide channel, exchanger or cotransporter in which halide addition, under appropriate conditions, alters sensor fluorescence. B. Screening protocol to measure transporter activity showing incubation with test compound followed by halide addition. An active transport modulator alters the kinetics of cell fluorescence change.

CFTR as a target to modulate epithelial fluid secretion

CFTR is a cAMP-regulated anion channel that transports anions, including chloride and bicarbonate, in response to an electrochemical driving force. CFTR is broadly expressed in secretory or absorptive epithelial cells in the airways, gastrointestinal tract, exocrine glands and reproductive organs, as well as in neurons, heart and immune cells (33). The most common CF-causing CFTR mutation is deletion of phenylalanine 508 (ΔF508, F508del), which is present as at least one allele in ~90% of CF subjects. The ΔF508 mutation causes defective CFTR folding, trafficking to the cell surface, and ion conductance (34,35). Some CFTR mutations, such as G551D, impair channel gating without effect on surface membrane targeting. CFTR is an attractive, druggable target with many potential clinical indications.

Inhibitors and activators of native, wildtype CFTR have potential clinical indications that are not related to cystic fibrosis. In the intestine, CFTR is expressed at the luminal membrane of epithelial cells, enterocytes (Fig. 5) (36). CFTR activation drives intestinal fluid secretion in enterotoxin-mediated bacterial secretory diarrheas, including cholera (caused by Vibrio cholerae) and Travelers diarrhea (caused by enterotoxigenic E. coli), and in some hereditary, bile acid-induced, drug-induced and tumor-related diarrheas (37,38). CFTR inhibition is predicted to reduce intestinal fluid losses by prevention of chloride and consequent fluid movement from blood into the intestinal lumen. Another potential indication for CFTR inhibitors is in autosomal polycystic kidney disease in which CFTR promotes fluid secretion into expanding cysts (39–41). Additional proposed indications, albeit less well-validated by available evidence, include cardiogenic pulmonary edema (42), rhinorrhea (43), secretory otitis media (44) and cancer-related ascites (45). As research tools, CFTR-selective inhibitors have been used extensively to demonstrate CFTR function and to pharmacologically create the CF phenotype in cell and animal models.

Figure 5.

Fluid absorption and secretion mechanisms in intestinal epithelial cells. Energy to drive transport is provided by a basolateral membrane sodium-potassium ATPase pump. Primary absorptive mechanisms include apical membrane sodium-glucose cotransport, chloride-bicarbonate exchange and sodium channels, whose contributions vary in different regions of intestine. Primary secretory mechanisms include apical membrane chloride channels and basolateral membrane potassium channels and sodium-potassium-chloride cotransporter. Figure adapted from ref. 38, used with permission.

CFTR inhibitors for secretory diarrheas

Many chloride channels transporters, including CFTR, are inhibited at high concentrations of non-selective blockers such as niflumic acid and disulfonic stilbenes. The first selective CFTR inhibitor, the thiazolidinone CFTR_inh-172, was identified by a screen of chemically diverse, synthetic small molecules in FRT cells cultured on 96-well plates that stably express an iodide-sensitive YFP together with human wildtype CFTR (Fig. 6A). FRT cells were selected because of their low intrinsic anion permeability, formation of a tight epithelium, rapid growth on plastic, and efficient transduction to stably express target genes. For screening, CFTR was activated by an agonist mix acting by different mechanisms, reasoning that any identified inhibitor candidate would likely target CFTR directly. Following CFTR activation, cells are incubated with test compounds at 25 μM for 10 min or more to allow for cellular uptake, followed by addition of iodide to the external solution. Iodide entry reduces YFP fluorescence over the first ~10 seconds, with CFTR inhibitors slowing the rate of fluorescence decrease (Fig. 6B). The original screening of 50,000 compounds produced only one active inhibitor, CFTR_inh-172, with IC₅₀ ~ 200–300 nM (46). CFTR_inh-172 has been used widely as a research tool. Follow-on synthetic chemistry yielded chemically related thiazolidinone inhibitors with improved inhibition potency and pharmacological properties including greater aqueous solubility (47). Mutation of arginine-347 in CFTR abolished CFTR_inh-172 inhibition with minimal effect on CFTR channel function (48), suggesting a potential location for CFTR_inh-172 binding. A more recent structural study of the CFTR-inhibitor complex by electron crystallography indicated its precise binding site in the CFTR pore, and consequent allosteric conformational changes in extracellular segments (49). CFTR_inh-172 or its analogs have demonstrated efficacy in experimental mouse models of cholera (50) and autosomal polycystic kidney disease (51).

Figure 6.

Discovery of thiazolidinone CFTR inhibitor CFTR_inh-172. A. Screening assay. FRT cells stably expressing human wildtype CFTR and YFP halide sensor were preincubated with an activator mix to fully activate CFTR. Following incubation with test compounds, iodide was added to cells in one well of a 96-well plate using an automated syringe pump during continuous measurement of cell fluorescence. B. Fluorescence curves from single wells showing responses to iodide in the absence of CFTR activators and in the presence of CFTR activators with inactive and active CFTR inhibitor candidates. C. (top) Chemical structure of CFTR_inh-172. (center and bottom). Short-circuit current in FRT cells expressing wildtype CFTR in response to CFTR activation by CPT-cAMP and additions of CFTR_inh-172. Figure adapted from ref. 46, used with permission.

A subsequent screen for CFTR inhibitors, which was biased to identify rapidly acting compounds with an extracellular site of action, identified the glycine hydrazide GlyH-101 (52). Whole-cell patch-clamp revealed voltage-dependent GlyH-101 inhibition, changing the CFTR current-voltage relationship from linear to strongly inwardly rectifying (Fig. 7A). Together with single-channel recordings showing flicker behavior with fast channel closures (Fig. 7B), the data supported a pore-blocking mechanism in which GlyH-101 binds within the CFTR pore near its extracellular entrance. An external pore blocking mechanism was proven by the inhibitory efficacy of a membrane-impermeable macromolecular conjugate containing a glycine hydrazide analog (MalH, malonic acid hydrazide) linked to polyethylene glycol (Fig. 7C) (53,54). Though structure data has not been reported for GlyH-101, mutagenesis and computational studies supported an external pore-occluding mechanism (55). GlyH-101 has been used in many studies as a research tool, notwithstanding its relatively low potency of 2–8 μM at physiological interior-negative membrane potentials and its imperfect CFTR selectivity.

Figure 7.

Glycine hydrazide CFTR inhibitor with an extracellular, pore-blocking inhibition mechanism. A. Whole-cell patch-clamp in FRT cells expressing wildtype CFTR showing a linear current-voltage relationship which becomes outwardly rectifying after inhibition by GlyH-101.

B. Single-channel patch-clamp recordings showing CFTR inhibition with channel flicker following GlyH-101 addition. C. Short-circuit current showing CFTR inhibition by a cell-impermeant conjugate consisting of hydrazide CFTR inhibitor linked to polyethylene glycol. Figure adapted from refs. 52 and 53, used with permission.

The unique, extracellular site of action of GlyH-101 presents the possibility of diarrhea treatment by an orally delivered drug with minimal systemic exposure, which might be accomplished using a non-absorbable drug or an absorbed, rapidly metabolized drug. However, a major limitation of an inhibitor that acts from the extracellular surface is that in secretory diarrhea a small molecule with micromolar potency, and hence binding off-times of milliseconds to seconds, is predicted to be convectively washed away from its CFTR binding site in intestinal crypts. Fig. 8A diagrams a mathematical model of the human intestine in which predicted inhibitor antidiarrheal efficacy was computed at different fluid secretion rates (56). The model predicted that at secretion rates typical of cholera, lumen drug concentration would need to exceed its binding affinity by 1000-fold, which is not feasible. Nevertheless, a chemical analog of GlyH-101, IOWH-032 (57), was tested for efficacy in human cholera. IOWH-032 was found to be safe but ineffective in reducing diarrheal fluid loss (58), probably because of its low inhibition potency of ~10 μM at physiological membrane potential and its predicted convective washout.

Figure 8.

Model predictions of the influence of fluid convection on the efficacy of an externally acting CFTR inhibitor in treating secretory diarrhea. A. (left) Fluid convection out of intestinal crypts produced by epithelial cell secretion, with drug entry into crypts by diffusion from the intestinal lumen. (right, top) Intestine modeled as a cylinder containing crypts using anatomically derived crypt dimensions and density. (right bottom) Example of computed result for inhibitor concentration, relative to its binding affinity (C_o/K_d), along the longitudinal axis of the intestine. B. Percentage inhibition of CFTR-mediated fluid secretion as a function of C_o/K_d at different lumen flow rates. The highest flow rate of 7 × 10⁻² μL/cm²/s, which is typical in cholera, predicts the requirement of very high luminal inhibitor concentrations to reduce CFTR activity and hence intestinal fluid secretion. Figure adapted from ref. 56, used with permission.

To address the washout issue, a non-absorbable conjugate containing MalH linked to a lectin was synthesized, reasoning that tetravalent binding of the lectin moiety to the dense extracellular glycocalyx on enterocytes would improve inhibitor binding affinity and prevent convective washout. A MalH-lectin conjugate showed strong CFTR inhibitory potency with IC₅₀ ~50 nM (Fig. 9A) and was efficacious in reducing fluid secretion and improving survival in animal models of cholera including a closed-intestinal loop model (Fig. 9B) and a suckling mouse model produced by oral administration of cholera toxin (Fig. 9C) (59). A fluorescently labeled MalH-lectin remained bound in the intestinal lumen for >6 hours after washout, whereas washout of unconjugated MalH occurred within minutes. Though MalH-lectin conjugates are potential drug development candidates, such protein-based biologics are probably not suitable for cholera therapy as they are expensive to manufacture and have poor stability in a hot, humid environment.

Figure 9.

Antidiarrheal efficacy of a lectin-conjugated CFTR inhibitor that is entrapped by the surface glycocalyx of intestinal epithelial cells. A. Short-circuit current in FRT cells expressing wildtype CFTR in response to addition of a MalH CFTR inhibitor linked to the lectin concanavalin A. B. Closed-loop model of intestinal fluid secretion in mice in which loop fluid accumulation, quantified by loop weight-to-length ratio, was measured after injection of loops with saline (PBS) without or with cholera toxin and indicated amounts of the inhibitor-lectin conjugate. C. Survival of neonatal mice administered vehicle or cholera toxin orally. Some cholera toxin-treated mice were orally administered 125 pmol of the CFTR inhibitor-lectin conjugate. Figure adapted from ref. 59, used with permission.

To identify alternative CFTR inhibitor scaffolds as potential drug development candidates, additional large-scale screening of diverse compounds collections not already tested yielded the pyrimido-pyrrolo-quinoxalinedione PPQ-102, which inhibited CFTR with IC₅₀ ~ 90 nM and showed efficacy in a mouse model of autosomal dominant polycystic kidney disease (60). A synthetic chemistry program, involving systematic modifications in different regions of the PPQ-102 molecule, yielded the benzopyrimido-pyrrolo-oxazinedione (R)-BPO-27 which had IC₅₀ ~ 4 nM for CFTR inhibition in a cell-based assay (Fig. 10A) and <1 nM in cell-detached patches (61,62). Patch-clamp indicated an inhibition mechanism involving stabilization of the channel closed state (63). A recent high-resolution structure study revealed that (R)-BPO-27 sterically blocks CFTR channel function by binding to the inner vestibule of the CFTR pore (64).

Figure 10.

High-potency (R)-BPO-27 CFTR inhibitor. A. (left) Inhibitor structure showing a single chiral carbon. (right) Short-circuit current in FRT cells expressing CFTR showing inhibition by (R)-BPO-27. B. (left) Intestinal closed-loop model in mice showing fluid accumulation in loops after injection of cholera toxin and indicated amounts of (R)-BPO-27. ** P < 0.01. (right) Photos of intestinal loops after removal from mice. Figure adapted from refs. 62 and 65, used with permission.

(R)-BPO-27 was effective in mouse models of enterotoxin-mediated secretory diarrhea, including an intestinal closed loop model (Fig. 10B) (65), and models of tyrosine kinase inhibitor diarrhea (66) and bile acid diarrhea (67). (R)-BPO-27 also inhibited CFTR in primary human enteroid cultures with low nM IC₅₀ (65) and was effective in reducing intestinal fluid secretion and mortality in mice administered live Vibrio cholerae (68). A BPO-class CFTR inhibitor was out-licensed to Vanda Pharmaceuticals and is currently in human clinical trials for treatment of cholera.

CFTR potentiators and correctors for cystic fibrosis

We originally introduced the concept of small molecule modulators of CF-causing mutant CFTRs, and reported the first potentiators (69) and correctors (70) of ΔF508-CFTR identified from small molecule screens. Potentiators treat defective mutant CFTR channel gating whereas correctors treat mutant CFTR misfolding and intracellular retention. The screens used FRT cells expressing human ΔF508-CFTR and a YFP sensor (Fig. 11A). For potentiator screening cells were initially incubated at reduced temperature for 24 hours to promote CFTR movement to the plasma membrane, as ΔF508-CFTR is a temperature-sensitive mutant. Cells were incubated with test compounds at 2.5 μM together the cAMP agonist forskolin. A low concentration of 2.5 μM was chosen for the screen based on pilot studies showing many activate potentiators. Active potentiators increase iodide influx producing a rapid drop in fluorescence (Fig. 11B). The initial screen identified several classes of potentiators, including a tetrahydrobenzothiophene with EC₅₀ of 600 nM (69). Modification of this scaffold by Galapagos gave a chemically related potentiator GLPG1837 that entered clinical testing (71). A subsequent FRT cell screen identified phenylglycine and sulfonamide potentiators with EC₅₀ down to 40 nM (72). We also introduced the concept of potentiator combinations acting in synergy to improve on the efficacy of single potentiators for certain CF-causing CFTR mutations. Several combination potentiators, called ‘co-potentiators’, were identified in a synergy screen (73). Independent cell-based screening by Vertex Pharmaceuticals using a membrane potential-sensitive dye identified early versions of the-now approved potentiator VX-770 (ivacaftor) (74).

Figure 11.

Original identification of small molecule modulators (potentiators and correctors) of Fdel508 (ΔF508) CFTR. A. Screening protocols. YFP fluorescence measured in cells stably expressing human ΔF508-CFTR and YFP halide sensor. (top) The assay to identify potentiators (compounds that correct mutant CFTR gating) involved 24-h incubation of cells at reduced temperature to promote ΔF508-CFTR trafficking to the plasma membrane, followed by incubation with test compound and a cAMP agonist, and then iodide addition. (bottom) The assay to identify correctors (compounds that correct mutant CFTR trafficking to the plasma membrane) involved 24-h incubation with test compounds, followed by addition of potentiator and cAMP agonist, and then iodide. B. YFP fluorescence curves from single wells of a 96-well plate for the potentiator assay, showing curve for inactive (or no) compounds, the reference potentiator genistein at high concentration, and active compounds from screening. Structure of potentiator ΔF508_act-02. Figure adapted from ref. 69, used with permission.

Correctors of certain CF-causing mutant CFTRs, including ΔF508-CFTR, improve their folding and stability, promoting transport to the plasma membrane. For corrector screening, FRT cells expressing ΔF508-CFTR and YFP were incubated with test compounds for 24 h, following by addition of a potentiator and cAMP agonist prior to iodide addition (Fig. 11A). Several correctors were identified in the initial screen including the compound corr-4a, which was shown to restore chloride conductance in primary cultures of human airway epithelial cells from ΔF508 CF subjects (70). We also introduced the concept of corrector combinations acting in synergy by different mechanisms to increase on the efficacy of single correctors and identified effective compounds using a synergy screen (75). Correctors and potentiators identified by our laboratory were out-licensed to Biomarin Pharmaceutics for development but did not reach the clinical trials stage.

Compound screening and optimization by Vertex Pharmaceuticals yielded the now-approved correctors VX-809, VX-661 and VX-445 (76,77). The Vertex strategy was to first obtain approval for the potentiator VX-770 for the CFTR G551D gating mutation and demonstrate clinical efficacy (78). Approvals then followed for two-drug corrector-potentiator combinations for the CFTR ΔF508 mutation, which showed limited though approvable efficacy (79), and most recently for triple combinations of two correctors and one potentiator with high efficacy for the ΔF508 and various other CF-causing CFTR mutations (80.81). For most CF patients the newer Vertex drugs have largely transformed a fatal disease into a manageable, chronic disease (82,83).

Activators of wildtype CFTR for epithelial hyposecretion

Activators of wildtype CFTR, which increase fluid secretion in many epithelial tissues, have a number of potential clinical indications, including non-CF inflammatory lung disorders, constipation, dry eye disease and potentially cholestatic liver disease and chronic pancreatitis. Potentiators of mutant CF-causing CFTRs are generally relatively weak activators of wildtype CFTR.

In some inflammatory lung disorders associated with acquired CFTR dysfunction, such as smoking-related chronic obstructive pulmonary disease (COPD) and bronchiectasis, CFTR activation has been proposed to improve lung function by increasing airway surface liquid hydration. A potentiator of CF-causing mutant CFTRs, icentacaftor, which also activates wildtype CFTR, was found to be safe when administered to COPD subjects and showed possible benefit, though the prescribed primary clinical endpoint was not met (84).

Early on, we carried out a small molecule screen to identify activators of wildtype CFTR using FRT cells expressing CFTR and a YFP halide sensor (85). Several novel activators emerged from activator screens, most recently including phenylquinoxalinones and triazines (86). The massive secretory diarrhea in cholera provided a rationale for use of CFTR activators in constipation, recognizing potential on-target side effects including diarrhea and fluid hypersecretion outside of the intestine. Following structure-activity studies, the efficacy of a phenylquinoxalinone activator was demonstrated in rodent models of constipation (86).

Dry eye disease is a potentially more tractable indication for CFTR activators because compounds could be applied topically, by eyedrop, with minimal systemic exposure. CFTR is expressed in corneal and conjunctival epithelia where it drives fluid secretion by physiological mechanisms similar to those in the intestine (87,88). A triazine CFTR activator, which was identified by screening following a medicinal chemistry program (89), was effective in increasing tear fluid volume and reducing inflammation in experimental models of dry eye (90,91) and is currently in human clinical trials.

The theoretical question arises whether the ocular surface epithelia, without changing lacrimal gland secretion, have the intrinsic capacity, by prosecretory and/or anti-absorptive mechanisms, to substantially increase tear fluid volume in dry eye disease. To address this, we developed a mathematical model of ocular surface fluid balance, which accounted for lacrimal gland secretion, evaporation, nasolacrimal drainage and epithelial transport (Fig. 12A) (92). Model predictions, which included tear fluid volume, were made for various changes in epithelial transport parameters simulating drug effects. Fig. 12B shows predictions for ENaC inhibition, CFTR activation and basolateral K⁺ channel activation, alone and in combinations, on the percentage normalization of tear fluid volume under dry eye conditions, with 100% normalization representing healthy, non-dry eyes. Significant, albeit modest effects were predicted, with combination therapy producing the best correction. An interesting prediction of the modeling was quite substantial normalization of tear fluid volume or osmolality by increasing cell water permeability or paracellular ion conductance. Though the strength of physiology models such as this one relies on model assumptions and parameter selection, such computations can support the potential efficacy of therapeutic interventions and suggest novel approaches, which are testable in animal models and ultimately in human clinical trials.

Figure 12.

Model predicting efficacy of drugs targeting ocular surface epithelial transporters in dry eye disease. A. (left) Diagram showing ocular surface epithelium in contact with a tear fluid layer. Water and solutes enter the tear fluid layer by lacrimal gland secretion and epithelial secretion, and exit the tear fluid layer by nasolacrimal drainage, evaporation (water only) and epithelial absorption. Model parameters include fluxes of water and individual ions into and out of the tear fluid and cytoplasmic compartments. (right) Higher resolution of ocular surface epithelium showing major transport routes. B. Predicted efficacy of modulating the activity of selected ocular surface transport processes on tear film volume Computations done under dry eye conditions produced by reduced lacrimal secretion and increased evaporation. The abscissa is the percentage normalization of tear film volume compared to healthy, non-dry eye, with 100% corresponding to tear film volume in non-dry eye. Changes in transporter activities, compared to baseline (without drugs) are 100% reduction for ENaC, 20-fold increase for CFTR, and 5-fold increase for basolateral K⁺ channel, with changes applied individually and in combinations. Symbols: [X]_c and [X]_tear, concentration of solute X in cytoplasm and tear fluid; J_v, volume (water) flux; J_s, solute flux; ψ_ap and ψ_bl, apical and basolateral membrane potentials.

Small molecule modulators of calcium-activated chloride channels

Decades of research indicate broad functional expression of plasma membrane calcium-activated chloride channels (CaCCs) in the airway, kidney, gastrointestinal and glandular epithelia, smooth muscle, sensory neurons, photoreceptors, immune cells, and others (93). Prior to the identification of TMEM16A (ANO1) as the protein responsible for CaCC activity in many cell types, we carried out a phenotype-based small molecule screen, seeking to identify compounds that blocked ATP and carbachol-stimulated iodide influx in human intestinal HT-29 cells using halide sensor YFP-H148Q/I152L introduced by lentiviral transduction (94). Screening produced several classes of inhibitors that fully blocked iodide influx in HT-29 and T-84 cells in response to multiple calcium agonists, including thapsigargin, without altering intracellular calcium concentration. The 3-acyl-2-aminothiophene carboxylic acid CaCC_inh-A01, which has IC₅₀ ~1 μM for inhibition of CaCC currents, has been used widely in physiology research though its action on chloride channels is somewhat non-selective.

Following its identification in 2008, we became interested in identifying modulators of the CaCC TMEM16A (anoctamin 1, ANO1; originally called DOG1 for Discovered On Gastrointestinal tumors-1). TMEM16A is expressed in epithelial cells in the airways, exocrine pancreas, liver, and salivary gland, and in smooth muscle and sensory neurons (95). TMEM16A is also highly expressed in some cancers including gastrointestinal stromal tumors, head and neck squamous cell carcinoma, and breast and esophageal cancers. Based on its tissue expression, TMEM16A inhibition has been speculated, albeit without direct evidence, to be useful in the treatment of some cancers, hypertension, neuropathic pain, itch, overactive bladder, and viral secretory diarrheas. There are conflicting opinions about whether TMEM16A inhibition or activation would be beneficial in cystic fibrosis; whereas TMEM16A activation might offer an alternative chloride channel to substitute for defective CFTR it would also cause undesired mucus hypersecretion.

Though CaCC_inh-A01 inhibits TMEM16A function, we were interested in identifying more potent and selective inhibitors to pharmacologically investigate TMEM16A functions and as potential drug development candidates. We identified the first TMEM16A inhibitors, including the aminophenylthiazole T16_inh-A01, in a target-based screen of ~110,000 synthetic small molecules using FRT cells coexpressing human TMEM16A and the YFP iodide sensor (96). Other laboratories subsequently identified inhibitors with different chemical scaffolds and properties, including Ani9, MONNA and niclosamide (97). Further screening by our lab identified 2-acylaminocycloalkyl-thiophene-3-carboxylic acid arylamides (AACTs) as potent and selective inhibitors of TMEM16A (98). Targeted synthesis of 48 AACT analogs produced TM_inh-23 with IC₅₀ of ~30 nM (Fig. 13A), in which inhibition was shown by patch-clamp to be direct and reversible.

Figure 13.

Inhibitor of calcium-activated chloride channel TMEM16A (ANO1) reduces blood pressure in hypertensive rats. A. TM_inh-23 inhibition of short-circuit current in FRT cells expressing TMEM16A following TMEM16A activation by the calcium agonist ionomycin. B. Mean arterial blood pressure (MAP) in spontaneously hypertensive rats before and for 8 hours after administration of 10 mg/kg TM_inh-23 (mean ± S.E.M., 3–6 rats per group, * P < 0.05, ** P < 0.01). C. MAP in rats treated for 4 days with 10 mg/kg TM_inh-23 twice daily (mean ± S.E.M., 4 rats per group, * P < 0.05). Figure adapted from ref. 99, used with permission.

We investigated the actions of TM_inh-23 on two physiological processes – blood pressure, based on TMEM16A expression in vascular smooth muscle, and gastric motility, based on TMEM16A expression in interstitial cells of Cajal, the pacemaker cells that regulate gastrointestinal motility. TM_inh-23 fully inhibited calcium-activated TMEM16A chloride current with nanomolar potency in primary cultures of rat vascular smooth muscle cells and reduced vasoconstriction produced by the thromboxane mimetic U46619 (99). Blood pressure measurements by tail-cuff or telemetry showed that a single dose of TM_inh-23 in spontaneously hypertensive rats produced marked reductions in blood pressure lasting for 4–6 hours (Fig. 13B), with little effect in normotensive rats. Chronic TM_inh-23 treatment of spontaneously hypertensive rats for five days produced sustained blood pressure reductions (Fig. 13C). These studies support TMEM16A as a target for antihypertensive therapy and demonstrated the efficacy of a TMEM16A inhibitor with a novel antihypertensive mechanism of action.

To assess gastric motility in mice, gastric emptying was measured from the quantity of phenol red in stomach after oral gavage (Fig. 14A) and independently by scintigraphy following an oral bolus of [^99mTc]-diethylenetriamine pentaacetic acid (Fig. 14B) (100). Both methods showed markedly slowed gastric emptying in TM_inh-23-treated mice. Whole-gut transit was not affected by TM_inh-23 treatment. In ex vivo strips of mouse gastric antrum TM_inh-23 reversibly inhibited spontaneous and carbachol-stimulated rhythmic contractions (Fig. 14C). As a consequence of the delayed gastric emptying, TM_inh-23 reduced the blood sugar spike in mice following an oral but not intraperitoneal glucose load. These results supported the involvement of TMEM16A in gastric emptying and suggested the potential use of TMEM16A inhibition in disorders of accelerated gastric emptying, such as dumping syndrome, and potentially to improve glucose tolerance in diabetes and perhaps to produce satiety in obesity.

Figure 14.

Slowed gastric emptying in mice treated with TMEM16A (ANO1) inhibitor TM_inh-23. A. Percentage gastric emptying at 20 min following gavage of phenol red in adult mice treated with metaclopramade (10 mg/kg) and TM_inh-23 (10 mg/kg) as indicated (mean ± S.E.M., 5–6 mice per group, * P < 0.05, *** P < 0.001). B. Radionucleotide scans, with x-ray overlap, at indicated times after gavage of 99mTc-TDPA without or with 10 mg/kg TM_inh-23. C. Gastric smooth muscle contractions in ex vivo strips of gastric antrum showing effects of TM_inh-23 addition and washout. Figure adapted from ref. 100, used with permission.

Though the experimental animal studies are interesting, there are substantial challenges to advance TMEM16A inhibitors for clinical use, including the wide tissue distribution of TMEM16A and the existence of many related TMEM16-family proteins with broad cellular functions. In addition to inhibitors, we and others have identified small molecule TMEM16A activators acting by different mechanisms (101,102). Given the challenges to develop a TMEM16A modulator with systemic exposure, perhaps targeted tissue delivery may be possible, for example delivery of a metabolically unstable compound to the airways by nebulization or to vascular smooth muscle cells by an intravenous route. Short-term systemic use of TMEM16A modulators is another possibility, for example for treatment in acute hypertensive crisis.

Urea transport inhibitors as salt-sparing diuretics for refractory edema

Kidney urea transporters (UTs) represent a novel class of targets whose inhibition is predicted to produce a unique, salt-sparing diuretic action with urinary losses of urea > salt – an action that we called ‘urearetic’ (103). Based on the physiology of UTs, including studies from knockout mice, the urearesis produced by UT inhibitors is predicted to be effective in the treatment of hyponatremias associated with volume overload and edema, including congestive heart failure, cirrhosis and nephrotic syndrome, volume-sensitive hypertension, and syndrome of inappropriate secretion of antidiuretic hormone (SIADH) (104–106). Compared to conventional diuretics such as furosemide and thiazides, which target salt transporters in kidney tubules, UT inhibitors may minimize electrolyte abnormalities such as hypokalemia, and potentially be effective, used alone or in combination with conventional diuretics, in conditions in which salt transport-blocking diuretics alone are not.

Urea is the end-product of nitrogen metabolism, being generated mainly in the liver and excreted by the kidney. Urea concentration in urine is much higher than in the blood and urea itself has been recognized for centuries as an osmotic diuretic. UTs are expressed in selected epithelial cells in kidney tubules (UT-A isoforms, SLC14A2) and in descending vasa rector microvessels (UT-B, SLC14A1) (Fig. 15A) where they facilitate the passive, concentration gradient-driven transport of urea across epithelial and endothelial cell membranes. The formation of a concentrated urine by the kidney relies on countercurrent multiplication and exchange mechanisms in which salt and urea are transported and recycled to generate a hypertonic inner medulla. Reabsorption of urea in the inner medullary collecting, facilitated by UT-A1 and UT-A3, facilitates urea reabsorption during antidiuresis to maintain a hypertonic medullary interstitium. Mathematical models of the urinary concentrating mechanism have clarified the roles and mechanisms of UT function in urinary concentration (107), which are supported by UT knockout studies in mice (105), providing the rationale to develop UT inhibitors.

Figure 15.

Discovery of small molecule inhibitors of urea transporter UT-B. A. Distribution of UT-A and UT-B urea transporters in kidney showing UT-A isoforms in kidney tubule epithelial cells in selected regions of the nephron and UT-B in descending vasa recta blood vessels. B. Assay for discovery of UT-B inhibitors. Erythrocytes, which natively express UT-B and water channel AQP1, are preloaded with acetamide, a urea analog that is transported relatively slowly by UT-B. Upon dilution into PBS, cells initially swell because of water influx and then return to normal size as acetamide exits. With a UT-B inhibitor, cell swelling is unopposed by acetamide exit, resulting in cell lysis which is measured by hemoglobin absorbance. C. UT-B inhibition in erythrocytes by UTB_inh-14 as measured by a stopped-flow light scattering assay. D. ‘Urearetic’ action of UTB_inh-14 in mice. Mice were pretreated for 1 h with 300 μg UTB_inh-14 (or vehicle alone) prior to administration of antidiuretic vasopressin analog DDAVP. Serial urine osmolalities measured (mean ± S.E.M., 6 mice per group, * P < 0.01). Inset. Four-hour urine volume after DDAVP administration. Figure adapted from ref. 108, used with permission.

Small molecule inhibitors of urea transporter UT-B

Until recently, the available UT inhibitors included urea analogs with millimolar potencies and the non-selective membrane intercalating compound phloretin. Because UT-mediated urea transport across cell membranes is very rapid, and because there is no optical sensor of urea concentration with rapid response, assaying urea transport in a high-throughput screening format has been challenging.

We developed the first high-throughput screening assay for rapid identification of UT inhibitors, which produced compounds with inhibition potency a million-fold better than urea analog inhibitors (95). The assay utilized erythrocytes, which natively express UT-B, and a single time-point readout of erythrocyte lysis by near-infrared light absorbance. The assay involved pre-loading erythrocytes with acetamide, a urea analog that is transported by UT-B at a slower rate than urea itself, such that its equilibration time is similar to that of osmotic water transport (Fig. 15B). Upon dilution in an acetamide-free solution, the large, outwardly directed gradient of acetamide causes erythrocyte swelling with minimal lysis because of the protective effect of UT-B-facilitated acetamide efflux. With a UT-B inhibitor cell swelling is unopposed, resulting in cell lysis. The assay was robust and sensitive, with a statistical Z’-factor of ~0.6. Results from the acetamide lysis screen correlated well with urea transport inhibition measured by stopped-flow light scattering, the gold standard assay for erythrocyte urea transport.

An initial screen of 50,000 synthetic small molecules using human erythrocytes produced dozens of UT-B inhibitors belonging four chemical classes (103). Subsequent screening against rodent erythrocytes and testing of hundreds of chemical analogs of active compounds gave many UT-B inhibitors, one example being the triazolothienopyrimidine UTB_inh-14 which by stopped-flow light scattering had IC₅₀ ~10 nM for inhibition of human UT-B and 25 nM for mouse UT-B (Fig. 15C) (108). UTB_inh-14 was selective, producing little inhibition of UT-A even at high concentrations. Interestingly, other UT inhibitors, some with small chemical modifications of UT-B selective inhibitors, showed poor UT-A vs. UT-B selectivity. Mechanistic studies suggested a competitive inhibition mechanism for UTB_inh-14 inhibition of UT-B urea transport. Follow-on medicinal chemistry produced analogs of UTB_inh-14 with low nanomolar potency, good metabolic stability, and accumulation in mouse kidneys and urine upon systemic administration (109).

UTB_inh-14 efficacy was tested in mice following pharmacological studies to establish its dosage, route and timing to produce predicted therapeutic concentrations in kidney and urine. In one set of studies mice were given UTB_inh-14 intraperitoneally one hour prior to administration of the antidiuretic agent dDAVP (Fig. 15D). UTB_inh-14 produced a mild diuresis over several hours following DDAVP administration, reducing urine osmolality and increasing urine volume. Urine osmolalities in UTB_inh-14-treated wildtype mice were similar to those in UT-B knockout mice in which UTB_inh-14 had no significant effect. The modest diuretic action of selective UT-B inhibition is consistent with the prediction for knockout mouse experiments and modeling that tubule epithelial UT-A is the major determinant in urinary concentration. Therefore, though useful as research tools to study renal and extrarenal UT-B physiology, UT-B selective inhibitors are not attractive candidates for development of a urearetic for human use.

Small molecule inhibitors of urea transporter UT-A

High-throughput identification of UT-A urea transport posed a greater challenge than UT-B urea transport, as available cell lines do not natively express UT-A isoforms, UT-A-mediated cellular urea equilibration is rapid, and measurement of cell volume can be difficult. Our screening assay for identification of UT-A inhibitors involved measurement of cell volume changes in response to urea addition in MDCK cells stably expressing UT-A1, AQP1 and a YFP chloride sensor (Fig. 16A) (110). AQP1 confers high cellular water transport to make water, and hence volume, equilibration rapid. The YFP chloride sensor provides an instantaneous measure as cell volume since cytoplasmic chloride varies inversely with cell volume. Urea addition to a well in a 96-well format drives osmotic water efflux and cell shrinking, which is followed by urea (and water) entry with return toward the initial cell volume. UT-A inhibition increases the initial drop in fluorescence upon urea addition and slows the return of fluorescence to baseline. The assay was robust, and results agreed well with transepithelial measurements of UT-A-facilitated urea transport using filter-grown UT-A expressing MDCK cells in which urea concentration was measured enzymatically in serial fluid samples.

Figure 16.

Discovery of small molecule inhibitors of urea transporter UT-A. A. Assay for discovery of UT-A inhibitors. FRT cells were stably transfected with UT-A1, AQP1 and GFP. (left) AQP1 and UT-A1 immunofluorescence and YFP fluorescence. (right) Assay method showing extracellular addition of urea to the cell layer, which produces initial volume decrease following by return to original cell volume as urea and water enter cells. Changes in cell volume alter cytosolic chloride concentration and YFP fluorescence. A UT-A inhibitor increases the initial reduction in cell volume and slows the recovery phase. B. UTA_inh-E02 inhibition of UT-A1 and UT-B mediated urea transport. C. ‘Urearetic’ action of UTA_inh-E02 in rats. Rats were pretreated for 1 h with 5 mg UTA_inh-E02 (or vehicle alone) prior to administration of DDAVP. Urine volumes and osmolalities measured (mean ± S.E.M., 4 rats per group, * P < 0.05). Figure adapted from refs. 110 and 111, used with permission.

Screening of 100,000 synthetic small molecules produced several classes of inhibitors with low micromolar IC₅₀, one of which, UTA_inh-E02, is shown in Fig. 16B (111,112). UT-A1 inhibition by UTA_inh-E02 was reversible, non-competitive and showed ~100-fold greater selectivity for UT-A vs. UT-B. Other inhibitors emerging from the screen and follow-on testing of analogs were not selective, as was seen for some compounds from the UT-B screen.

UTA_inh-E02 was administered to rats by intraperitoneal injection under conditions that produced predicted therapeutic concentrations for a few hours (111). Urine was collected in 3-hour intervals before and after dDAVP administration for determination of volume, osmolality and urea and ion concentrations. Urine volumes and osmolalities were similar in vehicle and UTA_inh-E02-treated groups prior to dDAVP administration (-3 to 0 h). The reduced urine volume and increased urine osmolality at 0–3 h following dDAVP was prevented by UTA_inh-E02, with little residual difference at 3–6 h in which UTA_inh-E02 was largely cleared. UTA_inh-E02 was also effective in producing a urearesis in hydrated rats given free access to water. Urinary urea and electrolyte measurements showed preferential loss of urea > salt, supporting a urearetic action. In agreement with these results, a salt-sparing diuretic action was reported by the Yang group in mice and rats for a thienoquinolin UT inhibitor with IC₅₀ of 0.32 μM for UT-A inhibition (113), which was identified using the UT-B erythrocyte lysis assay (114). The same group studied the pharmacological properties of a diarylamine UT inhibitor, compound 25a, reporting 15% oral bioavailability, metabolic stability, and inhibition against UT-A and UT-B isoforms (115). Recent structural studies of inhibitor-bound UT-A by electron crystallography revealed potential binding sites for several UT inhibitors (116).

In addition to the synthetic small molecule screens, we also carried out UT inhibitor screens of natural compounds, and approved and investigational drugs, to potentially identify existing drugs or compounds that could be repurposed as urearetics. The screens identified natural products nicotine, sanguinarine and an indolcarbonylchromenone, albeit with relatively weak IC₅₀ of 10–20 μM for inhibition of UT-A1 urea transport (117). Testing of urea analogs was also done with the rationale that several urea analogs are approved drugs, including urea itself, hydroxyurea, and several drugs containing the urea backbone including sulfonylureas, anti-cancer kinase inhibitors, and some older sedatives. The urea analog screen identified dimethylthiourea (DMTU), which reversibly inhibited UT-A1 and UT-B urea transport with IC₅₀ of 2–3 mM. DMTU has been used at high doses in experimental animal models as a hydroxyl radical scavenger and was previously shown to inhibit urinary concentration (118). In rats, a single intraperitoneal dose of 500 mg/kg DMTU gave a predicted peak plasma concentration of 9 mM and urine concentration of 20–40 mM. Rats treated chronically with DMTU showed a sustained 3-fold reduction in urine osmolality and increase in urine output. The DMTU-treated rats developed mild hypokalemia without abnormalities in other serum chemistries. Though DMTU is not a clinical development candidate, the rat data are useful in predicting effects of combined UT-A and UT-B inhibition.

Prospects for advancement of UT inhibitors as urearetic-type diuretics

The advancement of UT inhibitors to the clinic is uncertain, and will depend on their properties, the strength of proof-of-concept studies in animal models, and the perceived unmet need and market size. Screening studies demonstrate that urea transporters are readily druggable targets, with many active compounds identified, including optimized compounds with low nanomolar inhibitory potency and high UT isoform selectivity. With additional medicinal chemistry, if needed, the likelihood is high in generating lead candidates with suitable pharmacological properties for human clinical testing. However, one area of theoretical concern is UT inhibitor toxicity due extrarenal on-target effects. In addition to renal vasa recta microvessels, UT-B is expressed in erythrocytes, urinary bladder, intestines, inner ear, heart and brain, whereas UT-A isoforms (produced by alternative splicing) are expressed mainly in kidney with limited extrarenal expression, at least at the transcript level, in liver, colon and testis. Of note, rare humans with UT-B (erythrocyte Kidd antigen) loss-of-function mutations manifest a mild urinary concentrating defect without extrarenal problems (119); humans with UT-A loss of function mutations have not been reported. Given the broad tissue distribution of UT-B, and the predominant role of UT-A in urinary concentration, it is logical for drug development to focus on UT-A-selective inhibitors. With regard to animal models, though inhibitors have been demonstrated to have salt-sparing urearetic action in pharmacological models of antidiuresis, convincing animal models are not available to test UT inhibitors in clinically relevant conditions, such as congestive heart failure or cirrhosis associated with hyponatremia and diuretic-refractory edema. From the commercial perspective, advancement of UT inhibition is perceived as risky, in part because of the widespread and generally safe and effective use of salt transporter-blocking thiazide and loop diuretics. Nevertheless, based on their physiology and available animal models data, UT inhibitors could address an unmet medical need with a moderate market size and set of unique clinical indications.

Aquaporins as attractive albeit elusive drug targets

There are 13 mammalian aquaporins with a broad tissue distribution in fluid transporting epithelia, astrocytes, adipocytes, myocytes, epidermal cells, erythrocytes, lens fiber cells and many tumors (120,121). Structurally, aquaporin monomers of ~30 kDa molecular size contain six membrane-spanning helical domains that form a narrow internal pore, with monomers arranged in a tetrameric configuration. Some aquaporins transport water exclusively, while others transport water along with small polar molecules such as glycerol and hydrogen peroxide, and gases including carbon dioxide and ammonia. A wide variety of physiological functions have been ascribed to aquaporins. One class, based on their water transport function, includes urinary concentration, brain water balance, glandular fluid secretion and cell migration. For the subclass of aquaporins that also transport glycerol, called aquaglyceroporins, functions include skin hydration, fat and liver metabolism, cellular immunity and wound healing.

Potential indications of aquaporin modulators

Many indications of aquaporin modulators have been proposed (122–124). The most clearcut and compelling indications are in kidney, brain and cancer, in which aquaporin inhibition by small molecules or genetic knockdown is predicted to produce a water diuresis (‘aquaretic’ action) (125), reduce brain water accumulation in ischemic stroke (126), and reduced tumor angiogenesis (127) and metastases, respectively. Other proposed indications, albeit less substantiated, include inflammation, metabolic syndrome, glaucoma, skin disorders, epilepsy, pain and tumor cell proliferation. Augmenting aquaporin function, for example by gene delivery, has potential indications in salivary gland dysfunction and nephrogenic diabetes insipidus, and potentially in cholestatic liver disease, wound healing and obesity (128). Humans or mice with loss of function of AQPs 1–3 manifest nephrogenic diabetes insipidus with defective urinary concentration (129–133). Mutations in AQP0 (MIP, major intrinsic protein of lens fiber), which has structural and transporting roles in the lens, cause congenital cataracts (134). AQP4, the aquaporin expressed in brain astrocytes, is the target of pathogenic IgG1 autoantibodies in the autoimmune disease neuromyelitis optica spectrum disorder (NMOSD) (135).

Effective and selective aquaporin inhibitors could be useful as research tools and clinical development candidates. However, more so than for other epithelial membrane transporter targets, there are substantial challenges in developing aquaporin inhibitors. The wide tissue distribution of aquaporins requires compounds with high selectivity, a difficult task due to the many highly homologous aquaporin isoforms. Plugging a narrow water pore with a small molecule is problematic as a tiny pore would exclude drug-size molecules. The aquaporins are small, tight structures with limited extramembrane peptide loops for drug binding. Further, the high concentration of water, 55 molar, would by mass action compete for binding of inhibitors with a competitive inhibition mechanism. Several heavy metal containing compounds inhibit water permeability of multiple aquaporins, such as mercuric chloride, p-mercuribenzenesulfonate and phenanthroline-coordinated gold, though they are in general toxic and non-specific in their actions as they indiscriminately modify cysteine sulfhydryl residues on proteins. For the glycerol-transporting aquaglyceroporins, AQPs 3, 7 and 9, which contain a relatively wide pore that allows passage of glycerol and other polar molecules, several inhibitor candidates have been identified, including DFP00173, Z443927330 and RG100204, which are reported to have activity in models of inflammation and tumor growth (136–138).

Assaying aquaporin function

Reliable assay of aquaporin function is important for identification of aquaporin inhibitors. Whereas fluorescent sensor and electrophysiological methods are available to identify modulators of ion channels, water movement does not produce an electrical signature, is generally fast, and requires an osmotic challenge. Osmotically-driven transport across cell membranes alters cell volume, whose kinetics determine water permeability. Optical readouts of cell volume, depending on cell type and assay format, can utilize include light scattering, phase-contrast microscopy or interferometry in unlabeled cells; and in labeled cells, fluorescence quenching, total internal reflection and confocal signal detection (139).

Erythrocytes, which express AQP1 water channels, are the classic example of a cell type with high and pharmacologically inhibitable water permeability. Stopped-flow light scattering has been the gold standard method to measure erythrocyte water permeability, in which an erythrocyte suspension is rapidly, in ~1 ms, subjected to an osmotic challenge in a stopped-flow apparatus in which the time course of cell volume change is monitored by 90 degree scattered light intensity (140). Fig. 17A shows light scattering curves for human erythrocytes subjected to a 150-mM NaCl gradient in the absence and presence of the mercurial HgCl₂. Osmotically driven water efflux reduces cell volume and increases light scattering.

Figure 17.

Measurement of rapid, aquaporin-facilitated osmotic water permeability in erythrocytes. A. Stopped-flow light scattering measurement in which an erythrocyte suspension is subjected to an osmotic gradient by rapid, <1 ms, mixing with a hyperosmolar solution, increasing osmolality from 300 to 600 mOsm. Resultant cell shrinking increases 90-degree light scattering. Light scattering curves shown for cells incubated with indicated concentration of AQP1 inhibitor HgCl₂.

B. Measurement of osmotic water permeability using droplet microfluidics. A cell suspension in a microfluidic channel was mixed with an anisomolar solution. Individual droplets generated by oil infusion are passed through a bumpy channel to facilitate mixing and then deposited into a measurement region in which time after mixing is encoded by distance from the deposition point.

C. Erythrocytes were labeled with the volume marker calcein and subject to a 150 mM NaCl gradient using the set-up in panel B. (left) Relative fluorescence in the measurement zone showing reduced fluorescence away from the deposition point in cells exposed to the NaCl gradient. (right) Deduced calcein fluorescence as a function of time after mixing in the absence of a gradient (PBS) and with the NaCl osmotic gradient for erythrocytes from wildtype and AQP1 knockout mice. Figure adapted from ref. 141, used with permission.

For efficient water permeability measurement in suspended cells such as erythrocytes, we devised an alternative, microfluidic method that, unlike the stopped-flow method, does not require specialized instrumentation, uses sub-microliter solution quantities, and replaces challenging kinetic measurements by a single fluorescence image capture using a standard microscope (141). The microfluidic channel shown in Fig. 17B was designed for rapid mixing of a cell suspension with a solution of different composition, with optical measurement of cell fluorescence over time after mixing. The channel consists of generation, mixing and measurement regions. The generation region contains inlets for infusion of the cell suspension, a solution of different composition to impose, for example, an osmotic gradient, and a central inlet to inject an aqueous fluid stream of the same composition as the cell-containing solution to prevent pre-mixing prior to droplet formation. Aqueous droplets containing the cell suspension and second solution are created by infusion of oil through two additional ports. The mixing region consists of a bumpy channel to facilitate rapid solution mixing within droplets. A triangular measurement area allows droplet observation during their radial movement in which radial distance from the origin is related to time after mixing. Fig. 17C shows a water permeability measurement in calcein-labeled erythrocytes from wildtype vs. AQP1 knockout mice after mixing cell suspensions with a hyperosmolar, NaCl-containing solution. The reduced fluorescence in the measurement area results from slowed cell shrinking and slowed consequent calcein fluorescence quenching by hemoglobin (left panel). The deduced kinetics of cell volume (right panel) indicated a >10-fold reduced water permeability in the AQP1-deficient erythrocytes, in agreement of deduced permeability coefficients measured by the stopped-flow light scattering method.

Various measurement approaches are suitable for small molecule screening to identify aquaporin inhibitors. A simple and sensitive method utilizes cytoplasmic calcein fluorescence in which calcein fluorescence is quenched instantaneously by cytoplasmic proteins (142). Fig. 18A shows fluorescently labeled mouse astrocytes after cytoplasmic staining with calcein. Calcein fluorescence responds to changes in extracellular osmolality, increasing as cells swell because of cytoplasmic dilution, and decreasing as cells shrink. Fig. 18B shows time course data from single experiments, revealing reduced water permeability in AQP4-deficient astrocytes and increased water permeability with higher temperature. A chloride-sensing YFP can be substituted for calcein, albeit with reduced sensitivity and signal size. For erythrocyte AQP1, we devised a simple and sensitive method as the inverse the acetamide lysis method for UT inhibitors (103) – an AQP1 inhibitor protects rather than promotes lysis. Alternative functional screens for aquaporin inhibitor discovery include high-content imaging of cell shape, total internal reflectance and phase-contrast microscopy. For aquaporins that transport solutes such as glycerol or hydrogen peroxide, screens can be devised to measure transported solutes rather than water. Binding screens offer another alternative in which binding to purified, immobilized aquaporin is measured rather than aquaporin function.

Figure 18.

Osmotic water permeability in primary cultures of mouse cortical brain astrocytes. A. Adherent cells were stained with calcein as a volume marker, in which increased cell volume reduces calcein fluorescence quenching (increases fluorescence) as a consequence of dilution of cytosolic proteins. B. Calcein fluorescence in astrocytes cultured from wildtype and AQP4 knockout mice in response to changes in extracellular osmolality from 300 to 150 mOsm, and then back to 100 mOsm. Measurements done at 12 and 37 °C. Figure adapted from ref. 142, used with permission.

Prospects for small molecule aquaporin inhibitor drugs

Notwithstanding the challenges noted above, there is no theoretical reason why identification of selective, small molecule aquaporin inhibitors is not possible for the water-only aquaporins. Inhibitors may act, for example, as allosteric modulators by binding to unique sites on transmembrane helices away from the water pore. Investigation of the inhibition mechanism of the aquaglyceroporin inhibitors mentioned above may provide clues to identify inhibitors of the non-glycerol transporting aquaporins. Using the calcein and erythrocyte lysis assays, we screened several hundred thousand synthetic small molecules at 25 μM for inhibition of AQP1 and AQP4 water transport, but failed to identify active compounds. We also screened approved and investigational drugs, as well as a biased set of synthetic small molecules selected by computational chemistry. A number of compounds have been asserted to inhibit AQP1 or AQP4 water permeability, including several antiepileptics and diuretics, carbonic acid inhibitors such as acetazolamide and its analog TGN-020, dimethylsulfoxide, triethylammonium and others (143). However, retesting of these putative inhibitors by robust water transport assays has failed to confirm bona fide water transport inhibition (140, 144–146). Reported effects of these claimed aquaporin inhibitors might be explained by aquaporin-independent actions such as osmotic clamp effects of dimethylsulfoxide, solute carrier transport inhibition by diuretics, potassium channel inhibition by triethylammonium, and carbonic anhydrase inhibition by acetazolamide analogs. A recent abstract reported preliminary data on an AQP2 inhibitor (147), with predicted aquaretic action similar to that of vasopressin antagonists which block the antidiuretic signal in kidney collecting duct epithelial cells to translocate AQP2 to the plasma membrane. A safe and effective aquaporin inhibitor is anticipated to attract considerable commercial interest.

Aquaporumab, an engineered monoclonal antibody for a disease of AQP4

NMOSD is an inflammatory demyelinating disease of the central nervous system that affects spinal cord and optic nerve to a greater extent than brain, often producing impaired vision and motor function (148,149). Most NMOSD patients are seropositive for polyclonal IgG1 autoantibodies against AQP4, called AQP4-IgG. Cell culture, animal models and human studies indicate that NMOSD pathogenesis is primarily initiated by AQP4-IgG binding to AQP4 on astrocyte cell membranes, resulting in astrocyte injury by complement and cellular mechanisms, with secondary inflammation, blood-brain barrier disruption, and oligodendrocyte and neuronal injury (150). Currently approved NMOSD drugs include immunosuppressive agents that target B cells, IL-6 receptors or the complement system (151,152).

Because binding of AQP4-IgG to AQP4 on the astrocyte plasma membrane is a primary initiating event in NMOSD pathogenesis, its inhibition is a logical therapeutic strategy. Notwithstanding the limitations of a small molecule to block protein-protein interactions, we screened approved and investigational drugs and nutraceuticals using an ELISA assay of antibody binding to AQP4-transfected cells (153). The screen identified several blockers of AQP4-IgG binding to AQP4 that showed efficacy in animal models of seropositive NMOSD, albeit with low potency and limited CNS penetration. Subsequently, an antibody-based approach was developed, which we called aquaporumab, in which a high-affinity, monoclonal anti-AQP4 antibody was engineered to block the binding of pathogenic human AQP4-IgG antibodies to AQP4 (154) (Fig. 19A). Aquaporumab was generated from a high-affinity human monoclonal AQP4-IgG identified by sequence analysis of plasma cells from cerebrospinal fluid of a NMOSD patient. Fc mutations L234A/L235A were introduced to neutralize its complement and cell-mediated effector functions, and mutations were made in the Fab region following affinity maturation to increase its AQP4 binding affinity (155). Because of the relatively large size of the IgG1-class aquaporumab antibody compared to AQP4, a sufficiently high concentration of aquaporumab sterically blocks binding of the lower-affinity polyclonal AQP4-IgG in NMOSD patients. Aquaporumab blocked complement-dependent cytotoxicity produced by pathogenic human antibodies with IC₅₀ ~100 pM (Fig. 19B). Aquaporumab efficacy in reducing NMOSD pathology was also demonstrated in spinal cord slice and mouse models of NMOSD. Because of its selective targeting, aquaporumab may be safe and suitable for monotherapy or combination therapy with approved immunosuppressive drugs for treatment of acute disease exacerbations or chronic disease suppression.

Figure 19.

Aquaporumab anti-AQP4 antibody for therapy of AQP4 autoantibody-seropositive neuromyelitis optica spectrum disorder (NMOSD). In NMOSD polyclonal anti-AQP4 IgG1 antibodies injure astrocytes in the central nervous system by activation of immune mechanisms including complement-dependent cytotoxicity. Aquaporumab is a non-pathogenic, engineered IgG1-class antibody lacking Fc effector functions that blocks binding of pathogenic anti-AQP4 antibodies to AQP4 on astrocytes. A. (left) To-scale diagram of AQP4 antibody and membrane-associated AQP4 water channels. (right) Aquaporumab antibody structure showing variable and constant regions in heavy and light chains, and effector function-neutralizing Fc mutations. B. Cytotoxicity in AQP4-expressing cells produced by pathogenic anti-AQP4 autoantibody and complement injury. Aquaporumab reduces complement-dependent cytotoxicity as shown by Alamar blue assay (left) and live/dead cell staining (right). Figure adapted from refs. 154 and 155, used with permission.

Further preclinical development of aquaporumab requires investigation of its pharmacokinetics, CNS penetration and potential toxicity on AQP4-expressing tissues outside of the CNS. Despite its unique, non-immunosuppressive mechanism of action, the practical challenge in commercial development of aquaporumab is the small number of NMOSD patients and the recent approval of several repurposed immunosuppressive antibodies that are effective in most patients (156). The novel paradigm of a blocking, non-pathogenic engineered antibody may be applicable to many other neurological and autoimmune disorders such as myasthenia gravis, Lambert-Eaton syndrome, and several autoimmune encephalitides and neural hyperexcitability syndromes. An interesting and unrelated potential application of aquaporumab is for therapy of tumors that overexpress AQP4, including gliomas (157) and various epithelia-derived tumors. Tumor cell killing could be accomplished using an aquaporumab antibody-drug conjugate containing a cytotoxic payload, or an Fc-modified aquaporumab with enhanced antibody-dependent cellular cytotoxicity effector function (158).

Solute carrier-26 (SLC26) transporters as epithelial drug targets

The SLC26 family of membrane proteins consists of at least 10 anion transporters, some functioning as exchangers and some as channels, which transport various anion including chloride, bicarbonate, iodide, sulfate, and oxalate (159–161). Hereditary diseases caused by loss of function mutations in SLC26 genes include kidney stones (SCL26A1 and SLC26A6), chondrodysplasias (SLC26A2), chloride-losing diarrhea (SLC26A3), Pendred syndrome (hearing loss, SLC26A4), hypothyroidism (SLC26A7) and male infertility (SLC26A8). SLC26A9 is a prominent modifier gene in CF in which polymorphisms are associated with altered disease phenotype (162). Based on information from human diseases and knockout mice, DRA and pendrin emerge as the most attractive drug targets for inhibitor development. SLC29A9 is a potential target for CF therapy, though the desired effect, increasing chloride transport to compensate for defective CFTR, would require SLC29A9 upregulation or activation. Until recently, the only SLC26 inhibitors were of low potency and non-specific, including disulfonic stilbenes, niflumic acid and NPPB (5-nitro-2-(3-phenylpropylamino)benzoic acid).

SLC26A3 (DRA) inhibitors for constipation and hyperoxaluria

SLC26A3 (DRA, down-regulated in adenoma) functions as an electroneutral exchanger of chloride for bicarbonate and oxalate, and can transport other anions including iodide. DRA is expressed mainly at the luminal membrane of intestinal epithelial cells in the colon where it facilitates chloride, fluid and oxalate absorption (Fig. 5) (163). Humans and mice with DRA loss of function manifest chloride-losing diarrhea and reduced urinary oxalate, providing a rationale for DRA inhibition to treat constipation and oxalate-related kidney disorders including oxalate kidney stones and hyperoxaluria (164,165).

A robust screen to identify DRA inhibitors was established using FRT cells stably expressing human DRA and a YFP iodide sensor (166). An initial screen followed by testing of analogs of hits produced DRA inhibitors of several chemical classes. The dimethylcoumarin DRA_inh-A250 fully and reversibly inhibited DRA-mediated Cl⁻ exchange with HCO₃⁻, I⁻ and SCN⁻ with IC₅₀ ~ 0.2 μM, without effect on the closely related SLC26 proteins PAT1 (SLC26A6) or pendrin (SLC26A4), or the major intestinal ion transporters. In isolated colonic loops DRA_inh-A250 administration strongly blocked fluid absorption without changing luminal pH, and increased stool output and hydration in a mouse model of constipation produced by loperamide. A medicinal chemistry program produced lead candidate DRA inhibitors, one of which, DRA_inh-270, inhibited DRA anion exchange with IC₅₀ of 25 nM (Fig. 20A) (167). DRA_inh-270 at 10 mg/kg normalized stool output and hydration in the loperamide mouse model of constipation (Fig. 20B). Further analysis of DRA_inh-270 properties indicated good stability and favorable pharmacokinetics following oral administration, without detectable toxicity.

Figure 20.

DRA (SLC26A3) inhibitor for therapy of constipation. A. Concentration-dependent inhibition of DRA-mediated exchange of chloride for iodide by DRA_inh-A270. B. DRA_inh-A270 normalizes stool hydration in a mouse model of constipation produced by loperamide. Mice were administered loperamide and indicated concentrations of DRA_inh-A270. Water content of collected stool measured by wet-to-dry weight ratios (mean ± S.E.M, 4–6 mice per group, ** P < 0.01). Figure adapted from ref. 167, used with permission.

A small molecule screen was also done for the closely related anion exchanger SLC26A6 (PAT1, putative anion transporter 1), which is expressed in intestine mainly at the luminal membrane of enterocytes lining the ileum where it facilitates chloride and fluid absorption (168). The motivation for the screen was to develop a research tool to investigate the differential functions of DRA vs. PAT1 in fluid absorption in different intestinal regions. The screen identified the PAT1-selective pyrazolo-pyrido-pyrimidin one inhibitor PAT1_inh-B01, with IC₅₀ ~350 nM (169). PAT1_inh-B01 inhibited fluid absorption in mid-jejunum by 50%, which increased to >90% when administered together with DRA inhibitor DRA_inh-A270. PAT1_inh-B01 blocked fluid absorption in ileum by >80% whereas DRA_inh-A270 had no effect; in colon PAT1_inh-B01 was ineffective whereas DRA_inh-A270 completely blocked fluid absorption. The pharmacological results supported the conclusions from prior knockout studies about the pro-absorptive, segment-specific roles of PAT1 vs. DRA. A PAT1 inhibitor used chronically, though predicted to be effective for constipation treatment, may promote the formation of oxalate kidney stones as seen in humans with loss-of-function PAT1 mutations (170). There may be undesired on-target effects outside of the intestine as well because PAT1 is also expressed in kidney, pancreas, glandular epithelia and heart. However, for short-term application, a PAT1 inhibitor could be useful for treatment of hyposecretory disorders of the small intestine including CF-related meconium ileus and distal intestinal obstruction syndrome.

Reduced urinary oxalate excretion with DRA loss-of-function suggested a second indication for DRA inhibitors – prevention of oxalate kidney stones and treatment of enteric hyperoxaluria in which pathological oxalate absorption by the intestine can produce kidney injury. Enteric hyperoxaluria is commonly seen following gastric bypass surgery, inflammatory bowel disease, intestinal resection and chronic pancreatitis (171,172). Oxalate, which has no known biological functions in humans, enters the body by ingestion of oxalate-containing food and is cleared by the intestine and kidney (Fig. 21A) (173). In intestine, oxalate is secreted in the small intestine and absorbed in the colon. Inhibition of oxalate absorption in colon by a DRA inhibitor is predicted to reduce urinary oxalate excretion and hence kidney exposure.

Figure 21.

DRA (SLC26A3) inhibitor for therapy of oxalate kidney stones. A. Oxalate dynamics showing entry by dietary intake and hepatic metabolism, and clearance through stool and urine. Inhibition of colonic DRA reduces oxalate absorption, reducing kidney exposure to oxalate.

B. Oxalate nephrolithiasis protocol in which mice were placed on a high oxalate-low calcium diet to cause oxalate hyperabsorption. Mice were treated twice-daily with DRA_inh-A270, and blood/kidneys harvested as indicated. C.-F. DRA_inh-A270 reduces urinary oxalate, measured by oxalate/creatinine ratios (C.), maintains renal function, measured by serum creatinine (D.), reduces oxalate crystals in kidneys, measured by cross-polarization microscopy (E.) and prevents renal injury, assessed by scoring of hematoxylin/eosin-stained paraffin sections (F.). Mean ± S.E.M, 10 mice per group, * P < 0.05 ** P < 0.01, *** P < 0.001. Figure adapted from ref. 174, used with permission.

DRA_inh-270 was tested in an established model of hyperoxaluria and oxalate nephropathy in which mice were fed a high-oxalate low-calcium diet for two weeks (Fig. 21B). The reduced dietary calcium increases the oral bioavailability of oxalate, producing enteric hyperoxaluria with intrarenal calcium oxalate crystal deposition and progressive renal injury (174). During the high oxalate diet mice were administered DRA_inh-270, or control vehicle, twice daily. Fig. 21C–F shows that compared to control mice, the DRA_inh-270-treated mice had remarkably reduced urinary oxalate excretion, calcium oxalate crystal deposition and renal injury, the latter as assessed by serum creatinine and renal histology. These findings support the potential utility of DRA inhibitors to treat oxalate nephrolithiasis and enteric hyperoxaluria, major unmet medical needs in large patient populations that are currently without effective drug therapy.

DRA inhibition therapy of hyperoxaluria, and some forms of constipation, would require long-term chronic administration and hence possible undesired effects including diarrhea as well as on-target actions outside of the intestines. In addition to colon, DRA is expressed in prostate, testis, pancreas and sweat glands. Humans with DRA loss-of-function mutations, in addition to chloride-losing diarrhea, manifest varying degrees of male subfertility. There are also potential links between DRA polymorphisms and intestinal inflammation, hyperuricemia and tumorigenesis. Clinical development of a systemically absorbed DRA inhibitor such as DRA_inh-270 would require attention to these potential liabilities.

Motivated by the expression of DRA at the lumen-facing membrane of colonic epithelial cells, we reasoned that a DRA inhibitor targeting the extracellular surface of DRA could afford the unique opportunity to create an orally administered, non-absorbable inhibitor with minimal systemic exposure. Additional small molecule screening and analog testing produced thiazolo-pyrimidin-5-one and 3-carboxy-2-phenylbenzofuran DRA inhibitors with an extracellular site of action as deduced from kinetic washout studies and supported by molecular docking (175). In vivo efficacy was demonstrated in a mouse constipation. As was done for glycine hydrazide CFTR inhibitors (55), it may be possible to synthesize non-absorbable, externally acting DRA inhibitors by conjugation with macromolecular moieties such as polyethylene glycol.

SLC26A4 (pendrin) inhibitors for inflammatory lung disorders

Another potential drug target in the SLC26 gene family is pendrin (SLC26A4), an electroneutral exchanger of chloride for several anions of physiological significance including bicarbonate, iodide and thiocyanate. Pendrin is expressed in non-sensory epithelial cells in the inner ear, thyroid follicular cells, airway surface epithelial cells, kidney distal tubule beta-intercalated cells and adrenal medulla chromaffin cells (176). Pendrin inhibition is predicted to be of potential use in inflammatory lung disorders, including cystic fibrosis, asthma, acute lung injury, chronic obstructive pulmonary disease and chronic rhinosinusitis, by increasing the thickness of the airway surface liquid (ASL) layer lining air-exposed surface epithelial cells (177,178). As depicted in Fig. 22A, pendrin activity tonically reduces ASL volume by absorption of chloride, which is exchanged for bicarbonate and converted to water and CO₂. Pendrin inhibition is predicted to increase ASL volume by blocking this tonic pro-absorptive mechanism. Pendrin expression in humans and animal models of inflammatory lung diseases is greatly upregulated, up to 100-fold, and pendrin knockout in mice is associated with improved outcomes in models of lung inflammation (179). Another potential use of pendrin inhibitors is in hypertension and edema, by a mechanism involving inhibition of chloride reabsorption in distal tubules. Pendrin inhibitors may also be useful in some form of hyperthyroidism and thyroid tumors by blocking iodide entry into thyroid cells.

Figure 22.

Pendrin (SCL26A4) inhibitor for therapy of inflammatory airways diseases.

A. Pendrin inhibition hydrates the airway surface liquid layer by inhibition of apical surface chloride/bicarbonate exchange, reducing chloride influx into surface epithelial cells.

B. Concentration-dependent inhibition of pendrin-mediated exchange of chloride for iodide, nitrate or thiocyanate by PDS_inh-A01. C. PDS_inh-A01 increases the thickness of the airway surface liquid layer in IL-13-treated primary cultures of human airway epithelial cells from normal and cystic fibrosis (CF) subjects grown on a porous support. IL-13 simulates airway inflammation, producing pendrin upregulation. Airway surface liquid thickness measured by confocal microscopy after staining with a red-fluorescent dextran. Figure adapted from ref. 180, used with permission.

Motivated by these potential indications and the need for good research tools to study pendrin biology, we developed a cell-based screen to identify pendrin inhibitors in which FRT cells stably expressing human pendrin and a YFP iodide sensor were subject to an iodide gradient (180), similar to the DRA screen. Pendrin-facilitated chloride/iodide exchange allows iodide entry into cells, resulting in reduced YFP fluorescence which is blocked by pendrin inhibition. A small molecule screen with follow-on testing of chemical analogs produced several classes of pendrin inhibitors that selectively and reversibly blocked chloride/anion exchange, with one example shown in Fig. 22B. A subsequent medicinal chemistry study identified analogs with greater pendrin inhibitory potencies (181).

Cell culture and animal data support the potential utility of pendrin inhibitors in inflammatory airways diseases (182,183). Primary cultures of human airway epithelial cells from non-CF and CF subjects were grown on porous supports and incubated with the pro-inflammatory cytokine IL-13 to upregulate pendrin expression. The IL-13 treated cells showed strong chloride/bicarbonate exchange that was blocked by pendrin inhibition (180). Pendrin inhibition increased airway surface liquid thickness in IL-13-treated non-CF and CF cells without changing ASL pH, and was without significant effect in untreated cells in which pendrin expression was relatively low (Fig. 22C). Subsequent screening by Park et al. (182) using the FRT cell assay produced a different pendrin inhibitor that in mice was effective in models of ovalbumin-induced asthma and LPS-induced acute lung injury (183,184).

Motivated by the expression of pendrin in kidney distal tubule epithelia, and the finding of salt wasting in knockout mice lacking both pendrin and the sodium-chloride cotransporter NCC (185–187), we investigated the potential diuretic action of a pendrin inhibitor (188). Inhibitor administration to mice at predicted therapeutic doses, as determined in pharmacokinetic studies, did not affect urine output, osmolality, salt excretion or acid-base balance. However, pendrin inhibition showed a synergistic effect when used together with a maximal dose of the loop diuretic furosemide, producing significantly greater urine output and salt excretion than that produced by furosemide alone. Therefore, although pendrin inhibition alone does not have significant diuretic effect, combination therapy of a pendrin inhibitor with a conventional diuretic has potential applications for treatment of hypertension and edema, perhaps including diuretic-resistant edema.

Though the identification of selective, small molecule pendrin inhibitors and demonstration of their efficacy in cell and animal models provides a rationale for their further development, there are several challenges. Humans with pendrin loss-of-function mutations have Pendred syndrome, characterized by sensorineural deafness and often vestibular dysfunction and hypothyroidism with goiter (189,190). Although impaired hearing and vestibular functions are significant concerns, it is unclear whether these manifestations in Pendred disease are more developmental vs. functional. Thyroid dysfunction is another concern with pendrin inhibition, though it is not known why only a subset of Pendred subjects develop thyroid problems, and thyroid hormone replacement therapy may mitigate impaired thyroid hormone secretion. Metabolic alkalosis is another potential concern based on pendrin action to conserve bicarbonate in distal tubules. Further animal testing will be needed to clarify these issues. For inflammatory lung diseases, one potential approach to reduce on-target extrapulmonary effects is to deliver a relatively metabolically unstable pendrin inhibitor by nebulization to minimize systemic exposure. Another possible approach it to identify externally acting pendrin inhibitors, as was done for the close pendrin homolog DRA (175), Delivery of a non-absorbable pendrin inhibitor conjugate by the nebulized route may inhibit pendrin function in surface airway epithelial cells without effect in the inner ear and thyroid gland.

Opportunities and perspective

Significant unmet needs and opportunities remain for small molecule discovery in epithelial transport physiology, as selective modulators and approved drugs exist for only a small fraction of epithelial membrane transporters. In recent years, most approved drugs have been concentrated in the areas of oncology, neurology, immunology, infectious disease and rare diseases, with few directed at epithelial biology. An immediate opportunity lies in building on targets of already approved drugs, improving on drug efficacy and pharmacodynamics, minimizing on- and off-target side effects, and expanding clinical indications. For example, whereas many effective diuretics with distinct targets are in routine use, there remains a need for drugs for more efficacious therapies to treat edema and fluid overload in patients with heart failure and volume-sensitive hypertension that are refractory to current agents. Diuretics with salt-sparing action may be particularly useful, such as ‘urearetics’ that target urea transporters or ‘aquaretics’ that target aquaporins. Other novel diuretic targets with drug candidates at various stages of development include modulators of pendrin chloride-bicarbonate exchange, ROMK potassium channels, ClC-K chloride channels, aldosterone synthase and WNK-SPAK signaling. Another example is in cystic fibrosis therapeutics (83,191). Although triple combination CFTR modulators have markedly improved outcomes in cystic fibrosis, they generally achieve incomplete restoration of native CFTR function in patients with responsive mutations and may be associated with adverse effects. Emerging agents aim to complement existing therapies by targets aspects of CFTR biology not fully addressed by current modulators, including stabilizers of CFTR domain folding, amplifiers of CFTR expression, inhibitors of chaperone-CFTR interactions, and read-through agents for CFTR variants containing premature stop mutations (192).

Substantial opportunities remain for small molecule discovery targeting epithelial transport processes. Promising yet underexplored targets include transporters of phosphate, calcium, magnesium, glucose, amino acids, peptides, and monocarboxylates; TRP cation channels; ClC anion channels, sodium-proton exchangers, TMEM16 anion channels, acid-sensing cation channels; multi-drug toxin extrusion channels; and paracellular permeability proteins such as claudins. Potential therapeutic applications span secretory diarrheas, dry eye, chronic kidney disease, acid-base disorders, hypertension, heart failure, pancreatitis, inflammatory lung disorders, autoimmunity associated with exocrine gland dysfunction, and neurological disorders associated with altered cerebrospinal fluid dynamics.

Although not discussed here, many potential targets involve epithelial cell receptors and intracellular signaling pathways that modulate epithelial transport, such as modulators of cyclic nucleotide, calcium, epidermal growth factor, and Wnt-beta-catenin signaling. In general, well-designed screens of sufficiently large and diverse compound libraries will likely yield inhibitors of defined molecular targets or cellular phenotypes. However, while such efforts often generate valuable research tools, identification of compounds with the pharmacokinetic, safety and selectivity profiles required for clinical development is far less certain. Activators of constitutively active transporters and small molecules that disrupt protein–protein interactions pose particular challenges. Notwithstanding these considerations, given the breadth of physiologically important yet pharmacologically underexploited epithelial transport mechanisms, continued innovation in discovery strategies is likely to yield new therapeutic targets and candidate drugs addressing unmet medical needs in epithelial fluid and solute disorders.