A Repurposed Drug Screen for Compounds Regulating Aquaporin 5 Stability in Lung Epithelial Cells

Abstract

Aquaporin 5 (AQP5) is expressed in several cell types in the lung and regulates water transport, which contributes to barrier function during injury and the composition of glandular secretions. Reduced AQP5 expression is associated with barrier dysfunction during acute lung injury, and strategies to enhance its expression are associated with favorable phenotypes. Thus, pharmacologically enhancing AQP5 expression could be beneficial. Here, we optimized a high-throughput assay designed to detect AQP5 abundance using a cell line stably expressing bioluminescent-tagged AQP5. We then screened a library of 1153 compounds composed of FDA-approved drugs for their effects on AQP5 abundance. We show compounds Niclosamide, Panobinostat, and Candesartan Celexitil increased AQP5 abundance, and show that Niclosamide has favorable cellular toxicity profiles. We determine that AQP5 levels are regulated in part by ubiquitination and proteasomal degradation in lung epithelial cells, and mechanistically Niclosamide increases AQP5 levels by reducing AQP5 ubiquitination and proteasomal degradation. Functionally, Niclosamide stabilized AQP5 levels in response to hypotonic stress, a stimulus known to reduce AQP5 levels. In complementary assays, Niclosamide increased endogenous AQP5 in both A549 cells and in primary, polarized human bronchial epithelial cells compared to control-treated cells. Further, we measured rapid cell volume changes in A549 cells in response to osmotic stress, an effect controlled by aquaporin channels. Niclosamide-treated A549 cell volume changes occurred more rapidly compared to control-treated cells, suggesting that increased Niclosamide-mediated increases in AQP5 expression affects functional water transport. Taken together, we describe a strategy to identify repurposed compounds for their effect on AQP5 protein abundance. We validated the effects of Niclosamide on endogenous AQP5 levels and in regulating cell-volume changes in response to tonicity changes. Our findings highlight a unique approach to screen for drug effects on protein abundance, and our workflow can be applied broadly to study compound effects on protein abundance in lung epithelial cells.

Keywords: Aquaporin 5; Niclosamide; lung epithelial cells; proteasome; repurposed drugs.

FIGURE 1

Development of an assay to measure AQP5 Abundance. (A) Overview of AQP5-HiBiT stable cell line generation. HiBiT blotting shows AQP5-HiBit construct response to cycloheximide (CHX) and MG-132 in pooled cells. (B) AQP5-HiBiT signal in cells were treated with vehicle or MG-132 (n = 128 wells/condition) used for Z′ score calculation. (C) Overview of screening procedure. AQP5-HiBit cells were plated in 384-well plates and treated with 1153 compounds from the Selleck FDA drug library. AQP5-HiBiT luminescence values were recorded. (D) Drug screen results sorted by ranked AQP5-HibiT luminescence, with compounds of interest colorized as indicated.

FIGURE 2

Niclosamide, candesartan, and Panobinostat show dose-dependent accumulation of AQP5-HiBiT. (A) Heatmap showing breakdown of compound response classified by drug class. (B) Dose response curves of selected proteasome inhibitors (bortezomib, carfilzomib) and E1 inhibitor MLN2238, along with top-hit compounds candesartan, Niclosamide, and Panobinostat. (C,D, and E) Dose response curves for Niclosamide, Panobinostat, and candesartan in AQP5-HiBiT (EC50 values of 3.967, 1.109, and 5.687 μM respectively).

FIGURE 3

Cell viability after treatment with Niclosamide, candesartan, and Panobinostat. (A,B) Viability of AQP5-HiBiT cells in response to 6 h (A) and 24 h (B) Niclosamide at various doses. (C,D). Viability of AQP5-HiBiT cells in response to 6 h (C) and 24 h (D) Panobinostat at various doses. (E,F) Viability of AQP5-HiBiT cells in response to 6 h (E) and 24 h (F) candesartan at various doses.

FIGURE 4

AQP5 is degraded in the proteasome. (A) Immunoblot analysis of accumulation of AQP5 in Beas-2B cells in response to a MG-132 at various time points. (B) Time course immunoblot analysis of AQP5 with or without MG-132 co-treatment. (C) Decay curve of AQP5 abundance in Beas-2B treated with cycloheximide, with or without cotreatment with MG-132. Densitometry data are mean ± SEM of 3 independent experiments. (D) Normalized AQP5-HiBiT luminescence in response to 6-h treatment with CHX, carfilzomib (CFZ), or cotreatment. N = 96 wells/condition. Graph is representative of repeated independent experiments. (E) Normalized AQP5-HiBiT luminescence in response to 6-h treatment with CHX, Niclosamide, or cotreatment. N = 32 wells/condition. Graph is representative of repeated independent experiments. (F) Normalized AQP5-HiBiT luminescence in response to 6-h treatment with CHX, candesartan, or cotreatment. N = 48 wells/condition. Graph is representative repeated independent experiments. (G) Normalized AQP5-HiBiT luminescence in response to 6-h treatment with CHX, Panobinostat, or cotreatment. N = 48 wells/condition. Graph is representative of repeated independent experiments. (D–G). *p <0.05, **p <0.01, *** p <.001, ****p <0.0001 by 1-way ANOVA with correction for multiple comparisons by Tukey’s method.

FIGURE 5

Niclosamide reduces AQP5 ubiquitination and increases endogenous AQP5 levels in A549 cells and primary HBE cells: (A) AQP5-HiBiT cells were treated with CFZ and either) vehicle or Niclosamide (200 nM for 16 h, followed by TUBES pulldown to capture ubiquitinated substrates. Pull-down fractions were separated by electrophoresis and HiBiT luminescence was captured on the nitrocellulose membrane. Short exposure (left) shows a diminished primary band at ∼30 kDa in Niclosamide-treated cells, representing mono-ubiquitinated AQP5. A long exposure (right) shows diminished higher molecular weight bands (arrows) in Niclosamide-treated cells, representing multi- and poly-ubiquitinated AQP5. (B) AQP5 immunoblot in Beas-2B and A549 cells. (C). AQP5 immunoblot in Vehicle or Niclosamide (2 μM)-treated A549 cells at 6 h. (D) Densitometry of AQP5 in vehicle vs. Niclosamide treated cells from n = 3 independent experiments. * p = 0.03 by unpaired t-test. (E) AQP5 or AQP4 immunoblot in Vehicle or Niclosamide (200 nM)-treated primary human bronchial epithelial cells at 16 h. (F,G) Densitometry of AQP5 or AQP4 in vehicle vs. Niclosamide-treated cells from n = 6 wells/condition. * p = 0.04, ns = not significant by unpaired t-test.

FIGURE 6

Niclosamide Effect on AQP5 abundance in response to tonicity changes. (A) Normalized AQP5-HiBiT luminescence in response to 6-h treatment with hypotonic media (∼150 mOsm), and hypertonic media (∼400 mOsm). N = 48 wells/condition. Graph is representative of repeated independent experiments. (B) Normalized AQP5 HiBiT luminescence response to 6-h treatment with hypotonic media, Niclosamide, or combination. N = 48 wells/condition. Graph is representative of repeated independent experiments. *p <0.05, **p <0.01, *** p <.001, ****p <0.0001 by 1-way ANOVA with correction for multiple comparisons by Tukey’s method. (C) Calcein in A549 cells in response hypertonic stimuli of 400, 420, and 450 mOsm. (D) Calcein fluorescence in vehicle or Niclosamide-treated (2 μM, 6 h) A549 cells. Dots represent average values of n = 3 wells, with error bars displaying SEM values at a given time-point.