Discovery of a Small Molecule Drug Candidate for Selective NKCC1 Inhibition in Brain Disorders

Abstract

Aberrant expression ratio of Cl- transporters, NKCC1 and KCC2, is implicated in several brain conditions. NKCC1 inhibition by the FDA-approved diuretic drug, bumetanide, rescues core symptoms in rodent models and/or clinical trials with patients. However, bumetanide has a strong diuretic effect due to inhibition of the kidney Cl- transporter NKCC2, creating critical drug compliance issues and health concerns. Here, we report the discovery of a new chemical class of selective NKCC1 inhibitors and the lead drug candidate ARN23746. ARN23746 restores the physiological intracellular Cl- in murine Down syndrome neuronal cultures, has excellent solubility and metabolic stability, and displays no issues with off-target activity in vitro. ARN23746 recovers core symptoms in mouse models of Down syndrome and autism, with no diuretic effect, nor overt toxicity upon chronic treatment in adulthood. ARN23746 is ready for advanced preclinical/manufacturing studies toward the first sustainable therapeutics for the neurological conditions characterized by impaired Cl- homeostasis.

Keywords: Down syndrome; GABAergic transmission; KCC2; NKCC1; NKCC2; autism; brain disorders; chloride homeostasis; drug discovery; small molecule.

Conflict of interest statement

A.C. and L.C. are named as co-inventors on the following granted patent: US 9,822,368; EP 3083959; JP 6490077; A.C. and L.C. are named as co-inventors on the patent application WO 2018/189225. A.S., M.B., J.A.O., A.C., M.D.V., and L.C. are named as co-inventors on patent application IT 102019000004929.

© 2020 The Authors.

Figures

Graphical abstract

Figure 1

In Vitro Selection of the Selective NKCC1 Inhibitor ARN23746 as a Lead Compound (A) Schematic representation of the intervention point in bumetanide’s structure for synthesizing novel bumetanide analogs. (B) Quantification of the inhibitory activity of bumetanide (Bume) and bumetanide analogs (10, 100 μM) in NKCC1-(left) or NKCC2-(right) transfected HEK293 cells in the Cl− influx assay. Data are presented as a percentage of the respective control DMSO. Data represent mean ± standard error of the mean (SEM) from 3–4 independent experiments (Kruskal-Wallis one way ANOVA on Ranks, NKCC1 10 μM: H = 84.898, DF = 6, p < 0.001; NKCC1 100 μM: H = 86.799, DF=6, p < 0.001; NKCC2 10 μM: H = 40.700, DF = 6, p < 0.001; NKCC2 100 μM: H = 70.569, DF = 6, p < 0.001, Dunn’s post hoc test, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). (C) Representation of the ligand-based computational strategy to discover novel molecular scaffolds that inhibit NKCC1. The obtained bumetanide pharmacophore (1) consists of three H-bond acceptor (HBA) features (red spheres), three H-bond donor (HBD) interactions (blue spheres), one lipophilic feature (green sphere), and one stacking feature (brown sphere) anchored around the central aromatic core. Ligand disposition was then implemented by superimposing other known unspecific NKCC1 inhibitors (2), revealing shared features and different dihedral dispositions of substituent around the central aromatic core. This model was used as a search filter for the virtual screening (3) of our internal chemical collection (∼15,000 compounds). Results generated from in vitro testing of the 165 initial hits (4) were then used to retrain the model (5) and perform a second more specific screening of our chemical library and commercial chemical libraries (∼135,000 compounds). This iterative computational cycle led to hit compounds (6) ARN22393 and ARN22394. (D) Quantification of the inhibitory activity of the indicated compounds (10, 100 μM) in NKCC1-transfected HEK293 cells (Cl− influx assay). Data are presented as a percentage of the respective control DMSO. Data represent mean ± SEM from 3–4 independent experiments (Kruskal-Wallis one way ANOVA on Ranks, 10 μM: H = 37.119, DF = 3, p < 0.001; 100 μM: H = 33.724, DF = 3, p < 0.001, Dunn’s post hoc test, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). (E) Chemical structures of NKCC1 inhibitors with novel scaffold. (F) Left, example traces obtained in the Cl− influx assay on NKCC1-transfected HEK293 cells for each compound (100 μM). The arrow indicates the addition of NaCl (74 mM) to initiate the NKCC1-mediated Cl− influx. Right, quantification of the NKCC1 inhibitory activity of the indicated compounds (10, 100 μM) in experiments such as those on the right. Data are presented as a percentage of the respective control DMSO. Data represent mean ± SEM from 3–4 independent experiments (10 μM: one way ANOVA, F(4,84) = 33.048, p < 0.001, Dunnett's post hoc test, ∗p < 0.05, ∗∗∗p < 0.001; 100 μM: Kruskal-Wallis one way ANOVA on Ranks, H = 50.796, DF = 4, p < 0.001, Dunn’s post hoc test, ∗∗∗p < 0.001. (G) Left, example traces obtained in the Ca2+ influx assay on 3DIV hippocampal mouse neurons for each tested compound (100 μM). The arrows indicates the addition of GABA (100 μM) and KCl (90 mM). Right, quantification of the effect of the indicated compounds (10, 100 μM) in the Ca2+ influx assay on 3DIV neurons. Data are presented as a percentage of the control DMSO. Data represent mean ± SEM from three independent experiments. 10 μM: one-way ANOVA, F(5,161)= 77.184, p < 0.001, Dunnett's post hoc test ∗∗p < 0.01, ∗∗∗p < 0.001; 100 μM: Kruskal-Wallis One ANOVA on Ranks, H = 134.681, DF = 5, p < 0.001, Dunn’s post hoc test, ∗∗∗p < 0.001). (H) Amplitude change average and single cell data points of GABA-evoked currents obtained by voltage-clamp recordings of 12–20 DIV hippocampal mouse neurons before (gray example recordings in the inset above) and after (black example traces) bath application of the indicated drugs (10 μM). Data are presented as mean ± SEM (Paired t test, ∗p 

Figure 2

ARN23746 Restores [Cl−]i in DS Neurons, Does Not Inhibit NKCC2 in HEK Cells and Shows Improved Brain Penetration In Vivo in Comparison to Bumetanide (A) Left, representative pseudo-color images (colored scale below) of the [Cl−]i measured with the MQAE Cl−-sensitive dye, in WT and Ts65Dn hippocampal neurons at 15 DIV, after treatment with control DMSO (0.1%) and the indicated compounds (10 μM). Scale bar: 70 μm. Right, quantification of the indicated compounds (10 μM) in modulating [Cl−]i in experiments such as those on the left. Data represent mean ± SEM from three independent experiments (two-way ANOVA, Finteraction (2,30)= 3.815, p = 0.033, Tukey’s post hoc test, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). (B) Left, example traces obtained in the Cl− influx assay on NKCC2-transfected HEK293 cells for each compound (10 μM). The arrow indicates the addition of NaCl (74 mM) to initiate the NKCC2-mediated Cl− influx. Right, quantification of the NKCC2 inhibitory activity in experiments such as those on the right. Data are presented as a percentage of the respective control DMSO. Data represent mean ± SEM from four independent experiments (Kruskal-Wallis one way ANOVA on Ranks, H = 16,962, DF = 2, p < 0.001, Dunn’s post hoc test, ∗∗∗p < 0.001). (C) Left, example traces obtained in the Tl influx assay on KCC2-transfected HEK293 cells for each compound (10 μM). The arrows indicate the additions of TlSO4 and NaCl. Right, quantification of the KCC2 inhibitory activity in experiments such as those on the left. Data are presented as a percentage of the respective control DMSO. Data represent mean ± SEM from 4 independent experiments (one way ANOVA, F (2,51) = 10.676, p < 0.001, Dunnett's post hoc test ∗∗∗p < 0.001). (D) Quantification of the level of bumetanide and ARN23746 in the brain at diverse time points (5, 15, 30, 60, 120 min) after injection in C57BL/6N mice (BioLASCO Taiwan). (Two-way ANOVA, Ftreatment (1,50) = 6.510, p = 0.014, Tukey’s post hoc test, ∗p < 0.05.). (E) Quantification of the ratio between brain and plasma concentration of bumetanide and ARN2346 5 min after the injection (two-tailed t test, t = 7.915, p 

Figure 3

ARN23746 Rescues Cognitive Impairment in the Ts65Dn Mouse Model of Down Syndrome with no Diuretic Effect (A) Schematic cartoon of the experimental protocol for the treatment of adult WT and Ts65Dn mice to assess the diuretic effect. (B) Quantification of the mean ± SEM and single animal cases of the urine volume collected for 120 min after mice were treated with the indicated drugs (two-way ANOVA, Ftreatment (2,69) = 10.516, p < 0.001,Tukey’s post hoc test, ∗p < 0.05, ∗∗p < 0.01). (C) Schematic cartoon of the experimental protocol for the treatment of adult WT and Ts65Dn mice with ARN23746 for in vivo efficacy assessment of improved cognitive impairment in DS mice. (D) Top, schematic representation of the T-maze test. Bottom, quantification of the mean ± SEM and single animal cases of correct choices in mice treated with the indicated drugs (two-way ANOVA, Finteraction (1,46) = 5.475, p = 0.024, Tukey’s post hoc test, ∗∗p < 0.01, ∗∗∗p < 0.001). (E) Top, schematic representation of the OL test. Bottom, quantification of the mean ± SEM and single animal cases of discrimination index in mice treated with the indicated drugs (two-way ANOVA, Finteraction (1,45) = 15.523, p < 0.001, Tukey’s post hoc test; ∗∗∗p < 0.001). (F) Top, schematic representation of the NOR task. Bottom, quantification of the mean ± SEM and single animal cases of the discrimination index in mice treated with the indicated drugs (two-way ANOVA, Ftreatment (1,46)= 7.154, p = 0.010, Tukey’s post hoc test, ∗ p < 0.05, two-way). (G) Top, schematic representation of the CFC test. Bottom, quantification of the mean ± SEM and single animal cases of the freezing response in mice treated with the indicated drugs (two-way ANOVA on Ranks, Ftreatment (1,45)= 4.425, p = 0.041, Tukey’s post hoc test, ∗ p < 0.05, ∗∗p < 0.01). See also Figure S2; Table S4.

Figure 4

ARN23746 Rescues Social Deficits and Repetitive Behaviors in the Valproic Acid Mouse Model of Autism, with No Diuretic Effect (A) Schematic cartoon of the experimental protocol for the treatment of control (CTR) and VPA mice with bumetanide or ARN23746 to the assess the diuretic effect. (B) Quantification of mean ± SEM and single animal cases of the urine volume collection for 120 min after mice were treated with the indicated drugs (two-way ANOVA on Ranks, Ftreatment (2,63) = 11.635, p < 0.001, Tukey’s post hoc test, ∗p < 0.05, ∗∗p < 0.01) (C) Schematic cartoon of the experimental protocol for the treatment of young adult WT and VPA mice with ARN23746 for in vivo efficacy assessment of improved sociability and repetitive behaviors in ASD mice. (D) Top, schematic representation of the three-chamber test. Bottom left, quantification of the mean ± SEM and single animal cases of sociability index in mice treated with the indicated drugs (two-way ANOVA, Fcondition (1,42) = 9.575, p = 0.003, Tukey’s post hoc test, ∗∗p < 0.01; ∗∗∗p < 0.001). Bottom right, quantification of the mean ± SEM and single animal cases of social novelty index in mice treated with the indicated drugs (two-way ANOVA, Fcondition (1,42)= 8.195, p = 0.007, Tukey’s post hoc test, ∗p < 0.05, ∗∗p < 0.01). (E) Top, schematic representation of the male-female interaction test. Bottom quantification of the mean ± SEM and single animal cases of the interaction time in mice treated with the indicated drugs (two-way ANOVA, Ftreatment (1,42) = 6.351, p = 0.016, Tukey’s post hoc test, ∗∗p < 0.01). (F) Top, schematic representation of the marble burying test. Bottom, quantification of the mean ± SEM and single animal cases of the number of marbles buried by mice treated with the indicated drugs (two-way ANOVA, Fcondition (1,56) = 6.727, p = 0.012, Tukey’s post hoc test, ∗p < 0.05, ∗∗p < 0.01). (G) Top, schematic representation of the grooming test. Bottom quantification of the mean ± SEM and single animal cases of the grooming time for mice treated with the indicated drugs (two-way ANOVA on Ranks, Finteraction (1,59)= 5.019, p = 0.029, Tukey’s post hoc test, ∗p < 0.05, ∗∗p < 0.01). (H) Top, representative immunoblots for NKCC1 and KCC2 on extracts of membrane-enriched protein fractions from prefrontal cortices of individuals diagnosed with ASD and controls. Bottom left, quantification of KCC2 in samples from individuals with ASD in comparison to age- and sex-matched controls (two-tailed t test, t = 2.128, p = 0.0492). Bottom right, quantification of the NKCC1/KCC2 expression ratio in samples from individuals with ASD in comparison to age- and sex-matched controls (Mann-Whitney rank sum test, p =