Clioquinol as a new therapy in epilepsy: From preclinical evidence to a proof-of-concept clinical study

Abstract

Objective: Drug-resistant epilepsy (DRE) affects >25 million people worldwide and is often associated with neuroinflammation. Increasing evidence links deficiency or malfunctioning of the enzyme phosphoglycerate dehydrogenase (PHGDH), which converts 3-phosphoglycerate to generate serine and the neurotransmitter glycine, with (drug-resistant) epilepsy. Moreover, PHGDH, which is primarily expressed in astrocytes within the brain, has been identified as a critical enzyme in driving macrophage polarization toward an anti-inflammatory state. Hence, PHGDH activators may be beneficial for treating DRE by exhibiting both antiseizure and anti-inflammatory activity. The objective of this study was to identify such PHGDH activators.

Methods: We screened a drug repurposing library for PHGDH activators and assessed their antiseizure and anti-inflammatory properties using various zebrafish and mouse epilepsy models and explored the mechanistic consequences of activating PHGDH in a cell line, in astrocytes, and in zebrafish heads. Finally, we assessed the efficacy of clioquinol as add-on treatment in three severe DRE patients in a clinical open pilot proof-of-concept study.

Results: We identified haloquinolines from a drug repurposing library as potent activators of PHGDH. The most promising haloquinoline clioquinol can increase the catalytic activity of PHGDH up to 2.5-fold, thereby increasing de novo glycine biosynthesis and resulting in reduced glutamate levels. Moreover, we show that clioquinol has PHGDH-dependent antiseizure activity as well as anti-inflammatory properties in vivo using various zebrafish and mouse epilepsy models. Finally, we demonstrate the efficacy of clioquinol as add-on treatment in severe DRE patients; two patients showed a 37%-47% reduction in seizure frequency, and all three patients noted a positive impact on quality of life and seizure severity.

Significance: Increasing activity of PHGDH is a promising new approach to treat DRE.

Keywords: de novo glycine biosynthesis; mode of action; therapeutic intervention; translational neuroscience.

FIGURE 1

Phosphoglycerate dehydrogenase (PHGDH) activation by haloquinolines. (A) Schematic representation of de novo glycine/serine biosynthesis. Clioquinol (CQ) activation of PHGDH results in increased 3‐phosphoglycerate to 3‐phosphonooxypyruvate and downstream glycine synthesis, involving phosphoserine aminotransferase 1 (PSAT1)'s conversion of glutamate to α‐ketoglutarate. (B) PHGDH activity with or without 25 μmol·L⁻¹ chloroxine (CH), 25 μmol·L⁻¹ broxyquinoline (BX), or 25 μmol·L⁻¹ CQ, alone and for CQ also in combination with 10 μmol·L⁻¹ of the PHGDH inhibitor CBR‐5884. PHGDH activity was measured during 1 h with optical density (OD) = 450 nm as readout in a colorimetric assay, representing nicotinamide adenine dinucleotide (NADH) generated. Data are shown as mean area under the curve (AUC) ± SD in at least three biological repeats. Data shown are mean ± SD (n ≥ 5 in two biological repeats). Statistical differences: **p < .01, ***p < .001, ****p < .0001 by one‐way analysis of variance with Dunnett multiple comparisons test. (C) V _max (K _cat) of serine PHGDH in the presence of different concentrations of CQ. V _max was calculated based on colorimetric change due to NADH‐induced resazurin to resorufin conversion at increasing (.5–5 mmol·L⁻¹) 3‐phosphoglycerate and 100‐fold excess nicotinamide adenine dinucleotide. Data are mean ± SD of three biological repeats. AMPAR, α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid receptor; GLT1, glutamate transporter 1; GLUT1, glucose transporter 1; mGluR, metabotropic glutamatergic receptor; NMDAR, N‐methyl‐D‐aspartate receptor; PSP, phosphoserine phosphatase; SHMT, serine hydroxymethyltransferase.

FIGURE 2

Impact of clioquinol (CQ) on the phosphoglycerate dehydrogenase‐driven de novo glycine biosynthesis on 4T1 breast cancer cells. 4T1 breast cancer cells were incubated with 0 μmol·L⁻¹ (n = 9), 5 μmol·L⁻¹ (n = 9), and 7.5 μmol·L⁻¹ CQ (n = 9) in serine‐ and glycine‐free medium in three biological repeats. Intracellular levels of serine, glycine, and glutamate were measured using gas chromatography–mass spectrometry, and the fraction of glucose in serine and glycine upon ¹³C₆ glucose tracing was calculated and normalized to vehicle (0 μmol·L⁻¹) ± SD. M2 and M3 refer to the number of ¹³C‐labeled carbon atoms in glycine and serine, respectively. Statistical differences: *p < .05, **p < .01, ***p < .001 by Kruskal–Wallis test followed by Dunn multiple comparisons test (serine and glycine) and one‐way analysis of variance with Dunnett multiple comparisons test (glutamate).

FIGURE 3

Antiseizure activity of clioquinol (CQ) in ethyl ketopentenoate (EKP) and Dravet syndrome (scn1Lab ^−/−) zebrafish epilepsy models. (A, D) Locomotor activity of larvae in (A) response to EKP and (D) Dravet syndrome, with or without 1 μmol·L⁻¹ CQ. (A, D) Activity expressed as (A) mean actinteg units per 5 min ± SD during 30 min, relative to EKP‐only treatment; or (D) normalized lardist per 100 s ± SD, relative to vehicle (VHC; 1% dimethylsulfoxide). (B, C) Electrophysiological activity of larvae expressed as mean normalized power spectral density (PSD) within a 10–90 Hz frequency range per larva ± SD, relative to VHC. (E, F) Mean cumulative duration of epileptiform events (ms/10 min recording; E) and frequency of epileptiform events/10 min recording ± SD (F), relative to scn1Lab ^−/− . For each experiment, more than 10 larvae were used in at least three biological repeats. (G–I) Glutamate levels in heads of wild‐type (G) and EKP‐treated (H) larvae, and in Dravet syndrome zebrafish (I), with or without 1 μmol·L⁻¹ CQ. Results are expressed as normalized glutamate or glycine amount per 10 homogenized heads ± SD. Statistical differences: *p < .05, **p < .01, ***p < .001, ****p < .0001 by Kruskal–Wallis test followed by Dunn multiple comparisons test (A–F), Student t‐test (G, H), and one‐way analysis of variance with Tukey multiple comparisons test (I).

FIGURE 4

Antiseizure and anti‐inflammatory activity analysis of clioquinol (CQ) in the mouse 6‐Hz (44 mA) focal seizure and self‐sustained status epilepticus (SSSE) seizure models, respectively. (A) Drug‐resistant focal seizures were induced by electrical stimulation through the cornea, 60 min after intraperitoneal injection of vehicle (VHC; n = 8), CQ (5 mg/kg, n = 8), and CQ (10 mg/kg, n = 7). Mean seizure durations (±SD) are depicted. Statistical differences: *p < .05 by Kruskal–Wallis test followed by Dunn multiple comparisons test. (B, C) The hippocampal mRNA expression of anti‐inflammatory genes and phosphoglycerate dehydrogenase in CQ‐treated SHAM and SSSE animals is represented as dot plots with median represented by a line. Statistical differences: *p < .05, **p < .01, ***p < .001, ****p < .0001 by two‐way analysis of variance with Šídák multiple comparison test. ns, not significant.