Olanzapine: A potent agonist at the hM4D(Gi) DREADD amenable to clinical translation of chemogenetics

Abstract

Designer receptors exclusively activated by designer drugs (DREADDs) derived from muscarinic receptors not only are a powerful tool to test causality in basic neuroscience but also are potentially amenable to clinical translation. A major obstacle, however, is that the widely used agonist clozapine N-oxide undergoes conversion to clozapine, which penetrates the blood-brain barrier but has an unfavorable side effect profile. Perlapine has been reported to activate DREADDs at nanomolar concentrations but is not approved for use in humans by the Food and Drug Administration or the European Medicines Agency, limiting its translational potential. Here, we report that the atypical antipsychotic drug olanzapine, widely available in various formulations, is a potent agonist of the human M4 muscarinic receptor-based DREADD, facilitating clinical translation of chemogenetics to treat central nervous system diseases.

Fig. 1. Electrophysiology-based screen of hM4D(Gi) activation.

(A) Left: Representative traces of Kir3.1 and Kir3.2 currents with (+CNO 1 μM, bottom) and without (baseline, top) hM4D(Gi) agonist application. Middle: Mean current measured during the time indicated by the gray area in the left panel, plotted against holding voltage. The red line indicates the calculation of the membrane leak conductance, obtained from a linear fit between 0 and +50 mV. Right: Leak-subtracted Kir3.1/Kir3.2-mediated currents, together with a linear fit to currents at negative potentials (blue). The slope of the current-voltage relationship (k) was used for subsequent analysis of hM4D(Gi) activation. (B) Left: CZP, C21, and PLP act as potent agonists of hM4D(Gi). All data are shown normalized to CNO (1 μM) as a positive control and fitted by a Hill equation. Right: EC₅₀ of CZP, C21, and PLP (CZP: EC₅₀ = 61 ± 19 nM, n = 6; PLP: EC₅₀ = 40 ± 10 nM, n = 9; C21: EC₅₀ = 20 ± 4 nM; *P < 0.05, one-way ANOVA with Bonferroni post hoc test).

Fig. 2. Shape- and 2D similarity–based identification of hM4D(Gi) agonists.

(A) Overview of the 3D- and 2D-based virtual screen. C21 was used as the query compound for both the 2D similarity search of the ChEMBL database and the generation of a 3D shape–based model, because it showed the lowest EC₅₀ at hM4D(Gi) of known agonists. For detailed virtual screening results, see table S1. (B) hM4D(Gi)-dependent potentiation of Kir3.1/Kir3.2-mediated currents measured for selected hit compounds is normalized by 1 μM CNO as positive control. The TanimotoCombo (TC) score (for 3D screen), similarity (S) index (for 2D screen), the screening method (3D or 2D; separated by dotted line), and the tested concentration are indicated. For detailed structures of selected drugs, see table S2.

Fig. 3. OZP is a potent agonist at hM4D(Gi).

(A) Left: Dose-response curves for CZP and OZP at hM4D(Gi). The inset shows the efficacy of OZP (100 nM) and CZP (1 μM), normalized to 1 μM CNO as a positive control (CZP/CNO = 1.14 ± 0.06; n = 6; OZP/CNO = 1.01 ± 0.06; n = 6; P = 0.12, Student’s t test). Right: EC₅₀ for CZP at hM4D(Gi), and OZP at hM4D(Gi) and at the codon-optimized hM4D(Gi)_opt [CZP: hM4D(Gi) EC₅₀ = 61 ± 19 nM, n = 6; OZP: hM4D(Gi) EC₅₀ = 5 ± 2 nM, n = 6, P < 0.01 in comparison to CZP hM4D(Gi); OZP: hM4D(Gi)_opt EC₅₀ = 7 ± 2 nM, n = 6, P < 0.01 in comparison to CZP hM4D(Gi); one-way ANOVA with Bonferroni post hoc test]. ns, not significant. (B) Docking poses of OZP (left, blue sticks) in comparison to CZP (right, rose sticks) in the homology model of active hM4D(Gi) (gray cartoon and sticks). Ionic interactions with D112 and hydrogen bonds are highlighted by dotted and dashed lines, respectively. The methyl group of OZP is highlighted in the right panel. The Y113C and A203G mutations are highlighted in orange, and residues 434 to 443 are not depicted for clarity. (C) Left: Latency to fall of animals injected with AAV2/8-hCamKII-hM4D(Gi)_opt [before OZP: 149 ± 14 s; after OZP (0.1 mg/kg, i.p.): 111 ± 11 s; n = 15; **P = 0.002, paired Student’s t test], those injected with AAV2/8-hCamKII-GFP (before OZP: 180 ± 21 s; after OZP: 169 ± 25 s; n = 15), and sham-injected animals (before OZP: 212 ± 21 s; after OZP: 189 ± 20s; n = 6; **P < 0.01, Student’s paired t test; ^#P < 0.05, repeated-measures ANOVA with LSD post hoc test). Middle: Normalized latency to fall of AAV2/8-hCamKII-hM4D(Gi)_opt–injected animals treated with either OZP (0.1 mg/kg) (0.76 ± 0.06; n = 15) or CZP (0.1 mg/kg) (1.03 ± 0.08; n = 15; **P = 0.033, Student’s paired t test). Right: Representative confocal fluorescence images of mouse brains injected with either AAV2/8-hCamKII-hM4D(Gi)_opt (DREADD, −1.3 mm from bregma) or AAV2/8-hCamKII-GFP (GFP, −1.2 mm from bregma) (scale bars, 1 mm). (D) Latency to fall after OZP (0.1 mg/kg, i.p.) in animals injected with either AAV2/5-hCamKII-hM4D(Gi) or AAV2/5-hCamKII-empty bilaterally into striatum [AAV2/5-hCamKII-hM4D(Gi), 19 ± 4 s; n = 6; AAV2/5-hCamKII-empty, 44 ± 8 s; n = 6; P = 0.014, Student’s unpaired t test].