Direct competition binding between a vesicle glycoprotein 2A (SV2A) PET ligand and a γ-secretase-modulator tool compound

Introduction: We investigated the mechanism of action of our first potent, orally bioavailable, brain penetrable γ-secretase-modulator (GSM) tool compound RO7019009 in a mouse model for amyloidosis. To examine if modulation of γ-secretase not only reduces the load of Aβ deposition as major hallmark of Alzheimer’s Disease (AD) but also rescues synaptic loss, both APPSL70 and C57Bl/6 mice were scanned with the synaptic vesicle glycoprotein 2A (SV2A) tracer [¹⁸F]-UCB-H. SV2A is an abundant protein in the synaptic vesicles across the nervous system. Thus, [¹⁸F]-UCB-H is considered as a tracer candidate for quantifying the synaptic density in preclinical and clinical applications. However, we found surprising results suggesting an interference of the GSM and [¹⁸F]-UCB-H-PET signals. This finding led to subsequent experiments to investigate further the interaction of GSM RO7019009 and the tracer.

Methods: 57 female APPSL70 and 36 C57Bl/6 mice treated with the GSM for six months starting at an age of six months were scanned at six, nine, and twelve months of age with [¹⁸F]-UCB-H. Volume of distribution (Vt) images were calculated by using the blood input curve of the heart. Mixed effects models were used to determine the effects of the therapy. In vitro binding was used to test a possible direct interaction between the GSM study drug and [³H]-UCB-J, a close analog of UCB-H. Furthermore, we probed the possible direct interaction between the GSM study drug and the [¹⁸F]-UCB-H tracer in C57Bl/6 mice receiving a short-term 48 h application of the GSM study drug.

Results: We encountered an unexpected decrease in the SV2A-PET Vt images in vivo, detectable at nine months of age in both C57Bl/6 (p = 0.0365) and APPSL70 (p = 0.0005) animals treated with the GSM study drug compared to the animals receiving placebo, as well as at twelve months of age in both C57Bl/6 (p = 0.0037) and APPSL70 (p = 0.0005) animals. The subsequent in vitro radioligand competition binding experiments performed on C57BL/6 cortex membranes revealed a direct GSM-tracer interaction of the GSM study drug with the SV2A radiotracer with an IC₅₀ of 950 ng/ml; for reference, the GSM drug exposure achieved in the chronic in vivo study was 593.9 ng/ml ± 156.9 ng/ml. This suggests a possible relevant interference of the study drug with the SV2A µPET readouts. An acute treatment of C57Bl/6 mice also showed a significant reduction of Vt in a SV2A-PET study (p = 0.0112). Of note, quantification of SV2A protein in brain lysates of mice treated with RO7019009 revealed no significant treatment-related changes of SV2A.

Conclusions: This study demonstrates competitive binding of the GSM study drug with the SV2A radioligand [³H]-UCB-J in vitro. This direct interaction between drug candidate and tracer most probably explains the reduction of [¹⁸F]-UCB-H Vt values in an in vivo PET study in mice which were treated with the GSM tool compound. The results highlight the importance of considering a direct interaction between study drugs and tracers even for highly refined compounds.