The Design, Synthesis and Evaluation of Novel CB2 Ligands

The pathophysiology of brain injury and a variety of neurodegenerative diseases involves an inflammation cascade that causes neuronal loss and damage in the central nervous system (CNS) over time. A diagnostic tool that could allow earlier detection and intervention of neuroinflammation is currently missing. Under inflammatory conditions, an increase in the expression of cannabinoid receptor 2 (CB2) is observed in microglia that are the resident macrophages in CNS. Thus, CB2 represents a potential imaging biomarker and a therapeutic target for neuroinflammatory diseases. The aim of this work was to develop specific 18F-radioligands (T1/2 = 109.7 min) targeted at CB2 for the detection of activated microglia in vivo by positron emission tomography (PET). The first compound series consisted of fluorinated benzimidazole 5-sulfones. Several derivatives were found to show low nanomolar CB2 potency and excellent selectivity over CB1 subtype. Selected ligands with high CB2 binding affinity were radiolabelled and isolated with non-decay corrected (ndc) radiochemical yield (rcy) of 2040%, with high radiochemical purity (>95%), molar activity of 200 GBq/mol and good formulation stability. The radioligand candidates were observed to show low brain uptake in healthy rats, as expected in baseline conditions. The second series comprised of fluorinated azaindoles that were found to display restricted structure-activity relationship, with few analogues showing low nanomolar CB2 activity and sufficient subtype selectivity. The first ligand candidate was labelled using Cu-mediated 18F-fluorination with 10% rcy (ndc), high radiochemical purity (>95%) and formulation stability. The radioligand showed rapid uptake in CB2-rich spleen and had promising uptake in healthy rat brain. To summarise, the compounds investigated showed excellent in vitro and promising in vivo properties. As the next step, the CB2-specificity will be studied in greater detail in suitable neuroinflammation models