The design strategy for our probes encompassed the structural integration of a pharmacophore moiety with a fluorescent reporter unit, as illustrated in Scheme 1. This approach yielded probes of significantly reduced size, adhering to all established criteria of drug-likeness.^39,40 These criteria included a low molecular weight (MW), a balanced number of hydrogen bond donors and acceptors (HBD, HBA), low lipophilicity (clogP), a minimal number of rotatable bonds (nRotB), and a small topological polar surface area (tPSA) (see Table 1 and SI, Table S2). These are essential characteristics to facilitate membrane permeation,⁴¹ reduce nonspecific protein binding, and improve solubility and overall bioavailability.^39,40 These physicochemical properties are particularly important for addressing intracellular enzymes, such as MAGL.

Previously published⁴²⁻⁴⁶ structure–activity relationship (SAR) studies of MAGL inhibitors and structural data provide insights into the capacity of the flexible lipophilic binding pocket to harbor various apolar structural motifs. We conducted ligand superimposition and docking studies, which revealed that replacing the aromatic pharmacophore structures in MAGL inhibitors with a fluorescent BODIPY moiety could retain the key interactions toward MAGL. This led to the construction of structures with BODIPY connected to the cyclic amine moieties as a versatile linchpin for the development of reversible and irreversible MAGL probes (Schemes 2 and 3). We combined this with several privileged noncovalent amphiphilic headgroups^43,47−49 that bind to the glycerol-binding subpocket. Additionally, for covalent MAGL inhibitors, we used reactive carbamate structures with fluorinated alcohols as leaving groups that targeted catalytic Ser122.^27,50 This approach is a modification of the reverse design principle that uses high-quality drug ligands or components thereof in the development of chemical tool compounds.⁵¹

In order to achieve the desired BODIPY fluorescent probes, we established a comprehensive modular strategy for the synthesis of 8-alkylene connected constructs. This strategy employs a Liebeskind–Srogl cross-coupling (LSCC)⁵² between vinyl boronic acids and 8-thiomethyl-BODIPY building blocks as the central step (Scheme 2). The synthesis commenced with the construction of vinyl boronic acids obtained via Boron-Wittig olefination to form the corresponding vinyl boronates.^53,54 They were connected to their respective headgroups after N-Boc-deprotection via urea, amide, or carbamate formation. This was followed by the deprotection (if applicable) and reduction of the vinylic double bond under mild conditions. Interestingly, the presence of the conjugated alkene moiety caused a negligible quantum yield of the BODIPY unit while exhibiting high MAGL inhibitory potencies (SI, Table S1: S11, S16, S21).⁵² Hydrogenation of the double bond using triethylsilane and Pd/C in methanol was crucial for liberating the bright fluorescent properties of the BODIPY unit. Compounds 1 and 1a required an additional PMB-deprotection step (Scheme 2B), which was achieved using DDQ in DCM/H₂O. Direct C–H arylation at the 3-position of the 8-thiomethyl BODIPY building block was achieved using in situ-generated benzene diazonium. Functionalization at the 3-position with a pyrrol-2-yl moiety led to a strong bathochromic shift of the fluorophore. The desired building block was obtained in one step by microwave-assisted oxidative nucleophilic substitution with neat pyrrole.⁵⁵

Restraining conformational flexibility to the active conformation by spirocyclic moieties was found to be an effective strategy to enhance affinity (see SI, Table S1, S1 vs S5–S6)^56,57 and led to the discovery of the [3.3]-spirocyclic BODIPY motif in 1. The modular construction design allowed for successive variations of the headgroup by several other structural distinct moieties (Scheme 2C).

We found that the replacement of the headgroup interacting with the amphiphilic subpocket of the active site of MAGL occurs in a modular and independent manner (Scheme 2C). This phenomenon is attributed to the stabilization of the binding conformation by the central carbonyl forming a bond with the oxyanion hole of the serine hydrolase.⁴⁵ The most potent and promising compounds (Scheme 3) underwent comprehensive characterization, encompassing biochemical activity, selectivity, cellular potency, drug-likeness, and physicochemical properties (Table 1), as well as photophysical properties (SI Table S4). All compounds depicted in Scheme 3 exhibited subnanomolar inhibition of human MAGL (hMAGL) in a RapidFire mass spectrometric (MS) assay measuring the hydrolysis of the native 2-AG substrate.

Interestingly, substitution of the BODIPY unit at the 3-position by phenyl (SI, Table S1: S22) or pyrrol-2-yl (1a, 2a, 4a, 5a) moieties was well-tolerated, with at least on-par inhibitory potency.

Large species-dependent activity differences of MAGL inhibitors are often hampering the translation of preclinical data to clinical development. For instance, the widely used MAGL tool compound, JZL184, showed severely reduced potency on rat MAGL, complicating investigations in rodents.^12,61 Therefore, we tested our probes’ inhibitory activity against human, mouse, rat, and cynomolgus monkey MAGL orthologs to examine their applicability in translational research using the RapidFire MS native substrate assay monitoring the direct conversion of the endogenous substrate 2-AG.²⁶ As illustrated in Table 1, all probes exhibited consistently high potency against MAGL across all tested species. Only 3, 4, and 5a showed minor (less than 4-fold) losses against rat or cynomolgus monkey orthologs. To assess cellular target engagement we performed a nanoluciferase (Nluc)-based bioluminescence resonance energy transfer (nanoBRET) assay using live HEK293 cells.⁶³ Here, 5 showed the best cellular potency, with an IC₅₀ of 85 nM, while 1–4 showed activity in the 200–700 nM range, indicating a high affinity for MAGL and sufficient permeability. The observed variations in cellular target occupancy (i.e., between 2 and 5) could potentially originate from the complex interplay of additional factors, such as cellular permeability, (serum)protein binding, and membrane accumulation of the probe. Consistent with these results, all inhibitors meet Lipinski’s rule of five³⁹ and the “Veber rules”⁴⁰ and showed favorable tPSA and solubility (Table 1). In addition, the choice of the headgroup and substitution of the BODIPY fluorophore significantly influenced these physicochemical properties. For reference, the only published selective MAGL fluorescent labeling probe, LEI-463, significantly exceeded these descriptor boundaries (MW = 1087 g/mol, clogP = 9.28, HBA/HBD = 6/1, tPSA = 108 Ų, and nRotB = 22). The probes’ physicochemical descriptors, computed lipophilicities, and solubilities are similar to highly optimized MAGL inhibitor drug candidates (SI, Figure S2 and Table S2).

We confirmed both, the reversibility of the urea-containing compounds 4 and 5 through a surface plasmon resonance (SPR) experiment (SI, Figure S13), as well as the covalency of probe 1 through mass spectrometry analysis of labeled MAGL protein (SI, Figure S14). The residence time of 4 and 5 on hMAGL was determined as 37 and 44 min, respectively.