Mechanism of Carbamate Inactivation of FAAH: Implications for the Design of Covalent Inhibitors and In Vivo Functional Probes for Enzymes

Characterization of the Mechanism of FAAH Inhibition by Carbamates

Several carbamate inhibitors of FAAH have been described in both the scientific [20] and patent [29] literature. The provocative behavioral effects of these agents, including reductions in pain sensation [19, 29] and anxiety [20], have provided support for FAAH as a potential therapeutic target. Nonetheless, the mechanism by which carbamates inhibit FAAH remains obscure, thus complicating efforts to rationally design second-generation agents with superior potency and selectivity. Carbamates typically inhibit serine hydrolases by an irreversible (or slowly reversible) mechanism involving carbamylation of the serine nucleophile [30]. Initial modeling studies with the carbamates URB597 and URB532 suggested that their O-biaryl substituents reside in the acyl chain binding (ACB) channel of FAAH, serving as a more rigid structural mimetic of the arachidonoyl chain of the FAAH substrate anandamide [28] (Figure 1A). However, this orientation of binding would place the N-alkyl substituents of carbamates in the cytoplasmic access (CA) channel of FAAH in positions more suitable for serving as leaving groups (akin to the ethanolamine of anandamide), possibly resulting in carbonylation of the enzyme’s nucleophile (Figure 2A). More recent chemical reactivity studies have raised questions about this model, as the extent of electrospray ionization-induced cleavage of the C(O)-O bond in a series of FAAH carbamate inhibitors was found to correlate with their potency of inhibition [31], suggesting that O-aryl esters may serve as leaving groups in FAAH-carbamate reactions (carbamylation, Figure 2A). Finally, both of the aforementioned models assume that carbamates inhibit FAAH through a covalent mechanism of action, a premise that has not yet been directly tested.

To examine the mechanism of carbamate inhibition of FAAH, purified recombinant enzyme was treated with URB532 or URB597 for 1 hr, after which the protein samples were subjected to tryptic digestion and MALDI-TOF mass spectrometry (MS) analysis. In each case, a new mass signal was observed for the inhibitor-treated sample compared to untreated FAAH controls that corresponded to the mass of the FAAH tryptic peptide amino acids (AA) 217-243, which contain the FAAH nucleophile S241 [32], modified by the C(O)-N-alkyl substituent of the inhibitor (Figures 2B-2D). No evidence of C(O)-O aryl modification of FAAH was detected in either the URB532 or URB597 reaction. Liquid chromatography-tandem electrospray MS (LC-MS) analysis assigned the specific site of carbamylation to S241 with high confidence (Xcorr = 6.04; ΔCN = 0.54).

These MS results indicated that URB532 and URB597 inhibit FAAH by carbamylation of the enzyme’s serine nucleophile. As a corollary, the O-biaryl substituents of these agents would be expected to reside in the CA (rather than the ACB) channel of the FAAH active site, where they would be susceptible to enzyme-catalyzed protonation to enhance their function as leaving groups[21, 33] (Figure 1B). To further test this prediction, we synthesized a series of URB597 derivatives in which the N-cyclohexyl substituent was replaced with various N-alkyl groups that mimicked the structures of acyl chains of known FAAH substrates/inhibitors. These carbamates, which included N-oleyl (JP23) and N-(6-phenyl)hexyl (JP83) analogs, were found to inhibit FAAH with equal or greater potency than URB597 (Table 1). These data indicate that the URB597 O-biaryl substituent can be accommodated in the CA channel of the FAAH active site. Indeed, it appears that this group displays positive binding interactions with the CA channel, as its replacement with a phenyl moiety (JP87) significantly reduced inhibitor potency (Table 1).

Generation of a Click Chemistry Probe to Evaluate the Proteome-Wide Reactivity of FAAH-Directed Carbamates In Vivo

Covalent inhibitors, such as carbamates, are generally regarded with skepticism as potential drugs due to their inherent reactivity [34], which is perceived to compromise target selectivity in vivo. Nonetheless, this assertion often goes untested, as general methods to determine the in vivo target selectivity of covalent inhibitors are lacking. The discovery that URB597 inhibited FAAH via covalent carbamylation of the enzyme’s active site inspired us to synthesize an N-alkynyl derivative of this agent, JP104, which we hypothesized could be used to directly identify the targets of FAAH-directed carbamates in vivo. In this approach, tissues from animals treated with JP104 are reacted with a complementary azide-modified reporter tag (e.g., azido-rhodamine [RhN₃], azido-biotin [35, 36]) under CC conditions to afford the corresponding triazole products (Figure 3A). Initial in vitro experiments revealed that JP104 was a potent FAAH inhibitor in vitro (Table 1) and selectively labeled this enzyme in brain membrane extracts (Figure S1A; see the Supplemental Data available with this article online). To further confirm the specificity of JP104 for FAAH, we treated the JP104-labeled brain proteome with a rhodamine-tagged fluorophosphonate (FP-Rh), which serves as a general activity-based protein profiling (ABPP) probe for the serine hydrolase superfamily[37-39]. JP104 completely blocked FP-Rh labeling of FAAH at 100 nM, but it did not inhibit the labeling of any other brain serine hydrolases at concentrations up to 10 μM (Figure S1B).

We next tested whether click chemistry (CC) methods could be used to identify the targets of JP104 in vivo. Mice were treated with JP104 (10 mg/kg, intraperitoneal [i.p.]) for 1 hr, after which they were sacrificed and their tissues removed for CC analysis. Consistent with in vitro studies, FAAH was the primary in vivo target of JP104 observed in brain tissue (Figure 3B, left panel). In contrast, several additional targets of JP104, including multiple membrane-associated liver and kidney proteins, were detected in peripheral tissues (Figure 3B, right panels). These JP104-labeled proteins were also observed in in vivo-labeled, soluble proteomes from kidney and liver (Figure S2), as well as in FAAH^-/- tissues (Figure 3C), indicating that they do not represent FAAH or potential FAAH isoforms. Dose-response profiles identified a concentration range for JP104 (0.25-1.0 mg/kg) over which significant FAAH inhibition could be achieved without extensive modification of additional targets (Figure 4). For example, at 1 mg/kg JP104, FAAH labeling was ∼80% of maximum in the brain, whereas none of the liver and kidney targets were modified to greater than 20%. The nearly complete inactivation of brain FAAH in vivo by 1 mg/kg JP104 was confirmed by competitive ABPP studies with FP-Rh (Figure 3B, left panel). In contrast, JP104 did not significantly reduce the intensity of FP-Rh signals in liver and kidney proteomes (Figure 3B, right panels), which likely reflects the presence of several comigrating serine hydrolase activities in the 60-65 kDa range in these tissues (only a subset of which are modified by JP104). These results indicate that carbamate-based inhibitors can be designed that selectively target FAAH in vivo, albeit over a somewhat limited concentration range.

Identification and Characterization of In Vivo Targets of JP104

The molecular characterization of off-target sites of reactivity for FAAH-directed carbamates would greatly assist medicinal chemistry efforts aimed at improving the selectivity of these agents. With this goal in mind, we reacted liver cytoplasmic proteomes from JP104-treated mice (10 mg/kg, i.p.) with an azide-modified biotin tag under CC conditions. Biotinylated proteins were then enriched by avidin chromatography, separated by SDS-PAGE, and subjected to trypsin digestion and LCMS analysis. In the mass range of 60-65 kDa, two candidate targets of JP104—carboxylesterase 6 (CE6) and esterase 31 (Est31)—were identified. These enzymes were confirmed as specific targets of JP104 by recombinant expression in COS7 cells (Figure 5A). Using competitive ABPP [39], in which transfected COS7 cell proteomes were first pretreated with varying concentrations of JP104 (0.01-10 μM) and then incubated with FPRh (100 nM), we estimated IC₅₀ values of 0.05 and 2 μM for CE6 and Est31, respectively (Figure 5B and Table S1). CE6 was also inhibited by URB597 with similar potency (IC₅₀ = 0.2 μM, Figure 5B), while Est31 was not labeled by this carbamate in vitro (Table S1). To test whether CE6 was targeted by URB597 in vivo, mice were treated with this inhibitor (10 mg/kg, i.p.) for 1 hr, and the JP104-labeling profiles of excised tissues were compared to those of vehicle-treated animals. In addition to blocking JP104 labeling of FAAH in brain, URB597 significantly reduced the labeling of the 60-65 kDa CE6/Est31 bands in liver (Figure 5C). Given that only CE6 is sensitive to inhibition by URB597 in vitro, we speculate that the labeling signals observed for the 60-65 kDa bands in vivo are predominantly derived from this enzyme. Finally, dose-response studies with URB597 revealed that this inhibitor and JP104 displayed similar efficacies for inhibiting FAAH (Figure S3), indicating that URB597 also shows a moderate (5- to 10-fold) range of selectivity for this enzyme over CE6 in vivo.

Synthesis of Inhibitors

See Supplemental Data.

MS Analysis of Carbamate Labeling Site of FAAH

Inhibitor labeling reactions were conducted at room temperature with 10 μM recombinant purified FAAH (∼30 μg) prepared as described previously [45] and 200 μM URB597 or URB532 in 125 mM Tris$HCl (pH 9.0), 0.01% Triton X-100 in a total volume of 50 μl. Any noncovalent interactions between probe and enzyme were disrupted by gel filtration through a Biorad 10DG column, and the desired fractions were concentrated to 50 μl by using an Amicon Ultra centrifugal filter device, MWCO 10,000. Samples were then denatured in 8 M urea and subjected to reduction, alkylation, and tryptic digestion according to established procedures [46]. Samples (50 μl after concentration under reduced pressure) were desalted by C₁₈ ZipTip (Millipore) eluting in 4 μl (3:1 acetonitrile/water with 0.1% trifluoroacetic acid) and were analyzed by MALDI-TOF Reflectron mass spectrometry (Applied Biosystems Voyager STR) with 1:1 matrix α-cyano-4-hydroxycinnamic acid:sample (1 μl total volume). For tandem MS analysis, recombinant purified FAAH (∼30 mg) in 125 mM Tris$HCl (pH 9) was reacted with URB597 as described above and was then directly subjected to the tryptic digestion protocol. The resulting peptide mixture (50 μl after concentration under reduced pressure) was analyzed by reverse-phase liquid chromatography-electrospray ionization tandem MS (Finnigan LCQ Deca ion trap MS; Thermo Finnigan). The gradient elution began with 100% buffer A (95:5 vol/vol water/acetonitrile containing 0.1% formic acid) for 10 min, followed by an increase to 15% buffer B (80:20 vol/vol acetonitrile/water containing 0.1% formic acid) in 5 min. The gradient increased linearly to 50% buffer B in 75 min, followed by an increase to 100% buffer B in 10 min, and ending with 100% buffer B for 10 min. Tandem mass spectra were searched against a database containing the rat FAAH sequence by using a modified version of the SEQUEST algorithm. Search results were filtered and grouped with DTASelect [47].

Preparation of Mouse Tissue Proteomes for CC Analysis

Mouse tissues were Dounce-homogenized in 10 mM sodium/potassium phosphate buffer (pH 8.0) (PB), followed by a low-speed spin (890 × g) to remove debris. The supernatant was then subjected to centrifugation at 100,000 × g for 45 min to provide the cytosolic fraction in the supernatant and the membrane fraction as a pellet, which was washed and resuspended in PB buffer by sonication. The total protein concentration in each fraction was determined by a protein assay kit (Bio-Rad). Samples were stored at -80°C until use.

In Vitro Analysis of Inhibitor Potency

Inhibitor analysis was carried out in vitro by using two different as-says. First, a radioactive substrate assay was employed by using purified, recombinant FAAH as described previously [32, 45]. Briefly, FAAH activity was determined in the presence of varying concentrations of inhibitors (0.01-100 μM, 50× DMSO stocks) by following the conversion of [¹⁴C]oleamide to oleic acid by thin-layer chromatography. Inhibitor potency was also examined by using competitive ABPP as described previously [39]. Briefly, membrane proteomes of mouse tissues were diluted to 1 mg/ml and preincubated with varying concentrations of inhibitors (10 nM-100 μM, 50× DMSO stocks) for 10 min at room temperature prior to the addition of a rhodamine-tagged fluorophosphonate (FP-Rh; [38]) ABPP probe at a final concentration of 100 nM in a 50 μl total reaction volume. Reactions were quenched after 10 min with one volume 2× SDS loading buffer (reducing with protease inhibitor), run on SDS-PAGE, and visualized with a flatbed fluorescence scanner. Dose-response curves obtained from both methods from three trials at each inhibitor concentration were fit with Prism software (GraphPad) to obtain IC₅₀ values with 95% confidence intervals.

In Vitro CC Analysis of Proteins with JP104

Proteome samples (43 μl of 1 mg/ml protein in PB) were treated with 100 nM JP104 (unless otherwise indicated) (1 μl of a 5 μM DMSO stock) for 1 hr at room temperature, followed by CC reaction with 25 μM rhodamine-azide (RhN₃, 1 μl of a 1.25 mM DMSO stock) as previously described [35, 36]. Samples were analyzed separately by competitive ABPP with 1 μM FP-Rh instead of the CC reaction with RhN₃. Reactions were quenched with an equal volume (50 μl) of 2× SDS loading buffer (reducing with protease inhibitor), run on SDS-PAGE, and visualized by in-gel fluorescence scanning with a Hitachi FMBio IIe flatbed laser-induced fluorescence scanner (MiraiBio, Alameda, CA).

In Vivo Labeling in Mice

Mice (male C57BL/6 or FAAH^-/- [5]), weighed between 20 and 30 g and were between the ages of 12 and 18 weeks. Mice were given intraperitoneal (i.p.) injections of JP104 (0, 0.25, 1, and 10 mg/kg) or URB597 (0, 0.1, 0.2, 1, and 10 mg/kg) in vehicle (18:1:1 saline:emulphor:EtOH). The mock injections (0 mg/kg) consisted solely of vehicle. After 1 hr, the mice were sacrificed by CO₂ asphyxiation and decapitation, and relevant tissues were flash frozen (with liquid N₂) immediately upon removal. Tissues were processed as described above to isolate membrane and cytosolic proteomes. The CC reaction and/or FP-Rh labeling were carried out as described above, and reactions were analyzed by SDS-PAGE. The average of three independent samples at each concentration was reported.

Protein Isolation and Identification from Mouse Liver

To identify off-target sites of reactivity in the liver, trifunctional probes with biotin, Rh, and an azide [36] were used to affinity purify in vivo-labeled enzymes for identification by mass spectrometry techniques. The soluble liver proteome (2.5 ml, 3 mg/ml protein) from a mouse treated with JP104 (10 mg/kg, i.p.) was subjected to PD-10 size-exclusion chromatography to remove excess probe and was then fractionated by Q-Sepharose chromatography (0-2 M NaCl gradient). Fractions containing desired material were combined, and a 480 μl aliquot was subjected to CC reaction with the tri-functional azide probe. The excess CC reagents were removed in a series of washes as previously described [36]. Tagged proteins were enriched by using avidin-agarose beads, and the affinity-isolated proteins were separated by SDS-PAGE as previously described [36]. Gel bands were identified by in-gel fluorescence scanning, excised, and subject to in-gel trypsin digest [36]. The resulting peptides were analyzed by μlC-MS/MS, and the results were searched against public databases by using SEQUEST to identify probe-labeled proteins as carboxylesterase 6 (CE6; GI♯ 20071857) and esterase 31 (Est31; GI♯ 2494382). These off-target proteins were further verified by purchasing expressed sequence tags (Open Biosystems), sequencing, and transient transfection into COS7 cells as described previously [39]. Cytosolic fractions from transfected cells were isolated by centrifugation and were analyzed in vitro as described above to confirm labeling with JP104. IC₅₀ values were determined via competitive ABPP as described previously [39].