Peroxide-Dependent MGL Sulfenylation Regulates 2-AG-Mediated Endocannabinoid Signaling in Brain Neurons

SUMMARY The second messenger hydrogen peroxide transduces changes in cellular redox state by reversibly oxidizing protein cysteine residues to sulfenic acid. This signaling event regulates many cellular processes, but has been never shown to occur in the brain. Here we report that hydrogen peroxide heightens endocannabinoid signaling in brain neurons through sulfenylation of cysteines C201 and C208 in monoacylglycerol lipase (MGL), a serine hydrolase that deactivates the endocannabinoid 2-arachidonoyl-sn -glycerol (2-AG) in nerve terminals. The results suggest that MGL sulfenylation may provide a presynaptic control point for 2-AG-mediated endocannabinoid signaling.


INTRODUCTION
Activation of metabotropic glutamate receptor-5 (mGluR 5 ) in dendritic spines stimulates release of the lipid-derived endocannabinoid transmitter, 2-AG, which crosses the synapse in a retrograde direction to engage CB 1 cannabinoid receptors on axon terminals and suppress glutamate release (Castillo et al., 2012).This negative feedback mechanism is thought to modulate excitatory synaptic transmission throughout the central nervous system (CNS) (Katona and Freund, 2008).Along with mGluR 5 , another checkpoint for endocannabinoid signaling may be provided by MGL, a presynaptic serine hydrolase that cleaves 2-AG into arachidonic acid and glycerol, terminating its biological actions (Dinh et al., 2002;Gulyas et al, 2004).Genetic and pharmacological interventions that either enhance (Jung et al, 2012) or interrupt (Hohmann et al., 2005;Long et al, 2009) MGL activity are known to affect the pool of bioactive 2-AG, but whether posttranslational MGL modifications influence 2-AG signaling remains undetermined.
The catalytic site of MGL is partially covered by a flexible 'cap' domain that controls the entry of substrates into the enzyme's active site (Labar et al, 2010;Bertrand et al, 2010;Schalk-Hihi et al, 2011)(Figure 1A).This domain harbors two cysteine residues (C201 and C208) that mediate the interaction of MGL with thiol-reacting inhibitors such as benzisothiazolinones (King et al, 2009) and N-acyl-substituted maleimides (Saario et al, 2005).A third cysteine (C242) located within the active site, in close proximity of the catalytic nucleophile S122, directly influences catalysis (Labar et al, 2010;Saario et al, 2005).These residues might be targeted by the second messenger, hydrogen peroxide (H 2 O 2 ), which modifies protein function (Patel and Rice, 2012) through reversible oxidation of cysteine's thiol groups (-SH) to sulfenic acid (-SOH) (Dickinson and Chang, 2011;Lo Conte and Carroll, 2013).Because ionotropic glutamate receptors stimulate H 2 O 2 production (Patel and Rice, 2012) and H 2 O 2 influences the activities of several functionally important neuronal proteins (Rice, 2011), we hypothesized that peroxide-dependent sulfenylation of the regulatory cysteines, C201 and C208, might inhibit MGL and, by doing so, heighten 2-AG-mediated endocannabinoid signaling.If confirmed experimentally, this mechanism would provide a presynaptic control point for 2-AG-mediated endocannabinoid signaling at central excitatory synapses.

Hydrogen peroxide inhibits MGL
The flexible 'cap' domain of MGL, which contains the regulatory cysteines C201 and C208, controls the entry of substrates into the enzyme's active site (Labar et al, 2010;Bertrand et al, 2010;Schalk-Hihi et al, 2011) (Figure 1A).We hypothesized that peroxide-dependent sulfenylation of these two residues might inhibit MGL and thus heighten 2-AG-mediated signaling.As a first test of this idea, we exposed recombinant rat MGL to concentrations of H 2 O 2 that are known to reversibly affect protein function without causing oxidative damage (Rice, 2011).H 2 O 2 inhibited MGL with a median effective concentration (IC 50 ) of 8.2 ± 1.3 μM (Figure 1B).The inhibition was rapid (Figure 1C), displayed noncompetitive Michaelis-Menten kinetics (Figure 1D), and was fully reversible upon enzyme dilution (Figure 1E).By contrast, even at very high concentrations, H 2 O 2 did not affect the activity of recombinant diacylglycerol lipase-α (DGL-α) (Figure 1B), the serine hydrolase that catalyzes mGluR 5operated 2-AG formation in spines (Castillo et al., 2012;Katona and Freund, 2008).

Role of regulatory cysteines C201 and C208
We used site-directed mutagenesis to determine whether the three regulatory cysteine residues present in MGL -C201, C208 and C242 -contribute to peroxide-dependent enzyme inhibition.Replacing either C201 or C208 with alanine caused a rightward shift in the potency of H 2 O 2 (Figure 1B and Table S1).Importantly, an even more pronounced shift was observed when both C201 and C208 were mutated (Figure 1B and Table S1), which is suggestive of a cooperative interaction between the two residues.By contrast, replacing C242 with alanine or serine had no effect on H 2 O 2 -induced inhibition of MGL activity (Figure 1F).The inhibitory actions of H 2 O 2 were also apparent when the oxidant was applied on intact mouse Neuro-2a cells.In these experiments, we used H 2 O 2 at concentrations of 30-300 μM, in consideration of the fact that only 1-10% of externally applied H 2 O 2 crosses the cell membrane (Rice, 2011).In native Neuro-2a cells, H 2 O 2 inhibited MGL activity in a concentration-dependent manner (Figure S1A) and this effect was paralleled by an increase in intracellular 2-AG content (Figure S1B).In Neuro-2a cells that overexpressed the C201/C208 double mutant form of MGL, H 2 O 2 only weakly inhibited MGL activity (Figure S1C) and did not increase 2-AG levels (Figure S1D).Together, these findings indicate that H 2 O 2 inhibits MGL at concentrations similar to those demonstrated to be active on other biologically relevant targets, such as protein tyrosine phosphatase 1B (Denu and Tanner, 1998) and endothelial growth factor receptors (Paulsen, 2011).Inhibition occurs through an allosteric mechanism that involves the redox-sensitive cysteines C201 and C208, which respond cooperatively to H 2 O 2 -induced oxidation.

Hydrogen peroxide causes C201 and C208 sulfenylation
To determine whether MGL is a substrate for cysteine sulfenylation, we incubated the purified recombinant enzyme with H 2 O 2 in the presence of dimedone -a chemoselective probe that reacts with short-lived sulfenyl groups in cysteines to form stable thioethers (Figure 2A, inset) (Pan and Carroll, 2014) -and then digested the protein with trypsin.Liquid chromatography/mass spectrometry (LC/MS) analyses of the tryptic digest showed that dimedone was exclusively bound to the peptides containing C201 (SEVDLYNSDPLICHAGVK) (Figure 2A,B) and C208 (VCFGIQLLNAVSR) (Figure 2A,C).No other dimedone-bound peptides were observed (Figure S2).Further highresolution LC/MS studies revealed that C201 was mainly present as thiol or sulfenic acid, although minor quantities (1-7%) of sulfinic (-SO 2 H) and sulfonic (-SO 3 H) acids were also detected (Figure 2D).C208 was less sensitive than C201 to oxidation and yielded approximately equal amounts of sulfenic, sulfinic and sulfonic acids (Figure 2E).Tandem MS analyses of the two dimedone-labeled tryptic fragments of MGL allowed us unequivocally to assign C201 and C208 as the sites of sulfenylation induced by H 2 O 2 (Figures 3A,B).Lastly, using biotin-1,3-cyclopentanedione (BP1), a biotin-containing dimedone derivative that can be used for affinity precipitation (Qian et al, 2011), we were able to document the presence of sulfenylated MGL in intact Neuro-2a cells overexpressing the enzyme and exposed to H 2 O 2 (300 μM) (Figure 2F).While we found no evidence of sulfenyl amide modification of C201 and C208, it is possible that this unstable chemical species was also transiently generated (Salmeen et al, 2003).Formation of a sulfenyl amide at C208 would explain why this residue is less prone to sulfenylation than is C201.Despite this unanswered question, the results outlined above clearly identify MGL residues C201 and C208 as targets for peroxide-dependent sulfenylation.

MGL sulfenylation interrupts 2-AG degradation in neurons
To determine whether cysteine sulfenylation influences 2-AG deactivation in intact cells, we incubated primary cultures of rat cortical neurons with H 2 O 2 (30-300 μM) and measured MGL activity in intact cells using a dedicated in situ assay.H 2 O 2 inhibited MGL in a concentration-dependent manner (Figure 4A) and this effect was paralleled by an increase in intracellular 2-AG content (Figure 4B).By contrast, we observed no change in the levels of other lipids that are biogenetically or functionally related to 2-AG, including 1-steroyl-2arachidonoyl-sn-glycerol (a 2-AG precursor), arachidonic acid (a product of 2-AG hydrolysis), or anandamide (an endocannabinoid neurotransmitter that is not hydrolyzed by MGL) (Figure 4C).The presence of sulfenylated MGL in H 2 O 2 -treated neurons was evaluated using affinity precipitation combined with western blot and LC/MS analyses.We incubated primary neurons in cultures with vehicle or H 2 O 2 (300 μM), exposed them to the biotin-linked probe, BP1, and then concentrated sulfenyl-containing proteins by affinity precipitation using streptavidin-agarose beads.Blots probed with a selective anti-MGL antibody confirmed the presence of BP1-bound sulfenylated MGL in neurons exposed to H 2 O 2 , but not control neurons (Figure 4D).In a separate experiment, we subjected the affinity precipitates to shotgun LC-MS/MS proteomics.MGL peptides bearing BP1 bound to C201 (Figure 4E) or C208 (Figure 4F) were detected in precipitates of intact neurons treated with H 2 O 2 .Despite a low signal-to-noise ratio, due to the naturally low abundance of MGL, signals matching the predicted accurate mass values and charge states were detected in neurons treated with H 2 O 2 (red traces), whereas no such signal was detected in control cells (black traces).Tandem MS data of the same peptides, matching the expected primary sequence, are reported in Figure S3.Additional evidence that the redox status can affect MGL activity was obtained by blocking glutathione (GSH) biosynthesis with the GSH synthase inhibitor, L-buthionine sulfoximine (BSO) (Griffith and Meister, 1979).Exposing cortical neurons to BSO depleted intracellular GSH stores (Figure 5A), lowered MGL activity (Figure 5B) and increased 2-AG content in a concentration-dependent (Figure 5C) and time-dependent manner (Figure S4).Confirming the generality of this response, experiments with cultures of mouse Neuro-2a cells yielded similar results (Figure S5).

MGL sulfenylation enhances 2-AG-mediated signaling
The data reported above suggest that MGL sulfenylation impairs 2-AG degradation, but does it also enhance 2-AG signaling?To address this question, we exploited the fact that CB 1 receptor activation protects neurons from oxidative damage (Nagayama et al, 1999;Panikashvili et al, 2001).We incubated Neuro-2a cells with a high concentration of H 2 O 2 (300 μM) and assessed cell damage using three complementary methods: lactate dehydrogenase (LDH) release into the medium, 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) reduction and caspase-3 activity.In addition to native Neuro-2a cells, we assessed the effects of H 2 O 2 on Neuro-2a cells overexpressing MGL (Figure 6) or the 2-AG-synthesizing enzyme DGL-α (Figure 7).As expected, MGL overexpression decreased cellular 2-AG levels, which were partially restored by treatment with the MGL inhibitor JZL184 (1 μM) (Figure 6A).Along with this effect, MGL overexpression enhanced H 2 O 2 -induced toxicity (Figure 6B-D), which was (a) heightened by the CB 1 antagonist rimonabant (but not by the CB 2 antagonist AM630) and (b) tempered by MGL blockade with JZL184 (Figure 6B-D).Conversely, DGL-α overexpression increased 2-AG levels in Neuro-2a cells (Figure 7A), and this response was accompanied by a substantial decrease in H 2 O 2 -induced toxicity (Figure 7B-D).The cytotoxic effect of H 2 O 2 was reinstated by treatment with the CB 1 antagonist rimonabant (but not by the CB 2 antagonist AM630) (Figure 7B-D), suggesting that it was mediated by 2-AG-dependent activation of CB 1 receptors.Consistent with this view, in primary neuronal cultures, H 2 O 2 toxicity was enhanced by CB 1 , but not CB 2 , receptor blockade (Figure S6).Treatment with JZL184 alone did not affect cytotoxicity in Neuro-2a cells (LDH release %: vehicle control, 3.46 ± 0.59; JZL184, 1.73 ± 2.60, n=3).These observations indicate that peroxidedependent MGL deactivation accrues the cellular pool of 2-AG that is involved in signaling at CB 1 receptors.

SIGNIFICANCE
Current models conceptualize 2-AG-dependent endocannabinoid transmission at excitatory synapses as being primarily, if not exclusively, driven by postsynaptic mGluR 5 activation and consequent 2-AG production (Castillo et al., 2012).Our findings expand this view by identifying sulfenylation-dependent MGL inhibition as a presynaptic mechanism through which the local redox state regulates 2-AG-mediated signaling.Physiopathological signals that stimulate H 2 O 2 production might enhance the ability of 2-AG to modulate synaptic activity by temporarily interrupting 2-AG deactivation.The results also raise two questions.The first pertains to the sources of H 2 O 2 involved in MGL regulation.Identifying such sources will require additional work, but likely candidates include mitochondria respiration (Sies, 2014) and ionotropic glutamate receptor-operated activation of NADPH oxidase (Brennan et al., 2009;Paulsen et al, 2011).The second question is connected to the first and concerns the context in which the redox control of MGL activity might be operational.Dysfunctional redox states such as brain ischemia are associated with rises in the levels of H 2 O 2 and other reactive oxygen species (Armogida et al, 2012).In those conditions, MGL sulfenylation might act as an intrinsic neuroprotective mechanism by potentiating 2-AG signaling at CB 1 receptors.Nevertheless, localized foci of heightened H 2 O 2 production (Mishina et al, 2011) might be sufficient to deactivate MGL even under physiological conditions, particularly at synapses that experience high-frequency synaptic activity and use glutamate as a neurotransmitter (Patel and Rice, 2012).In that context, MGL sulfenylation may strengthen endocannabinoid-mediated retrograde transmission by lowering the presynaptic degradation of 2-AG generated in postsynaptic spines.Activation of mitochondrial CB 1 receptors (Benard et al., 2012) might stimulate respiration and reactive oxygen species production, further enhancing this negative feedback loop.Finally, allosteric MGL modulators that exploit this regulatory process, such as benzisothiazolinone derivatives (King et al, 2009), might find therapeutic application in stroke, neurodegeneration and chronic neuropathic pain.

Enzyme assays
In vitro MGL activity was measured as described (King et al, 2007).Briefly, we transiently transfected HeLa cells with plasmid DNA encoding recombinant rat MGL using Superfect reagent (Qiagen, Valencia, CA).We harvested cells in ice-cold Tris-HCl (50 mM, pH 8.0) containing 0.32 M sucrose.We prepared homogenates by sonicating cells for 1 min on ice followed by 3 freeze-thawing cycles.Homogenates were incubated with various agents for 10 min at 37°C in assay buffer (50 mM Tris-HCl, pH 8.0 containing 0.5 mg/ml fatty acidfree BSA).The enzyme substrate 2-OG (10 μM), which we use in preference to 2-AG to increase signal-to-noise ratio in the assay, was added to the mixture and incubated for 10 additional min at 37°C.Reactions were stopped by adding chloroform:methanol (2:1, vol/ vol), containing heptadecanoic acid (5 nmol/sample) as internal standard.After centrifugation at 2,000 × g at 4°C for 10 min, the organic layers were collected and dried under N 2 .The lipid extracts were suspended in chloroform:methanol (1:3, vol/vol) and analyzed by liquid chromatography-mass spectrometry (LC-MS).Diacylglycerol lipase (DGL) activity was measured in vitro as described (Jung et al, 2007).Rapid dilution assays were performed as described (King et al, 2009;Copeland, 2005).

MGL activity assay in situ
MGL activity in primary cortical neurons was determined as described (Marrs et al, 2010) with minor modifications.Neurons were prepared using pregnant Wistar rats at embryonic day 18-20 (Stella and Piomelli, 2001).After 10 days in culture, cells were treated with the indicated reagents or vehicle for 30 min in B-27-supplemented Neurobasal medium (Life Technologies, Grand Island, NY).Cells were rinsed once with medium and incubated with substrate 1-HG (1 μM) for 10 min.After rinsing twice with ice-cold PBS, cells were harvested and lipids were extracted as described above, using heptadecanoic acid as internal standard.Non-specific hydrolysis of 1-HG was measured in the presence of excess 1(3)oleoyl-sn-glycerol (100 μM), and baseline value was subtracted from each data point.Neuro-2a cells were partially differentiated by overnight serum deprivation and cultured in complete DMEM medium supplemented with 0.15% fatty acid free BSA.

Site-directed mutagenesis
Mutagenesis studies were performed using a QuikChange II XL Site-Directed Mutagenesis Kit (Strategene, La Jolla, CA) following manufacturer's instructions.We used rat MGL-pEF6/V5-His plasmid DNA as a template (King et al, 2009) and verified all plasmids by DNA sequencing.

MGL purification
Recombinant rat MGL was expressed in E. coli and purified as described (King et al, 2007) with minor modification.

Sulfenic acid trapping in vitro
A sample (200 μl) of MGL solution (40 μM) was transferred in a NanoSep 3K microcentrifuge tube (Pall Corporation, NY, USA) and buffer-exchanged (40×) versus HEPES (50 mM, pH 7.4), 300 mM NaCl 0.1% Triton X-100.The buffer was degassed with N 2 for 2 h prior to use.The sample was split into four 50-μl aliquots and incubated with dimedone at a final 20 mM concentration.H 2 O 2 was added to final 0.1, 1 and 10 μM concentrations.Incubation without H 2 O 2 was used as negative control.Only open-air oxidation was allowed for this control.The samples were incubated at 37°C for 1 h with shaking.Reactions were stopped adding 1 mL of cold acetone, and tubes were centrifuged at 14,000 rpm for 10 min.Acetone was removed and protein pellets were dried under N 2 .The precipitate was further washed with 1 mL of methanol and then was dissolved in 50 μL of ammonium bicarbonate (50 mM, pH 8).Trypsin (1 μL of a 1 μg/μL solution) was added to each tube and samples were incubated at 37°C overnight.Samples were centrifuged at 10,000 × g for 10 min and diluted with 0.1% formic acid in water containing 1% acetonitrile for LC-MS analysis.

LC-MS/MS analyses of peptides
Tryptic peptides were separated using an Acquity UPLC BEH C18 (1 × 100 mm, 1.7 μm) column coupled to a Synapt G2 QTof mass spectrometer (Waters Inc.Milford, MA, USA).The following gradient conditions were used: A = 0.1% formic acid in water, B = 0.1% formic acid in acetonitrile.After 1 min at 97% A, a linear gradient was applied to 45% A in 12 min.then to 0%A in next 3 min, followed by 2 min at 0% A and reconditioning to 97% A in 0.1 min.Total run time was 19 min.Injection volume was set at 5 μL.MS and MS/MS analyses were performed in the positive ESI mode.The capillary voltages was set at 3 kV.The cone voltage was set at 30 V. The source temperature was 120°C.Desolvation gas and cone gas (N 2 ) flow were 600 and 20 L/h, respectively.Desolvation temperature was 400°C.Data were acquired in MSe mode with MS/MS fragmentation performed in the trap region.Low energy scans were acquired at fixed 4 eV potential and high energy scans were acquired with an energy ramp from 25 to 35 eV.Scan rate was set to 0.4 s per spectrum.Scan range was set to 50 to 1600 m/z.Leucine enkephalin (2 ng/ml) was infused as lock mass for spectra recalibration.LC-MS/MS data were analyzed using Proteinlynx software to map MGL peptides, calculate MGL sequence coverage and quantify naïve and oxidized peptides.Additional de-novo sequencing of dimedone adducts at cysteine 201 and 208 was manually performed using Biolynx software to verify the peptide sequence and assign the modified residue.

Sulfenic acid trapping in intact cells
Neuro-2a cells were transfected with plasmids encoding MGL-V5 or the empty vector (pEF6) and, two days after transfection, were incubated with H 2 O 2 for 1 h.Rat cortical neurons were treated with H 2 O 2 for 1 h after 8 days in vitro.Following treatment, cells were scraped in lysis buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, and a mixture of protease inhibitors (Roche Diagnostics) in the presence or absence of 1 mM biotin-1,3-cyclopentanedione (BP1, KeraFAST, Boston, MA) (Qian et al, 2011).We incubated the mixtures on ice for 30 min with gentle mixing every 5 min.After centrifuging at 1,000 × g for 10 min at 4 °C, the supernatants were mixed with 10 μl of streptavidin-agarose (Pierce) at a protein concentration of 0.5 mg/ml, and incubated at 4°C overnight.The affinity precipitates were collected by brief centrifugation and washed 6 times with lysis buffer without BP1.For Western blot analyses, proteins in the precipitates were eluted by adding 20 μl of Laemmli sample buffer and incubating at 74°C for 20 min.For LC-MS/MS analyses, washed beads were stored in dry ice.Prior to LC-MS/MS, the beads were allowed to reach room temperature and were suspended in 25 ml of 2% SDS, 6 M urea and 2 M thiourea in PBS.Samples were kept 15 min at room temperature and then incubated at 96°C for 15 min to cleave the biotin-streptavidin bond (Rybak et al., 2004, with slight modifications).Samples were allowed to reach room temperature and were centrifuged for 10 min at 4000 × g.Supernatants were collected and the beads were further washed with 25 ml of water and incubated again at 96°C for 15 min.After centrifug ation (10 min, 4000 × g) the supernatants were pooled and the beads discarded.The resulting 50 ml of supernatant were added with 2 ml of 100 mM DTT and incubated at 50°C for 30 min.Iod oacetamide (100 mM, 4 ml) was added and the samples were incubated at room temperature for 30 min.After reduction and alkylation, released proteins were precipitated by adding 1 ml of ice-cold acetone and then centrifuged for 10 min at 4000 × g at 4°C.The supernatant was disc arded and the pellets were dried under N 2 .
Samples were suspended in 40 ml Rapigest (Waters Inc.) previously dissolved in 100 mM ammonium bicarbonate, pH 8, at a final 1 mg/ml concentration.Trypsin was then added (1 ml, dissolved at 0.5 mg/ml in 0.1% formic acid) and the samples were incubated overnight at 37°C.The following day, samples were added with 1 ml of TFA, incubated at 37°C for 1 h and centrifuged to remove hydrolyzed Rapigest (10 min at 4000 × g).Supernatants were collected (30 ml) and added with 10 ml of formic acid 0.1% in water/acetonitrile (97:3).An aliquot (5 ml) was loaded on a nano-UPLC chromatographic system equipped with a T3 C18 reversed-phase column (75 mm × 250mm).Peptides were eluted with a linear gradient of acetonitrile in water (both containing 0.1% formic acid) from 3 to 50% in 120 min.Flow rate was set to 350 nl per min.Eluted peptides were analyzed in positive ion mode by highresolution tandem mass spectrometry on a Synapt G2 qTOF mass spectrometer (Waters, Milford MA, USA).Tandem mass spectra were acquired in Data Dependent Acquisition mode (DDA), by selecting charge states ranging from 2 to 5 in the 300-1200 m/z range.

Figure 3 .
Figure 3. Tandem mass analyses of cysteine-containing tryptic fragments of MGL treated with hydrogen peroxide (A) Tandem mass spectra of peptide SEVDLYNSDPLICHAGVK, showing that sulfenylation (detected as DMD adduct) occurred on C201.(B) Tandem mass spectra of peptide VCFGIQLLNAVSR, showing that sulfenylation (detected as DMD adduct) occurred on C208.