Specific Inter-residue Interactions as Determinants of Human Monoacylglycerol Lipase Catalytic Competency

The serine hydrolase monoacylglycerol lipase (MGL) functions as the main metabolizing enzyme of 2-arachidonoyl glycerol, an endocannabinoid signaling lipid whose elevation through genetic or pharmacological MGL ablation exerts therapeutic effects in various preclinical disease models. To inform structure-based MGL inhibitor design, we report the direct NMR detection of a reversible equilibrium between active and inactive states of human MGL (hMGL) that is slow on the NMR time scale and can be modulated in a controlled manner by pH, temperature, and select point mutations. Kinetic measurements revealed that hMGL substrate turnover is rate-limited across this equilibrium. We identify a network of aromatic interactions and hydrogen bonds that regulates hMGL active-inactive state interconversion. The data highlight specific inter-residue interactions within hMGL modulating the enzymes function and implicate transitions between active (open) and inactive (closed) states of the hMGL lid domain in controlling substrate access to the enzymes active site.

The serine hydrolase monoacylglycerol lipase (MGL) functions as the main metabolizing enzyme of 2-arachidonoyl glycerol, an endocannabinoid signaling lipid whose elevation through genetic or pharmacological MGL ablation exerts therapeutic effects in various preclinical disease models. To inform structure-based MGL inhibitor design, we report the direct NMR detection of a reversible equilibrium between active and inactive states of human MGL (hMGL) that is slow on the NMR time scale and can be modulated in a controlled manner by pH, temperature, and select point mutations.

Kinetic measurements revealed that hMGL substrate turnover is rate-limited across this equilibrium. We identify a network of aromatic interactions and hydrogen bonds that regulates hMGL active-inactive state interconversion. The data highlight specific inter-residue interactions within hMGL modulating the enzymes function and implicate transitions between active (open) and inactive (closed) states of the hMGL lid domain in controlling substrate access to the enzymes active site.
A member of the serine hydrolase superfamily, monoacylglycerol lipase (MGL) 2 is largely responsible for the catalytic inactivation of the endocannabinoid signaling lipid, 2-arachidonoylglycerol, and regulates a fatty acid network that promotes tumorigenesis (1,2). On the basis of the preclinical efficacy of MGL genetic or pharmacological ablation against pain, inflammation, and cancer, MGL is considered an attractive therapeutic target (3). Published crystal structures of modified/ liganded forms of hMGL (4 -6) indicate that the enzyme has a typical lipase structure that includes a lid subdomain (residues 151-225) that can assume open or closed states and control, through its gating dynamics, substrate access to the active site (7)(8)(9)(10)(11)(12)(13). The open state has been observed in the absence and presence of bound ligand (4,5), whereas the hMGL closed conformation has only been reported in complex with a reversible inhibitor (6). Interspecies conservation of the overall hMGL lid architecture and dynamics has been suggested from x-ray studies and molecular dynamics simulations of bacterial MGL (14 -16).
Collective data from x-ray crystallographic studies of hMGL and bacterial MGL variants have allowed inference that the lid subdomain shows a high degree of conformational plasticity along a coordinate of catalytic activity and suggest the existence of a stochastic equilibrium that transitions the enzyme between open-and closed-lid conformations. Nonetheless, there are limits with which even a suite of static x-ray maps per se can reflect the conformational flexibility of a protein and structural influences on protein function (11). The concern is underscored by the important roles that protein structural transitions play in enzyme catalysis and the influence of targeted ligands thereon as prospective drugs (12,13).
These considerations led us to probe hMGL structure-function correlations by using the catalytically active enzyme. In this regard our previous work has demonstrated the effects of important interactions on hMGL structure, substrate affinity, and/or activity. These interactions include hMGL membrane association (14), binding of designer active site-directed inhibitors (15,16), and covalent or mutational modifications of amino acid residues at or near the hMGL catalytic triad (Ser-122-His-269 -Asp-239). The structural effects of these interactions are not restricted to the hMGL lid domain but are also to discrete non-lid enzyme regions as well (14,15). Perhaps most strikingly, we identified by NMR a strong hydrogen bond network and provisionally implicated His-269 and Asp-239 of the catalytic triad and neighboring Leu-241 and Cys-242 residues therein. Not only did this hydrogen bond network influence catalytic activity but active site-directed inhibitors of different types were observed to alter this population of hydrogen bonds in concert with their inhibitory activity (15).
The influence of noncovalent amino acid interactions in regulating protein conformational transitions, protein ligand binding architecture, and ligand pharmacological activity is a topic of great current interest. For hMGL, the factors leading to conformational transitions of the enzyme are not well defined, and the effects of specific intramolecular amino acid interactions on the hMGL catalytic efficiency have not been measured. Unlike other lipases (8,11), hMGL shows a high level of activity and conformational flexibility as an isolated enzyme without association to a lipid surface (17). It can, therefore, be utilized to determine important molecular details such as understanding hydrogen bond networks within an enzyme molecule that affect protein architecture and catalysis.
We demonstrate here simultaneous NMR detection of both active and inactive hMGL states in solution and report the unambiguous assignment of downfield NMR resonances that are highly sensitive to the equilibrium between these two states. These resonances of equal integral intensity represent the hydrogen-bonding pattern of the active conformer that is predominant at neutral pH. Significant changes in the intensities of the peaks with pH, temperature, or point mutations are indicative of alterations of this hydrogen-bonding pattern when the equilibrium is shifted toward the inactive state. The well-appreciated role of hydrogen bonds as determinants of protein structure suggests that the hMGL inactivation observed reflects enzyme conformational changes that have their origin in the altered hydrogen-bonding network. The ability of the perturbations in the downfield region of the hMGL NMR spectra to be elicited in a predictable manner from different changes to physical conditions implies the existence of a regulated equilibrium between distinct enzyme conformations. The activity of hMGL may be modulated by structural determinants that can transition the enzyme between open-closed conformations.
For enzyme purification, 3 g (wet weight) of cells were resuspended in 20 ml of lysis buffer (20 mM sodium phosphate, 200 mM NaCl, 1 mM DTT, pH 7.4) and 20 ml of xTractor buffer (Clontech, Mountain View, CA) supplemented with lysozyme (0.1 mg/ml) and DNase I (25 g/ml) (Fisher). After 20 min the lysate was disrupted on ice by three 1-min sonication cycles, each consisting of 1-s sonication bursts at a 50-watt power level separated by a 5-s interval (Vibra-Cell 500 W, Sonics, Newtown, CT). The resulting cell lysate was centrifuged at 20, 000 ϫ g for 25 min at 4°C. hMGL was isolated by incubating the resulting supernatant with 4.0 ml (bed volume) pre-equilibrated Talon metal affinity resin (Clontech) for 1 h at 4°C. This suspension was transferred to a gravity-flow column and allowed to settle. The resin was washed twice with 20 ml of lysis buffer containing 25 mM imidazole, and His-tagged hMGL was eluted with 12 ml of lysis buffer containing 300 mM imidazole. Protein purity was evaluated using Any kDa Mini-PROTEAN TGX SDS-PAGE (Bio-Rad). Protein samples were denatured at 70°C for 5 min in Laemmli buffer containing 5% ␤-mercaptoethanol, resolved on SDS-PAGE gels, and stained with Coomassie Blue (Fisher). Before enzyme assays and NMR experiments, purified hMGL samples were dialyzed for 12 h to ensure thorough imidazole removal using a membrane with a molecular mass cutoff of 10,000 -12,000 Da. Enzyme concentration was determined spectrophotometrically using the molar extinction coefficient ⑀ 280 24,910 M Ϫ1 cm Ϫ1 .
Our sample preparation protocol routinely yields sol-hMGL in the active conformation. Over time the active conformer undergoes spontaneous conformational switching to an inactive state. The spontaneous transition between the two states is extremely slow, suggestive of a high energy barrier for the switch in our sol-MGL variant. The active conformer is stable for several weeks at room temperature. At 310 K the transition rate is faster. Incubation of the sample at 310 K for several days resulted in a complete transition to the inactive form. This enabled us to prepare distinct sol-hMGL conformers for NMR analysis.
Enzyme Assays-The endogenous hMGL substrate, 2-arachidonoylglycerol (2-AG), was used for the determination of catalytic parameters (K m , V max , K cat ). Hydrolysis of 2-AG to arachidonic acid (AA) by hMGL was monitored and quantified by HPLC. Briefly, 280 l of assay buffer (50 mM Tris-HCl, 5 mM MgCl 2 , 1 mM EDTA, and 0.1% BSA, pH 7.4) was preincubated with 15 l of each 2-AG stock dilution (final 2-AG concentrations, 13-400 M), and the reaction was initiated by the addition of 5 l of purified hMGL (18 ng-1 g). Aliquots (50 l) were taken immediately at the start of the incubation and after 20 min diluted 4-fold by volume with chilled acetonitrile to quench enzyme activity and centrifuged at 20,000 g for 5 min at 4°C. Supernatant (20 l) was injected directly into a Waters Alliance 2695 HPLC system for analysis. In an 8-min run, 2-AG eluted at 3.0 min and AA at 4.0 min, allowing the reaction to be followed by either substrate (2-AG) turnover or product (AA) formation. A gradient elution profile of 5% B (water/acetonitrile/orthophosphoric acid ϭ 54/40/6%) to 100% A (acetonitrile) at a 1 ml/min flow rate was used for separation on a ZOR-BAX Eclipse XDB-C18 reverse-phase (4.6 ϫ 50 mm, 3.5 m) column (Agilent Technologies, Santa Clara, CA). Analytes were quantified with external standards. The rate of AA formation was determined by subtracting the AA concentration at t ϭ 0 from that at t ϭ 20 min. Initial rates of 2-AG hydrolysis at the various substrate concentrations were determined. V max and K m values were then estimated by fitting the initial-rate data to the Michaelis-Menten equation using nonlinear regression with GraphPad Prism 5.0 (San Diego, CA). All assays were performed in triplicate.
Sample Preparation and NMR Spectroscopy-Samples used for NMR analyses were 0.1-0.4 mM hMGL in 20 mM sodium phosphate, 200 mM NaCl, 1 mM DTT, 0.02% sodium azide, 95% H 2 O, 5% D 2 O at specified pH values. Sodium 2,2-dimethyl-2silapentane-5-sulfonate was added (ϳ20 M) as an internal chemical shift reference (␦ ϭ 0.00 ppm). Sample volume was 0.6 ml. The pH was adjusted as desired by the addition of microliter amounts of 0.1 M HCl or NaOH and measured using a Wilmad Labglass pH electrode (3-mm outer diameter ϫ 180-mm length) inserted into the protein solution in the 5-mm NMR tube at room temperature before and after NMR data collection. NMR spectra were recorded at 16 pH values between pH 6.5 and pH 12 for sol-hMGL because the enzyme precipitated below pH 6.
Ligand binding experiments were carried out using a 50 mM stock solution of paraoxon (diethyl 4-nitrophenyl phosphate), dissolved in DMSO-d 6 (18). Typically, no more than 5 l of paraoxon solution was added to each enzyme sample to achieve a concentration of paraoxon twice that of the enzyme.
One-dimensional 1 H NMR spectra were acquired at 700 MHz with a Bruker AVANCE II NMR spectrometer equipped with a 5-mm triple resonance inverse probe at 37°C. For optimal detection of downfield exchangeable proton resonances, the 3-9-19 WATERGATE (19) pulse sequence (p3919fpgp) with gradients and additional flipback pulse was used. This pulse sequence employs a binomial-like pulse train that provides null excitation at the water frequency. The center of the maximal excitation region was 13.9 ppm, and the calculated delay for binomial water suppression was 39 s at 700 MHz. In combination with a flipback pulse, the 3-9-19 sequence significantly prevents unwanted attenuation of downfield resonances from spin diffusion and chemical exchange with water. Routinely, 8-K scans were accumulated. The 1 H, 15 N Fast-HSQC (20) spectrum of uniformly 15 N-labeled sol-hMGL was recorded using a spectral width of 30 ppm and 140 ppm in 1 H and 15 N dimensions, respectively. The 1 H transmitter was set to the frequency of the water resonance, and the 15 N carrier frequency was set to 175 ppm to achieve maximum intensity for the histidine side chain resonances. Regular 1 H, 15 N HSQC and 1 H-13 C HSQC NMR spectra were recorded using standard pulse sequences supplied with the AVANCE 700 spectrometer. All NMR data were processed with TopSpin software (Bruker). For one-dimensional spectra, exponential multiplication (broadening factor lb ϭ 20 Hz) was applied.

Results
hMGL Variants Facilitating NMR Experiments-Recombinant wild-type hMGL required detergents during purification to maintain stability and prevent aggregation in concentrated solutions Ͼ100 M. Aggregation of wt-hMGL resulted in significant broadening of NMR resonance peaks and protein precipitation. To avoid the need for detergents and obviate enzyme aggregation/precipitation, a DNA construct expressing a soluble hMGL variant with two leucines substituted by two serines in the lid subdomain (double mutant L169S,L176S ) (sol-hMGL) was expressed and used in this study (6). These substi-tutions resulted in a modest decrease in enzyme catalytic efficiency (k cat /K m ) from 2.2 ϫ 10 5 to 2.5 ϫ 10 4 M Ϫ1 s Ϫ1 and no change in substrate affinity (respective K m values: 25 Ϯ 6 and 22 Ϯ 4 M). Thus, hMGL function was not significantly compromised by these mutations, suggesting that overall enzyme conformation was also not adversely affected. This construct (310 amino acids, 34.1 kDa) formed the basis for introducing additional mutations to allow residue assignment of downfield NMR resonances. Sol-hMGL and mutants exhibited greater solubility and stability, overcoming aggregation issues and providing sharp NMR resonances (data not shown).
hMGL NMR Downfield Spectral Features-A striking feature of the hMGL one-dimensional 1 H NMR spectrum is the downfield region (12-18 ppm) that exhibits four well resolved labile proton resonances (Fig. 1A). These signals are detectable with 15 N chemical shifts in the range of 160 -180 ppm, as shown in the partial 1 H, 15 N HSQC spectrum (Fig. 1B), and consequently correspond to nitrogen-bonded imidazole resonances from histidines (21). The extreme downfield chemical shift of these protein resonances is a consequence of their participation in hydrogen-bonding interactions with neighboring residues (22).
hMGL NMR Amino acid Resonance Assignments-1 H NMR signals for catalytic His protons involved in hydrogen bonds in the active sites of serine proteases have been observed in the range of 12-19 ppm (23). On this basis we had provisionally assigned one signal in the downfield region to an H-bond between the catalytic His-269 and Asp-239 of hMGL. To assign experimentally and definitively hMGL His resonance peaks and define intramolecular details affecting hMGL function, we conducted single-point mutations of specific amino acid residues. Our rationale was based upon the concept that disappearance of a resonance in the downfield region of the NMR spectrum upon substitution of a given His residue by Ala would provide definitive mutagenesis-based resonance assignment (Fig. 1A).
To assign the catalytic His residue, a variant of sol-hMGL was prepared by substituting His-269 with alanine. As a consequence of removal of the side-chain H-bond donor from the active site in this H269A mutant, only the downfield NMR signal at 14.9 ppm disappeared, whereas all other signals in the sol-hMGL downfield region remained unchanged (Fig. 1A). Thus, the peak observed at 14.9 ppm corresponds to the His-269 H ␦1 proton that is hydrogen-bonded to Asp-239 in the hMGL Ser-122-His-269 -Asp-239 catalytic triad ( Fig. 2A).
This chemical shift of catalytic histidine resonances in serine proteases has been shown to be highly dependent upon histidine protonation state and enzyme conformation (23). Seemingly minor changes in pH can readily change the His charge state and, consequently, the chemical shift of H ␦1 proton. In serine proteases, a doubly protonated (positively charged) active site His imidazole ring demonstrates that H ␦1 resonances usually shifted downfield to ϳ17-19 ppm (23). To investigate this phenomenon in hMGL and support the His-269 H ␦1 assignment, we attempted to titrate hMGL samples to a more acidic pH. Because this approach caused considerable protein precipitation, we alternatively utilized a sol-hMGL variant in which the catalytic Ser-122 is substituted with a cysteine residue (16). The proton NMR spectrum of the sol-hMGL-S122C mutant contained a new resonance at 18 ppm, and the signal at 14.9 ppm was absent (Fig. 1C). This diagnostic downfield shift provides strong evidence supporting the protonation of His-269 in the S122C mutant and formation of an imidazole-thiolate ion pair between the His-269 and Cys-122 in the modified hMGL active site (24). Gradual titration to a higher pH elicited a decrease in the peak intensity at 18 ppm, and this resonance disappeared completely at pH 10.6 (data not shown). This observation reflects deprotonation of His-269 in the basic environment. Titration back to a neutral pH restored the peak, demonstrating the reversibility of His-269 protonation. Moreover, the substitution of His-269 to alanine in the S122C mutant resulted in the disappearance of the peak at 18 ppm (Fig. 3A). The NMR spectra of the sol-hMGL H269A and S122C variants thus provide multiple lines of evidence allowing unambiguous assignment of the enzyme's catalytic His-269 H ␦1 proton at 14.9 ppm when His269 is neutral and at 18 ppm when His-269 is protonated. Furthermore, we performed enzymatic assays on all hMGL mutants used for assignment, and the H269A mutation was alone in completely abolishing activity, consistent with our assignment ( Table 1).
The resonance at 12.8 ppm was assigned to the His-49 H ␦1 proton based on the spectrum obtained from the H49A sol-hMGL variant (Fig. 1A). This histidine residue is located in the loop connecting ␣1 and ␤3 (Fig. 2B). The loop includes the Ala-51 residue, which is directly involved in the formation of the oxyanion hole together with the backbone amide group of Met-123 (6). Consequently, His-49 is involved in a hydrogenbonding network to the oxyanion hole.
Substitution of His-54 with alanine caused the NMR signal at 13.9 ppm to disappear, providing definitive assignment of this resonance (Fig. 1A). His-54 is located in the loop connecting ␣1 and ␤3 (Fig. 2C). The observation of the His-54 resonance in this downfield region suggests that in the hMGL active state there is a hydrogen bond between His-54 and Asp-197 located in the enzyme's lid subdomain (Fig. 2C). The H54A mutation also resulted in the disappearance of the His-269 H ␦1 resonance at 14.9 ppm and a dramatic loss in hMGL catalytic efficiency by 4 orders of magnitude (Table 1). These results from the H54A variant indicate that this hydrogen bond is important to enzyme function. In marked contrast, the H49A mutation preserves the H269A resonance, and there is a relatively modest 5-fold reduction in catalytic efficiency in that sol-hMGL variant ( Table 1).

JOURNAL OF BIOLOGICAL CHEMISTRY 2559
tion also preserves the downfield His-269 resonance and elicits a 20-fold reduction in catalytic efficiency (Table 1). Additional evidence for the unambiguous assignment of the resonance at 15.9 ppm to the hydrogen-bonded pair His-103-His-75 is provided by the spectrum of the H75A mutant (Fig. 3B). Mutation of either the donor or acceptor residue to Ala eliminates hydrogen-bonding between them and leads to the absence of a downfield-shifted (15.9 ppm) resonance. The resonances assigned represent protons from amino acid residues that are structurally essential or critical to hMGL catalysis. They can be used as unique probes to monitor functionally important events.
pH Effects on Sol-hMGL NMR Spectral Features-NMR titration experiments of sol-hMGL in the 7.0 -12 pH range revealed an exchange between two distinct conformations that is slow on the NMR time scale (Fig. 4). As the pH is increased from 7.5, there is a gradual decrease in the peak intensity of His-269 (14.9 ppm) until it completely disappeared at pH 11. A new signal appeared concomitantly at 12.90 ppm that is likely the His-269 resonance. This significant upfield shift, in a slow-   exchange manner, suggests a modulation of catalytic triad geometry (Fig. 4A). Increasing pH also elicits a similar decrease in the intensity of the His-54 (13.9 ppm) resonance due to its shifting well outside of the downfield region, likely as a result of the breaking of the hydrogen bond between His-54 and Asp-197 (Fig. 4A). A similar effect occurs in the spectra of the H54A mutant (Fig. 1A). These data correlate with a significant decrease in the catalytic efficiency of the H54A variant, suggesting a critical role for this hydrogen bond in maintaining hMGL catalytic competency ( Table 1). The resonance from His-103 (15.99 ppm) exhibited the formation of a second component shifted slightly downfield that gradually increased in intensity at the expense of the original peak. The two slightly separated peaks reach equal intensity at pH 9.6 until, at pH 11, there is only a single component observed at the slightly downfield chemical shift. A similar observation was made in the case of His-49 (12.8 ppm) where the peak at 12.8 ppm gradually decreased in intensity, and a new peak appeared at 12.4 ppm. Remarkably, the same slow exchange regime was observed in the upfield region of the spectrum for separated methyl group resonances such as the peak clearly observable at Ϫ0.52 ppm (Fig. 4B). Therefore, perturbations observed in both downfield and upfield chemical shift regions are likely indicative of one global process within the enzyme. All spectral changes were reversible with respect to pH (data not shown). These observations are consistent with a pH-dependent reversible equilibrium between two distinct hMGL conformations that are in slow exchange on the NMR time scale. The downfield signals were integrated and normalized to the intensity of the peak corresponding to the hydrogenbonded pair His-103-His-75 and are expressed as the percent conformation as a fraction of their respective maximal intensi-ties across the pH range studied (Fig. 4, C and D). The data demonstrate that hMGL populates active and inactive conformations in solution with an equilibrium constant K eq ϳ 1 at pH 9.6.
hMGL Global Conformational Changes-15 N HSQC and 13 C HSQC spectra of sol-hMGL in the active and inactive states were obtained. The active conformer of sol-hMGL undergoes a slow spontaneous transition to the inactive conformer after exposure to 37°C for long periods of time. This finding allows us to perform NMR experiments on the individual conformers at identical concentration, temperature, and pH ensuring that spectral differences were reflective of conformational changes. Superimposed HSQC spectra (Fig. 5) for individual forms demonstrate significant chemical shift perturbations for backbone as well as side-chain resonances. This is a clear indication that hMGL structural alterations have a global impact upon conformational transition from active to inactive form. Moreover, these spectra provide unambiguous experimental evidence that chemical shift perturbations observed in the downfield region are related to the global conformational changes. Thus, hMGL downfield proton NMR resonances can be used as a sensitive probe for quantification of conformational equilibrium for this enzyme.
His-272 Mutational and Temperature Effects on the Sol-hMGL NMR Spectra and Enzymatic Activity-To explore this equilibrium further, we performed a mutagenesis study on the hMGL His-272 residue that forms an aromatic cluster along with Tyr-58 and Arg-57 (Fig. 6). Substitution of His-272 with tyrosine resulted in little change in the downfield resonances, suggesting that there is minimal perturbation to the interactions in the aromatic cluster (Fig. 7A). In contrast, the H272A mutation elicited a decrease in the intensity of the His-269 and His-54 resonances (Fig. 7A), and the resultant spectrum closely resembles the spectra of the sol-hMGL enzyme at higher pH values (Fig. 4A, pH 9.7). Likewise, the H272S mutation resulted in greater reduction of the His-269 and His-54 resonance intensities (Fig. 7A), and the spectrum is comparable to that of sol-hMGL obtained at pH 11. We characterized the activities of these mutants toward the native hMGL substrate, 2-AG. The H272Y and H272A mutants evidenced a 12-fold reduction of   sol-hMGL catalytic efficiency, whereas the H272S mutation caused a 60-fold reduction in the enzyme's catalytic efficiency (Table 1). These effects correlate with the retention of the His-269 and His-54 resonances in the downfield region of H272Y and H272A spectra and the loss of these resonances from the downfield region in the H272S spectra.
These observations are consistent with a shift in the conformational equilibrium of the sol-hMGL-H272S mutant toward an inactive form. Thus, the downfield pattern of resonances is indicative of the conformational equilibrium and resulting hydrogen-bonding pattern. The catalytic efficiency of these mutants provides confirmation that the inactive conformation is characterized by the absence of the His-269 resonance at 14.9 ppm and His-54 resonance at 13.9 ppm. Fig. 7B shows the response of the sol-hMGL-H272S mutant to temperature. As the temperature is decreased, the intensity of the His-269 and His-54 resonances increases reversibly. This is additional evidence supporting the proposition that these resonances are sensitive to conformational changes in hMGL, with reduced temperatures favoring the active conformation. For the H272Y variant the major sol-hMGL population is in the active conformation, whereas for the H272A variant the major population is in an inactive conformation. For the H272S mutant, the equilibrium is almost completely shifted toward the inactive conformation (Fig. 7A).
Active-site Paraoxon Covalent Binding-To test the accessibility of the hMGL active site to ligands, we performed binding experiments with paraoxon, a low molecular weight organophosphate compound that covalently binds to the serine residue of serine hydrolases (Fig. 8) (18). The spectrum of sol-hMGL displays downfield resonances consistent with a major population of active enzyme. Likewise, the spectrum of hMGL-H272A also contains these resonances, although their intensities indicate a decrease in the population of active enzyme. For both of these hMGL variants, paraoxon elicits significant changes to the downfield NMR resonances that are immedi-ately observable, demonstrating that this ligand had access to the active site (Fig. 8A). The spectra of the H54A display downfield resonances consistent with a major population of the inactive form of the enzyme. Paraoxon did not immediately alter the NMR spectrum of the H54A mutant (Fig. 8B), suggesting that the binding site is inaccessible to small-molecule ligands in the H54A mutant.

Discussion
NMR spectroscopy has proven to be an important tool for probing the mechanism of action of the catalytic triad in serine proteases (23). Often, active-site labile protons exchange with water protons too rapidly in solution to be observed by NMR. The catalytic triad of serine hydrolases, however, is typically buried within the protein and thus shielded from direct solvent exposure, and the catalytic histidine protons are involved in hydrogen-bonding. These factors slow the exchange rate of labile histidine protons in serine hydrolases, improving the opportunity for detecting discrete downfield-shifted NMR resonances (22,23).
Previous NMR studies on serine proteases other than hMGL have focused mainly on histidine protons in the catalytic site (25,26). Our success in modifying hMGL to increase its solubility along with the use of appropriate solvent suppression schemes (19,20) has allowed us to observe several hydrogenbonded histidine protons outside the catalytic site that offer insights to overall enzyme conformational changes. The enzyme variants we engineered through defined sol-hMGL point mutations allowed us not only to obtain definitive amino acid resonance assignments but also to define inter-residue hydrogen bond interactions that correlate with changes in sol-hMGL catalytic efficiency.
The pH titration and temperature data show a similar pattern in that the more significant perturbations occur for the His-269 and His-54 resonances (Figs. 4 and 7). The inactive form of hMGL is thus characterized by a reduced number of peaks in the downfield NMR region. These perturbations likely reflect conformational changes affecting the geometry of the catalytic triad that predispose the enzyme toward an inactive form. Reversibility of the temperature and pH effects suggest a dynamic equilibrium between two distinct enzyme conformations: active and inactive. The downfield spectral effects correlate with chemical shift perturbations of backbone and side chain resonances seen in the HSQC spectra (Fig. 5) that indicate a global conformational rearrangement. Different physical factors can, therefore, impact the same global rearrangement that can modulate the hydrogen bond pattern of sol-hMGL, as reflected in the downfield region of the enzyme's NMR spectra.
For instance, the H54A mutation impacts the His-269 H ␦1 resonance at 14.9 ppm (Fig. 1) with a significant loss of catalytic efficiency ( Table 1). The His-54 residue of hMGL is located at a substantial distance from the catalytic triad (4,6) such that this mutation could not have had a direct effect on His-269. Among the His-272 mutants, the H272S mutation in sol-hMGL most compromised the enzyme's catalytic efficiency, consistent with the conformational transition to the inactive form reflected by the absence of the His-269 and His-54 resonances in the down- field region (Fig. 7A). This suggests that His-272 is a key residue that stabilizes the active conformation of hMGL through a series of aromatic interactions. The impressive overall congruence among the spectral changes elicited by sol-hMGL point mutations, pH, and temperature variations would indicate that the spectral changes observed are due to the same conformational transitions.
A lid subdomain is one of the distinguishing structural features of many lipases. The lid acts as a regulatory domain that gates substrate accessibility to the active site, giving rise to the operational designations of "open" (i.e. substrate-accessible) and "closed" (i.e. substrate-inaccessible) lipase structural states (8,27). Both static x-ray studies (4 -6) and our experimental data have identified an hMGL lid domain that can associate with the cell membrane in vivo and modulate access of endocannabinoid lipid-signaling substrates to the hydrophobic channel leading into the enzyme's active site. In view of the functional role of the hMGL lid domain in controlling substrate entry to the enzyme's catalytic core, it is tempting to speculate that at least some of the hydrogen bond interactions we have defined may be involved in conformational changes in the lid subdomain reflecting the lid's open and closed states.
In this regard there are two noteworthy structural features deduced from static x-ray data that implicate our data to the open and closed forms of the enzyme. There is a significant conformational rearrangement of the Arg-57 side chain between the open and closed states, which is in the vicinity of the aromatic cluster that includes His-272 (Fig. 6) (5,6). In addition, the hydrogen bond between the His-54 -Asp-197 residues occurs between the lid subdomain and the core of the enzyme, and this may also be impacted by the conformational rearrangement of Arg-57 in this region of the enzyme (Fig. 2C) (5,6).
This led us to believe that formation or breaking of this hydrogen bond may be associated with a global conformational change responsible for the regulation of the equilibrium between the open and closed states. Support of this hypothesis was provided by the paraoxon binding studies. The H54A mutant showed no changes occurring to the downfield resonances, whereas similar experiments demonstrated immediate changes to the sol-hMGL and H272A spectra (Fig. 8). We interpret this as the inability of the ligand to enter the binding pocket due to a drastic shift in the equilibrium of the H54A mutant toward a closed conformation.
Moreover the changes that occur to the wild type and H272A mutant upon the addition of paraoxon (Fig. 8) provide evidence that complex formation is proportional to the fraction of enzyme in the active conformation. It is possible, therefore, that the active conformation, as characterized by the presence of the His-54 and His-269 resonances, represents an open form of the enzyme that allows a ready access to the binding site.
Conversely, the absence of the His-54 and His-269 resonances as seen in the H54A, H272S, and high pH value spectra is associated with greatly reduced catalytic efficiencies (Fig. 4A) and may indicate a shift in equilibrium toward the closed conformation. With this interpretation, our data suggest the existence of a His-54 -Asp-197 hydrogen bond in the open state, as indicated by the presence of the His-54 resonance, whereas it is broken in the closed state, as indicated by the absence of the His-54 resonance. This differs from x-ray structures of the open form where His-54 is shown as the N ␦1 -H tautomer and does not display this hydrogen-bonding (4,5), whereas the structure of the closed form indicated His-54 is in the more common N ⑀2 OH tautomer and is hydrogen-bonded to Asp-197 (Fig. 2C) (6). This difference may be explained by common difficulties in deriving the exact conformational state of histidine side chains from deposited x-ray structures. The analysis of histidine sidechain conformations and hydrogen-bonding with proximal donors or acceptors can be obscure as there are three different protonation states of His, and three rotameric states may be generated through flipping of the imidazole ring (28). In addition, NMR results have previously differed from crystal structures in this regard, due to altering pK a values of histidine side chains in crystals compared with solution (29).
Closer examination of the crystal structures shows that in the open form, His-272 interacts with Tyr-58 via an aromatic-aromatic (-) interaction and simultaneously with the Arg-57 cationicONH 2 ϩ group via a cation-interaction, thereby establishing a network of aromatic interactions involving three residues (Fig. 6A). The geometry of this cluster of residues is optimal foras well as cation-andinteractions. Our results provide experimental evidence that this network may be stabilizing the open conformation of hMGL.
In the closed form the guanidinium group of Arg-57 is flipped toward the Asp-197 residue from the lid subdomain, pointing away from the His-272 imidazole ring and disrupting the cationinteraction (Fig. 6B). This group now forms a direct hydrogen bond with Asp-197, and the NH1 of Arg-57 is an H-bond donor to O ␦2 of Asp-197. Thus the Arg-57 side chain rotates out of the active site and adopts the alternate position during the conformational switch from the open to the closed form of hMGL.
The importance of the cationinteraction between the Arg-57 and His-272 side chains in stabilizing the active/open state is shown by the reduced catalytic efficiency of the H272Y, H272A, and H272S mutants (Table 1). Conservative H272Y and nonconservative H272A substitutions demonstrate a 12-fold reduction, whereas the nonconservative mutation H272S demonstrates a 60-fold reduction in the catalytic efficiency of the enzyme. The drop in catalytic efficiency is mainly caused by a decrease in k cat values for all His-272 mutants, whereas the K m values are essentially unchanged.
Removal of one hydrogen bond between His-54 H ⑀2 and O ␦2 of Asp-197 in the lid subdomain also results in a population shift toward the inactive/closed conformation (Fig. 1A) and significant reduction of k cat (Table 1). Clearly, the hMGL open conformation is highly pre-organized by aromatic interactions and hydrogen-bonding. These interactions are destabilized by perturbations induced by point mutations. Thus the hMGL turnover is rate-limited by the equilibrium constant of the preexisting open-closed transition. Aromatic residues proximal to the active site in addition to hydrogen bonds serve in the stabilization and/or regulation of the active site geometry.
In conclusion, our NMR study provides several lines of experimental evidence that hMGL exists in solution in a dynamic conformational equilibrium between active and inac-tive forms that is slow on the NMR time scale and can be modulated by pH, temperature, and specific point mutations. This study is further illustrative of a rare example of a conformationally flexible lipase about which NMR can correlate structure and function. Our results reveal a pre-existing communication route among amino acid side chains mediated by hydrogen-bonding and aromatic interactions and involving His-54, Arg-57, Tyr-58, and His-272 residues that are responsible for regulating hMGL activity. It is likely that these conformational transitions involve opening and closing of the lid domain. The data presented invite further experiments to ascertain the direct contribution of lid-gating dynamics to the transitions observed and the extent to which lid motion may be stabilized by substrate, product, and/or active sitedirected inhibitor. Along with such information, the present study provides insight into the regulation of hMGL function useful to design novel chemical strategies for selective inhibition of this therapeutic target.
Author Contributions-A. M. was the principal investigator and conceived the study. D. R. J. prepared the manuscript. S. P. interpreted the data and prepared the manuscript. N. Z. provided the mutagenesis of hMGL and interpreted the data. G. R. performed the expression and purification of MGL mutants, enzyme assays, and NMR experiments. I. K. performed the expression and purification of MGL mutants, enzyme assays, and NMR experiments. S. T. participated in all aspects of the work and prepared the manuscript.