cDNA Cloning, Tissue Distribution, and Identification of the Catalytic Triad of Monoglyceride Lipase

Monoglyceride lipase catalyzes the last step in the hydrolysis of stored triglycerides in the adipocyte and presumably also complements the action of lipoprotein lipase in degrading triglycerides from chylomicrons and very low density lipoproteins. Monoglyceride lipase was cloned from a mouse adipocyte cDNA library. The predicted amino acid sequence consisted of 302 amino acids, corresponding to a molecular weight of 33,218. The sequence showed no extensive homology to other known mammalian proteins, but a number of microbial proteins, including two bacterial lysophospholipases and a family of haloperoxidases, were found to be distantly related to this enzyme. By means of multiple sequence alignment and secondary structure prediction, the structural elements in monoglyceride lipase, as well as the putative catalytic triad, were identified. The residues of the proposed triad, Ser-122, in a GXSXG motif, Asp-239, and His-269, were confirmed by site-directed mutagenesis experiments. Northern blot analysis revealed that monoglyceride lipase is ubiquitously expressed among tissues, with a transcript size of about 4 kilobases.

The sequential hydrolysis of stored triglycerides in adipose tissue is the result of a combined action of two lipases, hormone-sensitive lipase and monoglyceride lipase (MGL 1 ; EC 3.1.1.23). Hormone-sensitive lipase catalyzes the first and ratelimiting step, the hydrolysis of triglycerides, and also the subsequent hydrolysis of di-and monoglycerides (1). Hormonesensitive lipase has a marked, although not absolute, preference for the primary ester bond of glyceride substrates. It has been shown that MGL is required to obtain a complete degradation of monoglycerides to fatty acids and glycerol, i.e. in the absence of MGL there is an accumulation of monoglycerides (mainly 2-monoglycerides) (2). The main physiological role for MGL is probably to assure complete hydrolysis of monoglycer-ides formed during the lipolysis of stored triglycerides of the adipocyte. Another role for the enzyme could be to catalyze the hydrolysis of 2-monoglycerides formed as a result of lipoprotein lipase-catalyzed hydrolysis of triglycerides from chylomicrons and very low density lipoproteins. Lipoprotein lipase has monoglyceride-hydrolyzing activity, with an absolute preference for the primary ester bond (3). This lipase could therefore catalyze the hydrolysis of 1(3)-monoglycerides, which are formed through isomerization from 2-monoglycerides. However, since the rate of isomerization at pH 7.4 is low, it is more likely that a substantial fraction of the 2-monoglycerides, formed through the action of lipoprotein lipase, is transported into the adipocyte and hydrolyzed by MGL (4). It should be pointed out that besides these two enzymes, there is no evidence for any other monoglyceride-hydrolyzing activity of adipose tissue.
MGL has been extensively purified from rat adipose tissue in our laboratory (5). The limited amounts of purified enzyme obtained have been used to study some of its enzymological and biochemical properties. These studies have shown that MGL is a 32.9-kDa protein. Nonionic detergent is strictly required to purify MGL from adipose tissue and to keep it stable in aqueous solution in the purified state, indicating that the enzyme has an amphiphilic character, as has also been shown for hormone-sensitive lipase (6). With regard to enzymological properties (5), MGL appears to be a specific monoacylglycerol hydrolase, hydrolyzing the 1(3)-and 2-ester bonds at equal rates. Inhibition by micromolar concentrations of diisopropylfluorophosphate indicates the presence of a reactive serine at the active site, as is the case for many lipases and esterases. Its activity is also extremely sensitive to inhibition by p-chloromercuribenzoic acid and mercury chloride, suggesting the presence of one or several essential sulfhydryl groups.
To date, many lipases and esterases have been cDNA cloned. Several of the known sequences show identity to each other, indicating evolutionary relationships and allowing the description of different families of lipases/esterases. However, no cDNA containing the entire coding sequence for a specific monoglyceride-hydrolyzing enzyme from any species or tissue has yet been described.
With regard to the three-dimensional structure of lipases, much has been learned in the recent years through x-ray crystallographic studies of several microbial lipases and one mammalian lipase, pancreatic lipase. The structures of these lipases show that they share a similar three-dimensional fold, called the ␣/␤-hydrolase fold (7). This fold consists of a central ␤sheet, surrounded by a variable number of ␣-helices. In turns between ␤-strands and ␣-helices, the catalytic triad, composed of a serine, a carboxylic acid, and a histidine, is found. The serine of the catalytic triad is, with a few exceptions, found within a GXSXG consensus sequence. The order of the residues of the catalytic triad in the primary sequence is serine followed by the carboxylic acid and the histidine in all lipases where the primary sequence as well as the residues of the catalytic triad are known (8,9).
Despite the fair number of known lipase structures, including some structures solved of lipases complexed to substrates and substrate analogues, and several molecular models of lipases, little has been learned regarding the structural basis for substrate specificity of lipases. As part of a long term goal to understand the relationship between structure and substrate specificity, we decided to cDNA clone the MGL of adipose tissue. As stated above, this enzyme has been shown to be a specific monoacylglycerol hydrolase with no preference for either isomer. MGL should therefore be a good model enzyme for this type of study.
We describe in this report the complete amino acid sequence of mouse adipose tissue MGL. The sequence, which represents the first known MGL sequence, reveals that this enzyme is related to a number of microbial proteins that include esterases, lysophospholipases, and haloperoxidases. By means of sequence alignments and site-directed mutagenesis experiments, we have identified the residues of the catalytic triad and localized the secondary structure elements that constitute the ␣/␤-hydrolase fold of this lipase. Finally, we show that the cloned MGL is not specific for adipose tissue, but seems to be widely expressed among tissues.

EXPERIMENTAL PROCEDURES
Purification of Rat MGL-To obtain large quantities of pure enzyme protein, a modification of our previously described purification scheme was employed (5). In brief, fractions containing active MGL from the QAE-Sephadex chromatography step of multiple purifications of hormone-sensitive lipase from epididymal rat fat pads (from a total of 5000 rats) were pooled (1). This pool (6 liters), contained in a buffer with the nondialyzable, nonionic detergent C 13 E 12 2 and glycerol, was concentrated 8-fold by ultrafiltration (Filtrone Minisette, Omega 30K filter). The concentrate was divided into two portions, each of which was subjected to isoelectric focusing with an Ampholine gradient of pH 3.5-10, using a 440-ml column (LKB 8100 Ampholine column, Pharmacia Biotech). Peak activity fractions from both purifications were pooled and subjected to a second isoelectric focusing step. The peak containing the active MGL was pooled, dialyzed, and concentrated to 35 ml by ultrafiltration (PM-30 Diaflo membrane). This material was loaded into the bottom of a Sephadex G-200 superfine column (5 ϫ 80 cm, Pharmacia Biotech) and eluted with reverse flow (6 ml/h), to prevent the sample with high density and viscosity from passing through the column by gravity. MGL activity was eluted as a single peak, and the corresponding fractions were pooled and ultrafiltrated to 55 ml. Finally, this material was included in an isoelectric focusing column (LKB 8110, 110 ml) with an Ampholine gradient of pH 5-8. The purity of the 32.9-kDa MGL protein in the peak fractions was checked by SDS-PAGE according to Laemmli (10) and silver staining. MGL activity was measured using mono-[ 3 H]olein (MO) as substrate (11).
Internal Peptide Sequence Analysis-Forty l of fraction 55 ( Fig. 1), containing 100 pmol of MGL protein, were reduced and alkylated with dithiothreitol and iodoacetamide, respectively. The sample was subjected to SDS-PAGE (10), stained with 0.05% Coomassie Blue in 50% methanol, 7% acetic acid for 5 min and then destained in the same solution without dye. The band containing MGL was excised, transferred to an Eppendorf tube, and subjected to in-gel digestion according to Hellman et al. (12). In brief, the gel piece was washed with 0.2 M ammonium bicarbonate and 50% acetonitrile and then completely dried. During rehydration, 0.5 g of modified trypsin (sequence grade; Promega) was added, and 0.2 M ammonium bicarbonate was given in small aliquots until the gel piece was immersed. After overnight incubation, the supernatant was saved and combined with extractions from the gel piece. Generated peptides were isolated by reversed phase HPLC on a RPC C2/C18 SC 2.1/10 column in a SMART System (Pharmacia Biotech). Peptides were sequenced on a model 470A sequencer (Applied Biosystems), following the manufacturer's instructions.
cDNA Clone Isolation and Sequencing-Total RNA was prepared from mouse adipose tissue by guanidinium-isothiocyanate extraction (13), and poly(A) ϩ RNA was further isolated using the Dynabeads mRNA purification kit (Dynal). A double-stranded cDNA library was prepared from the poly(A) ϩ RNA (1 g), using the Marathon cDNA amplification kit (CLONTECH). Sense and antisense oligonucleotides, corresponding to four of the tryptic peptides obtained (peptides 2, 3, 4, and 7, Fig. 2), were designed in accordance with codon frequency usage data (14). PCR amplifications with all possible combinations of oligonucleotides were performed using the Long Template PCR system (Boehringer Mannheim) with the following cycling parameters: 94°C for 2 min, followed by 30 cycles, each consisting of denaturation at 94°C for 10 s, annealing at 55°C for 30 s, and elongation at 68°C for 3 min. The PCR products were purified by agarose gel electrophoresis and the WIZARD PCR Preps DNA purification system (Promega). Sequencing 2 C 13 E 12 is a heterogeneous preparation of an alkyl polyoxyethylene ether detergent with the indicated average composition, where C represents alkyl carbons and E represents oxyethylene units (Berol 058, Berol kemi AB, Stenungssund, Sweden).
FIG. 1. Purification of rat adipose tissue MGL. Enzyme activity was measured for the fractions eluted from the last isoelectric focusing step in the purification scheme (A). Fractions corresponding to the peak of enzyme activity were analyzed for purity of the 32.9-kDa MGL protein by SDS-PAGE and silver staining (B). Fraction 55 (circled in A and underlined in B) was used for tryptic peptide sequence analysis and N-terminal sequencing.
of the products was carried out using the ABI PRISM Dye Terminator Cycle Sequencing Core kit (Perkin-Elmer) and a model 373A sequencer (Applied Biosystems). The largest fragment obtained, a 519-bp fragment, was used as a probe to screen a mouse adipocyte gt11 cDNA library (CLONTECH). One million recombinant phage clones were transferred to nylon membranes (Stratagene) and hybridized overnight at 42°C with the 519-bp fragment, internally radiolabeled with [ 32 P]dCTP. Each filter was washed twice in 2 ϫ SSC, 0.1% SDS at room temperature, followed by two washes in 0.1 ϫ SSC, 0.1% SDS at 55°C. Positive phage clones were purified, and inserts were released with EcoRI and subcloned into pBluescript SK (Stratagene). Sequence determination of both strands was performed as described above.
The open reading frame and the deduced amino acid sequence were determined using the GeneWorks program (IntelliGenetics). Homology searches were performed against several DNA and protein data bases using a BLAST program (15).
Transient Expression of MGL in COS Cells-The coding part of the MGL cDNA sequence (Fig. 3) was amplified by PCR using 27-oligomers, containing XmaI restriction enzyme recognition sites, as primers. Vent DNA polymerase (New England Biolabs) was used in the PCR amplification with 30 cycles, each consisting of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 1 min. The PCR product was digested with XmaI and purified using the WIZARD PCR Preps DNA purification system. After subcloning into the eukaryotic expression vector pCI-neo (Promega), sequencing of both strands was performed to ensure the absence of PCR mistakes. Transfection, harvesting, and homogenization was performed as described previously (16), using 5 g of DNA/culture dish. MGL lipase activity of the recombinant protein was measured using MO as substrate (11) and esterase activity using p-nitrophenyl butyrate (PNPB) as substrate (17). Total protein in the homogenates was estimated according to Bradford with bovine serum albumin as standard (18). Furthermore, activity measurements against substrates of cholesterol oleate, triolein, or 1(3)-monooleoyl-2-O-oleylglycerol (a diglyceride analogue), at pH 7.0 and 8.0, were performed as described (1,19,20) with modifications (21).
Identification of the Catalytic Site Serine, Aspartic Acid, and Histidine by Site-directed Mutagenesis-Mutations in the MGL cDNA sequence, encoding Ser or His to Ala and Asp to Asn (S122A, H269A, H272A, H284A, H292A, D239N, and D243N; see Fig. 3) were constructed using the PCR overlap extension technique (22), with 21oligomers as mutagenic primers. The constructs were subcloned into pCI-neo, sequenced, and expressed in COS cells, and homogenates of harvested cells were analyzed for MO-and PNPB-hydrolyzing activity (see above).
Northern Blot Analysis-Poly(A) ϩ RNA from rat adipose tissue, ovary, and adrenal gland was isolated using the Dynabeads mRNA direct purification kit (Dynal). The mRNA (1 g) was electrophoresed under denaturing conditions in 2.2 M formaldehyde and blotted onto a nylon membrane (Stratagene). This blot and a rat multiple tissue Northern blot (2 g poly(A) ϩ RNA/lane, CLONTECH) were hybridized with a probe corresponding to the coding part of the MGL cDNA, internally radiolabeled with [ 32 P]dCTP, using the ExpressHyb System (CLONTECH). Membranes were analyzed by digital imaging using a Fujix Bas 2000 (Fuji).

RESULTS
Cloning of Mouse MGL cDNA-Purified MGL was obtained in large quantities by upscaling and slightly modifying the original procedure described by us (5). The original procedure included detergent solubilization of a pH 5.2 precipitate from a fat-free 110,000 ϫ g infranatant of a rat adipose tissue homogenate, ion-exchange chromatography, gel exclusion chromatography, and finally two sequential isoelectric focusing steps, using a pH 6 -8 gradient. For the present purification, our starting material consisted of pooled fractions containing MGL activity from the QAE chromatography step (corresponding to the ion exchange chromatography step mentioned above) of several purifications of hormone-sensitive lipase (see Fig. 1 in Ref. 1), which had been saved and stored at Ϫ20°C. Concentration of this material (6 liters) in a buffer with a nondialyzable, nonionic detergent and glycerol to a volume of 35 ml was achieved by ultrafiltration and successive preparative isoelectric focusing steps in a matrix-free sucrose gradient. The MGL protein, but not the detergent, was concentrated and purified (see "Experimental Procedures"). The last two purification steps were gel exclusion chromatography and isoelectric focus- The tryptic peptide sequences obtained from the purified rat enzyme are underlined. The serine, the two aspartic acids, and the four histidines, which were mutated to probe the residues of the catalytic triad, are marked with circles, where the closed circles indicate the residues predicted to form the catalytic triad. The HG dipeptide is boxed.
ing with a pH 5-8 gradient, essentially according to the original procedure (see Figs. 2 and 5 in Ref. 5). Peak MGL activity fractions (Fig. 1) from the last isoelectric focusing step showed sufficient purity (Ͼ75%) to allow recovery of the 32.9-kDa MGL protein from SDS-PAGE for in-gel trypsinization and peptide sequencing.
The purified protein was cleaved with trypsin, and seven of the peptides obtained after HPLC separation were sequenced (Fig. 2). The Marathon cDNA amplification kit was used for PCR amplification of MGL sequences from mouse adipose tissue mRNA, with specific primers designed from the tryptic peptides of rat MGL. The largest fragment obtained was 519 bp long (the sense oligonucleotide derived from peptide 4 and the antisense oligonucleotide derived from peptide 3; Figs. 2 and 3). The deduced amino acid sequence was found to contain peptides 1,5, and 7 thus confirming the identity of the cDNA. Upon screening a mouse fat cell cDNA library with the 519-bp fragment as a probe, several positive clones were identified. The three largest inserts from these phage clones were subcloned into pBluescript SK and sequenced. The sequences were found to be overlapping, and together they represent the entire coding region. The nucleotide sequence and the predicted amino acid sequence of mouse MGL are presented in Fig. 3. The seven tryptic peptides of rat MGL were all identified in the mouse MGL sequence. The identity of these peptides to the deduced amino acid sequence of mouse MGL is 95%, indicating a high degree of conservation between rat and mouse MGL. The ATG (nucleotides 1-3) is suggested to be the translation initiation codon, since it is the first and only ATG in an open reading frame before the first identified tryptic peptide. In addition, the assignment of this ATG as the translation initiation codon is supported by some preliminary results from N-terminal se- FIG. 4. Amino acid sequence alignment of MGL with several related microbial proteins. The sequences have been divided into three groups according to their homology. The first group (top) includes non-heme haloperoxidases and two bacterial esterases; the second group (middle) includes MGL and four hypothetical proteins deduced from DNA sequences in GenBank TM ; and the last group (bottom) includes two bacterial lysophospholipases. The alignments were done with the Pileup program from the GCG package (26) and were manually adjusted to gather the gaps in loop regions of the proteins, by comparison with the crystal structure of BPA2_STRAU (24). The secondary structure elements in this protein are indicated. A secondary structure prediction was obtained for MGL with the PredictProtein PHD program (25), using the alignment between MGL and the hypothetical cowpox protein as the input data. All of the predicted secondary structure elements are indicated in the MGL sequence (␤-strands are underlined, and ␣-helices are boxed). Identical residues or conservative substitutions between MGL and at least four other proteins are indicated with a dark gray background (identical residues are highlighted in boldface type). Amino acids considered as similar are: Asp and Glu (D and E); Arg and Lys (R and K); Ile, Val, and Leu (I, V, and L); Phe, Tyr, and Trp (F, Y, and W); and Ser and Thr (S and T). The residues of the putative catalytic triad are marked with an arrowhead. The area shadowed in light gray delimits a highly variable region, in which the sequences are very divergent. Sequences are named with the Swiss-Prot identification when available. PRXC_PSEPY, chloroperoxidase from Pseudomonas pyrrocinia; PRXC_STRLI, chloroperoxidase from Streptomyces lividans; ESTE_PSEFL, arylesterase from Pseudomonas fluorescens; BPA1_STRAU and BPA2_STRAU, bromoperoxidase A1 and A2, respectively, from S. aureofaciens; YKJ4_YEAST, hypothetical 35.5-kDa protein from S. cerevisiae; PLDB_ECOLI, lysophospholipase L2 from E. coli; PLDB_HAEIN, probable lysophospholipase L2 from H. influenzae. The hypothetical proteins deduced from GenBank TM entries are as follows. PseuPut, esterase from P. putida (27); Cowpox, gene M5L from cowpox virus (accession number X94355); AraTha, lysophospholipase isolog from A. thaliana (accession number U95973); MycGen, magnesium-chelatase homologue from M. genitalium (accession number L43967). quencing of rat MGL, showing that proline is the N-terminal residue, followed by glutamic acid (data not shown).
Based on the deduced amino acid sequence and the preliminary assignment of proline as the N-terminal amino acid (assuming the same processing of the mouse and rat enzyme), mature mouse adipocyte MGL is predicted to be composed of 302 amino acids and to have a molecular weight of 33,218. This is very similar to the molecular mass of 32.9 Ϯ 0.4 kDa for rat adipocyte MGL, determined by SDS-PAGE (5).
Two lipase motifs were identified in the primary sequence, the active site serine motif GXSXG and the HG dipeptide. The latter is found in many lipases, 70 -100 amino acids N-terminal of the catalytic site serine (23). Data base searches revealed sequence identities between MGL and a number of microbial proteins. 3 An amino acid sequence alignment for some of these proteins is shown in Fig. 4.
Transient Expression of MGL in COS Cells and Probing of the Catalytic Triad by Site-directed Mutagenesis-The MGL coding sequence was subcloned into the eukaryotic vector pCIneo for transient expression in COS cells. Homogenates of COS cells transfected with the MGL/pCI-neo construct exhibited high levels of MO-and PNPB-hydrolyzing activity compared with COS cells transfected with the pCI-neo vector alone (Fig.  5). As expected, MGL exhibited practically no catalytic activity against a diglyceride, a triglyceride, and a cholesterol ester substrate (Ͻ2% compared with MO-hydrolyzing activity; data not shown).
The alignment shown in Fig. 4 suggests that the catalytic triad of MGL is formed by Ser-122, Asp-239, and His-269. To verify this triad, the three residues were individually mutated, and the mutant proteins were expressed in COS cells. A number of control mutations were analyzed in parallel. These included all histidines present downstream from the active site serine and Asp-243, which is highly conserved in the MGL subfamily and in the lysophospholipases (Fig. 4). As shown in Fig. 5, mutating Ser-122, Asp-239, or His-269 completely abolished both the lipase and esterase activity of MGL, whereas all of the other mutant proteins retained catalytic activity.
Tissue Distribution of MGL mRNA-A number of rat tissues, including adipose tissue, adrenal gland, ovary, heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testis, were analyzed for the presence of MGL mRNA by Northern blot analysis (Fig. 6). A single mRNA transcript of ϳ4 kilobases was identified in adipose tissue and all other tissues examined. DISCUSSION In the present study, we have isolated and characterized the cDNA for mouse MGL by screening an adipocyte cDNA library (Fig. 3). The deduced protein sequence for mature MGL is 302 amino acids long, corresponding to a molecular weight of 33,218. To our knowledge, this is the first described sequence of a specific monoglyceride-hydrolyzing enzyme.
The closest relative to MGL found in the data base is a hypothetical protein, encoded by the genome of cowpox virus (GenBank TM accession number X94355). The sequence of this protein shows more than 40% identity with MGL. Initial data base searches with BLAST (15) revealed a number of bacterial and yeast proteins that show between 20 and 25% sequence identity with MGL at the amino acid level. These include an esterase from Pseudomonas putida, a hypothetical 35.5-kDa protein from Saccharomyces cerevisiae, one protein each from Arabidopsis thaliana and Mycoplasma genitalium, and one lysophospholipase each from Escherichia coli and Hemophilus influenzae. Furthermore, the esterase from P. putida is, in fact, more closely related to a family of haloperoxidases. Thus, MGL shows a distant evolutionary relationship to esterases, lysophospholipases, and haloperoxidases (Fig. 4). The three-dimensional structure for one member of the haloperoxidase family, the bromoperoxidase from Streptomyces aureofaciens (BPA2_STRAU), has been solved by x-ray crystallography (24), showing that these proteins have the ␣/␤-hydrolase fold char- 3 Recently, an entry has been deposited in GenBank TM (U67963), which most likely corresponds to the human MGL. However, it was not identified as such but rather as a human homologue to an ectyromelia virus protein that shows sequence similarity to E. coli lysophospholipase.  6. Tissue distribution of rat MGL mRNAs. A rat multiple tissue Northern blot (2 g of poly(A) ϩ mRNA/lane; CLONTECH) and a blot containing 1 g of mRNA from adipose tissue, adrenal gland, and ovary, respectively (obtained with an mRNA direct purification kit (Dynal), electrophoresed in a 2.2 M formaldehyde, 1% agarose gel, and blotted onto a nylon membrane) were hybridized with a 32 P-labeled MGL cDNA, corresponding to the complete coding region. MGL mRNA size was estimated by comparison with RNA size standards (Promega). kb, kilobases. acteristic for lipases and esterases (7). By comparing the amino acid sequence of the BPA2_STRAU protein and MGL (Fig. 4), the secondary structure elements that constitute the ␣/␤-hydrolase fold of MGL could be located in the primary sequence. When an alignment of MGL and the cowpox protein was used as the input data, all of these elements were correctly predicted by the PredictProtein PHD secondary structure prediction program (25). The information provided by Fig. 4 enabled us to build a partial three-dimensional model for MGL, by standard homology modeling techniques (not shown). In addition, the catalytic triad was identified, not only of MGL but also of all the related proteins. The results from the site-directed mutagenesis experiments (Fig. 5) were in complete agreement with the triad suggested by the alignment. These experiments were of particular relevance for the unambiguous identification of the aspartic acid of the triad, since the aspartic acid is not conserved in the hypothetical protein of the cowpox virus, which has an asparagine in that particular position (Fig. 4). Furthermore, the two lysophospholipases have a glutamic acid instead of an aspartic acid. This conserved substitution is commonly found among other families of lipases/esterases, e.g. the carboxylesterase B family (8), and only requires slight rearrangements of the side chains to allow the correct geometry of the triad. On the other hand, the presence of a highly conserved aspartic acid in the MGL group and in the lysophospholipases (Asp-243 in MGL, Figs. 3 and 4), raised the possibility that this was the residue involved in the catalytic triad. From a structural point of view, this would only mean a longer connecting loop from ␤7. However, as shown in Fig. 5, Asp-239 is essential for MGL activity, whereas Asp-243 is not.
From the alignment shown in Fig. 4 and from the known structures of BPA2_STRAU (24) and other esterases and lipases, the structures of these proteins can be clearly divided into two conceptual modules: a central core harboring the essential elements of the ␣/␤-hydrolase fold, including the catalytic machinery, and an external region located between strands ␤6 and ␤7 (shadowed in light gray in Fig. 4). The core module shows a degree of homology compatible with the maintenance of the ␣/␤-hydrolase fold. On the contrary, the other module displays a high degree of variability among the different proteins. This modular division of the structure has provided an efficient way to generate a large superfamily of hydrolytic enzymes with a very broad substrate specificity.
The MGL transcript was observed in all tissues examined (Fig. 6), indicating that MGL functions as a widespread intracellular monoglyceride-hydrolyzing enzyme. An intriguing observation is that the MGL coding sequence only represents approximately 25% of the transcript length.
In conclusion, we have cloned and described the primary sequence of adipose tissue MGL, provided structural information based on sequence comparison with a distantly related family of enzymes, identified and probed the residues of the catalytic triad, and shown that MGL mRNA is constitutively expressed in the body. The described cDNA sequence of mouse MGL provides, for the first time, possibilities to perform structure-function relationship studies of this enzyme. Furthermore, after large scale expression and purification, structure determination should be feasible. This will hopefully provide further insight into the relationships between structure and substrate specificity.