Simplified assays of lipolysis enzymes for drug discovery and specificity assessment of known inhibitors

Lipids are used as cellular building blocks and condensed energy stores and also act as signaling molecules. The glycerolipid/ fatty acid cycle, encompassing lipolysis and lipogenesis, generates many lipid signals. Reliable procedures are not available for measuring activities of several lipolytic enzymes for the purposes of drug screening, and this resulted in questionable selectivity of various known lipase inhibitors. We now describe simple assays for lipolytic enzymes, including adipose triglyceride lipase (ATGL), hormone sensitive lipase (HSL), sn-1-diacylglycerol lipase (DAGL), monoacylglycerol lipase, α/β-hydrolase domain 6, and carboxylesterase 1 (CES1) using recombinant human and mouse enzymes either in cell extracts or using purified enzymes. We observed that many of the reported inhibitors lack specificity. Thus, Cay10499 (HSL inhibitor) and RHC20867 (DAGL inhibitor) also inhibit other lipases. Marked differences in the inhibitor sensitivities of human ATGL and HSL compared with the corresponding mouse enzymes was noticed. Thus, ATGListatin inhibited mouse ATGL but not human ATGL, and the HSL inhibitors WWL11 and Compound 13f were effective against mouse enzyme but much less potent against human enzyme. Many of these lipase inhibitors also inhibited human CES1. Results describe reliable assays for measuring lipase activities that are amenable for drug screening and also caution about the specificity of the many earlier described lipase inhibitors.

HEK 293T cells were cultured in DMEM high glucose, 4 mM L -glutamine without sodium pyruvate (HyClone, Logan, UT), supplemented with 10% FBS (Gibco, Grand Island, NY). For lipase expression, 9 × 10 5 cells were cultured overnight in a 10 cm Petri dish and then transfected with 20 g of lipase expression plasmids using 20 l of lipofectamine 2000 (Invitrogen, Carlsbad, CA) in 3 ml of Opti-MEM media (Invitrogen) and 10 ml of above DMEM. After 24 h, transfection medium was replaced by DMEM. Cells were harvested 72 h posttransfection in the corresponding lysis buffers described below, for different lipases after two washes in PBS.

ATGL assay
Human and mouse ATGL were expressed separately in 293T cells. Cell extracts were prepared in 1 ml of buffer A (50 mM HEPES, pH 7.2, 100 mM NaCl, 5 mM CaCl 2 , 0.5 mM DTT, 2% DMSO, 0.1% Triton X-100), by three cycles of freeze (in liquid nitrogen)/thaw sonication at high power in a cup horn sonicator (Misonix Sonicator 3000; Qsonica LLC, CT) for 5 min, followed by centrifugation at 1,000 g for 5 min at 4°C, and the supernatant was used as the source of ATGL after measuring protein concentration. Aliquots containing 2-4 µg/µl of proteins were stored at Ϫ 80°C. ATGL expression was verifi ed by Western blotting using antibodies against human (Cell Signaling #2138) or mouse ATGL (Cell Signaling #2439). ATGL activity was assayed using EnzChek lipase substrate (Life Technologies) ( 21 ). Control cell extracts prepared from empty vector-transfected cells were used to ascertain that the lipase activity measured is due to the overexpressed ATGL. Assays were performed in 96-well opaque black plates (Corning #3915) containing 30 g of ATGL cell extract in 90 l of buffer A, to which 5 l of test inhibitor (diluted in 30% DMSO to appropriate working stock concentrations to achieve different inhibitor concentrations) was added to give the required fi nal concentration. After 30 min preincubation at room temperature with 700 rpm orbital shaking, 5 µl of 20 M EnzChek lipase substrate working solution was added to each well to a fi nal concentration of 1 µM to start the reaction at 37°C. EnzChek lipase substrate stock solution (1 mM) was prepared in DMSO and then diluted 1:50 in buffer A, just before use. Final concentration of DMSO was kept at 5% (v/v) in all the wells. Fluorescence (excitation 485 nm; emission 510 nm) was recorded every 30 s for 60 to 90 min with 2 s of shaking preceding each reading. ATGL activity was calculated by subtracting background activity (no enzyme added), using the linear portion of the velocity curve, after the fi rst 15 min of the reaction. Endogenous ATGL activity in extracts prepared from non-or empty vector-transfected cells was found to be low and did not contribute signifi cantly to the activity measured with ATGL enzyme extracts.

HSL assay
Human and mouse HSL proteins were expressed separately in 293T cells. Cell extracts were prepared in PBS as described above for ATGL. Human and mouse HSL expression was verifi ed using HSL antibody (Cell Signaling #4107) in Western blots. Alternatively, we also tested commercially available purifi ed hHSL for some assays. The assay conditions were similar to those described below for ABHD6, except that potassium phosphate buffer was adjusted to pH 7.0 and 1 µg cell extract/well was used. Because HSL is capable of hydrolyzing 1-MAG, we used 1-S -arachidonoylthioglycerol as the substrate and measured the release of thioglycerol at 37°C using ThioGlo-1 (see ABHD6 Assay below for more details).

sn -1-DAGL assay
DAGL ␣ could not be expressed using the plasmid pCMV6-XL5-hDAGL ␣ from OriGene in 293T cells. The DAGL ␣ cDNA 2-arachidonoylglycerol, has been implicated in the control of pain mechanisms ( 13 ) and cancer cell proliferation ( 14 ). DAGL enzymes have been found to be important in the production of 2-arachidonoylglycerol and in the control of appetite and pain ( 15 ) and also in the regulation of Ca 2+ infl ux into cells ( 16 ). We recently showed that 1-MAG is a coupling factor linking glucose metabolism to insulin secretion in pancreatic ␤ -cells and that ABHD6 specifi cally controls ␤ -cell 1-MAG levels ( 4 ).
Various compounds have been used to inhibit particular lipolytic enzymes, assuming that the inhibitor is specifi c. However, no systematic studies have been done comparing the specifi city of many of these inhibitor compounds and whether their effectiveness differs among species. These issues make a signifi cant impact on our understanding of lipid metabolism in different organisms and also on the discovery of drugs targeting these enzymes for a possible use against human diseases. It is necessary to have reliable assay procedures to measure the activity of an enzyme before embarking on identifying its physiological role and also for the development of drugs targeting the particular enzyme.
In the present study, we describe reliable assays of lipolytic enzymes using recombinant human and mouse enzymes either in cell extracts or using purifi ed enzymes. Our fi ndings reveal that some of the inhibitors available commercially as "specifi c" against a particular enzyme are in fact not specifi c and also that certain inhibitors discovered using rodent enzymes are ineffective against their human counterparts.

ABHD6 assay
hABHD6 was expressed in 293T cells by transfecting with pCMV6-AC-hABHD6 plasmid. Cell extracts were prepared in PBS as described above for ATGL. ABHD6 expression was verifi ed using ABHD6 antibody (kindly provided by Dr. J. Mark Brown, Cleveland, OH) in Western blots. Aliquots containing 4 µg/µl of protein were stored at Ϫ 80°C. ABHD6 assay was based on the hydrolysis of 1-S -arachidonoylthioglycerol to release 1-thioglycerol, which spontaneously reacts with ThioGlo-1 to form a fl uorescent adduct ( 23 ). Assays were performed in 96-well opaque black plates containing 90 µl of 50 mM potassium phosphate buffer, pH 7.2 and 5 µl of test inhibitor in 30% DMSO. The plates were then preincubated for 30 min at room temperature, after which 5 µl of premixed 100 µM 1-S -arachidonolythioglycerol and 0.26 mM ThioGlo-1 in 20% DMSO was added to all but the gain and background wells. These wells received the Thio-Glo-1 only. The plates were immediately placed in the plate reader at 37°C and shaken for 10 s, and fl uorescence was continuously recorded at excitation 380 nm and emission 510 nm for 45 min at 30 s intervals. Sensitivity of the detection was adjusted via modifying photomultiplier tube gain from wells containing 1 µM thioglycerol only. Final DMSO concentration in all wells was kept at 2% (v/v). Another control to measure endogenous ABHD6 activity was also set up using cell extract prepared from empty vector-transfected 293T cells. Activity was calculated from the linear portion of the curve after background correction.
In order to compare the overall sensitivity of ABHD6 assays using the artifi cial substrate 1-S -arachidonoylthioglycerol and the natural substrate 1-palmitoylglycerol, extracts from hABHD6overexpressing 293T cells were used, and extracts from untransfected 293T cells were used as control. The assay system in a fi nal volume of 100 l contained 10 µg protein of extracts, 50 mM Tris-Cl pH 7.2, without or with 5 M WWL70 and either 5 µM 1-S -arachidonoylthioglycerol or 5 or 50 µM 1-palmitoylglycerol. Assay with 1-S -arachidonoylthioglycerol was described above. Incubations with 1-palmitoylglycerol were at 37°C for 30 min with constant mixing. Reactions were terminated by 5 µl 70% perchloric acid, and after complete mixing, pH was brought to 5-7 using 2 M KOH. Tubes were kept on ice 30 min and centrifuged for 10 min at 10,000 g . Supernatants were processed for measuring glycerol, released during hydrolysis of 1-palmitoylglycerol, by radiometric method ( 24 ).

CES1 assay
Purifi ed human CES1 (hCES1) was used to assay CES1 activity. The assay conditions were similar to those described above for ABHD6, with potassium phosphate buffer adjusted to pH 7.5 and an enzyme concentration of 60 ng/well purifi ed hCES1. 1-S -arachidonoylthioglycerol was used as a substrate, and the release of thioglycerol at 37°C was measured with ThioGlo-1. CES1 shows robust activity with many carboxyl esters including 1-MAG.

FAAH assay
Assay was performed using the FAAH inhibitor screening assay kit from Cayman Chemicals according to the manufacturer's protocol. Release of the fl uorescent product, 7-amino-4-methylcoumarin (AMC), from the hydrolysis of AMC-arachidonoyl amide, mediated by FAAH at 37°C, was recorded using a plate reader with an excitation fi lter of 340-360 nm and an emission fi lter of 450-465 nm. Inhibitors were preincubated for 30 min at indicated concentrations with the enzyme prior to the addition of substrate to initiate the reaction. Change in fl uorescence with time was taken as enzyme activity and the extent of inhibition was calculated from the initial rates. Positive control for FAAH inhibition was used using JZL195. Final DMSO concentration was kept at 2% (v/v) in the assay. harbors relatively long 5 ′ and 3 ′ untranslated regions (UTRs; 104 and 2,570 bp, respectively). In addition, the 5 ′ UTR has several out of frame ATG start sites. We observed a similar expression problem when using adenoviral vector with the same DAGL ␣ cDNA. We realized that the 5 ′ UTR interfered with protein expression. Thus, we have used a PCR-based strategy to remove both 5 ′ and 3 ′ UTRs, keeping only the coding sequence and the natural ATG start site. The coding sequence (3,129 bp long) from the pCMV6-XL5-hDAGL ␣ was amplifi ed by PCR using DAGL forward primer 5 ′ CAT CTA GAG CCA TGC TGC CCG GGA TCG TGG T and the reverse primer 5 ′ CAC TCG AGC TAG CGT GCT GAG ATG ACC A. The amplifi ed product was purifi ed and subcloned into pIRES2-EGFP (enhanced green fl uorescence protein) plasmid (Clontech). DNA sequencing was performed to confi rm in frame ligation of the construct. The plasmid pIRES2-DAGL ␣ thus generated was used to transfect 293T cells, and cell extracts were prepared in 300 µl of 0.25 M sucrose, 20 mM HEPES, pH 7.0 (buffer B). To this end, cells were lysed by three rapid cycles of freeze/thaw/sonication 2 min (10 s on/off cycles) at high power in ice-cold water in a cup horn sonicator, and the extract was centrifuged at 51,000 g for 30 min at 4°C. The membrane pellet was suspended in 200 µl buffer B by sonication, and the protein content was measured. Aliquots containing 2-5 µg/µl of protein were stored at Ϫ 80°C. DAGL ␣ expression was verifi ed using hDAGL ␣ antibody (Abnova #PAB11515) in Western blots. DAGL ␣ activity was assayed by following the hydrolysis of p-nitrophenylbutyrate (pNPB), as described earlier ( 22 ). Endogenous DAGL ␣ activity in extracts prepared from pIRES2-EGFP-transfected cells was found to be low and did not contribute signifi cantly to the activity measured with DAGL ␣ enzyme extracts. The assay system premix in a 96-well black plate with clear bottom (PerkinElmer Viewplate-96 F TC), in a volume of 90 l per well, contained 0.25 M sucrose, 50 mM HEPES, pH 7.3 (buffer C), and 10 g DAGL ␣ enzyme (cell membranes) extract to which 5 l of test inhibitor (diluted in 30% DMSO) was added. After 10 min preincubation at room temperature with 700 rpm orbital mixing and 20 min incubation at 37°C with mild shaking, 5 µl of freshly prepared 7.88 mM pNPB (in 70% DMSO in buffer C) was added to each well to start the reaction. The plates were shaken rapidly, and change in absorbance at 405 nm was measured every 30 s for 45 min. DAGL ␣ activity was calculated by subtracting background activity (wells without enzyme extract) using the linear portion of the velocity curve. The fi nal concentration of DMSO was adjusted to 5% in all wells. Activity was also measured using membrane extracts prepared from EGFPexpressing cells and was found to be negligible.
DAGL ␣ was also assayed using EnzChek lipase substrate described above for the ATGL assay. The assay system in a 96-well black plate was similar to the one used with pNPB above. However, reactions were started with 5 µl of 100 M EnzChek lipase substrate solution to a fi nal concentration of 5 µM. In parallel wells, CaCl 2 was added at 5 mM fi nal concentration. Enzyme was preincubated with 6 M KT109 DAGL inhibitor where indicated, prior to substrate addition. Fluorescence was recorded as described for the ATGL assay above.

MAGL assay
hMAGL was assayed using purified recombinant enzyme (Cayman Chemical). The assay conditions were similar to those described below for ABHD6 with potassium phosphate buffer adjusted to pH 7.4 and an enzyme dilution of 1:7,500 in potassium phosphate buffer. Final concentration of MAGL enzyme protein in the assay was 0.025 ng per well. 1-Sarachidonoylthioglycerol was used as the substrate, and the release of thioglycerol at 37°C was measured with ThioGlo-1 (see below). which is available as purifi ed recombinant enzyme commercially. We also used extracts of cells overexpressing either hHSL (118.3 kDa form; Fig. 1A ) or mouse HSL (87.3 kDa), for assessing inhibitor sensitivity. hHSL could be rapidly assayed in 96-well format, using 1-S -arachidonoylthioglycerol, an analog of 1-MAG, as substrate ( Fig. 1C ). Similar results were obtained with the mouse enzyme (not shown). Both human and mouse HSL enzymes were inhibited by Cay10499, although to different degrees (see below). This assay is adaptable for high-throughput screening.
ABHD6 assay. We recently demonstrated ( 4 ) that the previously suggested MAG hydrolase ABHD6 ( 6 ) is indeed a MAG hydrolase. Even though an assay procedure for measuring the activity of ABHD6 was described earlier, it was based on the measurement of glycerol release with 1-and 2-arachidonoylglycerol as substrates using a tedious coupled enzyme assay ( 5 ) and is not amenable for high-throughput screening. We currently describe a simple single-step assay using cell extracts containing recombinant hABHD6 ( Fig. 1A ) that does not involve any product isolation and is adaptable for high-throughput screening. The ABHD6 assay is based on the hydrolysis of 1-S -arachidonoylthioglycerol, an analog of 1-MAG, to release thioglycerol, which is detected by fl uorescent adduct formation with ThioGlo-1 ( Fig. 1E ). Recombinant hABHD6 is strongly inhibited by WWL70, an inhibitor described as an ABHD6-specifi c inhibitor ( 27 ). CES1 assay. hCES1 could also be assayed using 1-Sarachidonoylthioglycerol as substrate as this enzyme is capable of hydrolyzing a wide variety of carboxylic acid esters. Interestingly, we found that the MAGL inhibitor JZL184 was also a powerful inhibitor of CES1 ( Fig. 1F ), as recognized recently ( 28 ).
FAAH assay. A fl uorescence-based commercially available FAAH assay kit that uses recombinant human FAAH (hFAAH) and the hydrolysis of AMC-arachidonoyl amide was used. hFAAH showed total inhibition by JZL195 ( Fig. 1G ), a dual inhibitor of MAGL and FAAH ( 29 ), and also by JZL184, a MAGL inhibitor ( Table 1 ).
There is an ‫ف‬ 2-to 4-fold increase in corresponding lipase activity in extracts from cells overexpressing ATGL, ABHD6, DAGL ␣ , and HSL, measured using fl uorescence-based assays ( Fig. 1H ). Background activity measured in control empty vector-transfected cell extracts was used to correct the corresponding enzyme activity measured in overexpressed cell extracts while calculating the inhibitor specifi cities.

Expression of recombinant hDAGL ␣ and its assay
The commercially available plasmid and adenoviral constructs of hDAGL ␣ were not found to be effective for

Data analysis
Percentage of inhibition was calculated from lipase activity at initial rates measured using enzyme extracts treated with the inhibitors relative to the activity measured with vehicle-treated extracts. IC 50 values were determined using Prism version 5.01 (GraphPad Software, San Diego, CA) based on percentage of inhibition values. Results shown represent two to three separate experiments with triplicate observations.

RESULTS AND DISCUSSION
Simplifi ed assays for ATGL, HSL, MAGL, ABHD6, and CES1 Lipases are important in the metabolism of various lipids and thus play a critical role in the cell from membranogenesis to the generation of several signaling molecules. Several assays have been described before for different lipase enzymes, but many of them are cumbersome and use radioactive or custom-made substrates not available to all investigators. We have developed simplifi ed and reliable assays of different lipases using recombinant human and mouse enzymes either in the form of cell extracts or purifi ed enzymes. Even though the assays described here cannot be used directly for measuring the corresponding endogenous enzyme activities as many lipases have overlapping substrate specifi cities, they are adaptable for high-throughput screening of larger libraries of compounds. They may also be adapted for measuring endogenous lipase activity in case a highly specifi c inhibitor is available to distinguish its activity from other lipases. As we describe below, many such "specifi c" inhibitors previously described are indeed not specifi c.
ATGL assay. Using whole cell extracts from 293T cells overexpressing hATGL ( Fig. 1A ) or mouse ATGL, this enzyme activity could be assayed with EnzChek lipase substrate. We noticed that it is necessary to let the fl uorescence stabilize for the fi rst 15 to 20 min (not shown). The actual enzyme activity is recorded only after this initial stabilization, which is refl ected by steady increase in fl uorescence for at least 45 min. ATGL activity was almost completely inhibited by the panlipase inhibitor orlistat ( Fig. 1B ). Similar results were noticed with mouse ATGL, also expressed in 293T cells (data not shown). We also examined the need for added ATGL activator, CGI58 ( 3 ), for measuring ATGL activity. Coexpression of CGI58 along with ATGL or addition of separately expressed CGI58 to ATGL assay had no signifi cant effect on the measured ATGL activity (not shown), which suggested that under the present assay conditions, ATGL is fully active. This also simplifi es the assay procedure signifi cantly. This assay, which does not involve any radioactive materials or reaction product extraction, could be easily adapted to high-throughput screening for the discovery of selective inhibitors of hATGL. This assay could also be adapted for measuring ATGL activity in various tissues where the enzyme is well expressed.
HSL assay. hHSL exists in multiple isoforms ( 25 ), and in the present study, we used the long form (118.3 kDa), DAGL ␣ expression vector (pIRES2-DAGL ␣ ), using only the coding sequence from the commercial plasmid of DAGL ␣ without the 5 ′ and 3 ′ UTRs. Using 293T cells, we Extracts from cells not transfected (control for hABHD6 assays) or vector transfected (control for hDAGL, hHSL, and hATGL assays) were used as control cell extracts. Enzyme activities were measured as described in Materials and Methods using fl uorescence-based assays. N = 4 for all the assays, and all the activities were corrected for background (change in fl uorescence without added enzyme) in the corresponding assays. Results are expressed as fold increase above corresponding control extracts and are mean ± SEM; *** P < 0.001, compared with control. The actual activities (RFU × 10 type of assay procedures ( 27 ). Using the ABHD6 assay described here, we could screen a small library of 120 compounds for identifying selective inhibitors for this enzyme (not shown). The 120 compound library is a focused library developed based on the structure of WWL70, a known ABHD6 inhibitor. While most of these 120 compounds were inhibitory to ABHD6, identifi cation of the compounds that showed selectivity toward ABHD6 was achieved by testing the inhibitory potency of these compounds against other lipases using the assays described in the present work.

hCES1 activity and inhibition
Using the assay described here, we measured a K m of 1.29 M for 1-S -arachidonoylthioglycerol for purifi ed hCES1 ( Fig. 4A ). The classical MAGL inhibitor JZL184 ( 26 ) showed a strong inhibitory potency for CES1, with an IC 50 of 30 nM ( Fig. 4B ). In order to adapt this assay for cell extracts, it is necessary to eliminate other contaminating hydrolase activities. Screening of several inhibitors, supposedly specifi c for individual lipases, surprisingly revealed that all of these also inhibit CES1 ( Table 1 ). Thus, caution must be exercised when using these "specifi c" inhibitors for dissecting out the role of individual lipases/hydrolases in biological processes.

Inhibitor specifi cities of different lipases of human and murine origin
Using the new lipase assays, we examined the effectiveness of various commercially available and also recently described inhibitors against these enzymes ( Table 1 ). Some of the inhibitors were not found to be as specifi c as previously reported.

ATGL inhibitors.
A panlipase inhibitor, orlistat, was able to strongly inhibit hATGL, hDAGL ␣ , hMAGL, could express hDAGL ␣ in ample quantities ( Fig. 1A ), after transfection with pIRES2-DAGL ␣ . We noticed that both the 5 ′ and 3 ′ noncoding UTRs in the original DAGL ␣ cDNA somehow interfere with the expression of DAGL ␣ ; thus, they need to be removed. The recombinant hDAGL ␣ could hydrolyze the artifi cial substrate pNPB effectively, as described earlier ( 22 ), with a K m of 1.2 mM ( Fig. 2A ). This activity is suppressed by the recently described DAGL inhibitor ( 20 ) KT109 ( Fig. 2B ), indicating that the activity being measured is indeed due to DAGL ␣ . There was negligible activity with extracts prepared from control plasmid transfected cells (not shown). It is important to note that in order to minimize the contaminating activities, DAGL ␣ is measured in the cell membrane extracts and not in the whole cell extracts. Contrary to the Ca 2+ dependence of DAGL ␣ activity described earlier ( 7 ), we noticed that the enzyme activity with either pNPB or EnzChek lipase substrate as substrates is actually lower in the presence of added Ca 2+ ( Fig. 2C, D ). DAGL ␣ has been shown to act on EnzChek fl uorogenic lipase substrate ( 30 ). The reason for this apparent discrepancy on calcium dependency is not clear but may relate to the ability of calcium to complex and precipitate the released fl uorescent fatty acid or the yellow-colored pNPB. The recombinant DAGL ␣ cell membrane extracts showed good linearity of enzyme activity with increasing protein concentration ( Fig. 2E ).

ABHD6 activity and inhibition
Recombinant hABHD6 showed high affi nity for 1-Sarachidonoylthioglycerol, with a K m of 2.38 M ( Fig. 3A ). The recombinant ABHD6 also showed increased activity with increase in enzyme protein concentration (whole cell extract) ( Fig. 3B ), and this activity is strongly inhibited by WWL70, with an IC 50 of 55 nM ( Fig. 3C ), which is similar to the earlier reported values ( ‫ف‬ 70 nM), using different  ( 20 ) 0 0 0 0 9 0 0 9 0 Activities of recombinant human lipase enzymes were measured using either cell extracts with overexpressed enzymes (hATGL, hHSL, hDAGL ␣ , hABHD6) or purifi ed proteins (hMAGL, hFAAH, hCES1) as described in Materials and Methods. The dashes indicate "not determined." Values shown are average of two to six determinations. a specifi c ATGL inhibitor, ATGListatin, using recombinant mouse ATGL. We tested both of these inhibitors and surprisingly found that they do not affect hATGL ( Table 1 ). Furthermore, we found that hCES1 could be inhibited strongly by WWL64, whereas ATGListatin showed hABHD6, and hCES1. Orlistat's effect on hHSL was modest, and hFAAH was not inhibited. A recent study on serine hydrolase inhibitors ( 18 ) using rodent tissues showed that WWL64 is a specifi c inhibitor of patatin-like phospholipase domain containing 2 /ATGL. Another study ( 17 ) reported  hABHD6, as noted before ( 20 ), and also hCES1 ( Tables 1  and 2 ). The widely used DAGL inhibitor, RHC20867, actually barely inhibited hDAGL ␣ at 5 M but inhibited hHSL, hMAGL, hABHD6, and hCES1 to different degrees ( Table 1 ). DAGL activity with either pNPB or EnzChek lipase substrate was inhibited by KT109, and the presence or absence of Ca 2+ did not affect this inhibition ( Fig. 2C, D ) MAGL inhibitors. JZL184 was found to be relatively more specifi c to MAGL, even though it also strongly inhibited FAAH and CES1 ( Table 1 ), and to a lesser extent ABHD6, as noted earlier ( 26 ). However, JZL195, the dual inhibitor of MAGL and FAAH, also inhibited ABHD6, CES1, and, to a lesser extent, HSL.
ABHD6 inhibitors. WWL70 and KT195 were found to be relatively more specifi c to ABHD6, with only CES1 being inhibited among other lipases tested. moderate inhibition. Neither WWL64 nor ATGListatin affected other lipases tested.
HSL inhibitors. Similarly, we also examined the specifi city of known HSL inhibitors. Cay10499, which was previously shown to also inhibit hMAGL ( 31 ), was found to inhibit all the human lipases tested ( Table 1 and Table 2 ). An HSL inhibitor, WWL11, discovered using the activitybased protein profi ling approach against mouse HSL ( 18 ), was found to have much less effect on hHSL but significantly inhibited hCES1. Another mouse HSL-directed inhibitor, Compound 13f, described recently ( 19 ), also showed much less inhibition of hHSL, but it inhibited both hABHD6 and hCES1 ( Table 1 ).
DAGL ␣ inhibitors. Two recently reported inhibitors against mouse DAGL ␤ , KT109 and KT172, were found to inhibit hDAGL ␣ effectively, but these compounds also inhibited  Activities of recombinant human lipase enzymes were measured using either cell extracts with overexpressed enzymes (hATGL, hHSL, hDAGL ␣ , hABHD6) or purifi ed proteins (hHSL, hMAGL, hCES1) as described in Materials and Methods. To generate IC 50 values, inhibitors were used at concentrations ranging from 0.5 nM to 5 µM. Values shown are average of two to six determinations. toward these inhibitors. Several recent studies have used some of these compounds to "specifi cally" target a given lipase in human and also rodent cells, without realizing their effi cacy as well as specifi city. Results from such studies can lead to erroneous conclusions and need to be interpreted cautiously. Thus, it is important to ensure that the inhibitors developed against rodent enzymes also work against human enzymes, if the compounds are to be used with human cells and tissues. This becomes particularly important for drug discovery applications.

Advantages of the described simplifi ed assays for lipases
The available procedures for measuring the reaction products (e.g., glycerol or fatty acids) of the lipases using their natural authentic substrates (such as acyl glycerides) are not as sensitive as the currently described methods. For example, ABHD6 assay using its "authentic" natural substrate, 1-palmitoylglycerol, requires measuring the reaction products glycerol or palmitate. We show that measurement of glycerol released in this reaction is much less sensitive and reliable as compared with measuring the activity of this enzyme (same amount of enzyme protein) using 1-S -arachidonoylthioglycerol as the substrate ( Table 4 ). In fact, signal-to-noise ratio for the measurement of glycerol produced from 1-palmitoylglycerol was only about 1.2 to 1.9 in comparison with 7 to 23 for the fl uorogenic assay with 1-S -arachidonoylthioglycerol. Measurement of palmitate requires its extraction, derivatization, and processing for HPLC, which is very laborious and cumbersome with low sensitivity and yields. Neither of these approaches is amenable for high-throughput screening protocols. The same logic applies for ATGL, HSL, and DAGL, which use acylglycerides as the substrates and produce fatty acids and glycerol.
Overall, the results show that many of the inhibitors described earlier against rodent enzymes do not effectively inhibit the corresponding human enzymes and also that several of the compounds lack specifi city. The IC 50 values ( Table 2 ) of the tested inhibitors for some of these human lipases (ATGL, HSL, DAGL ␣ , MAGL, ABHD6, and CES1) relate to their inhibitory effects seen at 5 M concentration ( Table 1 ). Thus, results obtained using these compounds to specifi cally target a given lipase in a human cell/ tissue system must be interpreted cautiously, taking into account of the presence/absence of other enzymes, which may also be targeted. The specifi city of these inhibitors against murine enzymes remains to be assessed.
As we noticed that some of the previously described inhibitors against mouse lipases were ineffective against the corresponding human enzymes, we directly examined this by comparing the inhibitor effi cacy against recombinant human and mouse ATGL and HSL ( Table 3 ). While orlistat and Cay10499 were inhibitory against both human and mouse enzymes, hHSL was relatively less inhibited by either of these compounds. The inhibitors developed against mouse ATGL (WWL64 and ATGListatin) and mouse HSL (WWL11 and Compound 13f) were highly effective in inhibiting the corresponding mouse enzymes but have much less effect on the human enzymes ( Table 3 ). On the other hand, DAGL ␣ inhibitors (KT109, KT172, and RHC20867) and MAGL inhibitors (JZL184 and JZL195) were inhibitory to mouse ATGL and HSL to a variable extent, even though these compounds showed no signifi cant effect on hATGL and hHSL.
The different effects of these compounds on mouse and human enzymes may be because of subtle but signifi cant differences in their amino acid sequences and reactivity  NT, extract from not transfected 293 cells; ABHD6 transfected, extract from ABHD6-overexpressing 293 cells. N = 4 for 1-palmitoylglycerol assays, and N = 5 for 1-S -arachidonoylthioglycerol assays. Background corrected values are shown. Results are mean ± SEM.