Assay and Inhibition of Diacylglycerol Lipase Activity

Abstract

A series of N-formyl-α-amino acid esters of β-lactone derivatives structurally related to tetrahydrolipstatin (THL) and O-3841 were synthesized that inhibit human and murine diacylglycerol lipase (DAGL) activities. New ether lipid reporter compounds were developed for an in vitro assay to efficiently screen inhibitors of 1,2-diacyl-sn-glycerol hydrolysis and related lipase activities using fluorescence resonance energy transfer (FRET). A standardized thin layer chromatography (TLC) radioassay of diacylglycerol lipase activity utilizing the labeled endogenous substrate [1″-¹⁴C]1-stearoyl-2-arachidonoyl-sn-glycerol with phosphorimaging detection was used to quantify inhibition by following formation of the initial product [1″-¹⁴C]2-arachidonoylglycerol and further hydrolysis under the assay conditions to [1-¹⁴C]arachidonic acid.

The catalytic sites of diacylglycerol lipases (DAGLs) are unique among the lipases and hydrolases in effecting good selectivity for the hydrolysis of the 1-acyl group of 1,2-diacyl-sn-glycerol substrates.¹ The DAGLα (120 KDa) and DAGLβ (70 KDa) isoforms^1,2 are both present during brain development³ where their presynaptic axonal location suggests an important role in this period of neuritogenesis, rapid cell growth, and plasticity.^4,5 However, the ultimate postsynaptic dendritic location of the α-isoform in the adult brain is consistent with the subsequent role of DAGLα in endocannabinoid paracrine retrograde signaling with a lesser role for the DAGLβ isoform.^1,5–13 Thus, DAGLs have an important role in the endocannabinoid system as they are primarily responsible for releasing endocannabinoid 2-arachidonoylglycerol (2-AG) from diacylglycerols, including 1-stearoyl-2-arachidonoyl-sn-glycerol that is the principal 1,2-diacyl-sn-glycerol component of brain and nerves,^14–16 for signaling at cannabinoid receptors.^17–24 Small molecules that inhibit DAGL would have major effects on lipid metabolism. For example, less DAGL-catalyzed biosynthesis of endocannabinoid 2-arachidonoylglycerol (2-AG) may reduce activation of cannabinoid receptors. The resulting reduction of receptor signaling would be distinct from the actions of cannabinoid receptor inverse agonists and may be of pharmacological utility.

The three lead inhibitors of DAGL activity include bis-oximinocarbamate RHC80267 1,^1,25 fluorophosphonate O-3841 2,²⁶ and tetrahydrolipstatin (3, THL)^1,27 (Figure 1) that inhibit hDAGLα with apparent IC₅₀ values of 65000 nM, 160 nM, and 60 nM, respectively. The transition-state-mimicking fluorophosphonate group (RP=OOEtF) of FP-fluorescein (see Supplementary data for structure) utilized for the identification of serine hydrolases does not react appreciably with DAGL,²⁷ unlike fluorophosphonate O-3841 2 (ROP=OMeF)²⁶ and its t-butyl analog O-5596.²⁸ Also, DAGL activity is not affected by phenylmethylsulfonylfluoride (PhCH₂SO₂F).¹ There have been several previously reported analytical methodologies for measuring DAGL activities including radio-TLC,^1,26,29,30 LC-MS,^2,27 and the use of general esterase reporter molecules.²

Figure 1.

Diacylglycerol lipase inhibitors.

We now report our studies of new inhibitors of DAGL activity that generally resemble diglycerides. We will also discuss the DAGL proteins utilized in these studies. Our work on assay conditions using a radiolabeled endogenous diglyceride substrate will be detailed. Finally, the development of novel fluorescent resonance energy transfer (FRET) reporter substrates for the in vitro assay of DAGL activity will be discussed including their utilizations for the assays of related lipases.

We first prepared new analogs of THL 3 and OMDM-188 4.³¹ THL 3 has an N-formyl-L-leucyl ester and OMDM-188 4 is the corresponding N-formyl-L-isoleucyl ester. The three other novel isoleucine diastereomers (5, 6, 7) as well as the (S)-α-aminobutyryl ester 8 were prepared via the reported method³¹ from the corresponding benzyloxycarbonyl protected α-amino acids 13 (see Scheme 1). The β-lactone analog 9 completely lacking both the N-formyl-α-aminoacyloxy and the 2-hexyl groups was prepared by treating racemic 3-hydroxypalmitic acid with N-phenyl-bis(trifluoromethanesulfonimide). The shorter chain separable trans-10 and cis-11 β-lactones were prepared according to the reported method.³² Also, a new series of racemic ether lipids derived from lead compounds RHC80267 1 and O-3841 2 was prepared (see Supplementary data).

Scheme 1.

Synthesis of analogs of THL 3 and OMDM-188 4.

We used diacylglycerol lipase (DAGL) protein from either cell lysates or membrane fractions that were prepared according to the previously reported method.^1,33 The Cravatt group at Scripps provided hDAGLα, mDAGLα, and mDAGLβ from overexpression by transient infection of HEK293T cultures in addition to cell lysate of the empty vector HEK293T control. Some experiments used a second commercially prepared plasmid to provide additional human α-isoform overexpressed in HEK293T using the same methodology. All attempts to overexpress human DAGLα were disappointing. The lipase activity of hDAGLα expressed in the human cell line was still sufficient for assay of newly synthesized inhibitors. At least 100 μg of total protein from crude cell lysates or at least 10 μg of total protein from membrane preparations was required per well. The proteins were utilized such that the substrate hydrolysis would proceed to the extent of about 5% in 20 min. The more readily expressed mDAGLβ isoform (that has a 79% homology¹ with the human isoform) or mDAGLα isoform (that has a 97% homology¹ with the human isoform) were also used to confirm inhibition of DAGL activities.

It was of particular importance for biological relevance and inhibitor evaluation that assays of DAGL activities utilize pure endogenous substrate such as radiolabeled [1″-¹⁴C]1-stearoyl-2-arachidonoyl-sn-glycerol ([¹⁴C]SAG).^29,34 Any radiolabeled 1-stearoyl-3-arachidonoyl-sn-glycerol present in the substrate, due to rearrangement of [¹⁴C]SAG to labeled 1,3-diglyceride, would be readily hydrolyzed by other enzymes in the relatively crude enzyme preparations and the result misinterpreted to be due to DAGL activity. The 1,2-diglyceride substrate [¹⁴C]SAG that we used contained less than 0.5% of the 1,3-diglyceride isomer (see Supplementary data).

Although the [¹⁴C]SAG substrate has previously been used for TLC-based assays of DAGL activity,^1,26,30 details of these TLC assays were not fully described. Also, the apparent IC₅₀ of THL 3 for hDAGLα has been reported by the Di Marzo group to range from 60 nM^1,25,26 to 1000 nM,³¹ postulated to be due to the effect of DAGL protein concentration. Therefore, we developed a standardized assay that contained 10% DMSO.^35,37 Higher DMSO concentrations did not increase rates of substrate conversion to fluorescent products in any of the in vitro assays. Non-denaturing detergents n-heptyl-βD-thioglucopyranoside and Triton X-100 at concentrations up to 0.2% were also used in some experiments (see Tables 1 and 3). Detergent use dramatically increases apparent DAGL specific activities and also decreases the apparent IC₅₀ for DAGL inhibitory compounds. The DAGL enzyme suspensions had a 15 minute preincubation period of covalent quasi-irreversible inactivation by THL 3 or other potential inhibitors. The substrate [¹⁴C]SAG was then added (20 μM final concentration) and residual DAGL activity was quantified after 20 min by quenching the reaction with 2:1 chloroform/methanol and vortexing to denature the protein and move all lipids out of the aqueous phase. Very little rearrangement of [¹⁴C]SAG (1,2-diglyceride) to 1,3-diglyceride occurred under the reaction and workup conditions as assessed by TLC using boric acid treated silica gel plates.

Table 1.

Assays of the inhibition of hDAGLα (radio-TLC assay with [¹⁴C]SAG substrate),³⁷ rFAAH,⁵⁷ and hMAGL⁶³ enzyme activities

Compound Number	Detail of N-formyl- α-amino ester	hDAGLα inhibition at 10 μM detergent free or (with Triton X-100) (%)	rFAAH inhibition at 10 μM (%)	hMAGL inhibition at 10 μM (%)
3 (THL)	L-leucyl	100, (100A)	6	47
4 (OMDM-188)	L-isoleucyl	(98A)	7	16
5	L-allo-isoleucyl	(95A)	3	1
6	D-isoleucylC	(78A)	28	39
7	D-allo-isoleucyl	(86A)	0	42
8	(S)-α-aminobutyryl	(99A)	12	45
9	none	(25A, 45B)	10	4
trans-10	none	(37B)	79	55
cis-11	none	(25B)	94D	100D
JZL184	NA	(37B)	97E	100F
URB597	NA	ND	100E	18
n-C16H33SO2F	NA	ND	100E	82
PMSF	NA	19	100E	5
RHC80267 (1)	NA	30	95E	22
SD41	NA	<0*	10	13

NA, not applicable; ND, not done

^A0.05% Triton X-100 present

^B0.015% Triton X-100 present

^C5% impurity in D-isoleucyl analog due to D-allo-isoleucyl analog

^Dcis-11 inhibits rFAAH only 13% at 1 μM, but inhibits hMAGL 66% at 100 nM

^E

8 pt rFAAH IC₅₀ (95% confidence)

JZL184	974 nM (784–1210)
URB597	4.9 nM (4.1–6.0)
n-C16H33SO2F	6.3 nM (4.5–8.7)
PMSF	833 nM (746–931)
RHC80267 (1)	2,240 nM (2010–2500)

^F8 pt hMAGL IC₅₀ (95% confidence)

JZL184

57 nM (53–62)

^*protein was pretreated with JZL184 to completely inhibit MAGL activity

Table 3.

Results of in vitro FRET-based screening using reporter compound 17 of the inhibition of lipase activities

Compound Number	Detail of N-formyl- α-amino ester	hDAGLα inhibition at 10 μM detergent free or (0.05% Triton X-100) (%)			Lipoprotein Lipase (Bacterial) inhibition at 10 μM (%)	Pancreatic Lipase (Porcine) inhibition at 10 μM (%)
3 (THL)	L-leucyl	92,	86*	(99)	92	98
4 (OMDM-188)	L-isoleucyl	92		(99)	84	99
5	L-allo-isoleucyl	92		(99)	61	98
6	D-isoleucylA	90		(98)	86	99
7	D-allo-isoleucyl	90		(98)	68	95
8	(S)-α-aminobutyryl	93		(100)	80	97
9	none	14		(72)	<0	38
trans-10	none		69*		67	96
cis-11	none		65*		33	94
JZL184	NA	59,	44*		36	36
URB597	NA	22,	7*		21	68
n-C16H33SO2F	NA	40,	94*		8	29
PMSF	NA		8*		<0	17
RHC80267 (1)	NA	70,	40*		<0	78
SD41	NA	17,	7*		<0	<0
JZL195	NA	84,	55*		<0	43

NA, not applicable

^A5% impurity in D-isoleucyl analog due to D-allo-isoleucyl analog

^*protein was pretreated with JZL184 to completely inhibit MAGL activity

TLC DAGL assays with [¹⁴C]SAG substrate (or LC-MS assays with pure unlabeled SAG substrate) could result in significant errors if subsequent hydrolysis of 2-AG was not considered. It was critical that evaluations of enzymatic hydrolysis of [¹⁴C]SAG substrate include the sums of radiolabeled 2-AG and free radiolabeled arachidonic acid released. The crude cell preparations that were used had considerable monoacylglycerol lipase (MAGL), fatty acid amide hydrolase (FAAH), and other lipase activities which further degraded the radiolabeled 2-AG as it was formed under the assay conditions. It was very clear from the controls and from experiments with poor inhibitors that the release of labeled 2-AG was followed by further hydrolysis to labeled arachidonic acid. Thus DAGL activity was calculated via the sum of [¹⁴C]2-AG plus [¹⁴C]AA released divided by the sum of [¹⁴C]2-AG, [¹⁴C]AA, and final [¹⁴C]SAG concentrations for each lane. Also, DAGL activity in this human cell line (HEK293T) was not adjusted for hDAGLα and hDAGLβ activities demonstrated to be present in cell lysates following the mock infections.

We observed that the conversion of [¹⁴C]2-AG to [¹⁴C ]AA was reduced in modified radio-TLC assays by pre-treatment of cell lysate with the highly selective MAGL inhibitor JZL184 at 10 μM for 15 min. Then, tenfold dilution to 1 μM for screening assay use in some experiments as noted in the data tables gave little JZL184 interference with hDAGLα activities in radioassays and fluorescent assays.

The use of 10 μM THL 3 resulted in complete inhibition of (human) hDAGLα activity for all protein preparations (Table 1). Using TLC under basic conditions^1,26 and phosphorimaging analysis,^34,41 the radioassays consistently showed the apparent IC₅₀ of THL 3 to be in the range of 10 to 100 nM. Analogs 4 – 8 were also all extremely potent inhibitors of hDAGLα in radio-TLC assays. The β-lactones 9 – 11 and other compounds including JZL184, PMSF, and RHC80267 1 were poor inhibitors of hDAGLα. The β-lactone SD41 and ether lipid analogs of O-3841 we synthesized (see Supplementary data) did not inhibit hDAGLα or mDAGLβ at 10 μM screening concentrations.

The β-lactones THL 3, OMDM-188 4, and new analogs 5 – 9, were assayed for the inhibition of (murine) mDAGLα at 10 nM to 10 μM concentrations (Table 2). The D-isoleucyl- and D-allo-isoleucyl-analogs (6 and 7) were clearly less potent than analogs of (S)-α-amino acids. Several analogs (THL 3, OMDM-188 4, and the new L-allo analog 5) showed good selectivity for diacylglycerol lipase over monoacylglycerol lipase and fatty acid amide hydrolase enzymes. Future studies can include shorter alkyl chain analogs similar to the fatty acid synthase inhibitors from the Romo group⁴² that might have better solubility and membrane penetration properties.

Table 2.

Radio-TLC assay with [¹⁴C]SAG substrate of the inhibition of mDAGLα activity

Compound Number	Detail of N-formyl-α-amino ester	mDAGLα inhibition at 10 nM (%)	mDAGLagr; inhibition at 100 nM (%)	mDAGLagr; inhibition at 1000 nM (%)	mDAGLagr; inhibition at 10000 nM (%)
3 (THL)	L-leucyl	55	72	92	98
4 (OMDM-188)	L-isoleucyl	69	80	96	96
5	L-allo-isoleucyl	60	61	88	100
6	D-isoleucylA	33	29	69	85
7	D-allo-isoleucyl	19	51	75	90
8	(S)-α-aminobutyryl	50	68	96	100
9	NA	ND	ND	ND	<0
JZL184	NA	ND	ND	ND	3
URB597	NA	ND	ND	ND	<0
n-C16H33SO2F	NA	ND	ND	18	64
JZL195	NA	ND	ND	ND	10

NA, not applicable; ND, not done

^A5% impurity in D-isoleucyl analog due to D-allo-isoleucyl analog

Ether lipid substrates 17 – 22 (Figure 2) for in vitro fluorescence resonance energy transfer (FRET) assay of DAGL and related lipase activities were developed from related lipase activity reporter molecules.^43–47 This series of novel ether lipid molecules was synthesized (Scheme 2) with the biomimetic stereochemistry at sn-2 and incorporated terminally functionalized sn-1 and sn-2 fatty acyl groups. The epoxide of (R)-(-)-glycidyl methyl ether (23) was opened with benzyl alcohol and sodium hydride followed by silyl protection of the secondary alcohol 24, and hydrogenolysis of the benzyl ether 25 in ethanol/ethyl acetate/acetic acid (1:1:0.2). Subsequent acylation of the primary hydroxyl group of 26, deprotection of the secondary alcohol t-butyldimethylsilyl group, followed immediately by acylation of the secondary hydroxyl of 28 gave ether lipid products 17 – 22 with only 5% impurity from acyl rearrangement.

Figure 2.

Newly synthesized reporter substrates for in vitro FRET-based DAGL and related lipase assays.

Scheme 2.

Synthesis of new ether lipid reporter compounds 17 – 22 for in vitro assays of DAGL and related lipase activities.

The FRET pairs initially studied were the pyrene and nitrobenzoxadiazole (NBD) fluorophors with either the dinitrophenyl or nitroxyl group quenchers. Excitation of the pyrene or NBD results in radiationless energy transfer to the quenchers when close enough and sufficiently well oriented. The fully extended distances between the fluorophors and quenchers were estimated (Schrödinger Suite 2010, in an aqueous environment with a dielectric constant of 80) to be 18 Å for 17, 18, and the NBD analog 20; and, to be 24 Å for the pyrenedecanoyl analog 19. The assays with NBD analog 20 had too much baseline instability as this fluorescent group is quite sensitive to the polarity of its environment. The pyrene was estimated to be 15 Å and 24 Å from the nitroxyl stable free radical quenching groups⁴⁸ of the 5-doxylstearoyl analog 21 and the 16-doxylstearoyl analog 22, respectively, and both compounds were poor substrates for enzymatic hydrolysis. The pyrene-dinitrophenyl FRET pairings 17 and 18 were stable to uncatalyzed hydrolysis at neutral pH, were the best enzyme substrates, and were readily utilized in a 96-well format. A convenient in vitro fluorometric esterase assay utilizing the BioTek Instruments Synergy HT Multi-Mode Microplate 96-well reader was developed that measured nanomolar concentrations of fluorescent reaction product.⁴⁹ Cell lysate or membrane preparations containing overexpressed DAGL catalyzed the hydrolysis of the reporter substrate, and a fluorescence response increased at a nearly linear rate for over two hours. Wells that had a 15 minute pre-incubation with DAGL inhibitors and that showed a concentration dependent attenuation of fluorescence response were identified as “hits.” At a screening concentration of 10 μM, compounds 3–8 were identified as potent inhibitors of hDAGLα (Table 3). However, any potential DAGL inhibitor identified from the in vitro FRET assay should then be submitted to the TLC assay with the radiolabeled endogenous diglyceride substrate to identify any false positives for DAGL inhibition. Fluorescence results could reflect inhibition of other hydrolytic enzymes in the crude cell preparations that hydrolyze the ether lipid substrates 17 – 22 to a much greater extent than the endogenous 1,2-diacyl-sn-glycerol substrate. Using highly selective enzyme inhibitors JZL184, URB597, and others, the enzymes responsible for false positives include monoacylglycerol lipase (MGL) and fatty acid amide hydrolase (FAAH). The ether lipid FRET-substrate 17 will be most useful for assays with DAGLs purified to homogeneity.

Additional assays were performed to establish the selectivity of DAGL inhibitors. Compounds 3 – 11 were assayed for binding to the CB1 (rat brain preparation) and CB2 (mouse or human receptor expressed in HEK293),⁵⁰ and none had a Ki below 1 μM in these competition binding experiments. Compounds 3 – 11 were also assayed for inhibition of endocannabinoid hydrolytic enzymes in fluorescence-based assays (Table 1). Assays of the inhibition of fatty acid amide hydrolase (rFAAH) used the reported coumarin amide reporter compound,^56,57 and assays of the inhibition of monoacylglycerol lipase (hMGL) used the 7-hydroxy-6-methoxy analog^62,63 of the reported coumarin ester.⁶⁶ The assays were validated with standard compounds including the selective FAAH inhibitor URB597⁵⁹ and the selective MAGL inhibitor JZL184.³⁶ Preliminary studies with [1″-¹⁴C] SAG using the purified rFAAH and hMAGL endocannabinoid hydrolytic enzymes in these fluorescent assays showed low activities of these enzymes for the 1,2-diglyceride substrate. Also, preliminary studies with ether lipid substrate 17 in the in vitro FRET-based assay confirmed the potent inhibition by THL 3^67,68 of commercially available homogeneous bacterial lipoprotein lipase and porcine triacylglycerol lipase enzymes (Table 3).⁶⁹ This new FRET-based methodology should be suitable for the assay of new inhibitors of human recombinant proteins including lipoprotein lipase, triacylglycerol lipase, and other related hydrolases to determine DAGL selectivity in vitro.

In summary, our structure-activity relationship (SAR) studies have demonstrated molecular features of inhibitors that result in inactivation of human and murine DAGLs at nanomolar inhibitor concentrations. The importance of a small (S)-N-formyl-α-amino group as a structural feature in targeting DAGLs was clearly demonstrated. The β-lactone was the most active of the covalently reactive quasi-irreversible inactivating functional groups that we tested for DAGL inhibition. In pilot experiments, we have looked at many factors that may affect assay results including protein concentration, substrate structure and concentration, length of the incubation period for enzyme inactivation, and the presence of co-solvent and detergent. An in vitro FRET-based screen was established for rapidly identifying inhibitors of DAGL activity. Although false positives can occur due to inhibition of other hydrolytic enzymes present in the cell lysate or membrane preparations used, the assay will be suitable for the preliminary screening of compound libraries. An improved and detailed radio-TLC assay of DAGL activity with the labeled endogenous substrate [1″-¹⁴C]1-stearoyl-2-arachidonoyl-sn-glycerol was utilized to unambiguously distinguish DAGL activity from the activities of MAGL and FAAH. Thus, methodologies were established to determine α/β-subtype selectivity as well as selective inhibition of DAGL over esterases, amidases, and other lipases.

Abbreviations

AA

arachidonic acid

DAGL

diacylglycerol lipase

FAAH

fatty acid amide hydrolase

FRET

fluorescence resonance energy transfer

MAGL

monoacylglycerol lipase

NBD

nitrobenzoxadiazole

[¹⁴C]SAG

[1″-¹⁴C]1-stearoyl-2-arachidonoyl-sn-glycerol

THL

tetrahydrolipstatin