PK/PD Disconnect Observed with a Reversible Endothelial Lipase Inhibitor

Abstract

Screening of a small set of nonselective lipase inhibitors against endothelial lipase (EL) identified a potent and reversible inhibitor, N-(3-(3,4-dichlorophenyl)propyl)-3-hydroxy-1-methyl-2-oxo-1,2-dihydropyridine-4-carboxamide (5; EL IC₅₀ = 61 nM, EL_HDL IC₅₀ = 454 nM). Deck mining identified a related hit, N-(3-(3,4-dichlorophenyl)propyl)-4-hydroxy-1-methyl-5-oxo-2,5-dihydro-1H-pyrrole-3-carboxamide (6a; EL IC₅₀ = 41 nM, EL_HDL IC₅₀ = 1760 nM). Both compounds were selective against lipoprotein lipase (LPL) but nonselective versus hepatic lipase (HL). Optimization of compound 6a for EL inhibition using HDL as substrate led to N-(4-(3,4-dichlorophenyl)butan-2-yl)-1-ethyl-4-hydroxy-5-oxo-2,5-dihydro-1H-pyrrole-3-carboxamide (7c; EL IC₅₀ = 148 nM, EL_HDL IC₅₀ = 218 nM) having improved PK over compound 6a, providing a tool molecule to test for the ability to increase HDL-cholesterol (HDL-C) levels in vivo using a reversible EL inhibitor. Compound 7c did not increase HDL-C in vivo despite achieving plasma exposures targeted on the basis of enzyme activity and protein binding demonstrating the need to develop more physiologically relevant in vitro assays to guide compound progression for in vivo evaluation.

Endothelial lipase (EL; gene nomenclature LIPG)^1,2 exerts pleiotropic effects on cardiovascular biology through its role in high density lipoprotein (HDL) catabolism,³⁻⁶ vessel wall inflammation,⁷⁻¹⁵ and subsequent effects on reverse cholesterol transport (RCT).^4,16−19 Inhibition of EL using neutralizing polyclonal antibodies in mice²⁰ or pharmacologically using irreversible small molecule inhibitors XEN445 (1)²¹ and compound 2(22) (Figure 1) has been reported to raise HDL-C levels in mice, whereas loss of function EL variants in humans (e.g., Asn396Ser) have been associated with increased HDL-C levels.²³ A recently published Mendelian randomization study showed no correlation between increased HDL-C levels resulting from a loss of function EL variant and the risk of myocardial infarction despite an expected 13% reduction of risk estimated from the amount of HDL-C increase associated with the loss of function allele.²⁴ This study throws doubt into the hypothesis that raising HDL-C levels by inhibiting EL enzymatic activity will decrease coronary artery disease (CAD). To fully investigate the effect of raising HDL-C levels on CAD through pharmacological inhibition of EL will require high quality drug molecules with potent in vivo efficacy.

Figure 1.

Examples of literature EL inhibitors.

EL is a member of the family of enzymes that includes lipoprotein lipase (LPL), hepatic lipase (HL), and pancreatic lipase (PL). In contrast to LPL and PL, which selectively hydrolyze triglycerides (TGs) and HL that hydrolyzes both TGs and phospholipids, EL shows a preference for the hydrolysis of the sn1 ester of phosphatidylcholines found in HDL producing lyso-phosphatidylcholines (LPCs) and free fatty acids (FFAs), resulting in lipid-depleted HDL particles that are cleared more rapidly from the circulation.²⁵

Several small molecule inhibitors of EL have been reported in the literature, e.g., anthranilic acids (XEN445, 1),²¹ thiocarbamates (2),²² sulfanylfuran ureas (3),^26,27 and phenyl boronic acids (4)²⁸ (Figure 1). We report herein our initial efforts to identify potent, reversible inhibitors of EL for the purpose of elevating HDL-C blood levels in vivo.

In conjunction with virtual and high throughput screens, nonselective in-house lipase inhibitors were screened for EL inhibition using a mixed vesicle substrate (EL assay; see Supporting Information). The screen led to the identification of N-(3-(3,4-dichlorophenyl)propyl)-3-hydroxy-1-methyl-2-oxo-1,2-dihydropyridine-4-carboxamide (compound 5, Figure 2) as a potent inhibitor of EL activity (EL IC₅₀ = 61 nM) containing a weakly acidic to neutral heterocycle (measured pK_a = 6.5). Compound 5 was also a potent inhibitor of HL (HL IC₅₀ = 16 nM) but was found to be highly selective against LPL (LPL IC₅₀ > 90,000 nM). When EL was inhibited with a high concentration of compound 5 (2 μM), enzymatic function was restored by dialyzing away unbound inhibitor from the assay buffer (data not shown), demonstrating the reversibility of inhibition. Further deck mining around compound 5 identified a structurally related EL inhibitor, N-(3-(3,4-dichlorophenyl)-propyl)-4-hydroxy-1-methyl-5-oxo-2,5-dihydro-1H-pyrole-3-cabox-amide (compound 6a, Figure 2), with EL IC₅₀ = 41 nM that contained a slightly more acidic heterocycle (measured pK_a = 4.8). Compound 6a was also not selective against HL (HL IC₅₀ = 76 nM) but had excellent selectivity versus LPL (LPL IC₅₀ > 37,000 nM). Pharmacokinetic (PK) profiling of compounds 5 and 6a in CD1 mice (Table 1) showed both compounds to have excellent oral bioavailability (82% and 76%, respectively). Owing to slightly better selectivity against HL and facile synthetic methods available for compound 6a, we focused our initial SAR efforts on this chemotype, seeking to improve potency, to maintain the favorable PK properties, and to establish a pharmacodynamic (PD) profile and PK/PD relationship so the EL mechanism of action could be evaluated pharmacologically in vivo.

Figure 2.

Initial EL hits from focused deck screening.

Table 1. Pharmacokinetic Parameters for Compounds 5 and 6a in CD1 Mice.

	compound 5		compound 6a
parameter	i.v.	p.o	i.v.	p.o.
dose (mpk)a	1.0	1.0	1.0	1.0
Cmax (μM)	25	2.0	32	8.2
AUCtotal (μM·h)	19	16	24	18
CL (mL/min/kg)	2.5		2.1
t1/2  (h)	4.9	2.9	3.4	3.2
F (%)		82		76

^aBolus administration i.v. and p.o. using 60% PEG400, 30% water, and 10% EtOH as vehicle. mpk = milligram per kilogram.

The synthesis of compounds 6a–g, 8–12, and 13a–d is reported in the Supporting Information. Synthesis of the in vivo candidate 7c and related compounds 7a, 7b, and 7d is shown in Scheme 1. Ethyl 1-ethyl-4-hydroxy-5-oxo-2,5-dihydro-1H-pyrrole-3- carboxylate (16)^29,30 was first O-methylated with trimethylsilyldiazomethane in the presence of DIPEA,³¹ followed by saponification of the ester. The acid (17) was coupled with amines (R³-NH₂) using the corresponding acid chloride generated with oxalyl chloride. The resulting amides were demethylated with boron trichloride to give the final products 7a and 7b. Compounds 7c and 7d were obtained by chiral HPLC separation of 7b. Compound 7c was also obtained from the homochiral amine synthesized using the method of Ellman.³² Excellent stereochemical induction was observed (dr = 99:1) when methylmagnesium bromide was combined with the (R)-N-tert-butanesulfinyl imine of 3-(3,4-dichlorophenyl)propanal at 60 °C leading to a high enantiomeric excess of the final product (see Supporting Information).

Scheme 1. Synthesis of Compounds 7a–d.

Reagents and conditions: TMSCH₂N₂, DIPEA, Et₂O, rt, 48 h;

NaOH, MeOH/H₂O (1:1), 70 °C, 1 h;

(i) oxalyl chloride, CH₂Cl₂, DMF_(cat), (ii) R³-NH₂, DIPEA, CH₂Cl₂, rt, 3 h;

BCl₃, CH₂Cl₂, rt, 5 h. See Supporting Information for individual compound yield.

The compounds were evaluated using a PED-A1/DMPG mixed vesicle assay (EL assay) and an HDL assay using purified HDL particles as enzyme substrate with detection of the product, linoleoyl-lyso-phosphatidylcholine, by LCMS (EL_HDL assay; see Supporting Information) in an effort to provide a bridging assay between in vitro and in vivo activity.

We first explored the SAR of the R¹ position of the 1-methyl-5-oxo-2,5-dihydro-1H-pyrrole core found on compound 6a. It was observed that extending the R¹ side chain with lipophilic or polar groups maintained or improved EL potency (see compounds 7a and 8–10, Table 2), ranging between 3.6 nM (compound 9, R¹ = benzyl) and 30 nM (compound 7a, R¹ = ethyl). Addition of a basic side chain (R¹ = N-ethylmorpholine) reduced EL potency by ca. 5-fold, whereas the acidic acyl sulfonamide 12 had similar potency to compound 6a. When these compounds were tested using the EL_HDL assay, a significant right-shift to reduced potency was observed giving EL_HDL IC₅₀/EL IC₅₀ ratios between 11- and 150-fold.

Table 2. R¹ Group SAR of 1-Methyl-5-oxo-2,5-dihydro-1H-pyrrole core (compounds 7a and 8–12).

^aIC₅₀ values are an average of at least two independent determinations.

^bSee Supporting Information for standard deviations.

^cND = not determined.

The SAR for substitution at the R² position was examined within the context of R¹ = CH₃ (Table 3). A limited set of analogs was examined (compounds 6b–6d). All were of similar potency to the original hit. These analogs did not improve EL_HDL potency and the ratio of EL_HDL to EL potencies ranged from 60- to 100-fold. We concluded that further exploration of this position was not warranted.

Table 3. R² Group SAR of 1-Methyl-5-oxo-2,5-dihydro-1H-pyrrole Core (Compounds 6b–d).

^aIC₅₀ values are an average of at least two independent determinations.

Replacing the C-4 hydroxyl substituent with amines (R⁴-NH₂) in the context of R¹ = ethyl and R³ = N-(3-(3,4-dichlorophenyl)prop-1-yl provided compounds 13a–d (Table 4). Modification at this position provided analogs that were slightly more potent or equipotent to 6a in the EL assay (IC₅₀ range from <10 to 59 nM); however, a significant right-shift in potency in the EL_HDL assay was observed, with EL_HDL/EL ratios ranging from 50- to >300-fold.

Table 4. R⁴ Group SAR of 1-Methyl-5-oxo-2,5-dihydro-1H-pyrrole Core (Compounds 13a–d).

^aIC₅₀ values are the average of at least two independent determinations.

Exploration of the SAR of the amide side chain (R³) was accomplished in two series where R¹ = CH₃ or ethyl (Table 5). It was generally observed that the most potent compounds in the EL assay contained extended lipophilic R³ groups and two chlorine atoms located at C-3 and C-4 on the terminal phenyl. Thus, compound 6e (R³ = phenyethyl) was 10-fold less potent than the bis-chloro compound 6f, and compound 6a was ca. 140-fold more potent than the des-chloro analog compound 6g. Further SAR development of the R³ side chain identified a key modification that provided an analog with more balanced EL and EL_HDL potency, compound 7b (EL IC₅₀ = 44 nM, EL_HDL IC₅₀ = 264 nM). When the two enantiomers were obtained in homochiral form, one enantiomer (compound 7c; EL IC₅₀ = 148 nM, EL_HDL IC₅₀ = 218 nM) was significantly more potent than the other (compound 7d) and was equally potent for mouse EL (mouse EL_HDL IC₅₀ = 100 nM). Importantly, the potency of compound 7c using human or mouse EL was relatively insensitive to increasing concentration of HDL (tested at 40, 100, 300, and 1000 μg/mL HDL; see Supporting Information) suggesting compound 7c is not competitive with HDL.

Table 5. R³ Group SAR of 1-Methyl-5-oxo-2,5-dihydro-1H-pyrrole Core (Compounds 6a, 6e–g, and 7b–d).

^aIC₅₀ values are an average of at least two independent determinations.

^bSee Supporting Information for standard deviations.

^cND = not determined.

When tested against a panel of related enzymes, compound 7c was found to be >250-fold selective for EL versus LDL, monoacylglycerol lipase (MAGL), and pancreatic lipase (PL). Using the HDL-based assay, compound 7c was 12-fold selective for EL vs HL (see Supporting Information).

Pharmacokinetic evaluation of compound 7c using male CD1 mice at a dose of 1.0 mpk i.v. or p.o. (Table 6 and Chart 1) showed it to have excellent oral exposure (oral AUC_total = 292 μM; 96% bioavailability) and C_max (15 μM) with a longer half-life (t_1/2 = 13 h) than compound 6a (t_1/2 = 3.2 h).

Table 6. Pharmacokinetic Data for Compound 7c in CD1 Mice.

parameter	i.v.	p.o.
dose (mpk)a	1.0	1.0
Cmax (μM)	39	15
Tmax (h)	0.05	0.5
AUCtotal (μM·h)	304	292
t1/2 (h)	13	13
F (%)		96
C 24 h (nM)		4310
Cfree 24 h (nM)b		99

^aDosing vehicle for i.v. and p.o. administration was 60% PEG400, 30% water, 10% ethanol. mpk = milligram per kilogram

^bAdjusted for mouse protein binding using 2.3% free fraction.

Chart 1. Mean Plasma Concentrations of Compound 7a in CD1 Micea.

^a Compound 7c dosed p.o. at 1 mpk in CD1 mice using 60% PEG400/30% water/10% ethanol as dosing solution.

When compound 7c was dosed at 10 and 50 mpk, systemic exposure was dose related but less than dose proportional for both (7.5-fold and 18-fold increase in AUC_total, respectively, vs 1.0 mpk dose; see Supporting Information). Protein binding in mice was 97.7%.

When compound 7c was administered to C57BL/6J mice once a day p.o. for 5 days at 10 and 50 mpk, high plasma exposures were observed in the PK arm of the experiment (see Supporting Information). Thus, on day 5 at 50 mpk, C_max = 238 μM (6 h postdose) and C_trough = 57 μM (24 h postdose). After adjusting for protein binding, C_max free (5.5 μM) and C_trough free (1.3 μM) were 55- and 13-fold, respectively, above the mouse EL_HDL IC₅₀. Despite the high exposures achieved, there was no effect on plasma HDL-C levels at either dose (Chart 2). A follow-up study in normal hamsters dosed once a day for 7 days at 50 mpk gave the same outcome. In this study, drug exposures were comparable to mouse. Accounting for protein binding in hamster (98.4%), C_max free = 5.1 μM (day 5, 6 h postdose), ca. 51-fold above the IC₅₀ for mouse EL (see Supporting Information).

Chart 2. . Plasma HDL Levels in C57BL/6J Mice Dosed with Compound 7ca.

^a Compound 7c was dosed at 10 and 50 mpk once daily p.o. using 60% PEG400/30% water/10% ethanol as dosing solution.

In summary, optimizing potency of screening hit 6a using the EL_HDL assay resulted in the identification of compound 7c (EL_HDL IC₅₀ = 218 nM; mouse EL_HDL IC₅₀ = 100 nM). The lack of a pharmacodynamic effect in two animal models of increasing plasma HDL-C levels, despite achieving levels of compound exposure that would predict good inhibition of the target enzyme, suggests factors other than protein binding may be involved in the inability of compound 7c to inhibit plasma EL. In addition, these results point to the need to develop other, more physiologically relevant in vitro assays to guide compound progression for in vivo evaluation. The results of these investigations will be reported separately.

Glossary

Abbreviations

DIPEA

diisopropylethylamine

DMPG

1,2-dimyristoyl-sn-glycero-3-[phosphor-rac-(1-glycerol)] sodium salt;

dr

diastereomeric ratio

PED-A1

N-((6-(2,4-DNP)amino)-hexanoyl)-1-(BODIPY FL C5)-2-hexyl-sn-Gly-cero-3-phosphoethanolamine.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.8b00138.

• Reaction scheme, experimental procedures and characterization data for compounds 6a–g, 7a–d, 8–12, and 13a–d, standard deviations of EL and EL_HDL IC₅₀ data for compounds 6a and 7c, dependence of EL_HDL potency on HDL concentration, mouse and hamster PK data, hamster PD data, biochemical assay protocols, and mouse in vivo protocol (PDF)

Author Contributions

The manuscript was written by J.J.H. and through contributions by all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.