Macrophage foam cells store excess cholesterol as cholesteryl esters, which need to be hydrolyzed for cholesterol efflux. We recently reported that silencing expression of carboxylesterase 1 (CES1) in human THP-1 macrophages [CES1KD (THP-1 cells with CES1 expression knocked down) macrophages] reduced cholesterol uptake and decreased expression of CD36 and scavenger receptor-A in cells loaded with acetylated low-density lipoprotein (acLDL). Here, we report that CES1KD macrophages exhibit reduced transcription of cytochrome P45027A1 (CYP27A1) in nonloaded and acLDL-loaded cells. Moreover, levels of CYP27A1 protein and its enzymatic product, 27-hydroxycholesterol, were markedly reduced in CES1KD macrophages. Transcription of LXRα (liver X receptor α) and ABCA1 (ATP-binding cassette transporter A1) was also decreased in acLDL-loaded CES1KD macrophages, suggesting reduced signaling through PPARγ–CYP27A1–LXRα. Consistent with this, treatment of CES1KD macrophages with agonists for PPARγ, RAR, and/or RAR/RXR partially restored transcription of CYP27A1 and LXRα, and repaired cholesterol influx. Conversely, treatment of control macrophages with antagonists for PPARγ and/or RXR decreased transcription of CYP27A1 and LXRα. Pharmacologic inhibition of CES1 in both wild-type THP-1 cells and primary human macrophages also decreased CYP27A1 transcription. CES1 silencing did not affect transcript levels of PPARγ and RXR in acLDL-loaded macrophages, whereas it did reduce the catabolism of the endocannabinoid 2-arachidonoylglycerol. Finally, the gene expression profile of CES1KD macrophages was similar to that of PPARγ knockdown cells following acLDL exposures, further suggesting a mechanistic link between CES1 and PPARγ. These results are consistent with a model in which abrogation of CES1 function attenuates the CYP27A1–LXRα–ABCA1 signaling axis by depleting endogenous ligands for the nuclear receptors PPARγ, RAR, and/or RXR that regulate cholesterol homeostasis.
Atherosclerosis is the excess deposition of lipid in the intima of arteries and the underlying pathological process responsible for most cardiovascular disease . Macrophages play a central role in atherosclerosis by engulfing oxidized low-density lipoprotein (oxLDL) particles via scavenger receptors, such as scavenger receptor-A (SR-A) and CD36 . Scavenger receptor-mediated uptake of modified LDL, unlike uptake of native LDL by LDL receptors, is not controlled by negative feedback regulation. Once internalized, the cholesteryl esters, which are abundant in LDL particles, are hydrolyzed by lysosomal acid lipase to yield free cholesterol . This form of cholesterol can be shuttled from macrophages to ApoA1 and high-density lipoprotein (HDL) particles via ATP-binding cassette transporters ABCA1 (ATP-binding cassette transporter A1) and ABCG1 (ATP-binding cassette transporter G1), respectively . Cholesterol effluxed onto ApoA1 and HDL particles is then transported to the liver where it can be metabolized to bile acids and subsequently excreted . The process of removing cholesterol from intimal lipid deposits by macrophages and its transport to the liver is referred to as macrophage reverse cholesterol transport (mRCT) and is thought to be responsible for the regression of atherosclerotic plaques .
In macrophages, not all free cholesterol generated by lysosomal acid lipase is promptly effluxed from the cell. Because free cholesterol is cytotoxic, it is re-esterified and stored as cholesteryl esters in lipid droplets within the cytoplasm . The cholesteryl esters in the droplets are later hydrolyzed by a neutral cholesterol ester hydrolase to again generate free cholesterol for efflux . As the plaque progresses and its volume increases, some macrophages take up more cholesterol than they are able to efflux thereby becoming foam cells. As the macrophage's ability to store the excess cholesterol becomes overwhelmed, the increased free cholesterol content results in endoplasmic reticulum stress-induced apoptosis as well as necrosis, thereby contributing to the accumulation of lipid in the subendothelial space . Thus, macrophages appear to have a beneficial role in the early stages of plaque formation, i.e. the removal of cholesterol, whereas at later stages they may enhance plaque formation when their capacity to process cholesterol is overwhelmed.
The two steps thought to be rate-limiting in cholesterol removal from atherosclerotic plaques by mRCT are the hydrolysis of cholesteryl esters stored in lipid droplets and ABCA1- and ABCG1-mediated transfer of intracellular cholesterol across the cell membrane to acceptor particles . Several candidate enzymes have been proposed to be responsible for the cytoplasmic neutral cholesteryl ester hydrolase activity, including carboxylesterase 1 (CES1) , which is a member of the serine hydrolase superfamily and an important xenobiotic detoxication enzyme in the liver . With regard to cholesterol metabolism, macrophage-specific transgenic expression of human CES1 in Ldlr−/− mice resulted in reduced atherosclerosis and increased mRCT , which is consistent with CES1 having a role in promoting cholesterol efflux from macrophages . We have also investigated the role of CES1 in cholesterol efflux from human THP-1 macrophages that were lipid loaded with acetylated low-density lipoprotein (acLDL) . The treatment of THP-1 foam cells with either diphenylethane-1,2-dione (benzil), a specific CES inhibitor, or paraoxon, a nonspecific CES inhibitor, both resulted in an increased retention of cholesteryl esters following a specified efflux period. This result was consistent with CES1 influencing cholesterol transport through a cytoplasmic neutral cholesteryl ester hydrolase activity. However, a limitation of this study is that benzil and paraoxon may interact with proteins other than CES1, raising the possibility of off-target effects. In addition, both inhibitors were added during the efflux period after the macrophages had been lipid loaded, so their effects (if any) on earlier steps in cholesterol metabolism were not assessed.
To address the issue of off-target effects of benzil and paraoxon, we knocked down the expression of CES1 in THP-1 cells [CES1KD (THP-1 cells with CES1 expression knocked down) cells] and assessed its impact on cholesterol efflux . Surprisingly, there was no difference in the fractional rate of cholesterol efflux between CES1KD cells and control cells; however, it was found that CES1 silencing reduced macrophage cholesterol uptake when compared with that in control cells following lipid-loading. Additionally, the transcription and expression of the scavenger receptors SR-A and CD36, the receptors responsible for most of the uptake of acLDL , were reduced in CES1KD foam cells when compared with control foam cells. However, the changes in the transcription and expression of SR-A and CD36 caused by CES1 knockdown were not observed in nonlipid-loaded macrophages (i.e. nonfoam cells) . In addition, a small decrease in ABCA1 expression was observed in CES1KD foam cells, but not in CES1KD macrophages, when compared with the appropriate control cells. Although the mechanism that controls SR-A transcription has not been described, it is known that the transcription of CD36 is regulated by the ligand-activated nuclear receptor PPARγ (peroxisome proliferator-activated receptor gamma) and its heterodimeric partner RXR [16,17]. Furthermore, the transcription of ABCA1 and ABCG1 is regulated by LXRα (liver X receptor α)/RXR [18–20]. This suggested that ablating CES1 expression in THP-1 macrophages and foam cells results in decreased signaling through the nuclear receptor heterodimers PPARγ/RXR and LXRα/RXR. PPARγ and LXR signaling are known to be linked; ligand-activated PPARγ was shown to directly activate the transcription of LXR in macrophages [16,17]. Moreover, Szanto et al.  showed that PPARγ could directly regulate the expression of CYP27A1 (cytochrome P45027A1), a cytochrome P450 enzyme that synthesizes the oxysterol 27-hydroxycholesterol (27-OHC), which is an endogenous ligand of LXRα . Additionally, RAR/RXR (retinoic acid receptor/retinoid X receptor) can also trans-activate CYP27A1 expression, suggesting a role for retinoids in the regulation of CYP27A1 as well .
In this study, we further investigated the role of CES1 in the regulation of cholesterol homeostasis in human THP-1 macrophages and primary human macrophages. The effect of ablating CES1 activity in both macrophages and foam cells on the transcription of many genes important in the regulation of intracellular cholesterol metabolism, including CYP27A1, is reported. We discovered that CES1 silencing profoundly reduced CYP27A1 transcript levels, CYP27A1 protein expression levels, and 27-OHC levels. This result and others reported here are consistent with a mechanism in which abrogation of CES1 expression alters the signaling of PPARγ/RXR and LXRα/RXR. We hypothesize that these effects are caused by the limited availability of CES1-derived endogenous ligands for these nuclear receptors.
Materials and methods
Cells, chemicals, and reagents
Human THP-1 (TIB-202) monocytes, high-glucose RPMI-1640 medium, and gentamicin sulfate solution (50 mg/ml) were purchased from the American Type Culture Collection (ATCC) (Manassas, VA). Primary human monocytes were obtained from Astarte Biologics (Bothell, WA). Low endotoxin containing fetal bovine serum (FBS) was purchased from Invitrogen (Carlsbad, CA). Acetylated LDL was from Intracel (Bethseda, MD) or Alfa Aesar (Ward Hill, MA). Antibodies against CYP27A1, LXRα, and GAPDH were purchased from Abcam (Cambridge, MA). Phorbol 12-myristate 13-acetate (PMA) was purchased from Sigma (St. Louis, MO), as were selective agonists of RXR (LG100268), RAR (TTNPB), RAR/RXR (9-cis-retinoic acid), and LXR (T0901317). The PPARγ agonist GW1929 and PPARγ antagonist GW9662 were purchased from Cayman (Ann Arbor, MI). The RXR antagonist HX531 was purchased from Fisher Scientific (Pittsburg, PA). The RAR inverse agonist BMS493 was from TOCRIS (Minneapolis, MN). The LXR antagonist GSK2033 was from Axon Medchem (Reston, VA). RIPA lysis buffer and protease inhibitors (PMSF, AEBSF, bestatin, pepstatin A, leupeptin hemisulfate, and aprotinin) were from Santa Cruz Biotechnology (Dallas, TX). Total RNA isolation kits and SYBR-green quantitative real-time qRT-PCR master mix were from Qiagen (Valencia, CA). cDNA synthesis reagents were purchased from Bio-Rad Laboratories (Hercules, CA). Primers for RT-PCR consisted of both pre-validated Quantitect primer assays (Qiagen) and self-designed custom oligonucleotide primers were from Invitrogen (Table 1). Lentiviral particles containing CES1, PPARγ, and scrambled control (nonspecific) short hairpin RNA (shRNA) constructs, puromycin hydrochloride, and polybrene were purchased from Santa Cruz Biotechnology. The serine hydrolase activity probes, fluorophosphonate (FP)-biotin and FP-TAMRA, were from Toronto Chemicals and Thermo-Pierce, respectively. WWL113 and methyl-β-cyclodextrin (MβCD) were from Sigma.
|Gene||Forward sequence||Reverse sequence|
|CES1||CES1_2_SG QuantiTect primer assay-QT01155581|
|CES2||CES2_1_SG QuantiTect primer assay-QT00037443|
|CES3||CES3_1_SG QuantiTect primer assay-QT00034692|
|PPARα||PPARa_1_SG QuantiTect primer assay-QT00017451|
|PPARβ/δ||PPARd_1_SG QuantiTect primer assay-QT00078064|
|DAGLβ||DAGLB_1_SG QuantiTect primer assay-QT00074319|
|CB1||CNR1_2_SG QuantiTect primer assay-QT02305702|
|CB2||CNR2_1_SG QuantiTect primer assay-QT00012376|
|GAPDH||GAPDH_1_SG QuantiTect primer assay-QT00079247|
|Gene||Forward sequence||Reverse sequence|
|CES1||CES1_2_SG QuantiTect primer assay-QT01155581|
|CES2||CES2_1_SG QuantiTect primer assay-QT00037443|
|CES3||CES3_1_SG QuantiTect primer assay-QT00034692|
|PPARα||PPARa_1_SG QuantiTect primer assay-QT00017451|
|PPARβ/δ||PPARd_1_SG QuantiTect primer assay-QT00078064|
|DAGLβ||DAGLB_1_SG QuantiTect primer assay-QT00074319|
|CB1||CNR1_2_SG QuantiTect primer assay-QT02305702|
|CB2||CNR2_1_SG QuantiTect primer assay-QT00012376|
|GAPDH||GAPDH_1_SG QuantiTect primer assay-QT00079247|
General cell culture conditions
Human THP-1 monocytes were cultured in RPMI-1640 medium containing 10% FBS, 2 mM l-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 4500 mg/l glucose, 1500 mg/l sodium bicarbonate, and 50 µg/ml gentamicin (complete growth medium) and maintained at 37°C at a density of 1 × 106 cells/ml with medium changes every 48–72 h. THP-1 monocytes were differentiated into macrophages by the addition of PMA (final concentration 100 nM) to complete growth medium for 72 h. Primary human monocytes were plated in complete growth medium and nonadherent cells removed the next day. The adherent cells were cultured for 10 days in complete growth medium. Culture medium was replenished with fresh medium approximately every 3 days.
CES1 and PPARγ expression were individually knocked down (CES1KD and PPARγKD, respectively) in THP-1 monocytes using lentiviral particles containing either a CES1 shRNA or a PPARγ shRNA construct, respectively, and subjected to puromycin selection, as described in detail previously for CES1 . Control cells were transduced using lentiviral particles containing scrambled shRNA constructs. Abrogation of CES1 expression and activity in THP-1 monocytes was verified by Western blot analysis, real-time PCR, and para-nitrophenyl valerate (pNPV) esterase activity assay. Silencing of PPARγ expression was verified by Western blot analysis and real-time PCR. Control, CES1KD, and PPARγKD monocytes were cultured in complete growth medium with 5 µg/ml puromycin hydrochloride. Only puromycin-resistant cells were used for subsequent experiments. In this report, wild-type cells, control cells, CES1KD cells, and PPARγKD cells refer to THP-1 cells that were nontransduced, transduced with scrambled shRNA, transduced with CES1-specific shRNA, or transduced with PPARγ-specific shRNA, respectively.
Activity-based protein profiling of wild-type, control, and CES1KD cell proteomes
Cell lysates (proteomes) of wild-type, control, and CES1KD cells (monocytes) were prepared in 50 mM Tris–HCl (pH 7.4) buffer by sonication (3 × 10 s, on ice), followed by protein quantitation (BCA reagent, Thermo-Pierce). Proteomes (1 mg/ml of lysate protein) were incubated with FP-TAMRA (2 µM) for 60 min at room temperature. The reactions were quenched by adding 6× SDS–PAGE loading buffer (reducing), followed by heating (90°C, 5 min). After electrophoresis on a 10% PAGE gel, the separated serine hydrolases covalently labeled by the FP-TAMRA probe were visualized using a Typhoon fluorescent gel imager (image shown in grayscale).
The ability of WWL113  to inhibit serine hydrolases in THP-1 wild-type monocytes was evaluated by treating cell lysate proteomes (1 mg protein/ml) with increasing concentrations of WWL113 (0.001–1 µM). After a 30 min preincubation with WWL113, cell proteomes were subsequently treated with FP-biotin (5 µM, 60 min, room temperature) and the reactions were quenched by the addition of 6× SDS–PAGE loading buffer (reducing) and heating (90°C, 5 min). Gel-based activity-based protein profiling (ABPP) was then performed as described previously .
In addition to ABPP, the in situ CES activity in THP-1 monocytes, following preincubation (30 min, 37°C) with either vehicle (DMSO, 0.1% v/v) or increasing concentrations of WWL113 (0.001–10 µM), was determined using the substrate pNPV as previously described . In this assay, intact living cells (105 cells/well) are incubated in PBS containing the inhibitor, then pNPV is added and the extent of its hydrolysis within the live cells is monitored for 10 min at 405 nm in a plate reader.
Quantitative real-time PCR analysis of mRNA expression
Control, CES1KD, and PPARγ THP-1 monocytes were differentiated into macrophages in complete growth medium with PMA (100 nM) for 72 h. Following differentiation, both control, CES1KD, and PPARγKD macrophages were incubated for either 24 or 48 h in complete growth medium (1% FBS instead of 10% FBS) with and without acLDL (50 μg/ml) and total RNA isolated. In addition, CES1KD macrophages were incubated for 24 h in complete growth medium (1% FBS) with 9-cis-retinoic acid (an RAR/RXR agonist, 1 µM), TTNPB (a specific RAR agonist, 1 µM), or LG100268 (a specific RXR agonist, 100 nM) and total RNA isolated. CES1KD and PPARγKD macrophages were incubated for 48 h in complete growth medium (1% FBS) with GW1929 (a PPARγ agonist, 10 µM). Finally, control macrophages were incubated for 24 h in complete growth medium (1% FBS) with GW9662 (a PPARγ antagonist, 10 µM), HX531 (an RXR antagonist, 100 nM), or BMS493 (a pan-RAR inverse agonist, 1 µM) and total RNA collected. Vehicle (DMSO) treatments were also utilized in all experiments. Total RNA was isolated using the RNeasy Plus Mini Kit (Qiagen) according to the manufacturer's protocol. Recovered RNA was quantified using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Waltham, MA), and cDNA was synthesized with an iScript Select cDNA Synthesis Kit (Bio-Rad) using oligo(dT) primers according to the manufacturer's protocol. Real-time PCR of cDNA products was performed on a Stratagene Mx3005P thermal cycler with Quantifast SYBR Green PCR master mix (Qiagen) using the QuantiTect primer assays or primers detailed in Table 1. The thermocycler program used for all target genes consisted of a 5 min hot start at 95°C prior to 40 cycles of 10 s at 95°C, followed by 30 s at 60°C, as recommended by the manufacturer. PCR product quality was assessed via dissociation curve analysis immediately following amplification. Differential expression of target genes was assessed by the ΔΔCt method  using GAPDH as the reference gene, and results are presented relative to control macrophages that were transduced with lentivirus containing a scrambled shRNA construct.
Western blot analysis of CES1, PPARγ, CYP27A1, and LXRα
Whole-cell lysates of control, CES1KD, and PPARγKD THP-1 macrophages (2 × 106 cells per sample), with and without 24-h acLDL loading, were prepared by sonication in RIPA lysis buffer containing protease inhibitors (PMSF, AEBSF, bestatin, pepstatin A, leupeptin hemisulfate, and aprotinin). Thirty microgram protein per sample, as determined by the bicinchoninic acid assay (Thermo Scientific), was separated on 10% SDS–PAGE gels prior to semidry transfer (20 V for 30 min) onto PVDF membranes. Membranes were blocked in 5% (w/v) nonfat dry milk in Tris-buffered saline containing Tween-20 (TBST: 10 mM Tris, 150 mM NaCl, 0.1% Tween-20, pH 7.4) for 1 h at room temperature and probed for GAPDH (Abcam 37168; final dilution 1 : 15 000) and CES1 (Abcam 68190; final dilution 1 : 25 000), PPARγKD (Abcam 19481; final dilution 1 : 500), CYP27A1 (Abcam 126785; final dilution 1 : 1000), or LXRα (Abcam 176323; final dilution 1 : 1000) overnight at 4°C. After washing with TBST, blots were probed with goat anti-rabbit IgG-HRP (Santa Cruz sc-2030; final dilution 1 : 15 000) for 1 h at room temperature. Following final washes, blots were visualized by enhanced chemiluminescence using Thermo Supersignal West Pico ECL reagent. The resulting films were scanned and densitometry analysis was performed using ImageJ v1.49a (NIH).
Biosynthesis of 27-OHC by monocytes and macrophages
A 100× working solution of ‘water-soluble’ cholesterol (155 mM MβCD/10 mM cholesterol, 15.5:1 mol : mol ratio) was prepared in water and mixed on a rotator overnight at room temperature. Cholesterol was completely dissolved under these conditions. The cholesterol/MβCD solution was sterile-filtered before adding to culture medium. Control and CES1KD monocytes were plated at a density of 2 × 106 cells per well in a six-well plate and treated with 100 nM PMA for 48 h to produce macrophages. After washing twice with serum-free medium, the cells were incubated in a 5% CO2 atmosphere at 37°C for 24 h in serum-free medium supplemented with ‘water-soluble’ cholesterol at a final concentration of 0.1 mM cholesterol/1.55 mM methyl-β-cyclodextrin. Negative controls included cells grown in medium that contained 1.55 mM methyl-β-cyclodextrin only. Assuming the Kd for the cholesterol·methyl-β-cyclodextrin complex is the same as that for 2-hydroxypropyl-β-cyclodextrin (Kd = 1.1 mM) , the concentration of free cholesterol in the culture medium was estimated to be 40 µM; thus, a sufficiently high concentration of cholesterol was available to the cells to be metabolized. After 24 h, 5β-cholestan-3α-ol (internal standard, 194 nmol) was added to each well as the internal standard. The samples (cells plus medium) were harvested together by scraping into a glass tube and extracted with an equal volume of heptane/isopropanol (3 : 2, v/v). The layers were separated by centrifugation (5 min, 1000×g). The top organic layer was transferred to a new glass tube and the organic extraction step was repeated. The pooled organic extracts were evaporated to dryness under a nitrogen gas stream. Samples were reconstituted in 100 μl of methanol for LC–MS/MS analysis.
LC–MS/MS analysis of 27-OHC
Samples were analyzed on an Acquity UPLC system (Waters, Milford, MA) coupled to a TSQ Quantum tandem mass spectrometer equipped with an atmospheric pressure chemical ionization (APCI) source (Thermo Fisher Scientific, San Jose, CA). Chromatographic separation was carried out using an Acquity UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 μm) equipped with a precolumn (2.1 mm × 5 mm, 1.7 μm) at 40°C. The mobile phases used were 2 mM ammonium acetate with 0.1% acetic acid (mobile phase A) and methanol with 0.1% acetic acid (mobile phase B). Mobile phase gradient conditions were as follows: hold at 90% A and 10% B for 1 min, linear increase of B to 98% over 16.5 min, hold at 98% B for 4.5 min, then decrease B to 3% in 3 min and re-equilibrate at the initial conditions for 2 min. The overall run time was 27 min and flow rate was 0.2 ml/min. 27-OHC was analyzed in positive ion mode and selected reaction monitoring (SRM). APCI source parameters were as follows: discharge current = 2.0 µA, vaporizer temperature = 450°C, sheath gas = 30, auxiliary gas = 5 units and capillary temperature = 190°C. Precursor-to-product ion transitions m/z 385.4 → m/z 161.3 for 27-OHC and m/z 371.5 → m/z 95.5 for 5β-cholestan-3α-ol (internal standard) were used for quantification. Scan times were 0.1 s per SRM, and the scan width was m/z 0.01. Optimum collision energy and tube-lens conditions were determined for each compound by using autotune software for each analyte by postcolumn infusion of the individual compounds into a 50% A/50% B blend of the mobile phase being pumped at a flow rate of 0.2 ml/min. Xcalibur software was employed for data acquisition and processing. For quantification, calibration standards were prepared using authentic standards ranging in concentration from 5 to 5000 nM, and three QC samples (low, medium, and high concentration) were prepared by adding stock 27-OHC dissolved in methanol to culture medium at 25, 250, and 2500 nM final concentrations (prepared in triplicate). Calibration curves were constructed using 1/× as a weighting factor for 27-OHC with a correlation coefficient (r2) of 0.999. Analyses of the QC samples were 93 ± 8%, 93 ± 8%, and 97 ± 4% of the target concentrations in the low, mid, and high QC samples (mean ± SD), respectively.
Intracellular total cholesterol levels following treatment with PPARγ/RXR and RAR/RXR agonists and an LXR antagonist
Control and CES1KD macrophages were pretreated for 24 h with DMSO vehicle or GW1929/9cRA (1 µM/10 µM) and GSK2033 (1 µM) in culture medium containing 1% FBS. The culture medium was then removed and replaced with fresh medium containing acLDL (50 µg/ml) and 0.2% (w/v) BSA. After loading the cells with cholesterol for 24 h, the acLDL-containing medium was removed and the cells were washed with 1× PBS. The cells were then overlaid with equilibration medium [containing 0.2% (w/v) BSA] and incubated for an additional 18 h. The medium and cells were harvested separately, and cells were scraped into 0.1 M potassium phosphate (pH 7.4), 0.05 M NaCl, 5 mM cholic acid, and 0.1% Triton X-100, and lysed by sonication. Following brief centrifugation to remove cell debris (1000×g, 5 min, 4°C), intracellular total cholesterol and free cholesterol levels in the supernatants were assayed using an Invitrogen Cholesterol Assay kit as per the manufacturer's instructions and normalized on cellular DNA content.
Biosynthesis of 2-arachidonoylglycerol, prostaglandins, and thromboxane B2 by monocytes and macrophages
In six-well plates, control and CES1KD monocytes (3 × 106 cells/well) in serum-free culture medium were pretreated with vehicle [DMSO, 0.1% (v/v)] or WWL113 (1 µM)  for 30 min, followed by treatment with 100 µM arachidonic acid (AA) to stimulate the cellular production of 2-AG (2-arachidonoylglycerol) . After 60-min incubation with AA, the cells and culture medium were harvested separately by centrifugation. The cell pellet was washed with PBS and suspended in 1.1-ml PBS, followed by removal of a 0.1-ml aliquot for protein determination. The remaining cell suspension and culture medium were each spiked with 2-AG-d8 (270 and 336 pmol, respectively). The cell suspension (in 1-ml PBS) was diluted with 1.5-ml of 2 : 1 (v/v) methanol/water and extracted with ethyl acetate (2 × 2 ml), while the culture medium was diluted with an equal volume of PBS and extracted using C18 SepPak columns. The organic extracts were evaporated to dryness under nitrogen. The residues were reconstituted in 100 μl of 1 : 1 (v/v) methanol/water and transferred to LC vials for LC–MS/MS analysis . For each sample, the amount of 2-AG in cells and medium were added together and the sum was normalized on the total protein content in each well (to control for well-to-well differences in cell numbers).
In a separate experiment using a stable deuterated isotope of 2-AG, wild-type THP-1 macrophages were incubated with 2-AG-d8 (2.5 µM) for either 0 or 4 h in serum-free medium. At each time point, ice-cold methanol (2 ml containing an internal standard 12-[[(cyclohexylamino)carbonyl]amino]-dodecanoic acid, 100 pmol) was added to each dish containing the cells and medium (this lyses the cells and terminates enzymatic reactions), and dishes were allowed to sit on ice for 15 min. The lysed cells and medium were then extracted together into ethyl acetate. Levels of AA-d8, PGE2-d7, and PGD2-d7 in the extracts were determined by LC–MS/MS.
Control and CES1KD macrophages were also treated with exogenous 2-AG (10 µM in serum-free medium) for 1 h. Culture medium and cells were then spiked with deuterated internal standards and extracted with ethyl acetate. Extracts were analyzed by LC–MS/MS for prostaglandins and AA as previously described . In another experiment, wild-type THP-1 macrophages were primed with LPS (1 µg/ml, 4 h), then the medium removed and cells gently washed with PBS. Fresh serum-free medium containing 2-AG (10 µM), with or without benzil (50 µM), was overlaid on the cells and incubation continued for 1 h. Culture medium and cells were spiked with deuterated internal standards and extracted into ethyl acetate. Extracts were analyzed by LC–MS/MS for prostaglandins and AA .
Results are presented as means ± SD. All statistical analyses were performed using either SigmaPlot version 11.0 or SAS 9.3. In some cases, data were transformed (log or square root) to ensure that normality and equal variance assumptions were met. For quantitative RT-PCR gene expression results, data were converted to linearized ΔΔCt values in accordance with the method published by Schmittgen and Livak . Statistical comparison between cell types and treatment groups was performed by two-way ANOVA with Student–Newman–Keuls post hoc testing. For Western blotting and some transcription analyses, statistical comparison between multiple groups was done using one-way ANOVA with Tukey's post hoc testing. Comparisons between two groups were analyzed using Student's t-test. In general, a value of P < 0.05 between groups was considered to be statistically different.
Silencing CES1 expression in human THP-1 monocytes/macrophages
CES1 is one of the most abundant metabolic serine hydrolases in THP-1 macrophages and its activity is potently inhibited by many drugs and environmental toxicants, including organophosphorus poisons (Supplementary Figure S1A). However, small molecules that inactivate CES1 often exhibit off-target effects that limit their use in mechanistic experiments. We previously reported, on the basis of reductions in mRNA, protein, and enzymatic activity, the stable ablation of CES1 in human THP-1 cells using a lentivirus encoding CES1 shRNA [15,23]. In the current study, we confirmed and extended these observations to include gel-based ABPP methodology using the fluorescent probe FP-TAMRA, which covalently reacts with the conserved catalytic serine residue in serine hydrolases. Several lines of evidence indicate that CES1 expression is selectively diminished in our CES1KD cells. First, CES1 mRNA levels were markedly lower in the CES1KD cells compared with the control cells, whereas CES2 and CES3 mRNA levels were not significantly altered by CES1 silencing (Supplementary Figure S1B). Relative to CES1 expression, CES2 and CES3 mRNA transcripts are present at very low levels in THP-1 cells (at least 24-fold lower; www.proteinatlas.org), and CES2 protein was previously undetectable by immunoblot analysis in this cell line . Second, immunoblot analysis indicated that CES1 protein expression was completely ablated in the CES1KD cells (Figure 1A). Third, a prominent band ∼60 kDa on the ABPP gel, denoted as CES1, was detected in both wild-type cells and control cells (which had been transduced with scrambled shRNA), but it was markedly diminished in intensity in the CES1KD cells (Figure 1B). Except for CES1, the profiles of serine hydrolase activities detected in the ABPP gel were, in general, qualitatively and quantitatively similar between wild-type, control, and CES1KD cells, suggesting that attenuated CES1 expression did not perturb the overall expression/activities of the other serine hydrolases detected by the ABPP probe.
CES1 silencing in THP-1 cells via lentivirus-mediated siRNA.
Silencing CES1 expression in human THP-1 macrophages alters the transcription of genes involved in cholesterol metabolism
We further discovered that silencing CES1 expression in THP-1 cells had profound effects on gene transcription in both macrophages and foam cells (loaded with acLDL for 48 h), as shown in Figure 2. As expected, CES1 mRNA levels were markedly reduced in the CES1KD macrophages relative to those in control macrophages regardless of acLDL loading status. Strikingly, in macrophages under basal conditions, CES1 silencing caused several genes that encode proteins involved in the metabolism, sensing, and signaling of lipids to be down-regulated, including CYP27A1, LXRα, DAGLβ (diacylglycerol lipase β), PPARγ, and RXRα. On the other hand, transcript levels of the scavenger receptors CD36 and SR-A were unchanged in CES1KD macrophages relative to those in control macrophages (<2-fold reduction; Figure 2 and Supplementary Figure S2), as we previously reported . CES1 silencing also had no effect on the transcription of ABCA1, ABCG1, CB1 (cannabinoid receptor 1), RARα, RARγ, NPC1, and NPC2 (Nieman Pick C1 and C2 proteins) in macrophages under basal conditions (Figure 2; NPC1 and NPC2 data not shown).
CES1 silencing alters the transcription of several genes involved in cholesterol metabolism.
CES1 silencing in macrophage foam cells also resulted in significant down-regulation of multiple genes involved in lipid homeostasis, including CYP27A1, CD36, SR-A, LXRα, ABCA1, DAGLβ, and StAR (steroidogenic acute regulatory protein; Figure 2; StAR data not shown). We previously reported the down-regulation of CD36 and SR-A when CES1 was knocked down . On the other hand, transcript levels of ABCG1, CB1, PPARγ, RXRα, RARα, RARγ, NPC1, and NPC2 were unaffected by CES1 silencing in foam cells (Figure 2; NPC1 and NPC2 data not shown). Interestingly, the expression of the PPARγ-regulated genes CYP27A1 and LXRα were consistently down-regulated in the CES1KD cells relative to those in control cells, regardless of their lipid-loading status. The expression of the PPARγ-regulated gene CD36 was also consistently down-regulated in the CES1KD cells relative to that of control cells after lipid loading, but was variable when the cells were not lipid loaded. Qualitatively similar results as those shown in Figure 2 were found when macrophages were loaded with acLDL for 24 h instead of 48 h (Supplementary Figure S2). Furthermore, there was a statistically significant cell type × acLDL treatment interaction term for CYP27A1 and CD36 (Figure 2 and Supplementary Figure S2), which strongly suggested that the two cell types exhibited a differential PPARγ-mediated induction response toward acLDL. PPARγ , RXRα , RARα , RARγ  DAGLβ , ABCG1 , and CB1  have been reported to exhibit changes in their mRNA levels that roughly correlate with changes in protein expression in various tissues. CD36 and SR-A mRNA transcript levels also correlate with their protein levels in control and CES1KD cells . Thus, assessing the mRNA levels of these genes is an appropriate surrogate measure of their protein levels.
Because our data suggested that cellular CES1 activity is linked to downstream PPARγ signaling, this nuclear receptor was silenced in the THP-1 cells to evaluate whether the pattern of gene expression in CES1KD cells mimics that of PPARγKD cells under basal and acLDL-loaded conditions. Knockdown of PPARγ in THP-1 monocytes/macrophages was verified by (i) the marked decrease in its mRNA levels (Figure 3), (ii) Western blot analysis (Figure 3, inset), and (iii) demonstrating that the induction of PPARγ-target genes LXRα and CD36 in control cells treated with a PPARγ agonist (GW1929) was absent from PPARγKD cells (data not shown). We further noted that PPARα and PPARδ expression were not altered by PPARγ silencing (Supplementary Figure S3). The induction patterns of CYP27A1, LXRα, and CD36 in control and PPARγKD macrophages (Figure 3), which had been treated with acLDL for 48 h, were similar to that of control and CES1KD macrophages. Furthermore, a statistically significant cell type × acLDL treatment interaction term for CYP27A1, LXRα, and CD36 was found (Figure 3), which indicates that PPARγKD macrophages do not respond to acLDL in the same way as control macrophages do. These genes were positively induced in control cells by acLDL loading, whereas their induction was blunted in PPARγKD cells, as was also observed in CES1KD cells. Thus, the CES1KD cells appeared to phenocopy the behavior of the PPARγKD cells when challenged with acLDL. In support of this, CES1KD macrophages  and PPARγKD macrophages (Supplementary Figure S4) both exhibited reduced cholesterol uptake compared with that of control macrophages when exposed to acLDL. This finding is consistent with the reduced amounts of CD36 transcript in both CES1KD and PPARγKD cells compared with that of control cells. In addition, activation of macrophages by LPS challenge resulted in attenuated levels of the lipid mediator PGE2 in both CES1KD and PPARγKD cells compared with that of control cells (Supplementary Figure S5). Together, these results suggest that a mechanistic link exists between CES1 and PPARγ.
PPARγ silencing in macrophages mimics the transcription profile of CES1KD macrophages.
Effect of CES1 silencing on CYP27A1-catalyzed biosynthesis of 27-OHC
Of the genes examined, the levels of CYP27A1 transcript were the most dramatically decreased by CES1 silencing (Figure 2). This reduction was confirmed at the protein level, on the basis of Western blots of control and CES1KD macrophage lysates for CYP27A1 protein (Figure 4A). Because CYP27A1 catalyzes the biosynthesis of the oxysterol 27-OHC, an important LXRα agonist, we assessed whether control cells and CES1KD cells differentially biosynthesize this oxysterol. After loading monocytes and macrophages with a water-soluble form of free cholesterol (cyclodextrin–cholesterol), the CES1KD cells produced significantly less 27-OHC than the control cells (Figure 4B), which suggest that CES1 deficiency might alter downstream LXRα signaling pathways. It was estimated that control macrophages produced ∼15× more 27-OHC than control monocytes did, which is likely due to the higher relative expression of CYP27A1 in macrophages vs. monocytes (Figure 4C). Cells incubated in the absence of exogenous-free cholesterol (negative controls) did not produce detectable amounts of 27-OHC in our assay (data not shown).
CYP27A1 protein expression and biosynthetic activity are reduced in CES1 knockdown cells.
Effect of nuclear receptor agonists on the transcription of CYP27A1, LXRα, and CD36 in CES1KD macrophages
Transcription of CYP27A1 and LXRα are regulated, in part, by the nuclear receptor heterodimers PPARγ/RXR and/or RAR/RXR [21,35]. Therefore, we next examined whether the addition of specific agonists of RXR (LG100268), RAR (TTNPB), PPARγ (GW1929), and RAR/RXR (9-cis-retinoic acid) to CES 1KD macrophages could restore the expression of CYP27A1 and LXRα to levels seen in control macrophages (Figure 5). CD36, whose expression is also regulated, in part, by PPARγ/RXR, was included as a positive control. Because CYP27A1 and LXRα are components of the PPARγ–CYP27A1–LXRα signaling axis, we also determined the effects of the nuclear receptor agonists on their protein expression. The results for CYP27A1, LXRα, and CD36 are described in turn.
Restoration of CYP27A1, LXRα, and CD36 expression in CES1 knockdown cells by nuclear receptor agonists.
The RAR and PPARγ agonists (TTNPB and GW1929, respectively) both increased CYP27A1 transcription in CES1KD macrophages relative to that of vehicle-treated CES1KD macrophages (Figure 5A,D). Furthermore, the combination of GW1929 and 9-cis-retinoic acid increased CYP27A1 transcription to a greater extent than GW1929 alone (compare Figure 5D and E). Moreover, treatment of CES1KD macrophages with either the RAR agonist (TTNPB) or the PPARγ agonist (GW1929) could partially (TTNPB, Figure 5B) or almost completely (GW1929, Figure 5F) restore CYP27A1 protein expression to levels seen in control cells. On the other hand, treatment with the RXR agonist LG100268 did not alter CYP27A1 mRNA and protein levels in CES1KD cells (Figure 5A,B, respectively).
Next, we examined the effects of nuclear receptor antagonists or inverse agonists on CYP27A1 transcription in control cells. The treatment of control macrophages with a PPARγ antagonist (GW9662), an RXR antagonist (HX531), or a pan-RAR inverse agonist (BMS439) caused CYP27A1 mRNA levels to be reduced relative to those in vehicle-treated control cells (Figure 6A–C). Furthermore, the combination of GW9662 and HX531 also reduced CYP27A1 mRNA levels compared with those seen in vehicle-treated control cells (Figure 6D).
Attenuation of CYP27A1 and LXRα transcription in control cells by nuclear receptor antagonists.
Together, these results suggested that CES1 has an important role in the regulation of CYP27A1 transcription through PPARγ/RXR and RAR/RXR signaling, and that the transcriptional changes in CYP27A1 are accompanied by corresponding changes in its protein levels.
Transcript levels of LXRα were also significantly reduced in CES1KD macrophages when compared with those in control macrophages (Figure 2). Therefore, the ability of specific agonists of PPARγ, RAR, and RXR to restore LXRα transcription and protein expression in CES1KD macrophages was determined (Figure 5A,C–E). The addition of an RXR agonist (LG100268, Figure 5A), PPARγ agonist (GW1929, Figure 5D), or combination of PPARγ agonist and nonselective RAR/RXR agonist (GW1929 and 9-cis-retinoic acid, respectively, Figure 5E) to CES1KD macrophages increased LXRα transcription relative to that of vehicle-treated CES1KD macrophages. Western blot analysis confirmed that the level of LXRα protein was reduced in the CES1KD macrophages when compared with that in control macrophages (Figure 5C). However, treatment of CES1KD cells with either LG100268, TTNPB, or GW1929 did not have a significant effect on LXRα protein expression when compared with vehicle-treated CES1KD cells (Figure 5C; data not shown for GW1929), although the LG100268-treated and GW1929-treated CES1KD cells exhibited significantly greater LXRα mRNA levels relative to that of vehicle-treated CES1KD cells (Figure 5A,D, respectively). The RAR agonist (TTNPB) also had no effect on LXRα transcription (Figure 5A).
Although the treatment of control macrophages with a selective PPARγ antagonist (GW9662) only slightly affected LXRα mRNA levels when compared with vehicle-treated control cells (Figure 6A), the selective RXR antagonist (HX531), either alone or in combination with GW9662, significantly reduced LXRα transcription levels (Figure 6B,D). The pan-RAR inverse agonist BMS439 also reduced LXRα transcript levels (Figure 6C).
Together, these findings suggested that CES1 silencing in THP-1 macrophages can affect LXRα transcriptional regulation. The results indicate that LXRα transcription is partially restored in CES1KD macrophages by both PPARγ and RXR agonists, whereas its transcription in control macrophages could be reduced by PPARγ, RXR, and RAR antagonists.
The transcript levels of CD36 were only modestly reduced in CES1KD macrophages when compared with those in control macrophages, whereas they were significantly reduced in CES1KD cells vs. control cells after acLDL loading (Figure 2 and Supplementary Figure S2). The addition of RXR agonist (LG100268) or PPARγ agonist (GW1929) to CES1KD macrophages increased CD36 transcription relative to that of vehicle-treated CES1KD macrophages (Figure 5A,D). Treatment with a combination of GW1929 and 9-cis-retinoic acid increased CD36 transcript levels ∼5-fold. On the other hand, the RAR ligand (TTNPB) had no effect on CD36 transcription (Figure 5A). Consistent with a deficit of PPARγ ligands in CES1KD cells, supplementation of CES1KD cells, exposed or not to oxLDL, with the prototypical PPARγ ligand rosiglitazone also restored CD36 expression to control cell levels, whereas it had no effect on the SR-A expression defect (Supplementary Figure S6).
Repair of cholesterol metabolism defect in CES1KD macrophages by pretreatment with PPARγ/RXR and RAR/RXR ligands
CES1KD macrophages exhibit less cholesterol uptake compared with that of control macrophages following exposure to acLDL , which is attributable to reduced CD36 and SR-A expression. This effect on total cholesterol level was confirmed (Figure 5G). Moreover, pretreatment of control and CES1KD macrophages with ligands that activate PPARγ/RXR and RAR/RXR (GW1929 and 9cRA, respectively) prior to cholesterol loading (acLDL) caused significant reductions in intracellular total cholesterol levels (Figure 5G). Despite the fact that acLDL exposure promotes net cholesterol accumulation in cells, the presence of PPARγ/RXR and RAR/RXR ligands will also up-regulate the LXR-dependent cholesterol efflux pathway (e.g. induction of the LXR/ABCG1 axis) that counteracts cholesterol influx during the loading period. Thus, in addition to PPARγ/RXR and RAR/RXR ligands, we also added an LXR antagonist (GSK2033) to block the confounding effect of cholesterol efflux caused by up-regulated ABCG1-mediated aqueous diffusion of free cholesterol to acLDL acceptors present in the extracellular milieu . As depicted in Figure 5G, this approach repaired the cholesterol influx deficit of the CES1KD cells. The level of intracellular total cholesterol in CES1KD cells pretreated with the triple combination of GW1929/9cRA/GSK2033 prior to cholesterol loading was significantly greater than that of CES1KD cells treated with DMSO, whereas it was not statistically different from that of control cells treated with DMSO, indicating that normalization of cholesterol influx had occurred in the CES1KD cells. As expected, control cells pretreated with GW1929/9cRA/GSK2033 exhibited more intracellular total cholesterol compared with that of control cells treated with DMSO (Figure 5G).
Inhibition of CES1 activity in human wild-type THP-1 cells with a small-molecule inhibitor blocks 2-AG catabolism and recapitulates the effects on CYP27A1 transcription seen in CES1KD cells
We next evaluated whether selective inhibition of CES1 activity in wild-type cells using a small molecule (WWL113; ) would have the same effect on CYP27A1 transcription as CES1 shRNA did. On the basis of gel-based competitive ABPP , we established that WWL113 was a selective and potent inhibitor of CES1 activity in wild-type THP-1 cell proteomes (Figure 7A). The IC50 for WWL113 toward CES1 in THP-1 cell lysates was estimated to be ∼0.01 µM. Furthermore, treatment of intact monocytes with WWL113 (1 µM, 1 h) could inhibit ∼85% of the pNPV hydrolytic activity in THP-1 cells, establishing that CES1 could be targeted by this inhibitor. Moreover, pretreatment of intact control monocytes with WWL113 (1 µM, 1 h) increased the level of endogenous 2-AG, a lipid mediator and a known CES1 substrate , compared with that of vehicle-treated control cells upon stimulation with exogenous AA (Figure 7B). We have previously shown that free AA is an effective stimulant of 2-AG biosynthesis in cells via a Nox-dependent mechanism (; Supplementary Figure S7). Furthermore, the amount of 2-AG was higher in vehicle-treated CES1KD cells than that in vehicle-treated control cells, which was expected since CES1 has a role in regulating 2-AG degradation. There was no difference, however, in 2-AG levels between the vehicle- and WWL113-treated CES1KD cells (Figure 7B), which is also consistent with CES1 having a role in 2-AG catabolism in THP-1 cells. When wild-type THP-1 monocytes were differentiated into macrophages in the presence of WWL113 (1 µM, 48 h), the level of CYP27A1 mRNA was found to be significantly lower in the inhibitor-treated cells than in vehicle-treated cells, whether or not the cells had been exposed to acLDL (Figure 7C). This result is consistent with the substantial down-regulation in CYP27A1 transcription seen in the CES1KD cells (Figure 2), although not to the same extent. Furthermore, treatment of primary human macrophages with WWL113 (1 µM, 48 h) also resulted in lower CYP27A1 mRNA levels (Figure 7D).
The pharmacological agent WWL113 is a selective inhibitor of CES1 activity in THP-1 wild-type cells and attenuates CYP27A1 transcription.
We previously reported that inactivation of DAGLβ, a 2-AG biosynthetic enzyme, by RHC-80267 decreased prostaglandin synthesis in LPS/ionomycin-stimulated THP-1 macrophages . This result was consistent with the attenuated 2-AG and prostaglandin levels in murine peritoneal macrophages when DAGLβ activity was ablated genetically or with a small-molecule inhibitor . We extended these findings by treating intact THP-1 wild-type macrophages with deuterium-labeled 2-AG-d8 for 4 h and observed the production of AA-d8, PGE2-d7, and PGD2-d7 (Figure 8A). This demonstrated that 2-AG is a direct source of precursor AA that is subsequently metabolized by macrophages to produce prostaglandins (the conversion of AA-d8 into PGE2-d7 and PGD2-d7 results in the loss of a single deuterium atom). WWL113-treated cells and CES1KD cells both exhibited reduced 2-AG-d8 hydrolytic activity compared with their respective controls (Supplementary Figure S8). Additional evidence that 2-AG levels are regulated by CES1 was noted when control and CES1KD monocytes were differentiated into macrophages for 48 h, and steady-state amounts of 2-AG in the cell supernatants were determined (Figure 8B); CES1KD supernatants had ∼2.5× more 2-AG than control supernatants. Consistent with the fact that 2-AG hydrolysis provides eicosanoid precursor, when control and CES1KD macrophages were incubated with exogenous 2-AG (10 µM, 1 h), the levels of PGE2 and PGD2 levels in CES1KD cells were lower than those in control cells (Figure 8C), although TxB2 levels were not significantly different. Interestingly, a small-molecule inhibitor (benzil) that reversibly inhibits CES1 in THP-1 cells  also reduced the levels of 2-AG-derived PGE2, PGD2, and PGF2α when LPS-primed wild-type THP-1 macrophages were treated simultaneously with it and 2-AG (Figure 8D; Wang et al. ). On the other hand, benzil had no effect on TxB2 levels (Figure 8D), similar to what was observed in CES1KD cells (Figure 8C).
THP-1 macrophages metabolize 2-AG yielding prostaglandins.
Silencing CES1 expression had no effect on LPS-stimulated induction of TNFα and IL-6 (Supplementary Figure S9). This result suggests that LPS-evoked signaling through TLR-4 and NF-κB remains intact. Consistent with this notion, expression of TLR-4 was actually somewhat increased in CES1KD cells compared with control cells both at baseline and following LPS treatment. In addition, mRNA levels of the prostaglandin synthetic enzymes PTGS1 (COX1), PTGS2 (COX2), PTGES3, and TBXAS1 were unchanged in CES1KD macrophages compared with controls (Supplementary Figure S10). Only PTGDS expression levels were reduced in the setting of CES1 silencing.
Effect of CES1 silencing on the transcription of cholesterol-25 hydroxylase, CYP46A1, and 15-lipoxygenase
Because knockdown of CES1 expression reduced the transcription of CYP27A1, we also examined the transcription of two other genes that encode oxysterol-synthetic enzymes, namely cholesterol-25 hydroxylase (CH25H), which synthesizes 25-OHC (25-hydroxycholesterol), and cholesterol-24 hydroxylase (CYP46A1), which synthesizes 24-OHC. Similar to 27-OHC, these oxysterols are also ligands for LXRα. Whereas CYP46A1 transcripts in cells were below the limit of detection by our RT-PCR assay, the level of CH25H transcript was slightly greater in CES1KD cells than in control cells (1.6-fold, P = 0.008, data not shown).
The effect of CES1 silencing on the transcription of 15-lipoxygenase (ALOX15), an enzyme responsible for the synthesis of hydroxyoctadecadienoic acids (HODEs) and hydroxyeicosotetraenoic acids (HETEs), which are oxygenated metabolites of linoleic acid and AA, respectively, was also examined. The rationale for examining ALOX15 is that HODEs and HETEs are endogenous PPARγ ligands, and thus, we wondered whether CES1 silencing affected the transcription of ALOX15. Although transcript levels of ALOX15 were above the limit of detection in both control and CES1KD macrophages, they were not statistically different from each other (2.1-fold greater in CES1KD macrophages, but P = 0.16, data not shown).
The PPARγ–CYP27A1–LXRα signaling axis is a feedforward pathway that regulates the transcription of many genes in macrophages involved in restoring cholesterol homeostasis . Here, we present evidence that the metabolic serine hydrolase CES1 may regulate macrophage cholesterol metabolism in part by signaling through this axis. We found that silencing CES1 expression in THP-1 macrophages resulted in a profound decrease in the transcription and expression of the oxysterol biosynthetic enzyme CYP27A1 and a concomitant decrease in the levels of 27-OHC, an endogenous oxysterol agonist for LXRα. Importantly, the inhibitory effect on CYP27A1 transcription was recapitulated in THP-1 wild-type cells and primary human macrophages treated with a pharmacological agent WWL113 that selectively inhibits CES1 activity. Furthermore, three lines of evidence indicate an important role for PPARγ in the disrupted CYP27A1 expression: (i) CYP27A1 transcription and expression in CES1KD cells could be restored, in part, by the addition of a PPARγ agonist (GW1929); (ii) CYP27A1 transcription was significantly attenuated in control cells treated with a PPARγ antagonist (GW9662); and (iii) the induction profile of genes (including CYP27A1) in CES1KD cells exposed to acLDL phenocopied that of PPARγKD cells. LXRα and CD36, like CYP27A1, are also regulated by the heterodimeric PPARγ/RXR nuclear receptors [21,40], and their transcription and expression were reduced by ablating CES1 expression when macrophages were lipid loaded, although not to the degree that CYP27A1 was. Altogether, these results suggested a mechanistic link between CES1 and PPARγ. It should be noted that CYP27A1 expression was also regulated, in part, by RAR, because a specific agonist for this nuclear receptor (TTNPB) could partially restore CYP27A1 transcription and expression in CES1KD cells (Figure 5A,B). This finding is also consistent with that of Szanto et al. , which suggested that CYP27A1 expression was postively regulated by RAR in addition to PPARγ.
The LXR-regulated ABCA1 transcript levels in CES1KD foam cells were also reduced when compared with those in control foam cells, which indicated that perturbations in PPARγ signaling in CES1KD cells appeared to lead to wide-range reprograming in gene expression, though ABCG1 levels were not reduced. A small reduction in ABCA1 expression, but a larger reduction in ABCG1 expression, was previously found when PPARγ signaling in cells was inhibited by conditional disruption of the PPARγ gene in mice . The difference in our results and those previously reported may be due to species differences. Indeed, different responses to PPARγ agonists have been previously described in murine and human macrophages [41,42]. We also noted some variability in CD36 mRNA levels in CES1KD macrophages that were not lipid loaded. CD36 transcriptional regulation is complex and is controlled by transcription factors other than PPARγ/RXR, including Nrf2  and TR4 . Subtle changes in signaling through these or other factors might account for the variability in CD36 transcription in CES1KD macrophages under basal conditions. In addition, Zhao et al.  reported that stable overexpression of CES1 in THP-1 cells resulted in increased levels of mRNA for SR-A, CD36, ABCA1, and ABCG1 when cells were lipid loaded (CYP27A1 and LXR expression was not assessed), which is the opposite of what we observed for these genes in CES1KD cells. Together, our results suggest that the loss of CES1 expression in THP-1 macrophages attenuates signaling pathways that operate through the ligand-activated nuclear receptors PPARγ/RXR and RAR/RXR. Functionally, this altered signaling results in defective cholesterol influx into the CES1KD macrophages (Figure 9); however, the influx deficit could be repaired by pretreatment of CES1KD macrophages with GW1929 and 9-cis-RA (in the presence of an LXR antagonist to block the confounding effects of the cholesterol efflux pathway) (Figure 5G). This result suggested that supplementing CES1KD cells with ligands that activate PPARγ/RXR and RAR/RXR overcomes the limited availability of endogenous ligands in CES1-deficient macrophages and normalizes cholesterol homeostasis.
Schematic pathway representing findings of the present study.
We next considered how ablation of CES1 activity might mechanistically alter the gene expression profile. CES1 is a notable serine hydrolase because it exhibits a very broad substrate specificity . It is, therefore, feasible that one of its physiological functions is to hydrolyze ester-containing cellular metabolites (pro-ligands) to generate endogenous ligands that affect signaling through the PPARγ/RXR heterodimer. CES1 can metabolize several classes of neutral lipid esters, including cholesteryl esters, glyceryl esters, and retinyl esters . For example, CES1 can hydrolyze the monoacylglycerol 2-AG liberating free AA, which is further metabolized in macrophages to prostaglandins [23,38]. Our results directly showed that CES1 can regulate 2-AG levels in THP-1 cells under different conditions (Figures 7B and 8B), and it is known that 2-AG-derived AA is utilized by COX (cyclooxygenase) enzymes to produce eicosanoids [39,46]. Consistent with this, CES1KD cells synthesized smaller quantities of several prostaglandins than control cells did upon cellular metabolism of exogenous 2-AG (Figure 8). An exception to this was TxB2 (the stable metabolite of TxA2), which suggests that a different pool of AA is used for TxA2/B2 synthesis than for PGE2 and PGD2. In addition, CES1KD macrophages produced less PGE2 than control macrophages when activated by LPS (Supplementary Figure S5). Cytokine expression (TNFα and IL-6), however, remained intact in the LPS-stimulated CES1KD cells (Supplementary Figure S9), which suggested that TLR-4-mediated signaling and downstream NF-κB activity are unaffected by CES1 silencing. Levels of COX-1 and COX-2 transcripts were also similar between control and CES1KD macrophages (Supplementary Figure S10). Although the PGD2 synthase transcript (PTGDS) was lower in CES1KD cells compared with control cells, transcript levels of the PGE2 and TxA2 synthases (PTGS3 and TBXAS1, respectively), which make the two most quantitatively important prostaglandins in THP-1 cells, were not different in the two cell types. Therefore, in general, the levels of the prostaglandin synthetic enzymes in CES1KD cells were unaltered compared with control cells, which indicated that the decrease in prostaglandin levels in CES1KD macrophages is caused by a deficit in the provision of AA substrate to COX-1 and COX-2 enzymes. The significance of this is that several COX-derived oxylipins are ligands for PPARγ [47–49], but these oxylipins were depleted in the CES1KD cells when compared with the control cells.
In addition to monoacylglycerols, other neutral lipid esters are substrates of CES1. For example, CES1 hydrolyzes triacylglycerols , generating free fatty acids that can undergo additional oxidative biotransformation . Furthermore, lipid glyceryl esters containing oxidized fatty acyl moieties are also substrates for CES1. For instance, prostaglandin glyceryl esters are hydrolyzed by both CES1 and CES2 , liberating prostaglandins that might act as PPARγ ligands.
The origin of PPARγ ligands in the setting of scavenger receptor-mediated uptake of oxLDL is presumed to be the oxidized fatty acid components of oxidatively damaged cholesteryl esters and phospholipids [16,52]. Macrophages, which have been lipid loaded with acLDL, respond in a similar way to those lipid-loaded with oxLDL, i.e. both loading procedures result in macrophage foam cells. Internalized acLDL likely provides an abundance of substrates for oxidative enzymes, such as COX and lipoxygenase, in addition to containing preformed oxidized fatty acyl moieties. Interestingly, cholesterol esters that contain oxidized fatty acyl moieties were reported to be better substrates for neutral cholesterol ester hydrolase than cholesterol esters containing unoxidized fatty acyl moieties [53,54]. This suggests the possibility that CES1 might liberate endogenous ligands for PPARγ by hydrolyzing neutral lipid esters that possess oxidized fatty acyl moieties. More work is required, however, to understand how CES1 in macrophages affects lipid glyceryl ester metabolism and downstream pathways that might furnish endogenous nuclear receptor ligands. These studies are underway in our laboratories.
CES1 may also have an underappreciated role in vitamin A metabolism. Macrophages are exposed to multiple sources of vitamin A, including retinol bound to the retinol-binding protein 4-transthyretin complex (RBP4-TTR), retinoic acid bound to albumin, retinyl esters in VLDL and LDL particles, retinyl esters in chylomicrons, and β-glucuronides of both retinol and retinoic acid . The exact role each has in the supply of vitamin A to macrophages is currently uncertain. However, CES1 may increase the availability of ligands for RAR and RXR via the hydrolytic liberation of retinol stored in retinyl esters. Support for this idea comes from the fact that two rat orthologs of human CES1, termed ES10 and ES4, exhibit retinyl palmitate hydrolase activity  and are present in large amounts in rat hepatocytes . It should be noted, however, that ES10 and ES4 are not expressed in hepatic stellate cells , where most vitamin A is stored as retinyl esters. Whether or not human CES1 can hydrolyze retinyl esters has not been determined. Retinol, once released, needs to be converted to retinoic acid to affect gene expression via RAR and/or RXR. A previous study suggested that macrophages in human intestine can biosynthesize retinoic acid from retinol, and that retinoic acid plays an important role in regulating the macrophage phenotype . Overall, the roles of retinoids and CES1 in the regulation of human macrophage gene transcription still need to be explored.
The potential for CES1 to regulate the availability of ligands for PPARγ, RAR, and RXR, and, via these ligand-activated nuclear receptors, to control CYP27A1 expression, might have broader implications (Figure 9). For example, CYP27A1-derived oxysterol, 27-OHC, may be involved in the negative feedback regulation of CES1 activity. We previously described the ability of some oxysterols and polyunsaturated fatty acids to potently inhibit CES1 activity . 27-OHC exhibited an apparent inhibitory constant (Ki) for CES1 ∼10 nM and showed a significant degree of structural specificity. For instance, 25-OHC, a regioisomer of 27-OHC, had little capacity to inhibit CES1. In addition, several fatty acids could also inhibit CES1 activity, with unsaturated fatty acids being more potent than saturated fatty acids. Interestingly, AA was the most potent fatty acid inhibitor (Ki = 1.7 µM). Furthermore, the PGD2-derived metabolite 15-deoxy-Δ12,14 prostaglandin J2, a known PPARγ ligand , was also an inhibitor of CES1. The ability of these lipids to inhibit CES1 suggests a potential self-limiting mechanism whereby CES1 hydrolyzes pro-ligand esters to liberate either preformed PPARγ ligands or fatty acids that are subsequently oxygenated to products that are PPARγ ligands (Figure 9). Release of PPARγ ligands then up-regulates CYP27A1 expression resulting in the increased synthesis of 27-OHC . Subsequently, fatty acids, released by the hydrolysis of neutral lipid esters, and 27-OHC might directly interact with CES1 and inhibit its enzymatic activity. If such a regulatory pathway is operating in macrophages, ablation of CES1 activity may well affect other PPARγ-mediated processes, e.g. macrophage polarization.
In conclusion, our results are consistent with a model in which CES1 abrogation results in decreased CYP27A1–LXRα signaling by potentially reducing the availability of endogenous ligands for PPARγ, RAR, and/or RXR. We plan to further investigate the role that CES1 plays in macrophage cholesterol homeostasis and its molecular mechanisms, including identification of its endogenous substrates. Finally, it should be pointed out that our interpretation of CES1's ability to regulate macrophage gene expression does not require it to directly act as a cholesteryl ester hydrolase, a catalytic activity that is intensely debated for CES1 [59–62].
ATP-binding cassette transporter A1
ATP-binding cassette transporter G1
activity-based protein profiling
acetylated low-density lipoprotein
atmospheric pressure chemical ionization
cannabinoid receptor 1
THP-1 cells with CES1 expression knocked down
diacylglycerol lipase β
fetal bovine serum
apparent inhibition constant for inhibitor
liver X receptor
macrophage reverse cholesterol transport
- NPC1 and NPC2
Nieman Pick C1 and C2 proteins
oxidized low-density lipoprotein
phorbol 12-myristate 13-acetate
peroxisome proliferator-activated receptor gamma
retinoic acid receptor
retinoid X receptor
selected reaction monitoring
steroidogenic acute regulatory protein
L.C.M., X.H., M.K.R., J.A.C. conceived the study, performed experiments, analyzed/interpreted data, and wrote manuscript. A.B. and J.H.L. performed experiments and analyzed/interpreted data.
This work was supported by the National Institutes of Health [grant numbers 1R15GM116129-01 and 1R15ES015348-02]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or Mississippi State University.
The Authors declare that there are no competing interests associated with the manuscript.
These authors contributed equally to this work.