The ATPase activity of ABCA1 is increased by cholesterol in the presence of anionic lipids

Abstract

Graphical Abstract

Open in new tabDownload slide

Abbreviations

 
• apoA-I

  apolipoprotein A-I

 
• CBB

  Coomassie Brilliant Blue

 
• CHAPS

  3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate

 
• DMEM

  Dulbecco’s modified Eagle’s medium

 
• DOPA

  1,2-dioleoyl-phosphatidic acid

 
• DOPC

  1,2-dioleoyl-phosphatidylcholine

 
• DOPE

  1,2-dioleoyl-phosphatidylethanolamine

 
• DOPG

  1,2-dioleoyl-phosphatidylglycerol

 
• DOPS

  1,2-dioleoyl-phosphatidylserine

 
• ER

  endoplasmic reticulum

 
• FBS

  fetal bovine serum

 
• HBSS

  Hank's balanced salt solution

 
• HDL

  high-density lipoprotein

 
• HRP

  horse radish peroxidase

 
• LMNG

  lauryl maltose neopentyl glycol

 
• MαCD

  methyl-α-cyclodextrin

 
• MβCD

  methyl-β-cyclodextrin

 
• PA

  phosphatidic acid

 
• PC

  phosphatidylcholine

 
• PE

  phosphatidylethanolamine

 
• PEI

  polyethyleneimine ‘MAX’

 
• PG

  phosphatidylglycerol

 
• POPS

  1-palmitoyl-2-oleoyl-phosphatidylserine

 
• PS

  phosphatidylserine

 
• SM

  sphingomyelin

ATP-binding cassette protein A1 (ABCA1) is a member of the ABC protein superfamily and a key player in the generation of high-density lipoprotein (HDL) (1,  2,). HDL carries excess cholesterol from peripheral tissues back to the liver, and plasma HDL levels are inversely related to the incidence of cardiovascular diseases (3,). ABCA1 is essential for HDL generation because missense mutations in ABCA1 cause severe hypercholesterolemia or Tangier disease (4–6,) and ABCA1 knockout mice exhibit a severe reduction in plasma HDL levels (7–9,). Human ABCA1 is a 2,261-amino acid membrane protein consisting of two transmembrane domains, two large extracellular domains (10) and two nucleotide-binding domains that provide energy for transport through the ATP binding/hydrolysis cycle.

ABCA1 mediates the generation of nascent HDL, which is a lipid disc composed of several hundred lipid molecules wrapped by apolipoprotein A-I (apoA-I), a lipid acceptor in serum (11–14,). Since cholesterol and phosphatidylcholine (PC) are the major components of nascent HDL (12,  14,), it is postulated that ABCA1 transports cholesterol and PC. ABC proteins require the energy from ATP binding and hydrolysis to drive substrate transport across the cell membrane. Typically, ATPase activities of ABC proteins are increased by transport substrates (15–17,), and this property is useful for identifying transport substrates. We and others previously reported that the ATPase activity of ABCA1 was enhanced by PC, but not cholesterol (18,  19). Consequently, it remains unclear whether cholesterol is a transport substrate of ABCA1, although ABCA1 is important for cholesterol clearance from peripheral tissues.

Dietary cholesterol is absorbed in the intestine and then transported to the liver via lipoproteins (20,). Dietary plant sterols, including stigmasterol and β-sitosterol, are also absorbed in the intestine, but most of them are secreted into the bile and intestinal lumen by the heterodimeric sterol transporter ABCG5/G8 (21,). Accumulation of plant sterols causes sitosterolemia, a rare disorder characterized by premature coronary atherosclerosis (22,). It has been suggested that ABCA1 contributes to the absorption of cholesterol and β-sitosterol in the intestine (23,  24,), although Tachibana et al. reported that apoA-I–mediated β-sitosterol efflux was not observed in Caco-2 cells (25). Therefore, the sterol specificity of ABCA1 and the physiological importance of ABCA1-mediated sterol absorption remain controversial.

The membrane phospholipid environment is important for the function of many membrane proteins (26,  27,). Phosphatidylserine (PS) is a major anionic lipid in mammalian cells and is involved in the modulation of G protein-coupled receptors and ion channels such as cystic fibrosis transmembrane conductance regulator, an ABC protein that functions as a chloride ion channel (28,  29,). Another anionic lipid, phosphatidylinositol 4,5-bisphosphate, binds to and activates many ion channels and ion transporters (30,). It has also been reported that cardiolipin regulates the activity of ABCB10 (31). Therefore, it is possible that specific phospholipids are required for the recognition of cholesterol by ABCA1.

In this study, we purified human ABCA1 expressed in human cells and analyzed the effects of various phospholipids and sterols on the ATPase activity of ABCA1. Our results suggest that ABCA1 recognizes cholesterol as a transport substrate in the presence of anionic lipids, including PS, and prefers cholesterol over plant sterols as a transport substrate.

Materials and Methods

Materials

FreeStyle™ 293-F cells, FreeStyle™ 293 expression medium, gentamicin, Opti-MEM™ I, fetal bovine serum (FBS), Hank's balanced salt solution (HBSS) and Amplex UltraRed were purchased from Thermo Fisher Scientific. Dulbecco's modified Eagle’s medium (DMEM) and Triton X-100 were purchased from Nacalai Tesque (Japan). ANTI-FLAG M2 affinity gel, 1 × FLAG peptide, 3 × FLAG peptide, MβCD, cholesterol, 7-dehydrocholesterol, stigmasterol, and lanosterol, phospholipase D and choline oxidase were purchased from Sigma-Aldrich. CHAPS, sodium cholate, β-sitosterol, campesterol, brassicasterol and ergosterol were purchased from FUJI-FILM Wako Pure Chemical (Japan). Polyethyleneimine ‘MAX’ (PEI) was purchased from PolyScience. Amicon Ultra was purchased from Merck Millipore. LMNG was purchased from Anatrace. MαCD was purchased from AraChem (Malaysia). Cholesterol oxidase and horse radish peroxidase (HRP) were purchased from Oriental Yeast (Japan). The other lipids were purchased from Avanti Polar Lipids. Recombinant apoA-I fused with 6 histidine residues at the C-terminus was prepared previously (32).

Construction of expression vectors

Human ABCA1 and the non-functional mutant ABCA1-K939M-K1952M (ABCA1-MM) were subcloned into pEGFP-N3 vectors as described previously (32), with a GFP tag at the C-terminus. Triple FLAG (DYKDDDDK×3) tag and 10 histidine residues were inserted between ABCA1 and the GFP tag using an In-Fusion HD cloning Kit (Clontech).

Cell culture and transfection

Suspension-adapted HEK293 cells (FreeStyle™ 293-F cells) were grown in FreeStyle™ 293 expression medium containing 5 μg/ml of gentamicin in a humidified incubator containing 8% CO₂ at 37°C. Cells were maintained in 5-L flasks and shaken at 100 rpm. BHK/ABCA1 cells were kindly provided by the late Dr. John Oram of the University of Washington (33,). BHK/ABCA1-MM cells were generated as described previously (34). These cells were cultured in DMEM containing 10% FBS at 37°C under 5% CO₂.

For protein expression, expression vectors and PEI were separately dissolved in Opti-MEM™ I at final concentrations of 20 and 80 μg/ml, respectively. Equal volumes of these two solutions were then mixed and incubated for 30 min. The PEI–DNA complex (250 ml) was added to the cell culture (2.5 l). The cells were incubated for 48 h after transfection.

Protein purification

Purification of ABCA1 and ABCA1-MM was performed as described previously (32), with some modifications. In brief, ABCA1-expressing cells were harvested and resuspended in buffer A (50 mM HEPES (pH 7.4), 150 mM NaCl, 50 mM KCl, 10% glycerol) containing 1.25% (w/v) CHAPS and protease inhibitors (cOmplete™ Protease Inhibitor Cocktail; Roche). After 30-min incubation at 4°C, insoluble materials were removed by centrifugation (45,000 × g, 30 min). The solubilized proteins were applied to anti-FLAG M2 affinity gel and were rotated for more than 18 h. The ABCA1-bound resin was washed three times with buffer A containing 0.5% (w/v) CHAPS and was then washed three times with buffer A containing 0.01% (w/v) LMNG. The resin was further washed three times with buffer A. The proteins were eluted with buffer A containing 0.14 mg/ml of 1 × Flag peptide and 0.14 mg/ml of 3 × Flag peptide. The eluates were concentrated by centrifugation (Amicon Ultra MWCO 50 kDa). The purified proteins were stored at −80°C. All purification steps were performed at 0°C to 4°C.

Protein reconstitution into liposomes and ATPase assay

Liposome flotation assay

Preparation of cyclodextrin–lipid complex

For preparation of MβCD–sterol complexes, dried sterols were added to 30 mM MβCD in reaction buffer at a final concentration of 3 mM, and the mixture was incubated at 80°C for 1 to 6 days until the initially precipitated sterols were completely dissolved. To analyze the effects of sterols on ATPase activity, proteoliposomes were mixed with MβCD–sterol complex and the mixture was incubated at 25°C for 10 min. For preparation of MαCD–POPS complex, dried POPS was added to 180 mg/ml MαCD in phosphate-buffered saline at a final concentration of 6 mM, and the mixture was incubated at 37°C for 1 h.

Cellular lipid efflux assay

BHK/ABCA1 cells or BHK/ABCA1-MM cells were plated on 6-well plates at a density of 4 × 10⁵ cells per well in 10%FBS/DMEM. After incubation at 37°C for 24 h, the culture medium was replaced by DMEM containing 0.02% bovine serum albumin (BSA) and 10 nM mifepristone. After incubation for an additional 24 h, the culture medium was replaced by DMEM containing 0.02% BSA and MαCD–POPS complex, and the cells were incubated for an additional 30 min. The cells were washed two times with DMEM and incubated in DMEM containing 0.02% BSA. The culture medium was replaced by DMEM containing 0.02% BSA and 10 μg/ml apoA-I, and the cells were incubated for an additional 2 h. The culture medium was collected and centrifuged at 21,500 × g at 4°C for 15 min. The lipids in the supernatants were extracted with chloroform/methanol (2:1). The lipid solution was dried and resuspended in reaction mixture (HBSS containing 0.1% Triton X-100 and 5 mM sodium cholate). PC and cholesterol measurements were performed as described previously (34), with the following modification. For PC measurements, 40 μl of the lipid solution was mixed with 50 μl of HBSS containing 420 mU/ml choline oxidase, 100 mU/ml phospholipase D, 400 mU/ml HRP and 25 μM Amplex UltraRed, and the mixture was incubated at 37°C for 30 min. For cholesterol measurements, 40 μl of the lipid solution was mixed with 50 μl of HBSS containing 400 mU/ml cholesterol oxidase, 400 mU/ml HRP, and 25 μM Amplex UltraRed, and the mixture was incubated at 37°C for 30 min. Subsequently, the fluorescent intensity (excitation, 535 nm; emission, 590 nm) was measured.

Measurement of sterols

After incubation of liposomes with MβCD–sterol complex at 25°C for 10 min, liposomes were separated by a polypropylene column (Thermo Fisher Scientific) filled with Sephadex G-100 (Cytiva) in reaction buffer. The sterol content of liposomes was determined with the cholesterol quantification method as described in the previous section (cellular lipid efflux assay). A calibration curve for each sterol was obtained for quantification.

Fig. 1

Purification of human ABCA1 and the effect of cholesterol on the ATPase activity of ABCA1. (A) CBB staining of purified ABCA1. Purified ABCA1 and ABCA1-MM were separated by SDS-PAGE and stained by CBB. (B) ATP dependence of the ATPase reaction of ABCA1. The ATPase activity of DOPS liposome-reconstituted ABCA1 or ABCA1-MM was measured with 0.039, 0.078, 0.16, 0.31, 0.63, 1.25, 2.5, or 5 mM of ATP and 2 mM excess of MgCl₂ over all concentrations of ATP in the presence or absence of 0.3 mM MβCD–cholesterol complex. The ATPase reaction was performed at 37°C for 30 min. (C) Time dependence of the ATPase reaction. Purified ABCA1 was reconstituted into liposomes consisting of DOPS, and the ATPase reaction was analyzed with 3 mM ATP and 5 mM MgCl₂ for 0, 5, 10, 20, 30, 45, 60, or 120 min in the presence or absence of 0.3 mM MβCD–cholesterol complex. Linear fitting was carried out in the first 60 min. The experiments were performed in triplicate, and the plots are represented after subtraction of the mean values without protein. Error bars indicate S.D.

Open in new tabDownload slide

Statistical analysis

The statistical analysis of differences between mean values was performed using the unpaired t-test. Multiple comparisons were evaluated using Tukey’s test following one-way ANOVA. All experiments were performed at least twice.

Results

Purification of human ABCA1 from FreeStyle 293-F cells

Human ABCA1 and a non-functional ATPase-deficient mutant (ABCA1-MM), both fused with 10× histidine residues, triple FLAG tags and GFP at the C-terminus, were transiently expressed via polyethyleneimine-mediated transfection in FreeStyle 293-F cells (16,), which are derived from HEK293 cells and are adapted to suspension culture. In ABCA1-MM, two lysine residues crucial for ATP hydrolysis (K939 and K1952) are replaced by methionine (18,). Previous studies showed that GFP-fused ABCA1 and FLAG-fused ABCA1 expressed in cells exhibited significant cholesterol efflux activity (32,  37,  38,), suggesting that FLAG tag and GFP fusion at the C-terminus do not affect the function of ABCA1. The transfected cells were solubilized with 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) and then ABCA1 was purified with an antibody against FLAG tag. Because detergents significantly affect the activity of ABC proteins (39,  40), it was preferable to reduce the concentration of detergent in order to minimize its effects on ABCA1 activity. For this reason, resin-bound ABCA1 was flushed with lauryl maltose neopentyl glycol (LMNG), which has an extremely low critical micelle concentration, and eluted without detergent. Coomassie Brilliant Blue (CBB) staining showed major bands at > 250 kDa, which is consistent with the predicted molecular mass of ABCA1 fused with GFP and triple FLAG tags (288 kDa), and the purity of purified ABCA1 and ABCA1-MM was estimated at between 70% and 80% (Fig. 1A). Using this method, approximately 1 mg of ABCA1 was purified from 2.5 l of cultured cells.

Cholesterol increased the ATPase activity of ABCA1 in PS liposomes

Because PS is an abundant phospholipid in the inner leaflet of the plasma membrane and is known to affect the function of membrane proteins, including G protein–coupled receptors and ion channels (28,  29,), we used PS liposomes to analyze the effects of cholesterol on the ATPase activity of ABCA1. We first reconstituted purified ABCA1 into liposomes with or without cholesterol, but cholesterol affected the efficiency of protein reconstitution and comparative analysis could not be performed (data not shown). Thus, after the reconstitution of ABCA1 into liposomes consisting only of phospholipids, we loaded cholesterol into liposomes using a complex with methyl-β-cyclodextrin (MβCD). We found that the addition of the MβCD–cholesterol complex increased the ATPase activity of ABCA1 in 1,2-dioleoyl-phosphatidylserine (DOPS) liposomes (Fig. 1B and C). The ATPase activity of ABCA1 followed Michaelis–Menten kinetics for ATP regardless of cholesterol (Fig. 1B). Without cholesterol, the values of K_m for ATP and V_max were 0.37 ± 0.05 mM and 30 ± 1.2 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively. When cholesterol was loaded with MβCD, the values of K_m for ATP and V_max were 0.23 ± 0.02 mM and 48 ± 2.2 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively. The enzymatic parameters suggest that cholesterol accelerates the rate of ATP hydrolysis in cellular ATP concentration (1–5 mM) (41,). The V_max of ABCA1-MM was significantly lower than that of wild-type ABCA1 (without cholesterol, 11 ± 1 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹; with cholesterol, 5.6 ± 0.7 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹), indicating that the observed activities were those of the functional ABCA1. Time-dependent measurement showed that the amount of hydrolyzed ATP linearly increased during the first 60 min in both the presence and absence of cholesterol (Fig. 1C), indicating that the cholesterol-induced increase in ATPase activity did not reflect the stabilization of ABCA1. Therefore, the ATPase reactions described in the sections below were performed for 30 min in the presence of 3 mM MgATP. Since substrate transport by ABC proteins is coupled with ATP binding and hydrolysis, and it is commonly observed that the transport substrates exhibit increased ATPase activity (15–17), these results suggest that ABCA1 recognizes cholesterol as a transport substrate in PS liposomes.

Fig. 2

Effect of phospholipids on the cholesterol-induced increase in the ATPase activity of ABCA1. Purified ABCA1 or ABCA1-MM was reconstituted into DOPC liposomes containing 40 mol% of the indicated phospholipids. MβCD–cholesterol complex was added at a concentration of 0 or 0.3 mM after reconstitution. The mixture was mixed with ATP and MgCl₂ at a concentration of 3 and 5 mM, respectively, and incubated for 30 min at 37°C. The plots show measurements after subtraction of the mean values without protein. The experiment was performed in triplicate, and mean values ± S.D. are shown. ^*P < 0.05; ^**P < 0.001; N.S., P > 0.05.

Open in new tabDownload slide

Cholesterol increases the ATPase activity of ABCA1 in the presence of anionic lipids

To investigate if the cholesterol-induced increase in ATPase activity depends on phospholipids, we used liposomes consisting of 1,2-dioleoyl-phosphatidylcholine (DOPC), 1,2-dioleoyl-phosphatidylethanolamine (DOPE), egg sphingomyelin (egg SM), DOPS, 1,2-dioleoyl-phosphatidic acid (DOPA), or 1,2-dioleoyl-phosphatidylglycerol (DOPG). To clarify the effects of phospholipid head groups, the acyl chain composition was standardized, except in the SM liposomes. The ATPase activities of ABCA1 reconstituted in liposomes composed of DOPC alone or DOPC containing 40 mol% of DOPE (DOPC/DOPE) or egg SM (DOPC/egg SM) were low, with values of 2.0 ± 0.2 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, 1.2 ± 0.7 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹ and 3.1 ± 0.5 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively (Fig. 2). The addition of the MβCD–cholesterol complex slightly increased the ATPase activity in DOPC and DOPC/DOPE liposomes (3.0 ± 0.4 and 3.0 ± 0.8 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively) and showed no effect in DOPC/egg SM liposomes (3.5 ± 1.0 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹). In contrast, the ATPase activities of ABCA1 reconstituted in DOPC liposomes containing 40 mol% of DOPS (DOPC/DOPS), DOPA (DOPC/DOPA), or DOPG (DOPC/DOPG) were high, with values of 16 ± 1.0 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, 17 ± 0.6 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹ and 21 ± 1.0 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively. Importantly, cholesterol increased the ATPase activity by 2-fold to 3-fold in DOPC/DOPS, DOPC/DOPA and DOPC/DOPG liposomes (49 ± 0.3 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, 47 ± 0.8 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹ and 41 ± 1.3 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively). The ATPase activity of ABCA1-MM was less than 4 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹ in the absence of cholesterol and was hardly enhanced by cholesterol. A liposome flotation assay revealed that the efficiencies of ABCA1 reconstitution into DOPC and DOPC/DOPS liposomes were 71 ± 6.2% and 78 ± 3.2%, respectively (Supplementary Fig. S1), suggesting that phospholipids modulate ATPase activity without significantly affecting reconstitution efficiency. Considering that DOPS, DOPA and DOPG are all negatively charged and have different head structures, these results suggest that the negative charge of the head groups, but not their phospholipid structure, is important for the ATPase activity of ABCA1 and the strong increase by cholesterol.

Effects of PS level and the acyl chain structures of PS and PC on the ATPase activity of ABCA1

Since PS is the most abundant anionic lipid in mammalian cells (42), we speculated that PS plays a key role in the function of ABCA1 in cells. Therefore, we further investigated the effect of PS on the ATPase activity of ABCA1 in detail. As shown in Fig. 3A, the ATPase activity of reconstituted ABCA1 increased linearly from 0% to 100% DOPS. When analyzed with cholesterol-containing liposomes, the ATPase activity increased linearly up to 50% DOPS, at which point it plateaued, indicating that membrane PS concentration significantly affects the function of ABCA1.

Fig. 3

PS and acyl chain dependence of the ATPase activity of ABCA1. (A) PS dependence of the ATPase activity of ABCA1. Purified ABCA1 was reconstituted into DOPC liposomes containing various amounts of DOPS, and the ATPase reaction was analyzed in the presence or absence of 0.3 mM MβCD–cholesterol complex. (B) The effects of acyl chain structures of PS and a crude mixture of PC on increase in the ATPase activity. Purified ABCA1 was reconstituted into DOPC liposomes or egg PC liposomes containing 0 or 40% of DOPS or POPS. MβCD–cholesterol complex was added at a concentration of 0, 0.05, 0.1, 0.2, or 0.3 mM. (C) The effects of a crude mixture of PS on increase in the ATPase activity. Purified ABCA1 was reconstituted into DOPC liposomes containing 0 or 40% DOPS or brain PS. MβCD–cholesterol complex was added at a concentration of 0, 0.05, 0.1, 0.2, or 0.3 mM. The ATPase reactions were carried out at 37°C for 30 min in the presence of 3 mM ATP and 5 mM MgCl₂. The experiments were performed in triplicate, and solid lines indicate mean values. The plots show measurements after subtraction of the mean values without protein. Error bars indicate S.D.

Open in new tabDownload slide

The above experiments were carried out using 1,2-dioleoyl-phospholipids (DOPS and DOPC) because the low phase transition temperatures of these phospholipids facilitate liposome formation (43,). PS generally contains saturated fatty acids at the sn-1 position in cells, and the degree of fatty acid saturation of phospholipids is important for membrane fluidity and cholesterol packing (44,  45). To test whether the increase by cholesterol occurs in PS containing saturated fatty acids at the sn-1 position, we analyzed the ATPase activity of ABCA1 using 1-palmitoyl-2-oleoyl-phosphatidylserine (POPS) (Fig. 3B). Cholesterol increased the ATPase activity of ABCA1 reconstituted into DOPC liposomes containing 40% POPS as observed in 40% DOPS in a dose-dependent manner, suggesting that saturated fatty acids at the sn-1 position of PS do not affect the cholesterol-induced increase. The ATPase activity was also enhanced by cholesterol when ABCA1 was reconstituted into liposomes consisting of DOPS and egg PC, the latter of which is a crude PC mixture with a physiological fatty acid distribution (32.7% of 16:0, 12.3% of 18:0, 32.0% of 18:1 and 17.1% of 18:2). Without PS, cholesterol did not increase the ATPase activity of ABCA1 reconstituted into DOPC or egg PC liposomes. Furthermore, cholesterol increased the ATPase activity of ABCA1 reconstituted into DOPC liposomes containing 40% brain PS, a crude PS mixture from porcine, as observed in 40% DOPS (Fig. 3C). These results suggest that the fatty acyl chains of PS and PC have little effect on the cholesterol-induced increase in ATPase activity, and that this increase also occurs in membranes with a physiological composition.

Effect of PS addition to BHK/ABCA1 cells on PC and cholesterol efflux

To clarify whether the characteristic of the ATPase activity shown in Fig. 3A is relevant to nascent HDL generation by ABCA1, we next analyzed the effects of PS loading on the lipid efflux from ABCA1-expressed cells. For the lipid efflux assay, we used baby hamster kidney (BHK)/ABCA1 cells, which can be induced to express ABCA1 with mifepristone (33,). Because it has been reported that methyl-α-cyclodextrin (MαCD) can load phospholipids into cells (46,  47), we added POPS to BHK/ABCA1 cells or BHK/ABCA1-MM cells as a complex with MαCD. As shown in Fig. 4, the amounts of PC and cholesterol efflux from BHK/ABCA1 cells increased 1.3-fold and 1.6-fold, respectively, by the addition of 0.3 mM MαCD–POPS complex. In contrast, the amounts of PC and cholesterol efflux from BHK/ABCA1-MM cells were low, and MαCD–POPS complex did not affect the lipid efflux. These results suggest that PS- and cholesterol-dependent ATPase activity of ABCA1 correlates with the lipid efflux to apoA-I.

Fig. 4

Cellular PC and cholesterol efflux after POPS loading. BHK/ABCA1 cells and BHK/ABCA1-MM cells were incubated with 0 or 0.3 mM MαCD–POPS complex and 0.02% BSA at 37°C for 30 min. The cells were incubated with 0.02% BSA at 37°C for an additional 30 min, and further incubated with 10 μg/mL apoA-I and 0.02% BSA at 37°C for an additional 2 h. After the incubation, amounts of PC and cholesterol in the medium were measured. The experiments were performed in triplicate, and solid lines indicate mean values. Error bars indicate S.D. ^*P < 0.05; N.S., P > 0.05.

Open in new tabDownload slide

Effect of the sterol structure on the increase in the ATPase activity of ABCA1

Whether or not ABCA1 transports non-cholesterol sterols remains controversial. Previous studies reported that ABCA1 exports non-cholesterol sterols, including β-sitosterol, to apoA-I (23,  48,) while another study showed that apoA-I–dependent β-sitosterol efflux was not observed in Caco-2 cells (25). Since cholesterol increased the ATPase activity of ABCA1 in PS-containing liposomes, it was presumed that ABCA1 sterol specificity can be analyzed in PS liposomes. To analyze the effect of sterol structure on the increase in ATPase activity, we added cholesterol, intermediates of mammalian cholesterol biosynthesis (desmosterol, zymosterol, 7-dehydrocholesterol, or lathosterol; Fig. 5A), or plant sterols (stigmasterol or brassicasterol) into DOPS liposomes using a complex with MβCD. The amounts of sterols delivered into liposomes were quantified after liposomes were separated from the MβCD–sterol complex by size-exclusion chromatography. As shown in Fig. 5B, cholesterol, desmosterol, zymosterol, 7-dehydrocholesterol, lathosterol, stigmasterol and brassicasterol were inserted into liposomes at almost comparable levels. We also examined other plant sterols (campesterol and β-sitosterol; Supplementary Fig. S2A), a fungal sterol (ergosterol), and another cholesterol precursor (lanosterol), but the amounts of these sterols that were inserted into liposomes were half that of cholesterol (Supplementary Fig. S2B), and the amount of inserted lanosterol could not be quantified due to the specificity of the cholesterol oxidase. Therefore, we compared the effects of the sterols shown in Fig. 5A. The ATPase activity of ABCA1 was enhanced 2-fold by desmosterol and zymosterol, as well as by cholesterol, whereas 7-dehydrocholesterol and lathosterol increased the ATPase activity only moderately, by 1.2-fold (Fig. 5C). In contrast, the other sterols, including stigmasterol and brassicasterol, showed no effect. Next, we analyzed the dose-dependency of these sterols. Cholesterol increased the ATPase activity in a dose-dependent manner, and the ATPase activity was slightly suppressed at a MβCD–cholesterol complex concentration of 0.4 mM (Fig. 5D). Desmosterol increased the ATPase activity to the same extent as cholesterol. When 7-dehydrocholesterol was added to liposomes, a maximal but slight increase in ATPase activity was observed at 0.05 mM (Supplementary Fig. S3). Stigmasterol showed little effect on the ATPase activity at all tested concentrations, and a slight decrease was observed for brassicasterol. Since brassicasterol and stigmasterol differ from cholesterol in their alkyl side chains (Fig. 5A), and 7-dehydrocholesterol and lathosterol, which showed weak induction of ATPase activity, differ from cholesterol in their sterol rings, both the side chain and the ring structure of cholesterol are important for the increase in the ATPase activity of ABCA1.

Fig. 5

Effect of sterols on the ATPase activity of ABCA1. (A) Chemical structures of sterols. Sterols have four core rings (A, B, C and D). The different moieties from cholesterol are shown within the red circles. (B) The amounts of liposome-incorporated sterols. DOPS liposomes were incubated at a concentration of 0.2 mg/ml with 0.3 mM MβCD–sterol complex, and then separated by size-exclusion chromatography. The amounts of sterols in liposomes were quantified with cholesterol oxidase. (C) Sterol-induced ATPase activity of ABCA1. Purified ABCA1 was reconstituted into liposomes consisting only of DOPS, and 0 or 0.3 mM MβCD–sterol complex was added. Red and cyan bars indicate 2-fold and 1.2-fold increase in ABCA1 ATPase activity, respectively. Green bars indicate non-significant to None (without sterol). The plots show measurements after subtraction of the mean values without protein. (D) Sterol-concentration dependence of the ATPase activity. Purified ABCA1 was reconstituted into DOPS liposomes, and MβCD–sterol complex was added at a concentration of 0, 0.05, 0.1, 0.2, 0.3, or 0.4 mM. The ATPase reactions were carried out at 37°C for 30 min in the presence of 3 mM ATP and 5 mM MgCl₂. The plots show measurements after subtraction of the mean values without protein. The experiments were performed in triplicate, and solid lines indicate mean values. Error bars indicate S.D. ^**P < 0.001 versus none; N.S., P > 0.05.

Open in new tabDownload slide

Discussion

ABCA1 mediates HDL generation to prevent excess cholesterol accumulation. However, it remains unclear whether ABCA1 selectively transports cholesterol. In this study, we found that cholesterol increased the ATPase activity of ABCA1 in the presence of anionic lipids, including PS, and that cholesterol was a more preferred substrate for ABCA1 compared to plant sterols.

The present study is the first to demonstrate a cholesterol-induced increase in the ATPase activity of ABCA1. Most of the transport substrates of ABC proteins increased these proteins’ ATPase activities. Indeed, the ATPase activity of ABCG1, which contributes to HDL maturation by exporting cholesterol, and that of sterol transporter ABCG5/G8, are enhanced by cholesterol (16,  49,). Thus, our findings suggest that ABCA1 recognizes cholesterol as a transport substrate. Currently, two models of nascent HDL formation have been proposed: (i) ABCA1 directly generates nascent HDL by loading cholesterol onto apoA-I (34,); and (ii) ABCA1 translocates phospholipids to create a specialized lipid domain on the plasma membrane, and apoA-I spontaneously solubilizes lipids from this domain (50). The observations in this study support the first model.

We previously reported that cholesterol suppressed the ATPase activity of ABCA1 purified from Sf9 cells, even in the presence of anionic lipids (18,). Quazi et al. also reported similar findings in an analysis of ABCA1 purified from HEK293 cells (19,). In these previous studies, ABCA1 was reconstituted into liposomes containing cholesterol. However, in the present approach, we used MβCD to add cholesterol into liposomes after protein reconstitution because cholesterol can decrease the efficiency of reconstituting some membrane proteins into liposomes (51). Our present method allowed for accurate analysis of the effects of cholesterol on the ATPase activity of ABCA1 by eliminating the influence of cholesterol on reconstitution.

The effects of different sterols on the ATPase activity of ABCA1 varied. Desmosterol, which differs from cholesterol in that it contains a double bond at C-24 of its side chain (Fig. 5A), increased the ATPase activity of ABCA1 to the same extent as cholesterol (Fig. 5D). In contrast, brassicasterol, which contains a methyl group at C24 and a double bond at C22 in the alkyl side chain (Fig. 5A), did not increase the ATPase activity, suggesting that the sterol binding site of ABCA1 is not large enough to accommodate the bulky side chains of plant sterols. Similarly, differences in sterol rings also affected increase in the ATPase activity. Zymosterol, which contains a double bond at C-8 in the sterol B ring, increased the ATPase activity to the same extent as cholesterol. In contrast, 7-dehydrocholesterol, which contains two double bonds at C-5 and C-7, and lathosterol, which contains a double bound at C-7, caused less than half the ATPase activity of cholesterol, suggesting that the spatial structures of sterol B rings are also important for the recognition of sterols. Cholesterol can interact with the aromatic amino acid residues of membrane proteins (52,). In the case of the human delta-type opioid receptor, the tyrosine residue can stack onto the B ring of cholesterol through a CH-π stacking interaction (53). The weak increase in the ATPase activity caused by 7-dehydrocholesterol, which has only one CH bond at C-7 due to the addition of a double bond at C-7, may suggest a CH-π interaction between ABCA1 and cholesterol B ring. Taken together, our findings suggest that ABCA1 specifically recognizes cholesterol through its alkyl side chain and ring structures.

Dietary cholesterol is absorbed from the intestinal tract and transported to the liver as lipoproteins. ABCA1 is also involved in cholesterol efflux (54,). Most plant sterols in enterocytes are returned into the gut lumen by ABCG5/G8 expressed on the apical membrane, but some enter the liver (55,). Several reports suggest that ABCA1 transports dietary plant sterols from the intestine to the liver (23,  24,). However, the previous study by Tachibana et al. did not identify β-sitosterol efflux by ABCA1 expressed in Caco-2 (25), and it remains unclear whether ABCA1 can transport plant sterols as efficiently as cholesterol. We found that stigmasterol and brassicasterol did not increase the ATPase activity of ABCA1, suggesting that ABCA1 prefers cholesterol over plant sterols as a transport substrate and that ABCA1 possesses a substrate binding site that is rather specific to cholesterol. Accordingly, it is conceivable that intestinal ABCA1 contributes to preventing the accumulation of dietary plant sterols in the body, at least in part.

Another important finding of this study is that anionic lipids are required for the ATPase activity of ABCA1. The ATPase activity of ABCA1 was approximately 20 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹ in the presence of 40% of PS, phosphatidic acid (PA), or phosphatidylglycerol (PG), and was increased by 2-fold to 3-fold by cholesterol (Fig. 2). In the absence of anionic lipids, however, ABCA1 exhibited neither significant ATPase activity nor an increase in activity by cholesterol. Given that the structures of the hydrophilic headgroups of PS, PA and PG are quite different, it is conceivable that their negative charges are important for the ATPase activity of ABCA1. The PA and PG comprise less than 2% of human cell membranes (42,). Additionally, both POPS, a physiological PS species, and a crude PS mixture from porcine brain exhibited a similar increase in ATPase activity as DOPS (Fig. 3B and C). Furthermore, the addition of POPS to cells expressing ABCA1 increased PC and cholesterol efflux (Fig. 4). Therefore, it is suggested that PS plays a pivotal role in HDL generation in cells. The subcellular distribution of PS differs between organelles. PS is highly enriched in the plasma membrane (~10–15%) (56,), where ABCA1 is localized to generate nascent HDL (57,), whereas PS levels are low in the endoplasmic reticulum (ER; ~ 4%) and mitochondria (~1%). In the plasma membrane, PS is exclusively concentrated in the inner leaflet (58,), where it is estimated to comprise ~ 20–30% of the membrane. In addition, it has been reported that PS can form a clustered domain in the inner leaflet of the plasma membrane (59). We found that in the presence of cholesterol, the ATPase activity of ABCA1 increased linearly by ~ 13-fold with PS concentrations between 0% to 30% (Fig. 3A), suggesting that ABCA1 functions efficiently in the plasma membrane whereas its activity is suppressed in PS-poor environments such as the ER. Therefore, PS may regulate ABCA1 function to prevent the disruption of lipid homeostasis in intracellular organelles.

How does PS regulate the ATPase activity of ABCA1? A simple explanation is that PS is a transport substrate of ABCA1. However, given that the amounts of PS incorporated into nascent HDL are considerably lower than those of PC or cholesterol (12,  14,), it is likely that PS is not transported to apoA-I. Therefore, it is conceivable that PS is a modulator of ABCA1 rather than a transport substrate. We found that PS addition to BHK/ABCA1 cells enhanced PC and cholesterol efflux (Fig. 4). In the living cells, PS is localized to the inner leaflet of the plasma membrane by phospholipid flippases (60). Consequently, the increase in ABCA1-mediated lipid efflux by PS addition suggests that PS in the inner leaflet side is important for the function of ABCA1. PS may trigger a conformational change of ABCA1 to promote ATP hydrolysis, but further studies are required to clarify the mode of activation by PS.

We purified ABCA1 from FreeStyle 293-F cells, which are suitable for the large-scale expression of human ABC proteins (16,  61,). ABCA1 was purified in the absence of detergents after flushing with LMNG on anti-FLAG beads. It can be inferred that a minimal number of detergent molecules are bound to the surface of ABCA1 molecules purified by this procedure, since LMNG exhibits an extremely low off-rate compared to classically used detergents (62,) and some membrane proteins are active even at LMNG concentrations much lower than its critical micelle concentration (63,). Purified ABCA1 showed significant ATPase activity of approximately 50 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹ in PS liposomes containing cholesterol (Fig. 2). Although the activity was lower than that reported previously (200–1000 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹) (18,  19,  64,), possibly due to the differences in the expression hosts (18,), or purification system (19,  64,), it was similar to that of ABCA4 and ABCG1 (90 and 100 nmol|$\cdot$|min⁻¹|$\cdot$|mg⁻¹, respectively) (16,  65,). We have previously shown that 135 ng of cellular ABCA1 exported about 0.15 nmol of PC and 0.21 nmol of cholesterol in 60 min (34,). That is, 1 mg of ABCA1 can transport about 19 nmol of PC and 26 nmol of cholesterol per minute. The stoichiometry of ABCA1-mediated ATP hydrolysis and lipid transport has not been determined. However, the lipid transport by ABCA1 (34,) is highly consistent with the ATPase activity in this study if ABCA1 hydrolyzes one to two ATP to transport a single molecule as reported for the drug transporter (66).

In conclusion, we showed that the ATPase activity of ABCA1 is increased by cholesterol in the presence of anionic lipids. We propose that ABCA1 selectively transports cholesterol and requires PS for cholesterol transport. In addition, this finding provides valuable insights into the mechanism of HDL generation and may have implications for the development of therapies targeting ABCA1 to treat disorders related to lipid metabolism.

Funding

This work was supported by AMED-PRIME Grant Number 20gm5910022h0004 (Y. K.); JSPS KAKENHI Grant Numbers 21H04713 (N. K.), 18H05269 (K. U.), 23H00326 (K. U.) and 22H02258 (Y. K.) and JSPS Research Fellow Grant Number 22 J14975 (K.S.)

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Data availability

All data are available within the article and its supplementary data files.

Supplementary data

Supplementary Data are available at JB Online.

Author contributions

K. S. and Y. K. planned the experiments. K. S. conducted the experiments. K.S., N. K., K. U. and Y. K. analyzed data. K. S. and Y. K. wrote the manuscript. All authors revised the manuscript and approved the final manuscript.

Acknowledgment

We thank Dr. Fumihiko Ogasawara of Kyoto University for helpful discussion.

References