Proline residues in scavenger receptor-BI's C-terminal region support efficient cholesterol transport

Approximately one in three deaths (30.8%) in the U.S.A. are cardiovascular disease-related [1]. Most cardiovascular disease events result from atherosclerosis where cholesterol accumulates in lesions along the artery wall, a process that can progressively occlude vessels and decrease blood flow (reviewed in ref. [2]). High-density lipoproteins (HDLs) protect against atherosclerosis by removing excess cholesterol from circulation via reverse cholesterol transport [3]. During reverse cholesterol transport, HDL accepts cholesterol from cells in the periphery (e.g. arterial macrophages) [4] via cholesterol efflux and transports it to the liver as cholesteryl esters (CEs) via non-endocytic selective uptake of HDL-CE [5]. Once internalized by the liver, CEs are eventually converted to bile and excreted from the body.

Bidirectional cholesterol transport between cells and HDL occurs by a receptor-mediated process involving HDL's high-affinity receptor, scavenger receptor-BI (SR-BI) [4,6]. SR-BI is an 82 kDa cell surface glycoprotein consisting of a large extracellular domain anchored by two transmembrane domains [7]. Since SR-BI is responsible for cholesterol transport, it is important to understand its role in the context of atherosclerosis. SR-BI's role in protecting against atherosclerosis is supported by mouse models in which overexpression of SR-BI in the liver reduces atherosclerosis [8], while whole-body deficiency of SR-BI accelerates atherosclerosis and raises HDL-cholesterol [9]. In humans, rare variants of SR-BI (S112F, T175A, and P297S) were identified in patients with high HDL-cholesterol [10,11], and these mutations significantly impaired in vitro SR-BI-mediated cholesterol transport [11,12]. Another rare variant of SR-BI (P376L) is associated with an increased cardiovascular disease risk in humans [13] and mimics phenotypes observed for mice deficient in SR-BI [9]. Altogether, these studies suggest a critical role for SR-BI in reverse cholesterol transport and protection against atherosclerosis; however, the precise mechanisms by which SR-BI facilitates cholesterol transport remain undefined.

The use of structure–function studies has expanded our knowledge of SR-BI cholesterol transport functions [14–18]. Homology models of SR-BI's extracellular domain provided physical evidence of a hydrophobic cavity that can spatially accommodate cholesterol molecules [19]. This evidence supports one hypothesis for SR-BI-mediated cholesterol transport, in which SR-BI transports HDL-CE into cells via a hydrophobic channel that facilitates the movement of CE down its concentration gradient [20]. SR-BI oligomerization may also play a role in cholesterol transport [21,22]. Gaidukov et al. [23] demonstrated that a glycine dimerization motif within the N-terminal transmembrane domain of SR-BI is required for lipid uptake and SR-BI homo-dimerization. We observed homo-dimerization of SR-BI's C-terminal region in live cells using fluorescence resonance energy transfer experiments [24]. Dimerization of the C-terminal region is likely mediated by multiple interactions and may include a potential glycine dimerization motif as well as a leucine zipper dimerization motif that partially spans the C-terminal transmembrane domain [18]. Mutation of all leucine residues within the leucine zipper motif to alanine decreased receptor oligomerization and SR-BI-mediated cholesterol transport [18].

Seeking to gain additional insights into how the C-terminal region of SR-BI contributes to its function, we solved the high-resolution NMR structure of an SR-BI peptide encompassing residues 405–475 [SR-BI(405–475)], which spans the C-terminal transmembrane domain and proximal residues in the extracellular domain [18]. The amino acid sequence and NMR structure of SR-BI(405–475) revealed four conserved proline (Pro/P) residues, including one proline (Pro-459) that induces a kink in the transmembrane α-helix. Proline residues are so-called helix-breakers that generate kinks in α-helices due to the lack of a hydrogen bond donor and steric constraints [25]. Nonetheless, proline residues have been observed in the transmembrane domains of integral membrane proteins [26], albeit less frequently than other amino acids. The role of Pro-459 and the importance of the Pro-459-induced kink in SR-BI function have not yet been described.

In this study, we examined the role of conserved proline residues in SR-BI(405–475) with the goal of better understanding the structural features that contribute to SR-BI-mediated cholesterol transport. We hypothesized that the conserved proline residues support SR-BI in a conformation that allows for efficient cholesterol transport. To test this, we generated four different Pro-to-alanine (Ala/A) mutations in full-length SR-BI receptors (P408A-, P412A-, P438A-, and P459A-SR-BI) and transiently expressed the mutant receptors in COS-7 cells to determine how the mutations affected cholesterol transport. Our findings indicate that SR-BI-mediated cholesterol transport requires key proline residues in the C-terminal region of SR-BI.

COS-7 cells were purchased from American Type Culture Collection (ATCC, Manassas, VA). Antibodies targeting SR-BI's C-terminal region (residues 450–509) or extracellular domain (residues 230–380) and anti-microtubule-associated protein 1A/1B-light chain 3B (LC3B) were obtained from Novus Biologicals, Inc. (Littleton, CO). Anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody was from Cell Signaling Technology (Danvers, MA). Anti-ubiquitin antibody was from Thermo Fisher Scientific (Waltham, MA). Peroxidase-conjugated anti-rabbit IgG and anti-mouse IgG were obtained from Amersham–GE Healthcare (Chicago, IL). Fluorescein isothiocyanate (FITC)-conjugated anti-rabbit IgG was purchased from BD Biosciences (Franklin Lakes, NJ). Human HDL was purchased from Alfa Aesar (Tewksbury, MA). [³H]-cholesterol and [¹²⁵I]-sodium iodide were from PerkinElmer (Waltham, MA). [³H]-cholesteryl oleyl ether (COE) was from American Radiolabeled Chemicals, Inc. (St. Louis, MO). Chloroquine diphosphate salt, carbobenzoxy-Leu-Leu-leucinal (MG-132), cholesterol oxidase (Streptomyces), and thin-layer chromatography standards (cholesterol, 4-cholesten-3-one, and cholesteryl oleate) were obtained from Sigma–Aldrich (St. Louis, MO). All other reagents were of analytical grade.

Four Pro-to-Ala mutations were generated at positions 408, 412, 438, or 459 in the coding region of murine SR-BI, which was previously cloned into a pSG5 expression vector [5] (Stratagene, Inc.). Top Gene Technologies (Pointe-Claire, Quebec, Canada) performed cloning, site-directed mutagenesis, and sequencing.

COS-7 cells were maintained in high-glucose DMEM (Dulbecco's modified Eagle's medium, Gibco) containing 10% (v/v) newborn calf serum, 2 mM l-glutamine, 50 units/ml penicillin, 50 µg/ml streptomycin, and 1 mM sodium pyruvate under standard conditions (37°C/5% CO₂). Upon reaching 60–70% confluency, cells were transiently transfected with cDNA encoding empty pSG5 vector, wild-type (WT), or SR-BI mutant receptors using FuGENE 6 (Promega) at a ratio of 3 : 1 (FuGENE 6:plasmid DNA), as previously described [5]. All experiments were performed 48 h post-transfection unless stated otherwise.

RNA was isolated using TRIzol (Invitrogen), adhering to the manufacturer's instructions. RT-PCR was performed to assess mRNA expression of SR-BI using the following murine-specific primers (Integrated DNA Technologies): 5′-CCGTCCCTTTCTACTTGTCTG-3′ (forward) and 5′-ACCTTTTGTCTGAACTCCCTG-3′ (reverse). Samples were separated on 2% agarose gels containing ethidium bromide and visualized on an AlphaImager (Alpha Innotech).

Total cellular proteins were harvested by incubation in radioimmunoprecipitation assay (RIPA) buffer (1% NP-40, 10 mM Tris [pH 8.0], 150 mM NaCl, 0.5% sodium deoxycholate, and 0.1% SDS) containing protease inhibitors (1 µg/ml pepstatin/leupeptin/aprotinin and 20 µg/ml phenylmethylsulfonyl fluoride), and the resulting lysates were cleared by centrifugation at 6010×g. Protein concentrations were determined by the Lowry method [27].

Cellular lysates (10–20 µg protein) were separated by 8%, 10%, or 12% SDS–PAGE and transferred to nitrocellulose or polyvinylidene difluoride membranes which were then blocked in 5% nonfat dry milk (w/v) in 1× Tris-buffered saline with 0.05% Tween-20 (TBST). Antibody incubations were performed in 1% milk/TBST at the following dilutions: C-terminal region of SR-BI (1 : 5000), GAPDH (1 : 5000), ubiquitin (1 : 1000), LC3B (1 : 1000), peroxidase-conjugated donkey anti-rabbit IgG (1 : 10 000), and peroxidase-conjugated sheep anti-mouse IgG (1 : 10 000). Bands were detected upon exposure to SuperSignal West Pico or SuperSignal West Pico PLUS (Thermo Fisher Scientific) chemiluminescent substrate.

Cells expressing empty vector, WT, or mutant SR-BI receptors were trypsinized, resuspended in DMEM, and centrifuged at 300×g. Pellets were then washed in phosphate-buffered saline (PBS) containing 0.5% bovine serum albumin (BSA) and incubated with anti-extracellular SR-BI antibody (1 : 100) for 1 h on ice. Cells were washed and incubated with FITC-conjugated anti-rabbit IgG (1 : 200) for 20 min on ice. After a final wash, cell pellets were resuspended in PBS/0.5% BSA, and 10 000 events were recorded on an Accuri C6 flow cytometer (BD Biosciences; Blood Research Institute, Medical College of Wisconsin) equipped with a 488 nm laser. Analysis was performed using CFlow Plus software (BD Biosciences). Live cell populations (gated from the forward vs. side scatter plot) were further gated into FITC-positive populations by arbitrarily setting empty vector-transfected cells such that ∼3% of cells were positive for FITC. Mean fluorescence intensity (MFI) of FITC-positive cells was obtained for each sample, and MFI of the secondary antibody-only control was subtracted from each sample to account for background staining. MFI for each sample was set relative to average WT-SR-BI levels from all transfections.

For proteasome inhibition experiments, cells expressing empty vector, WT-, or P408A-SR-BI were incubated with culture media containing 0, 3, 6, or 15 µM MG-132 for 24 h at 37°C. For lysosomal degradation studies, cells expressing empty vector, WT-, or P408A-SR-BI were treated with chloroquine (0, 10, 30, or 100 µM) for 16 h at 37°C. Following incubation with the inhibitors, cellular proteins were harvested in RIPA buffer and subjected to SDS–PAGE and immunoblot analysis for detection of SR-BI protein or control proteins (GAPDH, β-actin, ubiquitin, and LC3B), according to the protocols described above.

Human HDL was double-radiolabeled with [³H]-COE and [¹²⁵I] using recombinant CE transfer protein (Roar Biomedical) and dilactitol-tyramine, respectively, according to established methods [28]. Cellular association of [¹²⁵I]-HDL and uptake of [³H]-COE were assayed as previously described [28]. Radioactivity measurements were normalized to corresponding sample protein concentrations, as determined by the Lowry method, as well as specific activity of each radioisotope [specific activities of HDL preparations (mean ± SEM, standard error of the mean): [¹²⁵I] = 221.6 ± 21.2 DPM/ng protein, [³H]-COE = 619.6 ± 7.2 DPM/ng protein].

Immediately after seeding onto 12-well plates, cells transiently expressing empty vector, WT-, or mutant SR-BI receptors were pre-radiolabeled in DMEM containing 2 µCi/ml [³H]-cholesterol and 1 µg/ml acyl-CoA:cholesterol acyltransferase (ACAT) inhibitor (Sigma–Aldrich) for 24 h at 37°C. Cholesterol oxidase sensitivity assays were performed by washing excess [³H]-cholesterol with PBS and incubating cells in DMEM/0.5% BSA containing 0.5 U/ml cholesterol oxidase and 1 µg/ml ACAT inhibitor for 4 h at 37°C. Cellular lipids were extracted with isopropanol, mixed with standards (cholesterol, cholestenone, and cholesterol oleate), separated on thin-layer chromatography paper (Agilent Technologies) using a solvent mixture of 180 hexanes : 20 ethyl ether : 2 methanol (v/v/v), and lipids were visualized by exposure to iodine. Radioactivity corresponding to each lipid species was measured on a Tri-Carb 2100TR liquid scintillation counter (PerkinElmer).

Cells transiently expressing empty vector, WT-, or mutant SR-BI receptors were pre-radiolabeled with [³H]-cholesterol, as described above. Media were replaced with DMEM/0.5% BSA 48 h post-transfection, and cholesterol efflux assays were performed 72 h post-transfection. Cells were incubated with 50 µg/ml human HDL and 1 µg/ml ACAT inhibitor in DMEM/0.5% BSA for 4 h at 37°C. Media were collected, centrifuged at 1500×g, and supernatants were subjected to scintillation counting. After washing the cells with PBS, the remaining cellular lipids were extracted using isopropanol, and radioactivity was measured by scintillation counting.

Cells expressing empty vector, WT-, or mutant SR-BI were harvested in PBS containing protease inhibitors (1 µg/ml pepstatin/leupeptin/aprotinin and 20 µg/ml phenylmethylsulfonyl fluoride) with gentle scraping and sonicated briefly on ice for four cycles. Lysates (10 µg protein) were incubated with perfluorooctanoic acid (PFO) sample buffer (5% PFO, 100 mM Tris base, 20% glycerol, and 0.005% bromophenol blue) for 30 min at room temperature. Samples were loaded on 8% polyacrylamide gels lacking SDS in PFO running buffer (25 mM Tris base, 192 mM glycine, and 0.5% PFO), separated at constant 20 mA, transferred to nitrocellulose membranes, and immunoblotting was performed as described above.

P-values with respect to WT-SR-BI activity or expression were calculated using one-way analysis of variance (ANOVA) with Dunnett's or Sidak's multiple comparisons tests, where ***P < 0.001, *P < 0.05, and no significance (ns) P ≥ 0.05.

To study the importance of conserved proline residues in SR-BI's C-terminal transmembrane domain and proximal extracellular region, we began by generating multiple-species sequence alignments of SR-BI(405–475) using the Multiple Sequence Comparison by Log-Expectation (MUSCLE) algorithm [29] (Figure 1A). High amino acid conservation was observed for four proline residues (Pro-408, Pro-412, Pro-438, and Pro-459). Of those, two proline residues (Pro-412 and Pro-438) were fully conserved amongst all species, while Pro-408 and Pro-459 were highly conserved except in lower vertebrates such as salmon and zebrafish. The four conserved proline residues were mapped onto the NMR structure of SR-BI(405–475) (Figure 1B). In the NMR structure, Pro-408, Pro-412, and Pro-438 are located in the extracellular region, while Pro-459 generates a kink in the C-terminal transmembrane domain of SR-BI. For comparison, we mapped Pro-408, Pro-412, and Pro-438 onto a homology model of SR-BI's extracellular domain that was generated with MODELLER software [30] using the X-ray crystal structure of the luminal domain of lysosomal integral membrane protein 2 (LIMP-2) [19] as a template (Figure 1C). In the homology model, Pro-408 and Pro-438 are in flexible linker regions, while Pro-412 is within a β-strand.

To test the functional importance of conserved proline residues in SR-BI(405–475), we utilized COS-7 cells as an in vitro overexpression system. COS-7 cells express a low level of endogenous SR-BI and are amenable to transient transfection of plasmids encoding full-length SR-BI. We used various methods to confirm the successful transfection and expression of WT and Pro-to-Ala mutant SR-BI receptors. Robust SR-BI gene expression was verified by RT-PCR for all SR-BI-transfected conditions, while trace amounts of PCR product could be detected for empty vector-transfected cells, likely due to low endogenous SR-BI expression (Figure 2A). SR-BI protein expression in whole-cell lysates was determined by SDS–PAGE and immunoblotting with an antibody directed against the C-terminal region of SR-BI (Figure 2B). Expression of most proline mutants was comparable to WT-SR-BI with the exception of P408A-SR-BI, which showed markedly reduced expression of SR-BI and was only detectable at saturating film exposures (data not shown). The same decrease in P408A-SR-BI expression was observed when using an antibody that targets the extracellular domain of SR-BI, and P408A-SR-BI was not detected in the pellet after centrifugation of the cell lysate (data not shown). As cell surface localization of SR-BI is essential for its interaction with HDL and cholesterol transport functions, we used flow cytometry to analyze the cells that were stained for cell surface SR-BI expression (Figure 2C). We confirmed that all mutant SR-BI receptors (excluding P408A-SR-BI) are present at the plasma membrane of COS-7 cells at levels comparable to WT-SR-BI or greater (in the case of P412A-SR-BI) (Figure 2C). Consistent with total protein analysis, P408A-SR-BI cell surface expression was noticeably lower but did not reach statistical significance (Figure 2C).

Proline residues are involved in the rate-limiting steps of protein folding [31]. We hypothesized that P408A-SR-BI expression is decreased due to a protein folding defect, resulting in a misfolded protein that would be degraded by either the ubiquitin/proteasome system or the lysosome. If WT- or P408A-SR-BI is degraded by these pathways, we would expect SR-BI protein to accumulate in the presence of proteasome or lysosome inhibitors. To assess the proteasomal degradation of SR-BI, 24 h post-transfection with empty vector, WT-, and P408A-SR-BI, cells were treated with 0–15 µM MG-132 for 24 h and analyzed by immunoblotting. Increased total protein ubiquitination was observed in the presence of MG-132, confirming that proteasomal inhibition was successful (Figure 3A). However, we observed no increase in WT- or P408A-SR-BI in the presence of MG-132 (Figure 3B), suggesting that P408A-SR-BI is not degraded by the proteasome. To assess lysosomal degradation of SR-BI, transfected cells were treated with 0–100 µM chloroquine for 16 h (Figure 3C). Chloroquine inhibition was confirmed by an increase in LC3B-II expression (Figure 3C). Neither WT- nor P408A-SR-BI expression was increased by treatment with chloroquine (Figure 3D). Since cell surface protein expression of P408A-SR-BI appears to be defective, we chose to exclude this mutant from the functional analyses described below.

HDL binds to SR-BI expressed on liver cells, and SR-BI facilitates the transport of HDL-CE into cells for biliary excretion, effectively clearing cholesterol via reverse cholesterol transport. To determine the ability of the proline mutant SR-BI receptors to bind HDL and mediate CE uptake, cells expressing empty vector, WT-, or mutant SR-BI receptors were incubated with [¹²⁵I]/[³H]-COE-HDL for 1.5 h, as described in Materials and Methods. HDL cell association was decreased to levels of 4.6%, 43%, and 69% relative to WT-SR-BI (100%) in the presence of P412A-, P438A-, and P459A-SR-BI, respectively (Figure 4A). As expected due to the loss of HDL cell association, all three proline mutants also showed decreased [³H]-COE uptake (4.4%, 47%, and 40%, for P412A-, P438A-, and P459A-SR-BI, respectively, vs. WT-SR-BI) (Figure 4B).

SR-BI functions independently of HDL to redistribute membrane cholesterol [32]. To assess the ability of SR-BI to modulate membrane cholesterol distribution, cells transfected with empty vector, WT-, or mutant SR-BI were pre-labeled with [³H]-cholesterol and incubated with exogenous cholesterol oxidase for 4 h, and the resulting lipid species were separated by thin-layer chromatography. Cholesterol oxidase catalyzes the conversion of free cholesterol to cholestenone (Figure 5A). Kellner-Weibel et al. [32] have demonstrated that WT-SR-BI expression increases the production of cholestenone, presumably by modulating membrane cholesterol distribution and increasing accessibility to cholesterol oxidase. Following incubation with cholesterol oxidase, cells expressing P412A-SR-BI were unable to produce cholestenone, while P438A-SR-BI showed ∼76% less [³H]-cholestenone production than cells expressing WT-SR-BI, suggesting that these mutant receptors have impaired ability to redistribute membrane cholesterol (Figure 5B). Interestingly, mutation of Pro-459 did not result in loss of function (Figure 5B).

The initiating steps of reverse cholesterol transport include SR-BI-mediated cholesterol efflux to HDL particles [3]. Importantly, cholesterol efflux protects against foam cell formation in arterial macrophages [33]. To assess the ability of proline mutants to facilitate cholesterol efflux, cells transfected with empty vector, WT-, or mutant SR-BI were pre-labeled with [³H]-cholesterol and treated with 50 µg/ml HDL for 4 h. Measurements of [³H]-cholesterol levels in the media and in the remaining cellular lipids were used to calculate the percent of cholesterol efflux. As shown in Figure 6, P412A- and P438A-SR-BI effluxed free cholesterol to HDL at statistically lower levels than WT-SR-BI. On the other hand, the ability of P459A-SR-BI to promote free cholesterol efflux to HDL was similar to that of WT-SR-BI.

Oligomerization of SR-BI receptors may play an important role in cholesterol transport [21], and dimerization is mediated, at least in part, by the N- and C-terminal transmembrane domains [18,23,24]. To determine if the presence of proline residues in the C-terminal transmembrane domain and adjacent extracellular domain is required for SR-BI's ability to oligomerize, we harvested cell lysates in buffer containing PFO (a non-dissociative detergent) [34], separated samples by PFO-PAGE, and performed immunoblot analysis using the antibody directed against the C-terminal region of SR-BI. SR-BI dimers and higher-order oligomers were observed for all proline mutants, and the oligomerization profiles all appeared similar to that of WT-SR-BI (Figure 7).

To improve our understanding of the mechanisms underlying SR-BI-mediated cholesterol transport, we applied a structure–function approach, focusing on the functionally relevant C-terminal transmembrane domain and the proximal extracellular domain. Within these regions, we identified four conserved proline residues and assessed the functional importance of each by mutating individual proline residues to alanine. Our results indicate that Pro-412 and Pro-438 are critical for SR-BI-mediated cholesterol transport, Pro-459 is important for HDL cell association and COE uptake, and Pro-408 is required for SR-BI protein expression.

Unexpectedly, we observed a substantial decrease in P408A-SR-BI total protein expression (Figure 2B) without a corresponding decrease in mRNA expression (Figure 2A). Decreased protein expression was verified using antibodies targeting the C-terminal region of SR-BI (Figure 2B) and extracellular domain (data not shown); however, we cannot exclude the possibility that the P408A mutation alters antibody binding. Furthermore, we confirmed that P408A-SR-BI does not separate into the insoluble pellet following cell lysis (data not shown) and is not present as an immature (unglycosylated, 55 kDa) form of the receptor (Figure 2B). Since a key component of protein maturation and folding is proline cis–trans isomerization [35], we tested the hypothesis that Pro-408 is a critical mediator of SR-BI folding and that P408A-SR-BI is misfolded and rapidly degraded by either ubiquitin-proteasomal or lysosomal degradation. However, our results do not support the idea that P408A-SR-BI is misfolded and degraded by either of these pathways, as treatment with MG-132 (Figure 3A,B) or chloroquine (Figure 3C,D) does not lead to the accumulation of SR-BI protein. Indeed, it is possible that other protein degradation pathways could be responsible for the low levels of P408A-SR-BI, and this will require further investigation. Otherwise, there may be impairment upstream of protein degradation pathways, potentially at the translational level. Pro-408 is likely not critical for SR-BI expression in all species, as it is not conserved in lower vertebrates such as salmon and zebrafish (Figure 1A). The importance of Pro-408 for SR-BI expression in an in vivo system remains to be determined.

The two proline mutants that were defective in all SR-BI-mediated cholesterol transport functions (P412A- and P438A-SR-BI) are also the two most highly conserved proline residues in our study. P412A-SR-BI cell surface expression, as measured by flow cytometry using an extracellular antibody, is increased to ∼280% relative to WT-SR-BI levels (Figure 2C). Although Pro-412 is located outside of the region used to generate the extracellular antibody (residues 230–380), the possibility that P412A mutation disrupts the antibody-binding epitope cannot be ruled out. An increase in P412A-SR-BI expression could be explained as an attempt to compensate for the nonfunctional mutant receptor. Alternatively, increased expression could result from impaired receptor recycling, and this warrants further investigation. Despite seemingly enhanced cell surface expression, P412A-SR-BI remains impaired in its ability to bind HDL, mediate uptake of HDL-CE, efflux free cholesterol, and redistribute membrane cholesterol. Structurally, Pro-412 is located either within an α-helix, as in the NMR structure (Figure 1B), or positioned within a β-strand and oriented away from the core of the extracellular domain, as in the homology model (Figure 1C). The NMR structure may better represent the secondary structure surrounding Pro-412 because it is the only high-resolution structural information for SR-BI. Alternatively, the presence of this α-helix in the NMR structure could be an artifact of determining the structure of only the C-terminal segment of SR-BI in detergent micelles (described in ref. [36]). It remains unknown how mutation of Pro-412 affects SR-BI structure, but short- or long-range conformational changes could disturb the putative hydrophobic channel formed by the β-barrel-like structure (in the homology model) or the nearby three-helix bundle (apex of the homology model). The hydrophobic channel and three-helix bundle are thought to be involved in uptake of CE and HDL cell association, respectively. Pro-438 appears to be in a flexible linker region near the transmembrane domain (Figure 1B,C). If the hydrophobic channel directs cholesterol between the two transmembrane domains, a mutation at the extracellular junction of the C-terminal transmembrane domain could, in theory, impair SR-BI's cholesterol transport functions. Another hypothesis is that Pro-438 forms a flexible kink that enables potential membrane receptor (also known as juxtamembrane) interactions in the adjacent α-helix (residues 426–436) [18]. This hypothesis is currently being investigated by our laboratory. Notably, functional defects with P412A- and P438A-SR-BI are not due to impaired self-association of SR-BI receptors, as both mutants form dimers and higher-order oligomers (Figure 7). The mutant proline receptors were expected to dimerize, as all mutations were made outside of the transmembrane portion of the leucine zipper dimerization motif (residues 413–455).

Mutation of the transmembrane proline kink (Pro-459) to alanine disrupts HDL cell association and HDL-CE uptake, while free cholesterol efflux and membrane cholesterol redistribution are unaffected. It is unknown if HDL binding to SR-BI requires conformational changes involving one or both transmembrane domains, but this could explain why mutation of the flexible proline kink impairs HDL cell association. Despite observing defects in some of SR-BI's cholesterol transport functions, it is possible that the mutant receptor retains the transmembrane kink structure. A similar phenomenon was observed when mutating individual proline kink residues to alanine in bacteriorhodopsin [37]. In addition to forming transmembrane kinks, proline has the unique ability to undergo cis–trans isomerization. Proline cis–trans isomerization is proposed to act as a molecular switch to regulate signal transduction [38,39]. Therefore, we suspect that other known functions of SR-BI involving signal mechanisms, such as macrophage migration [40] and pro-survival signaling [41], could be impaired in the presence of P459A-SR-BI. It is unknown whether Pro-459 undergoes cis–trans isomerization, but we are investigating the hypothesis that conformational changes induced by cis–trans isomerization trigger signaling events mediated by SR-BI that lead to the activation of endothelial nitric oxide synthase [42]. In addition, these signaling events require SR-BI's C-terminal transmembrane domain [43], making the transmembrane proline kink an intriguing target for future investigations. SR-BI's N-terminal transmembrane domain also contains a proline residue (Pro-33), although it is predicted to be more proximal to the extracellular space than Pro-459. It is unknown if Pro-33 induces a kink in the N-terminal transmembrane domain or plays a role in SR-BI's athero-protective functions.

Our findings indicate that Pro-412, Pro-438, and, to some extent, Pro-459 are necessary for maintaining SR-BI in a conformation that facilitates proper cholesterol transport. The contribution of these specific proline residues to reverse cholesterol transport in vivo remains to be determined. Unexpectedly, we also observed that Pro-408 is required for SR-BI expression. Our findings add to a growing list of structural motifs critical to SR-BI cholesterol transport functions and highlight a previously unstudied proline kink within the C-terminal transmembrane domain of SR-BI.

ACAT

acyl-CoA:cholesterol acyltransferase

Ala/A

alanine

ANOVA

analysis of variance

BSA

bovine serum albumin

CE

cholesteryl ester

COE

cholesteryl oleyl ether

DMEM

Dulbecco's modified Eagle's medium

FITC

fluorescein isothiocyanate

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

HDL

high-density lipoproteins

LC3B

microtubule-associated protein 1A/1B-light chain 3B

MFI

mean fluorescence intensity

MG-132

carbobenzoxy-Leu-Leu-leucinal

NMR

nuclear magnetic resonance

PBS

phosphate-buffered saline

PFO

perfluorooctanoic acid

Pro/P

proline

RIPA

radioimmunoprecipitation assay

SEM

standard error of the mean

SR-BI

scavenger receptor-BI

TBST

Tris-buffered saline with 0.05% Tween-20

WT

wild type

S.C.P. and D.S. designed research; S.C.P. performed research; S.C.P. and D.S. analyzed data; S.C.P. and D.S. wrote the manuscript.

This work was supported by the National Institutes of Health grants R01HL58012 (D.S.) and F31HL138744-01 (S.C.P), with partial support from an American Heart Association grant AHA17PRE33650013 (S.C.P.).

The authors thank Kay Nicholson for excellent technical guidance and Darcy Knaack and Hayley Powers for critical review of this manuscript. The authors also thank Dr Brian Volkman for helpful discussions.

The Authors declare that there are no competing interests associated with the manuscript.