Molecular Brain Research Group, Robarts Research Institute, Department of Physiology and Pharmacology, Schulich School of Medicine & Dentistry, University of Western Ontario, London, Ontario, Canada
Address correspondence and reprint requests to Dr Rebecca Jane Rylett, Molecular Brain Research Group, Robarts Research Institute, 100 Perth Drive, London, ON, Canada N6A 5K8. E-mail: email@example.com
The sodium-coupled, hemicholinium-3-sensitive, high-affinity choline transporter (CHT) is responsible for transport of choline into cholinergic nerve terminals from the synaptic cleft following acetylcholine release and hydrolysis. In this study, we address regulation of CHT function by plasma membrane cholesterol. We show for the first time that CHT is concentrated in cholesterol-rich lipid rafts in both SH-SY5Y cells and nerve terminals from mouse forebrain. Treatment of SH-SY5Y cells expressing rat CHT with filipin, methyl-β-cyclodextrin (MβC) or cholesterol oxidase significantly decreased choline uptake. In contrast, CHT activity was increased by addition of cholesterol to membranes using cholesterol-saturated MβC. Kinetic analysis of binding of [3H]hemicholinium-3 to CHT revealed that reducing membrane cholesterol with MβC decreased both the apparent binding affinity (KD) and maximum number of binding sites (Bmax); this was confirmed by decreased plasma membrane CHT protein in lipid rafts in cell surface protein biotinylation assays. Finally, the loss of cell surface CHT associated with lipid raft disruption was not because of changes in CHT internalization. In summary, we provide evidence that CHT association with cholesterol-rich rafts is critical for transporter function and localization. Alterations in plasma membrane cholesterol cholinergic nerve terminals could diminish cholinergic transmission by reducing choline availability for acetylcholine synthesis.
The sodium-coupled choline transporter CHT moves choline into cholinergic nerve terminals to serve as substrate for acetylcholine synthesis. We show for the first time that CHT is concentrated in cholesterol-rich lipid rafts, and decreasing membrane cholesterol significantly reduces both choline uptake activity and cell surface CHT protein levels. CHT association with cholesterol-rich rafts is critical for its function, and alterations in plasma membrane cholesterol could diminish cholinergic transmission by reducing choline availability for acetylcholine synthesis.
Cholinergic neurons play an important role in cognitive processes through actions of the neurotransmitter acetylcholine (ACh) that is produced in and released from their nerve terminals. Hemicholinium-3 (HC-3) sensitive, sodium-coupled, high-affinity choline transporters (CHT) that are situated at the neuronal plasma membrane mediate the uptake of choline that serves as substrate for ACh synthesis (Black and Rylett 2012). Under most conditions, this is the rate-limiting step for ACh production (Haga and Noda 1973; Yamamura and Snyder 1973). Dysfunction of cholinergic neurons underlies the cognitive decline seen in several neurological disorders, including Alzheimer's disease (AD) and mild cognitive impairment. High-affinity choline uptake is disordered in AD (Rylett et al. 1983; Pascual et al. 1991; Slotkin et al. 1994), thus understanding the mechanisms that regulate CHT activity could provide insight into novel therapeutic strategies.
The activity and trafficking of membrane proteins can be regulated by the plasma membrane lipid environment. Proteins can be segregated into cholesterol and sphingolipid-rich microdomains in membranes termed lipid rafts (Allen et al. 2007; Lingwood and Simons 2010) that are involved in cellular events, including membrane organization, protein–protein interactions, and signal transduction (Simons and Ikonen 1997; Maekawa et al. 2003). Lipid rafts also have a role in trafficking of membrane proteins by facilitating their endocytosis in clathrin-independent caveolae (Laude and Prior 2004; Lajoie and Nabi 2007). Cholesterol can play a critical role in organization of membrane proteins, and influence their function or activity either indirectly through changes in membrane fluidity or directly by cholesterol–protein interactions (Mcintosh and Simon 2006; Hicks et al. 2012; Head et al. 2013). The disruption of lipid rafts by removal of cholesterol can significantly affect either the structure or function of some membrane proteins (Pike 2006). In relation to this study, several neurotransmitter transporters are localized to lipid rafts. One mechanism by which this can regulate their solute uptake activity is by altering their endocytosis and trafficking, thereby affecting the amount of cell surface transporter protein (North and Fleischer 1983; Butchbach et al. 2004; Helms and Zurzolo 2004; Magnani et al. 2004). However, specific cholesterol–transporter interactions have also been identified that are independent of association of the protein with lipid rafts. For example, reconstitution of 5-HT transporter and GABA transporter activity is not observed when cholesterol is replaced by other sterols that have similar effects on membrane fluidity (Scanlon et al. 2001; Shouffani and Kanner 1990), and depletion of membrane cholesterol can decrease 5-HT transporter affinity for its substrate (Scanlon et al. 2001).
CHT is active at the plasma membrane, but experimental evidence indicates that the majority of CHT proteins form a subcellular reserve pool within endocytic and synaptic vesicles (Ferguson et al. 2003; Ferguson and Blakely 2004; Ribeiro et al. 2003, 2006). Only a small proportion of total cellular CHT proteins are at the cell surface, with a constitutive cycle of endocytosis by a clathrin/dynamin-dependent mechanism and recycling back to the plasma membrane serving as a critical regulatory mechanism that controls cell surface CHT density (Ferguson et al. 2003; Ferguson and Blakely 2004; Ribeiro et al. 2003, 2006). Plasma membrane CHT levels are enhanced by depolarization-induced exocytosis of synaptic vesicles that deliver additional CHT proteins to the cell surface during excitation of cholinergic nerve terminals, thereby increasing the amount of choline transported into cholinergic nerve terminals to drive ACh synthesis (Ribeiro et al. 2007). CHT trafficking is also regulated by phosphorylation (Black et al. 2010; Gates et al. 2004), extracellular ligand concentration (Okuda et al. 2011), and by interaction with other proteins, such as amyloid precursor protein (Wang et al. 2007).
To date, there have been only limited, and conflicting, reports regarding the influence of cholesterol or cholesterol-rich lipid rafts on high-affinity choline uptake; one study shows that cholesterol reduction in isolated nerve endings (synaptosomes) decreased choline uptake (Waser et al. 1978), whereas another study reports that depletion of membrane cholesterol in synaptosomes did not significantly influence choline uptake (North and Fleischer 1983). A recent study suggests that depleting membrane cholesterol by treating synaptosomes with high concentrations of methyl-β-cyclodextrin (MβC) reduced high-affinity choline uptake (Kristofiková et al. 2008). Although these observations suggest that cholesterol has a role in regulation of CHT, the underlying mechanisms were not investigated and there has been no assessment of whether CHT proteins are partitioned in membrane lipid rafts. Thus, the goal of this study was to address these points and provide data on cholesterol-related changes to CHT activity. We used pharmacological approaches to modulate membrane cholesterol levels in neural cells, and found that treatments that reduced cholesterol caused a reduction in CHT activity and addition of cholesterol to cells resulted in enhanced CHT activity. We show for the first time that CHT proteins are enriched in lipid rafts, and that disruption of lipid rafts reduces cell surface CHT. These results suggest that membrane cholesterol and lipid rafts serve as an important regulator of CHT trafficking and activity by retaining functional CHT at the cell surface.
Materials and methods
Filipin, MβC, cholesterol–MβC complex (MβC–Chol), and cholesterol oxidase were from Sigma-Aldrich (St. Louis, MO, USA) and [methyl-3H]choline chloride (128 Ci/mmol) and [methyl-3H]hemicholinium-3 diacetate ([3H]HC-3) (169 Ci/mmol) were from Perkin-Elmer Life Sciences (Boston, MA, USA). Other chemicals were from Sigma-Aldrich at the highest purity available. SH-SY5Y human neuroblastoma cells were from American Type Culture Collection (Manassus, VA, USA), and Invitrogen (Burlington, ON, Canada) supplied AlexaFluor 647 cholera toxin subunit B conjugate (CTB), Zenon AlexaFluor 555 rabbit IgG and Zenon AlexaFluor 488 mouse IgG labeling kits, Amplex Red Cholesterol Assay kit, and culture media and reagents. Enhanced ChemiLuminescence immunoblot reagent was from GE Healthcare Life Sciences (Baie d'Urfé, QC, Canada) and Biodegradable Scintillant was from Amersham Canada Ltd. (Oakville, ON, Canada). Purified mouse anti-EEA1 antibody was from BD Biosciences (Mississauga, ON, Canada) and rabbit polyclonal flotillin-1 antibody from Santa Cruz Biotechnology, Santa Cruz, CA, USA. Polyclonal CHT antibody was raised in rabbits to the antigenic peptide DVDSSPEGSGTEDNLQ that is conserved at the carboxyl terminus of human and rat CHT (Genemed Synthesis, San Antonio, TX, USA); this peptide was conjugated to keyhole limpet hemocyanin carrier protein by an amino-terminal cysteine. CHT-specific IgG was affinity purified in our laboratory from crude anti-serum on NHS-Sepharose (Amersham) to which antigenic peptide was coupled as the binding element. Specificity of this antibody for detection of CHT was described previously (Pinthong et al. 2008).
Selection of cell line and cell culture
Full-length rat CHT cDNA ligated to pSPORT was a gift from Dr T. Okuda (Okuda et al. 2000); a FLAG epitope tag (DYKDDDDK) was added to the amino terminus by PCR and the resulting cDNA ligated to pcDNA3.1 (Pinthong et al. 2008). SH-SY5Y cells were transfected with this FLAG-CHT plasmid by Lipofectamine 2000. Stable transformants were selected using 500 μg/mL geneticin (G418) for 4 weeks, then grown in Dulbecco's modified Eagle medium, 10% fetal bovine serum 100 U/mL penicillin, and 100 μg/mL each of streptomycin and G418. SH-SY5Y cell differentiation was induced by addition of 10 μM all-trans-retinoic acid for 3 days; substantial morphological and biochemical differentiation of cells occurred during this time (data not shown).
[3H]choline uptake assay
Monolayers of cells were washed, then incubated at 37°C in Krebs-Ringer HEPES solution (KRH) [mM: NaCl, 124; KCl, 5; MgSO4, 1.3; CaCl2, 1.5; glucose, 10; HEPES-NaOH, 20, pH 7.4]. Vehicle or drug was added to cells for specified times, followed by incubation for 5 min with 0.5 μM [3H]choline (0.5 μCi/mL) in the absence or presence of 1 μM HC-3. Following this incubation, cells were placed on ice and washed with cold KRH, then lysed in 0.1 M NaOH. After 30 min lysis, aliquots of samples were analyzed for tritium content by liquid scintillation spectrometry and protein concentration using Bio-Rad protein dye (Bio-Rad Laboratories, Hercules, CA, USA). Each independent experiment consisted of triplicate plates of cells per treatment group, with results normalized to sample protein content and averaged. Specific choline uptake was the difference between total choline uptake and non-specific uptake in the presence of HC-3, with the resulting [3H]choline uptake data expressed as pmol/mg protein per 5 min ± SEM.
[3H]HC-3 binding assay
Monolayers of cells were washed and incubated in KRH at 37°C with addition of vehicle or drug for specified times, then washed with ice-cold KRH, and kept on ice for 10 min to stop protein trafficking activity. Cells were incubated with [3H]HC-3 (10 nM; 1 Ci/mmol) in the presence or absence of 1 μM unlabeled HC-3 for 1 h on ice. For kinetic analysis, HC-3 binding was measured over the range of 0.5–10 nM HC-3 with the specific activity of [3H]HC-3 held constant at 1 Ci/mmol. Following incubation, cells were washed rapidly with cold KRH to remove unbound HC-3 and lysed in 0.1 M NaOH for 30 min. Aliquots of lysates were used for quantification of tritium and protein content. Each independent experiment had triplicate determinations with HC-3 binding normalized to sample protein content, then averaged. Specific HC-3 binding was calculated as the difference between total and non-specific HC-3 binding, and values were expressed as fmol/mg protein ± SEM.
Lipid raft preparation and sucrose flotation gradients
SH-SY5Y cells stably expressing CHT were grown to confluence on 100-mm dishes prior to the preparation of membrane fractions. Cells were placed on ice and washed twice with HEPES-buffered saline solution (HBSS), then lysed for 30 min in sodium carbonate lysis buffer (0.5 M Na2CO3 with 1 mM AEBSF (4-[2-aminoethyl]benzenesulfonyl fluoride hydrocholoride), 10 μg/mL each of leupeptin and aprotinin, and 25 μg/mL pepstatin A) or Triton X-100 lysis buffer (0.5% Triton X-100 with 1 mM AEBSF, 10 μg/mL each of leupeptin and aprotinin, and 25 μg/mL pepstatin A). Cell lysates were collected and homogenized three times for 10 s using a Polytron tissue grinder. Homogenates were next sonicated three times for 20 s using a Sonic Dismembrator. Homogenates were adjusted to 40% sucrose, then added to ultracentrifuge tubes and overlaid with 30% sucrose and 5% sucrose solutions. Samples were centrifuged at 120 000 g for 16 h at 4°C using a Beckman SW 41 (Beckman Coulter Canada LP, Mississauga, ON, Canada) Ti rotor. Twelve or ten (for cells lysed in sodium carbonate or Triton X-100, respectively) 1 mL fractions were collected from the top to bottom of each sucrose gradient. Protein aliquots from each fraction were incubated for 10 min at 55°C with Laemmli buffer (2% sodium dodecyl sulfate, 10% glycerol, 62.5 mM Tris-HCl, pH 6.8, 2.5% β-mercaptoethanol, and 0.001% bromophenol blue), then separated on 7.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) gels and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked in 8% non-fat dry milk in wash buffer (phosphate-buffered saline, 0.15% Triton X-100) and incubated with either anti-CHT antibody, anti-EEA1 antibody or anti-flotillin antibody overnight at 4°C. After washing, membranes were incubated for 1 h with species-matched secondary antibodies in wash buffer containing 8% milk, and washed again. Immunoreactive proteins on membranes were detected by chemiluminescence using the Enhanced ChemiLuminescence kit. Immunopositive bands were quantified by densitometry using Scion Image software (NIH).
Purified nerve terminals [synaptosomes] were isolated from mouse forebrain using established methods (Rylett and Walters 1990). Young adult (3–4 months of age) C57B6 mice of either sex bred locally were used in these experiments in compliance with the Canadian Council on Animal Care (CCAC) and the ARRIVE (www.nc3rs.org.uk/ARRIVE) guidelines, and a protocol approved by the Animal Use Subcommittee at the University of Western Ontario. Briefly, tissues were homogenized in 0.32 M sucrose buffered with 5 mM HEPES, pH 7.4. Following differential centrifugation at 1000 g for 10 min and 12 000 g for 20 min to recover the crude P2 fraction, pellets were resuspended in buffered 0.32 M sucrose and layered onto discontinuous gradients comprised of 8.5, 13, and 20% Ficoll in buffered 0.32 M sucrose. Following centrifugation at 27 000 g for 45 min, fractions at the 8.5–13% and 13–20% interfaces were recovered as they contained purified synaptosomes; these were diluted 5-fold with 0.32 M sucrose, then centrifuged at 12 000 g for 30 min. Purified synaptosome pellets were lysed in sodium carbonate lysis buffer and the resulting homogenates used directly to isolate lipid rafts, as described above.
Digital images of fixed cells were acquired with a Zeiss LSM510-Meta laser-scanning confocal microscope (Carl Zeiss Canada Ltd., Toronto, ON, Canada) using a 63X oil-immersion objective and magnified three times, unless specified otherwise. FLAG-tagged CHT proteins were fluorescently labeled in live cells using rabbit anti-FLAG antibody complexed to either Zenon 555 dye (for EEA1 staining) or Zenon 488 dye (for flotillin staining). The fluorescently tagged antibody was added to medium bathing the cells where it could bind to the FLAG epitope located on the extracellularly oriented amino terminus of CHT proteins at the plasma membrane. After allowing internalization of Zenon-labeled CHT proteins, cells were formaldehyde fixed then counter-stained with AlexaFluor 647 CTB (0.2 ng/μL) and either anti-flotillin (1 : 200) or anti-EEA1 (1 : 100) antibodies. Following immunostaining, images were acquired using 488-nm excitation and 505- to 530-nm emission wavelengths for flotillin; 543-nm excitation and 560- to 615-nm emission for EEA1; and 647-nm excitation and 650-nm emission using a long-pass filter for CTB. FLAG-CHT was visualized using 488-nm excitation and 505- to 530-nm emission (for EEA1-labeled cells) or using 543-nm excitation and 560- to 615-nm emission (for flotillin-labeled cells). Images were processed and colocalization was analyzed in Imaris (version 7.0.0; Bitplane Scientific Software, South Windsor, CT, USA), then formatted in Adobe Illustrator and Photoshop (Adobe Systems Inc, San Jose, CA, USA).
SH-SY5Y cells expressing FLAG-tagged CHT were grown to confluence on 100-mm dishes, then lysed in sodium carbonate lysis buffer and lipid rafts were isolated on sucrose gradients, as described above. Total cholesterol levels were measured in pooled raft fractions (#4–6) and non-raft fractions (#9–12) using the Amplex Red Cholesterol Assay, according to manufacturer's instructions. Briefly, 50 μL of cholesterol standards (0–20 μM per well) or sample replicates was added to 96-well black plates. Fifty μL of working solution (300 μM Amplex Red reagent, 2 unit/mL horseradish peroxidase, 2 unit/mL cholesterol oxidase, and 0.2 unit/mL cholesterol esterase) was added to each well and plates were incubated in the dark for 30 min at 37°C. Fluorescence was measured (excitation 544 nm, emission 590 nm, SpectraMax M5 Molecular Devices, Palo Alto, CA, USA). Relative fluorescence units of samples were converted into μM cholesterol per μg sample protein using the standard curve, and data were expressed as a percentage of vehicle-treated controls.
Cell surface biotinylation assay
Cells plated on 100-mm dishes were washed with HBSS, then treated at 37°C with either vehicle or drug. After treatment, cells were placed on ice under cold HBSS to stop protein trafficking. Plasma membrane proteins were biotinylated at 4°C by incubating with 1 mg/mL sulfo-NHS-SS-biotin in HBSS for 1 h (Dale et al. 2004; Pinthong et al. 2008) with gentle agitation. Unbound biotin was quenched by washing and incubating cells in cold 100 mM glycine in HBSS. After two further washes with HBSS, cells were lysed on ice for 30 min in sodium carbonate lysis buffer and lipid rafts prepared on sucrose flotation gradients as described above. Twelve 1-mL gradient fractions were collected and pooled as either raft fractions (#4–6) or non-raft fractions (#9–12). Pooled fractions were diluted to 10 mL with lysis buffer (1% w/v Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, and pH 7.5) and pH adjusted to 7.5. Protein concentrations were measured using Bio-Rad dye and aliquots of cell lysates containing 30 μg protein subjected to SDS–PAGE to determine total CHT, flotillin, and EEA1 levels. Biotinylated proteins in sucrose density gradient fractions were separated from non-biotinylated proteins by Neutravidin bead pull-down from 500 μg total cellular protein from each sample. Beads were washed three times with lysis buffer to remove non-specific proteins. Proteins were eluted for 10 min at 55°C with Laemmli buffer, then separated on 7.5% SDS–PAGE gels and transferred to PVDF membranes. Immunoblots for CHT, flotillin, and EEA1 were carried out as described above and analyzed using a Chemidoc Imaging System (Bio-Rad Laboratories).
Internalization cell surface biotinylation assay
Cells plated on 100-mm dishes were washed twice with cold HBSS and placed on ice under cold HBSS to stop protein trafficking. Plasma membrane proteins were biotinylated at 4°C by incubating with 1 mg/mL sulfo-NHS-SS-biotin in HBSS for 1 h with gentle agitation, then unbound biotin was quenched by washing and incubating cells in cold 100 mM glycine in HBSS. Dishes were washed with warm HBSS, then incubated at 37°C with either vehicle or drug to allow constitutive internalization of CHT proteins (Ribeiro et al. 2005). CHT internalization was subsequently terminated by transferring dishes of cells to ice and replacing medium with cold HBSS. Residual cell surface biotin was stripped by incubating cells once for 15 min and twice for 30 min with freshly prepared 50 mM mercaptoethanesulfonic acid (MesNa) in TE buffer (150 mm NaCl, 1 mm EDTA, 0.2% bovine serum albumin, 20 mm Tris, and pH 8.6). Stripping efficiency was determined for each experiment on parallel sets of biotinylated cells that were maintained on ice and which did not undergo MesNa stripping to assess the amount of CHT at the plasma membrane. Cells were lysed in lysis buffer (1% w/v Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1 mM AEBSF, 10 μg/mL each of leupeptin and aprotinin, 25 μg/mL pepstatin A, and 700 U/mL DNase) and protein concentration was measured using Bio-Rad dye. Biotinylated proteins were separated from non-biotinylated proteins by Neutravidin bead pull-down from 500 μg protein from each sample. Beads were washed three times with lysis buffer to remove non-specifically bound proteins, then proteins eluted for 10 min at 55°C with Laemmli buffer, separated on 7.5% SDS–PAGE gels, and transferred to PVDF membranes. Immunoblots for CHT were carried out as described above and analyzed using a Chemidoc Imaging System.
Data are presented as mean ± SEM with n values representing the number of independent experiments performed on separate populations of cells. Each n value was obtained from the average of multiple sample replicates in each experiment. Replicate experiments were performed on cells cultured in successive passages as much as possible to minimize interexperiment variability; intraexperiment variability between replicate samples was minimal thus facilitating comparison of treatment effects. GraphPad Prism 5 (GraphPad Software, San Diego, CA, USA), InStat software (GraphPad Software) and Image Lab 4.1 (Bio-Rad Laboratories) were used for data analysis. Sigmoid and Michaelis–Menten equations were used to calculate kinetic parameters (Bmax and KD) of HC-3 binding. Data were assessed for statistically significant differences by unpaired Student's t-test, or between groups using repeated measures one-way anova with Tukey's post hoc multiple comparison test as appropriate, with statistical significance defined as p ≤ 0.05.
The initial experiments were designed to determine the effect of manipulation of membrane cholesterol content on HC-3-sensitive, high-affinity choline uptake activity in SH-SY5Y cells that stably express FLAG-tagged CHT. Three separate pharmacological approaches having different mechanisms of action were used to lower membrane cholesterol levels, and one approach was used to increase the membrane cholesterol level. Cells were treated with either filipin which binds to and sequesters free cholesterol within both lipid raft and non-raft areas of membrane and disrupts lipid rafts (Cremona et al. 2011), MβC which extracts unesterified cholesterol from plasma membrane resulting in the disruption of lipid rafts and direct cholesterol–protein interactions (Zidovetzki and Levitan 2007), or cholesterol oxidase which converts cholesterol to the inactive sterol cholest-4-en-3-one and does not alter membrane fluidity (Gimpl et al. 1997). Neither cell density of cultures nor cell morphology was visibly altered after drug treatments, and the level of total CHT protein was unchanged (data not shown). We predicted that if cholesterol is required for CHT function, then HC-3-sensitive choline uptake activity would be altered by these drugs. Figure 1 illustrates dose–response relationships for [3H]choline uptake in cells treated with filipin, MβC, and cholesterol oxidase. When cells were pre-treated with vehicle or filipin for 60 min, [3H]choline uptake was significantly reduced at both doses of filipin (2.5 or 5 μg/mL) (Fig. 1a). Similarly, treating cells with MβC for 30 min (0.1 or 0.3 mM) resulted in a statistically significant decrease in [3H]choline uptake at the higher concentration of the drug (Fig. 1b). Finally, when cells were treated with cholesterol oxidase for 30 min, a statistically significant reduction in [3H]choline uptake activity was observed at both concentrations of cholesterol oxidase (0.5 and 1 U/mL) compared to vehicle-treated control cells (Fig. 1c).
In the next set of experiments, we determined if CHT activity is altered by increasing the level of membrane cholesterol. SH-SY5Y cells were incubated with a cholesterol-saturated form of MβC (MβC–Chol) for 30 min (MβC–Chol complex concentration was 5 or 25 μM, calculated by weight of cholesterol) or with vehicle, then [3H]choline uptake was measured. Importantly, when cholesterol was added to cells, a statistically significant increase in choline uptake was observed at 25 μM MβC–Chol when compared to vehicle-treated cells (Fig. 1d).
MβC decreases both the Bmax and KD for [3H]HC-3 binding to CHT
Several studies suggest that membrane cholesterol and lipid rafts regulate neurotransmitter transporter activity by controlling the number of transporters at the cell surface (Butchbach et al. 2004; Jayanthi et al. 2004; Samuvel et al. 2005). However, cholesterol can also regulate the function of neurotransmitter transporters by modulating their solute binding activity (Shouffani and Kanner 1990; Scanlon et al. 2001; Jones et al. 2012). To establish whether the observed decrease in choline uptake following cholesterol depletion was owing to a change in ligand binding affinity (KD) or to a change in the number of cell surface CHT proteins (Bmax), kinetic analysis of binding of the non-transported competitive inhibitor [3H]HC-3 to CHT was performed following treatment of cells with either vehicle or 0.3 mM MβC (Fig. 2a). These data reveal that cell surface CHT levels were reduced by approximately 50% by MβC treatment, with Bmax for [3H]HC-3 binding decreased from 470.5 ± 115 to 220.7 ± 77 fmol [3H]HC-3 bound/mg protein in control and MβC-treated cells, respectively (Fig. 2b). This coincides with a small, but significant, MβC-mediated increase in binding affinity with the KD for [3H]HC-3 binding reduced from 8.5 ± 1.1 to 4.0 ± 0.6 nM in control and MβC-treated cells, respectively (Fig. 2c).
CHT proteins are associated with lipid rafts
The observation that choline uptake activity is modulated by changes in membrane cholesterol suggests that CHT proteins may be located in lipid rafts. Technically, lipid rafts are defined as membrane domains that are insoluble in cold non-ionic detergents, such as Triton X-100, and have a specific buoyant density during sucrose gradient centrifugation (Allen et al. 2007). However, non-ionic detergents can also solubilize proteins that are only weakly associated with lipid rafts, and also lipid and protein composition of rafts can differ between different extraction methods (Hooper 1999; Janes et al. 2000; Waugh et al. 1999). Therefore, we used a second approach to isolate lipid rafts and validate our findings using a non-detergent procedure that is based on pH and carbonate resistance of lipid raft domains (Song et al. 1996).
Importantly, we show here for the first time that CHT proteins are highly concentrated in lipid rafts prepared from SH-SY5Y cells that stably express the transporter (Fig. 3). Membrane lipid rafts from cells lysed in either 1 M Na2CO3 pH 11 or 0.5% Triton X-100 were isolated using sucrose density gradient ultracentrifugation (Di Guglielmo et al. 2003; Luga et al. 2009). Solubilized proteins remain in the bottom fractions of the gradient (within 40% sucrose), whereas proteins associated with lipid rafts float to the upper layers (between 5% and 30% sucrose). To confirm isolation of lipid rafts, we also analyzed the distribution of flotllin-1, a lipid raft protein (Lamb et al. 2002), and EEA1, a non-raft protein (Di Guglielmo et al. 2003), in gradient fractions. Thus, CHT proteins are found predominantly in lipid raft fractions, and this distribution profile is similar in cells lysed with either Na2CO3 (Fig. 3a and c) or Triton X-100 (Fig. 3b and d). Figure 3a and b show representative immunoblots for CHT protein, with blotting membranes stripped and reprobed with anti-flotillin antibodies and anti-EEA1 antibodies. The histograms displayed in Fig. 3c and d illustrate quantification of CHT protein distribution in sucrose density gradients by densitometry. Enrichment of CHT proteins in lipid rafts is seen with 85 ± 10% and 85 ± 6% of total CHT present in pooled flotillin-positive raft fractions, compared to 15 ± 6% and 15 ± 5% of total CHT present in EEA1-positive non-raft fractions from cells lysed with Na2CO3 and Triton X-100, respectively.
To confirm if CHT proteins are also present in lipid rafts in brain cholinergic nerve terminals, we prepared purified synaptosomes from mouse forebrain. Synaptosomes were lysed in 1 M Na2CO3, then membrane lipid rafts were isolated on sucrose density gradients and immunoblots performed for CHT, EEA1, and flotillin. Figure 3e shows a representative immunoblot of the distribution of these proteins in gradient fractions. Importantly, in confirmation of our findings using cultured neural cells, a substantial proportion of CHT protein was found in lipid raft fractions. Densitometric quantification of CHT protein on immunoblots revealed that 47 ± 5% is present in fraction #5 that also coincides with the greatest density of the lipid raft protein flotillin. This is compared to 53 ± 15% in EEA1 positive non-raft fractions #9–12 (Fig. 3f). Interestingly, the maximum CHT protein distribution in fractions #5 and 12 coincide with higher levels of free cholesterol content (Fig. 3g).
CHT colocalizes with the lipid raft markers flotillin and ganglioside GM1
Confocal imaging experiments where cell surface CHT proteins were fluorescently labeled in live SH-SY5Y cells allowed further assessment of distribution of the transporter in raft and non-raft membrane domains. We assessed the colocalization of CHT with CTB, which binds to ganglioside GM1 and is found concentrated in lipid rafts (Simons and Toomre 2000), along with either flotillin (raft marker) or EEA1 (non-raft marker). To identify colocalized proteins, a threshold fluorescence intensity was set that filters for the brightest 2% of pixels of CHT (red channel) that also fall within the brightest 2% of pixels of and GM1 (blue channel) and either flotillin (green channel) or EEA1 (green channel); this approach to visualize protein colocalization was validated previously (Lorenzen et al. 2010, Cuddy et al. 2012). The colocalized pixels are identified in a separate colocalization channel (shown as ‘Co-localized pixels’ images). We found that CHT colocalizes to a similar extent with both GM1 (Fig. 4a, colocalized pixels shown in purple) and flotillin (Fig. 4a, colocalized pixels shown in yellow), and that all three proteins are found together in a punctate distribution pattern indicative of lipid rafts (triple colocalized pixels panel not shown). In contrast, in cells stained for EEA1 and GM1, we found that CHT appears to be found more in GM1-positive compartments (Fig. 4b, colocalized pixels shown in purple) than colocalized with EEA1 (Fig. 4b, colocalized pixels shown in yellow).
Effect of cholesterol manipulating drugs on cholesterol level in SH-SY5Y cell membranes
We analyzed the effect of treatment of SH-SY5Y cells with either MβC or filipin on free and esterified cholesterol content in pooled lipid raft fractions (#4–6) and non-raft fractions (#9–12) obtained from sucrose gradients (Fig. 5). As expected, total cholesterol levels were higher in lipid raft fractions when compared to non-raft fractions, with about 65% of the cholesterol contained in rafts (data not shown). Filipin, which functionally depletes cholesterol by forming filipin–cholesterol complexes within the membrane, did not significantly alter cholesterol levels. In cells treated with 5 μg/mL filipin, cholesterol levels were 7.2 ± 0.5 and 3.9 ± 0.5 μmol/μg protein in raft and non-raft fractions, respectively, compared to vehicle-treated controls where cholesterol levels were 7.2 ± 0.4 and 3.9 ± 0.3 μmol/μg protein in lipid raft and non-raft fractions, respectively (Fig. 5a). In contrast, treatment of cells with 0.3 mM MβC, which extracts cholesterol from membranes, significantly reduced cholesterol content in raft fractions but not in non-raft fractions; in MβC-treated cells, cholesterol levels were 6.8 ± 0.4 and 3.7 ± 0.5 μmol/μg protein in raft fractions and non-raft fractions, respectively, compared to 7.7 ± 0.3 and 3.8 ± 0.4 μmol/μg protein in raft and non-raft fractions, respectively, in vehicle-treated cells (Fig. 5b). This corresponds to a cholesterol reduction of approximately 10% in lipid raft fractions, which compares with the findings of Foster et al. (2008) who showed that treatment of cells with a much higher concentration of MβC (5 mM) causing a 52% reduction in cholesterol in lipid raft fractions.
Lipid raft disruption decreases plasma membrane CHT protein levels
Kinetic analysis of [3H]HC-3 binding to CHT (Fig. 2) revealed that MβC treatment of cells decreases Bmax, suggesting that the number of CHT proteins at the plasma membrane is decreased. To confirm this and extend our studies on how lipid raft disruption alters CHT localization at the plasma membrane, we undertook cell surface protein biotinylation experiments. CHT proteins are located predominantly in subcellular endocytic compartments and synaptic vesicles (Ribeiro et al. 2006) that also contain lipid rafts. Since we wanted to examine the relative distribution of CHT within lipid rafts only at the cell surface, plasma membrane proteins were biotinylated using membrane impermeable sulfo-NHS-biotin at 4°C to ensure that labeled CHT proteins did not undergo endocytosis. Prior to protein biotinylation, cells were treated at 37°C with either 5 μg/mL filipin, 0.3 mM MβC or vehicle. Na2CO3-insoluble lipid rafts were isolated, then biotinylated proteins were recovered from the pooled raft and non-raft fractions for analysis.
A critical finding is that modification in membrane cholesterol by either filipin or MβC reduced the amount of CHT protein in plasma membrane lipid raft fractions, but not in non-raft fractions (Fig. 6). Representative immunoblots show the levels of cell surface (biotinylated) CHT protein, and the amount of total CHT protein, flotillin-1, and EEA1 in raft and non-raft fractions from cells treated with either filipin (Fig. 6a) or MβC (Fig. 6b). In vehicle-treated cells, plasma membrane CHT proteins are concentrated in lipid rafts with a distribution similar to that seen for total cellular CHT protein (Fig. 3). Cell surface CHT levels in raft fractions were significantly reduced by approximately 50% in both filipin- and MβC-treated cells when compared to vehicle-treated cells (p <0.05) (Figs. 6c and d). This does not reflect redistribution of CHT to non-raft areas of plasma membrane, as changes were not observed in CHT levels in non-raft fractions between vehicle- and drug-treated cells. Immunoblots for cell surface CHT were reprobed with anti-flotillin and anti-EEA1 antibodies to confirm the isolation of lipid rafts. As predicted, flotillin-1 levels in raft fractions were reduced in drug-treated cells confirming disruption of lipid rafts. No changes in EEA1 levels or distribution were observed between samples from vehicle- and drug-treated cells. Taken together, these results suggest that total cellular and cell surface CHT proteins are partitioned similarly between lipid raft and non-raft membrane domains in SH-SY5Y cells, with lipid raft disruption resulting in reduced cell surface CHT in lipid rafts.
It is important to note the presence of CHT-immunopositive bands with an apparent molecular mass of about 100 kDa in lipid raft fractions at higher exposures of the immunoblots for cell surface CHT proteins (Fig. 6a and b). These bands, which are not present in the non-raft fractions, likely represent CHT protein dimers. In both filipin- and MβC-treated cells, levels of the putative CHT dimers at the cell surface are significantly reduced by about 50% (p <0.05) (Fig. 6e and f).
Lipid raft disruption does not alter CHT internalization from the cell surface
CHT proteins at the cell surface undergo constitutive internalization by a clathrin-mediated mechanism into early endosomes, and these proteins can recycle back to the cell surface (Ribeiro et al. 2006). We tested the hypothesis that the decrease in plasma membrane CHT protein in filipin- and MβC-treated cells is because of an increase in internalization of CHT. SH-SY5Y cell plasma membrane proteins were biotinylated on ice, and then cells were incubated at 37°C with either vehicle, 0.3 mM MβC or 5 μg/mL filipin to allow endocytosis of proteins from the cell surface to subcellular organelles. Only internalized biotinylated proteins were measured as, following drug treatments, cells were returned to ice to terminate endocytosis and residual biotin remaining on cell surface proteins was stripped by membrane-impermeant MesNa. Figure 7 illustrates that neither filipin nor MβC significantly alters CHT internalization from the cell surface to subcellular organelles. The amount of internalized biotinylated CHT protein was normalized to total cell surface CHT levels from cells that were incubated on ice in the absence of either filipin or MβC (Fig. 7a and c, lane 2). A separate set of cells was incubated on ice throughout the procedure to ensure that MesNa efficiently stripped biotin from cell surface proteins (Fig. 7a and c, lanes 1). Immunoblots of cell lysates (Fig. 7a and b, lower panels) show that there were no changes in total CHT levels in cells as a consequence of the treatments. A statistically significant time-related increase in internalized CHT was observed in cells treated with either vehicle or filipin at 60 min when compared to 15 min, but no other statistically significant changes between groups were found (Fig. 7b and d).
Identification of putative cholesterol-binding motifs in CHT protein
Transmembrane proteins may bind cholesterol molecules, with the two most common sites for this interaction on the protein being the Cholesterol Recognition/interaction Amino acid Consensus sequence (CRAC domain) and the CARC domain that comprised an inverted CRAC domain sequence (Baier et al. 2011; Fantini and Barrantes 2013). The definition of both of these motifs is based on a triad of basic (K or R), aromatic (Y or F), and aliphatic (L or V) residues. The CRAC domain sequence is defined as L/V-(X)1-5-Y-(X)1-5-R/K, where X can be one to five residues of any amino acid. The CARC domain is the reverse sequence and the residue Y may be substituted by F. Analysis of the primary sequence of rat CHT reveals putative cholesterol-binding motifs, with these being conserved between rodent and human proteins. As illustrated in Fig. 8, candidate sequences for putative CRAC and CARC domains are found in transmembrane domains (TM) 4, 11, 12, and 13 of CHT. TM 4 contains the CARC sequence R-X5-F-X3-L that could bind cholesterol in the plasma membrane cytoplasmic leaflet. TM 13 contains a CRAC sequence V-X1-Y-X2-K, also on the bilayer cytoplasmic side. CRAC-like motifs, bearing close similarity to CRAC sequences, that could potentially bind cholesterol on the cytoplasmic leaflet are in TM 11 and 12 [TM 11, L-X3-F-X1-K or L-X4-F-X1-K; TM 12, V-X2-Y-X6-R]. In addition to cholesterol binding to CRAC and CARC motifs, transmembrane domains containing G-XXX-G motifs may also be engaged in binding cholesterol molecules (Barrett et al. 2012; Fantini and Barrantes 2013). Importantly, as shown in Fig. 8, TM 12 of CHT has a conserved G-XXX-G-XXX-G motif (G-AVA-G-YVS-G) that could potentially serve as both a cholesterol-binding motif and as a dimerization motif. This sequence of the protein also coincides with a CRAC-like domain.
We made several novel findings in this study that indicate that partitioning of CHT proteins into cholesterol-rich lipid rafts, and potentially direct interaction of CHT with cholesterol, may play a critical role in CHT localization at the cell surface and activity. We show for the first time in both SH-SY5Y cells and mouse brain cholinergic nerve terminals that CHT proteins are partitioned between membrane lipid raft and non-raft microdomains, and tend to be concentrated in lipid rafts. Microscopically, CHT proteins colocalize with the lipid raft-resident protein flotillin and the lipid raft marker ganglioside GM1, and to a lesser extent with the non-raft protein EEA1. Second, we find that choline uptake activity shows a dependence on membrane cholesterol levels. Thus, treatment of neural cells with the cholesterol interfering drugs filipin, cholesterol oxidase, or MβC causes a significant decrease in HC-3-sensitive choline uptake. Moreover, choline uptake is significantly increased by adding exogenous cholesterol to cells. Third, we determined that reducing cell membrane cholesterol levels and disrupting lipid rafts by extracting cholesterol with MβC decreased the number of cell surface CHT proteins when measured by [3H]HC-3 binding to CHT (decreased Bmax) and enhanced ligand binding affinity (decreased KD). Fourth, we observed that treatment of cells with filipin and MβC results in a decrease in plasma membrane CHT protein levels from lipid rafts, but not non-raft areas, measured by cell surface protein biotinylation. This is not related to an accelerated rate of CHT protein internalization. Finally, we identified putative cholesterol-binding motifs in CHT protein that are conserved between rodent and human.
We used four separate pharmacological approaches that alter membrane cholesterol by different mechanisms to assess how cholesterol may regulate CHT function and activity. Treatment of cells with cholesterol oxidase results in conversion of membrane cholesterol to its functionally inactive analogue steroid 4-cholesten-3-one, but does not alter the physical state of plasma membranes (Gimpl et al. 1997). This allowed us to test if the change in choline uptake activity in choline oxidase-treated cells is owing to altered membrane fluidity, rather than a change in membrane cholesterol levels. Based on our findings, it is unlikely that the decrease in choline uptake observed in these cells is because of an alteration of membrane fluidity. In other experiments, we treated cells with MβC which extracts cellular cholesterol and can disrupt lipid rafts and cholesterol–protein interactions in both raft and non-raft areas (Zidovetzki and Levitan 2007). Low concentrations of MβC (0.3 mM) significantly decreased cholesterol content measured in membrane lipid raft fractions and resulted in decreased choline uptake activity. However, it is not possible to determine from these experiments if the loss of CHT activity is caused by a direct or indirect effect of lowering cholesterol. Filipin, which binds to and chelates free cholesterol within the membrane resulting in disruption of lipid rafts, was not expected to lower cholesterol levels and our results are in agreement with previous reports with filipin and the polyene antifungal agent nystatin (Cremona et al. 2011; Jones et al. 2012). Filipin treatment did, however, reduce choline uptake activity, suggesting that disruption of lipid rafts does impact CHT function. Finally, we incubated cells with MβC–Chol to increase membrane cholesterol levels and this significantly enhanced choline uptake. Based on these experimental manipulations, it appears that CHT function requires the presence of membrane lipid rafts and may also be modulated by changes in cholesterol levels, indicating a potential role for direct interaction between cholesterol and CHT proteins.
Decreased choline uptake in cells treated with agents that reduce membrane cholesterol could be because of a loss of functional transporters at the cell surface, or be the result of conformational changes in CHT protein that alter solute binding affinity or impair solute translocation subsequent to binding. Kinetic analysis of binding of the inhibitor [3H]HC-3 to CHT in MβC-treated cells and quantification of cell surface protein biotinylation assays in cells treated with either filipin or MβC indicate that there is a reduction in CHT levels at the plasma membrane. Interestingly, the decrease in CHT protein is seen in lipid raft fractions prepared from either filipin- or MβC-treated cells and is not accompanied by an increase in CHT levels in non-raft fractions, indicating that down-regulation of choline uptake activity is not a result of CHT movement between these compartments. Indeed, some proteins display redistribution between these membrane compartments in response to lipid raft disrupting treatments, whereas other proteins do not (Allen et al. 2007). Our results are in agreement with those found for the dopamine transporter (DAT) by Foster et al. (2008) who showed that while MβC treatment reduced DAT cell surface levels, it did not affect partitioning of DAT proteins between lipid raft and non-raft fractions. Our finding of increased binding affinity (KD) of CHT for HC-3 in MβC-treated SH-SY5Y cells is in agreement with a report of increased HC-3 binding affinity to CHT observed in MβC-treated synaptosomes (Kristofiková et al. 2008), and is similar to reports of increased ligand binding affinity found for other neurotransmitter transporters, such as DAT in MβC-treated cells (Cremona et al. 2011; Jones et al. 2012). The change in binding properties of CHT in MβC-treated cells could indicate that the conformational state of CHT required for solute binding is modulated by either lipid raft association or by cholesterol binding. Previous studies show that CHT proteins may exist in more than one state or conformation and that binding sites in some states may be occluded under some conditions (Rylett 1986; Saltarelli et al. 1987; Ferguson et al. 1994). Alternatively, changes to the physical properties of lipid rafts may disrupt cholesterol-sensitive protein–protein interactions involving CHT or affect the ability of CHT to undergo post-translational modifications such as phosphorylation, both of which could affect solute binding or translocation (Foster et al. 2008; Jones et al. 2012).
Membrane cholesterol and/or lipid rafts could have a role in regulating the amount of CHT protein at the cell surface by altering its rate of internalization into endosomes and/or its recycling back to the cell surface. Several reports suggest that proteins can traffic to the plasma membrane and be selectively recruited to lipid raft microdomains (Michaely et al. 1999; Lindwasser and Resh 2001; McCabe and Berthiaume 1999). We showed previously that alterations in CHT internalization critically affect transporter function (Ribeiro et al. 2003; Pinthong et al. 2008; Black et al. 2010; Cuddy et al. 2012), while CHT recycling is required to maintain the level of transporter at the cell surface thereby regulating CHT activity to deliver sufficient choline to maintain ACh synthesis (Apparsundaram et al. 2005; Ribeiro et al. 2005, 2007). CHT undergoes endocytosis by a clathrin- and dynamin-dependent mechanism (Ferguson and Blakely 2004; Ribeiro et al. 2006), but there is no evidence that CHT internalizes by a clathrin-independent, caveolin-dependent process from lipid raft domains. In this study, we found that CHT protein internalization was not significantly altered by disruption of membrane cholesterol or rafts by either filipin or MβC, suggesting that the decrease in plasma membrane CHT levels was not owing to increased endocytosis; while not measured in these studies, reduced recycling of transporters to the cell surface could result in decreased cell surface levels of the protein. These experiments also reinforce that CHT proteins do not likely undergo internalization by caveolae since, as lipid rafts were disrupted, an increase in plasma membrane CHT protein would have been predicted. In the current studies, cells were treated with a relatively low concentration of MβC (0.3 mM) since we observed that higher concentrations of the drug (2 or 5 mM) used in other studies led to accumulation of CHT proteins at the cell surface (data not shown). MβC can inhibit clathrin-mediated endocytosis (Subtil et al. 1999) and as CHT undergoes rapid constitutive endocytosis, conditions that disrupt this process can result in substantial accumulation of the protein at the cell surface. Thus, we chose to use an MβC concentration that significantly decreased cholesterol levels in lipid rafts, and was less likely to perturb endocytosis.
Interestingly, we observed a CHT-immunopositive band with an apparent molecular mass of about 100 kDa in lipid raft fractions. These bands could represent homo-oligomers of CHT proteins, with recent work by Okuda et al. (2012) showing that CHT expressed in cultured cells can form homo-oligomers at the cell surface. Here, we provide evidence that these homo-dimers are located within lipid rafts, and that these are reduced by lipid raft disrupting treatments. This suggests that oligomerization may result in stabilization of CHT into the raft domain and be required for cell surface localization of CHT. Detection of CHT in cell lysates often reveals multiple protein species (Misawa et al. 2001; Okuda et al. 2002; Ribeiro et al. 2005) and it has been suggested that these represent varying glycosylated forms of CHT (Ferguson et al. 2003). CHT may undergo differential post-translational modifications, such as phosphorylation, between raft and non-raft areas of membrane, but it remains to be established if these impact either CHT activity or plasma membrane trafficking. For example, activation of protein kinase C (PKC) enhances internalization of the noradrenaline transporter through a dynamin- and clathrin-independent lipid raft-mediated process (Jayanthi et al. 2004). In comparison, PKC-induced internalization of DAT is dynamin and clathrin dependent, and is independent of lipid rafts (Foster et al. 2008).
Several lines of evidence support a connection between alterations in cholesterol levels and its disposition in aging (Martin et al. 2010) and AD (Poirier et al. 1995; Simons et al. 2001; Kivipelto 2001), and the organization of lipid rafts is disrupted in AD brains (Ledesma et al. 2003). Here, we provide the first evidence that association of CHT proteins with cholesterol-rich lipid rafts is critical for transporter function and localization. These data are important for understanding the normal function of widely distributed cholinergic neurons, and how they are altered in disorders that involve cognitive dysfunction owing to reduced cholinergic transmission. These studies will allow us to analyze responses to changes in the cholinergic nerve endings and to assess alterations in pathology to aide in the design of new therapies. Therapeutic strategies directed at CHT function may be useful for enhancing cholinergic transmission, for example, early in AD when ACh release is reduced but cholinergic neurons remain viable. Given the projected increases in our aging population and the prevalence of neurodegenerative disorders, these studies are both timely and critical.
This research was supported by a grant to RJR from Canadian Institutes for Health Research (CIHR). LKC is the recipient of a Queen Elizabeth II Graduate Scholarship in Science & Technology and Doctoral Award from Alzheimer Society of Canada. WWN is recipient of a Schulich Doctoral Research Scholarship. The authors do not have any conflicts of interest, financial or otherwise, to disclose in relation to publication of this manuscript.