HeLa cells were cultured in Dulbecco modified Eagle's medium (DMEM) with 10% heat-inactivated fetal calf serum (FCS) (Invitrogen, Cergy-Pontoise, France), 2 mM glutamine, 100 UI/ml penicillin, 100 µg/ml streptomycin (Sigma-Aldrich, St Louis, MO). Cells were treated with dynasore (Sigma-Aldrich) at 80 µM in the presence of 1% lipoprotein-deficient serum (LPDS) medium to avoid inactivation of dynasore by serum proteins. U18666A (Sigma-Aldrich) was used at 3 µg/ml in 1% LPDS medium.

Monocytes were isolated from human blood of normolipidemic donors by ficoll density gradient (Eurobio, les Ulis, France) and subsequently differentiated into human monocytes-derived macrophages (HMDM) by adhesion on plastic Primaria plates (Falcon BD, Franklin Lakes, NJ,) for 7 days in RPM1640 medium, supplemented with 10% heat-inactived FCS, 2 mM glutamine, 100 UI/ml penicillin, 100 µg/ml streptomycin, and 20 ng/ml hM-CSF (human macrophages colony-stimulating factor) (AbCys, Paris, France).

HeLa cells expressing a GFP tagged M6PR were described in ^[12]. Cells were cultured in Dulbecco modified Eagle's medium (DMEM) with 10% heat-inactivated fetal calf serum (FCS) (Invitrogen,), 2 mM glutamine, 100 UI/ml penicillin, 100 µg/ml streptomycin (Sigma-Aldrich,) completed with geneticin. They were treated with dynasore the same way that HeLa cells.

The mouse anti-human Lamp-1 antibody was from BD Transduction Laboratories. Primary antibodies were revealed by incubation with Alexa Fluor®488 donkey anti-mouse IgG (Invitrogen).

Human LDL (d = 1.019–1.063 g/ml) were separated from fresh plasma by sequential ultracentrifugation as described previously ^[9]. HeLa cells deprived of sterols for 48 h in 10% LPDS medium were incubated for the indicated times with LDL (200 µg/ml), supplemented with 50 µM pravastatin and 50 µM sodium mevalonate. HMDM were incubated with 50 or 100 µg/ml of acetylated LDL (AcLDL) prepared using the standard acetic anhydride-sodium acetate solution on ice as described ^[13].

LDL were labeled with DiI (1,1′-dioctadecyl-3,3,3′3′-tetramethyl-indocarbocyanine) as described previously ^[14]. Briefly, DiI were added to LDL to a final ratio of 300 µg for 1 mg LDL protein. After incubation for 18 h at 37°C under nitrogen and light protection, DiI-labeled LDL were isolated by ultracentrifugation and extensively dialyzed at 4°C against PBS (Eurobio). For DiI-AcLDL production, AcLDL were first labeled with DiI and acetylated.

LDL uptake was described in ^[9]. Briefly, cells grown in LPDS medium during 48 h were washed in PBS and incubated at 37°C with increasing concentrations of DiI-LDL (HeLa cells) or DiI-AcLDL (HMDM) as indicated. After 4 h, cells were treated by trypsin to remove cell surface bound fluorescent LDL. Cells were detached and washed twice in PBS at 4°C. The fluorescence of internalized DiI-LDL or DiI-AcLDL was measured by flow cytometry (emission at 585 nm) and expressed as mean fluorescence intensity.

After 48 h of culture in LPDS medium, cells were incubated for 6 h with 200 µg/ml LDL (HeLa cells) or 50 µg/ml AcLDL (HMDM) and then lysed in 0.2 M NaOH. Total cholesterol was extracted with methanol (2.5 ml) followed by hexane (5 ml). About 4.5 ml of the hexane phase was evaporated under vacuum and dissolved in mobile phase. Separation of FC and CE including cholesteryl docohexanoate (CDH), cholesteryl arachidonate (CA), cholesteryl linoleate (CL), cholesteryl myristate (CM), cholesteryl oleate (CO) and cholesteryl stearate (CS) was done by reverse phase HPLC on a C-18 column (25×0.46 cm length, 5-µm pore size, Sigma-Aldrich) by measuring the 205 nm absorbance after elution with acetonitrile/isopropanol (30/70, v/v) ^[15], ^[16]. The amount of proteins present in cell lysates and LDL preparations was measured using the bicinchoninic acid (BCA) method ^[17].

120 µM sodium myristate (Sigma-Aldrich) was added during cholesterol delivery to the cells. The ACAT enzyme activity was evaluated by quantifying the incorporation of myristate into cholesteryl myristate by HPLC as described above. The ACAT inhibitor Sandoz 58-035 (Sigma-Aldrich) (10 µg/ml) was added to the cholesterol loading medium. When ACAT activity is inhibited, CE are provided by pre-existing pools such as endocytosed LDL. Therefore, the difference in cholesterol esterification measured by HPLC with and without Sandoz 58-035 represents the specific amount of cholesterol esterified by ACAT.

After treatment with dynasore or U18666A, HMDM were equilibrated in RPMI1640 0.2% BSA for 16 h in the presence or not of 5 µg/ml 22-hydroxy-cholesterol (22OH-C) and 10 µM 9-cis retinoic acid (9cRA) (Sigma-Aldrich). Cellular cholesterol efflux to either 1 mg/ml methyl-β-cyclodextrin (MβCD), 10 µg/ml lipid-free apoA-I (Sigma-Aldrich) or 15 µg/ml high-density lipoprotein-phospholipids (HDL-PL isolated from normolipidemic human plasma by preparative ultracentrifugation) was measured during a 4 h chase period. Drugs were maintained during the equilibration and efflux periods. At the end of the efflux, the medium was collected and the cells lysed in 0.2 M NaOH. Cell and media were extracted and analyzed for free and esterified cholesterol mass by HPLC as described above. HDL samples were separately analyzed to allow correction for HDL cholesterol present in relevant media samples. Mass cholesterol efflux is expressed as the percentage of efflux (medium cholesterol over total cholesterol-medium and cells) ^[18].

Analysis of LDL-derived cholesterol trafficking was performed by immunofluorescence as described ^[9]. Briefly, HeLa cells and HMDM were grown on cover slips for 48 h in LPDS medium. To monitor LDL trafficking, cells were incubated with fluorescent LDL for indicated times. Cells were incubated with 200 µg/ml LDL or 50 µg/ml AcLDL and then stained with filipin (50 µg/ml) to detect FC as described ^[19]. Cells were fixed in 4% paraformaldehyde (PFA) in PBS. For immunolocalization studies, cells were incubated with primary and secondary antibodies diluted in PBS/BSA with saponin. Cover slips were mounted in MOWIOL, and cells were imaged with an epifluorescent Leica microscope.

Total RNA was extracted from HeLa cells using the RNeasy Mini kit (Qiagen, Courtaboeuf, France). One microgram of total RNA was transcribed to cDNA using random hexamers (Amersham, Orsay, France) and SuperScriptII reverse transcriptase (Invitrogen). Specific primers were described previously ^[9]. Real-time quantitative PCR analyses were performed using the Sybr® Green reagents kit (Applied Biosystems, Courtaboeuf, France) with an ABI PRISM 7900HT Sequence detector instrument (Applied Biosystems) according to the manufacturer's instructions. Amplification was carried out in a final volume of 10 µL with 20 ng of reverse transcribed total RNA, 150 nM (HMGCoAR, LDLR) or 300 nM (UBC and SREBF-2) of both sense and antisense primers (Eurogentec, Liège, Belgium) in the Sybr® Green PCR Master Mix (MESA GREEN qPCR, Eurogentec). The cycling conditions comprised 40 cycles at 95°C for 15 s and 60°C for 1 min. Relative quantification for a given gene, expressed as fold-variation over control at T0, was calculated after normalization to a reference gene (UBC) and determination of the CT (cycle threshold) difference between conditioned cells and control cells at T0 using the comparative CT method ^[20]. Efficiencies of the target and control amplification were very similar.

Non targeting siRNA (si scramble) and siRNA against dynamin2 (si Dyn2) ([gene: GGA CAU GAU CCU GCA GUUdTdT]) were described in ^[21]. Briefly, HeLa cells were transfected using RNAiMAX (Invitrogen) according to the manufacturer's instructions. Knockdown of Dyn2 was observed 48 h post-treatment. Cells transfected with scramble siRNA were used as a control.

Cells were lysed in 1% triton ×100 in PBS containing protease inhibitors. Proteins (30 µg per lane) were separated on a 3–8% NuPAGE in Tris acetate gels (Invitrogen) and transferred to a PDVF membrane (Bio-Rad, Hercules, CA). The membrane was incubated for 1 h in blocking solution (5% nonfat dry milk, 0.1% Tween 20 in PBS). Dynamin 2 was detected using mouse monoclonal anti-dynamin 2 (Invitrogen). A rabbit anti-β-actin polyclonal antibody was used as loading control (Sigma). Development was performed with chemiluminescent detection kit (ECL, Amersham Life Science).

We tested the effect of dynasore on HeLa cells, where dynamin was first reported to control CT trafficking ^[9]. Dynasore was tested at 80 µM, the concentration shown to efficiently block the internalization of the transferrin receptor ^[10]. We measured the uptake of fluorescently labeled human LDL by flow cytometry. Dynasore treatment led to a strong inhibition of LDL uptake at 10 µg/ml with less than 10% of the uptake level measured in untreated cells (Figure 1A). For comparison, the uptake of LDL was decreased by about 50% in our previous study with HeLa cells expressing the K44A inhibitory mutant of dynamin ^[9]. The continuous uptake of LDL for 6 hours in sterol-starved cells led to a 2.5 fold increase of the total amount of cellular cholesterol – free and esterified – as measured by HPLC in control cells. Despite the strong inhibition of LDL uptake, there was still an increase albeit more moderate in dynasore-treated cells (50±15 and 43±8 nmol/mg cell protein, respectively; Figure 1B). This is in agreement with our findings in HeLa cells that cholesterol can enter cells through other dynamin-independent endocytic pathways ^[9]. We next characterized dynasore in HMDM since they play a central role in the formation and progress of atherosclerotic plaques ^[22]. The uptake of AcLDL in HMDM was twice as much less efficient than the uptake of LDL in HeLa cells (Figure 1C). The effect of dynasore was less important in HMDM since the uptake of AcLDL uptake was decreased by 50% at 10 µg/ml and by 17% at 100 µg/ml. However, the absolute amount of endocytosed cholesterol was similar in HeLa cells and HMDM treated by dynasore. In agreement with the lower inhibition of AcLDL uptake, the measure of the total amount of cholesterol revealed no difference between control and dynasore-treated HMDM (Figure 1D).

We examined the intracellular distribution of FC derived from endocytosed LDL using filipin, a naturally fluorescent antibiotic that binds selectively to the FC fraction present in cellular membranes. We also visualized the intracellular distribution of DiI-LDL since we showed previously that DiI-LDL undergo the same endocytic and degradative processes than unlabeled LDL ^[9]. Cells were incubated with LDL for 24 hours to allow their distribution within the different endosomal compartments along the endocytic pathway. In control cells, FC was found into small endosomal structures within the cytoplasm of the cell (Figure 2A). A similar pattern was found for DiI-LDL (Figure 2B). Upon dynasore treatment, we could observe the accumulation of FC and Internalized DiI-LDL into numerous enlarged endosomes (Figures 2 A and B). We also examined the intracellular distribution of DiI-AcLDL in HMDM after 6 hours of internalization. Dynasore treatment led also to an increase of both the number and the size of the endosomal structures loaded with DiI-AcLDL (Figure 2C). This effect was enhanced when cells were first incubated with DiI-AcLDL for 24 hours before the addition of dynasore (Figure 2D). These endosomes were part of the late endosomal network since they were positive for the lysosomal associated membrane protein 1 (Lamp1), a marker of late endosomes and lysosomes (Figures 3 A and B). Thus, dynasore which blocks the GTPase activity of dynamin, causes the accumulation of endocytosed LDL-derived cholesterol in the late endocytic compartment and prevents its egress from this compartment in both HeLa and HMDM cells. This is in agreement with our previous study showing that dynamin controls the delivery of cholesterol from late endosomes to the ER in HeLa cells ^[9]. The abnormal endosomal accumulation of LDL was already observed after 15 min of dynasore treatment indicating that dynasore acts at the endoslysosomal level in the same order of time that it requires to inhibit the uptake of transferrin and LDL at the plasma membrane (data not shown) ^[10].

To confirm the specificity of dynasore treatment, we depleted endogenous dynamin 2, the major dynamin isoform expressed in HeLa cells using specific siRNA. Dyn2 siRNa treatment led to a strong down-expression of dynamin 2 as assessed by real time quantitative RT-PCR (data not shown) and western blot analysis (Figure 4A). Under this treatment, there was an accumulation of swollen endosomal structures loaded with LDL or FC, and positive for Lamp1 (Figure 4 B and C). These results faithfully reproduce the phenotype observed with dynasore treatment and therefore exclude dynasore side effects at the endosomal interface.

The delivery of FC from the late endolysosomal compartment to the ER is a key process in the transcriptional regulation of sterol-sensitive genes ^[1], ^[7], ^[9]. We therefore studied whether dynasore had an impact on this regulation. In contrast to the conditional K44A HeLa cell line that requires 48 hours to express the K44A dynamin mutant ^[9], dynasore is active within a few minutes. We thus determined the minimal amount of time required to measure an effect of dynasore on the expression of the LDLR gene, one of the major actors of the transcriptional control of cholesterol homeostasis ^[2]. After 48 hours of sterol starvation, cells were incubated with LDL in the presence or in the absence of dynasore. The kinetics of expression of the LDLR gene were monitored by real time quantitative RT-PCR analysis in HeLa cells (Figure 5A). As expected in cells that are in excess of exogenous cholesterol, a potent repression of the expression of the LDLR gene was measured as early as 5 hours after the addition of LDL and for as long as 24 hours in control cells. In contrast, no down-expression of the LDLR gene could be measured in cells treated with dynasore at similar times, and LDLR expression levels were similar to those observed in cells not supplemented with LDL. Seven hours after LDL addition, however, the effect of dynasore decreased progressively and the level of LDLR gene expression resumed to control values at 24 hours. Thus, we chose to analyze the effects of dynasore after 6 hours of LDL addition, which corresponds to a strong repression of the sterol-sensitive transcriptional response in control cells and to the maximal inhibition by dynasore. At 6 hours, dynasore also strongly inhibited the LDL-induced repression of another major sterol-sensitive gene, HMGCoAR, whereas the repression on SREBF-2 gene expression was not observed at this time in HeLa cells (Figure 5B). Similar results were obtained in dynasore-treated HMDM, with a strong inhibition of the down-regulation of sterol-sensitive genes (LDLR, HMGCoAR and to a lesser extent, SREBF-2) induced by AcLDL (Figure 5C). These data are in agreement with the accumulation of LDL and AcLDL in enlarged endosomes observed in both cell types (Figure 2) and are likely to reflect a defect in the delivery of LDL-derived cholesterol to the ER.

We next measured the fraction of esterified cholesterol by the ACAT enzyme as a marker of the amount of FC being delivered to the ER. Indeed, CE are generated from FC by the activity of the ACAT enzyme ^[7]. Since this enzyme is strictly localized in the ER membranes, the amount of cholesterol esterified by ACAT reflects the amount of FC delivery to the ER ^[23], ^[24]. Thus, we measured by HPLC the balance between the pools of free and esterified intracellular cholesterol. When HeLa cells were grown under sterol starvation, cholesterol was mainly detected as FC and esters represented less than 5% of total cholesterol (Figure 6A). After the addition of LDL, the total amount of CE represented more than 30% of total cholesterol. Dynasore treatment reduced this amount by about 15%. This moderate inhibition disagrees with the complete absence of sterol-sensitive genes repression in dynasore treated cells as observed above. Thus, we studied whether a fraction of the measured pool of CE may be independent from the ER-ACAT activity. Indeed, dynasore treatment leads to the abnormal endosomal accumulation of LDL, which are unlikely to be de-esterified by the lysosomal hydrolases and thus could contribute to the total intracellular pool of CE. Therefore, we measured the total amount of CE in HeLa cells in which ACAT activity was pharmacologically inhibited. Under this condition, we found that the amount of CE generated by ACAT accounted for only 38% of the total intracellular pool of CE (Figure 6A). When cells were treated with dynasore, this amount decreased to about 10%, which represents a 74% inhibition of LDL-derived cholesterol esterification. We could confirm this result by measuring the synthesis of cholesteryl myristate by ACAT, an ester that was not initially present in our cells. After addition of myristate, we found by HPLC that the production of cholesteryl myristate was decreased by 80% in cells treated with dynasore (Figure 6B).

In HMDM, dynasore treatment decreased the total amount of CE more efficiently than in HeLa cells (Figure 6C). However, the level of CE was initially lower in HMDM than in HeLa cells (11% versus 35%, respectively). This could be due to different kinetics of AcLDL uptake and the lower concentration used (50 µg/ml). Indeed, after 24 hours of incubation with AcLDL, the level of CE in HMDM was higher and represented 40% of total cholesterol (data not shown). Again, dynasore treatment led to a strong decrease (90%) of the amount of CE specifically generated by ACAT (Figure 6C) and of the incorporation of myristate into CE (82% inhibition) (Figure 6D). Together these findings demonstrate that the abnormal endosomal accumulation of LDL induced by dynasore treatment results in a strong defect of FC delivery to the ER.

Although reverse cholesterol transport (RCT) is a general peripheral process, macrophages are the primary cells that are overloaded with cholesterol in atherosclerotic lesions ^[25]. The efficiency of cholesterol efflux, the first step of RCT from macrophage foam cells, is determined not only by the activity of cholesterol transporters and extracellular acceptors, but also by the availability of cholesterol for efflux. It was therefore interesting to test whether the abnormal sequestration of large amounts of cholesterol in the endolysosomal compartment could affect the cholesterol efflux capacity of HMDM. Thus, we measured cholesterol mass efflux to either lipid-free apoA-I or to HDL (high density lipoproteins) in HMDM loaded with AcLDL (Figure 7A). Dynasore treatment resulted in a strong decrease of cholesterol efflux to apoA-I (40%) and HDL (70%). Similar results were observed when cholesterol efflux was stimulated with the natural LXR/RXR agonists, 22OHC/9cRA (Figure 7B). The two ABC (ATP binding cassette) members of cholesterol transporters, ABCA1 and ABCG1 have been involved in cholesterol efflux to lipid poor apoA-I and HDL, respectively ^[25], ^[26]. We thus examined the impact of dynasore treatment on ABCA1 and ABCG1 gene expression. As expected, the addition of AcLDL to HMDM led to an increase of ABCA1 and ABCG1 mRNA levels (1.9-fold and 3.5-fold, respectively). However, this up-regulation was completely abolished when HMDM were treated by dynasore (Figure 7C). Finally, we examined whether dynasore treatment could affect the amount of cholesterol associated with the plasma membrane, which is involved in the passive cholesterol efflux pathway. To address this question, we used MβCD, a high-affinity acceptor of cholesterol. Short time incubation with MβCD was shown to remove cholesterol from the plasma membrane ^[27]. Under these conditions, cholesterol efflux was significantly decreased (34%) in dynasore-treated cells (Figure 7D). Together, these results indicate that dynasore, by affecting intracellular cholesterol trafficking, has a major impact on several pathways implied in active and passive cholesterol efflux.

We tested whether dynasore could affect the trafficking of other cargos through the late endosomal network. We therefore studied the cation-independent mannose 6-phosphate receptor (M6PR), which constitutively recycles between late endosomes and the Golgi complex ^[28]. When HeLa cells expressing a GFP tagged M6PR were treated by dynasore, we observed that the M6PR was not present at the Golgi complex and was localized instead in dispersed endosomal structures (Figure 8). This result, in agreement with a previous study ^[28], indicates that the effect of dynasore was not restricted to the block of FC from the late endosomal network but affected also cargo trafficking at this interface.

These results led us to compare the effects of dynasore with U18666A, a hydrophobic amine that has been extensively used to study cholesterol homeostasis, and more particularly cholesterol intracellular trafficking ^[29]. In contrast to dynasore, we found that U18666A treatment led to a respective 20% and 40% increase of LDL uptake and AcLDL in HeLa cells and HMDM (Figure 9 A and C). Accordingly, we measured an increase in the total amount of cholesterol present in HeLa cells and HMDM after U18666A treatment (Figures 9 B and D). The inhibition of cholesterol egress from late endosomes and lysosomes is a well-known effect of U18666A ^[29], ^[30]. As expected, treatment with U18666A led to the formation of swollen endosomal structures loaded with FC in HeLa cells and HMDM (Figure 10). These structures were part of the endolysosomal network as confirmed by staining with Lamp1 (not shown). As a consequence, there was no response of sterol-sensitive genes to the addition of LDL or AcLDL in U18666A treated cells (Figures 11 A and B). In contrast to cells treated with dynasore, we observed a slight increase in the percentage of CE in U18666A-treated cells (Figures 11 C and D). The measure of CE in cells loaded with cholesterol in the presence the ACAT inhibitor revealed that the fraction of CE specifically generated by ACAT was 54% of the total intracellular pool of CE. When cells were treated with U18666A, the amount of CE generated by ACAT was decreased by 90%. Likewise the incorporation of myristate into CE was strongly inhibited by U18666A (Figure 11 E and F).

Finally, we also measured the effects of U18666A on cholesterol efflux. As shown in Figures 12 A and D, the cellular cholesterol efflux to apoA-I, HDL, and MβCD was decreased by 42%, 58%, and 37%, respectively when macrophages were treated with U18666A. A similar decrease was measured in AcLDL-loaded macrophages after stimulation by the LXR/RXR agonists, 22OHC/9cRA (Figure 12B). Like dynasore, U18666A completely blocked the increase of ABCA1 and ABCG1 mRNA levels in response to the addition of AcLDL in HMDM (Figure 12C).