Effects of cholesterol depletion and increased lipid unsaturation on the properties of endocytic membranes

DAF, decay accelerating factor; DHE, dehydroergosterol; ERC, endocytic recycling compartment; GPI, glycosylphosphatidylinositol; LDL, low density lipoprotein; LE/LY, late endosomes/lysosomes; M β CD, methyl- β -cyclodextrin; SCD1, stearoyl-CoA desaturase1; SE, sorting endosomes; Tf, transferrin; TX-100, Triton X-100. ABSTRACT Lipid analogs with dialkylindocarbocyanine (DiI) headgroups and short or unsaturated hydrocarbon chains (e.g., DiIC12 and FAST DiI) enter the endocytic recycling compartment efficiently, while lipid analogs with long, saturated tails (e.g., DiIC16 and DiIC18) are sorted out of this pathway and targeted to the late endosomes/lysosomes (Mukherjee, S, T.T. Soe, and F.R. Maxfield. 1999. J. Cell Biol . 144: 1271-84). This differential trafficking of lipid analogs with the same polar headgroup was interpreted to result from differential partitioning to different types of domains with varying membrane order and/or curvature. Here, we investigate the system further by monitoring the trafficking behavior of these lipid analogs under conditions that alter domain properties. There was a marked effect of cholesterol depletion on the cell-surface distribution and degree of internalization of the lipid probes. Furthermore, instead of going to the late endosomes/lysosomes as in control cells, long chain DiI analogs, such as DiIC16, were sorted to the recycling pathway in cholesterol-depleted cells. We confirmed that this difference was due to a change in overall membrane properties, and not cholesterol levels per se , by utilizing a CHO cell line that overexpressed transfected stearoyl-CoA desaturase 1, a rate-limiting enzyme in the production of monounsaturated fatty acids. These cells have a decrease in membrane order since they contain a much larger fraction of unsaturated fatty acids. These cells showed alteration of DiI trafficking very similar to cholesterol-depleted cells. Using cold Triton X-100 extractability of different lipids as a criterion to determine the membrane properties of intracellular organelles, we found that the endocytic recycling compartment has abundant detergent-resistant membranes, in contrast to the late endosomes and lysosomes. distinct change in the endocytic trafficking destinations of lipid analogs (as exemplified by the DiI analogs) upon cholesterol depletion. It interesting to note that almost all endocytic trafficking destinations are affected. Specifically, the internalization of disordered domain preferring lipids was most severely affected upon cholesterol depletion. Using CHO-SCD cells, in which the membrane fluidity is altered without a change in cholesterol content, we were able to show that the alteration in trafficking of the lipid analogs was because of changes in membrane order and not specifically due to the chemical entity, cholesterol.


INTRODUCTION
Lipids and proteins associated with the cell surface vary in their lateral and transbilayer distribution, as well as the rate at which they are internalized from the plasma membrane. Once inside the cell, they can potentially be delivered to a large variety of organelles by selective partitioning in a series of sorting steps associated with vesicle or tubule formation (1). Although many specific peptide motifs and protein-protein interactions that determine the distribution and trafficking of transmembrane proteins have been characterized (1,2), the principles underlying lipid sorting and trafficking remain relatively unclear. While the intracellular destinations and sorting decisions for a variety of lipids and lipid analogs have been investigated in the recent years (1,3), a coherent general set of sorting rules for lipids are yet to emerge.
Part of the difficulty in understanding lipid sorting and distribution in the cell arises from the fact that this is the result of a complex interplay between the specific chemistries of individual lipid molecules (e.g., their size, hydrophobicity, headgroup to acyl chain crosssectional ratio, charge on the headgroup, acyl chain unsaturation, etc.) as well as the biophysical properties of the membrane bilayer as a whole (e.g., its composition, thickness, tension, fluidity and curvature) [reviewed in (4,5)]. Adding to the complexity is the fact that a typical biological membrane is composed of hundreds of different lipid classes, with varying permutations of head groups and acyl chains, in addition to varying amounts of rigid, relatively planar structures like cholesterol. In addition, membranes contain a variety of proteins, both transmembrane and  peripheral, which vary in shapes, sizes, charge distribution, and propensity for aggregation among themselves or with other proteins and/or lipids. As a final measure of the complexity, many of these components are distributed non-randomly in the bilayer, varying both in the lateral and the transbilayer dimensions. Biophysical studies in model membrane systems of precise composition have clarified to a great extent, the ways in which classes of lipid molecules interact among themselves and with other classes of lipids or cholesterol (6). However, these studies are carried out in relatively simple, well-defined, two or three component systems. Thus, although there is no doubt that similar principles are at play in various biological membranes, these membranes are too complex to allow a simple extrapolation of the insights obtained from model systems.
In contrast to the plasma membrane, very little is known about the biophysical properties of most intracellular membranes. Although lateral membrane domains or "rafts" have been shown in several studies to exist on the plasma membranes of mammalian cells, whether such domains exist in endocytic organelles as well, and if they do, how their properties compare with the domains on the cell surface remains an open question. Glycolipid and cholesterol enriched 'rafts' have been proposed to play a role in biosynthetic protein and lipid sorting (7). Also, experiments have utilized the ability of BODIPY-labeled lipid analogs to form excimers in a concentration-dependent manner to suggest a redistribution of lipids within seconds after the initiation of endocytosis (8). Lipid rafts have been shown to exist as early in the biosynthetic pathway as the endoplasmic reticulum (7, 9).  In the current study, we have utilized several of fluorescent lipid analogs of the dialkylindocarbocyanine (DiI) series, whose trafficking behavior in normal fibroblasts was investigated previously (1). In the previous study, we found that, in general, lipids that have a propensity to partition into the more disordered lipid domains, or ones with a propensity to enter membranes of concave curvature, trafficked preferentially to the endocytic recycling compartment (ERC), while those with opposite propensities trafficked predominantly to the late endosomes/lysosomes (LE/LY). In order to explain the results of the above studies, we had proposed a working hypothesis based on differential partitioning of endocytosed membraneassociated molecules into coexisting membrane domains, as defined by varying fluidities and/or curvatures.
It is well known that both membrane fluidity (or 'membrane order'), as well as curvature, are strongly modulated by the amount of cholesterol present in the bilayer (10)(11)(12)(13). Thus, in this paper, we test our working hypothesis by following the trafficking of the DiI analogs in cells whose membrane properties have been altered by depleting the amount of cholesterol in the bilayer. In order to ensure that the effect of cholesterol depletion was an overall alteration of membrane structure and dynamics, rather than a specific cholesterol interaction effect, we utilized an alternate method to alter membrane fluidity, without changing its cholesterol content. This was achieved by utilizing a cell line (CHO-SCD1 cells) overexpressing an enzyme, stearoyl-CoA desaturase 1 (SCD1), which alters the amount of monounsaturated fatty acid chains in the membrane lipids (35).
In addition to following the trafficking itineraries of the lipid analogs in this study, we also investigate the membrane properties of intracellular endocytic organelles, namely the ERC and the LE/LY, by monitoring their insolubility in cold Triton X-100 (TX-100). Resistance to cold triton extraction is a criterion often used in cell biology research as phenomenological evidence for molecules residing in membrane domains, termed 'rafts', that are believed to exist in the so-called liquid ordered or L o phase (14,15).
Our interest in this study is to look for any general sorting rules for lipids that may emerge from these comparative studies, and also, to investigate the types of lateral lipid distributions (domains) that may occur in various endocytic organelles. In these studies, we are able to follow the fate of cholesterol directly, by using the fluorescent cholesterol analog, dehydroergosterol (DHE) (16,17). Hao

Materials
All fluorescent probes and anti-Alexa 488 were obtained from Molecular Probes Inc.
(Eugene, OR). Human Tf was obtained from Sigma Chemical Co. (St. Louis, MO). It was iron loaded and passed through a Sephacryl S-300 gel filtration system as described previously (18).
Alexa 488 was then conjugated to the iron-loaded Tf following manufacturers' instructions.
Labeled transferrin was dialyzed thoroughly to remove the unbound dye. Monoclonal antibody against DAF was provided by Dr. S. Tomlinson (Medical University of South Carolina, SC) (19). DHE-loaded MβCD was prepared as described previously (17). DiI-LDL was a gift from Dr. Ira Tabas (Columbia University, NY). Lipid analogs and free fatty acids were transferred as monomers from fatty acid-free BSA carriers (1). All tissue culture supplies were from GIBCO-BRL (Gaithersburg, MD). All other chemicals were from Sigma Chemical Co.

Cells and cell culture
TRVb-1 is a modified CHO cell line that lacks endogenous Tf receptor and expresses the human Tf receptor (20). DAFTb-1 cells are a derivative of TRVb-1 cells. In addition to the human Tf receptors, they also express the GPI-linked DAF (21). They were grown in bicarbonate buffered Hams F-12 medium supplemented with 5% fetal bovine serum (FBS), 100 U/ml penicillin, 100 µg/ml streptomycin and 200 µg/ml geneticin as a selection for the Hao et  10% FBS, 100 U/ml penicillin, 100 µg/ml streptomycin. 200 U/ml hygromycin was used as a selection for the transfected SCD1. All cells were kept in a 5% CO 2 environment in humidified incubators at 37°C. Cells for microscopy were grown for 2 days on 35 mm plastic tissue culture dishes whose bottoms were replaced with poly-D-lysine coated coverslips (22). All experimental manipulations as well as microscopy were carried out in these dishes.

Cholesterol depletion
Metabolic depletion -TRVb-1 cells were grown for 2 days in metabolic depletion medium [Hams F-12 medium similar to the growth medium but with 5% lipoprotein-deficient serum in place of FBS, supplemented with 200 µM mevalonate, and 10 µM mevastatin (23)] to block cholesterol synthesis and deplete cholesterol stores (24).  Fluorescence cross-over was measured using single-labeled samples of each probe, and images were corrected for background (26) and cross-over (1).

Confocal Microscopy
Confocal microscopy was performed using an Axiovert 100M inverted microscope

Cholesterol depletion interferes with the normal trafficking of the DiI analogs
Endocytic fates of the DiI derivatives were determined after their initial incorporation in the plasma membrane. Fluorescent transferrin (Tf), bound to its receptor (TfR), was used as a marker for the ERC (27,28). Exit from the ERC is the slowest step in the endocytic recycling itinerary of the TfR, resulting in the ERC being the most brightly labeled structure at steady state (18). It has been shown previously (29)   To be sure that the change in intracellular trafficking upon cholesterol depletion was related to the order preferences of the hydrocarbon tails, we confirmed these results with DiIC 18 , which has a higher affinity for the ordered domains than DiIC 16 (30). Cells labeled with DiIC 16 and DiIC 18 behaved identically, both under normal growth conditions, and upon cholesterol depletion (data not shown).
In order to avoid possible artifacts, cholesterol depletion was carried out in two different ways. The approach used for the cells shown in Figure 1 was to treat cells grown in normal medium with methyl-β-cyclodextrin (MβCD), an efficient cholesterol chelator known to form soluble inclusion complexes with cholesterol (31). In CHO cells, incubation with 10 mM MβCD for 30 min or longer reduced total cellular cholesterol by 40-50% (29,32). The other method to reduce cholesterol was metabolic depletion, in which the cells were grown for two days in medium containing lipid-depleted serum, an inhibitor of cholesterol biosynthesis (mevastatin) and supplemented with low levels of mevalonate in order to maintain basic metabolism of other essential isoprenoids (23,24). Both methods produced similar degrees of cholesterol depletion In CHO cells, the ERC is a collection of narrow tubular elements that appears as an unresolved central spot by epifluorescence microscopy (Fig. 1). We used confocal scanning  especially in the FAST DiI (Fig. 3D), and these patches on the plasma membrane were relatively stable over time (at least up to 20 min). We have previously reported patching of DiIC 12 and C 6 -NBD-SM into the more disordered regions of the plasma membrane upon cholesterol depletion (32). To verify that the patches of FAST DiI were indeed on the plasma membrane, we utilized a confocal microscope to optically section through a cholesterol-depleted cell labeled with FAST DiI (Fig. 4A-D). At each section, the fluorescent patches were seen at or near the cell border, consistent with patching on the cell surface. Very little FAST DiI was seen inside the cells.
The effect of cholesterol depletion on the DiI trafficking is reversible (Fig. 4 E,F). Cells were first incubated with MβCD for 1 hr. They were then labeled with FAST DiI for 2 min, chased for 30 min in either Medium 1 (Fig. 4E) or Medium 1 supplemented with cholesterolloaded MβCD (Fig. 4F). Repletion of cholesterol with an exogenous source resulted in a uniform plasma membrane distribution and intracellular accumulation for FAST DiI, demonstrating that the change in FAST DiI distribution upon cholesterol depletion treatment was truly an effect of reduction in cholesterol levels.

Quantification of the overlap of the distribution of different DiI analogs relative to Tf
Although Fig. 1 shows the redirection of DiIC 16 trafficking from the LE/LY to the ERC visually, we quantified this redirection to give statistical validity to our results. Fig 5A shows

Retention of disorder-preferring lipids on the plasma membrane of cholesterol depleted cells is due to impaired internalization
In the pulse-chase protocols used to examine lipid trafficking, the higher surface labeling with FAST DiI and DiIC 12 in cholesterol depleted cells could have been a consequence of either reduced internalization or increased recycling. In order to distinguish between these possibilities, we required a lipid analog that could be easily stripped from the plasma membrane. For this, we chose C 6 -NBD-SM, which has very similar membrane partitioning behavior and trafficking pattern as DiIC 12 and FAST DiI. We labeled the cells with C 6 -NBD-SM for 2 min at 37°C and then immediately back-exchanged on ice with defatted BSA to remove cell-surface probes (33,34). Fig. 6 shows intracellular accumulation of C 6 -NBD-SM after 2 minutes of internalization in normal and cholesterol depleted cells. It can be seen that C 6 -NBD-SM uptake  was severely impeded in cholesterol depleted cells. Fig 6G shows that this decrease in internalization upon cholesterol depletion is not accounted for by a reduction in incorporation of C 6 -NBD-SM into the plasma membrane. Upon cholesterol depletion, whereas the total amount of cell-associated C 6 -NBD-SM is reduced only marginally (~74% of normal), the amount internalized is reduced to ~17% of normal.

DiIC 16 trafficking is altered in CHO-SCD cells
SCD1 is a membrane-bound, rate-limiting enzyme in the biosynthesis of monounsaturated fatty acids. Since membrane fluidity is controlled in part by the ratio of saturated to unsaturated fatty acid chains in lipids, SCD1 plays an important role in regulating membrane biophysical properties (35). We wanted to confirm that the effects we were seeing on the trafficking of lipid analogs upon cholesterol depletion were truly due to a change in membrane biophysical (fluidity) properties and not due to a specific interaction with cholesterol itself. For this, we utilized a CHO cell line that overexpressed the SCD1 gene. CHO-SCD cell membranes have a dramatically different lipid acyl chain composition compared to their parental CHO cells. In particular, the SCD1 expressing cells showed a 71% increase in 18:1 (unsaturated) to 18:0 (saturated) fatty acid ratio in plasma membrane compared to the parental cells (36).
Interestingly, there were no significant differences between the parental CHO cells and the CHO-SCD cells, in terms of the amounts of total cholesterol, free cholesterol, cholesterol esters,  and phosphatidylcholine, the major phospholipid class in these cells (35). In this way, we were able to manipulate the membrane properties of the early endocytic pathway without directly altering cholesterol levels.
We examined DiIC 16 trafficking in CHO-SCD cells. Instead of being delivered to the LE/LY as in parental CHO cells (Fig. 7B), the majority of DiIC 16 was found in the ERC of SCD1 overexpressing cells (panels D-F). This alteration in DiIC 16 trafficking was very similar to that observed in cholesterol depleted cells (Fig. 1C-D). Other endocytic routes were not affected in CHO-SCD cells, as shown by normal delivery of DiIC 12 and Tf to the ERC (panels G-I) and LDL to the LE/LY (panels J-L). Interestingly, although the trafficking of DiIC 16 was changed in CHO-SCD cells, Tf internalization was relatively normal (Fig. 7F, I, L), suggesting that SCD1 overexpression did not affect clathrin-mediated endocytosis to the same extent as was seen in cholesterol depleted cells.

Altered DiIC 16 trafficking in cholesterol depleted or CHO-SCD cells are not reversed by addition of exogenous free long chain saturated fatty acids
We attempted to determine whether the DiI trafficking changes upon cholesterol depletion or in CHO-SCD cells could be reversed by increasing the membrane order, using an approach that did not involve adding back cholesterol to cells. For this, we chose long chain

Cold TX-100 solubility shows different profiles in the intracellular compartments
The above results show that modulation of membrane biophysical properties affects the   (16,17,39). On the plasma membranes of CHO cells expressing DAF and transmembrane human TfR (DAFTb-1 cells), both DAF and DHE have been shown to be resistant to cold TX-100 solubilization, where TfR is almost completely solubilized (21,32). All three molecules, Tf (bound to the TfR), DAF, and DHE, accumulate in the ERC at steady state (Fig. 8A, B and E, F) (17,23). Fig. 8G shows that after extraction with cold TX-100, Tf was completely extracted from the ERC. In contrast, a large fraction of DAF and DHE molecules in the ERC were retained after treatment with the detergent (panels C, D, H), similar to the detergent resistance in the plasma membrane.  In panels I and J, we determined whether DiIC 16 , which has been shown to be retained on the plasma membrane after cold TX-100 extraction (32), is also retained in its final endocytic Our results with cold Triton X-100 extraction of labeled cells showed that the ERC not only contains high amounts of cholesterol (16,17) but also forms "raft-like" domains, as defined by resistance to cold TX-100 extraction. The LE/LY, by contrast, do not contain these rafts, which is not surprising, since they do not contain much cholesterol (16,17,40).

Hypotheses for differential sorting at multiple endocytic destinations
Sorting at the plasma membrane: The plasma membranes of cells growing in normal media were always found to be In this paper, when we examine the internalization profiles of these phase-separated lipid analogs, we find that the ordered domain preferring lipid, DiIC 16 , appears to enter the cells quite efficiently, whereas the disordered domain preferring lipids, such as DiIC 12 or FAST DiI, are almost completely prevented from entering the cell when cellular cholesterol is reduced. Such prevention of internalization has previously been reported for some transmembrane proteins such as the TfR (29,41), whereas the internalization of ricin (41) appears to continue relatively unhindered. We do not understand the basis for this difference in internalization rates.
Sorting in the sorting endosome: We have shown previously that lipid analogs, varying solely in the length and degree of unsaturation of their hydrophobic tails, are delivered to different intracellular destinations following internalization by endocytosis (1). More precisely, lipids containing either short or unsaturated hydrocarbon chains (DiIC 12 and FAST DiI), which partition into the more disordered parts of the membrane bilayer, efficiently entered the ERC. In contrast, lipid analogs with long and saturated tails (DiIC 16 and DiIC 18 ), which preferentially enter the more ordered domains, were sorted out of this pathway and were instead targeted to LE/LY. Similar behavior is observed for short chain lipid analogs such as C 6 -NBD-SM, which recycle efficiently upon internalization (33,34), whereas some lipid analogs containing long, saturated acyl chains (e.g.,  Rh-PE and some glycosphingolipids) (42,43) are targeted to late endocytic compartments.
This differential trafficking can be rationalized by the propensities of the lipid analogs to partition into membrane domains of varying fluidity and/or curvature (1). Briefly, this hypothesis states that a membrane-bound molecule can be effectively excluded from the recycling pathway