Annexin VI Stimulates Endocytosis and Is Involved in the Trafficking of Low Density Lipoprotein to the Prelysosomal Compartment*

Annexins are calcium-binding proteins with a wide distribution in most polarized and nonpolarized cells that participate in a variety of membrane-membrane interactions. At the cell surface, annexin VI is thought to remodel the spectrin cytoskeleton to facilitate budding of coated pits. However, annexin VI is also found in late endocytic compartments in a number of cell types, indicating an additional important role at later stages of the endocytic pathway. Therefore overexpression of annexin VI in Chinese hamster ovary cells was used to investigate its possible role in endocytosis and intracellular trafficking of low density lipoprotein (LDL) and transferrin. While overexpression of annexin VI alone did not alter endocytosis and degradation of LDL, coexpression of annexin VI and LDL receptor resulted in an increase in LDL uptake with a concomitant increase of its degradation. Whereas annexin VI showed a wide intracellular distribution in resting Chinese hamster ovary cells, it was mainly found in the endocytic compartment and remained associated with LDL-containing vesicles even at later stages of the endocytic pathway. Thus, data presented in this study suggest that after stimulating endocytosis at the cell surface, annexin VI remains bound to endocytic vesicles to regulate entry of ligands into the prelysosomal compartment.

Annexins are a family of highly conserved proteins, which are characterized by their Ca 2ϩ -dependent binding to phospholipids (1). Each annexin consists of a conserved core domain with four or eight repeats (70 amino acids) and a nonconserved, short, NH 2 -terminal domain. More than 10 different family members, several of which exist as multiple isoforms, have been described in higher vertebrates (2). Since annexins are expressed in many tissues and are located in the same cellular compartments, the understanding of the distinct physiological role of each annexin still remains elusive (1,3). In recent years, the involvement of annexins in membrane traffic has emerged as one of their predominant functions (1,4). Several annexins including annexin I, II, IV, VI, VII, and XIIIb have been directly implicated in different steps of the intracellular trafficking pathways (5)(6)(7)(8)(9)(10)(11)(12)(13)(14) and, despite some controversy, essentially due to the variety of cells and antibodies used, they are all associated with the endocytic compartment.
The enrichment of annexin VI in rat liver endosomes (12,13), its polarized localization in the apical endosomes in rat hepatocytes (14) and WIF-B cells (15), and the colocalization with lgp120, a prelysosomal marker in normal rat kidney cells (15), indicate a potential role for annexin VI in the endocytic pathways of polarized and nonpolarized cells.
In support of this hypothesis, annexin VI has been found to bind ␤-spectrin at the cell surface, which in turn recruits and activates a calpain-like protease. This cascade of events seems to open the actin-cortical cytoskeleton to facilitate the initial steps of endocytosis (16,17). The complexity of these interactions has recently been pointed out by Michaely and co-workers (18) and involves also some other cytoskeleton proteins such as ankyrin that associates with clathrin to participate in the annexin VI-dependent, clathrin-mediated endocytosis.
However, the possible involvement of annexin VI in fusion events in the late endocytic pathway is not completely understood (15,19). The complex intracellular distribution of annexin VI in the various cell types analyzed (1-3, 12-15, 20) indicate an additional role of annexin VI, which may be responsible for the triggering of transient interactions with other structures that may regulate the entrance and the trafficking of different ligands.
In order to investigate the influence of annexin VI on endocytic processes, we studied the effect of overexpression of annexin VI in Chinese hamster ovary (CHO) 1 cells on LDL and transferrin (Tf) uptake and intracellular processing. Here we report that parallel overexpression of annexin VI and the LDL receptor increases the internalization and degradation of LDL. When cotransfected with the Tf receptor, annexin VI moderately stimulates endocytosis but has no effect on the recycling of Tf. In contrast, overexpression of annexin II does not facilitate a stimulatory effect on the endocytosis of both ligands. Cellular subfractionation and immunofluorescence analysis suggest that annexin VI enters the prelysosomal compartment while being associated with LDL-containing endocytic vesicles. These findings indicate that intracellular stores of annexin VI have the ability to respond to so far unknown signals (e.g. ligand, cytokines, calcium) (21) and then concentrate in endosomal vesicles directed to the prelysosomal compartment.
Cell Culture-CHO cells were grown in Ham's F-12 supplemented with 10% fetal calf serum, L-glutamine (2 mM), penicillin (100 units/ml), and streptomycin (100 g/ml) at 37°C, 5% CO 2 . The annexin VI-overexpressing CHO cell line CHOanx6 was grown in the presence of 1 mg/ml G418. 48 h prior to incubation with radioactive ligands, 1 ϫ 10 5 cells/well were plated on 12-well plates. Cells were transfected after 24 h and incubated with the various ligands the following day.
Recombinant DNA-A rat annexin VI cDNA clone was isolated from a rat liver cDNA library (Fisher rats, Uni-Zap TM XR, Stratagene) and subcloned into pGEMT (Promega). The coding region was polymerase chain reaction-amplified and generated a 2.0-kilobase pair annexin VI cDNA fragment (26) that was cloned into a mammalian expression vector (pcDNA3.1ϩ; Invitrogen) to generate pcDNAanx6. DNA sequence analysis using a Dye Terminator Cycle Sequencing reaction kit (PerkinElmer Life Sciences) together with T7, SP6 (Promega), and internal annexin VI primers confirmed the full-length and wild type rat annexin VI cDNA sequence in pcDNAanx6. All cloning procedures were performed according to standard protocols (30). The expression vectors encoding the human LDL receptor (pCMVhLDLR), human Tf receptor (pcD-TFR) (31), and human Rab5 (mRab5pF) (32) were kindly provided by Dr. F. Schnieders and Dr. T. Jentsch (University of Hamburg). The expression vector for human annexin II (pCMV5-EX-AII) (33) was a gift from Dr. V. Gerke. ␤-Galactosidase encoding pCMVSPORT-␤Gal was obtained from Life Technologies, Inc.
Transfections-The DNA used for transfections was purified with the plasmid purification kit from Qiagen. 2-3 ϫ 10 5 cells were transfected with 1 g of DNA and the FUGENE TM 6 Transfection Reagent (Roche Molecular Biochemicals) according to the instructions of the manufacturer. Control cells were transfected with pCMVSPORT-␤Gal. For cotransfections, 0.5 g of each plasmid was used. In these experiments, LDLR-and Tf receptor-transfected cells were cotransfected with ␤-galactosidase-, annexin VI-, annexin II-, or Rab5-encoding plasmids. Expression of ␤-galactosidase in 10 -25% of cells was confirmed by 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) staining, and coexpression of plasmids in transfected cells was confirmed by double immunofluorescence (data not shown).
To generate stable annexin VI-overexpresing cells, 1 ϫ 10 6 CHO cells were transfected with 10 g of pcDNAanx6 and the FUGENE TM 6 Transfection Reagent (Roche Molecular Biochemicals). G418 (1 mg/ml) was added 24 h after transfection, and 2 weeks after transfection G418-resistant colonies were isolated and examined for expression of annexin VI by Western blotting and immunofluorescence.
Immunoblot Analysis-For Western blotting, cells were solubilized in 1% Triton X-100, 50 mM Tris, 2 mM CaCl 2 , and 80 mM NaCl. 50 g of cell protein or 500 ng of purified annexin VI were resolved by 10% SDS-polyacrylamide gel electrophoresis (34) and transferred to Protran R Nitrocellulose membranes (Schleicher and Schü ll). Annexin VI was detected using polyclonal sheep (Abimed) and three different rabbit anti-annexin VI antibodies. Recombinant human LDL receptor expression was analyzed with the rabbit anti-LDLR antibody. Annexin II, Rab5 (data not shown), Rab4, and dynamin expression was confirmed with the antibodies described above. After incubation with peroxidaseconjugated secondary antibodies, the reaction product was finally detected using the ECL system (Amersham Pharmacia Biotech).
Radioactive Labeling of Ligands-LDL was radiolabeled with [ 125 I]Iodine by the iodine monochloride method (35). Transferrin and HRP were iodinated by the IODO-GEN method (36). Reproducibly 93-97% of the radiolabeled LDL was trichloroacetic acid-precipitable. The protein content of LDL in five different labeled preparations was 0.6 Ϯ 0.1 mg/ml, and the specific radioactivity was 180 -300 cpm/ng. The specific radioactivity of Tf in three different preparations was 250 -300 cpm/ng, and the specific radioactivity of HRP was 80 -100 cpm/ng.
Internalization of Ligands and Fluid Phase Markers-To determine the internalization of 125 I-LDL, 2-3 ϫ 10 5 CHO cells were washed with PBS 24 h after transfection and preincubated with serum-free medium (Ham's F-12, 1% BSA) for 30 min at 37°C. Cells were chilled on ice, and 125 I-LDL was added at 4 -5 g/ml (in triplicate). Cells were then incubated for 60 min or 24 h at 37°C. 50-fold excess of unlabeled LDL was added to every third sample prior to the incubation to determine nonspecific uptake of 125 I-LDL. After internalization, cells were washed three times with ice-cold PBS and subsequently with PBS/heparin (20 units/ml) to remove surface-bound LDL. Cells were lysed with 1 ml of 0.1 N NaOH, and the radioactivity was measured. An aliquot of the cell lysate was used to determine the cell protein concentration (37).
For endocytosis of 125 I-Tf and 125 I-HRP, cells were washed with PBS and incubated in Ham's F-12, 1% BSA for 30 min at 37°C. Cells were chilled on ice, and 125 I-Tf (2-3 g/ml) or 125 I-HRP (5 g/ml) was added (in triplicate). A 50-fold excess of unlabeled Tf was added to every third sample. Cells were then incubated at 37°C for various times. To stop internalization, cells were put on ice, and the medium was removed. Cells were washed four times with ice-cold PBS followed by a mild acid wash for 125 I-Tf-incubated cells as described earlier (38) and solubilized in 1 ml of 0.1 N NaOH. The radioactivity and protein concentration were determined.
Analysis of 125 I-LDL Degradation-Cells were incubated with 125 I-LDL as described above for 2, 6, or 24 h at 37°C. At each time point, 100 l of medium was removed, and 3 M ice-cold trichloroacetic acid was added to a final concentration of 10%. The sample was vortexed vigorously and incubated for 10 min on ice. After the addition of 250 l of 0.7 M AgNO 3 to remove free iodine (39), the sample was vortexed again and then centrifuged, and the amount of radioactivity in the remaining supernatant was determined. Cell lysates for protein concentration were prepared as described above.
Analysis of 125 I-Transferrin Recycling-Cells were incubated with 125 I-Tf for 60 min at 37°C as described above. A 50-fold excess of unlabeled Tf was added to every third sample prior to the incubation. After incubation cells were put on ice, the medium was removed, and cells were washed four times with ice-cold PBS. Cells were then incubated in Ham's F-12, 1% BSA at 37°C, and the radioactivity released in the medium was collected at various times. Cells were lysed in 1 ml of 0.1 N NaOH to determine cell protein concentration and the amount of 125 I-Tf remaining in the cells. Immunofluorescence-1 ϫ 10 5 cells were grown on chamberslides (Nunc). 24 h after transfection, cells were washed with cold PBS and fixed in 4% paraformaldehyde. For DiI-LDL uptake experiments, the transfected cells were preincubated with DiI-LDL (5 g/ml) for 30 min at 4°C, washed with PBS and incubated for 5 min at 37°C, fixed with 4% paraformaldehyde, and incubated with antibodies (15). To visualize the primary antibodies, we used immune adsorbed Cy3-or Cy2-conjugated F(abЈ) 2 donkey anti-rabbit, donkey anti-mouse F(abЈ) 2 fragments, or donkey anti-sheep F(abЈ) 2 fragments. Samples were washed extensively with PBS, and finally the chamberslides were mounted with Mowiol®. Confocal laser scanning microscopy was performed using a Leica TCS (Leica Lasertechnik, Heidelberg) instrument based on an inverted Leitz DMIRBE microscope interfaced with an argon-krypton laser adjusted at 488 and 568 nm. The images were converted to TIFF format and processed with Corel Photo-Paint 7.
Subcellular Fractionation-Early endosomes were prepared according to the protocol described by Gorvel et al. (40). Briefly, 4 -6 ϫ 10 7 CHOanx6 cells were used for each gradient. To label early or late endosomes, radiolabeled LDL was endocytosed for 5 or 120 min at 37°C. Cells were washed twice with PBS and then preincubated with Ham's F-12, 10 mM Hepes (pH 7.4) for 5 min at 37°C. The medium was removed, and cells were incubated in Ham's F-12, 10 mM Hepes containing 10 -20 g/ml 125 I-LDL plus 0.5 mg/ml cold LDL for 5 or 120 min at 37°C. Cells were put on ice, washed two times with cold PBS, 0.5% BSA, and collected in homogenization buffer (250 mM sucrose, 3 mM imidazole, pH 7.4, and protease inhibitors). The cells were pelleted, resuspended in 2 ml of homogenization buffer, and homogenized by 10 passages through a 22-gauge needle. Complete homogenization was confirmed under the phase microscope. The homogenate was centrifuged for 15 min at 4°C at 3400 cpm in an Eppendorf centrifuge. The postnuclear supernatant was brought to a final 40.2% sucrose (w/v) concentration by adding 62% sucrose (3 mM imidazole, pH 7.4) to postnuclear supernatant and loaded at the bottom of an SW40 centrifugation tube (Beckman Ultraclear). Then 35% sucrose, 25% sucrose, and finally homogenization buffer were poured stepwise on top of the postnuclear supernatant. The gradient was centrifuged for 90 min at 35,000 rpm, 4°C in a swing out Beckman SW40 rotor. After centrifugation, 1-ml fractions were collected from top to bottom, and the radioactivity was determined. Aliquots of each fraction were assayed for ␤-hexosaminidase activity as described (41). Finally, the samples were trichloroacetic acid-precipitated to determine the distribution of annexin VI, Rab4, and dynamin by Western blotting.

RESULTS
Overexpression and Subcellular Characterization of Annexin VI in CHO Cells-To study the role of annexin VI in the endocytic pathways, rat annexin VI was overexpressed in CHO cells. These cells contain only low amounts of endogenous annexin VI as judged by reverse transcriptase-polymerase chain reaction (data not shown), Western blotting, and immunofluorescence with four different annexin VI antibodies. Fig. 1 shows a representative Western blot demonstrating increased annexin VI expression in transfected cells (Fig. 1, lanes 2 and 5) compared with endogenous annexin VI levels in ␤-galactosidase-transfected controls (lanes 1 and 4; approximately 10 -15fold, compare lanes 4 and 5) using different antibodies. All four antibodies used positively recognize purified bovine liver annexin VI (Fig. 1, lane 3) and rat hepatocytic endosomal annexin VI (data not shown). In all experiments, the efficiency of trans-fection was approximately 10 -25% of the cells. Transient or stable transfected annexin VI overexpressing cells showed morphological (actin-cytoskeleton, data not shown) and functional features (receptor-mediated endocytosis, see below) comparable with the wild type CHO cells.
Annexin VI Does Not Affect the Recycling of Transferrin-To

FIG. 2. Recycling (A) of transferrin and immunofluorescence of annexin VI and Tf receptor (B) in CHO cells.
Analysis is shown of the transferrin cycle in 2-3 ϫ 10 5 CHO cells transfected with ␤-galactosidase (q), Tf receptor and ␤-galactosidase (E), Tf receptor and annexin II (ƒ), Tf receptor and annexin VI (f), and Tf receptor and Rab5 (). A, 24 h after transfection, cells were chilled on ice, and 125 I-Tf (2-3 g/ml Ϯ 50-fold excess cold transferrin) was added. Cells were incubated at 37°C for 60 min. After incubation, cells were put on ice, and the medium was removed. After extensive washing, cells were incubated at 37°C in Ham's F-12, 1% BSA, and the medium was collected after 5, 10, 15, 30, 60, and 120 min. Cells were lysed in 0.1 M NaOH, and the amount of recycled and internalized 125 Tf was measured. Total specific radioactivity was calculated, and the relative amount of recycled 125 I-Tf is given (percentage). Values represent the mean of three independent experiments with triplicate samples. The average S.D. for each experiment was Ϯ0.2 for each time point. B, 1 ϫ 10 5 CHO cells were grown on chamberslides and cotransfected with plasmids encoding rat annexin VI and human Tf receptor. 24 h after transfection, cells were fixed, permeabilized, and double immunolabeled with anti-annexin VI (green) and anti-Tf receptor (red). The superimposed image is shown. The arrows mark colocalization of annexin VI and Tf receptor within the cell (yellow). Primary antibodies were visualized by immunofluorescence as described under "Experimental Procedures." Images correspond to confocal projections (bar, 10 m).  1 and 4) and annexin VI encoding pcDNAanx6 (lanes 2 and 5). Whole cell extracts were prepared 24 h after transfection, and 50 g of cell protein were separated on 10% SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. 500 ng of purified annexin VI from bovine liver was analyzed in parallel (lane 3). Lanes 1-3 were incubated with sheep anti-annexin VI antibody (AB3718). Lanes 4 and 5 were incubated with rabbit anti-glutathione S-transferase-annexin VI. The positions of annexin VI (Anx6) and molecular weight markers are indicated.
investigate the role of annexin VI in the endocytotic pathway of Tf, we determined the uptake and recycling ( Fig. 2A) of radiolabeled Tf in transiently annexin VI-overexpressing CHO cells. Rab5, a member of the Rab protein family that has been demonstrated to stimulate receptor-mediated endocytosis (32,42), served as a positive control. Cells transfected only with annexin VI, annexin II, or Rab5 did not demonstrate a significant alteration of Tf internalization (data not shown). As expected (32,42), cells coexpressing Rab5 together with the Tf receptor endocytosed Tf slightly faster than Tf receptor-expressing controls. Cotransfection of annexin II together with the Tf receptor did not alter endocytosis of transferrin compared with Tf receptor-transfected cells. Coexpression of annexin VI and the Tf receptor revealed a modest stimulatory effect on Tf internalization (1.5-1.8-fold) at the various time points (5-60 min) (data not shown).
Subsequently, the recycling of Tf in controls and annexin VI-transfected cells was examined. Overexpression of annexin VI, annexin II, and Rab5 did not affect Tf recycling (data not shown). As observed by others (32), more than 40% of internalized Tf was recycled in Tf receptor-overexpressing cells after 30 min ( Fig. 2A). In cells cotransfected with annexin VI and Tf receptor, the recycling of Tf was not significantly increased compared with Tf receptor-transfected controls (p Ͼ 0.05 for t ϭ 20, 30 min; Fig. 2A). Similar results were obtained with Rab5 and annexin II, indicating that none of those proteins significantly affected the rates of recycling of transferrin ( Fig. 2A).
In order to identify the intracellular localization of annexin VI and Tf receptor, we then performed immunofluorescence experiments of cells cotransfected with annexin VI and the human transferrin receptor. As expected, overexpression of the Tf receptor shows some localization at the cell surface, as well FIG. 3. Internalization (A and B) of 125 I-LDL and LDLR expression (C) in CHO cells. A, CHO cells transfected with ␤-galactosidase (E), LDL receptor (q), and LDLR together with annexin VI () were chilled on ice, and 125 I-LDL (4 -5 g/ml Ϯ 50-fold excess cold LDL) was added. Cells were incubated at 4°C for 30 min, unbound 125 I-LDL was removed, and cells were incubated for 5-60 times at 37°C. Cells were lysed with 0.1 M NaOH, and the specific radioactivity of solubilized cells was determined. Values were calculated as -fold increase compared with ␤-galactosidase-transfected cells (1.0 at t ϭ 5 min) and represent the mean Ϯ S.D. of one of three representative and independent experiments with triplicate samples. B, CHO cells were cotransfected with increasing amounts (0, 0.4, 0.6, and 0.9 g) of annexin VI and a constant (0.1 g) amount of LDLR (ϩ LDLR)-encoding plasmids. The total amount of DNA was kept constant with ␤-galactosidase. 24 h after transfection, cells were incubated with 125 I-LDL at 37°C for 60 min, and the amount of internalized 125 I-LDL was determined as described above. The data represent the mean Ϯ S.D. of three independent experiments with triplicate samples. Dose-dependent increased annexin VI expression with increasing amounts of transfected annexin VI-encoding plasmid is shown by Western blot analysis. C, whole cell extracts were prepared 24 h after transfection of 2 ϫ 10 5 CHO cells that were transfected with recombinant plasmids encoding ␤-galactosidase (␤-Gal), the LDLR and ␤-galactosidase (LDLR), annexin VI (anx6), and LDLR and annexin VI (LDLR ϩ anx6). 50 g/lane were loaded. The ECL system was used to detect protein expression. The positions of LDLR and molecular weight markers are indicated. as a punctate staining of recycling vesicles with a more intense labeling in the perinuclear Tf recycling compartment (Fig. 2B). In contrast, overexpression of annexin VI results in a wide diffuse staining throughout the cell. Although annexin VI and the Tf receptor partially colocalize within the cell, annexin VI is almost excluded from the recycling compartment of the Tf receptor at the perinuclear region (Fig. 2B). These results correlate with the functional analysis of Tf recycling in these cells ( Fig. 2A) and further suggest that annexin VI is not involved in the trafficking of Tf recycling.
Annexin VI Stimulates Internalization and Degradation of Low Density Lipoprotein-The effect of overexpression of annexin VI on the uptake (Fig. 3) and degradation (Fig. 4) of LDL was studied. We first determined the accumulation of radiolabeled LDL in transiently annexin VI-overexpressing CHO cells (data not shown). Rab5 served again as a positive control (32). Overexpression of annexin VI or Rab5 alone did not result in a significant accumulation of 125 I-LDL compared with the control cells. Therefore, cotransfection of the above proteins together with the human LDL receptor was performed, and the accumulation of 125 I-LDL after 24 h was compared with cells that were transfected with the LDL receptor alone. LDLR overexpression resulted in the 2.4-fold accumulation of LDL, whereas cotransfection of LDLR together with annexin VI or Rab5 resulted in a 2.7-2.8-fold increase of 125 I-LDL accumulation compared with the ␤-galactosidase-expressing controls. We then compared the kinetics of 125 I-LDL internalization for the control (␤-galactosidase) and for cells transfected with the LDLR or cotransfected with annexin VI and LDL-R (Fig. 3A). These results demonstrate increased internalization of 125 I-LDL in cells cotransfected with annexin VI and LDLR compared with LDLR-transfected controls. Furthermore, LDL endocytosis is stimulated in a dose-dependent manner (approximately 1.5-2.0-fold) in cells cotransfected with increasing amounts of an-nexin VI and constant amounts of LDL receptor (Fig. 3B). Subsequent Western blot analysis in transfected CHO cells demonstrated that equivalent amounts of LDL receptor were expressed in cells cotransfected with the LDLR and annexin VI compared with cells transfected with LDLR only (Fig. 3C). Taken together, these results suggest that that annexin VI plays a stimulatory role in LDL receptor-mediated endocytosis.
In addition, overexpression of annexin VI and Rab5 alone did not result in a significant alteration of 125 I-LDL degradation compared with ␤-galactosidase-transfected control (Fig. 4). After 24 h, LDLR overexpression increased degradation of LDL 1.8-fold compared with ␤-galactosidase-expressing control cells (Fig. 4). However, when annexin VI or Rab5 were coexpressed with the LDLR, degradation of 125 I-LDL was stimulated 3-3.4fold after 24 h compared with the control cells (Fig. 4). Taken together, these results indicate that the stimulation of LDL internalization (Fig. 3, A and B) and degradation (Fig. 4) reflects an increased internalization and intracellular processing of LDL, since LDLR expression levels are not elevated in cells cotransfected with annexin VI or Rab5 together with the LDLR compared with cells transfected with LDLR only.
In order to identify specific functions of annexins during the receptor-dependent internalization of ligands, we then compared internalization and degradation of LDL in annexin VIand annexin II-overexpressing CHO cells. Similarly to the experiments described above annexin VI, annexin II, and Rab5 overexpression alone did not result in a significant alteration of 125 I-LDL internalization (data not shown), which correlates with the observation that Rab5 induces internalization of Tf only when coexpressed with the Tf receptor (41). Although the effect of annexin VI on LDL uptake is modest considering the vast overexpression of annexin VI in the transfected cells (see Figs. 2, 5, and 7), coexpression of annexin VI and the LDL receptor significantly stimulated 125 I-LDL internalization by  VI (a-c),  LDL receptor (d-f), and LDLR together with annexin VI (g-i). Immunofluorescence analysis is shown of 1 ϫ 10 5 CHO cells grown on chamberslides and transfected with annexin VI alone (a-c), cotransfected with human LDL receptor and ␤-galactosidase (d-f), and transfected with LDLR together with annexin VI (gi). 24 h after transfection, cells were incubated with DiI-LDL at 4°C for 30 min, and internalization of DiI-LDL was allowed for 5 min at 37°C (b, e, and h). Cells were then fixed, permeabilized, and immunolabeled in green for anti-annexin VI (a and g) and anti-LDLR (d). The superimposed images of both signals (annexin VI and DiI-LDL in c; LDLR and DiI-LDL in f; annexin VI and DiI-LDL in i) are shown (bar, 10 m).
50% compared with cells expressing the LDL receptor alone (p Ͻ 0.0005) in seven independent experiments with triplicate samples. In contrast, when annexin II was coexpressed with the LDL receptor, no significant change in 125 I-LDL endocytosis was observed. In these experiments, coexpression of annexin VI and the LDL receptor increased 125 I-LDL degradation by 70% compared with LDL receptor-expressing cells (p Ͻ 0.003). Coexpression of annexin II together with the LDL receptor did not significantly increase the rate of 125 I-LDL degradation compared with LDLR-transfected controls.
In order to analyze single transfected cells, we incubated CHO cells with DiI-labeled LDL for 5 min at 37°C that were transfected with either annexin VI, LDLR, or LDLR cotransfected with annexin VI (Fig. 5). Consistent with the previous findings annexin VI overexpressing cells did not reveal any enhanced DiI-LDL uptake compared with the neighboring nontransfected cells (Fig. 5, a-c). Similar results were obtained with annexin II-overexpressing cells (data not shown). The expression of the LDLR (Fig. 5, d-f) alone as well as the overexpression of annexin VI together with the LDL receptor ( Fig. 5, g-h) resulted in massive accumulation of DiI-LDL in the transfected cells. These results confirm that LDLR activity was not affected in nontransfected neighboring cells.
Finally, in order to determine a possible stimulatory effect of annexin VI on fluid phase endocytosis cells transfected with annexin VI, annexin II, or Rab5 were incubated with radiolabeled HRP for 120 min at 37°C. In three independent experiments with triplicate samples, the amount of internalized 125 I-HRP was determined. In these experiments, overexpression of annexin VI (31.1 Ϯ 16.7 ng of HRP/mg of cell protein), annexin II (30.3 Ϯ 14.6 ng/mg) or Rab5 (30.7 Ϯ 14.9 ng/mg) did not significantly affect the accumulation of 125 I-HRP compared with ␤-galactosidase-transfected controls (32.2 Ϯ 18.9 ng/mg). Furthermore, cells transfected with LDL receptor alone or cotransfected with annexin VI did not demonstrate increased HRP accumulation when incubated with 125 I-HRP in lipoprotein-deficient medium (data not shown). These results indicate that the increased internalization of 125 I-LDL or 125 I-transferrin in annexin VI and LDL-or transferrin receptor-cotransfected cells (Figs. 2-5) was not due to unspecific internalization mechanisms but rather reflects increased receptor-mediated endocytosis of ligands.
Annexin VI Is Involved in the Trafficking of LDL to the Prelysosomal Compartment-The increased degradation of 125 I-LDL in annexin VI-and LDLR-overexpressing cells (Fig. 4) indicates a potential role of annexin VI at later stages of the endocytic pathway. Due to the bright diffuse staining throughout the cell, it was difficult to characterize the intracellular location of annexin VI overexpression by immunofluorescence. Therefore, the intracellular location of annexin VI was analyzed using subcellular fractionation by sucrose gradients of a stable annexin VI-overexpressing CHO cell line. Fig. 6A shows the distribution of annexin VI in fractions collected from a sucrose gradient (8 -40%, w/v) designed to separate early and late endosomal fractions from plasma membrane and other membranes (40). Annexin VI was detected, by Western blotting, all along the gradient, being most abundant in the bottom region, where heavy membranes (e.g. plasma membranes) are found. Dynamin, a GTP-binding protein that is found at the plasma membrane and in clathrin-coated vesicles and that has recently been described to interact with annexin VI (20), is present predominantly in the bottom fractions of this gradient (Fig. 6A). In contrast, the early endosomal marker Rab4 peaks in the center fractions of the gradient (fractions 5-8). The prelysosomal content of the upper fractions (fractions 3 and 4) was confirmed by their increased ␤-hex- To assess the possible involvement of annexin VI in the endocytic pathway, we induced endocytosis by 125 I-LDL administration for 5 and 120 min (Fig. 6B). The identification of Rab4 and dynamin in the same fractions in control cells (0 min) and after 120-min LDL administration (Fig. 6A) as well as the almost identical profile of ␤-hexosaminidase activity throughout the sucrose gradient at all time points (Fig. 6C) confirmed the stability and reproducibility of this subcellular fractionation. After 5-min 125 I-LDL administration, internalized 125 I-LDL is mainly found in the early endosomal fractions of the gradient (Fig. 6B) Subsequently, 120 min after administration of radiolabeled LDL, 125 I-LDL accumulated in the late endosomal fractions (Fig. 6B, fractions 3 and 4). In parallel, a shift of annexin VI to these prelysosomal fractions (peak at 20 -22% sucrose) was detected (Fig. 6A).
To confirm the biochemical results, we compared the distribution of annexin VI, by confocal microscopy, with Rab4, Rab5, FIG. 6. Subcellular distribution of annexin VI in CHO cells. A, early and late endosomes were separated from plasma membrane containing heavy membranes by sucrose gradient cell fractionation as described (see "Experimental Procedures"). The distribution of annexin VI was studied by Western blotting with anti-annexin VI antibody without or after the administration of 125 I-LDL for 120 min. In the same fractions, under both conditions, the positions of Rab4 and dynamin were analyzed with anti-Rab4 and anti-dynamin antibodies, respectively. B, the intracellular distribution of internalized 125 I-LDL was determined in endosomal fractions of fractionated CHO cells incubated with radiolabeled LDL for 5 min (q) or 120 min (E). C, the activity of ␤-hexosaminidase in each subcellular fraction was determined in cells incubated without LDL (E) and with LDL for 5 min () and 120 min (q). LDLR, and Lamp1 (Fig. 7) in stably transfected CHO cells overexpressing annexin VI. In control cells (0 min) and as mentioned above, the pattern of distribution of annexin VI in CHO cells was largely diffuse throughout the cell (Fig. 7, upper  panel). However, after the administration of LDL (120 min), the staining becomes more punctate (vesicular) and more intense in the perinuclear region of the cells (Fig. 7, lower panel). In contrast, the intracellular distribution of the other proteins Rab4, Rab5, LDLR, and Lamp1 is not affected in such a manner after LDL administration ( Fig. 7; compare upper and lower panels at 0 and 120 min). Thus, overexpression of annexin VI results in a ligand-induced trafficking of annexin VI as a consequence of uptake and trafficking of ligands that use the receptor-mediated endocytosis pathway. DISCUSSION In this study, we have demonstrated that overexpression of annexin VI together with the LDL receptor stimulates endocytosis of low density lipoproteins, whereas fluid phase endocytosis and the recycling of transferrin were not affected in annexin VI-overexpressing cells. After the internalization of LDL, the overall intracellular distribution of annexin VI proteins throughout the cell was shown to redistribute in order to concentrate in late endosomal fractions. These ligand-induced translocations of annexin VI correlate with recent biochemical studies in smooth muscle cells, demonstrating that annexin VI may undergo redistributions between different cellular compartments in a calcium and concentration-dependent manner (43).
Most of our knowledge on annexins in membrane traffic is based on its biochemical identification in isolated membrane fractions, cell-free membrane fusion assays, and colocalization with endocytic markers, but only a limited number of experiments have analyzed the function of annexins using transfection systems (1,3,4). Overexpression of annexin VI or Rab5 alone, which served as a positive control in our experiments, did not alter LDL internalization rates, possibly indicating that the low number of receptors on the cell surface are rate-limiting in these cells. The necessity for high receptor expression in this experimental approach was demonstrated in a number of experiments for Rab5, a member of the Ras-related family of small GTPases, which can stimulate Tf internalization when cotransfected with the Tf receptor (32,42). Since annexin VI does not affect the fluid-phase endocytosis of HRP, it is unlikely that the increase of LDL internalization occurs via non-coatedpit-dependent mechanisms. Therefore, the stimulatory effect on ligand internalization rates in annexin VI-and LDL receptor-overexpressing cells is most likely due to increased internalization and budding of coated vesicles. These results correlate with the partial localization of annexin VI at the plasma membrane in transfected CHO cells, its co-purification with dynamin (20), and the stimulatory effect of purified annexin VI on the detachment of coated pits from purified plasma membranes in vitro (16,17). Although Smythe et al. (44) have questioned a role of annexin VI in endocytosis in human A431 squamous carcinoma cells, these findings could be explained by the development of an annexin VI-independent mechanism to allow coated vesicle formation (17).
In contrast to the stimulatory effect of annexin VI on receptor-mediated endocytosis, annexin VI does not seem to participate in the regulation of Tf receptor recycling. Neither overexpression of annexin VI alone nor cotransfection with the Tf receptor significantly affected the rates of Tf recycling in CHO cells. Furthermore, immunofluorescence analysis demonstrates the absence of annexin VI in the Tf receptor recycling compartment in these cells. Similar negative results have also been described by Smythe et al. (44) and indicate that annexin VI rather plays a specific role in the internalization but is not actively involved in the recycling of endosomes to the cell surface. Consistent with these findings, only 2% of the annexin VI was found in tubular endocytic structures in Lowicryl sections of rat liver. When isolated fractions were examined by electron microscopy, most of annexin VI was identified in the vacuolar structures of the early/sorting endosomes (compartment of uncoupling of receptor and ligands, or CURL) or in the vesicles of receptor-recycling compartment but very little in the tubular extensions where receptors for recycling concentrate (14). In addition, in normal rat kidney cells little colocalization was detected between annexin VI and fluorescein isothiocyanatetransferrin 30 min after internalization (15).
Annexin II, the other member of the annexin family analyzed here, is located at the plasma membrane and in the early endocytic compartment in a number of cell types (1,45,46). It promotes the homotypic fusion of early endosomal membranes in vitro (7) and has therefore been implicated in early endocytic FIG. 7. Immunofluorescence analysis of annexin VI in CHO cells. Immunofluorescence analysis of stably transfected CHO cells overexpressing annexin VI by confocal microscopy. 1 ϫ 10 5 cells grown on chamberslides and incubated with and without LDL (0 and 120 min) were fixed, permeabilized, and immunolabeled in green for anti-annexin VI and in red for anti-Rab4, anti-Rab5, anti-LDLR and anti-Lamp1 as indicated (bar, 10 m).
events. However, upon transient transfection of annexin II constructs into CHO cells, we did not observe any stimulatory effect on the LDL or transferrin endocytosis, even when cotransfected with either the LDL or transferrin receptor. Since annexin II forms a stable heterotetrameric protein complex with p11, a protein of the S-100 family (47), limiting concentration of p11 protein in annexin II-transfected CHO cells could result in low amounts of active annexin II 2 p11 2 complex.
When cotransfected with the LDL receptor, annexin VI also stimulates the degradation of LDL. Sucrose gradient and immunofluorescence analysis indicate that annexin VI is recruited to early endosomes after LDL internalization and most likely remains associated with LDL-containing vesicles entering the prelysosomal compartment. This ligand-induced shift of annexin VI into late endosomal fractions suggests that annexin VI-mediated interactions with cytoskeleton proteins not only participate in the budding of coated pits but also seem to play an important role during the delivery of ligands to lysosomes. In a different experimental approach, microinjection of a dominant negative annexin VI mutant not only reduced endocytosis of LDL but also resulted in the mislocalization of LDL-positive vesicles (17), indicating that transient interactions of annexin VI with the cytoskeleton guide endocytic vesicles along intracellular routes to the prelysosomal compartment. The partial localization of endogenous annexin VI in prelysosomes of normal rat kidney fibroblasts and polarized WIF-B hepatoma cells (15) further supports this observation. At the cell surface, annexin VI is thought to interact with membrane-bound spectrin and calpain to allow budding of clathrin-coated vesicles (17,18,48). In addition, annexin VI has recently been demonstrated to co-immunoprecipitate with dynamin, a GTPase essential for endocytic vesicles pinching off the plasma membrane, in endocytic and transferrin-positive vesicles (20). Therefore, similar annexin VI-dependent mechanisms could play a role in directing ligands from the cell surface to lysosomes.