Identification and characterization of the acidic pH binding sites for growth regulatory ligands of low density lipoprotein receptor-related protein-1.

The type V TGF-beta receptor (TbetaR-V) plays an important role in growth inhibition by IGFBP-3 and TGF-beta in responsive cells. Unexpectedly, TbetaR-V was recently found to be identical to the LRP-1/alpha(2)M receptor; this has disclosed previously unreported growth regulatory functions of LRP-1. Here we demonstrate that, in addition to expressing LRP-1, all cells examined exhibit low affinity but high density acidic pH binding sites for LRP-1 growth regulatory ligands (TGF-beta(1), IGFBP-3, and alpha(2)M(*)). These sites, like LRP-1, are sensitive to receptor-associated protein and calcium depletion but, unlike LRP-1, are also sensitive to chondroitin sulfate and heparin and capable of directly binding ligands, which do not bind to LRP-1. Annexin VI has been identified as a major membrane-associated protein capable of directly binding alpha(2)M(*) at acidic pH. This is evidenced by: 1) structural and Western blot analyses of the protein purified from bovine liver plasma membranes by alpha(2)M(*) affinity column chromatography at acidic pH, and 2) dot blot analysis of the interaction of annexin VI and (125)I-alpha(2)M(*). Cell surface annexin VI is involved in (125)I-TGF-beta(1) and (125)I-alpha(2)M(*) binding to the acidic pH binding sites and (125)I-alpha(2)M(*) binding to LRP-1 at neutral pH as demonstrated by the sensitivity of cells to pretreatment with anti-annexin VI IgG. Cell surface annexin VI is also capable of mediating internalization and degradation of cell surface-bound (125)I-TGF-beta(1) and (125)I-alpha(2)M(*) at pH 6 and of forming ternary complexes with (125)I-alpha(2)M(*) and LRP-1 at neutral pH as demonstrated by co-immunoprecipitation. Trifluoperazine and fluphenazine, which inhibit ligand binding to the acidic pH binding sites, block degradation after internalization of cell surface-bound (125)I-TGF-beta(1) or (125)I-alpha(2)M(*). These results suggest that cell surface annexin VI may function as an acidic pH binding site or receptor and may also function as a co-receptor with LRP-1 at neutral pH.

The type V TGF-␤ receptor (T␤R-V) 1 is a high molecular weight glycoprotein receptor, which co-expresses with type I, type II, and type III TGF-␤ receptors (T␤R-I, T␤R-II, T␤R-III) in most cell types (1)(2)(3). Many carcinoma cells express little or no T␤R-V (3,4). Their growth is not inhibited by TGF-␤ 1 , suggesting that T␤R-V may be involved in the growth inhibitory response to TGF-␤ 1 and that loss of T␤R-V may contribute to the malignant phenotype (3)(4)(5). Identification of T␤R-V as the insulin-like growth factor-binding protein-3 (IGFBP-3), which mediates the IGF-independent growth inhibitory response upon IGFBP-3 stimulation of responsive cells, has highlighted the likely importance of T␤R-V in mediating the growth inhibitory response to TGF-␤ 1 (5,6). Structural and functional analysis of T␤R-V seems to be required for elucidating the molecular mechanisms by which TGF-␤ 1 and IGFBP-3 induce growth inhibition in responsive cells (5)(6)(7). Unexpectedly, we recently found that the T␤R-V is identical to the low density lipoprotein receptor-related protein-1/activated ␣ 2 M receptor (LRP-1/␣ 2 M receptor) as determined by structural and functional analyses of purified T␤R-V (8). Genetic evidence and evidence of rescue experiments has revealed that T␤R-V/LRP-1 is required for cellular growth inhibition caused by IGFBP-3 and TGF-␤ 1 (8).
LRP-1 was identified by Herz et al. (9) through homologous sequence screening. It is best known as an endocytic receptor and has been shown to bind multiple ligands with distinct structures and to mediate their plasma clearance and cellular catabolism (10 -12). One of the prominent ligands is the activated form of ␣ 2 -macroglobulin (␣ 2 M*) (13). Hepatic LRP-1 appears to be responsible for plasma clearance of ␣ 2 M*. Recently, LRP-1 has been reported to mediate signaling in several cell types upon binding by ␣ 2 M*; ␣ 2 M* has also been shown to regulate cell growth of certain cell types (14 -17). The cytoplasmic domain of LRP-1 has also been shown to modulate the MEKK/JNK/c-Jun signaling cascade (18). The finding that T␤R-V is identical to LRP-1 has provided new insights into the role of LRP-1 in mediating growth regulatory signaling. Since the growth regulatory functions of IGFBP-3 and TGF-␤ 1 are biologically important (8), we investigated the interactions of the newly identified growth regulatory ligands for LRP-1 with LRP-1 in cells. We predicted that these investigations would help to disclose mechanisms by which LRP-1 regulates cellular proliferation upon stimulation by IGFBP-3 and TGF-␤ 1 . During these investigations, we discovered that all cell types examined exhibited low affinity but high density acidic pH binding sites for LRP-1 growth regulatory ligands and that these differ from LRP-1 (which has ligand binding activity with a neutral pH optimum). Since these acidic pH binding sites are likely to modulate ligand binding to LRP-1 at neutral pH, we characterized these acidic pH binding sites in responsive cells. In this communication, we demonstrate that the acidic pH binding sites share some properties (receptor-associated protein-and calcium depletion-sensitive ligand binding) with LRP-1 but differ from LRP-1 in their sensitivity to heparin and chondroitin sulfate and in their ability to bind non-LRP-1 ligands. We also show that annexin VI is the major protein purified from bovine liver plasma membranes by ␣ 2 M* affinity column chromatography at acidic pH and that annexin VI directly interacts with ␣ 2 M* in a pH-dependent manner with an optimum pH of 5 as demonstrated by dot blot analysis. We further demonstrate that cell surface annexin VI is involved in the binding of 125 I-TGF-␤ 1 or 125 I-␣ 2 M* at acidic pH. Cell surface annexin VI is also capable of mediating internalization and degradation of cell surface-bound 125 I-TGF-␤ 1 or 125 I-␣ 2 M* at acidic pH. Additionally, it is involved in the binding of 125 I-␣ 2 M* to LRP-1 at neutral pH and forms ternary complexes with 125 I-␣ 2 M* and LRP-1 at neutral pH as demonstrated by co-immunoprecipitation. Finally, we show that trifluoperazine and fluphenazine both inhibit binding of 125 I-TGF-␤ 1 or 125 I-␣ 2 M* to these acidic pH binding sites and to LRP-1 at neutral pH, and block cellular degradation after internalization of cell surface-bound 125 I-TGF-␤ 1 or 125 I-␣ 2 M*.
Internalization and Degradation of Cell Surface-bound 125 I-TGF-␤ 1 or 125 I-␣ 2 M*-Cells (8 ϫ 10 4 cells/well) in 48-well clustered dishes were incubated with 125 I-TGF-␤ 1 (100 pM) or 125 I-␣ 2 M* (2 nM) with or without 10 M trifluoperazine, fluphenazine, or promethazine in the presence and absence of 200-fold excess of unlabeled TGF-␤ 1 or ␣ 2 M* (to estimate nonspecific binding) in DMEM/25 mM HEPES, pH 7.4 containing BSA (1 mg/ml). After 2 h at 0°C, the cells were washed and incubated with DMEM/25 mM HEPES, pH 7.4 containing BSA (1 mg/ml) with or without 10 M trifluoperazine, fluphenazine or promethazine. After 1 h at 37°C, the medium was collected and precipitated with 10% trichloroacetic acid. The trichloroacetic acid-soluble radioactive material in the medium represented the cellular degradation products of 125 I-TGF-␤ 1 or 125 I-␣ 2 M*. The cells were then treated with trypsin (5 mg/ml), maintained for 20 min at 0°C, and centrifuged. The radioactivity in the supernatant and cell pellets represented cell surface-bound and internalized 125 I-TGF-␤ 1 or 125 I-␣ 2 M*, respectively. The experiments were performed in quadruplicate.
Cell Surface Localization of Annexin VI-Cells grown on coverslips in DMEM/25 mM HEPES, pH 7.4 containing 10% fetal calf serum were fixed with 3.7% formaldehyde in DMEM/25 mM HEPES, pH 7.4 (ice cold). After 1 h, the fixed cells were washed with DMEM/25 mM HEPES, pH 7.4 (ice cold). The coverslips were then blocked with BSA (5 mg/ml) in DMEM/25 mM HEPES, pH 7.4 on ice overnight. After washing with DMEM/25 mM HEPES, pH 7.4, the fixed cells were treated with antiannexin VI IgG or control IgG (1:75 dilution) in DMEM/25 mM HEPES, pH 7.4 containing BSA (5 mg/ml) at room temperature for 2 h. After washing, fixed cells were incubated with anti-rabbit IgG-fluorescein isothiocyanate conjugate (1:50 dilution) at room temperature for 1.5 h and then washed twice with ice-cold phosphate-buffered saline prior to visualize with a confocal fluorescent microscope.
Affinity Column Chromatography on ␣ 2 M*-Sepharose 4B-Bovine liver plasma membranes were subjected to Triton X-100 extraction according to published procedures (2), except that 50 mM HEPES/ acetate buffer, pH 6 (or pH 5), containing 0.15 M NaCl and 4 mM CaCl 2 was used. The Triton X-100 extracts were applied onto a column of ␣ 2 M*-Sepharose 4B (1.6 ϫ 20 cm) in 50 mM HEPES/acetate buffer, pH 6.0 (or pH 5.0), 0.15 M NaCl, and 0.1% Triton X-100 (HEPES/acetate buffer) containing 4 mM Ca 2ϩ . After extensive washing with HEPES/ acetate buffer containing 4 mM CaCl 2 , the column was eluted with 10 mM EDTA in HEPES/acetate buffer. The fractional volume of the eluents was 1 ml. An aliquot of fractions (EDTA eluents) was subjected to 7.5% SDS-PAGE under non-reducing and reducing conditions and silver staining. The concentrated flow-through fractions and peak fraction were analyzed by Western blot analysis using anti-annexin VI IgG (8).
MALDI-TOF Analysis-A 68-kDa protein purified from ␣ 2 M*-Sepharose 4B affinity column chromatography was subjected to 7.5% SDS-PAGE under reducing conditions, stained with Coomassie Blue, and digested with trypsin. MALDI-TOF analysis of the tryptic digests was carried out at Applied Biosystems (Foster City, CA).
Dot Blot Analysis-1 l of annexin I, II, III, IV, V, and VI (0.1 mg/ml in H 2 O) was applied onto nitrocellulose membranes and air-dried. The membranes were blocked with 5% BSA in phosphate-buffered saline. After 2 h at room temperature, the membranes were incubated overnight with 125 I-␣ 2 M* (10 nM) in 50 mM HEPES/acetate at pH 4, 5, 6 and 7 containing BSA (2 mg/ml), 2 mM Ca 2ϩ , and 0.15 M NaCl in the presence and absence of RAP (100 g/ml) or BAPTA (5 mM). The membranes were extensively washed with the respective binding buffers containing BSA (5 mg/ml) and analyzed by autoradiography.
Anti-annexin VI IgG Treatment of Cells-Mv1Lu, MEF, and PEA-13 cells (8 ϫ 10 4 cells/well) grown on 48-well clustered dishes were treated with various concentrations (0, 7.5, 15, and 30 g/ml) of anti-annexin VI IgG or control IgG in DMEM/25 mM HEPES/acetate buffer at pH 6.4 or 7.4 containing BSA (1 mg/ml) at 37°C for 2 h. The treated cells were then kept on ice and additional anti-annexin VI IgG or control IgG in the same DMEM/HEPES/acetate buffer was added to wells (0, 7.5, 15, and 30 g/ml). The binding assay at pH 6.4 or 7.4 (at 0°C) was started by adding 125 I-TGF-␤ 1 (100 pM) or 125 I-␣ 2 M* (1 nM) with or without 200-fold excess of unlabeled TGF-␤ 1 or ␣ 2 M* (to estimate nonspecific binding) to wells. After 2 h at 0°C, the cell-associated 125 I-TGF-␤ 1 or 125 I-␣ 2 M* was determined. The specific binding of 125 I-TGF-␤ 1 or 125 I-␣ 2 M* was estimated by subtracting nonspecific binding from total binding. The experiments were performed in duplicate.

LRP-1 Ligands Exhibit High Capacity Binding to Cells at
Acidic pH-LRP-1 is a 600-kDa type I membrane glycoprotein which has been shown to bind more than 35 ligands (10 -12). These ligands have different structures and functions but share two receptor binding properties: 1) Ca 2ϩ dependence and 2) inhibition by RAP. Since IGFBP-3 and TGF-␤ 1 are the newly identified ligands for LRP-1, we determined whether they share these two receptor binding properties with ␣ 2 M* in Mv1Lu cells. Mv1Lu cells are a standard model cell system for investigating TGF-␤ 1 and IGFBP-3 receptors and activities (4,5,(27)(28)(29). These cells were incubated with 6 nM 125 I-IGFBP-3, 1 nM 125 I-TGF-␤ 1 , or 10 nM 125 I-␣ 2 M* in the presence or absence of GST-RAP (15 g/ml), EGTA or BAPTA (5 mM), or 200-fold excess of unlabeled IGFBP-3, TGF-␤ 1 or ␣ 2 M* at pH 4, 5, 6, 7.4 (or 7.0), and 8.0. After 2.5 h at 0°C, the specific binding of 125 I-IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M* was determined. GST-RAP is a fusion protein of glutathione S-transferase and RAP, which inhibits binding of all known ligands to LRP-1 (9,10,19,30,31). The tetrasodium salt (but not the free acid form) of EGTA functions well as a chelator of Ca 2ϩ at acidic pH. BAPTA is a Ca 2ϩ chelator independent of pH (26). As shown in Fig. 1, A-C, 125 I-IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M* bound to Mv1Lu cells in a pH-dependent manner. The specific binding (GST-RAP-sensitive) of 125 I-IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M* exhibited a maximum at pH 5. The specific binding (GST-RAP-sensitive) of 125 I-IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M* at pH 7.4 was much less than at pH 5 ( Fig. 1, A-C). The pH profiles of the EDTA-or BAPTA-sensitive binding for these radioactive ligands were similar to those of the GST-RAP sensitive binding (data not shown). The apparent K d values for binding of IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M* to T␤R-V/ LRP-1 at pH 7.4 are known to be 6 nM, 50-400 pM, and 75 pM, respectively (7,8,32). These results suggest that Mv1Lu cells may possess low affinity, high density binding sites (which are GST-RAP-and Ca 2ϩ depletion-sensitive) for 125 I-IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M* with acidic pH optima. This suggestion is supported by Scatchard plot analysis of 125 I-␣ 2 M* binding to Mv1Lu cells at pH 5. As shown in Fig. 2A, 125 I-␣ 2 M* bound to Mv1Lu cells in a concentration-dependent manner at pH 5 with a saturating concentration of 120 nM. Scatchard plot analysis of the binding data revealed a single class of low affinity binding sites with an apparent K d of 54 nM and 1.5 ϫ 10 6 sites/cell (Fig. 2B). The K d values of the low affinity acidic pH binding sites for 125 I-IGFBP-3 and 125 I-TGF-␤ 1 were not determined. However, it is very possible that the K d values of the low affinity acidic pH binding sites for 125 I-IGFBP-3 and 125 I-TGF-␤ 1 are similar to the apparent K d of the acidic pH binding sites for 125 I-␣ 2 M*. We therefore focused on characterizing 125 I-TGF-␤ 1 and 125 I-␣ 2 M* binding to the acidic pH binding sites in all of the following experiments.
Since binding of LRP-1 ligands ( 125 I-IGFBP-3, 125 I-TGF-␤ 1 , and 125 I-␣ 2 M*) to the acidic pH binding sites requires the presence of Ca 2ϩ and is sensitive to GST-RAP, we suspected that LRP-1 itself might mediate binding. LRP-1 is known to have ligand binding activity with a neutral pH optimum. To exclude this possibility, we performed 125 I-TGF-␤ 1 or 125 I-␣ 2 M* binding at a varying pH using mouse embryonic fibroblasts (MEF) and LRP-1-deficient mouse embryonic fibroblasts (PEA-13 cells). As shown in Fig. 1, D and E, binding in both MEF and PEA-13 cells was maximal at pH 5, suggesting that such binding is mediated by a protein(s) other than LRP-1. These results are also consistent with the notion that LRP-1 does not have significant ligand binding activity at acidic pH and that LRP-1 unloads bound ligands in endosomes (due to acidic pH-induced ligand dissociation) following internalization.
The Acidic pH Binding Sites in Cells Have Broad Ligand Specificity-␣ 2 M is a plasma protease inhibitor that inhibits all four classes of proteases. ␣ 2 M is cleaved by the protease in the bait region (which is close to the thioester bond in the ␣ 2 M three-dimensional structure) and undergoes conformational changes, resulting in trapping of the protease (21)(22)(23)(24)(25). The protease-activated ␣ 2 M is termed ␣ 2 M*. ␣ 2 M* can be mimicked by methylamine-treated ␣ 2 M since methylamine also cleaves the same thioester bond, thus inducing the same conformational changes in ␣ 2 M (21-25). LRP-1 has been shown to bind ␣ 2 M* and native ␣ 2 M with different affinities (K d : 40-75 pM and 2 nM, respectively) (10 -14). To see if, like LRP-1, the acidic pH binding sites have different affinities for ␣ 2 M* and native ␣ 2 M, we determined the effects of increasing concentrations of unlabeled ␣ 2 M* and native ␣ 2 M on 125 Table I). We also determined the effects of increasing concentrations of unlabeled ␣ 2 M* and native ␣ 2 M on 125 I-␣ 2 M* binding to PEA-13 cells which are known to be deficient in LRP-1. As shown in Fig. 3B, unlabeled ␣ 2 M* and native ␣ 2 M also inhibited 125 I-␣ 2 M* binding to the acidic pH binding sites in a concentrationdependent manner with IC 50 values of 120 nM and Ͼ 400 nM, respectively. These IC 50 values were higher than those found in wild-type cells (MEF cells). This result suggests that in PEA-13 cells, the absence of LRP-1 may decrease the binding affinity of native ␣ 2 M or ␣ 2 M* for acidic pH binding sites. Alternatively, LRP-1 may collaborate with the acidic pH binding sites for ligand interactions at acidic pH.
The ligands of endocytic receptors such as transferrin, lactoferrin (a LRP-1 ligand) and LDL have also been shown to exhibit acidic pH binding in various cell types (33-35), but they have not been well characterized. To determine whether the acidic pH binding sites for ␣ 2 M* are also responsible for binding transferrin, lactoferrin and apoE at acidic pH in cells, we first examined the effects of these proteins on 125 I-␣ 2 M* binding to Mv1Lu, MEF, and PEA-13 cells. Cells were incubated with 2 nM 125 I-␣ 2 M* in the presence and absence of 10 M transferrin, lactoferrin, or apoE at pH 5.5 or pH 7.4 (for comparison). After 2.5 h at 0°C, the specific binding of 125 I-␣ 2 M* to cells was determined. At 10 M, all of these proteins completely blocked the specific binding (at pH 5.5) of 125 I-␣ 2 M* in Mv1Lu, MEF, and PEA-13 cells (data not shown). In contrast, none of these proteins had a significant effect on 125 I-␣ 2 M* binding (at pH 7.4) to Mv1Lu and MEF cells, which is mediated by LRP-1 (data not shown). Lactoferrin and ␣ 2 M* bind to distinct sites of LRP-1 and do not compete with each other for binding to LRP-1 (10,12). These results suggest that transferrin, lactoferrin and apoE may bind to the same acidic pH binding sites as ␣ 2 M* does. Alternatively, the acidic pH binding sites for these molecules may be different but overlapping. To further define the ligand specificity of the acidic pH binding sites, the effects of various concentrations of transferrin, lactoferrin, ␥-globulin, Pseudomonas exotoxin A (also a ligand of LRP-1) (36), or LDL on 125 I-␣ 2 M* binding (at pH 5 or 6) to Mv1Lu cells were determined. As shown in Fig. 4, A and B, increasing concentrations of lactoferrin, transferrin, and LDL correspondingly inhibited 125 I-␣ 2 M* binding to the acidic pH binding sites in Mv1Lu cells with IC 50 values of 0.05 M (pH 5), 0.5 M (pH 5), and 5 g/ml (pH 6), respectively. In contrast, ␥-globulin and Pseudomonas exotoxin at 0.5 M did not effectively inhibit 125 I-␣ 2 M* binding to the acidic pH binding sites (data not shown). These results indicate that the acidic binding sites are capable of binding ligands, which do not bind to LRP-1 (e.g. transferrin).
Cell Surface Annexin VI Is Involved in the Acidic pH Binding of 125 I-TGF-␤ 1 and 125 I-␣ 2 M* in Cells-The acidic pH binding sites on the cell surface identified herein may participate in the process of ligand endocytosis and degradation. They may be co-internalized with LRP-1/␣ 2 M receptors and possibly other unidentified receptors with ligand binding activities with neutral pH optima) and become activated in endosomes which have acidic luminal pH (pH 5.5-6.5) through acidic pH-induced conformational changes. They may also be present in endosomes where they function as intracellular cargo transporters, which target ligands for lysosomal degradation. To identify the membrane-associated protein(s) responsible for mediating the acidic pH binding, we decided to purify this protein(s) from Triton X-100 extracts of bovine plasma membranes by ␣ 2 M*-Sepharose affinity column chromatography at pH 5 or 6. Bovine liver  125 I-labeled ligand was determined. The specific binding (GST-RAP-sensitive binding) is estimated by subtracting nonspecific binding (GST-RAPinsensitive binding which was determined in the presence of 15 g/ml GST-RAP) from total binding (which was determined in the absence of GST-RAP). The nonspecific binding (GST-RAP-insensitive binding) was comparable to the nonspecific binding which was determined in the presence of 200-fold excess of unlabeled ␣ 2 M* and 15 g/ml GST-RAP. Each data point is the mean of quadruplicate determinations. plasma membranes were used as the starting material because they are rich in endocytic receptors such as LRP-1 (9, 10). If these acidic binding sites have collaborative interactions with the endocytic receptors in vivo, they should be abundant in tissues (e.g. liver) and cells that are rich in endocytic receptors. The Triton X-100 extracts (pH 5 or 6) containing 4 mM CaCl 2 of bovine liver plasma membranes were subjected to ␣ 2 M*-Sepharose 4B affinity column chromatography at pH 5 or 6. After extensive washing with HEPES/acetate buffer at pH 5 or 6 containing 0.1% Triton X-100 and 4 mM CaCl 2 , the column was eluted with HEPES/acetate buffer at pH 5 or 6 containing 10 mM EDTA and 0.1% Triton X-100 and the eluted fractions analyzed by silver staining. As shown in Fig. 5A, a 68-kDa protein was found in the EDTA eluent fractions (from pH 6 affinity column chromatography), as demonstrated by 7.5% SDS-PAGE under non-reducing conditions and silver staining. MALDI-TOF analysis of the tryptic digests of this 68-kDa protein revealed that the protein was bovine annexin VI (data not shown). Western blot analysis of the 68-kDa protein also supported the conclusion that it was annexin VI (Fig. 5B, lane  2). Under the experimental conditions (affinity column chromatography at pH 6), a very small amount of LRP-1 was found in the EDTA eluent fractions. LRP-1 is known to have ligand binding activity with a neutral pH optimum (10 -12). However, annexin VI appeared to be the major protein in the EDTA eluents of ␣ 2 M*-Sepharose 4B affinity column chromatography at pH 5 (data not shown) or 6. These results suggest that bovine annexin VI is an ␣ 2 M*-binding protein. To further define the direct interaction of annexin VI with ␣ 2 M*, we performed dot blot analysis. Pure annexin I, II, III, IV, and VI (0.1 g) were immobilized on nitrocellulose membranes. After blocking with BSA, the membranes were probed with 125 I-␣ 2 M* at pH 4, 5, 6, and 7 in the presence of 2 mM Ca 2ϩ . As shown in Fig. 6, 125 I-␣ 2 M* directly interacts with annexins in a pH-dependent manner. Binding of 125 I-␣ 2 M* to these annexins was maximum at pH 5. Quantitation of 125 I-␣ 2 M* bound to these annexins revealed that annexin VI bound more 125 I-␣ 2 M* than other annexins in the order of annexin VI Ͼ annexin IV Ͼ annexin III Ͼ annexin I Ͼ annexin II. Annexin V was also found to be as effective as annexin IV for binding 125 I-␣ 2 M* at pH 5 (data not shown). Binding of 125 I-␣ 2 M* to these annexins was abolished in the presence of RAP (100 g/ml) or BAPTA (5 mM) (data not shown). These results suggest that annexin VI as well as other annexins are capable of directly binding ␣ 2 M* with an optimum pH of 5.0. They are also consistent with the contention that cell surface annexin VI is involved in the acidic pH binding of ␣ 2 M* in cells.
Annexin VI has recently been shown to be a putative cell surface receptor for chondroitin sulfate (37). It was also reported to be capable of binding heparin (38,39). We, therefore, examined the effects of chondroitin sulfate and heparin on 125 I-␣ 2 M* binding to Mv1Lu cells at pH 5. As shown in Fig. 7A, chondroitin sulfate A, B and C were potent inhibitors of 125 I-␣ 2 M* binding at pH 5. Interestingly, chondroitin sulfate B and C were more potent than chondroitin sulfate A in inhibiting 125 I-␣ 2 M* binding to the cells. In contrast, heparin at 1 g/ml enhanced 125 I-␣ 2 M* binding to the acidic pH binding sites by 400% (Fig. 7B). At 10 g/ml, heparin inhibited 125 I-␣ 2 M* binding to these cells by Ͼ80% (Fig. 7B). Since annexin VI is known to bind chondroitin sulfate and heparin, these results are consistent with the contention that cell surface annexin VI is involved in the acidic pH binding of 125 I-␣ 2 M*. The molecular mechanism by which heparin at 1 g/ml increases 125 I-␣ 2 M* binding to the acidic pH binding sites is unknown. Heparin at 1 g/ml may increase 125 I-␣ 2 M* binding to the acidic pH binding sites by removal of inhibitor activity. Heparin has also been shown to increase and decrease binding of other ligands at low and high concentrations, respectively (40). At 10 g/ml, chondroitin sulfate A, B, and C and heparin did not significantly affect 125 I-␣ 2 M* binding to LRP-1 at neutral pH in Mv1Lu cells (data not shown).
To prove that cell surface annexin VI is involved in the acidic pH binding of LRP-1 ligands, we examined the effect of antiannexin VI IgG treatment at pH 6.4 and 7.4 (for subsequent binding assays at pH 6.4 and 7.4, respectively) on 125 I-TGF-␤ 1 binding to Mv1Lu cells or 125 I-␣ 2 M* binding to MEF and PEA-13 cells. The anti-annexin VI IgG was highly specific and did not react with other annexins as determined by Western blot analysis and Coomassie Blue staining (Fig. 8, A and B, respectively). Cells were treated with various concentrations of anti-annexin VI IgG or control IgG at pH 6.4 and 7.4 at 37°C for 2 h. The choice of pH 6.4 incubation (instead of pH 5.0 or 5.5) was to allow appropriate interaction of the antigen and antibody. The specific binding (at pH 6.4 and 7.4) of 125 I-TGF-␤ 1 or 125 I-␣ 2 M* was then determined. As shown in Fig. 9, increasing concentrations of anti-annexin VI IgG quantitatively blocked 125 I-TGF-␤ 1 binding at pH 6.4 in Mv1Lu cells (Fig. 9A), 125 I-␣ 2 M* binding at pH 6.4 in Mv1Lu, MEF, and PEA-13 cells (Fig. 9, B-D, respectively) and 125 I-␣ 2 M* binding at pH 7.4 in Mv1Lu and MEF cells (Fig. 9, E and F). Antiannexin VI IgG at 25 g/ml blocked binding of 125 I-TGF-␤ 1 to Mv1Lu cells by 50% (Fig. 9A) and at 30 g/ml completely blocked the 125   , cells were washed with binding buffer. The cell-associated 125 I-␣ 2 M* binding was determined. The specific binding was estimated by subtracting nonspecific binding. Scatchard plot analysis of the binding data was performed. 3-8 independent analyses were performed for estimating mean and S.D. PEA-13 cells (Fig. 9, B-D, respectively). Anti-annexin VI IgG (30 g/ml) also blocked 50 -60% of the 125 I-␣ 2 M* binding in Mv1Lu and MEF cells at pH 7.4 (Fig. 9, E and F); this was mainly mediated by LRP-1. Anti-annexin VI IgG (30 g/ml) exhibited only a slight inhibitory effect (15%) on 125 I-TGF-␤ 1 binding to Mv1Lu cells at pH 7.4 (data not shown); we presume this is due to the fact that 125 I-TGF-␤ 1 binding at pH 7.4 is mainly mediated by T␤R-I, T␤R-II and T␤R-III in Mv1Lu cells (1,3). The T␤R-V (LRP-1) is only responsible for a fraction of the 125 I-TGF-␤ 1 binding at pH 7.4 in Mv1Lu cells (3,7,8). In PEA-13 cells, which are deficient in LRP-1, anti-annexin VI IgG also completely inhibited 125 I-␣ 2 M* binding to cells at pH 6.4 (Fig. 9D). In the control experiments, treatment with antiannexin I IgG or anti-annexin II IgG (50 g/ml) did not affect 125 I-␣ 2 M* binding to Mv1Lu and MEF cells at pH 6.4 and 7.4 (data not shown). These results indicate that treatment of cells with anti-annexin VI IgG is capable of blocking 125 I-TGF-␤ 1 binding to the acidic pH binding sites (annexin VI) and also capable of inhibiting 125 I-␣ 2 M* binding to either the acidic binding sites (annexin VI) or LRP-1 (at pH 7.4). These results also suggest that cell surface annexin VI is involved in the acidic pH binding of 125 I-TGF-␤ 1 and 125 I-␣ 2 M* and in the neutral pH binding (to LRP-1) of 125 I-␣ 2 M*. The inhibition of 125 I-␣ 2 M* binding (at neutral pH) to LRP-1 by treatment of cells with anti-annexin VI IgG suggests that cell surface annexin VI may associate with LRP-1 and function as a coreceptor with LRP-1 at neutral pH. Alternatively, annexin VI may be located very close to LRP-1 at the cell surface. The cell surface localization of annexin VI was also shown by immunofluorescent staining of annexin VI at the cell surface of Mv1Lu, MEF, PEA-13, and Hep3B cells (Fig. 10, A, E, G, and  C, respectively). Annexin VI has also been localized at the cell surface of other cell types (37,38). To test the above possibilities, we performed co-immunoprecipitation of cell surfacebound 125 I-␣ 2 M* (at pH 7.4) in MEF cells and PEA-13 cells using anti-annexin VI IgG. As shown in Fig. 11, anti-annexin acidic pH binding sites in PEA-13 cells is as great as in wildtype MEF cells, PEA-13 cells, and MEF cells should be a good system for testing whether cell surface annexin VI is capable of mediating ligand (e.g. ␣ 2 M*) binding and internalization/degradation at acidic pH. PEA-13 and MEF cells were incubated with 125 I-␣ 2 M* at pH 6 or 7.4 (for MEF cells only) at 0°C for 2.5 h. PEA-13 cells did not exhibit specific binding of 125 I-␣ 2 M* at pH 7.4. This is consistent with the fact that they are deficient in LRP-1. These cells were then washed and warmed to 37°C. After 1 h at 37°C, the cell surface-bound, internalized and degraded (trichloroacetic acid-soluble) 125 I-␣ 2 M* were determined. As shown in Table II, MEF and PEA-13 cells were able to internalize and degrade 125 I-␣ 2 M* bound to the cell surface by 70% at pH 6.0. This cell surface binding (specific binding), internalization, and degradation of 125 I-␣ 2 M* could be blocked by preincubation of cells with anti-annexin VI IgG ( Fig. 9, C and D and data not shown). At pH 7.4, more than 90% of cell surface bound 125 I-␣ 2 M* underwent internalization/degradation (which was mediated by LRP-1) in MEF cells after an incubation time of 1 h (Table II). These results suggest that although cell surface annexin VI is less efficient than LRP-1 (which mediates internalization and degradation of 125 I-␣ 2 M* at pH 7.4), it is capable of mediating internalization and degradation of 125 I-␣ 2 M* at pH 6.
Specific Inhibitors Block Acidic pH Ligand Binding in Cells-Fluphenazine was previously shown to be an annexin VI binding compound as demonstrated by affinity column chromatography (41). Since fluphenazine and other phenothiazinerelated compounds, which are weak bases, are capable of entering cells and accumulating at high concentration in intracellular acidic compartments (e.g. endosomes) (42,43), it seemed possible that fluphenazine and similar compounds (e.g. trifluoperazine) may affect LRP ligand binding to annexin VI in the lumen of endosomes and prelysosomal compartments. To test this possibility, we examined the effects of several weak bases including trifluoperazine (a phenothiazine) (43), fluphenazine (another phenothiazine) (41), monodansylcadaverine (a transglutaminase inhibitor) (44), promethazine (another phenothiazine compound) (45), W-5 (a weak calmodulin antagonist) (46), W-7 (a potent calmodulin antagonist) (46), and verapamil (a calcium channel blocker) (47) on 125 I-␣ 2 M* binding to MEF cells and PEA-13 cells at pH 5.5 and pH 7.4. Among these compounds, trifluoperazine and fluphenazine were found to be the most potent inhibitors of 125 I-␣ 2 M* binding to MEF cells at pH 5.5 (Fig. 12A). Monodansylcadaverine and W-7 were less effective inhibitors. Promethazine, whose structure is homologous to trifluoperazine and fluphenazine, was not effective in blocking 125 I-␣ 2 M* binding to cells at pH 5.5. Verapamil and W-5 (100 M) were inactive in blocking 125 I-␣ 2 M* binding to cells at pH 5.5 (data not shown). Trifluoperazine and fluphenazine inhibited 125 I-␣ 2 M* binding in a concentration-dependent manner with IC 50 values of 65-75 M at pH 5.5. Trifluoperazine and fluphenazine also appeared to be effective in inhibiting 125 I-␣ 2 M* binding to LRP-1 at pH 7.4 (Fig. 12B). The IC 50 values of the trifluoperazine and fluphenazine were estimated to be 25-30 M (Fig. 12B). Interestingly, promethazine was almost as effective as trifluoperazine and fluphenazine for inhibiting 125 I-␣ 2 M* binding (at pH 7.4) to LRP-1 in MEF cells (Fig. 12B). These results suggest that trifluoperazine and fluphenazine are capable of blocking the binding of LRP ligands (e.g. ␣ 2 M*) to acidic pH binding sites or annexin VI and may be useful agents for defining the biological functions of the acidic  1 and 3) or control IgG (lanes 2 and 4). pH binding site-mediated or annexin VI-mediated binding in intracellular (endosomal) trafficking and degradation of internalized LRP ligands.
To determine the effect of trifluoperazine on 125 I-TGF-␤ 1 binding to the acidic pH binding sites, Mv1Lu cells were incubated with 100 pM 125 I-TGF-␤ 1 in the presence of various concentrations of trifluoperazine. After 2.5 h at 0°C, the specific binding of 125 I-TGF-␤ 1 was determined. As shown in Fig. 12C, trifluoperazine inhibited specific binding at pH 5 of 125 I-TGF-␤ 1 in a concentration-dependent manner with an IC 50 of 150 M. This result suggests that trifluoperazine also blocks 125 I-TGF-␤ 1 binding to the acidic pH binding sites (e.g. annexin VI) effectively.

Trifluoperazine and Fluphenazine Inhibit Cellular Degradation of 125 I-TGF-␤ 1 and 125 I-␣ 2 M* in Mv1Lu and MEF Cells-
Annexin VI has been implicated in the transport of LDL from early endosomes to later endosomes or prelysosomal compartments after internalization of LDL (48). We hypothesized that cell surface annexin VI (as an LRP-1 co-receptor or a component in the LRP-1 complex) and LRP-1-ligand complexes are co-internalized and enter early endosomes that have an acidic luminal pH, facilitating the strong interaction between annexin VI and LRP-1 ligands (e.g. ␣ 2 M*) and the dissociation of LRP-1 ligands from the LRP-1 complexes. Moreover, annexin VI, which is internalized or is already present in endosomes (48), may function as a cargo transporter, which carries the cargo (ligand) from early endosomes to late endosomes or the prelysosomal compartment (49 -53). In the prelysosomal compartment (in which the Ca 2ϩ concentration is low) (54), the cargo (LRP-1 ligands or other proteins) is unloaded from the annexin VI complex and then targeted to lysosomes for degradation. If this hypothesis is correct, trifluoperazine and fluphenazine, which are weak bases (like acridine orange) capable of entering cells and accumulating in the lumen of acidic endosomes (concentration Ͼ 100-fold that in medium) (55), should be able to block degradation of LRP-1 ligands by impairing their movement from early endosomes to late endosomes. We therefore examined the effect of 10 M trifluoperazine, fluphenazine, or promethazine on the internalization and degradation of cell surface-bound (specific binding) 125  These results support the hypothesis that annexin VI is involved in intracellular trafficking events leading to lysosomal degradation (48).

DISCUSSION
Acidic pH binding sites have been demonstrated in many cell types using various ligands, including IGFBP-3, vascular endothelial cell growth factor, transferrin, ApoE, and many others (33-35, 56, 57). However, they have not been well characterized. Here we demonstrate that LRP ligands IGFBP-3, TGF-␤ 1 , and ␣ 2 M* exhibit high capacity and low affinity acidic pH binding in MEF, PEA-13, and Mv1Lu cells. Unlike LRP-1, the acidic pH binding sites are sensitive to heparin and chondroitin sulfate. We also provide evidence to suggest that cell surface annexin VI is involved in the acidic pH binding of LRP-1 ligands (e.g. IGFBP-3, TGF-␤ 1 , and ␣ 2 M*) and other proteins. The evidence includes 1) Annexin VI is a major protein identified in Triton X-100 extracts of bovine liver plasma membranes, which binds to the ␣ 2 M*-Sepharose affinity column at acidic pH (pH 5 and 6) in a Ca 2ϩ -dependent manner. 2) Cell surface annexin VI is known to bind Ca 2ϩ , heparin and chondroitin sulfate (36 -39). The acidic pH binding of 125 2 Annexin VI is a member of a family of structurally homologous Ca 2ϩ -dependent phospholipid-binding proteins (37,58). It is abundant in rat liver endosomes (49,50,52), localized in the apical endosomes in rat hepatocytes, and colocalizes with Igp120, a prelysosomal marker in normal rat kidney cells (37,49 -52,59). It has been implicated in the budding of clathrin-coated pits from plasma membranes (37,59) and is involved in the trafficking of low density lipoprotein from endosomes to the prelysosomal compartment (48). It has also been shown to be able to form Ca 2ϩ channels and insert into membranes at acidic pH or in the presence of 4 mM GTP (38,60). Although annexin VI (like other types of annexins) lacks a signal sequence for secretion, it has been identified extracellularly where it can act as a receptor for chondroitin sulfate (37). The cell surface location of annexin VI may be due to its ability to insert into phospholipid bilayers (38,59). Here, using immunofluorescent staining, we demonstrate that annexin VI is localized at the cell surface of MEF, PEA-13 and Mv1Lu cells and other cell types. We also show that pretreatment of cells with anti-annexin VI IgG partially or completely prevents 125 I-TGF-␤ 1 binding to cells at pH 6.4 or 125 I-␣ 2 M* binding to cells at pH 6.4 and pH 7.4. Since LRP-1 is known to be responsible for ␣ 2 M* binding at pH 7.4, these results suggest that cell surface annexin VI may function as a receptor (at acidic pH) and a co-receptor (at pH 7.4) for LRP-1 ligands (e.g. ␣ 2 M*). This suggestion is supported by several observations: 1) The complete inhibition of 125 I-␣ 2 M* binding to the acidic pH binding sites by treatment of cells with anti-annexin VI IgG in Mv1Lu, MEF and PEA-13 cells indicates that cell surface annexin VI mediates the acidic pH binding of 125 I-␣ 2 M* in these cells. 2) Cell surface annexin VI is involved in ligand binding, internalization and degradation at acidic pH. 3) The partial inhibition of 125 I-␣ 2 M* binding (at pH 7.4) to LRP-1 by treatment of cells with anti-annexin VI IgG suggests that cell surface annexin VI may function as a co-receptor for only a fraction of LRP-1 on the cell surface. This suggestion is supported by the observation that 40% of LRP-1-bound 125 I-␣ 2 M* was immunoprecipitated by anti-annexin VI IgG. 4) Cellular heparan sulfate and chondroitin sulfate have been shown to be co-receptors for certain LRP-1 ligands (61)(62)(63)(64). Removal of heparan sulfate or chondroitin sulfate from cells by enzymic digestion appears to diminish the ability of the cells to internalize and degrade these LRP-1 ligands. Since annexin VI has been shown to bind heparin and chondroitin sulfate at the cell surface, we hypothesize that the heparan sulfate or chondroitin sulfate complex of cell surface annexin VI may serve as a co-receptor for these LRP-1 ligands. 5) The corresponding expression of both LRP-1 and annexin VI, as determined by Western blot analysis occurs in all cell types examined. 2 For example, fibroblasts (MEF and NIH 3T3 cells) exhibit 3-5-fold higher amounts of both LRP-1 and annexin VI than epithelial cells (mink lung epithelial cells). Carcinoma cells (e.g. HCT116 cells) that lack or express very low levels of LRP-1 also produce no or very little annexin VI (58). 2 Both annexin VI and LRP-1/T␤R-V are hypothesized to be candidates for tumor suppressor gene products (3-5, 7, 8, 58). 6) Cell surface annexin VI forms ternary complexes with 125 I-␣ 2 M* and LRP-1, as shown by co-immunoprecipitation (at pH 7.4) of annexin VI and 125 I-␣ 2 M* in MEF cells but not in PEA-13 cells. 7) ␣ 2 M* has been shown to regulate N-methyl-Daspartate receptor-mediated calcium influx in primary culture neurons (16). Since annexin VI and other annexin family mem-  bers are known to form calcium channels in membranes (37,59), they may play a role (as co-receptors) in the depletion of calcium ions from endosomes (possibly resulting in calcium influx) during endocytosis and in endosomal trafficking of ligands and their receptors (54). The calcium concentrations in extracellular compartments and late endosomes are in the range of mM and M, respectively (54). Depletion of calcium and acidification of endosomes are required for endosomal trafficking of internalized ligands and receptors (54), and 8) co-receptors (Grp 78 and midkine) for LRP-1 have recently been reported (17,65). However, we have no evidence to indicate the presence of either co-receptor in Mv1Lu and MEF cells.
The ligand binding activity of the acidic pH binding sites or cell surface annexin VI may play a role in tumor biology. Cumulative acquisition of genetic alteration via activation of proto-oncogenes to oncogenes and loss of tumor suppressor genes selects tumor cell clones with either proliferation or survival potential. The increase of nutrient and oxygen consumption in tumor cells leads to an extracellular microenvironment in tumors characterized by low oxygen and glucose levels and acidic pH ranging from pH 7.0 to as low as pH 5.8 (66 -70). The acidic microenvironment within solid tumors may contribute to changes in cellular physiology and responses of tumor cells. We hypothesize that the acidic pH binding sites or cell surface annexin VI in tumor cells may partially substitute for certain receptors (e.g. LRP-1, transferrin receptor, and LDL receptor, which have optimal activity at neutral pH) under such acidic conditions. Although the acidic pH binding sites (cell surface annexin VI) are less efficient than LRP-1 (at pH 7.4) in mediating ligand internalization and degradation, their high density in cells may enable them to function (at acidic pH) as a significant receptor class, comparable to LRP-1 or other receptors at pH 7.4 (Table II). If this hypothesis is correct, annexin VI should be important in animal pathophysiology. However, annexin VI-null mutant mice have been shown to exhibit normal phenotypes (38,59), suggesting that other annexin family members or other proteins may also be involved in the acidic pH ligand binding activity of cells. This possibility is supported by the observations: 1) A431 cells, which lack annexin VI, exhibit anti-annexin VI IgG-insensitive acidic pH ligand (␣ 2 M*) binding and internalization activity. 2 2) Other annexins (annexin I, III, IV, and V) are also capable of directly interacting with 125 I-␣ 2 M* with an optimal pH of 5 as demonstrated by dot blot analysis, and 3) fibronectin has recently been shown to mediate the acidic pH (pH 5.5) binding of VEGF in cells (57,70).
Trifluoperazine and fluphenazine, which are weak bases and have calmodulin antagonist activity, have been used as antipsychotic drugs. Their antipsychotic actions are believed to be mediated by their activity as dopamine receptor antagonists (71). Trifluoperazine has been shown to reversibly deplete 50% of cell surface ␣ 2 M* receptors at 30 M (43). Thioridazine, a phenothiazine derivative, has been reported to inhibit cellular degradation of 125 I-labeled epidermal growth factor (72). The mechanisms by which phenothiazine derivatives affect these cellular processes are unknown. The possible involvement of the weak basicity (raising the pH of endocytic vesicles or lysosomes) and calmodulin antagonist activity of these compounds at the concentrations generally used have been ruled out (43,72). Here we demonstrate that trifluoperazine and fluphenazine are effective inhibitors of 125 I-␣ 2 M* binding to the acidic pH binding sites (e.g. annexin VI) and to LRP-1 (at pH 7.4) with IC 50 values of 65-75 and 25-30 M, respectively. Since weak base compounds (e.g. acridine orange) are capable of entering cells and accumulating in the intracellular acidic compartments such as endosomes at a few hundred-fold higher concen-tration than that in medium (55), the acidic pH ligand binding inhibitory activity of trifluoperazine and fluphenazine may be pharmacologically significant. In our studies, treatment of cells with 10 M trifluoperazine or fluphenazine completely inhibits cellular degradation of cell surface-bound 125 I-TGF-␤ 1 or 125 I-␣ 2 M* following internalization. At 10 M in the medium, trifluoperazine or fluphenazine, both weak bases, should be able to accumulate in the lumen of endosomes at concentrations that are effective in inhibiting 125 I-TGF-␤ 1 -or 125 I-␣ 2 M*-annexin VI (or acidic pH binding site) complex formation in endosomes and subsequent lysosomal targeting. Promethazine, another weakly basic phenothiazine, appears to be ineffective in blocking cellular degradation of 125 I-TGF-␤ 1 or 125 I-␣ 2 M* under the same experimental conditions, suggesting that the inhibition of the 125 I-TGF-␤ 1 or 125 I-␣ 2 M* degradation by trifluoperazine or fluphenazine is specific and is likely due to its newly identified annexin VI or acidic pH ligand binding inhibitory activity.
Endosomal signaling has recently been shown to play a pivotal role in several ligand receptor-mediated signaling cascade systems (73,74). Inhibition of lysosomal targeting for degradation of ligands should logically enhance or prolong endosomal signaling mediated by the ligand-receptor complex. Trifluoperazine or related compounds may be useful agents for enhancing pharmacological actions of ligands, which are sensitive to these compounds and utilize endosomal signaling.