Annexin A2 Regulates β1 Integrin Internalization and Intestinal Epithelial Cell Migration*

Background: Intestinal epithelial wound closure requires dynamic regulation/turnover of cell-matrix adhesions. Results: Down-regulation of annexin A2 increases cell-matrix adhesion and delays β1 integrin internalization from the plasma membrane. Conclusion: Annexin A2 promotes wound closure by regulating internalization of β1 integrin in migrating epithelial cells. Significance: This study identifies a novel role of annexin A2 in regulating β1 integrin internalization required for intestinal epithelial cell migration. The gastrointestinal epithelium functions as an important barrier that separates luminal contents from the underlying tissue compartment and is vital in maintaining mucosal homeostasis. Mucosal wounds in inflammatory disorders compromise the critical epithelial barrier. In response to injury, intestinal epithelial cells (IECs) rapidly migrate to reseal wounds. We have previously observed that a membrane-associated, actin binding protein, annexin A2 (AnxA2), is up-regulated in migrating IECs and plays an important role in promoting wound closure. To identify the mechanisms by which AnxA2 promotes IEC movement and wound closure, we used a loss of function approach. AnxA2-specific shRNA was utilized to generate IECs with stable down-regulation of AnxA2. Loss of AnxA2 inhibited IEC migration while promoting enhanced cell-matrix adhesion. These functional effects were associated with increased levels of β1 integrin protein, which is reported to play an important role in mediating the cell-matrix adhesive properties of epithelial cells. Because cell migration requires dynamic turnover of integrin-based adhesions, we tested whether AnxA2 modulates internalization of cell surface β1 integrin required for forward cell movement. Indeed, pulse-chase biotinylation experiments in IECs lacking AnxA2 demonstrated a significant increase in cell surface β1 integrin that was accompanied by decreased β1 integrin internalization and degradation. These findings support an important role of AnxA2 in controlling dynamics of β1 integrin at the cell surface that in turn is required for the active turnover of cell-matrix associations, cell migration, and wound closure.

The gastrointestinal epithelium functions as an important barrier that separates luminal contents from the underlying tissue compartment and is vital in maintaining mucosal homeostasis. Mucosal wounds in inflammatory disorders compromise the critical epithelial barrier. In response to injury, intestinal epithelial cells (IECs) rapidly migrate to reseal wounds. We have previously observed that a membrane-associated, actin binding protein, annexin A2 (AnxA2), is up-regulated in migrating IECs and plays an important role in promoting wound closure. To identify the mechanisms by which AnxA2 promotes IEC movement and wound closure, we used a loss of function approach. AnxA2-specific shRNA was utilized to generate IECs with stable down-regulation of AnxA2. Loss of AnxA2 inhibited IEC migration while promoting enhanced cell-matrix adhesion. These functional effects were associated with increased levels of ␤1 integrin protein, which is reported to play an important role in mediating the cell-matrix adhesive properties of epithelial cells. Because cell migration requires dynamic turnover of integrinbased adhesions, we tested whether AnxA2 modulates internalization of cell surface ␤1 integrin required for forward cell movement. Indeed, pulse-chase biotinylation experiments in IECs lacking AnxA2 demonstrated a significant increase in cell surface ␤1 integrin that was accompanied by decreased ␤1 integrin internalization and degradation. These findings support an important role of AnxA2 in controlling dynamics of ␤1 integrin at the cell surface that in turn is required for the active turnover of cell-matrix associations, cell migration, and wound closure.
The intestinal epithelium functions as a selectively permeable barrier that allows for the absorption of nutrients and fluids while excluding bacteria and toxins from underlying tissue compartments. The epithelium is highly dynamic as it is actively turned over by proliferation of crypt epithelial cells, migration of cells along the crypt-luminal axis, and apoptosis at the luminal surface. Epithelial injury such as that observed in inflammatory diseases leads to fluid and electrolyte loss as well as exposure to luminal antigens that further exacerbate the inflammatory response. It is therefore critical that epithelial wounds efficiently reseal so as to reestablish the intestinal epithelial barrier and restore the physiologic state. The intestinal epithelium has a remarkable capacity to efficiently reseal wounds. An important mechanism of intestinal epithelial wound closure involves migration of the epithelial cell sheet that rapidly covers denuded surfaces (1)(2)(3).
Cell migration requires regulated cycles of polarized cell-matrix attachment at the leading edge of the cell with corresponding detachment at its trailing edge. These critical migratory events are mediated by dynamic turnover of integrin-mediated adhesions. Indeed, disassembly of integrin-based cell-matrix adhesions is important for mediating forward cell movement. Past studies have shown that integrin based adhesions form at the leading edge, mature, and then disassemble at the trailing edge. Focal adhesion disassembly requires endocytosis of integrins that are then recycled back to the plasma membrane at the leading edge of migrating cells or destroyed by lysosomal degradation (4 -7). A central component of such biological events is the coordinated intracellular movement of proteins allowing for reuse of non-damaged proteins. Studies from others and our group have demonstrated that ␤1 integrin plays an important role in IEC 2 cell-matrix adhesion and migration (8 -10). Additionally, we and others have shown endocytosis and trafficking of ␤1 integrin is lipid raft-mediated (11,12).
We have previously reported that the calcium-dependent phospholipid binding protein, annexin A2 (AnxA2), facilitates IEC migration (13). AnxA2 is a highly conserved protein in invertebrates and plants and is expressed in most cell types. AnxA2 contains four 70 -80-amino acid annexin core repeats that allow for Ca 2ϩ -dependent attachment to phosphatidylinositol 4,5-bisphosphate-rich membranes, a component of lipid rafts (14 -20). Additionally, AnxA2 can cap filamentous actin while associated with lipid rafts (21)(22)(23). Because AnxA2 regulates membrane-membrane and membrane-cytoskeletal interactions, it has been implicated in functioning as a regulatory protein for a wide range of membrane trafficking events that include endocytosis, endosome fusion, and exocytosis (24 -29).
Although the structure function relationship between AnxA2 and migration has been investigated, the mechanisms by which AnxA2 controls IEC migration remain poorly understood. Using an AnxA2 shRNA gene silencing approach, we identify a role for AnxA2 in regulating the internalization of cell surface ␤1 integrin that is needed for the turnover of focal adhesions, forward cell movement, and ultimately, wound closure.

EXPERIMENTAL PROCEDURES
Cell Culture and Constructs-Lentiviruses were generated in HEK293T cells by polyethyleneimine transfection (Open Biosystems). Titered lentiviruses were then used to transduce SK-CO15 and Caco2 human colonic cancer epithelial cells. Clonal lines were grown as described previously except for the addition of 2 g/ml puromycin (30). Cycloheximide (MP Biomedical) was used at the concentration of 50 g/ml for the given amount of time at 37°C before harvesting cells. Bafilomycin A1 (Acros) was used at the concentration of 30 nM for two hours at 37°C before fixation. For transient rescue experiments, a pShuttle eGFP plasmid containing a shRNA-resistant AnxA2 (Addgene) was electroporated into cells in solution L, program 15 (Lonza). For scratch wound rescue experiments a shRNA-resistant AnxA2-eGFP was subcloned from pShuttle into pLEX to generate a shRNA-resistant AnxA2-eGFP lentivirus. Cells were transduced with the lentivirus and sorted for eGFP expression by flow cytometry. Supplemental Table 1 contains the shRNA sequences and primers used for the creation of shRNA-resistant constructs.
Immunofluorescence Microscopy-Cells were fixed in 3.7% formalin for 20 min and permeabilized in 0.5% Triton X-100 for 3 min. Fixed cells were then incubated with 5% bovine serum albumin for 1 h before primary antibody incubation overnight. After secondary antibody incubation, cells were mounted in ProLong gold antifade agent (Invitrogen) before confocal microscopy (Zeiss LSM 510). The following primary antibodies were used: pPaxillin Y118 (Cell Signaling), ␤1 integrin clone Mab13 (BD Biosciences), AnxA2 clone 3E8-B6 (Sigma). Alexa Fluor-conjugated secondary antibodies were purchased from Invitrogen. Proliferation was monitored by a 1-h pulse with 5-ethynyl-2Ј-deoxyuridine (Invitrogen). Cells were fixed and permeabilized as described above before counterstaining with To-Pro 3-iodide (Invitrogen) to highlight nuclei. For Rab5a co-localization experiments, cells were transduced in suspension with Celllight early endosomes RFP, Bacmam (version 2.0, Invitrogen) before culturing. Focal adhesions in individual cells were quantified using To-Pro 3-iodide to highlight nuclei and NIH ImageJ software.
Scratch Wound Assay-Scratch wounds were generated using suction via a micropipette. Time-lapse imaging of wound closure was then performed. Images of the leading edge were captured every 2 min for 4 h at 37°C with an Electron Multiplying Charge-Coupled Device camera (Photometrics). For the 24-h time point, the bottom of the culture plate was scored by a marker perpendicular to the wound, and images were taken at this intersection immediately after wounding and 24 h later. NIH ImageJ was used to quantify wound areas.
Extracellular Matrix Adhesion Assay-Single cell suspensions were generated using a non-enzymatic cell dissociation buffer (Invitrogen). Equal concentrations of cells (10000 cells in 0.1 ml) were resuspended in optimem with 0.1% BSA and then incubated with BCECF-AM (Invitrogen) according to the manufacturer's protocol. Antibodies against ␤1 integrin, clone Mab13 (Becton Dickinson Biosciences) or clone TS2/16 (eBioscience), were dialyzed in PBS overnight and used at a concentration of 5 g/ml. Cells were incubated with these antibodies for 2 h prior to the adhesion assay. Antibody-treated cells were then allowed to adhere to a matrix-coated/BSAblocked 96-well plate for 30 min at 37°C. Plates were coated overnight with 10 g/ml of Matrigel TM (356234, BD Biosciences) or fibronectin (354008, BD Biosciences).
Real Time PCR-RNA was isolated using Trizol (Invitrogen) and treated with DNase I (Qiagen) prior to cDNA synthesis using an oligo(dT) primer and Superscript III (Invitrogen). SYBR Green master mix Applied Biosystems was used to perform real time PCR on a Bio-Rad iQ5 cycler. Cycle threshold values (C t ) were used to calculate the fold change in mRNA according to the 2ЈϪ⌬⌬C T method (31). GAPDH was used as a reference gene. Supplemental Table 1 includes PCR primer sequences.
Lipid Raft Isolation-Cells were harvested in HBSS ϩ supplemented with a protease inhibitor mixture (Sigma) before nitrogen cavitation at 200 psi for 15 min (32). The postnuclear supernatant was placed in the bottom of a 5-35% continuous sucrose gradient. Gradients were then centrifuged at 39,000 rpm for 19 h in a SW41 rotor (Beckman) to isolated light density lipid rafts (ϳ21% sucrose). Sucrose fractions were harvested and immunoblotted.
Cell Surface Biotinylation and Endocytosis-Pulse-chase biotinylation experiments were performed as described previously (33,34). Briefly cell surface proteins were biotinylated (0.5 mg/ml) on ice using NHS-SS-biotin (Pierce). Internalization of cell surface proteins was induced by placing the cells at 37°C for 2 h. Biotin was stripped from non-internalized proteins by treating cells twice with 20 mM 2-mercaptoethane sulfonate sodium (Genscript) for 5 min. Cells were then treated with 20 mM iodoacetamide (Sigma) for 15 min to quench any remaining 2-mercaptoethane sulfonate sodium. Biotinylated proteins from soluble lysates (200 g) were purified using monomeric avidin agarose (Pierce), eluted in a reducing SDS sample buffer, and subjected to immunoblot analysis.

RESULTS
AnxA2 Promotes Epithelial Cell Migration and Wound Closure-We have previously reported that AnxA2 is up-regulated in migrating IECs and controls cell motility and wound closure (13). To further identify the mechanisms by which AnxA2 regulates epithelial cell motility, we generated cell lines with stable down-regulation of AnxA2 using two different shRNA sequences (shAnxA2) and a control cell line with a nonsilencing shRNA target (shCtrl). Additionally, to verify our results, AnxA2 was stably down-regulated in two model IECs (SK-CO15, Caco2). Immunoblot analysis revealed a significant down-regulation of AnxA2 (Ͼ80% reduction) in shAnxA2 cells as compared with shCtrl cells (Fig. 1A). Furthermore, to determine whether any effects of the AnxA2 shRNA were specific to loss of AnxA2 protein, we transduced shAnxA2 cells with a shRNA-resistant eGFP-tagged AnxA2. As shown in Fig. 1A, the AnxA2 eGFP fusion protein was identified at 70 kDa using an AnxA2 antibody. Actin and tubulin were used as loading controls. We next tested the functional effects of stable AnxA2 down-regulation and rescue on IEC wound closure. Using an in vitro scratch wound resealing assay and time-lapse imaging, we observed that loss of AnxA2 resulted in a 4-fold delay in wound closure over a four hour time period and re-expression of AnxA2 was able to significantly rescue the delay in wound closure (p Ͻ 0.001) (Fig. 1B, supplemental Movies 1-3). Similarly, after 24 h of wound closure, shAnxA2 cells displayed a 2-fold decrease in wound closure that was abrogated in shAnxA2 cells re-expressing AnxA2 (supplemental Fig. 1A). These experiments support a role of AnxA2 in controlling intestinal epithelial wound closure.
Because wound closure can be mediated by cell proliferation in addition to cell migration, epithelial proliferation was evaluated by determining EdU incorporation in cells during wound closure. Despite decreased wound closure in shAnxA2 cells as compared with control cells, we observed increased proliferation in cells with down-regulated AnxA2 (supplemental Fig.  1B). These findings suggest that the decreased wound closure following AnxA2 down-regulation is mediated by an influence of this protein on epithelial cell migration.
Down-regulation of AnxA2 Increases Cell Matrix Adhesion-Because AnxA2 facilitates IEC migration that, in turn, requires dynamic turnover of cell-matrix associations, we next investigated the adhesion of epithelial cells to a complex extracellular matrix. Down-regulation of AnxA2 increased cell-matrix adhesion (40% increase) ( Fig. 2A, left panel). In an analogous manner, increased adhesion of shAnxA2 cells to fibronectin was also observed ( Fig. 2A, right panel). Because SK-CO15 cells were used for these assays, we verified adhesive properties of another model intestinal epithelial cell line, Caco2. Analogous to SK-CO15 cells, increased cell-matrix adhesion was seen in Caco2 cells with down-regulated AnxA2 (data not shown). To biochemically link functional cell-matrix adhesion to the activity of proteins that mediate cell-matrix adherence, we evaluated the phosphorylation/activation status of the ␤1 integrin scaffold protein, paxillin. Phosphorylated paxillin at tyrosine 118 (pPaxY118) has been previously reported in engaged cell-matrix adhesions (35,36). In keeping with the increased adhesive properties of shAnxA2 cells, immunoblot analysis revealed pPaxY118 and total Paxillin levels were increased 2-fold in shAnxA2 cells compared with shCtrl IECs (Fig. 2B). Immunofluorescence labeling and confocal microscopy of spreading cells corroborated the biochemical results. In shCtrl cells, pPaxY118 is visualized in short punctate basal structures representing focal adhesions. In shAnxA2 cells, the pPaxY118 containing cell-matrix adhesions were more prominent in size and staining intensity. Quantification of pPaxY118-positive adhesions per spreading cell revealed a 2-fold increase in the number adhesions in cells with downregulated AnxA2 (Fig. 2D). These findings support a role of AnxA2 in regulating cell-matrix adhesions in IECs migrating to close wounds.
Cell Surface ␤1 Integrin Stability Is Increased in Cells with Down-regulated AnxA2-Cell surface ␤1 integrin mediates IEC-matrix adhesion and plays an important role in cell movement. Because we had observed that AnxA2 down-regulation resulted in increased cell-matrix adhesion with a concomitant increase in pPaxY118 containing focal adhesions, we next determined whether the increase in cell-matrix adhesion in shAnxA2 cells was dependent upon ␤1 integrin function. Inhibitory (Mab13) but not non-inhibitory (TS2/16) ␤1 integrin antibodies abrogated the increase in cell-matrix adhesion in shCtrl and shAnxA2 cells ( Fig. 2A, right panel). Additionally, the inhibition of ␤1 integrin abrogated the difference in pPaxY118 in shCtrl and shAnxA2 cells (Fig. 2C). To better understand how AnxA2 regulates ␤1 integrin, we first exam-ined the steady state levels of ␤1 integrin mRNA and protein. IEC immunoblots revealed a 2-fold increase in ␤1 integrin protein in shAnxA2 cells as compared with shCtrl cells (Fig. 3A). The increased ␤1 integrin protein in shAnxA2 cells was not associated with a corresponding increase in ␤1 integrin mRNA (Fig. 3B), suggesting an influence of AnxA2 on ␤1 integrin protein dynamics/stability. We next determined whether AnxA2 regulates the post-translational stability of ␤1 integrin. As shown in Fig. 3C, the stability of ␤1 integrin protein was increased 4 h after cycloheximide treatment in shAnxA2 cells as compared with shCtrl cells. Additionally, ␤1 integrin stability was significantly increased after 8 h of cycloheximide treatment in shAnxA2 cells as compared with shCtrl cells (p Ͻ 0.05).
Given that down-regulation of AnxA2 resulted in increased stability of ␤1 integrin protein, we next determined ␤1 integrin subcellular localization by immunofluorescence labeling and confocal microscopy. In control cells, ␤1 integrin is visualized in intracellular punctate, "vesicle-like" structures that were prominent at the leading edge of migrating cells in addition to a small pool in the basolateral membrane. However, in migrating shAnxA2 cells, ␤1 integrin was localized predominantly in the basolateral membrane of migrating cells (Fig. 3D). These findings suggest that loss of AnxA2 results in increased ␤1 integrin at the cell surface. To further support these results, we biochemically analyzed cell surface ␤1 integrin levels. Cell surface biotinylation, avidin pulldown, and ␤1 integrin immunoblotting revealed a 6-fold increase in cell surface ␤1 integrin relative to total ␤1 integrin in shAnxA2 cells as compared with control cells (Fig. 3E). The AnxA2 effects observed were specific for ␤1 integrin as we did not observe a change in total and cell surface levels of another basolateral plasma membrane-associated protein, E-cadherin. The influence of AnxA2 loss on total and cell surface ␤1 integrin levels was also confirmed in another model intestinal epithelial cell line, Caco2 (Fig. 3F).
AnxA2 Promotes ␤1 Integrin Localization in Endosomes-Cell surface proteins undergo trafficking into early endosomes. Thus, to further characterize ␤1 integrin trafficking in cells with down-regulated AnxA2, we co-localized ␤1 integrin with Rab5a, a marker of early endosomes. To highlight early endosomes, a CellLight early endosome RFP, Bacmam (version 2.0, RFP-Rab5a) construct was expressed in migrating IECs. In shCtrl cells, a large pool of ␤1 integrin co-localized with early endosomes. However, in shAnxA2 cells, ␤1 integrin was distributed in the basolateral plasma membrane with minimal localization in RFP-Rab5a containing early endosomes (Fig.  4A). In keeping with the imaging data, subcellular fractionation of migrating epithelial cells by isopycnic sucrose density gradient sedimentation revealed co-sedimentation of AnxA2 with ␤1 integrin and an early endosome protein, early endosome antigen 1. Down-regulation of AnxA2 decreased co-sedimentation of ␤1 integrin with early endosome antigen 1 (data not shown). To ensure the specificity of the AnxA2 shRNA, we performed a rescue experiment to determine whether the expression of a shRNA-resistant eGFP-tagged AnxA2 in shAnxA2 cells would promote the redistribution of ␤1 integrin to endosomes. Immunofluorescence labeling and confocal microscopy revealed that in cells lacking eGFP-AnxA2, ␤1 integrin resided mainly in the basolateral membrane. In cells re-expressing AnxA2, as marked by GFP expression, ␤1 integrin redistributed to endosomal-like vesicle structures (Fig. 4B).
␤1 Integrin Internalization Is Delayed in IECs with Downregulated AnxA2-Because loss of AnxA2 increases the cell surface level of ␤1 integrin, we investigated whether AnxA2 functions to either increase the rate of ␤1 integrin internalization or decrease the rate of ␤1 integrin trafficking back to the cell surface. To distinguish between these two possibilities, we first tested the effect of bafilomycinA1 (BafA1) on ␤1 integrin localization in shCtrl and shAnxA2 cells. BafA1 inhibits the vacuolar-type H ϩ ATPases, retarding protein trafficking between endosomes needed for recycling (37). In shCtrl cells, BafA1 treatment increased the intracellular accumulation of ␤1 integrin. However, in BafA1-treated shAnxA2 cells, ␤1 integrin primarily localized in the plasma membrane, suggesting that loss of AnxA2 delays ␤1 integrin internalization (Fig. 5A). To determine the rate of ␤1 integrin internalization in shCtrl and shAnxA2 cells, biotin pulse-chase experiments were performed. After biotin labeling of cell surface proteins at 4°C, cells were transferred to 37°C to induce protein internalization. Cells were then placed back at 4°C, and biotin was chemically removed (stripped) from non-internalized cell surface proteins, leaving only internalized proteins biotinylated. Biotinylated proteins from resultant lysates were pulled down using avidin agarose and immunoblotted for ␤1 integrin. To ensure complete biotin stripping from non-internalized proteins, one group of cells was left at 4°C after labeling and then stripped and lysed (labeled strip). Additionally, to account for internalized and degraded biotinylated protein, cells were lysed directly after the internalization period. In shCtrl IECs, all cell surface ␤1 integrin was internalized after 2 h at 37°C. In contrast to shCtrl cells, the rate of ␤1 integrin internalization was significantly decreased in shAnxA2 cells (2-fold) (Fig. 5B). To better define how AnxA2 regulates the internalization of ␤1 integrin, we determined whether these proteins co-fractionate with the lipid raft associated protein caveolin-1. Lipid rafts from control and AnxA2 down-regulated cells were isolated by floatation in sucrose gradients. Indeed, immunoblot analysis identified AnxA2 and ␤1 integrin with caveolin-1 in lipid raft-containing sucrose fractions (Fig. 5C). To verify the specificity of this finding, we immunoblotted fractions for the transferrin receptor as it has been previously reported that this receptor does not associate with lipid rafts (32). In conclusion, these results suggest that AnxA2 and ␤1 integrin associate with IEC caveolin-1-containing lipid rafts that have been previously shown to mediate ␤1 integrin internalization from the cell surface (11,12).

DISCUSSION
Directed migration of a cohesive sheet of epithelial cells is needed to achieve mucosal wound closure. In this report, we identify a novel role of AnxA2 in controlling ␤1 integrin trafficking that is also required for cell-matrix adhesion and migration of the epithelial sheet. In a previous report, AnxA2 has been implicated in mediating cell-matrix adhesion (38). These authors demonstrated that phosphorylation of AnxA2 induces cell detachment and that this effect requires RhoA. In our current study, we show that AnxA2 modulates ␤1 integrin endocytosis from the cell surface, thereby promoting integrin trafficking and cell migration. In the absence of AnxA2, ␤1 integrin internalization into early endosomes is inhibited, thereby resulting in increased cell surface ␤1 integrin that is available FIGURE 3. Cell surface ␤1 integrin stability is increased following loss of AnxA2. A, control non-silencing shRNA (shCtrl) or shAnxA2 migrating epithelial cells were isolated and immunoblotted for ␤1 integrin, AnxA2, and actin. The bar graph shows densitometric quantification with mean Ϯ S.E. B, after purification of total RNA and cDNA synthesis, real time PCR was performed to determined the relative levels of ␤1 integrin mRNA in shCtrl and shAnxA2 cells. Relative ␤1 integrin mRNA levels in shCtrl versus shAnx2 cells was calculated using the 2Ϫ⌬⌬C T method. GAPDH was included as a reference gene. Results are presented as mean Ϯ S.E. C, shCtrl and shAnxA2 cells were treated with either dimethyl sulfoxide (DMSO; 1:500) or cycloheximide (Cyclohex.) for the stated amount of time, and cell lysates were immunoblotted for ␤1 integrin (␤1 int.), AnxA2, and tubulin. Results are mean Ϯ S.E. *, p Ͻ 0.05. D, immunofluorescence labeling and confocal microscopy of AnxA2 and ␤1 integrin in shCtrl and shAnxA2 migrating epithelial cells (24 h). Representative confocal images of leading edge cells adjoin the wound are shown. Scale bar, 20 m. E, cell surface proteins from shCtrl and shAnxA2 SK-CO15 cells were biotinylated on ice, captured with streptavidin, and immunoblotted to detect ␤1 integrin (␤1 int.), E-cadherin (E-Cad.), AnxA2, and actin. A non-biotinylated (No Bio.) sample was included as negative control. F, cell surface proteins from shCtrl and shAnxA2 Caco2 cells were biotinylated on ice, captured with streptavidin, and immunoblotted to detect ␤1 integrin (␤1 int.), AnxA2, and actin. for cell matrix adhesion. Because cells require remodeling of ␤1 integrin adhesions for movement, such a buildup of ␤1 integrin following loss of AnxA2 retards cell motility. This finding is analogous to a report indicating that down-regulation of sorting nexin 27 results in disruption of multidrug resistance protein 4 receptor internalization, causing increased steady state levels of this receptor (39). In a previous study, we reported that AnxA2 controls intestinal epithelial wound closure by modulating the small GTPase, RhoA (13). Our current findings are complementary to this report because Rho GTPases control F-actin organization that, in turn, modulates dynamics of membrane raft-associated proteins (40). Interestingly, in our previous report, we also observed that transient and incomplete down-regulation of AnxA2 did not significantly alter ␤1 integrin protein levels. In contrast, our current findings were derived from stable cell lines having greater than 90% down-regulation of AnxA2, which clearly show significantly increased ␤1 integrin protein levels. These differences can be explained either by limited ability to detect changes under conditions of low transfection efficiency or that there is a threshold for AnxA2 downregulation, beyond which massive loss of AnxA2 inhibits ␤1 integrin trafficking.
Efficient cell migration requires the internalization and recycling of integrins, however, the general mechanism of ␤1 integrin endocytosis in IECs remains incompletely understood. In our study, we observed that ␤1 integrin and AnxA2 co-fraction-ate with caveolin in lipid raft fractions, suggesting that in IECs, endocytosis of ␤1 integrin is mediated via lipid rafts. Furthermore, our results suggest that internalization of ␤1 integrin into early endosomes is facilitated by AnxA2. These findings are highlighted in a hypothetical model in which AnxA2 association with lipid rafts allows for the internalization of ␤1 integrin, thereby facilitating focal adhesion turnover in migrating cells (Fig. 6A). Following the loss of AnxA2, focal adhesions containing ␤1 integrin and paxillin accumulate at the plasma membrane resulting in increased cell-matrix adhesion and decreased cell migration (Fig. 6B). Interestingly, we observed that paxillin steady-state levels mimic that observed for ␤1 integrin. A functionally inhibitory ␤1 integrin antibody (Mab13) decreased paxillin protein levels. Thus, the association of paxillin with extracellular matrix engaged ␤1 integrin decreases the rate at which paxillin gets degraded in the cytosol. The phosphorylation of tyrosine 118 in paxillin was also enhanced in shAnxA2 cells. Because this phosphorylation is observed when ␤1 integrin is engaged with the extracellular matrix (35), this confirms that ␤1 integrin is fully functional after loss of AnxA2.
AnxA2 has been implicated in regulating intracellular protein trafficking by promoting endosomal fusion and exocytosis. These studies show that AnxA2 enhances the aggregation and fusion of vesicles (26,41,42). Interestingly, previous studies in cells with down-regulated AnxA2 have reported that epidermal growth factor and transferrin can be internalized but accumu- FIGURE 5. Down-regulation of AnxA2 delays ␤1 integrin internalization. A, after treating shCtrl and shAnxA2 cells with BafA1, immunofluorescence labeling of ␤1 integrin (white) and confocal microscopy were performed. Red arrows highlight intracellular accumulation of ␤1 integrin. Scale bar, 20 m. B, a pulse-chase biotinylation assay was performed in shCtrl and shAnxA2 cells to determine ␤1 integrin internalization rates. Cell surface proteins were biotinylated on ice (Cell surface; CS), and protein internalization was allowed to occur at 37°C for 2 h. Biotin from non-internalized proteins was chemically removed, and internalized biotinylated proteins were pulled down with avidin agarose and subjected to immunoblotting (Internalized). A stripping control (Strip) was performed to ensure the removal of biotin from non-internalized proteins. To control for protein degradation, a non-degraded control (Non-Deg.) was included after the internalization period. Percent internalization was quantified by adding the degraded fraction to the internalized fraction divided by the surface fraction. A non-biotinylated (No Bio.) sample was included as negative control. MESNA, Sodium 2-mercaptoethanesulfonate. The graph represents three independent experiments. Results indicate mean Ϯ S.E. *, p Ͻ 0.05. C, shCtrl cells were lysed by nitrogen cavitation, and membrane raft fractions were isolated by floatation in continuous sucrose gradient (5-30%). The fractions were subjected to immunoblot analysis of ␤1 integrin (␤1 int.), AnxA2, caveolin-1 (Cav-1), and transferrin receptor (TfR). FIGURE 6. Proposed model for annexin A2-mediated endocytosis of ␤1 integrin. A, in intestinal epithelial cells, AnxA2 associates with lipid rafts (Rafts) that contain ␤1 integrin (␤1). To promote internalization of rafts, the amino terminus of AnxA2 caps plasma membrane-associated polymerizing F-actin (45)(46)(47). Internalized ␤1 integrin can then be reused at the new leading edge of cells, allowing for efficient cell migration. B, in IECs lacking AnxA2, lipid raft (Raft)associated ␤1 integrin (␤1) builds up at the plasma membrane. Paxillin then becomes phosphorylated and stabilized as a consequence of the increased adhesion. Without being able to detach from the matrix, IECs lacking AnxA2 do not efficiently migrate to close wounds. late in late endosomes (41,43). Because we observed inhibition of cell surface ␤1 integrin endocytosis after down-regulation of AnxA2, it appears that the function of AnxA2 likely depends on its cargo in the vesicular compartments. Furthermore, our lossof-function study can only analyze the influence of AnxA2 on the earliest event in ␤1 integrin endocytic recycling, which we determined to be internalization. We therefore do not exclude a role of AnxA2 in controlling ␤1 integrin trafficking between other endosomal compartments such as ␤1 integrin recycling back to the plasma membrane in migrating epithelial cells.
Current knowledge of how AnxA2 functions in the endocytosis of lipid raft-associated proteins is very limited. A single study to date suggested that AnxA2 promotes the internalization of a cell surface receptor (44). To build a more comprehensive model on the role of AnxA2 in endocytosis, studies have analyzed key AnxA2 binding partners. These reports have shown that AnxA2 can simultaneously interact with actin and lipid rafts (14 -20). The role of actin in endocytosis and the involvement of lipid rafts in endocytosis have been reported previously (45,46). Actin-nucleating proteins attach to the plasma membrane, thereby promoting polymerization of actin at these sites. Vesicles are attached to filamentous actin (F-actin), which facilitates inward budding (47). Because AnxA2 can cap filamentous actin while being associated with lipid rafts, we propose that AnxA2 provides a link between F-actin and internalization of ␤1 integrin from cell surface rafts to early endosomes (see model in Fig. 6A). It is likely that PKC-mediated phosphorylation of AnxA2 plays a regulatory role in the endocytosis of lipid rafts and cell-matrix detachment because PKC has been reported to phosphorylate AnxA2 and also regulate the endocytosis of ␤1 integrin (38, 48 -50).
In summary, our findings implicate an important role of AnxA2 in controlling ␤1 integrin dynamics, epithelial migration, and wound closure.