Tetraspanin CD151 Promotes Cell Migration by Regulating Integrin Trafficking*

Regulation of cell migration is an important feature of tetraspanin CD151. Although it is well established that CD151 physically associates with integrins, the mechanism by which CD151 regulates integrin-dependent cell migration is basically unknown. Given the fact that CD151 is localized in both the plasma membrane and intracellular vesicles, we found that CD151 and its associated α3β1, α5β1, and α6β1 integrins undergo endocytosis and accumulate in the same intracellular vesicular compartments. CD151 contains a YRSL sequence, a YXXφ type of endocytosis/sorting motif, in its C-terminal cytoplasmic domain. Mutation of this motif markedly attenuated CD151 internalization. The loss of CD151 trafficking completely abrogated CD151-promoted cell migration on extracellular matrices such as laminin and diminished the internalization of its associated integrins, indicating a critical role for integrin trafficking in regulating cell motility. In conclusion, the YXXφ motif-mediated internalization of CD151 promotes integrin-dependent cell migration by modulating the endocytosis and/or vesicular trafficking of its associated integrins.

Tetraspanin CD151 is functionally linked to cancer metastasis, hemidesmosome formation, neurite outgrowth, and vascular morphogenesis (1)(2)(3)(4)(5). Although the mechanism remains obscure, cell movement appears to be the common theme for many of these events. The regulatory role of CD151 in cell migration is supported by the knock-out studies in mice (6,7) and knockdown experiments in vitro (8). Also, the association of CD151 with integrins, especially the laminin (LN) 2 -binding integrins, stands out as a prominent feature (3, 9 -11). CD151 is expressed in a variety of tissues and is especially abundant in epithelia, endothelia, cardiac muscle, smooth muscle, and megakaryocytes (12). Notably, the tissue distribution of CD151 is similar to the distribution of LN-binding integrins such as ␣3␤1, ␣6␤1, ␣6␤4, and ␣7␤1 (13), which is consistent with its ability to associate with these integrins and with a likely concomitant function during development (14,15). At the cellular level, CD151 is present on plasma membrane and in intracellular vesicles (1). On the cell surface, CD151 is localized in the basolateral membrane in endothelial and epithelial cells (1,16,17), whereas intracellularly, CD151 resides in endosomal/lysosomal vesicles (1).
Because CD151 directly associates with LN-binding integrins, it is not surprising that CD151 regulates cell-matrix adhesion. CD151 is needed for cell adhesion strengthening mediated by LN-binding integrin ␣6␤1 (18). CD151 also potentiates the ligand binding activity of integrin ␣3␤1 by stabilizing the activated conformation of this integrin (19). Furthermore, CD151 is a component of hemidesmosomes that anchor epithelial cells to basal lamina (3). Finally, CD151 was reported to up-regulate integrin ␣5␤1-dependent static adhesion to fibronectin (FN) (20). Interestingly, CD151 also plays a role in cell-cell adhesion (10). For example, CD151 promotes the formation of epithelial cell-cell contacts in a protein kinase C (PKC) and Cdc42-dependent manner (17). CD151 recruits integrin ␣3␤1 to cell-cell contacts to form a multimolecular complex that includes PKC, phosphotyrosine phosphatase , E-cadherin, and ␤-catenin (21). CD151 contains two extracellular loops and two short cytoplasmic tails. The large extracellular loop is required for the association with ␣3 and ␣6 integrins (22,23). The CD151 C-terminal cytoplasmic domain plays a key role in signaling. The deletion or exchange of the CD151 C-terminal cytoplasmic tail markedly impaired integrin ␣6␤1-dependent cell spreading and cellular morphogenesis (5,18). In addition, the C-terminal cytoplasmic tail of CD151 negatively regulates Ras activity (24). Moreover, the C-terminal tail determines integrin ␣6␤1-mediated adhesion strengthening to LN-1 (18). Despite several lines of evidence indicating an essential role of the C-terminal tail of CD151, it is still unclear how only eight amino acids at the C-terminal tail regulate these CD151-mediated activities.
The C-terminal cytoplasmic domain of CD151 contains a YRSL sequence that meets the criteria of the YXX-sorting motif, in which Y is tyrosine and X indicates any amino acid, and represents the amino acid residue with a bulky hydrophobic side chain (25,26). The YXX motifs in the cytoplasmic domain of transmembrane proteins are recognized by adaptor protein-2 complex, a core component of clathrin endocytic machinery (26,27). This led us to hypothesize that the YRSL motif drives CD151 endocytosis, is responsible for the func-tional importance of CD151 C-terminal cytoplasmic domain, and ultimately regulates CD151 activities such as integrin-dependent cell movement.

MATERIALS AND METHODS
Cells and Transfectants-NIH3T3 and Du145 cell lines were obtained from the ATCC and cultured at 37°C in 5% CO 2 in DMEM (Invitrogen) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Human dermal microvascular endothelial cells (HMEC-1) were obtained from the Centers for Disease Control and Prevention and cultured in ECBM-2 media (Cambrex).
CD151 wild type and the YRSL 3 ARSA mutant (termed the Yala mutant) were constructed into a eukaryotic expression vector pZeoSV2 (Invitrogen) as described in our earlier study (5). Briefly, the CD151 Yala mutant was generated by simultaneously replacing the Tyr and Leu residues in the YRSL sequence of wild type CD151 C-terminal cytoplasmic domain with alanine residues through PCR. The full-length cDNAs encoding human wild type and Yala mutant CD151 were constructed into pZeoSV2 vector, respectively, through the XbaI and HindIII sites. The CD151 coding information was confirmed by DNA sequencing. NIH3T3 cells were transfected with plasmid DNA using Lipofectamine2000 (Invitrogen) and selected with Zeocin (Invitrogen) at a concentration of 0.2 mg/ml. Hundreds of Zeocin-resistant clones were pooled, and CD151-positive clones were collected by flow cytometric cell sorting. The pooled and sorted CD151-positive clones were the CD151 wild type and Yala stable transfectants.
Flow Cytometry-NIH3T3 transfectant cells were detached with 2 mM EDTA/PBS, washed once with PBS, treated with DMEM supplemented with 5% goat serum at 4°C for 10 -15 min, and then incubated with primary mAb at 4°C for 1 h. After washes, cells were further labeled with FITC-conjugated second antibody at 4°C for 1 h and then analyzed on FACSCalibur flow cytometer.
Immunofluorescence and Confocal Microscopy-Cells were fixed with 3% paraformaldehyde at room temperature for 10 min, permeabilized with 0.1% Brij 98 at room temperature for 2-5 min, blocked with 20% goat serum at room temperature for 1 h or at 4°C overnight, and then incubated with primary mAbs at room temperature for 30 min, followed by staining with a secondary antibody at room temperature for 30 min. After each antibody incubation, the cells were washed with PBS five times; each wash lasted 15 min. For double staining, the cells were blocked with murine IgG and then further stained with Alexa 488-or Alexa 594-conjugated human CD151 mAb 5C11. For immunofluorescent analysis, the cells were examined with an Axiophot fluorescent microscope (Carl Zeiss), and images were captured using an Optronics digital camera. For confocal microscopic analysis, the coverslips were examined using LSM 510 laser scanning microscope (Carl Zeiss) equipped with 63ϫ/ 1.4 Plan-APOCHROMAT oil immersion objectives. Green fluorescent protein, FITC, or Alexa 488 and rhodamine or Texas Red were excited by the 488-and 543-nm lines of Kr-Ar lasers, respectively, and individual channels were scanned in a series to prevent cross-channel bleed through. Each image represents a stack of ϳ1.0 m "Z" optical sections of the cells.
Immunoprecipitation-Immuoprecipitations were carried out as described previously (29). Briefly, cells were lysed with 1% Brij 97 and/or 1% Nonidet P-40 lysis buffer at 4°C for 1 h. In some experiments, the cell surface was labeled with 0.5 mg/ml EZlink sulfo-NHS-LC biotin (Pierce) before cell lysis. After removing the insoluble material by 14,000 ϫ g centrifugation, the precleared lysates were incubated with primary mAb preabsorbed protein A-and G-Sepharose beads (Amersham Biosciences) from 3 h to overnight at 4°C. The precipitates were washed with the lysis buffer three times, dissolved in Laemmli sample buffer, heated at 95°C for 5 min, separated by SDS-PAGE, and then electrically transferred to nitrocellulose membranes (Bio-Rad). For immunoblot experiments, the membranes were sequentially blotted with primary Ab and horseradish peroxidase-conjugated second Ab (Sigma), followed by chemiluminescence (PerkinElmer Life Sciences). For biotinylation experiments, the membranes were blotted with horseradish peroxidase-conjugated extravidin (Sigma), followed by chemiluminescence.
Internalization Assay-The internalization of CD151 and integrins was examined with antibody uptake assay. Cells were incubated with fluorochrome-conjugated CD151 mAb and/ or integrin mAb at 4°C for 1 h. The unbound antibodies were removed by three washes with serum-free medium. The cells were then incubated at 37°C for internalization for typically 30 or 60 min or for various periods of times as specified in the particular experiments. The antibodies that bound to the cell surface and were not internalized were removed by acid washes with the 0.1 M glycine, 0.1 M NaCl buffer (pH 3.0). The acid wash typically removed ϳ80 -90% of the cell surface-bound antibodies. The cells were then fixed and examined with immunofluorescent or confocal microscopy.
The kinetics of integrin and CD151 internalization were determined following the established protocol (29 -33). Cells were incubated with reducible sulfo-NHS-SS-biotin at the saturated concentration (0.5 mg/ml) at 4°C to label the cell surface proteins. The labeling lasted for 1 h to allow maximal biotinylation of the cell surface proteins. The free biotin was removed by three washes with ice-cold PBS. The labeled cells were incubated at 37°C for various times for the internalization and then treated with a reducing solution for 20 min on ice to remove cell surface-bound biotin. For the internalization of CD151 and its associated integrin, the reducing solution contained 42 mM GSH, 1 mM EDTA, 75 mM NaOH, and 1% bovine serum albumin. After washing with PBS, the cells were lysed in 1% Nonidet P-40, 1% Brij 97 lysis buffer at 4°C for 1 h, and CD151 was immunoprecipitated as aforementioned. After SDS-PAGE and electric transferring, the biotinylated integrin or CD151 was detected by peroxidase-conjugated extravidin and visualized by chemiluminescence. The integrin and CD151 bands were quantitated by densitometric analysis as described previously (29). Integrin internalization at an indicated time point was calculated with the following formula: (the densitometry of internalized integrin at that time point Ϫ the densitometry of internalized integrin at time 0)/the total surface integrin ϫ 100%.
Cell Migration Assay-Transwell cell migration assay was performed as described previously (34). Briefly, cells were detached from culture plates with EDTA, washed once with PBS, and replated onto the inserts of 8-m pore size Transwell chambers (Costar). The undersides of the inserts were precoated with either FN (10 g/ml) or LN 1 (10 g/ml) at 4°C overnight and blocked with 0.1% heat-inactivated bovine serum albumin at 37°C for 45 min. The cell number was 8 ϫ 10 3 per insert filter. The media include 500 l of DMEM containing 1% fetal calf serum in the lower well and 300 l DMEM containing 0.1% heat-inactivated bovine serum albumin in the insert. After incubating at 37°C for 3 h, cells that had not migrated through inserts were removed, and cells that had migrated to the undersides of the inserts were fixed and stained with Diff-Quick (Baxter Merz & Dade). The cells that had migrated through the insert were counted. Data from independent experiments were pooled and analyzed using two-tailed Student's t test.
For wound healing assay, NIH transfectant cells were cultured in Lab-Tek chamber slides (Nunc, Rochester, NY). After the monolayers were formed, wounds were generated by scraping the monolayers with sterile pipette tips. The detached cells were rinsed away with PBS, and the wounded monolayers were replenished with the complete medium. After overnight culture at 37°C, the monolayers were fixed and photographed under inverted light microscope (Olympus, Center Valley, PA).

RESULTS
The YRSL Sequence in CD151 Cytoplasmic Domain Determines Its Trafficking-To determine whether the YRSL sequence of CD151 (Fig. 1A) is involved in regulating its intracellular trafficking and function, we mutated this motif in human CD151 molecule. The Tyr and Leu residues in CD151 YRSL motif were replaced with alanine residues simultaneously, and the resulting YRSL 3 ARSA mutant was designated as the Yala mutant. We transfected human CD151 WT and Yala mutant cDNA into NIH3T3 mouse fibroblast cells and established stable transfectants by pooling and sorting hundreds of CD151-expressing clones by flow cytometry. As shown in Fig. 1B, CD151 WT and the Yala mutant were stably expressed at comparable levels on the surface of NIH3T3 cells when cells reached confluence in culture. When cells were less confluent in culture, the cell surface level of CD151 on the Yala mutant was higher than that on the wild type. For example, when cells were at the 60 -70% confluent stage, the mean fluorescent intensity of CD151 was 134 in wild type and 189 in the Yala mutant. The expression levels of other cell surface proteins such as CD44 (Fig. 1B, top panel) were equivalent on all transfectants and not altered upon CD151 expression. The total cellular levels of CD151 proteins were equivalent between CD151 wild type and the Yala transfectants, with a slightly higher level in the wild type (densitometry of CD151 band in the wild type: that in the Yala mutant was 1:0.88) (Fig. 1B, bottom panel). The total CD151 proteins remained stable at different stages during cell culture.
The effect of the Yala mutation on CD151 subcellular localization was analyzed at the steady state (Fig. 1C). Typically, CD151 proteins are localized on the cell surface and in intracellular endosomes and lysosomes (1). 3 CD151 Yala mutant was distributed mainly at the cell surface and largely lost the vesicular staining observed with CD151 WT, suggesting that the YRSL motif is important for regulating CD151 localization. In NIH3T3-CD151 wild type transfectant cells, we confirmed that some of CD151-positive intracellular vesicles were EEA1-or Rab5-positive early endosomes (Fig. 1C, bottom panel).
We further analyzed the internalization of CD151 in the WT and Yala transfectants using an antibody uptake assay. After a 30-min endocytosis, a substantial fraction of WT CD151 was internalized and accumulated in intracellular vesicles, whereas CD151 Yala mutant was barely internalized in this time frame (Fig. 1D, top panel). The quantitative analyses showed the significant diminution of CD151 endocytosis in the Yala mutant (Fig. 1D, bottom panel). A small fraction of the CD151 Yala mutant proteins were internalized after prolonged incubation at 37°C, such as 2 h, and the internalized CD151 Yala mutant proteins remained at the cell peripheral area (data not shown), suggesting a deficiency in the trafficking following endocytosis. Thus, the YRSL motif is required for CD151 endocytic processes.
CD151 Trafficking Regulates Integrin-dependent Cell Migration-Regulation of cell motility is a prominent feature of CD151. We tested whether CD151 internalization is important for cell migration. We measured the directional motility of NIH3T3-CD151 transfectants on the substratum coated with FN or LN1, a ligand of ␣5 or ␣6 integrins, respectively. Because the ␣6 integrin expressed in NIH3T3 cells is the ␣6␤1 integrin (5), cell migration on LN1 is mediated primarily by integrin ␣6␤1. As shown in Fig. 2A, the expression of WT CD151 in NIH3T3 significantly enhanced cell migration on both FN and LN1, consistent with previous results (1,2,9,16). The ability of CD151 Yala mutant cells to migrate toward FN and LN1 was markedly reduced compared with that of CD151 WT and Mock transfectant cells. The diminished cell migration of the Yala mutant indicates that YRSL internalization signal is not only required for CD151 pro-migratory activity but also affects constitutive motility of NIH3T3 cells. We also used wound healing assay to assess the effect of the Yala mutation on cell migration. Consistent with the results obtained from the transwell cell migration assay, CD151 WT promoted wound healing, whereas CD151 Yala mutant exhibited delayed wound healing compared with the CD151 WT transfectant (Fig. 2B). Thus, CD151 endocytosis plays an essential role in regulating integrin-dependent cell migration.
Deficient Endocytosis of CD151 Does Not Alter the Integrin-CD151 Association-To determine why the loss of CD151 internalization affects integrin-mediated cell migration, we first assessed if CD151 Yala mutant altered the CD151-integrin complex. We compared the CD151integrin associations between NIH3T3-CD151 WT and Yala transfectants. CD151 immunoprecipitation profiles of surface-biotinylated NIH3T3 cells revealed that integrins coprecipitated equally well with CD151 WT and the Yala mutant under either a stringent lysis condition, i.e. 1% Nonidet P-40 ( Fig.  3A), or a mild lysis condition, i.e. 1% Brij 99 (data not show). Immunoblotting analysis indicated that the coprecipitated integrin under the stringent lysis condition was endogenous murine ␣3, and the ones under the mild conditions were ␣5 and ␣6 (Fig. 3B). The transmembrane protein CD44 (Fig. 3A) and housekeeping protein tubulin ( Fig.  3B) were included to indicate the equal loadings of the cell lysate. The blotting of CD151 immunoprecipitates with CD151 mAb ensured that the biotinylated ϳ30-kDa band in CD151 immunoprecipitate is indeed CD151 (Fig. 3A, bottom panel). Together, these results indicate that the Yala mutation did not alter the association of CD151 with cell surface integrins and that CD151 endocytosis is not required for CD151-integrin complex formation. In addition, the association of CD151 with endogenous murine tetraspanins such as CD9 was not altered by the Yala mutation (Fig. 3C). Thus, CD151 endocytosis does not appear to be critical for the partitioning of CD151 into tetraspanin webs (34) or tetraspanin-enriched microdomains (TEM) (36).

Integrin Is Internalized into CD151-containing Vesicular
Compartments-Based on the fact that CD151 physically associates with integrins, we investigated whether CD151-associated integrins could also be internalized. To address this issue, we used human dermal microvascular endothelial cells (HMECs) that express CD151 and ␣3␤1, ␣5␤1, and ␣6␤1 integrins. As shown in Fig. 4, ␣3␤1, ␣5␤1, and ␣6␤1 integrins can be detected in the intracellular vesicles of HMECs after inter-nalization. Moreover, a significant fraction of the internalized ␣3␤1, ␣5␤1, and ␣6␤1 integrins accumulated in intracellular vesicles that also contained internalized CD151 (Fig. 4). These results suggest that CD151 was cointernalized with its associated integrins, and CD151-integrin complexes traffic through, at least in part, the same endocytic compartments. In 84% cells examined, the endocytic vesicles positive for both CD151 and integrin ␣3 were detected. These vesicles were distributed throughout cytoplasm with a higher density in the perinuclear area. Approximately 33.3% of the intracellular vesicles positive for internalized CD151 also contained the internalized integrin ␣3 (Table 1, ImageJ analysis) (ImageJ, National Institutes of Health, Bethesda) (38). The staining of internalized ␣3 or CD151 in cointernalization experiments was less than the staining when the internalization was performed individually for ␣3 or CD151. This is probably caused by the steric hindering effect of the ␣3 mAb-␣3 binding on the CD151 mAb-CD151 binding and/or vice versa because of the tight association of ␣3 and CD151. Thus, the colocalization of internalized ␣3 or CD151 could be substantially underestimated in this cointernalization experiment. Endocytic vesicles that contained both internalized CD151 and internalized integrin ␣5 were found in 49% cells and were mainly located in the perinuclear area (Fig.  4). In intracellular vesicles, the internalized integrin ␣5 and internalized CD151 are partially colocalized with each other ( Table 1). The endocytic vesicles that contain both internalized CD151 and internalized integrin ␣6 were found in 80% of cells and were also primarily perinuclear (Fig. 4). In the majority of cells, the internalized integrin ␣6 and internalized CD151 were partially colocalized (Table 1, ImageJ analysis), whereas in ϳ10% of the cells integrin ␣6 and CD151 were fully colocalized after internalization. These integrin-and CD151-positive vesicles belong to an unidentified vesicular compartment. 3 The colocalization of internalized CD151 with internalized integrins was quantitated using both visual counting and ImageJ software (Table 1), and the results obtained from both measurements were consistent with each other. The extent of colocalization of CD151 with the internalized ␣3, ␣6, or ␣5 integrin also correlated with the stoichiometries and stability of the physical associations of CD151 with these integrins. We then analyzed the colocalization of internalized CD151 and internal-FIGURE 1. The YXX motif of CD151 is required for its internalization. A, schematic representation of CD151 YXX motifs. The C-terminal cytoplasmic domain of human CD151 contains eight amino acid residues, in which the YRSL sequence is a YXX internalization motif. The C-terminal cytoplasmic domains and partial 4th transmembrane domains of vertebrate CD151 proteins are aligned, and the YXX motifs are boxed. The asterisks and colons denote the identical and conserved residues among the aligned sequences, respectively. B, expression of CD151 WT and Yala mutant in NIH3T3 cells. Top panel, the Tyr and Leu residues were simultaneously replaced by an A residue, and the YRSL 3 ARSL mutant is designated as Yala. The NIH3T3-Mock, -WT, and -Yala transfectant cells were detached, incubated with a negative control mAb mouse IgG2b (solid line), a positive control mAb anti-mouse CD44 (black shading), and a human CD151 mAb (red shading), respectively, and then stained with a FITC-conjugated second Ab. The stained cells were analyzed by flow cytometry. The mean fluorescence intensity of CD151 staining for NIH3T3-Mock, -CD151 WT, and -CD151 Yala mutant was 8, 113, and 121, respectively. Bottom panel, NIH3T3 transfectant cells were lysed in 1% Nonidet P-40 lysis buffer, and CD151 proteins were detected via Western blot using human CD151 mAb 8C3. The tubulin levels were used as a loading control. C, YXX motif determines cellular distribution of CD151. NIH3T3 transfectant cells were plated on FN-coated coverslips and incubated in serum-free DMEM at 37°C for 3 h. Then cells were fixed, permeabilized, and sequentially stained with CD151 mAb and rhodamine-conjugated second Ab. Bottom panel, the colocalization of CD151 with early endosome markers EEA1 and Rab5 were assessed using Alexa594-conjugated CD151 mAb and FITCconjugated EEA1 mAb NIH3T3-CD151 wild type transfectant cells. Digital images were captured with a fluorescent microscope (ϫ63). Examples of vesicular colocalization are indicated with arrows. D, the YXX motif is required for CD151 internalization.   OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43

JOURNAL OF BIOLOGICAL CHEMISTRY 31635
ized CD71 (Fig. 4) and found colocalization in less than 20% of HMECs ( Table 1). The endocytosis of CD71, the transferrin receptor, was assessed by using transferrin as a tracer. Moreover, the extent of colocalization in those 20% cells is markedly lower than the degree of the colocalization between internalized CD151 and the integrins ( Table 1, Image J analysis).
CD151 Regulates the Endocytosis of Its Associated Integrin-Because CD151 and its associated integrins can be internalized into the same intracellular vesicular compartment and CD151 Yala mutant is still able to associate with integrins, the attenuation in CD151 internalization by the Yala mutation may result in the impaired endocytosis of its associated integrins. We thus investigated whether the endocytic defect of the CD151 Yala mutant affected the endocytosis of its associated integrins. A, NIH3T3-Mock, -CD151 WT, and -CD151 Yala transfectant cells were biotinylated and subsequently lysed in 1% Nonidet P-40 lysis buffer. Cell lysates were immunoprecipitated (IP) with CD151 mAb. The precipitated proteins were resolved by nonreducing SDS-PAGE. Following transfer to nitrocellulose, the membrane was blotted with extravidin, and the proteins visualized by chemiluminescence. The CD44 immunoprecipitates were included as protein loading controls (middle panel). The CD151 immunoprecipitates were blotted with CD151 mAb 8C3 to confirm the presence of CD151 in precipitates (bottom panel). B, CD151 immunoprecipitates were blotted with a pAb against human/murine integrin ␣3 cytoplasmic tail, a pAb against murine integrin ␣5, and a pAb against human/ murine integrin ␣6 cytoplasmic tail, respectively. For integrin ␣3 coprecipitation, cells were lysed in 1% Nonidet P-40; whereas for integrins ␣5 and ␣6 coprecipitations, cells were lysed in 1% Brij 98. The cell lysates were blotted for tubulin as a loading control. C, NIH3T3 transfectant cells were lysed in 1% Brij 98 lysis buffer, and the cell lysates were immunoprecipitated with human CD151 mAb 5C11. The precipitated proteins were resolved by nonreducing SDS-PAGE. Following transfer to nitrocellulose, the membrane was blotted with murine CD9 mAb KMC8, and the CD9 proteins were visualized by chemiluminescence. The cell lysates were blotted for tubulin to demonstrate the equal loading. For the cointernalization of integrin ␣3, ␣5, or ␣6 and CD151, HMECs grown on glass coverslips overnight were incubated with FITC-conjugated ␣3, ␣5, or ␣6 mAb and Alexa 594-conjugated CD151 mAb 5C11 at 4°C for 1 h. The FITC-conjugated mAbs against human integrins were integrin ␣3 mAb X8, integrin ␣5 mAb CBL497F, and integrin ␣6 mAb GoH3. For the control, cells were incubated with Alexa 488-conjugated transferrin and Alexa 594-conjugated CD151 mAb. After the unbound mAbs and transferrin were removed by washes with cold PBS, the cells were then incubated at 37°C for 1 h for internalization. The cells were treated with two consecutive acid washes to remove the cell surface uninternalized mAbs or transferrin. The confocal images were captured at the magnification ϫ63.
First, we measured the internalization kinetics of relevant integrins in NIH3T3 mock, CD151 wild type, and CD151 Yala transfectants. For the ␣3 integrin, because no workable murine integrin ␣3 mAb was available, we could not measure the kinetics of ␣3 internalization in the antibody uptake endocytosis assay. For ␣6 and ␣5 integrins, no significant differences were detected in endocytoses of the total cell surface integrin ␣6␤1 or ␣5␤1 between CD151 WT and the Yala transfectants (Fig.  5A), indicating that the Yala mutation did not alter the global endocytoses of these integrins. It is not surprising because only ϳ20% of total cell surface integrin ␣6 is associated with CD151 in NIH3T3 cells, and the stoichiometry for integrin ␣5 may be even lower. The attenuated endocytosis of CD151 may not result in marked changes in the global endocytosis of these integrins. We then assessed the endocytoses of CD151-associated ␣6 and ␣5 integrins in CD151 transfectants. As shown in Fig.  5B, the internalized integrins were probed with specific mAbs conjugated with green fluorochrome, whereas total cellular CD151 was stained in red to display the cellular distribution of CD151. The ␣6 and ␣5 integrins could still undergo endocytosis in the Yala mutant. However, the extent of internalizations and possibly subsequent trafficking of CD151-colocalized ␣6 and ␣5 integrins (Fig. 5B, shown in yellow) were significantly diminished in the Yala mutant. After acid washing the cell surface, uninternalized integrins were completely removed in WT cells but not in the Yala mutant cells, probably because more integrins are accumulated on the cell surface in the Yala mutant after 1 h of endocytosis. The CD151-colocalized, internalized ␣6 and ␣5 integrins mainly distributed in pericentriolar areas in CD151 WT cells, but CD151-colocalized ␣6 and ␣5 integrins were accumulated in cell periphery areas in the Yala cells (Fig.  5B). Because CD151-colocalized integrins are likely to be CD151-associated integrins, this result strongly suggests that the endocytosis and/or the trafficking immediately after endocytosis of CD151-associated ␣5 and ␣6 integrins becomes deficient when CD151 endocytosis is attenuated. No difference in the distributions of the internalized CD71 was found between NIH3T3-CD151 wild type and the Yala mutant (Fig. 5B), supporting the notion that the CD151 Yala mutation specifically alters the internalization of CD151-associated integrins.
To ensure that the endocytosis of CD151-associated integrins is indeed attenuated, we compared the endocytic kinetics of CD151-associated integrins between CD151 WT and the Yala transfectants. By labeling the cell surface with a reducible biotin, we determined the internalized integrin based on the inaccessibility of integrin to biotin release upon reduction. As shown in Fig. 5C, the internalization of CD151-associated integrin in the Yala mutant was indeed attenuated, compared with that in the WT transfectant. The internalization rate of the integrin that physically associates with CD151 Yala mutant was significantly lower than the rate of the integrin that physically associates with CD151 WT. The actual quantities of the CD151-associated, internalized integrins at each of the time points were also less in the Yala than in the WT transfectant, and the differences were particularly marked between 15 and 45 min. It has been demonstrated in earlier studies that, at 1% Nonidet P-40 lysis condition, the biotinylated ␣3 integrin is the only CD151-associated protein that is detectable in the biotinavidin blot (9,22). Because NIH3T3 transfectant cells were lysed with 1% Nonidet P-40 detergent in the internalization experiments, we predict that CD151-associated integrin was murine ␣3 integrin. Using a pAb that was raised with human ␣3 cytoplasmic domain but that cross-reacts with murine ␣3 cytoplasmic domain, we confirmed that the CD151-associated integrins were indeed ␣3 integrin by immunoblotting (Fig. 5C). CD151-associated total ␣3 integrin and the level of tubulin in cell lysates were included to ensure equal loading at each time point.
Finally, we analyzed the internalization kinetics of CD151associated ␣6 and ␣5 integrins in NIH3T3-CD151 wild type and -CD151 Yala transfectants. CD151-associated ␣6 and ␣5 integrins were re-immunoprecipitated using their specific mAbs, respectively, from the CD151 immunoprecipitates that were performed under 1% Brij 98 cell lysis condition. Similar to ␣3 integrin, we found that the internalization of CD151-associated ␣6 and ␣5 integrins was diminished in the Yala mutant, especially evident between 15 and 45 min (Fig. 5D), indicating that the Yala mutation also inhibits the endocytosis of CD151associated ␣6 and ␣5 integrins.
Besides interacting with integrins, CD151 also physically associates with tetraspanins such as CD9 and CD81 (35,39). To investigate if the defect in CD151 endocytosis affects the internalizations of tetraspanins, we analyzed the endocytosis of endogenous CD9 and CD81 in NIH3T3-CD151 transfectants. For CD9, we found that no CD9 was internalized in 60-min endocytosis experiments in ϳ70% WT and Yala cells. In ϳ30% cells, the internalized CD9 was detected in intracellular vesicles (Fig. 6A). In these cells, CD9 was internalized at similar degrees between CD151 WT and the Yala mutant cells (Fig. 6B), suggesting that CD9 endocytosis is not altered by the CD151 Yala mutation. The expression level of CD81 in NIH3T3 cells appears to be negligible because we barely detected any CD81 protein expression using a workable murine CD81 mAb (40). Thus, the effect of the CD151 Yala mutation on CD81 endocytosis cannot be assessed in NIH3T3 cells.

DISCUSSION
The YXX Motif-mediated Endocytosis and Trafficking of CD151-As predicted, the YXX motif is a crucial intrinsic determinant governing CD151 internalization. Mutation of the CD151 YXX motif resulted in diminished internalization and subsequent endocytic trafficking of CD151 and in increased  (41), A15, 4 and CD82 4 when the YXX motifs in the C-terminal cytoplasmic tail were mutated. Thus, using the YXX motifs to traffic between different cellular compartments appears to be a common mechanism for the tetraspanins that contain this motif. For the tetraspanins that do not contain the YXX motifs, such as CD9 and CD81, their endocytosis may depend on unidentified intrinsic sorting signals or on their directly associated tetraspanins that contain this motif, such as CD63, CD82, and/or CD151. For example, we found in this study that CD9 underwent endocytosis, but its endocytosis was not markedly altered by the CD151 Yala mutation. In this case, CD9 endocytosis may be determined by its own endocytosis signal that remains to be identified. In terms of the mechanism that governs CD151 endocytosis, we have demonstrated elsewhere that CD151 is internalized via a dynamin-and Arf6-independent but phosphatidylinositol 3-kinase-and actin-dependent endocytosis pathway. 3 Although the defect in endocytosis is obvious in the CD151 Yala mutant, the Yala effect was more profound when endocytosis was measured using antibody uptake than using biotinylation (data not shown). It is possible that the antibody causes the cross-linking of the cell surface CD151 or that the pH conditions during the cell surface stripping are different in these two kinds of endocytosis assays. The clustering of cell surface CD151, however, likely occurs physiologically, because the engagement of its associated integrins with multivalent ECM ligands or the reorganization of TEM can induce CD151 clustering. Also, if the ligand or counter-re-4 X. A. Zhang, unpublished data. FIGURE 5. The internalization of CD151-associated integrins is attenuated by the CD151 Yala mutation. A, comparisons of integrin internalization kinetics between NIH3T3-CD151 transfectants. NIH3T3-CD151 wild type and Yala cells were biotinylated with reducible NHS-SS-biotin at 4°C for 1 h, followed by thorough washes to remove free biotin. An aliquot of cells was lysed immediately in RIPA buffer. Other aliquots of cells were incubated at 37°C for various times to allow the internalizations of cell surface proteins, and the cells were then treated with a reducing solution at 4°C to remove all biotin remaining at the cell surface. After cell lysis, the cellular integrins ␣6 and ␣5 were immunoprecipitated from each aliquot, resolved in nonreducing SDS-PAGE, detected by peroxidase-conjugated extravidin, and visualized by chemiluminescence. The internalized ␣5 or ␣6 integrin bands were quantitated by densitometric analysis. The internal-to-surface ratios of ␣5 or ␣6 were calculated from the internalized ␣5 or ␣6 as a percentage of the steady-state surface ␣5 or ␣6 and plotted as a function of endocytosis time. There are no statistically significant differences in integrin ␣5 or ␣6 endocytoses between CD151 wild type cells and the Yala cells at any time point, respectively. B, effect of CD151 Yala mutation on integrins ␣5 and ␣6 internalization. NIH3T3 cells were plated on FN-or LN1-coated coverslips for 3 h at 37°C and then incubated with murine integrin ␣5 or ␣6 mAb or Alexa 488-conjugated transferrin for 1 h at 4°C. After the unbound Abs or transferrin were removed by three washes, the cells were incubated at 37°C for 1 h for internalization, followed with the acid washes to remove uninternalized Abs or transferrin. Then, for integrin groups, cells were fixed, permeabilized, and stained with Alexa 488-conjugated second Ab. After thorough washes and being blocked with murine IgG, the cells were further stained with Alexa 594-conjugated CD151 mAb. The images were acquired under confocal microscopy as described above and were the X-Z section stacks. C, endocytosis kinetics of CD151-associated ␣3 integrin. NIH3T3-CD151 WT and -CD151 Yala cells were biotinylated with reducible NHS-SS-biotin at 4°C for 1 h, followed by thorough washes to remove free biotin. An identical number of cells from each transfectant were used for biotinylation. For each transfectant, the biotinylated cells ware evenly divided into 7 aliquots. An aliquot of cells was lysed immediately in the lysis buffer containing detergents 1% Nonidet P-40 and 1% Brij97. Other aliquots of cells were incubated at 37°C for various times to allow the internalizations of cell surface proteins, and the cells were then treated with a reducing solution at 4°C to remove all biotin remaining at the cell surface. After cell lysis, integrin was coprecipitated with CD151 mAb 5C11 from each aliquot and resolved in nonreducing SDS-PAGE. The CD151-associated and internalized integrin was detected by peroxidase-conjugated extravidin and visualized by chemiluminescence. The total integrins that were coprecipitated with CD151 were detected with the pAb against integrin ␣3 cytoplasmic domain. The cell lysates from each time point were probed for an ␣-tubulin antibody to ensure equal loading. Surf denotes an aliquot of cells without the cell surface reduction treatment. The internalized CD151-associated ␣3 integrin bands were quantitated by densitometric analysis. The internal-to-surface ratios of ␣3 were calculated from the net internalized ␣3 as a percentage of steady-state surface ␣3 and plotted as a function of endocytosis time. Statistically, the internalization of integrin ␣3 in CD151 wild type cells at 30 min is significantly higher than that in the Yala cells, respectively (p value Ͻ0.01). D, endocytic kinetics of CD151-associated ␣5 and ␣6 integrins. The CD151-associated and internalized ␣5 or ␣6 integrins were collected from the CD151 immunoprecipitates by re-immunoprecipitation using murine ␣5 or ␣6 integrin mAb, respectively. CD151 immunoprecipitations (I.P.) were performed using biotinylated lysates prepared under the 1% Brij 98 lysis conditions and the CD151-associated integrin ␣5 or ␣6 was re-immunoprecipitated after the CD151-integrin ␣5 or ␣6 complexes were disrupted by 1% Nonidet P-40. After electrophoresis, extravidin blotting, and chemiluminescence, the internalized and CD151-associated ␣5 or ␣6 integrin bands were quantitated by densitometric analysis. The internal-to-surface ratios of CD151-associated ␣5 or ␣6 were calculated from the net internalized ␣5 or ␣6 as a percentage of the steady-state surface, CD151-bound ␣5 or ␣6, and plotted as a function of endocytosis time. Statistically, the internalizations of integrin ␣5 or ␣6 in CD151 wild type cells at 30 min are significantly higher than those in the Yala cells, respectively, (p values ϭ 0.027 and 0.014 for integrin ␣5 or ␣6, respectively). In the top panel, cell lysates from each time point were probed for an ␣-tubulin antibody to ensure equal loading. Surf denotes an aliquot of cells without the cell surface reduction treatment. ceptor of CD151 is present, the antibody may imitate the CD151 ligand or counter-receptor to elicit CD151 signaling and trigger its internalization. Thus, the CD151 internalization measured by either antibody uptake or biotinylation could reflect its endocytosis under different physiological settings.
Tetraspanin and integrin have been shown previously to colocalize in the vesicle-like intracellular structure in an earlier study (42). Upon PKC inhibition, integrin ␤1 can actually be trapped in an unidentified recycling compartment that contains tetraspanin CD81 (43). However, it is not known whether they are internalized via the same endocytic pathway and traffic through the same endocytic compartment. Our study for the first time demonstrates that tetraspanin and integrin are possibly delivered to the same vesicular compartment following endocytosis, although this remains to be confirmed by immunoelectron microscopy in the future studies. The colocalization of CD151 and its associated integrins in endocytic vesicles implies that they are either internalized through the same endocytosis pathway or trafficked to the same intracellular vesicular compartment after differential internalizations. Because CD151 physically associates with integrin, it is more likely that they share the same endocytosis pathway or internalization route. Thus, the cotrafficking of tetraspanin with integrin could be an immediate functional consequence of the well documented tetraspanin-integrin physical association.
CD151 Trafficking Regulates Cell Migration-CD151 regulates cell migration, but the mechanism remains elusive. Notably in this study, when the YXX motif was mutated, the WT CD151-promoted cell migration on FN and LN1 was completely inhibited. This reduction in cell migration correlated with the marked reduction in the internalization of the CD151 Yala mutant. These results strongly suggest that CD151 endocytosis and/or trafficking is needed for integrin-mediated cell migration. Meanwhile, our study is the first to demonstrate that the trafficking of tetraspanin can play a key role in regulating cell migration. Endocytosis of other transmembrane proteins such as MT1-MMP has also been found to regulate cell migration (44). For motility-related transmembrane proteins, it appears that their endocytosis takes part in the motility regulation. A role for vesicular trafficking in regulating cell migration had been postulated long ago (45). On the one hand, disrupted endocytosis is accompanied by a deficiency in cell migration (46). On the other hand, polarized exocytosis occurs at the leading edge of motile cells and may assist cells migrating forward (47). Regulating cell motility is a common feature of most tetraspanins, and tetraspanins such as CD63, CD82, A15, and CD151 are present in intracellular vesicles (39,48). Vesicular trafficking could therefore be a common point that tetraspanins act on for regulating cell motility.
How does CD151 internalization regulate cell migration? We postulated that the association of CD151 with tetraspanin webs (35) or glycosynapse (49) or TEM (36), which are mainly formed by tetraspanins, cell adhesion molecules, and glycosphingolipids, holds the key. CD151 internalization possibly induces endocytosis of its associated tetraspanin web or glycosynapse or TEM, and the latter in turn facilitates the recycling and reutilization of TEM components. Indeed, we demonstrated in this study that trafficking of integrin, a TEM component, was significantly affected by CD151 (see following). Likely, CD151 also regulates the trafficking of other TEM components. If it is true, the directionality of CD151 trafficking in migrating cells, the cotrafficking of TEM components with CD151, and the identity of CD151-containing vesicles become key issues to be addressed in future studies.
The Role of CD151 Trafficking in Integrin Function-Although the traction force generated from integrin-ECM engagement is crucial for propelling cells to move, integrin disengagement from ECM during cell migration is equally important as its engagement in maintaining motility (50). The ECMdisengaged integrins in the rear of migrating cells could be diffused laterally across the cell surface through the plasma membrane and/or internalized and redelivered to the cell surface (51). The observations of integrin endocytosis and recycling of integrin-containing vesicles at the cell rear or in circulation (30,52) led to the notion that integrins are internalized from plasma membrane at the rear of migrating cells and likely recycled back to plasma membrane at the leading edge to engage with ECM again (53). The vesicle trafficking of integrins therefore is postulated to regulate cell migration (54). Indeed, integrin distribution is polarized in migrating cells and maintained by the cycles of endocytosis from the rear and recycling to the front (55). Also during cell migration, integrin traffics via endocytic and recycling vesicles from the rear portion of a cell through the perinuclear region to the base of the leading lamellipodia (56 -58). The structural elements in integrin responsible for its trafficking in cell migration have been elucidated in recent studies. For example, the YXX motif in the integrin ␤2 subunit is essential for recycling of the internalized ␤2 integrins, and a mutation of this motif disrupted integrin ␤2-dependent cell migration (33,59). The trafficking pathways of integrin are also becoming better understood (37, 43, 59 -65). For example, the PKC-dependent internalization of integrin ␤1 in MCF-7 breast cancer cells requires dynamin activity, and the dominant-negative dynamin inhibits the PKC-promoted, integrin-mediated cell motility on collagen (60).
Regulation of cell migration via CD151 trafficking raised a question whether integrin trafficking is involved in this regulatory process. The direct association of CD151 with LN-binding integrins such as ␣3␤1, ␣6␤1, ␣6␤4, and ␣7␤1 strongly suggests that CD151 directly regulates the trafficking of these integrins. Although CD151 may not directly bind to FN receptor ␣5␤1 integrin (36), this integrin could be part of a tetraspanin web or glycosynapse or TEM and indirectly associates with CD151 (1,20). Therefore, the trafficking of ␣5␤1 integrin could also be affected by CD151 when CD151 is expressed in fibroblast cells such as NIH3T3 in which ␣5␤1 is one of the major integrins. Indeed, the endocytosis (and/or the trafficking immediately following the endocytosis) of CD151-associated integrins such as ␣3␤1, ␣6␤1, and ␣5␤1 is markedly attenuated by the CD151 Yala mutation. Consistently, the endocytosis of integrin ␣3␤1 was impaired when CD151 was silenced, correlating with the diminished cell migration (8). Although recent studies have shown that tetraspanins regulate integrin endocytosis (8,29), our study is the first to identify the structural element in tetraspanin that determines the trafficking of its associated inte-grins and to provide a mechanism for how CD151 regulates integrin trafficking. The fact that CD151 alters only the trafficking of its associated integrins, which are the integrins in the activated conformation (19) or the integrins actively involved in cell migration (this study), strongly suggests that CD151 links integrin activation to integrin trafficking in migrating cells. Based on these observations, we extrapolate that activated integrins participate in cell migration through vesicle trafficking.
Besides the highly stoichiometric association of CD151 with integrin ␣3␤1, the associations of CD151 with other integrins have relatively low stoichiometries (36). Especially in NIH3T3 transfectants, a portion of integrins is preoccupied by endogenous murine CD151, resulting in even lower stoichiometry of human CD151-murine integrin associations. Thus, it is not surprising that the global trafficking of ␣5␤1 and ␣6␤1 integrins was not substantially altered by human CD151 in NIH3T3 transfectant cells. Although we could not determine the effect of the CD151 Yala mutation on the global internalization of murine integrin ␣3 in NIH3T3 transfectants, we predict that the total ␣3 integrin trafficking will be significantly attenuated because of the highly stoichiometric association of CD151 with integrin ␣3␤1, which is different from ␣5␤1 and ␣6␤1 integrins.
Because the cytoplasmic domains of many integrin ␤ subunits contain NPXY endocytosis signals, the trafficking of integrins could also be regulated by their own sorting signals. By associating with tetraspanin web or glycosynapse or TEM, the trafficking paths and/or dynamics of integrins could be modified by the intrinsic trafficking signals of their associated tetraspanins such as the YRSL motif of CD151. Thus, the trafficking of integrins and other TEM components are guided by their intrinsic trafficking signals, whereas tetraspanin adds another layer of complexity and a new dimension to the trafficking of its associated TEM components such as integrin. It remains to be tested in a future study whether CD151 provides or alters the trafficking directionality of integrins in migrating cells. Given the fact that CD151 expression is frequently up-regulated in invasive and metastatic cancers (2), this notion actually points to a novel and potential mechanism that the altered trafficking pattern of integrins in cancer cells, because of the increased level of CD151, may facilitate cell movement and consequently promote the invasiveness and metastatic potential of cancers. Also, given the fact that pathological angiogenesis becomes deficient in CD151-null mice (7), the role of CD151 in regulating the trafficking of integrins in endothelial cell movements needed for blood vessel formation and remodeling is an intriguing issue remaining to be tested.