A critical role for tetraspanin CD151 in alpha3beta1 and alpha6beta4 integrin-dependent tumor cell functions on laminin-5

Abstract

The basement membrane protein laminin-5 supports tumor cell adhesion and motility and is implicated at multiple steps of the metastatic cascade. Tetraspanin CD151 engages in lateral, cell surface complexes with both of the major laminin-5 receptors, integrins alpha3beta1 and alpha6beta4. To determine the role of CD151 in tumor cell responses to laminin-5, we used retroviral RNA interference to efficiently silence CD151 expression in epidermal carcinoma cells. Near total loss of CD151 had no effect on steady state cell surface expression of alpha3beta1, alpha6beta4, or other integrins with which CD151 associates. However, CD151-silenced carcinoma cells displayed markedly impaired motility on laminin-5, accompanied by unusually persistent lateral and trailing edge adhesive contacts. CD151 silencing disrupted alpha3beta1 integrin association with tetraspanin-enriched microdomains, reduced the bulk detergent extractability of alpha3beta1, and impaired alpha3beta1 internalization in cells migrating on laminin-5. Both alpha3beta1- and alpha6beta4-dependent cell adhesion to laminin-5 were also impaired in CD151-silenced cells. Reexpressing CD151 in CD151-silenced cells reversed the adhesion and motility defects. Finally, loss of CD151 also impaired migration but not adhesion on substrates other than laminin-5. These data show that CD151 plays a critical role in tumor cell responses to laminin-5 and reveal promotion of integrin recycling as a novel potential mechanism whereby CD151 regulates tumor cell migration.

Figure 1.

Efficient silencing of CD151 expression in A431 epidermal carcinoma cells. A431 cells were transduced with retroviral vectors encoding three different CD151 shRNA constructs. Stably transduced cells were selected for 2 wk in puromycin and maintained as uncloned populations. (A) CD151 expression in wild-type parental A431 cells and A431 cells bearing CD151 shRNA constructs was evaluated by flow cytometry. Isotype neg., background staining of wild-type cells with nonimmune mouse IgG1; A431 WT, staining of wild-type cells with 5C11 anti-CD151 antibody; A431 sh1-sh3, staining of cells bearing different CD151 shRNA constructs with the 5C11 antibody. The sh3 construct efficiently silenced CD151 expression in >95% of the cells in the A431 sh3 population. (B) A431 wild-type, sh1, sh2, and sh3 cells were lysed in 1% Triton X-100 and lysate protein concentrations were normalized by BCA assay. CD151 (lanes 1–4) or α3 integrin (lanes 5–8) were immunoprecipitated with the 5C11 antibody or the A3-X8 anti-α3 integrin antibody, followed by blotting with the 11B1.G4 anti-CD151 mAb. (C) The same cell populations as in B were cell surface–labeled with biotin and extracted in 1% Triton X-100 followed by CD151 or α3β1 immunoprecipitation. Biotin-labeled α3β1 integrin was detected by blotting with Extravidin-HRP. Bottom panel in lanes 5–8 shows a shorter exposure time.

Figure 2.

CD151 silencing impairs α3β1 integrin–dependent motility on laminin-5. (A–C) Wild-type A431 cells were plated in serum-free medium on laminin-5 in the absence of antibody (A), the presence of A3-IIF5 anti-α3 integrin function blocking antibody (B), or the presence of GoH3 anti-α6 function blocking integrin antibody (C). Motility was monitored by time-lapse microscopy. Each gray line represents the cumulative distance traveled by an individual cell versus time. In each field, all cells that could be tracked for at least 1 h were quantified and used to calculate the average velocity (indicated by black arrows). (D) A431 sh3 cell motility on laminin-5 was measured by time-lapse microscopy as in A–C above. For a side by side comparison of wild-type and sh3 cell motility, see Supplementary Video, Figure 2.mov. (E) For two independent trials, average velocities were determined separately for each hour of the assay. A431 sh3 cells were slower than wild-type cells during all 3 h in both trials; ∗p < 0.0001, unpaired t test. (F) For two independent trials, average velocities of all cells that remained free of cell–cell contact for at least 1 h were determined during the contact-free interval and compared with the average velocity of the entire cell population. sh3 cells were slower than wild-type cells for both contact-free and total cell populations; ∗p < 0.0001, unpaired t test. (G) Results of several independent motility experiments performed as in A–D. Velocities ± SEM are depicted; n = the number of independent experiments. Velocity of A431 sh3 cells was consistently reduced by ∼50% compared with wild-type, sh1, and sh2 cells; ∗p < 0.01 (for sh3 vs. wild-type or sh1 cells), and ∗p < 0.05 (for sh3 vs. sh2 cells), ANOVA with Bonferroni post-tests. (H) Results of several experiments performed in the presence of the GoH3 function-blocking anti-α6 integrin antibody (n = number of independent trials). sh3 cell velocity was significantly reduced compared with wild-type cells; ∗p < 0.03, unpaired t test.

Figure 3.

CD151-positive A431 sh3 escapers show wild-type motility on laminin-5. (A) A431 sh3 cells were stained with anti-CD151 antibody, 5C11, and subjected to fluorescence-activated cell sorting. The gates used to isolate CD151-silenced cells (NEG SORT) and CD151-positive cells (ESCAPERS) are indicated. (B) Velocities of CD151-silenced A431 sh3 cells and CD151-positive A431 sh3 escapers were measured by time-lapse microscopy, as for Figure 2. Significantly faster than CD151-silenced cells; ∗p < 0.0001, unpaired t test. Results are representative of two separate trials that produced similar results.

Figure 4.

Silencing CD151 disrupts α3β1 integrin association with tetraspanin-enriched microdomains. (A) A431 wild-type, sh1, sh2, or sh3 cells were lysed in Brij 96, and α3 integrin was immunoprecipitated with the A3-X8 antibody. The amount of α3 in each immunoprecipitate was assayed by blotting with the D23 anti-α3 polyclonal antibody, and the amounts of coprecipitating EWI-2 or CD81 were assayed by blotting with 8A12 or biotinylated M38 monoclonal antibodies, respectively. (B) Cells were extracted with Brij 96, and α3 integrin was immunoprecipitated as in A. α3 integrin was blotted with the D23 polyclonal antibody, and coprecipitating CD9 was assayed by blotting the C9BB mAb. (C) Equal numbers of A431 wild-type and sh3 cells were cell surface–labeled with biotin and then extracted on ice with Brij 98 for 20 min. The Brij 98–soluble fraction was gently removed, and Brij 98–insoluble material was recovered by extracting a second time with 1% Triton X-100, 0.1% SDS. α3β1 integrin was immunoprecipitated from each fraction with the A3-X8 mAb, followed by blotting with Extravidin HRP. (D) Results of three separate experiments performed as in C. Data are presented as the mean ratio of Brij 98 soluble to Brij 98 insoluble material ± SEM Significantly different from wild type; ∗p < 0.0001, unpaired t test.

Figure 5.

Adhesion to laminin-5 is impaired at early time points but not later time points for CD151-silenced cells. (A) A431 wild-type or sh3 cells, 2 × 10⁵, were plated on ice in laminin-5 or BSA-coated wells, allowed to settle, and placed at 37°C for various periods of time. Nonadherent cells were removed by rinsing, and adherent cells were stained with crystal violet, solubilized, and quantified using a plate reader. Values are reported as the fraction of cells input as measured by cells attached to PLL control wells. Experimentally determined values for total cells input varied by <10% for the two cell types. Points represent the mean ± SEM for four wells per condition. Significantly different from wild type; ∗p < 0.01; Significantly different from wild type; ∗∗p < 0.005, unpaired t test. The experiment was repeated twice with essentially identical results. (B) A431 wild-type or sh3 cells, 2 × 10⁵, were plated in laminin-5 or BSA-coated wells and allowed to attach for 30 or 120 min. Wells were rinsed, stained, and quantified as in A. For some wells function-blocking anti-α3 and α6 integrin antibodies A3-IIF5 and GoH3 were added in combination, 10 μg/ml each. Two independent trials are shown. A431 sh3 cell adhesion was significantly reduced compared with wild-type cells after 30 min on laminin-5 and after 120 min on BSA; ∗p < 0.02; ∗∗p < 0.005; ^†p < 0.0005, unpaired t tests. (C) A431 wild-type and sh3 cells were plated on BSA-coated coverslips for 2 h and then fixed and stained with the 6F12 anti-laminin-5 mAb, followed by Cy2-conjugated goat anti-mouse secondary antibody. Left panels, DIC images; right panels, laminin-5 staining. Negative control staining for this experiment was shared with the experiment in Figure 7 (see Figure 7M).

Figure 6.

Lateral and trailing edge defects in CD151-silenced A431 sh3 cells migrating on laminin-5. (A) Individual frames from time-lapse videos comparing morphology of A431 wild-type and sh3 cells migrating on laminin-5. Rapidly migrating wild-type cells typically display smooth trailing edges and a compact but highly active leading lamellipodium. A431 sh3 cells frequently display persistent adhesive contacts that arise at their lateral and trailing edges (white arrows). Elapsed time is indicated to left of each pair of panels. Time 0′ in this series corresponds to the 20-min time point in Supplementary Video, Figure 6.mov, which shows a side be side comparison of wild-type and sh3 morphology. Lateral/trailing edge adhesive contacts of sh3 cells persist for several minutes and appear to come under tension before snapping back into the cell soma within 1–2 min. (B) The frequency of lateral and trailing edge adhesive events, such as those depicted in A, that lasted 4 min or longer was measured for A431 wild-type and sh3 cells in four separate 3-h videos. Significantly different from wild-type cells; ∗p < 0.05%, unpaired t test. (C) The mean duration of lateral and trailing edge adhesive events for A431 wild-type and sh3 cells was measured in four separate 3-h videos. Significantly different from wild-type cells; ∗p < 0.01%, unpaired t test. (D) In two independent trials, A431 wild-type and sh3 cells migrating on laminin-5 were tracked for 2 h, and directional persistence for each cell was calculated during each hour. Persistence was defined as the ratio of net distance traveled divided by total distance traveled. The average persistence of sh3 cells was significantly reduced compared with wild-type cells; ∗p < 0.03; ∗∗p < 0.0001, unpaired t test.

Figure 7.

Localization of α3β1 integrin and CD151. Unpermeabilized A431 wild-type or sh3 cells were stained for α3 integrin with the A3-X8 mAb (A and B) or the 5C11 anti-CD151 mAb (C and D) followed by Cy2 goat anti-mouse secondary antibody. For codistribution studies, cells were permeabilized with 0.1% saponin and stained with the D23 polyclonal anti-α3 integrin antibody, which binds to the α3 cytoplasmic tail, followed by Alexa 594 goat anti-rabbit secondary (E and F) and the 5C11 anti-CD151 mAb followed by Cy2 goat anti-mouse secondary (G and H). Overlays are shown in I and J. Arrows indicate regions of α3 and CD151 codistribution. In negative control experiments, cells were stained with the D23 antibody followed by Cy2 goat anti-mouse (K), 5C11 followed by Alexa 594 goat anti-rabbit (L), or nonimmune mouse IgG followed by Cy2 goat anti-mouse (M).

Figure 8.

Impaired trafficking of α3β1 integrin in CD151-silenced cells migrating on laminin-5. (A) A431 wild-type and sh3 cells migrating on laminin-5 were chilled to 4°C and labeled on ice with reducible biotin. Cells were then returned to 37°C for various periods to allow internalization of labeled integrin, after which they were stripped with alkaline glutathione (GSH) to remove any label remaining on the cell surface. Cells were then lysed in 1% Triton X-100, and α3β1 integrin was immunoprecipitated using the A3-X8 antibody and visualized by blotting with Extravidin-HRP. T, total α3β1 as measured from immunoprecipitates of cells that had been labeled and held on ice in the absence of glutathione stripping. To control for efficiency of immunoprecipitation in each sample, β-tubulin (β-tub) was simultaneously immunoprecipitated and then detected by immunoblotting using the anti-tubulin mAb, DM1B. (B) Results of two independent trials performed as in A are plotted as fraction of α3β1 internalized versus time. Values were obtained by semiquantitative densitometry of transilluminated films. Data were fit to pseudo first-order kinetics using Prism software (GraphPad, San Diego, CA). (C) In a third trial, samples were prepared as in A and B, but biotinylated and total α3 were detected simultaneously on the same blot. Biotinylated α3 was visualized with IRdye 800 avidin and total α3 was visualized with the D23 anti-α3 antibody followed by blotting with Alexa 680 goat anti-rabbit. The blot was analyzed with a Li-Cor infrared gel imager in the 700- and 800-nm fluorescent channels. Lower panel, overlay of two top panels. (D) The blot shown in C was quantified with Li-Cor software, and the ratio of α3 biotin counts to total α3 counts was determined for the 10- and 30-min time points.

Figure 9.

Reexpression of CD151 in A431 sh3 cells reverses defects in α3 and α6 integrin functions. A CD151 cDNA (CD151 Rx) containing silent mutations that defeat RNAi was used to restore CD151 expression in A431 sh3 cells (Materials and Methods for details). (A) Flow cytometry using the 5C11 anti-CD151 antibody to stain A431 wild-type cells, A431 sh3 cells, or A431 sh3 cells reconstituted with CD151 (Rx). neg, con., wild-type cells stained with an isotype-matched negative control antibody. (B) A431 wild-type, sh3, and sh3 Rx cells were cell surface–labeled with biotin and extracted with 1% Triton X-100. CD151 or α3 integrin were immunoprecipitated with 5C11 or the A3-X8 anti-α3 integrin antibody, respectively, and α3β1 integrin was detected by blotting with Extravidin-HRP. (C) A431 wild-type, sh3, or sh3 Rx cells were plated in laminin-5– or BSA-coated wells, in the presence or absence of 10 μg/ml function-blocking antibodies. After 25 min, nonadherent cells were removed by rinsing, and adherent cells were stained and quantified as in Figure 5. Function blocking antibodies were anti-α6 integrin, GoH3, and anti-α3 integrin, A3-IIF5. Values represent the means ± SEM of eight wells per condition from two experiments that yielded very similar results. Values are reported as the fraction of total input cells measured in PLL-coated-wells, as in Figure 5. Significantly different from either wild-type or sh 3 Rx cells (∗p < 0.001) ANOVA with Bonferroni post-tests. The A431 sh3 cell defect in α6 integrin–dependent adhesion shown in the third set of C is also illustrated by time-lapse assay in Figure 9.mov, in which cells treated with A3-IIF5 anti-α3 antibody are plated on laminin-5. (D) A431 wild-type, sh3, and sh3 Rx cells motility on laminin-5 was measured in multiple separate time-lapse assays as in Figure 2 (at least 5 trials per cell type). Significantly different from either wild-type (∗p < 0.001) or sh3 Rx cells (∗p < 0.05), ANOVA with Bonferroni post-tests.

Figure 10.

CD151 silencing impairs migration but not adhesion on α3β1 integrin–independent substrates. (A) A431 wild-type, sh3, and sh3 Rx cells were plated in wells coated with 40 μg/ml collagen I (Col I), 20 μg/ml vitronectin (VN), 50 μg/ml fibronectin (FN), or BSA, in the presence or absence of various function blocking monoclonal antibodies (10 μg/ml). After 25 min, nonadherent cells were removed by rinsing, and adherent cells were stained and quantified as in Figure 8. Function blocking antibodies were anti-α2 integrin, A2-IIE10, and anti-αv integrin, 69-6-5, and anti-α5 integrin, P1D6. (B) A431 wild-type and sh3 cell motility on collagen I and fibronectin was measured in time-lapse assays as in Figure 2. Velocity on collagen I during the first hour of the assays was also calculated separately (1st h). In additional experiments, velocity on collagen I in the presence of 10 μg/ml A3-IIF5 anti-α3 integrin function-blocking antibody was measured. Significantly different from wild-type cell motility under the same conditions, ∗p < 0.0002, unpaired t test. Results in B are representative of two independent trials that yielded similar results. (C) A431 wild-type, sh3, and sh3 Rx cells were plated in wells coated with 5 μg/ml fibronectin or collagen I and adhesion was measured as in A. Results show the mean ± SEM for 3 independent trials.