PKCα regulates β1 integrin‐dependent cell motility through association and control of integrin traffic

The antibodies employed in this study, 12G10 and mAb13, distinguish at least two major sub‐populations of integrins: the 12G10+ population represents activated and ligand‐occupied integrins and the mAb13+ population represents unoccupied integrins (Mould et al., 1995, 1998). Consistent with their ability to detect these different conformers, immunocytochemical analysis of 12G10 and mAb13 epitope expression in MCF7 cells revealed overlapping but distinct compartments (Figure 1A). For the purposes of discussion, the 12G10‐positive population is referred to as activated, while the mAb13‐positive population is referred to as ligand unoccupied. The 8E3 monoclonal is a pan‐β1 integrin antibody.

To investigate the effect of PKCα expression on β1 integrin we first determined the localization of β1 integrin in relation to that of green fluorescent protein (GFP)–PKCα. In untreated MCF‐7 cells activated β1 integrin was found predominantly in a perinuclear/pericentriolar compartment with tubulovesicular structures distributed through the cytoplasm (Figure 1B). There was also some plasma membrane staining with some cells displaying a concentration at extended filopodia (see arrows, Figure 1B). This localization compared with a punctate distribution of GFP–PKCα that only partly overlapped with the β1 integrin compartments. Treatment of cells with the PKC activator TPA caused the redistribution of PKCα to the plasma membrane and to a perinuclear region (two examples are shown in Figure 1C). Coincident with this, the β1 integrin also redistributed to the plasma membrane and to a distinct perinuclear compartment. Consequently, the main site of co‐localization between the two proteins following TPA stimulation of PKCα is at the plasma membrane.

The location of the β1 integrin was investigated using markers to various defined compartments (Figure 2A and B). The lack of any apparent co‐localization with giantin, p63, Rab‐7 and β‐COP indicates that the perinuclear/pericentriolar structure is not Golgi, endoplasmic reticulum or the intermediate compartment. Treatment with brefeldin A (5 μg/ml) at 37°C for 20 min did not cause redistribution or dispersal of the integrin‐containing structures, consistent with a non‐Golgi localization (data not shown). In breast carcinoma cells such as MCF‐7 and MDA‐MB‐231, the steady‐state distribution of activated β1 integrin was, however, found to co‐localize with that of both the endogenous transferrin receptor and fibronectin (Figure 2A). Incubation of cells at 20°C for 30 min led to a dramatic redistribution of activated β1 integrin to the plasma membrane (Figure 2A). However, treatment with nocodazole (5 μg/ml) at 37°C for 2 h resulted in the dispersal of these structures to form numerous, punctate vesicles, which were smaller and less tubular than the integrin‐containing vesicles in untreated cells. Plasma membrane staining by 12G10 was also reduced in nocodazole‐treated cells.

This behaviour of β1 integrin is reminiscent of that described for recycling receptors (Gruenberg and Maxfield, 1995). To assess this directly, we compared the localization of β1 integrin with that of the transferrin receptor, which is known to reside within the pericentriolar, endocytic recycling compartment. Localization of these proteins was detected in parallel through the uptake and intracellular trafficking of Cy3.5‐labelled 12G10 and FITC‐transferrin (Tf), following pulse–chase labelling of MCF‐7 cells. During the initial 10 min of chase at 37°C, both 12G10–Cy3.5 and FITC–Tf were found predominantly at the plasma membrane and juxta‐membrane vesicular structures (Figure 2B). After 20 min, these labelled structures were found to concentrate in a perinuclear location. The near‐complete co‐localization of both 12G10–Cy3.5 and FITC–Tf throughout the entire time course of pulse–chase suggests that both activated integrin and transferrin receptors traffic through a similar endocytic, recycling compartment.

The precise localization of β1 integrin was examined by immunoelectron microscopy after staining with anti‐β1 integrin mAbs followed by rabbit anti‐mouse IgG1/protein A–gold. Of the two breast carcinoma cell lines that were investigated, MCF‐7 and MDA‐MB‐231, the latter gave a more distinct staining with the pan‐β1 integrin antibody 8E3 probably due to a difference in the levels of β1 integrin expression between the two cell lines (Figure 3A). Staining was found mainly at the filopodial extensions of the plasma membrane as well as in large intracellular vesicular structures, which can be morphologically classified as multivesicular bodies (MVBs). In contrast, there was no detectable labelling of other membranous structures including the mitochondria, Golgi, trans‐Golgi, endoplasmic reticulum and the nucleus. Figure 3A (upper panel) shows the corresponding immunofluorescence image of MDA‐MB‐231 stained with 8E3 (total integrin pool). As for MCF‐7 cells, MDA‐MB‐231 cells showed co‐localization of 12G10‐positive integrin with the transferrin receptor (Figure 3B). As shown, this recycling compartment was also positive for the marker Rab11. By immunogold‐electron microscope (EM) staining, the activated (12G10‐immunoreactive) integrin pool was also localized to the filopodia and endosomes within the MVBs. The staining of 12G10 immunogold labelling was specific but weak compared with that of 8E3.

Because of the partial overlap in localization of GFP–PKCα and activated β1 integrin we determined whether these proteins in fact form a complex. To assess this possibility in situ, we employed fluorescence lifetime imaging microscopy (FLIM) to determine the extent of fluorescence resonance energy transfer (FRET) between the GFP–PKCα (donor) and β1 integrin (12G10–Cy3.5 acceptor). The spatial separation (R₀) between GFP and Cy3.5 at which the resonance energy transfer efficiency is 50% is 5.7 nm. Detectable energy transfer can, however, occur up to 9 nm (Ng et al., 1999). FRET results in a shortening of the GFP (donor) fluorescence lifetime which is measured by two independent parameters, the phase shift (τ_p) and relative modulation depth (τ_m) (reviewed by Bastiaens and Squire, 1999; Ng et al., 1999). A detailed description of these parameters and the biophysical principles of FLIM is, however, beyond the scope of this article and can be found in the aforementioned review (Bastiaens and Squire, 1999). For the interpretation of the results, it is taken that the β1 integrin is present in excess of the GFP–PKCα, i.e. it is not limiting for FRET. This is shown here by the finding that there is no correlation between the intensity of integrin staining and the occurrence of FRET (see for example the 12G10 direct images and the lifetime images in Figure 4). GFP–PKCα lifetime measurements for unstained, transfected MCF‐7 cells showed a fairly uniform average lifetime [(τ_m + τ_p)/2 = <τ>] (Figure 4A). When these cells were stained additionally with Cy3.5‐labelled 12G10 (12G10–Cy3.5) post‐fixation, it was evident that τ for GFP was decreased in the interior of the cell with a punctate pattern of unaltered GFP–PKCα lifetime at the cell periphery (Figure 4A). Thus PKCα is in part complexed to this population of β1 integrin. Upon TPA treatment for 20 min, the extent of complex formation is increased with a further reduction in the GFP–PKCα lifetime within the cell interior, where up to 44% FRET efficiency was detected. Owing to the non‐confocal nature of the images, a direct correlation between co‐localization and FRET between PKCα and the 12G10‐reactive β1 integrin in the same juxta‐membrane vesicular structures after TPA stimulation is more readily observed near the cell margins. In the centre of the cells, the lifetime presented represents the average of donor lifetimes at both the plasma membrane and intracellular vesicles throughout the whole depth of the cell, making it harder to correlate the co‐association and FRET data visually. The specific changes in τ_m and τ_p are shown graphically in the bottom panels, illustrating the decreases in both modulation (m) and phase (p) lifetimes.

The demonstration of FRET between GFP–PKCα and activated β1 integrin provides direct evidence of complex formation in vivo. Next, we addressed the question of whether the formation of the PKCα–β1 integrin complex is conformation dependent by examining the interaction between GFP–PKCα (donor) and the subset of β1 integrin which is immunoreactive with the Cy3.5‐labelled mAb13 (acceptor), i.e. the ligand‐unoccupied population (Mould et al., 1996). FRET was observable in some cells as illustrated in Figure 4B). However, while FRET from GFP–PKCα to the 12G10‐recognized integrin was uniformly observed among all the fields of cells studied, only a minority of cells (two of seven and four of 12 before and after TPA treatment, respectively) demonstrated a significant interaction between PKCα and the mAb13‐positive integrin, hence the heterogeneity of lifetime distribution in the 2D histogram. It is evident that PKC activation with TPA does not alter the extent of complex formation with this integrin population.

In order to assess the requirements for PKCα–β1 integrin interaction, a GFP–PKCα regulatory domain construct (GFP–PKCα^reg) was employed. In the absence of any PKCα stimulus, the lifetime of the GFP–PKCα^reg is not influenced by staining fixed cells with 12G10–Cy3.5 (Figure 5A). Thus the regulatory domain itself has little constitutive capacity to interact with activated β1 integrin. However, on TPA treatment, the GFP–PKCα^reg becomes complexed to the activated β1 integrin. In contrast, Figure 5B shows interaction between GFP–PKCα^reg and the population of ligand‐unoccupied β1 integrin. The extent of this interaction is reduced upon TPA treatment, as the regulatory domain associates increasingly with the 12G10‐positive conformer (see Figure 5A.

To corroborate these findings, we determined whether a β1 integrin–PKCα complex could be demonstrated by co‐immunoprecipitation. Immunoprecipitation of the full‐length PKCα or its regulatory domain, with the antibody MC5 and Western analysis with Mab13 reveals that β1 integrin can be co‐immunoprecipitated with PKC (Figure 6). The coprecipitation of β1 integrin with PKCα prior to activation is consistent with the complex formation (PKCα–12G10 and to a lesser extent, PKCα–mAb13) evinced by FLIM. Based upon the FLIM data, the β1 integrin that is coprecipitated with GFP–PKCα^reg is presumably in the ligand‐unoccupied conformation. Interestingly, coprecipitation may occur in part through post‐extraction interaction of these proteins, since the regulatory domain can interact with β1 integrin in cell extracts. Thus a pull‐down using immobilized PKCα^reg can coprecipitate β1 integrin (Figure 6). This observation indicates that PKCα^reg contains a region sufficient for β1 integrin binding.

To determine the influence of PKCα on cell surface expression of β1 integrin, fluorescence‐activated cell sorting (FACS) analysis was performed (Figure 7. A comparison between GFP–PKCα and empty vector‐transfected cells showed that there was an increase in the cell surface expression of β1 integrin associated with PKCα overexpression. This effect was most evident in comparing the mean fluorescent intensities for vector‐ (61 arbitrary fluorescence units) and PKCα‐ (135 arbitrary fluorescence units) transfected cells. The upregulation of 12G10‐positive integrin on the surface of GFP–PKCα‐transfected cells was initially detected in cells FACSorted according to the GFP fluorescence (Figure 7A) and subsequently confirmed by two‐colour FACS analysis (GFP versus surface labelling by an anti‐β1 integrin antibody) on whole cell populations in more than three independent experiments. The increase in cell surface expression of β1 integrin was evident on detection with either 12G10 (activated subset) or 8E3 (total pool). The percentage of β1 integrin molecules that were 12G10 positive was increased in GFP–PKCα‐expressing cells compared with vector controls from 65% (34/52) to 81% (67/83), indicating that there is also an increase in activated β1 integrin compared with the bulk population (Figure 7C). Transient transfection with GFP–PKCα^reg increased the proportion of β1 integrin that was 12G10 positive to 74% (40/54) from 57%, in the vector‐transfected control (21/37), along with a moderate increase of total cell surface integrin expression (Figure 7B). The notion that activated β1 integrin expression is influenced by the regulatory domain of PKCα, is further supported by the lack of any apparent change in 12G10 expression when a PKCα construct without the regulatory domain (CD2–PKM) is expressed (compare with vector‐only profiles).

Under all conditions, TPA treatment of MCF‐7 cells caused a redistribution of β1 integrin from the cell surface. Thus treatment led to a modest decrease in 8E3 staining (52 to 41% and 83 to 52% in GFP and GFP–PKCα‐expressing cells, respectively) and a greater concomitant effect on 12G10‐positive β1 integrin (2‐ and 2.4‐fold decreases in GFP and GFP–PKCα‐positive cells, respectively). The greater change in the 12G10‐positive β1 integrin subset relative to the total pool (8E3‐positive) indicates that the active conformer is removed more readily from the cell surface.

To determine whether 12G10‐positive β1 integrin was internalized following PKC activation, the uptake of Cy5‐labelled fibronectin (FN–Cy5) into GFP–PKCα‐expressing cells was monitored by immunofluorescence after a 20‐min pulse labelling at 37°C in the presence of 400 nM TPA, followed by cell fixation and antibody (12G10–Cy3.5) staining. TPA treatment of GFP–PKCα‐expressing cells led to a time‐dependent relocation of both GFP–PKCα and 12G10‐positive β1 integrin. Initial movement was to the cell surface, a process taking 5–10 min (Figure 8A). On further exposure (20 min) internalization occurred and the distinct perinuclear locations of the β1 integrin and PKCα became evident. Throughout the time course of the experiment, FN–Cy5 taken up by the cell was co‐localized with the activated integrin (an example is shown in Figure 8B, left upper panel). The temporal behaviour varied within populations of unsynchronized cells; however, the sequence of events and patterns of responses are typical. Following TPA treatment, all cells become rounded up, as evident from the Z‐series projection (Figure 8A). Evidence that the β1 integrin was internalized in response to TPA was provided by the finding that an extensive and time‐dependent uptake of fibronectin was observed (Figure 8A). This occurred in GFP–PKCα‐expressing cells and was not seen at this level of detection in untransfected cells (an example is shown in Figure 8b, left upper panel).

The PKCα‐dependent uptake of fibronectin was inhibited by the PI3K inhibitor LY294002, by chelation of cytosolic Ca²⁺ with BAPTA or by a dominant negative (K44A) mutant of dynamin I (Figure 8B). Furthermore, it was not supported by GFP–PKCα^reg despite this protein's ability to form a stable association with β1 integrin upon TPA treatment in vivo (see above). Similarly, a PKCα kinase‐dead mutant (PKCα^kin−) was unable to promote fibronectin uptake. Hence the PKCα‐driven internalization of activated integrin receptors required its full catalytic potential. As observed for the 12G10–Cy3.5‐positive β1 integrin staining (Figure 2B), in GFP–PKCα‐expressing cells, FN–Cy5 trafficked through the same endocytic compartment as FITC–Tf following its internalization. Figure 8B shows the endocytic vesicles which contained both ligands after a 20‐min pulse–chase at 37°C in the presence of 400 nM TPA, followed by cell fixation.

The evidence presented above indicates that PKCα is recruited to β1 integrin receptor complexes. This association is in part dependent upon the regulatory domain. Conversely the cell surface 12G10‐positive integrin can be internalized in a PKCα kinase activity‐dependent manner. To assess how this PKC‐dependent behaviour may influence integrin‐dependent function, we analysed the migratory behaviour of MCF‐7 cells in a modified Transwell chamber assay. GFP–PKCα was transfected into cells and the proportion of transfected cells migrating or not migrating on different matrix substrates was scored (Figure 9). Among GFP–PKCα‐transfected cells, there was an enrichment of GFP–PKCα‐positive cells in the migrating population when cells were plated on the β1 integrin substrates: fibronectin (FN), laminin (Lam) and collagen (Coll). In Figure 9, the percentage of cells in the upper chamber of each transfected well was given the arbitrary value of unity while that of the migrated cells was expressed as a proportion of the non‐migrating transfected percentage, for comparison of results from independent experiments. In a typical experiment, 6% (SD 2%, No. of fields = 6) of cells in the upper chamber expressed GFP–PKCα while the percentage of expressors among the migrated cells in wells coated with fibronectin was enriched to 17% (SD 6%). Vitronectin is not a ligand for β1 integrin and no enrichment of GFP–PKCα‐positive cells was observed among the migrating population. One of the vitronectin receptors αvβ5 is, however, present in MCF‐7 cells (Meyer et al., 1998). Cells transfected with GFP vector alone showed neither positive nor negative enrichment among the migrating and non‐migrating populations. When the dominant negative dynamin I (K44A) construct was co‐transfected with GFP–PKCα (3:1 ratio), the effect of PKCα expression on migration was blocked. This indicates that the active cycling of β1 integrin contributes to the migratory response. Consistent with this, neither the kinase‐defective PKCα nor the PKCα regulatory domain increased the ability of transfected cells to migrate.

Human breast carcinoma cells (MCF‐7 and MDA‐MB‐231 cells) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum at 37°C, in a 10% CO₂ atmosphere. In addition, the MCF‐7 cultures were grown in media supplemented with insulin (10 μg/ml). Cells were transfected using Lipofectant (Life Technologies). Transfected cells were stimulated as indicated in the text and figure legends, 48 h after transfection.

Myc‐ and GFP‐tagged PKCα constructs were obtained by subcloning the PKCα coding sequence into the EcoRI sites of a modified pCDNA3 vector (Clontech) and a pZeoSV‐derived vector (Invitrogen), permitting fusion of a myc coding sequence (ATGGAGCAGAAGCTCATATCGGAGGAGGACCTAGGGCCCGAATTC) and the coding sequence of GFP, respectively, at the N‐terminus of PKCα. In order to fuse PKCα in‐frame with the myc and GFP tags, the p‐Babe vector (Alvaro et al., 1997) containing human PKCα was used as a template in a PCR using 5′ primer 5′‐GGGAATTCCAGCAAGCTTGGTTGGGGGGGGGACC‐3′ (for insertion in pCDNA3) or 5′‐GGGAATTCCAGCAAGCTTGTTGGGGGGGGGACC‐3′ (for insertion in pZeoSV) and a 3′ primer 5′‐CATATCGCAGGTGTCACATTTCATCCCTTGATGGAT‐3′ corresponding to the 391–420 bp of the PKCα coding sequence. The 455/4 bp PKCα N‐terminal amplified sequences were digested at their unique BanI restriction site. The resulting 265/4 bp sequences corresponding to the N‐terminal sequence of PKCα were then ligated to the C‐terminal part of PKCα obtained from a combined BanI and EcoRI digestion of the original p‐Babe vector. The resulting ligation product containing the full‐length sequence of PKCα was digested with EcoRI and ligated to the EcoRI site of the corresponding vectors. The integrity of the PCR amplified N‐terminal sequence of PKCα was verified by sequencing. The construction of the CD2–PKM plasmid has been described previously (Garcia‐Paramio et al., 1998). The GFP‐tagged PKCα regulatory domain (with variable domain V3) construct was kindly provided by D.Joubert (INSERM U469, Montpellier, France). The dominant negative (K44A) mutant of dynamin I and the full‐length chicken β1 integrin plasmids were generous gifts of S.Schmidt (Scripps, La Jolla, CA) and Dr A.Horwitz (University of Illinois, Urbana, IL), respectively.

MC5 is a murine mAb that recognizes the V3 region of PKCα (Young et al., 1988). The monoclonal antibody M4 (a gift of S.Jaken, Lake Placid, NY), is directed against the PKCα regulatory domain and was used as ascites fluid. The characterization of the mouse anti‐human integrin mAb 12G10, which recognizes an activation epitope within the putative A domain region has been described previously (Fogerty et al., 1990; Mould et al., 1995, 1998). The rat anti‐human β1 antibody mAb13 recognizes the ligand‐unoccupied forms of the integrin and blocks ligand binding (Humphries, 1996; Mould et al., 1996). 8E3 is a pan‐beta;1 integrin murine mAb. Direct conjugation of proteins/antibodies to the fluorophores Cy3 and Cy3.5/Cy5 (Amersham Life Science) was performed at pH 9 as described previously (Bastiaens and Jovin, 1998). Anti‐Rab 7 polyclonal antiserum was a gift from M.Zerial (EMBL, Heidelberg). Rabbit anti‐Rab11 was kindly provided by Dr S.Tooze (ICRF, London). Goat anti‐transferrin receptor and anti‐fibronectin polyclonal antibodies were from Santa Cruz Biotechnology. mAb MC5 was coupled to protein G–Sepharose (Pharmacia) at pH 9 (in 0.2 M sodium borate buffer) using dimethyl pimelimidate according to the manufacturer's protocol.

Myc‐tagged PKCα or GFP–PKCα^reg was immunoprecipitated from the whole cell lysates of MCF‐7 cells using 10–20 μg of protein G‐coupled MC5 as described previously (Kanner et al., 1989), except that 1% (w/v) n‐octyl β‐d‐glucopyranoside was used instead of NP‐40. The immunoprecipitates were denatured in sample buffer (Laemmli, 1970), separated on an 8% polyacrylamide gel under reducing conditions and transferred electrophoretically to PVDF membrane. Blots were then probed with the rat anti‐integrin mAb13 (2–4 μg/ml) followed by a donkey anti‐rat–HRP conjugate (Amersham) in the presence of 0.5 μg/ml mouse IgG to block immunoreactivity with the precipitating antibody. The lower part of the same membrane was probed with MC5 or M3 (used at 2 μg/ml and 1/1000 dilution of ascites, respectively) to detect PKCα or its regulatory domain. Incubation with primary antibodies was performed overnight at 4°C. Detection was with ECL (Amersham) according to recommended procedures.

For in vitro pull‐down assays, immunoprecipitates containing GFP–PKCα^reg protein were washed thoroughly with modified RIPA buffer (Kanner et al., 1989), then left to tumble overnight at 4°C in the same buffer (negative control) or in freshly lysed detergent extracts of MCF‐7 cells alone or cells which had been transiently transfected with a full‐length chicken β1 integrin construct (Reszka et al., 1992). The GFP–PKCα^reg protein complexes were then washed three times with RIPA buffer, once with Tris–saline buffer (pH 7.4) and subsequently resuspended in Laemmli sample buffer.

For fibronectin uptake studies, MCF‐7 cells transiently transfected with various PKCα constructs were pretreated with cycloheximide (10 μg/ml) for 2–4 h to inhibit the synthesis of new proteins, including integrin receptors. Cells were then incubated at 4°C with media containing 50 μg/ml Cy5–labelled fibronectin, with or without 50 μg/ml FITC‐labelled transferrin (Sigma) for 30 min, washed with phosphate‐buffered saline (PBS), then stimulated with TPA (400 nM) at 37°C for various lengths of time before paraformaldehyde fixation, permeabilization and staining with 12G10–Cy3.5. The pulse–chase experiments were performed as previously described (Mayor et al., 1993). Briefly, untransfected MCF‐7 cells on coverslips were serum starved for 30 min, pulsed for 5 min with Cy3.5‐labelled 12G10 and FITC–Tf at 37°C, then the excess ligand/antibody was washed off with medium at 4°C. The pulsed cells were chased for various periods of time (up to 80 min) at 37°C, fixed in 2% (w/v) paraformaldehyde for 15 min at room temperature, then examined by confocal microscopy.

Immunocytochemical staining was performed as described elsewhere (Kiley et al., 1997) except for the following modifications. Cells were permeabilized with 0.2% (v/v) Triton X‐100/PBS following fixation in 4% (w/v) paraformaldehyde. Primary antibodies were diluted 1:200 in PBS containing 1% BSA, except for the fluorophore‐conjugated antibodies which were used at 1:20–1:50. The secondary conjugates used were Cy5‐conjugated donkey anti‐goat IgG (1:200) and Texas Red/Cy3‐conjugated donkey anti‐rabbit IgG (1:400) (Jackson ImmunoResearch Laboratories, West Grove, PA). Confocal images were acquired on a confocal laser scanning microscope (model LSM 510, Carl Zeiss Inc.) equipped with a 63×/1.4Plan‐APOCHROMAT oil immersion objective. Each image represents a 2‐dimensional projection of sections in the z‐series, taken across the depth of the cell at 0.2 μm intervals.

Cells were scraped from culture plates and fixed in 4% (w/v) paraformaldehyde for 1 h, followed by 2% (w/v) paraformaldehyde overnight at 4°C. Cells were processed and ultrathin sections were collected as described (Gorlich et al., 1995). Sections were incubated in the following order: anti‐beta;1 integrin mAb (12G10/8E3, at 500 or 100 μg/ml respectively), rabbit‐anti mouse IgG1 (1/100), protein A–gold 10 nM particles (1/100). Antibody reagents were diluted in PBS containing 1% (w/v) BSA. After antibody labelling, sections were examined using a Jeol 1010 microscope.

A detailed description of the FLIM apparatus used for FRET determination in this work can be found elsewhere (Squire and Bastiaens, 1999). The lifetime instrument performs phase and modulation based imaging fluorimetry by microscopy. All images were taken using a Zeiss Plan‐APOCHROMAT 100×/1.4NA phase 3 oil objective and the homodyne phase sensitive images recorded at a modulation frequency of 80.218 MHz. For experiments involving the anti‐integrin IgG–Cy3.5 donor/acceptor FRET pair, the donor (GFP–PKCα or GFP–PKCα^reg) was excited using the 488 nm line of a Argon/Krypton laser and the resultant fluorescence separated using a combination of dichroic beamsplitter (Q 505 LP; Chroma Technology Corp.) and narrow band emitter filter (BP514/10; Lys & Optik). Acceptor images (anti‐integrin mAb–Cy3.5) were recorded using a 100 W mercury arc lamp (Zeiss Attoarc) as a source of sample illumination combined with a high Q Cy3 filter set (exciter: HQ 535/50, dichroic:Q 565 LP, emitter: HQ 610/75 LP; Chroma Technology Corp.).

Transiently transfected MCF‐7 cells were detached from plates using EDTA then, without fixation, stained with either an anti‐beta;1 integrin mAb or mouse IgG isotype control, followed by an anti‐mouse IgG–RPE conjugate. Cells were analysed on the same day by a FACScan, as previously described (Ng et al., 1995). All antibody incubation and washing steps were carried out at 4°C. Cell purification by FACSorting according to GFP fluorescence was performed using Becton Dickinson FACS Vantage.

MCF‐7 cells were transiently transfected with a vector control or various PKCα constructs. After 20 h, cells were detached from culture plates with trypsin, washed three times with serum‐free medium supplemented with glutamine and 0.5% (w/v) BSA (migration buffer), then replated at 10⁶ cells/ml on to the inserts of 8 μm pore size Transwell chambers (Costar, Cambridge, MA) according to the manufacturer's instructions. The underside of the inserts was precoated by filling the lower chamber with either 0.1% (w/v) BSA or different integrin substrates (vitronectin 5 μg/ml; fibronectin, laminin and collagen, 10 μg/ml) and incubation overnight at 4°C. After 20 h, cells were trypsinized from both the top and underside of the inserts and then washed three times in migration buffer. The cell suspensions obtained, along with the fluid from either the upper or lower chambers, were centrifuged and cytospins recovered on coverslips before mounting on a glass slide. The percentage of GFP‐, GFP–PKCα‐ or GFP–PKCα^reg‐transfected cells that had migrated through the insert was compared with that of the non‐migrating population that remained in the upper chamber. For the untagged kinase‐defective PKCα, immunofluorescence staining with a PKCα C‐terminus‐specific polyclonal antibody was performed to ascertain the percentage of transfected cells in each chamber. The cells in six or more low power fields per slide were counted with the aid of a fluorescence microscope. Data from three independent experiments were pooled and analysed by two way analysis of variance using the statistical software MINITAB.