Cas-family proteins have been implicated as signaling intermediaries in diverse processes including cellular attachment, motility, growth factor response, apoptosis and oncogenic transformation. The three defined Cas-family members (p130Cas, HEF1/Cas-L and Efs/Sin) are subject to multiple forms of regulation (including cell-cycle- and cell-attachment-mediated post-translational modification and cleavage) that complicate elucidation of the function of specific Cas proteins in defined biological processes. To explore the biological role of HEF1 further, we have developed a series of cell lines in which HEF1 production is regulated by an inducible promoter. In this system, HEF1 production rapidly induces changes in cellular morphology and motility, enhancing cell speed and haptotaxis towards fibronectin in a process partially dependent on intact ERK and p38 MAPK signaling pathways. Finally, cDNA expression array analysis and subsequent studies indicate that HEF1 production increases levels of mRNA transcripts encoding proteins that are associated with motility, cell transformation and invasiveness, including several metalloproteinases, MLCK, p160ROCK and ErbB2. Upregulation of such proteins suggests mechanisms through which misregulation of HEF1 may be involved in cancer progression.
Integrin-dependent signaling contributes to cellular decisions to initiate diverse programs such as proliferation, apoptosis, differentiation and migration ( Giancotti and Ruoslahti, 1999; Miyamoto et al., 1995; Ruoslahti and Reed, 1994; Sheetz et al., 1998). Hence, integrin signaling affects a variety of physiological processes including development, tissue remodeling, wound healing and tumor cell growth and metastasis ( Petersen et al., 1998). Efforts to understand the mechanisms through which integrin signaling regulates these cellular processes have focused on analysis of focal adhesions. Focal adhesion sites consist of integrin receptors clustered following their engagement by extracellular ligand and an associated complex of intracellular proteins including actin filaments, actin-binding and -cross-linking proteins, and a number of tyrosine kinases, phosphatases and docking/adaptor proteins ( Dedhar and Hannigan, 1996). In addition to providing a physical scaffold connecting cells to basement membranes, focal adhesions also act as signaling centers, which generate and convey information from the cell periphery to downstream effector molecules. Thus, focal adhesions integrate the mechanical signals derived from morphological changes with the chemical signals triggered by receptor engagement ( Chen et al., 1997; Chicurel et al., 1998; Huang et al., 1998).
The contributions of a number of kinases and docking/adaptor proteins to the signaling capacity of focal adhesions have been elucidated. For example, focal adhesion kinase (FAK) and family members are tyrosine kinases that localize to focal adhesion sites, undergo autophosphorylation following integrin receptor engagement and contribute to focal adhesion regulation ( Schlaepfer and Hunter, 1998). Members of the Src family of tyrosine kinases localize with FAK and phosphorylate components of focal adhesions ( Schlaepfer and Hunter, 1998). Significantly, modulation of FAK signaling affects both cell motility and the induction of apoptosis, suggesting that these cellular processes have components in common ( Frisch et al., 1996; Hungerford et al., 1996; Ilic et al., 1995). Substrates of FAK and Src ( Hanks and Polte, 1997), which include actin binding proteins such as paxillin and adaptor proteins such as Crk and the Cas (Crk-associated substrate) family of signaling proteins, have become the targets of scrutiny as FAK/Src effectors in these dual processes.
The Cas family of adaptor proteins ( O’Neill et al., 2000) includes p130Cas ( Sakai et al., 1994), human enhancer of filamentation 1 (HEF1; also known as CasL) ( Law et al., 1996; Minegishi et al., 1996) and Efs (also known as Sin) ( Ishino et al., 1995; Alexandropoulos and Baltimore, 1996). Members of this family were initially identified as components of viral transformation signaling pathways ( Ishino et al., 1995; Kanner et al., 1990; Sakai et al., 1994) and/or as modulators of cell growth and morphology ( Law et al., 1996). Intriguingly, recent clinical studies have indicated that enhanced Cas family expression correlates with significant differences in cancer progression in humans, whereas induction of p130Cas overexpression enhances resistance to the action of anti-estrogens ( Brinkman et al., 2000; van der Flier et al., 2000). The Cas proteins have a conserved domain structure composed of an N-terminal SH3 domain, a substrate domain containing multiple tyrosine motifs that are recognized by SH2 domain proteins following phosphorylation, a serine-rich region and a C-terminal dimerization motif ( O’Neill et al., 2000). HEF1, p130Cas and Efs localize to focal adhesion sites via interaction of their SH3 domains with FAK ( Law et al., 1996; Ohba et al., 1998; Polte and Hanks, 1995; Tachibana et al., 1997) and contribute to the assembly of signaling complexes downstream of the integrin receptor following ligand binding ( O’Neill et al., 2000). An important question is whether the function of discrete Cas family members at focal complexes is equivalent or whether individual Cas proteins are associated with promotion of different biological effects.
A number of studies characterizing HEF1 and p130Cas have underscored the potential for functional divergence between these Cas family proteins. HEF1 and p130Cas are differentially regulated; HEF1 is produced at maximal levels in cells of epithelial and lymphoid origin ( Law et al., 1996; Law et al., 1998; Minegishi et al., 1996), whereas p130Cas is produced ubiquitously ( Sakai et al., 1994). Moreover, HEF1 is also regulated in a cell cycle dependent manner and is processed at the G2/M boundary by caspases to truncated isoforms that localize to distinct subcellular compartments ( Law et al., 1998). Most notably, we have recently found that HEF1 overproduction mediates apoptosis in epithelially derived cell lines, including MCF7 and HeLa cells ( Law et al., 2000), which is contrary to the pro-survival activity described for p130Cas ( Almeida et al., 2000; Cho and Klemke, 2000). Finally, recent studies in lymphoid cells showed that HEF1 (CasL) expression contributes to T cell migration induced by ligation of CD3 and β1 integrin ( Ohashi et al., 1999; van Seventer et al., 2001). Taken in sum, these results demonstrate that HEF1 differs from p130Cas in both the manner in which it is regulated and its spectrum of effector function.
We have begun to elucidate the mechanisms underlying HEF1-induced cellular responses by performing cDNA array analyses to identify downstream transcriptional targets that are upregulated as a consequence of HEF1 overproduction. Using a tetracycline-regulated HEF1-producing MCF7 cell line, we find a dramatic effect of HEF1 overproduction on cell morphology and motility, characterized by the development of a crescent shape, enhanced ruffling and increased cell spreading. HEF1-induced populations contain more highly motile cells and demonstrate increased haptotaxis towards fibronectin. Using DNA array analysis, we find that this enhanced motility is accompanied by upregulation of a set of genes associated with enhanced migration and invasion, including those encoding myosin light chain kinase (MLCK), p160ROCK, eight matrix metalloproteinases (MMPs) and ErbB2. Overall, these data suggest that the spectrum of biological effects attributable to HEF1 is complex and potentially includes promigratory and prometastatic activity.
Materials and Methods
The complete coding sequence of HEF1 was cloned in the pBPSTR1 retroviral vector ( Paulus et al., 1996) downstream of a tetracycline responsive element to create pBPSTR1-HEF1. A second plasmid, pTet-tTAK, encodes the tTA protein, a fusion of the Tet repressor DNA binding domain and the transcriptional activation domain of VP16 (Gibco/BRL).
To prepare stable, regulated clonal cell lines, MCF7 breast adenocarcinoma cells were transfected with pBPSTR1-HEF1, pTet-tTAK and MSCVhygroR (which provides a hygomycin resistance gene; kindly provided by J. Testa) using Lipofectamine™ (Gibco/BRL). Transfected cells were selected in media containing 2 μg ml–1 puromycin (to retain the tetracycline-regulated pBPSTR1-HEF1), 400 μg ml–1 hygromycin and 1 μg ml–1 tetracycline (to repress HEF1 production during selection). Cell lines were derived from isolated single colonies, expanded and examined for inducible HEF1 production. Unless otherwise stated, experiments were carried out in DMEM plus 10% FBS.
Induction of HEF1, cell lysis, immunoprecipitation and western analysis
Cells were plated at low cell density (∼600,000 cells per 100-mm culture dish) in the presence (uninduced) or absence (induced) of tetracycline for up to 24 hours prior to lysis, as noted in the figure legends. Adherent monolayers were washed twice with phosphate buffered saline and then lysed in Triton X-100 lysis buffer (50 mM HEPES (pH 7.5), 50 mM NaCl, 5 mM EDTA, 1% Triton X-100, 50 mM NaF, 10 mM Na4P2O7) supplemented with 1 mM sodium orthovanadate, 0.1 mM phenylmethylsulfonyl fluoride, 1 μg ml–1 aprotinin and 1 μg ml–1 leupeptin. The protein concentration of total cell lysates was quantitated using a BCA protein determination kit (Pierce). Total cell lysate was separated by SDS-PAGE and transferred to polyvinyl difluoride membranes (Immobilon). Membranes were blocked using 5% fat-free milk, probed with rabbit polyclonal antisera specific for HEF1 [αHEF1-SB-R1 ( Law et al., 1998)] or p130Cas [Transduction Labs; cross-reactive with HEF1, as noted by Law et al. ( Law et al., 1998)] and developed using a chemiluminescent system (NEN). As indicated, membranes were also probed with the following antibodies, following the manufacturer’s protocols: mouse monoclonal antibody specific for ERK (MAPK; Transduction Laboratories, Lexington, KY) and rabbit polyclonal antisera specific for p38 kinase (Santa Cruz Biotechnology, Santa Cruz, CA) and activated forms of MAPK and p38 kinase (Promega Corporation, Madison, WI).
For p130Cas immunoprecipitation, cells were harvested in PTY buffer ( O’Neill and Golemis, 2001) at the indicated time points and immunoprecipitated with antibody to p130Cas using Protein-G/Sepharose (Gibco/BRL). Precipitates were subjected to SDS-PAGE and transfer (see above), and tyrosine phosphorylation was assessed using primary antibody 4G10 (Upstate Biotechnology) and secondary antibody and development as described above, except using bovine serum albumin (BSA) as blocking agent. Subsequently, blots were stripped and reprobed with antibody to p130Cas or HEF1, as described above.
Cell spreading analysis
Cells were initially plated at ∼60% confluence in the presence or absence of tetracycline for 18 hours. Cells were then detached by incubation in PBS + 5mM EDTA for 15 minutes at 37°C, re-plated onto either uncoated glass coverslips in Dulbecco’s modified Eagle’s medium (DMEM) plus fetal bovine serum (FBS) or human fibronectin (FN) (Gibco/BRL) coated coverslips (6 μg ml–1) in serum-free DMEM and maintained in either inducing or non-inducing conditions for the indicated times prior to fixation in 3.5% paraformaldehyde. In each field, cell area measurements were determined using Inovision ISEE™ software to outline the perimeters of individual cells and to calculate the number of pixels encompassed.
Cells cultured on coverslips were fixed in 3.5% paraformaldehyde, permeabilized with 0.2% Tween-20 and blocked with 0.1% BSA in Tris buffer (10 mM Tris (pH 7.5), 150 mM NaCl). Cells were incubated with anti-HEF1 rabbit antisera (αHEF1-SB-R2) ( Law et al., 1998) or anti-paxillin mouse monoclonal antibodies (Transduction Labs) as primary and either rhodamine-conjugated anti-rabbit antibodies (Molecular Probes), biotin-conjugated anti-rabbit antibodies plus Texas-Red-conjugated streptavidin (Vector Laboratories) or dichlorotriazinylaminofluorescin (DTAF)-conjugated anti-mouse antibodies (Jackson Immunological Labs) as secondary antibody. FITC- or TRITC-conjugated phalloidin (Molecular Probes) was included in a final incubation to visualize actin. A Bio-Rad MRC 600 laser scanning confocal microscope (Cell Imaging Facility, Fox Chase Cancer Center) was used to analyze images.
For measurements of haptotaxis, 10,000 cells per 35 mm-well were plated onto the porous membrane (top well) of a modified Boyden chamber (tissue-culture treated, 8-μm pores, Transwell™; Costar, Cambridge, MA). Both top and bottom of the Boyden chamber contained DMEM with or without tetracycline. Soluble human plasma FN (Gibco/BRL) was added (4 μg ml–1, as indicated) to the bottom wells just before cell plating to coat the underside of the porous membrane. Cells on the upper side of the membrane were removed by scraping. Cells attached to the bottom membrane were fixed and stained with modified Giemsa stain. For measurements of haptotaxis in the presence of pharmacological inhibitors, the above procedure was scaled down. Briefly, 2000 cells per well (24-well Transwell plates, 8-μm pores) were added to the top chamber of a modified Boyden chamber in the absence or presence of the following compounds: 25 μM PD98059 (Sigma), 25 μM SB202190 (Sigma) and DMSO (in which PD98059 and SB202190 were dissolved) as control. Migratory cells in five to ten randomly selected fields (10× objective) per condition were counted.
For the speed analysis, cell lines were plated at low cell density in DMEM plus 10% FBS with or without tetracycline for 4-6 hours prior to the start of time-lapse video microscopy imaging. Phase contrast images were recorded at 5-minute intervals for calculation of cell speed for 18-24 hours. Cells were tracked for 70 intervals using Isee and Nanotrack imaging software, and the results were analyzed using Excel™.
Atlas array analysis and RT-PCR confirmation
Total RNA was purified from HEF1.M1 and CM1 cells that were uninduced or induced for 9 hours, treated with RNase-free DNase I (Gibco/BRL) and used to synthesize 33P-labeled cDNA probes using the protocols and cDNA-probe synthesis kit provided with the Clontech Atlas 1.2 Human Cancer gene arrays. Each of the obtained cDNA probes were hybridized in parallel to an Atlas 1.2 Human Cancer array filter for ∼14 hours, washed and exposed to BioMax film with an LE transcreen (Kodak) for 24-72 hours, according to the manufacturer’s instructions. The obtained autoradiographic array images were scanned at 16 bits per pixel and 1200 dpi (25 μm) resolution, exported as 8-bit bitmap files and images processed using the Arrayexplorer© software ( Patriotis et al., 2001) and further analysed in Excel. Each gene array data set was normalized on the basis of the expression values of nine housekeeping genes included in the array (details available in Clontech Atlas array manual). Array data sets were subjected to pair-wise correlation analysis to establish reproducibility between experiments. The ratios between the gene intensities were calculated for each pair of data sets and the genes undergoing significant change in expression (greater than twofold) were identified.
For reverse-transcription PCR (RT-PCR), total RNA was isolated from HEF1.M1 cell populations that were either uninduced or induced to express HEF1 for 9 hours and DNase treated as described above. Following the protocol outlined in the Advantage RT-for PCR Kit (Clontech) with minor modifications, cDNA was generated from these samples and normalized using quantitative competitive template (QCT) RT-PCR with primers specific for actin using the Gene Express System 1A. Parallel PCRs were performed on a panel of dilutions of the two cDNA samples that were spiked with a constant amount of actin competitive template (CT). Following normalization for actin template levels, specific PCR analyses were performed using primer pairs specific for MLCK, p160ROCK, MDA7 and disintegrin/metalloprotease. For direct analyses of proteins nominated by mRNA analysis, lysates were generated from parallel populations of HEF1.M1 cells (uninduced or induced for ∼20 hours) and analyzed by immunoblotting using mouse monoclonal antibodies for human c-ErbB2 (NeoMarkers, Union City, CA), MMP1 and MMP14 (Chemicon International, Temecula, CA).
Establishment of cell lines that inducibly produce HEF1
To examine the consequences of HEF1 overproduction, we created stable cell lines derived from MCF7 breast adenocarcinoma cells. As endogenous HEF1 is produced at significant levels in MCF7 cells and many cell lines of epithelial origin ( Law et al., 1998), it appeared likely that HEF1 protein partners would be present in this context. A series of stable MCF7 clonal cell lines that expressed a tetracycline-regulatable HEF1 transgene or carried the parental vector were generated and characterized. Immunoblots of total cell lysates derived from representative HEF1-inducible clones (HEF1.M1-M.5) probed with HEF1-specific antibodies (αHEF1-SB-R1) showed that transgene expression was tightly regulated, as demonstrated by the appearance of full length HEF1 protein following removal of tetracycline ( Fig. 1A). By contrast, the parental vector clones CM1 and CM2 demonstrated no change in HEF1 production levels in the presence or absence of tetracycline ( Fig. 1A). To enhance visualization of tetracycline-induced HEF1 expression, the western blot shown was exposed relatively briefly. For this reason, endogenous HEF1 is not well visualized in this experiment.
To determine the kinetics of HEF1 production, total cell lysates were isolated from the HEF1.M1 clone at the indicated time intervals following induction upon tetracycline removal and analyzed on immunoblots probed with αHEF1-SB-R1 antibodies to evaluate HEF1 protein levels ( Fig. 1B). By six hours post-induction, HEF1 levels were significantly enhanced in induced HEF1.M1 lysates relative to those of uninduced HEF1.M1 lysates. Maximal HEF1 production was achieved by 9 hours of induction and maintained over the course of an induction spanning 24 hours. By contrast, lysates derived from the uninduced HEF1.M1 clone and the mock induced CM1 clone produced low levels of HEF1 protein, reflecting the expression of the endogenous protein, at all time intervals examined ( Fig. 1B, lanes 1 and 9, respectively).
Because HEF1 is closely related in sequence to p130Cas and enhanced levels of HEF1 may compete with p130Cas for shared interactive partners, characterizing the status of p130Cas in the context of HEF1 induction constituted an important control. First, in the control CM1 cells and in HEF1.M1 and HEF1.M2 cells, we have analyzed p130Cas levels in cells induced for 0-24 hours or left uninduced. Levels of p130Cas remain constant throughout the experiment, whereas levels of HEF1 increase in the HEF1.M1 and HEF1.M2 cell lines ( Fig. 1C). Second, although the similar migration of p130Cas species in samples prepared at different time points suggested that phosphorylation of p130Cas was not affected by HEF1 induction, we tested this point directly. The p130Cas and HEF1 proteins were immunoprecipitated from CM1, HEF1.M1 or HEF1.M2 cells that were induced for 0, 9 or 24 hours or left uninduced. Tyrosine phosphorylation was assessed using antibody to phosphotyrosine; blots were stripped and reprobed to compare phosphotyrosine levels to levels of immunoprecipitated p130Cas or HEF1. As shown, levels of phosphotyrosine were constant for p130Cas ( Fig. 1D), whereas robust tyrosine phosphorylation of induced HEF1 was observed (results not shown).
HEF1 production induces crescent morphology and cell spreading
We previously demonstrated that HEF1 overproduction mediates apoptosis in epithelial cells ( Law et al., 2000), whereas separate reports have shown that modulating HEF1 levels can contribute to T-cell migration ( Ohashi et al., 1999; van Seventer et al., 2001). To address the question of whether HEF1 also regulates epithelial cell shape and motility or whether cell-type-specific differences influence the spectrum of HEF1 activities, we characterized the MCF7-based cell lines for HEF1-dependent changes in cell morphology, substrate-dependent adhesion and movement. As prolonged overproduction of HEF1 induces apoptosis in MCF7 cells ( Law et al., 2000), we focused on the first 24 hours after tetracycline removal for this analysis.
Morphological changes consequent on HEF1 production were evident within 4-6 hours of induction, concomitant with the increase in HEF1 protein levels ( Fig. 1B; Fig. 2). Phase contrast microscopy of HEF1.M1 and HEF1.M2 following 18 hours of HEF1 induction revealed that the cells had undergone dramatic morphological changes. These were typified by the appearance of crescent-shaped cells with large leading edge lamellipodia, enhanced ruffling and a pronounced trailing edge ( Fig. 2, compare A to B and C to D), similar to the morphology typifying highly motile cells ( Cooper and Schliwa, 1986). Analysis of time-lapse video microscopy images taken from six experiments indicated that between 47% and 75% of the examined population of cells were crescents by 18-20 hours post-induction (data not shown). By contrast, crescent-shaped cells were not detected in uninduced HEF1 cells ( Fig. 2A,C) and uninduced or mock induced CM1 cells ( Fig. 2E,F). Qualitatively similar morphological changes were observed with other inducible HEF1 lines (results not shown). In addition, we performed a titration of tetracycline removal for the HEF1.M1 and HEF1.M2 lines, comparing samples incubated with 1 μg ml–1, 0.5 μg ml–1, 0.25 μg ml–1 and 0 μg ml–1 for HEF1 induction and morphological phenotype. Graded changes in HEF1-dependent phenotypes were observed, with reduced phenotypes at 0.25 μg ml–1 versus 0 μg ml–1 and marginally detectable phenotypic differences at 0.5 μg ml–1.
HEF1 and other Cas family proteins localize to focal adhesions in adherent cells ( Law et al., 1996; O’Neill et al., 2000). Immunofluorescence analysis with antibody to HEF1 ( Fig. 3A,D,G) and the focal adhesion proteins paxillin ( Fig. 3B,E) and FAK (not shown) indicated that HEF1 resided predominantly at focal adhesions in the induced HEF1.M1 cell line and was not mislocalized owing to overproduction. Most focal adhesions observed were found in the leading edge lamellipodia ( Fig. 3D-F). Visualization of HEF1 ( Fig. 3G) and the actin cytoskeleton ( Fig. 3H) revealed that HEF1 was concentrated at the distal ends of actin stress fibers, coinciding with focal adhesion sites ( Fig. 3I, arrowheads). HEF1-producing cells also exhibited reorganization of the actin cytoskeleton, with actin bundles arranged radially at the lamellipodial front and in stress fibers radiating out from the perinuclear region to the leading edge ( Fig. 3H,I). Based on observed paxillin and FAK staining, focal adhesions were more prominent in HEF1-producing cells than in uninduced HEF1 lines ( Fig. 3, compare B and E) or in either uninduced or mock induced vector control cell lines (data not shown). Together, these results suggest that enhanced levels of HEF1 might contribute to the recruitment of other focal adhesion component proteins. Finally, to evaluate the contribution of increased HEF1 production to cell spreading, HEF1.M1 cells and control cells were induced for 18 hours to facilitate the production of high levels of HEF1 and then replated and allowed to spread on FN-coated coverslips in serum-free media under non-inducing or inducing conditions ( Fig. 4). HEF1-producing cells were more highly spread at earlier time points ( Fig. 4, compare E and F, 2 hours after plating) and maintained an enhanced degree of spread relative to uninduced or control cells at 6 hours post-plating ( Fig. 4, compare I and J).
Cell spreading and focal complex assembly depend on contributions from both serum factors and extracellular matrix (ECM) components that engage integrin receptors ( Clark et al., 1998; Hotchin and Hall, 1995). To evaluate the contribution of serum factors and integrin engagement to HEF1-enhanced cell spreading, HEF1.M1 cells and controls were induced for 18 hours and then replated for an additional 6 hours with induction on uncoated coverslips in media containing serum or on FN-coated coverslips in serum-free media. Quantification of mean cell area for induced and uninduced populations revealed that HEF1 production resulted in a 2.2-times (uncoated coverslips) or a 1.75-times (FN-coated coverslips) increase in cell area ( Fig. 4, right panel). Analysis of controls demonstrated that the mean cell area of control uninduced HEF1.M1 cells was similar to that of uninduced or mock induced CM1 cells. These results indicate that FN is sufficient to enable HEF1-dependent spreading in the absence of any other serum factors.
HEF1 production enhances cell motility
The HEF1-dependent shape changes described above result in a morphology typical of highly motile cells. To examine the effect of HEF1 production on cellular motility directly, we followed two approaches. First, we measured the speed of movement of HEF1-induced versus control cells ( Fig. 5). To this end, phase contrast video images of uninduced and induced cell lines were recorded at 5-minute intervals using a CCD camera, compiled and analyzed to determine cell speed. The average speed of HEF1-producing HEF1.M1 cells (3.33 nm second–1±0.16 standard error) reflected a 26% increase over that of uninduced cells (2.65 nm second–1±0.11 standard error). By contrast, parallel analyses of the CM1 clone demonstrated that the average speed of these cells following removal of tetracycline (2.23 nm second–1±0.11 standard error) was similar to that of cells maintained in the presence of tetracycline (2.52 nm second–1±0.11 standard error) or the uninduced HEF1.M1 line. Moreover, detailed analyses showed that HEF1 production correlated with a greater proportion of cells in a population traveling at higher speeds ( Fig. 5A,B). For example, uninduced HEF1.M1 and parental vector cells never achieved the maximum speed of 6.0-7.0 nm second–1 attained by induced HEF1.M1 cells.
Interestingly, the quantitation of cell speed revealed that HEF1.M1 and CM1 cells moved in an oscillating fashion, with bursts of speed interspersed with slower phases. Such periodic oscillations in cell speed and associated internal force generation have previously been characterized in migrating neutrophils, macrophages and fibroblasts ( Galbraith and Sheetz, 1997; Ehrengruber et al., 1996; Sheetz et al., 1999). To determine whether HEF1-mediated enhancement of cell speed was a consequence of changes in the periodicity of this oscillating pattern, we quantified the highly motile phases of individual cells over the course of the assay. A highly motile phase was defined as a 5-minute interval during which a cell traveled with a speed equal to or greater than 2 nm second–1. These analyses revealed that the production of HEF1 did not alter the duration or frequency of the highly motile phase of individual cells. Together, these data demonstrate that HEF1 production facilitated faster movement of cells during motile phases but did not alter the oscillating pattern of movement.
Cas family proteins function as downstream components of integrin receptor signaling ( O’Neill et al., 2000). Therefore, we sought to determine whether HEF1 production enhanced FN-mediated haptotactic responses in MCF7 cells. To this end, we performed Boyden chamber assays using HEF1.M1, HEF1.M2 and CM1 cells, in the presence and absence of tetracycline ( Fig. 6A). All cell lines assessed exhibited FN-mediated haptotaxis. However, induction of HEF1 production specifically correlated with an approximately sixfold enhancement in migration. Similarly, induction of HEF1 in the HEF1.M2 clone (which produces lower levels of HEF1 than HEF1.M1 upon induction) conferred a three- to fivefold enhancement in haptotaxis compared with that of uninduced HEF1.M2 cells (data not shown).
MAPK and p38 kinase are activated by HEF1 but are not needed to enhance migration
Increasing cellular levels of HEF1 induces activation of JNK ( Law et al., 2000). Because activation of ERK ( Fincham et al., 2000; Glading et al., 2000; Klemke et al., 1997; Zeigler et al., 1999) and p38 kinase ( Matsumoto et al., 1999) has also been implicated in cellular motility, we investigated whether HEF1 production also activated these signaling cascades. HEF1 overproduction results in substantial activation of these kinases, as demonstrated by the appearance of phosphorylated forms of ERK (42/44 kDa) and p38 (48 kDa; Fig. 6B). To probe the requirement for these signaling pathways in HEF1-dependent haptotaxis, we examined the effect of MAPK-pathway-specific inhibitors in the Boyden chamber assay ( Fig. 6C). Both PD98059 (which inhibits ERK1/2 signaling) and SB202190 (which inhibits p38 signaling) reduced migration in cells overproducing HEF1 by ∼50%. We separately confirmed that treatment with PD98059 did not reduce levels of HEF1 (results not shown). Finally, to determine whether these compounds were inhibiting the basic cellular migration machinery or a HEF1-specific signaling pathway, we also examined the effects of PD98059 and SB202190 on the migration of cells uninduced for HEF1 production. We found that basal levels of cellular migration were also reduced by ∼50% in uninduced HEF1.M1 cells (data not shown), arguing that the inhibition of the MAPK pathways affects the basic migratory machinery rather than HEF1-specific pathways that hyperactivate cellular motility.
Downstream consequences of HEF1 overproduction: transcriptional induction of genes associated with motility
At present, little is known about the downstream targets whose expression is altered as a consequence of Cas family engagement. To begin to address this point, we compared the transcriptional profiles of uninduced or induced HEF1.M1 and CM1 cells. Because we were interested in changes in transcriptional response that might reflect HEF1 enhancement of cellular motility, mRNA was harvested 9 hours following tetracycline removal. cDNA prepared from these different mRNA populations was labeled with 33P-dATP and hybridized in parallel to Clontech Human Cancer 1.2 Atlas arrays, each containing 1176 immobilized cDNA fragments corresponding to genes functioning in diverse cellular processes. Pair-wise analysis of data obtained for each of the four experimental parameters (HEF1.M1 and CM1, with or without tetracycline) indicated that the profiles of hybridization were extremely similar for any two samples, such that the coefficient of correlation (R2) for all genes assessed was >0.90 ( Fig. 7). These data indicated that the variation between independently prepared samples was minimal and revealed that the induction of HEF1 altered the expression of a specific, limited set of genes.
Increased transcript levels of a significant number of genes encoding proteins known to be associated with cell migration, metastasis and transformation correlated with HEF1 overproduction ( Table 1). Multiple members of the matrix metalloproteinase family (MMP1, MMP8, MMP12, MMP13, MMP14) and disintegrin and a number of motility-associated kinases, including MLCK, the Nck-interacting kinase (NIK) and the Rho-binding kinase p160ROCK, were upregulated in response to HEF1 induction. The ephrin receptors (Eph-R5, -R1 and -R4) and specific ephrin-related ligands, proteins known to function in axonal guidance ( Holder and Klein, 1999), were also upregulated in this assay. Transcripts encoding extracellular matrix components, including a fibronectin precursor, collagen and heparin sulfate proteoglycan, were also elevated in level, as were transcripts encoding some transforming growth factor receptors. Notably, the transcript encoding the ErbB2 receptor, which is implicated in mammary cell differentiation and tumorigenesis, was upregulated.
To confirm these results, representative genes predicted to be transcriptionally amplified in response to HEF1 production were further examined by quantitative RT-PCR ( Fig. 8A) and/or western blot analysis ( Fig. 8B). PCR reactions performed on actin-normalized template with primers specific for MLCK, metalloproteinase and p160ROCK confirmed the array results ( Fig. 8A). Total cell lysates prepared from HEF1.M1 cells induced to produce HEF1 for 20 hours clearly contained higher levels of MMP1, MMP14 and ErbB2 than lysates derived from parallel cultures of uninduced HEF1.M1 cells ( Fig. 8B). Taken together, these confirmatory experiments supported the array results and, significantly, demonstrated that HEF1 production ultimately leads to increased levels of the MMP1, MMP14 and ErbB2 proteins.
Finally, because we have shown that the treatment of cells with ERK1/2 and p38 inhibitors diminished general haptotaxis towards FN but not HEF1-specific enhancement of cell motility ( Fig. 6), we evaluated whether treatment with these compounds inhibited HEF1-mediated induction of the MMP1 or MMP14 proteins. We found that treatment with PD90859 did not diminish HEF1 induction of either MMP1 or MMP14 (data not shown), arguing that the upregulation of these genes was not dependent on the activity of the ERK1/2 MAPK pathway. Although treatment with SB202190 did reduce the levels of MMP1 and MMP14 proteins, it also reduced the level of HEF1 by a similar degree, suggesting that induction of these MMPs was not dependent on p38 signaling. This last result agrees with a model that posits that the HEF1-dependent effect on migration proceeds through activation of the JNK pathway ( Dolfi et al., 1998).
Signaling through integrin receptors is a central coordinating element in biological processes including embryonic development, tumor metastasis, wound repair and inflammation. This broad spectrum of integrin-dependent activity reflects the fact that modulation of integrin signaling contributes to the decision of individual cells to proliferate, differentiate, migrate or undergo apoptosis. In this study, we focus on one defined downstream component of the integrin signaling machinery, the docking protein HEF1 ( Law et al., 1996; Minegishi et al., 1996). We have previously determined that prolonged overproduction of HEF1 induces apoptosis ( Law et al., 2000). Here, using a controlled inducible production system for HEF1 in vitro, we characterize a spectrum of early HEF1-mediated cellular responses including regulation of cell shape and enhancement of cellular motility. Using gene array filters, we identify a number of transcripts associated with extracellular matrix remodeling, cell motility and growth factor response that are rapidly upregulated in response to increased levels of HEF1. We confirm that a number of these upregulated transcripts correspond to upregulated proteins, including the matrix metalloproteinases MMP1 and MMP14, and the ErbB2 receptor tyrosine kinase. Together with our previous results ( Law et al., 2000), these data suggest that HEF1 may function in a complex manner potentially to promote or inhibit tumor growth in vivo.
Considerable evidence suggests that p130Cas ( Sakai et al., 1994) and the related molecules HEF1/CasL ( Law et al., 1996; Minegishi et al., 1996) and Efs/Sin ( Alexandropoulos and Baltimore, 1996; Ishino et al., 1995) are control points for information processing at focal adhesions ( O’Neill et al., 2000). Initial work focused on characterizing modifications of Cas family proteins that occur in response to transient or permanent changes in cellular adhesion status, such as those occurring during attachment ( Petch et al., 1995; Vuori and Ruoslahti, 1995), migration ( Cary et al., 1998; Klemke et al., 1998), mitosis ( Law et al., 1998; Yamakita et al., 1999) or apoptosis ( Kook et al., 2000; Law et al., 2000; Weng et al., 1999). These studies have indicated that specific modification of the Cas family proteins by phosphorylation and, in some cases, defined proteolytic cleavage occurs in response to discrete biological stimuli related to adhesive status.
As the targeting of these molecules for modification suggests that they are components of cellular signaling pathways that sense adhesive status, more recent work has explored whether cell-adhesion-related processes, such as motility and apoptosis, are affected by altering Cas protein expression. In particular, integration of information concerning Cas family signaling from integrin receptor to nuclear response is a goal of current research. This work has begun to define an axis of signaling that links Cas family proteins to the GTPases Ras, Rac1, and Rap1 and Src, and eventually leads to activation of MAPK and JNK cascades ( Alexandropoulos and Baltimore, 1996; Almeida et al., 2000; Burnham et al., 2000; Cheresh et al., 1999; Cho and Klemke, 2000; Law et al., 2000; Xing et al., 2000). Transcriptional regulation by Efs/Sin and p130Cas of some targets known to be downstream of ERK and SAPK signaling has been demonstrated [such as genes regulated through serum response elements ( Alexandropoulos and Baltimore, 1996; Hakak and Martin, 1999)]. Additional studies focused on defining proteins that interact with Cas family members have identified proteins such as Id2 ( Law et al., 1999), CHAT/Nsp3/SHEP1 ( Dodelet et al., 1999; Lu et al., 1999; Sakakibara and Hattori, 2000) and CIZ ( Nakamoto et al., 2000), which provide additional potential links to transcriptional response. At present, knowledge of the ultimate targets of transcriptional activation and repression by Cas family proteins is minimal. However, p130Cas signaling through its binding partner CIZ has been linked to the transcriptional upregulation of MMP-1, MMP-3 and MMP-7 by a direct mechanism ( Nakamoto et al., 2000) via binding of CIZ to a (G/C)AAAAA(A) consensus present in the promoter of these and other ( Furuya et al., 2000) potential target genes.
The primary goal of the current study was to use a well-controlled system to delineate a framework of HEF1-dependent activities that would allow determination of whether HEF1 and Cas acted in parallel, opposed or wholly distinct signaling processes in epithelial cells. A second goal was to identify HEF1-responsive targets that might be responsible for mediating the biological functions of the protein. The HEF1-responsive lines described here provide a useful system with which to identify downstream targets whose transcript levels are altered in response to HEF1 production. In order to address its role in mediating migration, we focused on changes in transcriptional regulation that occur relatively early (9 hours after tetracycline removal; 3 hours after the first detectable elevation of HEF1 levels) in the cellular response to HEF1 production. At this and later time points, levels and phosphorylation of p130Cas are unchanged ( Fig. 1), allowing us to ascribe observed biological effects specifically to the activity of HEF1.
The initial candidates we have isolated through analyses using cDNA arrays are intriguing. For example, the Rho-associated kinase and effector p160ROCK ( Fujisawa et al., 1996) has been shown to function in the control of cellular motility during developmentally significant processes such as neuronal outgrowth ( Bito et al., 2000). MMPs remodel the ECM by digesting constituent proteins, thereby promoting cellular migration and invasiveness in vivo. These are significant downstream consequences of HEF1 induction, as these proteins are required for many developmental processes and in metastasis of cancerous cells ( McCawley and Matrisian, 2000; Vu and Werb, 2000). MMP1 has previously been reported to be a downstream target of p130Cas activation, induced through the action of CIZ ( Nakamoto et al., 2000). MMP14 (also known as MT1-MMP) is a novel Cas-family target and is interesting because it is a member of a family of structurally distinct, membrane-associated metalloproteinases that function both as classical metalloproteinases that directly degrade the ECM and as enzymes that cleave and therefore activate other metalloproteinases in zymogen form ( Apte et al., 1997; Murphy et al., 1999). Scrutiny of the recently described promoter region of MMP14 ( Lohi et al., 2000) reveals at least two matches to the proposed consensus for CIZ binding near the transcriptional start site of the gene, suggesting that this transcript may be coordinately regulated with other MMPs. The detected elevation in levels of ErbB2 transcripts is also of considerable interest, insofar as transcriptional upregulation of ErbB2 is a frequent marker of poor prognosis for breast cancer ( Bates and Hurst, 1997), and ErbB2 overproduction has been shown to promote Cas/Crk coupling and cell invasion ( Spencer et al., 2000). These results emphasize that HEF1 signaling may function in a manner conducive to the promotion of cancer and may reciprocally regulate the factors shown elsewhere to regulate Cas family members. It is also intriguing to note that HEF1 production is correlated with an increase in the transcript levels of several genes encoding ECM components ( Table 1), a finding consistent with a recent study that linked the upregulation of a number of ECM proteins to metastatic capacity ( Clark et al., 2000) and also interesting in light of the report that CIZ induces type 1 collagen ( Furuya et al., 2000).
In contrast to this work, which identifies and characterizes HEF1-dependent changes in cell shape and motility, other work from our laboratory has indicated that sustained induction of HEF1 over 24-48 hours results in the induction of apoptosis and cellular detachment ( Law et al., 2000; O’Neill and Golemis, 2001). Insight into the means by which one protein leads to such disparate effects may be derived from a study of HEF1 processing because, at late time points after exogenous overproduction of HEF1, HEF1 is cleaved by caspases and subject to selective degradation by the proteasome such that full-length HEF1 is replaced by a C-terminally derived p28 species ( Law et al., 2000; O’Neill and Golemis, 2001). These forms are not produced solely as a consequence of overproduction; rather, analysis of the processing of endogenous HEF1 during three discrete processes involving cell rounding, including mitosis ( Law et al., 1998), apoptosis ( Law et al., 2000) or detachment ( O’Neill and Golemis, 2001) reveals that similar replacement of full-length with truncated protein forms occurs. Finally, we have shown that expression of the p28 cleavage product is sufficient to produce cell rounding and apoptosis. Based on these results and the present study, it seems likely that it is the change in relative levels between the full-length and cleaved forms of HEF1 that governs the shift from enhanced cell spreading and hypermotility to apoptosis. Indeed, we have found that the hypermotile period described in this study is followed by increasing elongation of HEF1-producing cells to narrow crescents and, finally, a sudden release of cell-matrix contacts, as cells recoil to a rounded, presumably pre-apoptotic, form. This proposal is consistent with others that link cellular morphology, tension and death ( Chicurel et al., 1998). One mechanism by which this transition may occur is by competition of accumulating p28 for partners of the residual full-length HEF1 and p130Cas proteins, disrupting their docking function at focal adhesions ( O’Neill et al., 2000) and promoting anoikis ( Frisch and Ruoslahti, 1997; Hungerford et al., 1996). Although we currently favor this mechanism, an alternative possibility is that the released p28 protein can interact with partner molecules such as the Id2 protein, inducing signaling relevant to apoptosis ( Law et al., 1999).
Finally, studies during the past year provide the first clinical evidence that perturbation of Cas family expression can result in significant differences in cancer progression in humans ( Brinkman et al., 2000; van der Flier et al., 2000). For example, p130Cas was re-isolated as the product of the BCAR1 (breast cancer resistance 1) locus, whose overexpression is associated with resistance to anti-estrogens. Strikingly, the recent cloning of BCAR3, a separate gene with properties similar to BCAR1 in altering response to estrogens, has revealed its product to be identical to AND-34 ( Cai et al., 1999; Gotoh et al., 2000) and Nsp2 ( Lu et al., 1999). AND-34 was isolated based on its interaction with p130Cas and HEF1, whereas Nsp2 was defined as an adaptor protein linking integrin signaling and JNK activation. Of further interest, the BCAR3/AND-34/Nsp2 protein is closely related to the CHAT/SHEP1/Nsp3 protein ( Dodelet et al., 1999; Lu et al., 1999; Sakakibara and Hattori, 2000) noted above, which is known to interact with the C terminus of Cas family members and to activate JNK signaling. Together with the data in the current study, these findings begin to reveal a profile of Cas-family function that initiates at focal adhesions and ultimately alters the transcriptional regulation of a number of genes, thereby influencing a spectrum of cellular processes.
We are grateful to J. Boyd for expert help with microscopy and image rendering. We thank R. Stoyanova and B. Gruver for advice and help, and the labs of A. Klein-Szanto and G. Adams for reagents. We thank B. Zahorchak for the GeneExpress reagents. We are very grateful to T. Coleman, M. A. Sells, J. van Seventer, G. Kruh and J. Chernoff for reviewing the manuscript. This work was supported by NIH grants RO1 CA63366 (to E. A. G.) and R29-CA73676 (to C. P.), and core funds CA-06927 (to Fox Chase Cancer Center). S. J. F. was supported by NIH grant T32 CA09035 and later by Department of Defense Breast Cancer Research Program grant DAMD 17-00-1-0249, managed by the US Army Medical Research and Materiel Command. G. M. O’N. was supported by a W. J. Avery Fellowship provided by the Connelly Foundation.
- Accepted September 24, 2001.
- © The Company of Biologists Limited 2002