The signaling pathways linking integrins to nuclear events are incompletely understood. We have examined intracellular signaling by the α6β4 integrin, a laminin receptor expressed in basal keratinocytes and other cells. Ligation of α6β4 in primary human keratinocytes caused tyrosine phosphorylation of Shc, recruitment of Grb2, activation of Ras and stimulation of the MAP kinases Erk and Jnk. In contrast, ligation of the laminin‐ and collagen‐binding integrins α3β1 and α2β1 did not cause these events. While the stimulation of Erk by α6β4 was suppressed by dominant‐negative Shc, Ras and RhoA, the activation of Jnk was inhibited by dominant‐negative Ras and Rac1 and by the phosphoinositide 3‐kinase inhibitor Wortmannin. Adhesion mediated by α6β4induced transcription from the Fos serum response element and promoted cell cycle progression in response to mitogens. In contrast, α3β1‐ and α2β1‐dependent adhesion did not induce these events. These findings suggest that the coupling of α6β4 integrin to the control of cell cycle progression mediated by Shc regulates the proliferation of basal keratinocytes and possibly other cells which are in contact with the basement membrane in vivo.
In addition to promoting cell adhesion and contributing to the organization of tissues and organs, basement membranes exert complex and often divergent effects on the survival, proliferation and differentiation of epithelial cells (Adams and Watt, 1993; Lin and Bissel, 1993). The influences of basement membranes on epithelial cells are likely to be mediated by the ability of laminin‐binding integrins to induce intracellular signaling (Giancotti and Mainiero, 1994; Clark and Brugge, 1995; Schwartz et al., 1995), but the mechanisms involved are incompletely understood.
To elucidate the effects of laminins on epithelial cells, we have focused on the α6β4 integrin. Cell adhesion assays with α6β4‐transfected K562 cells and radioligand binding studies with purified recombinant α6β4 have indicated that this integrin is a receptor for various laminin isoforms and binds with the highest apparent affinity to laminins 5 and 4 (Niessen et al., 1994; Spinardi et al., 1995). In accordance with its role as a basement membrane receptor, α6β4 is expressed in epithelial cells (Kajiji et al., 1989), Schwann cells (Sonnenberg et al., 1990; Einheber et al., 1993) and a subset of endothelial cells (Kennel et al., 1992; Klein et al., 1993) and thymocytes (Wadsworth et al., 1992).
The α6β4 integrin has a distinctive structure and subcellular localization. The large cytoplasmic domain of the β4 subunit, which is characterized by two pairs of type III fibronectin (Fn)‐like domains separated by a 142 amino acid sequence (connecting segment), does not contain any region of homology with the cytoplasmic domains of other known integrin β subunits (Hogervorst et al., 1990; Suzuki and Naitoh, 1990). Furthermore, while β1 and αv integrins are concentrated in focal adhesions and linked to the actin filament system, the α6β4 integrin is found in hemidesmosomes both in vivo and in cultured cells (Carter et al., 1990a; Stepp et al., 1990). These observations suggest that the unique cytoplasmic domain of β4 interacts with cytoskeletal elements of hemidesmosomes, thereby linking α6β4 to the keratin filament system.
Our previous studies have indicated that the association of α6β4 with the hemidesmosomal cytoskeleton is mediated by the cytoplasmic domain of β4 and specifically by a region which includes the first pair of type III Fn‐like repeats and the connecting segment (Spinardi et al., 1993). In accordance with the hypothesis that α6β4 plays a crucial role in the assembly of hemidesmosomes and their linkage to the keratin filament system, we have observed that the introduction of a truncated tail‐less β4 subunit into cells possessing endogenous α6β4 integrins and hemidesmosomes results in a dominant‐negative effect on hemidesmosome assembly (Spinardi et al., 1995). Since the tail‐less integrin binds efficiently to extracellular ligand, its dominant‐negative effect is likely to result from its ability to co‐cluster with the endogenous wild‐type receptor and block a signal necessary for hemidesmosome assembly. In accordance with this hypothesis, recent studies have revealed that α6β4 is associated with an intracellular tyrosine kinase. Mutagenesis experiments have provided evidence that the phosphorylation of a tyrosine activation motif (TAM) located in the connecting segment controls the association of α6β4 with the hemidesmosomal cytoskeleton, presumably via the recruitment of a signaling molecule containing two tandem Src homology 2 (SH2) domains (Mainiero et al., 1995).
In the epidermis and other stratified epithelia, the expression of α6β4 is restricted to the basal cell layer which contains cells endowed with proliferative capacity (Kajiji et al., 1989). It is known that keratinocytes exit the cell cycle and begin their differentiation program when they detach from the basement membrane to migrate to the upper epidermal layers (Hall and Watt, 1989). In fact, this process can be replicated in vitro by depriving cultured keratinocytes of anchorage to their endogenously produced extracellular matrix (Green, 1977), which is particularly rich in laminin 5 (Carter et al., 1991; Rousselle et al., 1991). Furthermore, squamous carcinoma cells endowed with high proliferative potential often express elevated levels of α6β4 (Kimmel and Carey, 1986; Wolf et al., 1990). Finally, the basal keratinocytes of β4 knock‐out mice display signs of degeneration even in areas of epidermis where no significant detachment from the basement membrane is observed (Dowling et al., 1996). These observations suggest that α6β4 may provide epithelial cells with a signal important for cell survival and cell cycle progression.
What is the mechanism by which the α6β4 integrin transduces biochemical signals capable of affecting cell proliferation? Immunoprecipitation and GST fusion protein binding experiments have indicated that ligation of α6β4 results in the association of the adaptor protein Shc with tyrosine‐phosphorylated β4. Shc is then phosphorylated on tyrosine residues, presumably by the integrin‐associated kinase, and combines with the other adaptor protein Grb2 (Mainiero et al., 1995). These observations raise two key questions. First, what are the biochemical consequences of the recruitment of Shc and Grb2 to α6β4? Second, what is the biological significance of α6β4 signaling in epithelial cells?
In this study, we provide evidence that the α6β4 integrin stimulates the Ras–Erk and Rac–Jnk mitogen‐activated protein kinase (MAP kinase) signaling pathways via Shc and thereby controls immediate‐early gene expression and keratinocyte proliferation in response to laminin.
Ligation of α6β4 causes activation of the Ras–Erk signaling pathway
To examine the intracellular signaling pathways activated by the α6β4 integrin in a physiologically relevant cellular context, we elected to use primary human keratinocytes. These cells express high levels of α6β4, α2β1 and α3β1 and adhere to laminin 5 through α6β4 and α3β1 (Xia et al., 1996) and to collagen I through α2β1 (Carter et al., 1990b). Incubation of suspended keratinocytes with polystyrene beads coated with the anti‐β4 monoclonal antibody (Mab) 3E1 as well as adhesion to laminin 5‐coated dishes caused tyrosine phosphorylation of the 52 and 46 kDa mol. wt isoforms of Shc (the 66 kDa isoform of Shc is expressed at very low levels in keratinocytes) and thereby recruitment of Grb2 (Figure 1A). No significant tyrosine phosphorylation of Shc and recruitment of Grb2 was observed in keratinocytes treated with beads coated with the anti‐α3β1 Mab P1B5 (not shown) or the control anti‐MHC Mab W6.32 and in keratinocytes adhering to collagen I or poly‐l‐lysine‐coated dishes (Figure 1A). The inability of collagen I to induce tyrosine phosphorylation of Shc and recruitment of Grb2 was not a consequence of insufficient adhesion because the keratinocytes spread equally well on laminin 5 and collagen I under our experimental conditions. Furthermore, we observed that antibody‐mediated cross‐linking of α2β1 does not induce tyrosine phosphorylation of Shc and recruitment of Grb2 (not shown). These findings, which are consistent with previous results (Mainiero et al., 1995; Wary et al., 1996), indicate that ligation of α6β4, but not α2β1 and α3β1, can promote signaling mediated by Shc in primary keratinocytes.
We next examined the role of cytoplasmic domain of β4 in the recruitment and tyrosine phosphorylation of Shc. Rat bladder 804G cells expressing either a recombinant full‐length human β4 subunit (clone A) or a tail‐less version (clone L) were cross‐linked in suspension with the anti‐β4 Mab 3E1 or the control anti‐MHC Mab W6.32. Immunoprecipitation with the 3E1 Mab followed by immunoblotting with anti‐Shc antibodies revealed that the recruitment of all three isoforms of Shc by α6β4 requires the cytoplasmic domain of the β4 subunit (Figure 1B). To examine if recruitment to α6β4 was a prerequisite for tyrosine phosphorylation of Shc, clone A and clone L were either cross‐linked in suspension with the 3E1 or the W6.32 Mab or plated onto dishes coated with the same antibodies. Immunoprecipitation with anti‐Shc antibodies followed by immunoblotting with anti‐P‐Tyr antibodies indicated that ligation of wild‐type α6β4 induces tyrosine phosphorylation of the 52 kDa isoform of Shc (the major isoform recruited to activated α6β4 in these cells). In contrast, ligation of tail‐less α6β4 did not cause efficient tyrosine phosphorylation of Shc (Figure 1C). These results are consistent with the hypothesis that the recruitment of Shc to activated α6β4 is mediated by tyrosine phosphorylation of the cytoplasmic domain of β4. They also suggest that this event is necessary for subsequent tyrosine phosphorylation of the adaptor protein. Since in a number of experiments tail‐less α6β4 was able to induce a modest level of tyrosine phosphorylation of Shc, it is possible that an additional indirect mechanism contributes to the activation of Shc by α6β4.
Since Grb2 is stably associated with the Ras‐GTP exchange factor mSOS, the recruitment of Grb2 to the plasma membrane mediated by Shc is likely to bring mSOS in close proximity to its target Ras (Schlessinger, 1994). Ras‐GTP loading experiments were therefore performed to examine if ligation of α6β4 resulted in activation of Ras. After growth factor starvation and in vivo labeling with [32P]orthophosphate, primary human keratinocytes were detached and either incubated in suspension with anti‐β4 or anti‐MHC beads or replated on dishes coated with laminin 5 or anti‐β4 Mab. As a control, adherent keratinocytes were either left untreated or were stimulated with epidermal growth factor (EGF). As shown in Figure 2, chromatographic analysis of nucleotides bound to Ras indicated that adhesion to laminin 5‐ or anti‐β4 Mab‐coated dishes results in an accumulation of GTP on Ras comparable with that caused by EGF. In suspended cells, however, antibody‐mediated ligation of α6β4 did not cause activation of Ras (Figure 2). The results of this experiment indicate that α6β4‐mediated cell adhesion causes activation of Ras. They also suggest that, in contrast to the sequential recruitment of Shc and Grb2 which can occur in suspended cells treated with anti‐β4 beads, optimal activation of Ras by α6β4 requires adhesion and/or spreading on a substratum coated with α6β4 ligands.
We next examined if ligation of α6β4 resulted in activation of the MAP kinase Erk. Growth factor‐starved keratinocytes were detached and either kept in suspension or plated on dishes coated with the anti‐β4 Mab 3E1, the anti‐MHC Mab W6.32, laminin 5, collagen I or poly‐l‐lysine. As a control, adherent keratinocytes were treated with EGF. As shown in Figure 3A, immunoprecipitation and in vitro kinase assays indicated that adhesion to laminin 5‐ or anti‐β4 Mab‐coated dishes causes activation of Erk to a level similar to that induced by EGF. In contrast, adhesion to collagen I, poly‐l‐lysine or anti‐MHC Mab did not result in significant activation of Erk (Figure 3A). Adhesion to anti‐α3 Mab‐coated dishes also did not cause Erk activation (not shown). To examine the kinetics of Erk activation by α6β4, keratinocytes were plated on laminin 5 for various times and subjected to Erk immunoprecipitation and kinase assay. As shown in Figure 3B, we detected a significant activation of Erk as early as 5 min after plating the keratinocytes on laminin 5. The level of Erk activity peaked at 30 min of adhesion and declined thereafter. These results indicate that ligation of α6β4 causes a significant and relatively persistent activation of the MAP kinase Erk and are in agreement with the observation that ligation of the laminin‐ and collagen‐binding integrins α3β1 and α2β1 does not induce this event (Wary et al., 1996). Since adhesion mediated by α3β1 and α2β1 is known to cause activation of focal adhesion kinase (FAK), these findings are also consistent with the notion that activation of FAK is not sufficient for stimulation of Erk in response to integrin ligation (Wary et al., 1996).
The activation of Erk mediated by α6β4 requires Shc, Ras and Rho
The mechanism of Erk activation in response to α6β4 ligation was examined by testing the effect of various dominant‐interfering mutant proteins. Since transient transfection of primary keratinocytes is very inefficient, we elected to use HeLa cells which express levels of α6β4, α3β1 and α2β1 comparable with those of primary keratinocytes. HeLa cells were transfected with a hemagglutinin (HA)‐tagged Erk2 vector in combination with different concentrations of cDNAs encoding dominant‐negative Shc (317F), Ras (N17), RhoA (N19), CDC42 (N17) and Rac1 (N17). As shown in Figure 4, immunoprecipitation of HA‐Erk2 followed by in vitro kinase assay indicated that the activation of Erk in response to α6β4 ligation is suppressed by dominant‐negative Shc (Figure 4A), Ras and RhoA (Figure 4B), but not by dominant‐negative Cdc42 and Rac1 (Figure 4C). Although at the highest concentration tested dominant‐negative Shc, Ras and RhoA completely inhibited the activation of Erk by α6β4, at the intermediate and lowest concentration tested the inhibitory activity of the three dominant‐negative mutants differed, perhaps as a consequence of the different relative abundance of their target protein or their different mechanism of action. These results indicate that Shc couples the α6β4 integrin to the Ras–Erk signaling pathway and suggest that full activation of Erk in response to α6β4 ligation requires the activity of both Ras and Rho.
Ligation of α6β4 causes activation of the Rac–Jnk signaling pathway
We next examined if ligation of α6β4 stimulated the MAP kinase Jnk. Growth factor‐starved primary keratinocytes were detached and either kept in suspension or plated on dishes coated with anti‐β4 Mab, anti‐MHC Mab, laminin 5, collagen I or poly‐l‐lysine. As a control, adherent keratinocytes were exposed to UV light. Jnk was precipitated by using a GST–Jun fusion protein and its activity examined by in vitro kinase assay. As shown in Figure 5, the binding of Mab 3E1 or laminin 5 to α6β4 caused an activation of Jnk comparable with that induced by UV stimulation. In contrast, adhesion to dishes coated with collagen I, poly‐l‐lysine, the control Mab W6.32 (Figure 5) or the anti‐α3β1 Mab P1B5 (not shown) did not result in significant activation of Jnk. Time‐course experiments indicated that the kinetics of activation of Jnk in response to laminin 5 were similar to those observed for Erk (not shown). These results indicate that ligation of α6β4 causes activation of Jnk, and suggest that ligation of the collagen‐binding integrin α2β1 does not induce this event.
The activation of Jnk mediated by α6β4 requires Ras, PI‐3K and Rac
The mechanism of Jnk activation in response to α6β4 ligation was examined by testing the effect of various dominant‐interfering mutant proteins and the phosphoinositide 3‐kinase (PI‐3K) inhibitor Wortmannin. HeLa cells were transfected with a Flag‐tagged Jnk1 vector in combination with different concentrations of cDNAs encoding dominant‐negative Ras (N17), Rac1 (N17), Cdc42 (N17) and RhoA (N19). As shown in Figure 6A, precipitation of Flag‐Jnk1 followed by in vitro kinase assay indicated that the activation of Jnk in response to α6β4 ligation is suppressed by dominant‐negative Ras and Rac1, but not by dominant‐negative RhoA, and very modestly by dominant‐negative Cdc42. These results indicate that the activation of Jnk by α6β4 requires the activity of both Ras and Rac. Although it has been suggested that Raf can activate Jnk by acting on the MAP kinase kinase kinase MEKK1 (Lange‐Carter and Johnson, 1994), recent results indicate that the predominant mechanism by which Ras activates Jnk involves Rac, and not Raf (Kosravi‐Far et al., 1995; Minden et al., 1995; Qiu et al., 1995; Joneson et al., 1996). The requirement for Rac in our system is consistent with this hypothesis. Since PI‐3K is a downstream target effector of Ras (Rodriguez‐Viciana et al., 1994) and has been implicated in the activation of Rac (Nobes et al., 1995; Klippel et al., 1996), we tested if inhibition of PI‐3K interfered with the activation of Jnk by α6β4. As shown in Figure 6B, the activation of Jnk in response to ligation of α6β4 was completely suppressed by as little as 50 nM Wortmannin, a concentration at which the inhibitor has very little effect on signaling molecules other than PI‐3‐K (Wymann et al., 1996). Taken together, the results of these experiments suggest that α6β4 activates the Rac–Jnk signaling pathway via Ras and PI‐3K.
Induction of Fos SRE‐dependent transcription by α6β4
We next examined if the coupling of α6β4 to Shc played a role in the control of immediate‐early gene expression. Since Erk regulates transcription from the Fos serum response element (SRE) by phosphorylating the ternary complex factors Elk‐1 and SAP‐1, and Rho family proteins cooperate with this Erk function by acting on the serum response factor (Treisman, 1995), we examined the effect of α6β4 ligation on the Fos SRE. HeLa cells were transiently transfected with the Fos‐SRE–Luc vector, which contains the Fos SRE promoter element linked to the luciferase reporter gene. Upon plating on dishes coated with poly‐l‐lysine, collagen I or laminin 5, the cells were either left untreated or exposed to EGF. They were then subjected to luciferase assay. As shown in Figure 7, while adhesion to laminin 5 in the absence of EGF caused elevation of Fos SRE‐dependent transcription, adhesion to poly‐l‐lysine or collagen I under the same conditions did not induce this activity. This suggests that ligation of α6β4, but not α2β1, is sufficient to promote transcription from the Fos SRE in the absence of mitogens. Treatment with EGF induced a significant elevation of Fos SRE activity in HeLa cells adhering to laminin 5, but caused a remarkably modest effect in cells attaching to poly‐l‐lysine or collagen I. This result suggests that ligation of α6β4 is required for optimal induction of Fos SRE‐dependent transcription in response to EGF.
To examine if laminin 5 is able to induce transcription from the Fos SRE in normal untransformed cells and to demonstrate the role of α6β4 in this process, we transfected NIH 3T3 fibroblasts with a retroviral vector encoding human β4 and isolated stable cell lines. Immunoprecipitation and fluorescence‐activated cell sorting (FACS) analysis indicated that the recombinant β4 subunit associated with endogenous α6 and was regularly exported to the cell surface. Clones 8 and 18 displayed the highest levels of recombinant β4 on the cell surface and were examined further. FACS analysis indicated that the level of expression of recombinant β4 in these two clones approximated 40% of that of endogenous β4 in primary keratinocytes. Since the NIH 3T3 cells do not express α6β4 and adhere to laminin 5 through α3β1, we examined the function of recombinant α6β4 by comparing the kinetics by which the β4 transfectants and control cells adhered to laminin 5‐coated dishes. The results showed that the β4 transfectants adhered to laminin 5 with faster kinetics than the controls. However, both types of cells became equally spread by 30 min of plating (not shown).
To analyze Fos SRE‐dependent transcription in reponse to laminin 5, control and β4‐transfected NIH 3T3 cells were transiently transfected with the Fos‐SRE–Luc plasmid. Upon plating on dishes coated with poly‐l‐lysine or laminin 5, the cells were either left untreated or exposed to basic fibroblast growth factor (bFGF), insulin and platelet‐derived growth factor (PDGF). The results of luciferase assays indicated that adhesion to laminin 5 causes elevation of Fos SRE‐dependent transcription in the β4 transfectants even in the absence of mitogens and this induction is potentiated by mitogen treatment (Figure 7). The ability of laminin 5 to induce Fos SRE‐dependent transcription was dependent on α6β4 expression, because adhesion to laminin 5 did not cause this effect in control cells even after mitogen treatment (Figure 7). These results indicate that ligation of α6β4, but not α3β1, is sufficient to induce Fos SRE‐dependent transcription, and confirm that α6β4 cooperates with mitogens to cause optimal induction of this activity.
Adhesion mediated by α6β4 promotes cell cycle progression
To examine if α6β4 signaling played a role in cell cycle progression, primary keratinocytes were growth factor starved and then plated in the presence of EGF on plastic wells coated with laminin 5, collagen I or poly‐l‐lysine. Entry into the S phase was examined by 5′‐bromo‐2′‐deoxyuridine (BrdU) incorporation and anti‐BrdU staining. As shown in Figure 8, a significant fraction of keratinocytes plated on laminin 5 entered into the S phase during the 22 h of the assay. In contrast, only a modest percentage of cells plated on collagen I or poly‐l‐lysine entered into S during the assay. In the absence of mitogens, a similarly small percentage of cells plated on collagen I or poly‐l‐lysine entered into S phase (not shown). This fraction may consist of unsynchronized cells, which have already passed the G1–S boundary at the time of plating. In addition, because the keratinocytes acquired and maintained a well‐spread morphology on collagen I, their inability to enter into S on this substratum is not the result of insufficient spreading. The results of these experiments indicate that physical attachment and spreading on the extracellular matrix is not sufficient for progression of keratinocytes through G1 in response to EGF, and suggest that this process requires ligation of a specific integrin, such as α6β4.
To examine the relative roles of α6β4 and α3β1 in keratinocyte proliferation, we tested the effect of inhibitory anti‐β4 and anti‐β1 antibodies. Growth factor‐starved keratinocytes were plated on laminin 5 and exposed to EGF in the presence of the inhibitory anti‐β1 Mab 4B4, the inhibitory anti‐β4 Mab A9 or the control anti‐MHC Mab W6.32. As shown in Figure 8, exposure to the anti‐β4 Mab completely suppressed keratinocyte entry into S. In contrast, treatment with the anti‐β1 or anti‐MHC Mab did not inhibit keratinocyte proliferation on laminin 5. To control the efficacy of the anti‐β1 Mab 4B4, G0 syncronized primary human fibroblasts were plated on a mixed substrate consisting of poly‐l‐lysine and fibronectin and exposed to mitogens in the presence of the 4B4 or W6.32 Mab. In accordance with the recent observation that a class of β1 integrins, which include the α5β1 fibronectin receptor, is linked to the Ras–Erk pathway and the control of cell cycle progression by Shc (Wary et al., 1996), plating of the primary fibroblasts on fibronectin/poly‐l‐lysine promoted cell cycle progression, and exposure to anti‐β1 Mab 4B4 blocked this process without inducing detachment (Figure 8). These results suggest that the ability of laminin 5 to promote keratinocyte cell cycle progression is mediated by α6β4, and not by α3β1.
We next examined the ability of control and β4‐expressing NIH 3T3 cells to progress through G1 on laminin 5. While only a modest percentage of control cells progressed through G1 when plated on laminin 5 for 22 h, a significant fraction of β4 expressors entered into S under the same conditions (Figure 8), suggesting that ligation of α6β4 is sufficient to promote progression through G1 in response to mitogens. Taken together, the results of these assays indicate that ligation of α6β4 is required and sufficient to promote keratinocyte proliferation in response to laminin 5.
Although the notion that cell adhesion to the extracellular matrix regulates gene expression is supported by considerable experimental evidence, the signaling pathways linking integrins to nuclear events are not well known. In particular, the mechanisms by which integrin‐dependent signals regulate cell cycle progression in normal epithelial cells are not fully understood. The results of recent studies have defined the membrane‐proximal events induced by ligation of the α6β4 integrin, a laminin receptor involved in various morphogenetic processes (Giancotti, 1996). Upon binding to extracellular ligand, α6β4 becomes phosphorylated on tyrosine residues by the action of an integrin‐associated kinase and thereby combines sequentially with the adaptor proteins Shc and Grb2 (Mainiero et al., 1995). The results of the present study provide clear evidence that these receptor‐proximal events result in the activation of Ras and of two distinct MAP kinase signaling pathways which regulate immediate‐early gene expression. In contrast, other integrins, such as the α2β1 and α3β1 collagen and laminin receptors, do not induce these events. Since α6β4‐mediated adhesion promotes keratinocyte progression through G1 in response to growth factor treatment, while α2β1‐ and α3β1‐mediated adhesion does not, we propose that the linkage of α6β4 to Ras signaling mediated by Shc participates in the control of cell cycle progression in normal epithelial cells.
The adaptor protein Shc contains two separate domains involved in the recognition of tyrosine‐phosphorylated sequence motifs: an N‐terminal phosphotyrosine‐binding (PTB) domain and a C‐terminal Src homology 2 (SH2) domain (Pawson, 1995). GST fusion protein binding experiments have suggested that both domains can interact with the tyrosine‐phosphorylated β4 tail (Mainiero et al., 1995). The results of this study are in accordance with the notion that the recruitment of Shc to α6β4 is mediated by the cytoplasmic domain of β4 and suggest that this event is important for the subsequent tyrosine phosporylation of Shc, which is presumably mediated by the integrin‐associated kinase. Upon phosphorylation, Shc combines with the other adaptor protein Grb2. Since Grb2 is constitutively associated with the Ras‐GTP exchange factor mSOS, the recruitment of Grb2 to the plasma membrane potentially links α6β4 to Ras. In accordance with this hypothesis, our current results show that ligation of α6β4 results in a significant activation of Ras.
In contrast to the recruitment of Shc and Grb2 which could be observed in suspended keratinocytes cross‐linked with anti‐α6β4 antibodies, full activation of Ras required physical attachment and/or spreading on a substratum coated with α6β4 ligands. Since it has been suggested that proper targeting of the Grb2–mSOS complex to Ras may require an interaction of the Grb2 SH3 domains with the cortical cytoskeleton (Bar‐Sagi et al., 1993), it is possible that such targeting is defective in suspended keratinocytes cross‐linked with anti‐α6β4 antibodies. Alternatively, since the pleckstrin homology domain of mSOS may bind to phosphatidylinositol(4,5)bisphosphate (PtdInsP2) in the plasma membrane (Lemmon et al., 1996) and it is known that PtdInsP2 levels decline in suspended cells (McNamee et al., 1993), it is possible that Ras activation by α6β4 requires a threshold concentration of PtdInsP2 in the plasma membrane which is not available in suspended keratinocytes. Future studies will be required to resolve this issue.
The results of this study indicate that ligation of α6β4 results in the stimulation of both Ras–Erk and Rac–Jnk MAP kinase signaling pathways. The activation of Erk by α6β4 was suppressed by dominant‐negative versions of both Shc and Ras, indicating that the coupling to Ras mediated by Shc is the major mechanism by which α6β4 controls Erk activation. Interestingly, Erk activation was also inhibited by dominant‐negative RhoA. This result, which is in agreement with the recent observation that Rho activity is required for full activation of Erk in response to various extracellular stimuli (Hill et al., 1995), suggests that this G protein also participates in signaling by α6β4. The activation of Jnk by α6β4 was inhibited by dominant‐negative Ras and Rac1, but not by dominant‐negative RhoA and Cdc42. In addition, it was suppressed by nanomolar concentrations of the PI‐3K inhibitor Wortmannin. Since there is evidence that PI‐3K is a downstream target effector of Ras and is involved in the activation of Rac (Rodriguez‐Viciana et al., 1994; Nobes et al., 1995; Klippel et al., 1996), it is likely that α6β4 stimulates the Rac–Jnk pathway via Ras. Thus, the coupling of α6β4 to Ras mediated by Shc leads to the activation of both Ras–Erk and Rac–Jnk signaling pathways.
In accordance with the observation that Erk stimulates transcription of the immediate‐early gene fos (Treisman, 1995), the results of our study indicate that adhesion mediated by α6β4 is sufficient to promote transcription from the Fos SRE. Interestingly, while treatment with mitogens caused a significant elevation of Fos SRE activity in cells plated on the α6β4 ligand laminin 5, it was ineffective in cells adhering to the α2β1 ligand collagen I, indicating that the expression of Fos in response to mitogens requires ligation of a specific integrin, such as α6β4. Future studies will be required to examine further the mechanism by which α6β4 controls immediate‐early gene expression. For example, it is known that Rho family proteins can activate the Fos promoter by stimulating the serum response factor (Hill et al., 1995). The ability of α6β4 to stimulate Rac may thus contribute to the activation of Fos promoter in response to laminin 5. In addition, since it is well established that Jnk controls the activity of the Jun promoter (Karin, 1995), it is likely that α6β4 also regulates the expression of the immediate‐early expression gene Jun. Taken together, these observations suggest that α6β4 is a crucial regulator of immediate‐early gene expression.
What is the biological significance of α6β4 signaling? The results of our cell proliferation analysis indicate that α6β4 signaling promotes transit through G1 in keratinocytes and other α6β4‐expressing cells exposed to mitogens. In this respect, α6β4 appears to be functionally distinct from other integrins, such as α3β1 and α2β1, which do not appear to be able to do so. In fact, it is quite remarkable that keratinocytes plated on the α2β1 ligand collagen I adhere and spread but do not enter into the S phase despite being exposed to otherwise mitogenic concentrations of EGF. We have observed recently that a class of β1 and αv integrins, which include α1β1, α5β1 and αvβ3, but not α2β1, α3β1 and α6β1, are also linked to the Ras–Erk pathway and the control of cell cycle progression by Shc (Wary et al., 1996). Taken together, these observations suggest that the ability of extracellular matrix to control cell proliferation, thereby mediating anchorage‐dependent cell growth, depends on its composition and the repertoire of integrins on the responding cell.
What is the mechanism by which α6β4 signaling regulates cell proliferation? Previous studies have indicated that the cell cycle of normal cells contains an anchorage‐dependent transition in early–mid G1. In fact, adhesion to the extracellular matrix is required for translation of cyclin D1 in cells exposed to mitogens, suggesting that integrin‐ and growth factor‐dependent signals converge prior to the induction of cyclin D1 to control progression through G1 (Fang et al., 1996; Zhu et al., 1996). Our results suggest that these signals are integrated before the induction of immediate‐early gene expression. The simplest hypothesis is that in normal cells growth factor receptors and specific integrins cooperate to activate MAP kinase beyond the threshold level required for immediate‐early gene expression. Since most dominant oncogenes, including Shc (Pelicci et al., 1992), induce neoplastic transformation by constitutively activating the Ras–MAP kinase pathway, this model also explains why neoplastic cells usually display anchorage‐independent growth.
In conclusion, the results of this study indicate that the coupling of α6β4 integrin to Ras–Erk and Rac–Jnk pathways mediated by Shc regulates immediate‐early gene expression and cell cycle progression in response to mitogens. Since the major keratinocyte integrins, in addition to α6β4, are α2β1 and α3β1, which are not coupled to Shc, the signaling function of α6β4 may explain why cell proliferation is restricted to the basal layer in the epidermis and other stratified epithelia. In addition, since exit from the cell cycle is a prerequisite for differentiation, our results may also explain why the onset of keratinocyte differentiation is coupled to the detachment from the basement membrane (Hall and Watt, 1989). Finally, the ability of α6β4 to stimulate cell growth suggests that its overexpression in squamous carcinoma (Kimmel and Carey, 1986; Wolf et al., 1990) may contribute to tumor progression.
Materials and methods
Antibodies and extracellular matrix molecules
The specificity of anti‐β4 Mab 3E1 was described previously (Giancotti et al., 1992). The polyclonal antiserum to the β4 ectodomain was generated by immunizing a rabbit with a GST fusion protein comprising amino acids 31–217. The inhibitory anti‐β4 Mab A9 was obtained from Tom Carey (Comprehensive Cancer Center, University of Michigan at Ann Arbor). Hybridomas producing the anti‐α1 Mab TS2/7 and anti‐β1 Mab TS2/16 were obtained from ATCC (Rockville, MD). The anti‐α2 Mab P1E6, anti‐α3 Mab P1B5 and anti‐α5 Mab P1D6 were from Gibco‐BRL (Gaithersburg, MD). The anti‐β1 Mab 4B4 was from Coulter (Hialeah, FL). The anti‐MHC Mab W6.32 reacts with human and cultured rodent cells. The anti‐FLAG M2 Mab and anti‐HA peptide tag Mab 12CA5 were purchased from Eastman Kodak Company (New Haven, CT) and Boehringer Mannheim (Indianapolis, IN), respectively. The anti‐Shc Mab and the recombinant horseradish peroxidase (HRP)‐conjugated anti‐P‐Tyr Mab RC20 were from Transduction Laboratories (Lexington, KY). Anti‐Erk2 and anti‐Grb2 polyclonal antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Human fibronectin and collagen I were purchased from Gibco‐BRL. Laminin 5 matrices were prepared as described previously (Sonnenberg et al., 1993; Spinardi et al., 1995).
Cell lines, constructs and transfections
HeLa cells were cultured in Dulbecco's modified minimal essential medium (DMEM) with 10% fetal calf serum (FCS). Primary human keratinocytes were cultured in keratinocyte serum‐free medium (K‐SFM) supplemented with bovine pituitary extract (50 μg/ml) and human recombinant EGF (5 ng/ml) (Gibco BRL). NIH 3T3 cells expressing recombinant human β4 were maintained in DMEM containing 10% bovine calf serum (BCS) and 3 mM l‐histidinol (Sigma, St Louis, MO). Primary dermal fibroblasts were obtained from Clonetics (San Diego, CA) and cultured in fibroblast basal medium (FBM) supplemented with 2% FCS, 2 ng/ml bFGF and 10 μg/ml insulin.
To generate pLXSHD‐β4, the full‐length human β4 cDNA was subcloned in the EcoRI site of pLXSHD, a derivative of the Moloney leukemia virus‐derived retroviral vector pLXSN containing as a selection marker the histidinol dehydrogenase gene (Dusty Miller and Rosman, 1989). The recombinant virus was produced by transiently transfecting 293‐T cells with 10 μg of pLXHD‐β4 and 10 μg of packaging‐defective ecotropic virus (Landau and Littman, 1992). Subconfluent NIH 3T3 cells were infected with a dilution of the culture supernatant collected 48 h after transfection. Cell lines expressing recombinant human β4 were isolated by l‐histidinol selection and identified by FACS analysis. Rat 804G cells expressing a recombinant wild‐type human β4 subunit were previously described (Spinardi et al., 1993). To generate a cytomegalovirus (CMV)‐driven eukaryotic expression vector encoding a tail‐less β4 subunit, the 2.3 kb EcoRI–ScaI fragment of β4 cDNA was ligated into EcoRI–BamHI‐digested pRK‐5. The 2.3 kb EcoRI–XbaI fragment of the resulting plasmid was then ligated into EcoRI–XbaI‐linearized pRc‐CMV. The recombinant β4 subunit encoded by this vector is truncated immediately after Lys734, which marks the boundary between the transmembrane and intracellular domains of the polypeptide. Clone L cells were generated by transfecting parental 804G cells with the above vector according to previously published protocols (Spinardi et al., 1993). FACS analysis was used to verify that clone A and clone L cells had comparable levels of expression of recombinant β4. Metabolic labeling with [35S]methionine/cysteine (Translabel, ICN, Costa Mesa, CA) and immunoprecipitation were used to verify the correct assembly of recombinant β4 with endogenous α6 in both NIH 3T3 and 804G transfectants.
The CMV promoter‐based expression vectors encoding HA‐tagged Erk2, dominant‐negative p52Shc (Y317F) and dominant‐negative Ras (N17) were obtained from Edward Scolnik (NYU School of Medicine). The vectors encoding GST–Jun and Flag‐tagged Jnk1 were described previously (Hibi et al., 1993; Derijard et al., 1994). The cDNAs encoding dominant‐negative RhoA (N19) and Rac1 (N17) (Khosravi‐Far et al., 1995) were subcloned in pcDNA3. Dominant‐negative Cdc42 (N17) in pCMV5 was obtained from Jonathan Chernoff (Fox Chase Cancer Center, Philadelphia, PA). The Fos‐SRE–Luc reporter plasmid was from Joseph Schlessinger. Vectors were transiently transfected in HeLa and NIH 3T3 cells by the lipofectamine method (Gibco‐BRL).
To obtain ligation of integrins in the absence of any co‐stimulus, the cells were growth factor starved for 36 h, detached, and resuspended in serum‐free medium. The cells were then either incubated in suspension with polystyrene beads (2.5 μm diameter, IDC, Portland, OR) coated with anti‐integrin Mabs (Mainiero et al., 1995; Wary et al., 1996) or plated onto dishes coated sequentially with affinity‐purified goat anti‐mouse IgGs and anti‐integrin Mabs or extracellular matrix proteins. At the coating concentrations used, the cells attached and spread equally well on laminin 5 and collagen I, attached and partially spread on the anti‐β4 Mab 3E1, and attached without spreading on poly‐l‐lysine and the control Mab W6.32. As a positive control for Jnk activation, adherent cells were exposed to UV radiation as previously described (Hibi et al., 1993; Derijard et al., 1994). At the end of the incubation, the cells were extracted and subjected to biochemical analysis.
To immunoprecipitate Shc and α6β4, primary human keratinocytes were extracted in Triton lysis buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1% Triton X‐100) containing 1 mM sodium orthovanadate, 2.5 mM sodium pyrophosphate, 25 mM sodium fluoride, 0.01% aprotinin, 4 mg/ml pepstatin A, 10 mg/ml leupeptin and 1 mM phenylmethanesulfonyl fluoride (PMSF) (all from Sigma) for 30 min on ice. Immunoprecipitation, SDS–PAGE and immunoblotting analysis were performed as previously described (Giancotti and Ruoslahti, 1990; Mainiero et al., 1995). Nitrocellulose‐bound antibodies were detected by chemiluminescence with ECL (Amersham Life Sciences, Little Chalfont, UK).
To examine Erk activity, cells were extracted with NP‐40 lysis buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1% NP‐40, 1 mM EDTA) containing phosphatase and protease inhibitors for 30 min on ice. Endogenous and recombinant tagged Erks were immunoprecipitated with anti‐Erk2 or anti‐HA Mab, respectively, and subjected to in vitro kinase assay. The kinase reaction was initiated by adding to the beads 25 μl of kinase buffer (25 mM Tris pH 7.5, 12.5 mM β‐glycerophosphate, 7.5 mM MgCl2, 20 μM cold ATP, 0.5 mM sodium orthovanadate) containing 5 μCi of [γ‐32P]ATP (4500 Ci/mmol, ICN Biomedicals Inc.) and 2.5 μg of myelin basic protein (Sigma). After 30 min of incubation at 30°C, the samples were boiled in sample buffer and separated by SDS–PAGE.
To analyze the activation of Jnk, cells were extracted for 30 min on ice with modified Triton lysis buffer (25 mM HEPES pH 7.5, 300 mM NaCl, 0.1% Triton X‐100, 0.2 mM EDTA, 20 mM β‐glycerophosphate, 1.5 mM MgCl2, 0.5 mM dithiothreitol) containing phosphatase and protease inhibitors. Endogenous Jnk was precipitated with 3 μg of GST–Jun fusion protein coupled to glutathione–agarose beads. After washing, the beads were incubated with 25 μl of kinase buffer containing 5 μCi of [γ‐32P]ATP. Recombinant Flag‐tagged Jnk1 was immunoprecipitated with the anti‐Flag Mab M2 and incubated with 25 μl of kinase buffer containing 2.5 μg of GST–Jun and 5 μCi of [γ‐32P]ATP. After 30 min of incubation at 30°C, the samples were boiled in sample buffer and separated by SDS–PAGE.
To estimate Ras activation, primary human keratinocytes were starved for 48 h in K‐SFM and labeled for 12 h with [32P]orthophosphate (0.5 mCi/ml, ICN) in phosphate‐free DMEM supplemented with 0.1% phosphate‐free BCS. After stimulation, the cells were extracted and the samples subjected to Ras‐GTP loading assay as described previously (Gale et al., 1993). Nucleotides bound to Ras were analyzed by TLC on PEI‐cellulose plates in 0.75 M K2HPO4, pH 3.5. Radioactivity in GDP and GTP was estimated by Phosphorimager analysis.
To measure trascription from the Fos SRE, HeLa cells and NIH 3T3 transfectants were transiently transfected with the reporter plasmid Fos‐SRE–Luc. After 24 h of growth factor starvation, the cells were detached and plated on dishes coated with laminin 5, collagen I or poly‐l‐lysine for 30 min. The HeLa cells were then either left untreated or exposed to 50 ng/ml EGF, 20 ng/ml PDGF‐BB, 2 ng/ml bFGF and 10 μg/ml insulin for 10 min. The NIH 3T3 transfectants were either left untreated or exposed to 10 ng/ml PDGF‐BB, 5 ng/ml bFGF and 10 μg/ml insulin for 10 min. Luciferase activity in cell lysates was estimated as previously described (Brasier et al., 1989).
Measurement of cell cycle progression
To monitor progression through G1 and entry into S phase, the cells were starved by incubation in medium devoid of serum and growth factors for 48 h, detached and plated at low density on microtiter plates or glass coverslips coated with 10 μg/ml poly‐l‐lysine, laminin 5, collagen I or a mixture of 10 μg/ml fibronectin and 10 μg/ml poly‐l‐lysine. The keratinocytes were incubated in K‐SFM supplemented with 5 ng/ml human recombinant EGF. The NIH 3T3 transfectants were incubated in FBM supplemented with 20 ng/ml PDGF, 2 ng/ml bFGF and 10 μg/ml insulin. Primary dermal fibroblasts were incubated in FBM with 2 ng/ml bFGF and 10 μg/ml insulin. The media were supplemented with 10 μM BrdU and, when indicated, with the inhibitory anti‐β1 Mab 4B4, the inhibitory anti‐β4 Mab A9 or the control anti‐MHC Mab W6.32. After 22 h of incubation, the cells were fixed in 70% ethanol, 50 mM glycine, pH 2.0 for 30 min at −20°C and stained with anti‐BrdU Mab and alkaline phosphatase‐conjugated secondary antibodies (Boehringer Mannheim, Indianapolis, IN). The percentage of labeled nuclei was determined by scoring at least 500 cells from five different microscopic fields.
We thank Dafna Bar‐Sagi, Tom Carey, Jonathan Chernoff, Maria Galisteo, Ed Scolnik and Jossie Schlessinger for constructs and antibodies, Mitchell Yeon for expert technical assistance, and Giuseppe Pintucci for Phosphorimager analysis. We are especially grateful to Michael Dans for generating clone L cells. This work was supported by NIH grant CA 58976 and DAMD grant 17‐94‐J4306. F.M. was supported by a fellowship from the American Italian Foundation for Cancer Research. C.M. is on leave of absence from the Istituto Nazionale della Nutrizione (Rome, Italy). F.G.G. was the recipient of awards from the Lucille P.Markey and the Irma T.Hirschl Charitable Trusts and is an Established Investigator of the American Heart Association.
- Copyright © 1997 European Molecular Biology Organization