¹Ludwig Institute for Cancer Research, 91 Riding House Street, London W1W 7BS, UK

²University College London Breast Cancer Laboratory, Department of Surgery, Royal Free and University College Medical School, Charles Bell House, 67-73 Riding House Street, London W1W 7EJ, UK

Correspondence to: J F Timms, E-mail: jtimms@ludwig.ucl.ac.uk

Most breast cancers arise from luminal epithelial cells and 25-30% of these tumours overexpress the ErbB-2 receptor. Herein, a non-transformed, immortalized cell system was used to investigate the effects of ErbB-2 overexpression in luminal epithelial cells. The phenotypic consequence of ErbB-2 overexpression is a shortening of the G1 phase of the cell cycle and early S phase entry, which leads to hyperproliferation. We show that this effect was mediated through the up-regulation of cdk6 and cyclins D1 and E, and enhanced degradation and relocalization of p27^Kip1. These changes were effected predominantly through enhanced MAPK signalling, resulting in cdk2 hyperactivation. PI3K signalling also participated in cell cycle progression, since PI3K and MAPK coordinately regulated changes in cyclin D1 and cdk6 expression. Cdk4 activity was not required for cell cycle progression in these cells, and was constitutively inhibited through its association with p16^INK4A. MAPK-dependent induction of p21^Cip1 was also necessary for G1 phase progression, although its degradation by the proteasome was required for S phase entry. These data provide new insights into the complex molecular mechanisms underlying mitogenic cell cycle control in luminal epithelial cells, the cell type relevant to primary breast cancer, and show how ErbB-2 overexpression subverts this normal control.

Oncogene (2002) 21, 6573-6586. doi:10.1038/sj.onc.1205847

ErbB-2; cell cycle; cdk inhibitors; cdk2; luminal epithelial

Introduction

The overexpression of ErbB family members and their ligands has been implicated in the progression of many different types of cancer (reviewed in Yarden and Sliwkowski, 2001). Notably gene amplification and overexpression of ErbB-2 occurs in 25-30% of invasive ductal carcinomas of the breast. These cancers arise from the luminal epithelial cell-type, and are associated with poor prognosis (Harari and Yarden, 2000; Slamon et al., 1987). ErbB-2 overexpression is known to enhance tumour cell proliferation, metastasis and chemoresistance, although the exact mechanisms by which these effects are mediated are not yet fully understood.

It has been proposed that ErbB-2 overexpression results in the constitutive activation of ErbB receptor heterodimers, and indeed, ErbB-2 is the preferred heterodimerization partner for the other ErbB family members (Graus-Porta et al., 1997; Karunagaran et al., 1996; Tzahar et al., 1996). Thus, although ErbB-2 has no known ligands, all EGF family growth factors can induce its activation via heterodimeric receptor interactions. The consequence of increased receptor activation is the increased recruitment of signalling intermediates involved in such cellular processes as proliferation, migration, and survival. However, it has also been proposed that increased activation of ErbB-2-containing heterodimers may also allow cells to evade normal receptor inactivation processes, thereby prolonging signal duration and promoting growth factor independence and survival (Harari and Yarden, 2000).

The major downstream pathways triggered by activated ErbB receptors are the mitogen-activated protein kinase (MAPK), stress-activated protein kinase (SAPK), phosphatidylinositide 3'-kinase (PI3K), p70S6 kinase (p70S6K) and protein kinase C (PKC) pathways (Yarden and Sliwkowski, 2001). The activation of the MAPK and SAPK pathways have been shown to be enhanced and prolonged in ErbB-2 overexpressing cells, suggesting that they may confer a major transforming signal (Ben-Levy et al., 1994; Karunagaran et al., 1996). This appears to be mediated by increased recruitment of the adaptors Grb2 and Shc (upstream activators of Ras/MAPK signalling) to activated ErbB-2 (Dankort et al., 1997, 2001). In addition, the PI3K pathway has also been implicated in ErbB-2-mediated evasion of apoptosis (Zhou et al., 2000), and in ErbB-2-induced invasion of mammary epithelial cells (Ignatoski et al., 2000).

One of the major cellular processes affected by ErbB-2 signalling is the cell cycle. Progression through G1-phase and entry into S phase is regulated by the activation of G1 phase cyclin-dependent kinases, namely cdk2, cdk4 and cdk6. These phosphorylate and inhibit the retinoblastoma gene product (pRb), releasing E2F transcription factors for induction of genes required for S phase progression and DNA synthesis (Sherr and Roberts, 1999). Cdk activation is regulated by specific phosphorylation and dephosphorylation events, and by binding to cyclins, D-type cyclins in the case of cdk4 and cdk6, and cyclin E in the case of cdk2. It is the mitogen-dependent expression of these cyclins that is critical for cdk activitiy and cell cycle progression. Two families of cdk inhibitor, the INK4 and Kip/Cip proteins, also influence G1/S phase progression (reviewed in Sherr and Roberts, 1999). INK4 proteins act by preventing cyclin D-cdk4/6 association, and are involved in such cellular processes as senescence, development and antigrowth factor signalling, whilst Kip/Cip proteins inhibit pre-existing cyclin-cdk complexes. Despite their apparent inhibitory roles, recent evidence also suggests that p21^Cip1/WAF1 and p27^Kip1 (p21 and p27 hereafter) are actually required for cyclin D-cdk complex assembly and activation (Cheng et al., 1999; Labaer et al., 1997; Sherr and Roberts, 1999; Weiss et al., 2000). It is likely therefore, that specific threshold levels and/or the sub-cellular localization of these molecules play a critical role in determining their inhibitory capacities.

Cyclin D1 overexpression has been reported in many breast tumours and appears to be a major downstream target of ErbB-2-mediated transformation (Bartkova et al., 1994; Fredersdorf et al., 1997; Weinstat-Saslow et al., 1995). Indeed, transgenic mice overexpressing cyclin D1 in mammary gland exhibit proliferative abnormalities and an increased incidence of breast tumours (Wang et al., 1994). More importantly, cyclin D1 was recently shown to be necessary for both activated ErbB-2 and Ras-induced malignant transformation of murine mammary epithelial cells (Yu et al., 2001). These findings reinforce the link between ErbB-2 overexpression, enhanced Ras/MAPK signalling and cyclin D up-regulation in promoting breast epithelial cell transformation. Cyclin D1 expression is increased by transcriptional up-regulation through the MAPK and/or PI3K/Akt pathways (Gille and Downward, 1999; Lenferink et al., 2001; Takuwa et al., 1999), although the exact contribution each pathway makes is unclear. Cyclin D1 is also regulated at the post-transcriptional level via several PI3K/Akt-dependent mechanisms (Diehl et al., 1998; Dufourny et al., 2000; Muise-Helmericks et al., 1998). While there is little evidence to suggest that cyclin E expression is directly regulated by MAPK or PI3K signalling, the initial formation of cyclin D-cdk4/6 complexes may drive formation of activated cyclin E-cdk2 complexes through the sequestration of cdk inhibitors (Musgrove et al., 1996; Perez-Roger et al., 1999; Sherr and Roberts, 1999).

ErbB-2 overexpression has been correlated with down-regulation of p27 with concomitant enhancement of G1 phase cyclin-cdk activity (Catzavelos et al., 1997; Fredersdorf et al., 1997; Porter et al., 1997). This correlation has been attributed to enhanced degradation or redistribution/sequestration of p27 (Lane et al., 2000; Lenferink et al., 2001; Yang et al., 2000). The signalling pathways responsible for increased p27 degradation in tumour cells are unclear, but up-regulation of both MAPK and PI3K signalling has been implicated (Busse et al., 2000; Hoshino et al., 2001; Lenferink et al., 2001; Yang et al., 2000). These pathways may act via cdk2, which phosphorylates p27, directing it for ubiquitin-mediated degradation (Sheaff et al., 1997; Vlach et al., 1997). Recent evidence also suggests that p27 is transcriptionally regulated via PI3K/Akt-mediated inhibition of forkhead transcription factors (Dijkers et al., 2000; Medema et al., 2000). The role of p21 in cellular transformation is less clear. It can be induced by both p53-dependent and independent mechanisms, and has been shown to mediate cell cycle arrest in response to stimuli that cause DNA damage, differentiation, senescence or apoptosis (Deng et al., 1995; Macleod et al., 1995; Stein et al., 1999). Moreover, the upregulation of p21 has been linked to chemoresistance in ErbB-2 overexpressing cells (Yu et al., 1998). p21 is also induced in some cell types in response to mitogens or matrix attachment through either MAPK (Bottazzi et al., 1999) or PI3K-dependent mechanisms (Zhou et al., 2001). The functional significance of this induction is unclear, but it may be required as a positive regulator of cell cycle progression by promoting the assembly of cyclin D-cdk complexes (see above).

In this study, we investigated the cellular effects of ErbB-2 overexpression on proliferation and cell cycle regulation in a model cell system of human mammary luminal epithelial cells.

Results

ErbB-2 overexpression in breast luminal epithelial cells results in enhanced proliferation and cell cycle progression

In order to better understand the effects of ErbB-2 overexpression on the growth and cell cycle of mammary luminal epithelial cells, we undertook a biological and biochemical comparison of 'normal' parental HB4a luminal epithelial cells and the ErbB-2 overexpressing C3.6 clone. Firstly, we confirmed that ErbB-2 overexpression results in increased proliferation (Figure 1a) (Harris et al., 1999). C3.6 hyperproliferation was more noticeable when cells were seeded at higher densities, suggesting that ErbB-2 overexpression may also suppress contact growth inhibition in cultured cells. The C3.6 cells also showed enhanced proliferation when treated with low doses of EGF and HRG1 (EGFR- and ErbB-3-specific ligands, respectively) (Figure 1b). This demonstrates that ErbB-2 overexpression enhances proliferative signalling in response to signalling through distinct ErbB receptors. The observed proliferative response was equivalent for the two growth factors, although lower than that induced by serum.

In order to characterize the effects of ErbB-2 overexpression on the luminal epithelial cell cycle, we first carried out an analysis of DNA content, comparing exponentially growing (10% FCS), serum-starved, and mitogen-treated (HRG1, EGF or 10% FCS) cells. Results in Figure 1c show that under virtually all conditions, there were a higher proportion of C3.6 cells in S- and G2/M-phase. As expected, serum-starvation increased the number of cells in G0/G1, with a concomitant decrease of cells in G2/M. However, a larger proportion of the C3.6 cells could not be rendered quiescent compared to the HB4a cells. These data suggest that constitutive receptor signalling can drive the C3.6 cells through the cell cycle even in the absence of mitogens. After 24 h of stimulation with EGF, HRG1 or 10% FCS, a higher proportion of the C3.6 cells had entered S phase compared with the HB4a cells (Figure 1c). The responses to EGF and HRG1 were similar, and less potent than serum. Notably, a higher proportion of the C3.6 cells had a DNA content of greater than 4n (accounting for the remaining percentage of gated cells in the analysis). This result may indicate that additional rounds of DNA replication without cellular division can occur in response to ErbB-2 overexpression. A very low and equivalent sub-G1 DNA content in both cell lines indicated minimal apoptosis under these conditions (data not shown). Taken together, these data show that ErbB2 overexpression in breast luminal epithelial cells results in early S phase entry following mitogen treatment, and that significant cell cycling can occur in the absence of mitogens.

ErbB-2 overexpression enhances MAPK signalling, but not PI3K signalling

We next examined the effect of ErbB-2 overexpression on the activation of key intracellular signalling pathways triggered in response to EGF compared to HRG1 stimulation. These two ligands were used to test the effect of signalling from different activated heterodimer receptor complexes. ERK1/2 activation and expression of c-myc were both elevated in C3.6 cells, and were higher following EGF treatment compared with HRG1 (Figure 2a). We also observed enhanced activation of p90RSK1, a downstream target of MAPK signalling (data not shown). Conversely, Akt activation was found to be lower in EGF-treated C3.6 cells compared with the HB4a, although it was equivalent in the two cell types treated with HRG1 (Figure 2a). Taken together, these data show that ErbB-2 overexpression enhances signalling through the MAPK, but not PI3K pathway.

Since the nature and extent of downstream signalling in response to EGF-like ligands is likely to be determined by the levels of all four ErbB receptor family members, we assessed there relative expression levels in the two cell lines under different growth conditions (Figure 2b). EGFR levels were approximately fivefold lower in the C3.6 cell line, and decreased in both cell lines in response to EGF, but not HRG1 treatment. Conversely, ErbB-3 expression was slightly higher in the C3.6 cells, and both HRG1 and EGF treatment resulted in ErbB-3 down regulation, although the response was more rapid with HRG1 (compare 4 and 8 h treatments in Figure 2b). We were unable to detect ErbB-4 expression in these cell lines. These data not only suggest that both EGFR and ErbB-3 are down regulated by ligand-induced internalization/degradation, but also that ErbB-2 can modulate the expression of the other ErbB family members. We are currently investigating this phenomenon.

Regulation of cell cycle proteins in response to ErbB-2 overexpression

To further characterize the molecular determinants of the shortened G1 phase in the ErbB-2 overexpressing cells, we analysed the expression levels of key cell cycle regulators. Expression of the cdk inhibitor p27 was lower in the C3.6 cells, and was reduced even further in response to EGF, HRG1 or serum treatment (Figure 3). This effect was completely blocked by co-treatment of cells with a potent proteasome inhibitor, PS-341 (Adams et al., 1999); (Figure 3b). This indicates that p27 is targeted to the proteasome in response to mitogen treatment, a process that is augmented by ErbB-2 overexpression. Surprisingly, the related inhibitor p21 was transiently induced; reaching maximal levels 3-5 h post-stimulation, although levels had returned to basal prior to S phase entry (18-24 h). This reduction was proteasome-dependent, as shown by co-treatment with PS-341 (Figure 3b). ErbB-2 overexpression had little effect on p21 induction, although higher levels were retained in the HB4a cells following its degradation (Figure 3c).

Cyclin D1 was also induced in response to EGF, HRG1 and serum treatment (Figure 3), with levels returning to basal by 15 h (prior to S phase entry), and in a proteasome-dependent manner. Notably, the induction was more rapid, and its expression higher in C3.6 cells following growth factor treatment. Cyclin D1 was also detectable at low levels in serum-starved C3.6 cells, which could be indicative of growth factor-independent signalling. Cyclin D3 is also expressed in these cells, but its expression level was unaffected by ErbB-2 or mitogen treatment (data not shown). We were unable to detect cyclin D2 in these cells. Cyclin E and cdk6 expression levels were also higher in the C3.6 cells and were shown to increase upon mitogen treatment, reaching maximal levels around S phase entry (see Figure 3). Cdk2 and cdk4 expression was unaffected by growth factor treatment or ErbB-2 overexpression.

A comparison of the effects of EGF and HRG1 treatment showed that EGF is more potent than HRG1 in inducing p21, cyclin D1 and cdk6 expression (Figure 3c), and this likely reflects the increased potency of EGF in activating MAPK signalling (Figure 2a). The kinetics of mitogen-induced p27 degradation and cyclin E expression were similar for the two ligands, although re-expression of p27 was suppressed for longer in HRG1-treated C3.6 cells. In addition, cyclin D1 expression was sustained in HRG1-treated C3.6 cells up to at least 34 h, albeit at low levels. Taken together, these data suggest that EGF induces a more potent activation of cell cycle signalling, whereas HRG1 induces a more sustained signalling. Importantly, both of these effects are enhanced by ErbB-2 overexpression.

We also examined the levels of cell cycle regulators and activation of MAPK and PI3K pathways in another ErbB-2 overexpressing clone, C5.2 (Figure 3d). This analysis was included to rule out the possibility that clonal variation may have led to the observed alterations in cell signalling components. Clone C5.2 cells have an even higher level of ErbB-2 expression than C3.6, and a concomitantly increased proliferative rate (data not shown and Harris et al., 1999). This analysis of exponentially growing cells showed that ERK1/2 activation, c-myc, cyclin D1, cyclin E and cdk6 expression all correlated with ErbB-2 overexpression, whilst p27 levels were reduced in the ErbB-2 overexpressing clones. Interestingly, we also observed increased Akt activation in the exponentially growing C5.2 cells (Figure 3d). These data suggest that very high levels of ErbB-2 expression may lead to constitutive PI3K and Akt activation.

Inhibition of intracellular signalling pathways

To further delineate the pathways involved in aberrant ErbB-2-dependent cell cycle regulation, we treated cells with various inhibitors prior to mitogenic stimulation. Treatment with LY294002 (a PI3K inhibitor) only very slightly abrogated mitogen-dependent p27 degradation (Figures 3b and 4). Its expression (over 3 h of treatment) was also unaffected by PD098059 (a MAPK pathway inhibitor) or rapamycin (an inhibitor of p70S6K-mediated translational regulation) (Figure 4b). EGF-induction and subsequent degradation of p21 were also unaffected by LY294002 or rapamycin, but its induction was partially inhibited by PD098059 treatment. LY294002 and PD098059 were both found to abrogate EGF-induced cyclin D1 expression, and simultaneous treatment with both inhibitors reduced levels to those seen in unstimulated cells. Rapamycin had no effect on cyclin D1 levels. Notably, LY294002 also significantly inhibited cdk6 expression, pointing to the existence of a novel PI3K-dependent signalling pathway (Figure 3b). These data confirm that both PI3K- and MAPK-dependent signalling co-ordinately regulate the expression of key cell cycle proteins downstream of ErbB receptors.

ErbB-2 overexpression results in cdk2 and cdk6 hyperactivation

We have shown that ErbB-2 overexpression alters the expression of G1/S phase regulators in response to mitogens, but the functional consequences of this are unclear. We therefore examined the catalytic activities of various cyclin-cdk complexes by in vitro kinase assay. Using histone H1 as a substrate, we observed elevated cdk2 activity in serum-starved C3.6 cells, suggestive of constitutive signalling (Figure 5a). Under such conditions, when p21 levels are low, the observed difference in cdk2 activity seemed to correlate more with p27-cdk2 association, although it was difficult to visualize p27 in cdk2 immunoprecipitates due to cross-reactivity with the co-migrating IgG light chain (Figure 5a).

Cdk2 activity was increased at the onset of S phase (23 h) in response to EGF in both cell lines, but was fivefold higher in the ErbB-2 overexpressors (Figure 5a). Cyclin A- and cyclin E-associated cdk2 activities were also elevated in these cells (data not shown). Surprisingly, cdk2 activity was reduced at intermediate time points, correlating with maximal p21 induction and p21-cdk2 complex formation (Figure 5a,b). Indeed, by comparing p21 levels in cdk2 and p21 immunoprecipitates under conditions of immunodepletion, we showed that 30-50% of total cellular p21 became associated with cdk2 (Figure 5a, lower panel). Thus, induction of p21 mediates a mid G1 phase inhibition of cdk2 activity by direct association.

Cdk6-associated kinase activity (measured using GST-pRb as a substrate) did not change substantially with EGF treatment, although its activity was elevated in the C3.6 cells (Figure 5b). This reflects the increased expression of both cdk6 and its binding partner, cyclin D1, in these cells. Notably, cdk6 activity was not inhibited during p21 induction, despite a detectable association of p21 and cyclin D1 during this period (Figure 5c). Cdk4 kinase activity was barely detectable in either cell type under any of the conditions tested, despite its expression (Figure 3b) and our ability to immunoprecipitate the protein (data not shown). Interestingly, the extent of cyclin E-cdk2 association was reduced at 23 h stimulation, when cdk2 activity was maximal. This shows that the extent of cyclin E-cdk2 association does not necessarily reflect the kinase activity, but also suggests that p21 may stabilize an inactive cyclin E-cdk2 complex.

To further investigate the mechanisms of cdk2 activation, we again employed various inhibitors. The MEK inhibitor PD184352 had little effect on cdk2 activity (Figure 5c). However, a dose of 50 nM proved sub-optimal for MAPK inhibition, and repeat experiments with higher concentrations and other MEK inhibitors were shown to substantially reduce S phase cdk2 activity, and in a dose-dependent manner (Figure 5d). This effect may be mediated through p27, since serum-induced p27 degradation was also abrogated by inhibition of MEK in a dose-dependent manner. This finding demonstrates that MAPK signalling plays a prominent role in cdk2 regulation, possibly through its effects on p27 degradation. LY294002 also reduced cdk2 activity independently of p21-cdk2 association (Figure 5c,d). However, this was concomitant with a reduced cdk6 expression, with little effect on p27 (Figure 5d). PS-341 inhibition produced the largest reduction of cdk2 activity, reducing it to below pre-stimulation levels (Figure 5c). This inhibition was mediated via increased p21-cdk2 association (a consequence of blocking p21 degradation), and resulted in a potent G1 phase arrest (data not shown). Together, these inhibitor studies show that both MAPK- and PI3K-dependent signalling, acting through p27 and cdk6, respectively, and together through cyclin D1, are necessary for full cdk2 activity, but also that a proteasome-dependent degradation of p21 is also required for S phase entry.

p27 determines cdk2 activity in serum-starved and exponentially growing luminal epithelial cells

To further assess the role of p27 in regulating cell cycle progression in luminal epithelial cells, we compared its association with specific cyclins and kinases, and measured kinase activities in serum-starved and exponentially growing cells; conditions under which p21 expression is relatively low (Figure 6a). Importantly, much less cdk2 was associated with p27 in C3.6 cells under both growth conditions. This association correlated with the relative expression of p27 in the two cell types, and was inversely correlated with cdk2 activity (Figure 6a,b). Thus, p27 is a major regulator of cdk2 activity under these conditions.

There was little detectable association between p27 and either cyclin D1, cdk4 and cdk6 in either cell type, despite near immunodepletion of p27 from lysates (Figure 6a,c). Moreover, we detected a significant p27-associated kinase activity towards GST-Rb (Figure 6b). Thus, p27 cannot account for reduced cdk4 and cdk6 activity in HB4a cells. We also observed appreciable p21-associated kinase activity towards GST-Rb (but not H1) in exponentially growing cells (Figure 6b and data not shown). Thus, p21 does not potently inhibit cdk4 and cdk6 in these cells either. Extrapolating from this, our data supports the notion that ErbB-2-enhanced cdk6 activity is a consequence of increased cdk6 expression (and cyclin D1), and not the altered interaction with p21 or p27. The relative inactivity of cdk4 in both cell lines appears attributable to binding and inhibition by the cdk-inhibitor p16^INK4A, which did not complex to cdk6 (Figure 6c). Cdc2-associated kinase activity was unaffected by ErbB-2 overexpression (Figure 6b).

ErbB-2 overexpression results in nuclear exclusion of p27, but does not affect p21

In order to assess whether cellular localization plays a part in p21 and p27 regulation, and to examine if ErbB-2 overexpression affects localization, we immunostained fixed cells for the endogenous proteins. Anti-p21 staining was predominantly nuclear under all conditions tested, with no detectable differences between the two cell lines (Figure 7a). The number of cells with strong nuclear staining of p21 was dramatically increased after 4 h EGF treatment or when cells were treated with PS-341, conditions under which p21 levels are high (Figure 4a). Thus, p21 appears to act predominantly in the nucleus, and its localization is unaffected by ErbB-2 overexpression.

In contrast, there was a striking difference in p27 localization and staining intensity between the cell lines. Most serum-starved HB4a cells had intense nuclear staining, whereas p27 appeared to be excluded from the nucleus in the C3.6 cells (Figure 7b). Overall staining was lower in the C3.6 cells, presumably because there is less p27 expressed in these cells. Growth in 10% serum reduced nuclear staining in both cell types, although ~10% of HB4a cells still retained nuclear staining. Thus, mitogen exposure not only promotes p27 degradation, but also results in its relocalization from the nucleus, a process that is enhanced by ErbB-2 overexpression.

Discussion

In this study we have used a model human mammary luminal epithelial cell system to investigate the molecular mechanisms underlying ErbB-2-mediated transformation. This system is a relevant model because the most common form of breast cancer, infiltrating ductal carcinoma, arises from the luminal epithelial cell type (Harris et al., 1999). Moreover, most established breast tumour cell lines have multiple genetic aberrations, including ErbB-2 overexpression. By contrast, the present model consists of a non-transformed, immortalized cell line with a strictly luminal phenotype which has been specifically engineered to overexpress ErbB-2, but in other aspects should be identical to the parental strain. This permits a cleaner analysis of the specific effects of enhanced ErbB-2 levels on luminal epithelial cell function and phenotype.

In agreement with previous studies using other cell systems, we show that ErbB-2 overexpression enhances mitogen-dependent proliferation in luminal epithelial cells. This results from a shortening of the G1 phase of the cell cycle, and consequently early S phase entry (Figure 1). Moreover, our data suggests that ErbB-2 overexpression enhances cell cycling and proliferative signalling in the absence of mitogens, indicative of constitutive ErbB receptor activation. Stimulation with EGFR and ErbB-3 specific ligands were found to induce similar proliferative responses and S phase entry, and these effects were enhanced by ErbB-2 overexpression. Hence, ErbB-2 augments signalling from both EGFR and ErbB-3 receptors in these cells, supporting the notion that ErbB-2 is the preferred heterodimerization partner for the other ErbB family members (Graus-Porta et al., 1997; Karunagaran et al., 1996; Tzahar et al., 1996). While we demonstrate that ErbB-2 overexpression results in a proliferative advantage, this alone appears to be insufficient for transformation or immortalization of luminal epithelial cells. This is shown by a potent arrest of both cell types following temperature-inactivation of the large T-Ag in these cells (unpublished data). Moreover, these cells do not form tumours in nude or SCID mice (Harris et al., 1999). Thus, it is likely that additional mutagenic events, such as pRb, p53 or p16 inactivation, are required for breast tumour formation that would be potentiated by ErbB-2-mediated hyperproliferation.

Although we (and others) observed that HRG1 promoted cellular proliferation, some studies have shown that this ligand can cause growth arrest and apoptosis in ErbB-2 overexpressing tumour cell lines (Daly et al., 1997, 1999). The reason for this discrepancy is unclear, but a recent study showed that only human breast cancer cell lines expressing ErbB-4 exhibit HRG-dependent antiproliferative responses, and independently of ErbB-2 expression level (Sartor et al., 2001). ErbB-4 is not detectable in the luminal cell lines used here, perhaps explaining the observed positive responsiveness to HRG1.

Our analysis of down-stream signalling events in response to EGF and HRG1 treatment shows that MAPK signalling, and not PI3K/Akt signalling, plays the crucial role in promoting the ErbB-2-mediated hyperproliferation and cell cycle progression of luminal epithelial cells. It has been reported that ErbB-2 overexpression results in constitutive activation of Akt (Hermanto et al., 2001; Zhou et al., 2000). The reason for this discrepancy is unclear, but is likely to be due to the different cell types used in these other studies. However, we did observe increased Akt activation in exponentially growing C5.2 cells (Figure 3d). Since these cells express even higher levels of ErbB-2, it appears that the level of receptor determines the extent of Akt activity. We have also carried out a comparison of ErbB receptor family expression and Akt phosphorylation in multiple breast tumour cell lines (data not shown). This analysis revealed that Akt activation correlated more with ErbB-3, rather than ErbB-2 expression, in line with ErbB-3's ability to recruit PI3K in response to its activation.

EGF and HRG1 were equipotent in promoting proliferation and S-phase entry, and both mitogens produced enhanced MAPK activation in the ErbB-2 overexpressing cells. Despite this, EGF stimulation tended to produce a lower level Akt activation in the C3.6 cells compared to HB4a, but equivalent activation in response to HRG1 stimulation (Figure 2a). This difference is likely to be due to the differential expression of the three ErbB receptor family members in these cells (Figure 2b). Since only ErbB-3 possesses PI3K recruitment sites, we speculate that EGFR-ErbB-3 heterodimers are formed in response to EGF in the HB4a cells, thus promoting a more robust PI3K/Akt activation. This is compared to EGF-induced activation of EGFR-ErbB-2 heterodimers in the C3.6 cells, which would be less potent for PI3K/Akt activation. HRG1 should also favour EGFR-ErbB-3 heterodimers in the HB4a cells, but ErbB-2-ErbB-3 heterodimers in the C3.6 cells; complexes that appear to be equipotent for PI3K/Akt activation. From this, we propose that enhanced MAPK signalling in the C3.6 cells in response to both ligands, must originate from phosphorylated ErbB-2 receptors directly, irrespective of the dimerization partner. In support of this, a previous study using these same cell lines showed that ErbB-2 overexpression enhances Grb2 and Shc association (Harris et al., 1999), an event required to activate downstream Ras/MAPK signalling.

The connection between the ErbB-2-mediated, enhanced MAPK signalling and cell cycle progression was investigated further by analysing the activity and expression of key G1 phase cell cycle regulators. Our data indicate that C3.6 hyperproliferation results from the sustained down-regulation of p27 acting in concert with up-regulation of cyclin D1, cyclin E and cdk6. Previous studies have implicated cyclin D1 up-regulation and/or p27 down-regulation in the progression of breast cancer (see Introduction), and we show that this holds true in the luminal epithelial cell type from which most breast cancers arise. Our finding that up-regulation of cdk6 also plays a role in ErbB-2-mediated hyperproliferation appears to be novel, and warrants further investigation. The finding that indole-3-carbinol, which selectively abolished cdk6 expression and induced growth inhibition in MCF-7 breast cancer cells, further supports the notion that cdk6 is crucial to breast epithelial cell proliferation (Cover et al., 1998). While ErbB-2-mediated hyperproliferation appears to depend upon MAPK signalling, our inhibitor studies show that PI3K/Akt signalling is also required for cdk2 activity (Figure 5). Thus, in luminal epithelial cells, signalling through both MAPK and PI3K is required for cyclin D1 and cdk6 induction, suggesting that both pathways co-ordinately regulate cell cycle progression.

Importantly, the changes observed in these cell cycle regulators in response to ErbB-2 overexpression resulted in elevated cdk2 activity, which was the most distal downstream response that we identified. Cdk2-associated kinase activity was higher in the C3.6 cells under conditions of serum-starvation, long-term mitogen stimulation, and exponential growth, and was inversely correlated with p27-cdk2 association under these conditions, in agreement with previous reports (Lane et al., 2000; Yang et al., 2000). Thus, p27 expression is a major determinant of ErbB-2-enhanced cdk2 activity and cell cycle progression in luminal epithelial cells. However, in these cells, assembly of cyclin D-cdk complexes is also an absolute requirement for cdk2 activity, since blocking cyclin D1 or cdk6 expression with inhibitors, abrogates cdk2 activation (Figures 4 and 5). It has been suggested that up-regulation of D-type cyclins, cdk6 and cdk4, and their complex formation, contribute to the evasion of antigrowth signals by sequestration of cdk inhibitors away from cyclin E-cdk2 complexes (Sherr and Roberts, 1999), and that ErbB-2 potentiates cyclin E-cdk2 activity through the regulation of p27 sequestration proteins (Lane et al., 2000). However, we detected no significant association between p27 and cyclin D1, cdk6 or cdk4 in these cells (Figure 6), suggesting that expression of p27, and not its sequestration by cyclin D-cdk complexes, determines cdk2 activity and proliferative capacity.

The mechanisms of p27 regulation are not fully understood, but herein, we show that proteasome-mediated degradation plays a critical role in regulating p27. Using specific inhibitors, we revealed that MAPK signalling, and not PI3K signalling, is required for this process (Figures 4 and 5). The exact mechanism by which MAPK signals p27 destruction is unclear, but it may do so by promoting cdk2 activity, resulting in direct p27 phosphorylation at Thr187 and targeted proteasomal degradation (Sheaff et al., 1997; Vlach et al., 1997). However, MAPK may also directly phosphorylate this site (Lenferink et al., 2001). In these cells, another level of p27 regulation exists, since we show that mitogenic stimulation (here meaning growth in serum) results in nuclear exclusion of p27, a process that is enhanced by ErbB-2 overexpression (Figure 7b). As well as partitioning p27 from its nuclear targets, relocalization may potentiate proteasomal degradation within the cytosol. We are currently investigating whether this process is regulated by the p27 phosphorylations we have identified in 2D immunoblotting experiments (data not shown).

A novel finding of this study is that induction of cdk6, and not cdk4, regulates G1-phase progression in this cell type. Cdk6 activity was higher in the C3.6 cells, but the kinase was not associated with p27 or p16 in either cell type. Thus, ErbB-2-enhanced cdk6 kinase activity appears to be a direct consequence of its increased expression. In contrast to cdk6, cdk4 expression was equivalent to both cell types, remained unchanged during mitogenic stimulation, and had very low kinase activity towards GST-pRb. This inactivity appeared to be a direct consequence of binding and inhibition by the cdk inhibitor p16. This demonstrates that cdk4 does not play a significant role in mediating cell cycle progression in mammary luminal epithelial cells, but also shows that p16 preferentially binds to cdk4, and not cdk6, in these cells.

The transcription factor c-myc was also up regulated by ErbB-2 overexpression, consistent with the elevation of MAPK signalling (Fiddes et al., 1998; Sears et al., 1999). C-myc has been shown to promote cyclin E-cdk2 activation by enhancing cyclin E gene transcription, but also by inhibiting binding of p27 to newly formed cdk2 complexes (Perez-Roger et al., 1997). Moreover, c-myc is required for transcriptional up-regulation of cyclin D1 (Amati et al., 1998; Daksis et al., 1994), and positively regulates cyclin D-cdk4 and -cdk6 activities (Mateyak et al., 1999). Thus, c-myc up-regulation, in response to enhanced MAPK signalling, is likely to be a critical mediator of cell cycle progression in C3.6 cells, supporting the notion that it cooperates with ErbB-2 in promoting mammary cancer (Hynes and Lane, 2001).

The cdk inhibitor p21 was induced during G1 phase in response to our mitogen treatment, suggesting that it is required for normal cell cycle progression in these cells. The induction was predominantly MAPK-dependent, and independent of p53-mediated transcription; p53 is inactivated in these cells by the SV40 large T-Ag. Surprisingly, p21 induction resulted in a potent cdk2 inhibition in mid G1 phase, a direct consequence of p21-cdk2 complex formation. However, cdk2 activity was restored prior to S phase entry in response to the proteasomal degradation of p21. Notably, the HB4a cells retained a higher level of p21 at this point, which may contribute to the lower S phase activity of cdk2 in these cells. p21 degradation was critical for S phase entry, since blocking it to promote cdk2-p21 association produced a potent G1 arrest.

Mitogenic p21 induction has been observed in several cell types, and studies suggest it is required for assembly of active cyclin D-cdk4/6 complexes (see Introduction). This is likely to be the case in our cell system, since we detected mitogen-induced cyclin D1-p21 association, without finding inhibition of cdk6 activity. Moreover, our measurement of a significant p21-associated GST-pRb kinase activity, further supports the notion that p21 acts to promote assembly of activated cyclin D1-cdk6 complexes. Although p21 invoked a temporal inhibition of cdk2, it did not appear to play a significant role in ErbB-2-mediated hyperproliferation in these cells. This contradicts a recent report which showed that ErbB-2 can relieve the growth-inhibitory properties of p21 by causing it to relocalize to the cytoplasm in response to phosphorylation by Akt (Zhou et al., 2001). In the luminal epithelial cells used here, endogenous p21 was found to be nuclear, and to remain in the nucleus following mitogen stimulation or proteasomal inhibition (Figure 7a). PI3K inhibition also had no effect on p21 induction or degradation in these cells, and we can only speculate that these conflicting data are attributable to the different cell types used in each study.

In conclusion, our data show that ErbB-2 overexpression gives cells a proliferative advantage, primarily through the elevation of MAPK and c-myc signalling, to promote cdk2 activity and enhanced G1-S phase transition. This study suggests that cdk2 activity is an ideal readout of proliferative capacity in response to oncogenic transformation and a prime candidate for anti-cancer therapies. Indeed, PS-341, which potently blocked cdk2 activity in these cells, is currently in Phase I clinical trials (Adams et al., 2000). Our data reveal that the enhanced cdk2 activation occurs because of up-regulation of cyclin D1, cyclin E and cdk6, and the degradation and relocalization of p27. These changes were dependent to a certain extent upon elevated MAPK signalling. While there are anti-cancer agents currently in use that inhibit MAPK signalling directly, e.g. PD184352 (Sebolt-Leopold et al., 1999), or indirectly, by blocking the ErbB-2 receptor itself, e.g. Herceptin (Mendelsohn and Baselga, 2000), blocking the receptor or MAPK signalling may interfere with other vital cell functions. This study suggests that inhibiting downstream molecules, such as cdk2 and cdk6 directly, may more specifically target the abnormal proliferative capacity of ErbB-2 overexpressing cells, and thus be better weapons in the fight against cancer.

Materials and methods

Cell culture, mitogen stimulation and inhibitor treatment

The parental HB4a cell line was established as described (Stamps et al., 1994). The ErbB-2 overexpressing variants C3.6 and C5.2 were derived from HB4a by stable co-transfection with full-length normal human ErbB-2 cDNA (derived from the established breast cancer cell line BT474) under the control of the MMTV-LTR promoter and SV40 polyadenylation signals (Harris et al., 1999). Cells were maintained in RPMI-1640 with 10% FCS, 2 mM glutamine, 100 IU/ml penicillin, 100 g/ml streptomycin (GIBCO-BRL) and 5 g/ml hydrocortisone and insulin (Sigma, Poole, UK) at 37°C in a 10% CO₂ humidifier incubator.

For stimulations, cells were starved for 36-48 h in media containing 0.1% FCS and no insulin. Cells were stimulated with either 10% FCS (Sigma), or 1 nM EGF or 1 nM HRG1 (both from R&D Systems) for the times indicated. The following inhibitors from Calbiochem were used: rapamycin, an inhibitor of p70S6K-mediated translational regulation; PD098059, PD184352, UO126, inhibitors of MEK and ERK1/2 signalling; and LY294002, an inhibitor of PI3K. The proteasome inhibitor PS-341 (Adams et al., 1999; from Dr Julian Adams, PostScript) and the translation inhibitor, cycloheximide (Sigma) were also used. Inhibitors or DMSO (control) were added 30 min prior to growth factor addition, and were used at the concentrations shown.

Proliferation assays and flow cytometry

Cells were seeded by 10% FCS-containing media in triplicate into 12-well plates at the densities shown, or at 2.5´10⁴ cells per well. At the indicated times, cells were trypsinized and viable cells (Trypan blue-excluding) counted using a haemocytometer. For flow cytometric analysis of DNA content, cells were trypsinized, fixed in 70% methanol at -20°C, and stained for 30 min at room temperature with 40 g/ml propidium iodide (Calbiochem) plus 500 g/ml RNAase (DNase-free; Sigma) in PBS. 1´10⁴ cells from each sample were analysed for DNA content using a Becton Dickinson FACSCaliber instrument. The percentage of gated cells in G0/G1, S and G2/M phases were determined using CellQuest software.

Immunoblotting, immunoprecipitation and in vitro kinase assays

Cells were washed in PBS and scraped into lysis buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 1% NP40, 1 mM EDTA, 2 mM sodium orthovanadate, 100 g/ml AEBSF, 17 g.ml aprotinin, 1 g/ml leupeptin, 1 g/ml pepstatin, 5 M fenvalerate, 5 M BpVphen and 1 M okadaic acid). Lysates were clarified by centrifugation at 14 000 g for 10 min at 4°C and protein concentrations determined using the Coomassie Protein Assay Reagent (Pierce).

Proteins were immunoprecipitated from 200-800 g cell lysate using 2-5 g antibody for 3-16 h at 4°C. Immune complexes were washed three times in lysis buffer prior to further analysis. Cell lyates (30 g) and immunoprecipitates were separated by SDS-PAGE, electrophoretically transferred to Immobilon P membrane (Millipore), and probed with appropriate primary antibodies (Ab) and then HRP-conjugated secondary Ab. Immunoreactive bands were visualized with the enhanced chemiluminescence system (NEN Life Sciences).

Immunoblotting antibodies were: anti-p27 mAb (K25020) and anti-ERK2 mAb (both from Transduction Laboratories). Anti-EGFR pAb (1005), anti-neu/ErbB-2 mAb (9G6), anti-ErbB-3 pAb (C-17), anti-phosphotyrosine (PY-99), anti-p27 pAb (C-19), anti-p21 pAb (M-19), anti-cdc2 pAb (C-19), anti-phospho-ERK1/2 mAb (E-4), anti-Rsk1 pAb (C-21), anti-cyclin B1 pAb (H-433), anti-cyclin E pAb (C-19), anti-cdk2 pAb (M2), anti-cdk4 pAb (H-22), anti-cdk6 pAb (C-21) and anti-pRb pAb (C-15) (all from Santa Cruz). Anti-phospho-Akt (Ser473) pAb, anti-Akt pAb, anti-phospho-SAPK/JNK (Thr183/Tyr185) pAb, anti-phospho-p90RSK (Thr360/Ser364) pAb and anti-phospho-GSK-3/(Ser21/9) pAb (from NEB/Cell Signalling Technology). Anti-GSK3/ mAb (4G-1E) (Upstate Biotechnology) and anti-cyclin D1 mAb (Pharmingen). Anti-c-myc mAb (9E10) was a gift from Dr Julian Downward (Imperial Cancer Research Fund), anti-p53 mAb 421 was a gift from Dr Parmjit Jat (Ludwig Institute for Cancer Research, London, UK). Anti-cyclin D1 pAb, anti-cyclin D2 mAb (DCS 3.1), anti-cyclin D3 mAb (DCS 28) and anti-p16 mAb (DCS 50.2) hybridoma supernatant were generous gifts from Dr Gordon Peters (Imperial Cancer Research Fund, London, UK). Anti-p21 mAb (0001.2U) and anti-p27 mAb (008.AA) hybridoma supernatants were generous gifts from Dr Eric Lam (Imperial College, London, UK).

Antibodies for immunoprecipitation and in vitro kinase assay were as follows: anti-p21 mAb (0001.2U), anti-p27 mAb (008.AA), anti-cdk2 pAb agarose-conjugate (M2), anti-cdk4 pAb (H-22), anti-cdk6 pAb (C-21), anti-cyclin E pAb (C-19), anti-cyclin A pAb (C-19) and anti-cdc2 pAb (C-19). All antibodies were pre-bound to Protein A or Protein G Sepharose prior to immunoprecipitation. Anti-p21 and anti-p27 mAbs were purified and cross-linked to protein G-Sepharose using dimethyl pimelimidate (Harlow and Lane, 1988). For kinase assays, immunoprecipitates were further washed twice in kinase buffer (50 mM HEPES, pH 7.4, 10 mM MgCl₂, 10 mM MnCl₂, 10 mM -glycerophosphate, 1 mM DTT, 0.1 M protein kinase inhibitor (Sigma, P-8140)). Reactions were initiated by addition of kinase buffer containing 20 M cold ATP, 5 Ci [³²P]ATP and either 1 g histone H1 (Boehringer Mannheim) or 1 g recombinant pRb (amino acids 792-928) fused to GST (Meyerson and Harlow, 1994). Reactions were incubated for 30 min at 37°C, terminated with sample buffer, separated by SDS-PAGE, transferred to membranes and phosphoimaged.

Immunofluorescence

Mitogen or inhibitor treated cells grown on cover slips were fixed in 4% formaldehyde, permeabilized with 0.2% Triton X-100 and stained with mAbs against p21 (0.001.2U, used undiluted) or p27 (008.AA, used undiluted) or K25020, used at 1 : 300). Anti-mouse FITC conjugated secondary antibody was used for detection at 1 : 200. Images were collected with a Coolview 12 integrating cooled CCD camera (Photonic Science, UK) mounted over an Axiophot microscope fitted with a ´63 NA 1.4 oil immersion objective (Zeiss, UK).

Acknowledgements

We would like to thank Drs Parmjit Jat and Karin Barnouin for critical reading of the manuscript. We are also grateful to Drs Eric Lam, Gordon Peters, Peter Elliot, Julian Adams and Julian Downward for providing reagents.

Adams J, Palombella VJ, Elliott PJ. (2000). Invest. New Drugs, 18: 109-121. MEDLINE

Adams J, Palombella VJ, Sausville EA, Johnson J, Destree A, Lazarus DD, Maas J, Pien CS, Prakash S, Elliott PJ. (1999). Cancer Res., 59: 2615-2622. MEDLINE

Amati B, Alevizopoulos K, Vlach J. (1998). Front. Biosci., 3: D250-D268. MEDLINE

Bartkova J, Lukas J, Muller H, Lutzhoft D, Strauss M, Bartek J. (1994). Int. J. Cancer, 57: 353-361. MEDLINE

Ben-Levy R, Paterson HF, Marshall CJ, Yarden Y. (1994). EMBO J., 13: 3302-3311. MEDLINE

Bottazzi ME, Zhu X, Bohmer RM, Assoian RK. (1999). J. Cell Biol., 146: 1255-1264.

Busse D, Doughty RS, Ramsey TT, Russell WE, Price JO, Flanagan WM, Shawver LK, Arteaga CL. (2000). J. Biol. Chem., 275: 6987-6995. MEDLINE

Catzavelos C, Bhattacharya N, Ung YC, Wilson JA, Roncari L, Sandhu C, Shaw P, Yeger H, Morava-Protzner I, Kapusta L, Franssen E, Pritchard KI, Slingerland JM. (1997). Nat. Med., 3: 227-230. MEDLINE

Cheng M, Oliver P, Diehl JA, Fero M, Roussel MF, Roberts JM, Sherr CJ. (1999). EMBO J., 18: 1571-1583. Article MEDLINE

Cover CM, Hsieh SJ, Tran SH, Hallden G, Kim GS, Bjeldanes LF, Firestone GL. (1998). J. Biol. Chem., 273: 3838-3847. MEDLINE

Daksis JI, Lu RY, Facchini LM, Marhin WW, Penn LJ. (1994). Oncogene, 9: 3635-3645. MEDLINE

Daly JM, Jannot CB, Beerli RR, Graus-Porta D, Maurer FG, Hynes NE. (1997). Cancer Res., 57: 3804-3811. MEDLINE

Daly JM, Olayioye MA, Wong AM, Neve R, Lane HA, Maurer FG, Hynes NE. (1999). Oncogene, 18: 3400-3451.

Dankort D, Maslikowski B, Warner N, Kanno N, Kim H, Wang Z, Moran MF, Oshima RG, Cardiff RD, Muller WJ. (2001). Mol. Cell Biol., 21: 1540-1551. MEDLINE

Dankort DL, Wang Z, Blackmore V, Moran MF, Muller WJ. (1997). Mol. Cell Biol., 17: 5410-5425. MEDLINE

Deng C, Zhang P, Harper JW, Elledge SJ, Leder P. (1995). Cell, 82: 675-684. MEDLINE

Diehl JA, Cheng M, Roussel MF, Sherr CJ. (1998). Genes Dev., 12: 3499-3511.

Dijkers PF, Medema RH, Pals C, Banerji L, Thomas NS, Lam EW, Burgering BM, Raaijmaker JA, Lammers JW, Koenderman L, Coffer PJ. (2000). Mol. Cell Biol., 20: 9138-9148. Article MEDLINE

Dufourny B, van Teeffelen HA, Hamelers IH, Sussenbach JS, Steenbergh PH. (2000). J. Endocrinol., 166: 329-338. MEDLINE

Fiddes RJ, Janes PW, Sivertsen SP, Sutherland RL, Musgrove EA, Daly RJ. (1998). Oncogene, 16: 2803-2813.

Fredersdorf S, Burns J, Milne AM, Packham G, Fallis L, Gillett CE, Royds JA, Peston D, Hall PA, Hanby AM, Barnes DM, Shousha S, O'Hare MJ, Lu X. (1997). Proc. Natl. Acad. Sci. USA, 94: 6380-6385. MEDLINE

Gille H, Downward J. (1999). J. Biol. Chem., 274: 22033-22040. Article MEDLINE

Graus-Porta D, Beerli RR, Daly JM, Hynes NE. (1997). EMBO J., 16: 1647-1655. Article MEDLINE

Harari D, Yarden Y. (2000). Oncogene, 19: 6102-6114.

Harlow E, Lane DP. (1988). Antibodies: a laboratory manual 1st edn New York: Cold Spring Harbor Laboratory Publications.

Harris RA, Eichholtz TJ, Hiles ID, Page MJ, O'Hare MJ. (1999). Int. J. Cancer, 80: 477-484. Article MEDLINE

Hermanto U, Zong CS, Wang LH. (2001). Oncogene, 20: 7551-7562.

Hoshino R, Tanimura S, Watanabe K, Kataoka T, Kohno M. (2001). J. Biol. Chem., 276: 2686-2692. MEDLINE

Hynes NE, Lane HA. (2001). J. Mam. Gland Biol. Neoplasia, 6: 141-150.

Ignatoski KM, Maehama T, Markwart SM, Dixon JE, Livant DL, Ethier SP. (2000). Br. J. Cancer, 82: 666-674.

Karunagaran D, Tzahar E, Beerli RR, Chen X, Graus-Porta D, Ratzkin BJ, Seger R, Hynes NE, Yarden Y. (1996). EMBO J., 15: 254-264. MEDLINE

LaBaer J, Garrett MD, Stevenson LF, Slingerland JM, Sandhu C, Chou HS, Fattaey A, Harlow E. (1997). Genes Dev., 11: 847-862. MEDLINE

Lane HA, Beuvink I, Motoyama AB, Daly JM, Neve RM, Hynes NE. (2000). Mol. Cell Biol., 20: 3210-3223. MEDLINE

Lenferink AE, Busse D, Flanagan WM, Yakes FM, Arteaga CL. (2001). Cancer Res., 61: 6583-6591. MEDLINE

Macleod KF, Sherry N, Hannon G, Beach D, Tokino T, Kinzler K, Vogelstein B, Jacks T. (1995). Genes Dev., 9: 935-944. MEDLINE

Mateyak MK, Obaya AJ, Sedivy JM. (1999). Mol. Cell Biol., 19: 4672-4683. MEDLINE

Medema RH, Kops GJ, Bos JL, Burgering BM. (2000). Nature, 404: 782-787. Article MEDLINE

Mendelsohn J, Baselga J. (2000). Oncogene, 19: 6550-6565. Article MEDLINE

Meyerson M, Harlow E. (1994). Mol. Cell Biol., 14: 2077-2086. MEDLINE

Muise-Helmericks RC, Grimes HL, Bellacosa A, Malstrom SE, Tsichlis PN, Rosen N. (1998). J. Biol. Chem., 273: 29864-29872.

Musgrove EA, Sarcevic B, Sutherland RL. (1996). J. Cell Biochem., 60: 363-378.

Perez-Roger I, Kim SH, Griffiths B, Sewing A, Lane H. (1999). EMBO J., 18: 5310-5320. Article MEDLINE

Perez-Roger I, Solomon DL, Sewing A, Land H. (1997). Oncogene, 14: 2373-2381.

Porter PL, Malone KE, Heagerty PJ, Alexander GM, Gatti LA, Firpo EJ, Daling JR, Roberts JM. (1997). Nat. Med., 3: 222-225. MEDLINE

Sartor CI, Zhou H, Kozlowska E, Guttridge K, Kawata E, Caskey L, Harrelson J, Hynes N, Ethier S, Calvo B, Earp IIIHS. (2001). Mol. Cell Biol., 21: 4265-4275.

Sears R, Leone G, DeGregori J, Nevins JR. (1999). Mol. Cell, 3: 169-179. MEDLINE

Sebolt-Leopold JS, Dudley DT, Herrera R, Van Becelaere K, Wiland A, Gowan RC, Tecle H, Barrett SD, Bridges A, Przybranowski S, Leopold WR, Saltiel AR. (1999). Nat. Med., 5: 810-816. Article MEDLINE

Sheaff RJ, Groudine M, Gordon M, Roberts JM, Clurman BE. (1997). Genes Dev., 11: 1464-1478. MEDLINE

Sherr CJ, Roberts JM. (1999). Genes Dev., 13: 1501-1512. MEDLINE

Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL. (1987). Science, 235: 177-182. MEDLINE

Stamps AC, Davies SC, Burman J, O'Hare MJ. (1994). Int. J. Cancer, 57: 865-874. MEDLINE

Stein GH, Drullinger LF, Soulard A, Dulic V. (1999). Mol. Cell. Biol., 19: 2109-2117. MEDLINE

Takuwa N, Fukui Y, Takuwa Y. (1999). Mol. Cell Biol., 19: 1346-1358. MEDLINE

Tzahar E, Waterman H, Chen X, Levkowitz G, Karunagaran D, Lavi S, Ratzkin BJ, Yarden Y. (1996). Mol. Cell Biol., 16: 5276-5287. MEDLINE

Vlach J, Hennecke S, Amati B. (1997). EMBO J., 16: 5334-5344. Article MEDLINE

Wang TC, Cardiff RD, Zukerberg L, Lees E, Arnold A, Schmidt EV. (1994). Nature, 369: 669-671. MEDLINE

Weinstat-Saslow D, Merino MJ, Manrow RE, Lawrence JA, Bluth RF, Wittenbel KD, Simpson JF, Page DL, Steeg PS. (1995). Nat. Med., 1: 1257-1260. MEDLINE

Weiss RH, Joo A, Randour C. (2000). J. Biol. Chem., 275: 10285-10290.

Yang HY, Zhou BP, Hung MC, Lee MH. (2000). J. Biol. Chem., 275: 24735-24739. MEDLINE

Yarden Y, Sliwkowski MX. (2001). Nat. Rev. Mol. Cell. Biol., 2: 127-137. Article MEDLINE

Yu D, Jing T, Liu B, Yao J, Tan M, McDonnell TJ, Hung MC. (1998). Mol. Cell, 2: 581-591. MEDLINE

Yu Q, Geng Y, Sicinski P. (2001). Nature, 411: 1017-1021. Article MEDLINE

Zhou BP, Hu MC, Miller SA, Yu Z, Xia W, Lin SY, Hung MC. (2000). J. Biol. Chem., 275: 8027-8031. Article MEDLINE

Zhou BP, Liao Y, Xia W, Spohn B, Lee MH, Hung MC. (2001). Nat. Cell Biol., 3: 245-252. Article MEDLINE

Figure 1 ErbB-2 overexpression enhances proliferation and cell cycle progression of luminal epithelial cells. (a) Effect of cell density on proliferation of HB4a and C3.6 cells. Cells were seeded in 10% FCS-containing media at the densities shown and counted after 4 days. (b) Proliferation of HB4a and C3.6 cells in response to HRG1, EGF and 10% FCS. Cells were seeded, allowed to recover for 24 h, serum-starved for 48 h, then stimulated with 1 nM EGF, 1 nM HRG1 or 10% FCS. Cells were counted 24 h prior to stimulation, at the zero time point, and following stimulation for the indicated times. All values are the mean of triplicate counts. Error bars represent the standard deviation of the mean. (c) ErbB-2 promotes mitogen-stimulated S phase entry. Exponentially growing (10% FCS), serum-starved (for 24 and 48 h), and serum-starved cells (for 48 h) treated with 1 nM HRG1, 1 nM EGF or 10% FCS for 24, 36 and 48 h were analysed for DNA content (see Materials and methods). The percentage of gated cells with G0/G1, S and G2/M phase DNA contents is represented. Similar results were obtained in two independent experiments

Figure 2 Comparison of mitogen-stimulated signal cascade activation and receptor levels in normal and ErbB-2 overexpressing cells. (a) Activation of MAPK and PI3K signalling pathways in response to EGF and HRG1 stimulation (1 nM of each for the indicated times) were analysed by immunoblotting with phosphospecific Abs to ERK1/2 and Akt. The levels of these proteins and c-myc were also compared by immunoblotting. (b) Receptor levels were compared by immunoblotting lysate from exponentially growing (10% FCS), serum-starved (0.1% FCS), EGF-stimulated and HRG1-stimulated cells (both 1 nM) for the indicated times. All blots are representative of at least three independent experiments

Figure 3 Comparison of mitogen-stimulated cell cycle regulator levels in normal and ErbB-2 overexpressing cells. (a) Long-term EGF-stimulated responses. Serum-starved cells were stimulated with EGF (1 nM), lysed at the indicated times and specific proteins detected by immunoblotting. Anti-ERK2 blots were reprobed with an anti-phosphoERK1/2 Ab, however, residual non-phosphoERK2 staining was still visible on these 'reprobes'. (b) Serum-stimulated responses and the effect of PI3K and protesome inhibition. Serum-starved cells were stimulated with 10% FCS, lysed at the indicated times, and proteins detected by immunoblotting. Cells were also co-treated with 10 M LY294002 (LY) or 0.1 M PS-341 (PS) for 23 h, and were added 30 min prior of FCS addition. (c) Effect of ErbB-2 overexpression on cell cycle regulator levels in response to serum withdrawal, EGF and HRG1 treatment. Levels of the indicated proteins were compared by immunoblotting lysates from randomly growing (-48 h/10% FCS), 24 and 48 h serum-starved (-24 h and 0 h/0.1% FCS), and EGF- and HRG1-treated cells (both used at 1 nM for the indicated times). All blots are representative of at least three independent experiments. (d) Expression of cell cycle regulators and signal activation in independent cell clones. Exponentially growing HB4a, C3.6 and C5.2 cells were lysed and the levels of the indicated proteins were compared by immunoblotting

Figure 4 Effect of inhibitors on EGF-regulated cellular responses. (a) Time-course of effects of LY294002 and PS-341 on p27, p21 and cyclin D1 levels in EGF-stimulated HB4a cells. Serum-starved HB4a cells were co-treated with EGF and inhibitor (added 30 min prior to stimulation). Cells were lysed at the indicated times and protein levels compared by immunoblotting. Serum-starved (0.1% FCS) and randomly growing cells (10% FCS) were also analysed as controls. (b) Effect of inhibitors on protein levels 3 h post EGF stimulation. Serum-starved HB4a and C3.6 cells were pre-treated for 30 min with inhibitors: 10 M LY294002 (LY), 50 M PD098059 (PD), 50 nM rapamycin (Rap), a combination of LY294002 and PD098059 (LY/PD), 10 mg/ml cycloheximide (CHX), or DMSO (Ctrl). Cells were stimulated with 1 nM EGF for 3 h, lysed and protein levels determined by immunoblotting. Non-treated (-) cells were run as control. MAPK and PI3K inhibition was monitored with phospho-specific ERK1/2 and Akt Abs. All blots are representative of three independent experiments

Figure 5 ErbB-2 overexpression enhanced cdk2 and cdk6 kinase activity. (a) Enhanced cdk2 activity depends upon p21- and p27-association. Cdk2-associated in vitro histone H1 kinase activities were compared in serum-starved and EGF-stimulated cells. Levels of cdk2-associated p21 and p27 were determined by immunoblotting anti-cdk2 immunoprecipitates (IPs). The lower panel shows p21 levels in anti-p21 and anti-cdk2 IPs under conditions of immunodepletion. (b) ErbB-2 overexpression elevates cdk2 and cdk6 activity, but not cdk4 activity. Cdk2-, cdk6- and cdk4-associated in vitro kinase activities were compared from starved and EGF-stimulated cells using GST-pRb as substrate. p21 and cyclin E levels were also compared by immunoblotting anti-cdk2 IPs and total cell lysates (TCLs) (lower panel). (c) Effects of inhibitors in EGF-induced cdk2 activity. Cdk2-associated histone H1 kinase activity was compared in EGF-stimulated cells co-treated with LY294002, PD184352 and PS-341 for the indicated times. Inhibitors were added 30 min prior to EGF treatment. Cdk2-associated p21 and cyclin E were compared by immunoblotting anti-cdk2 IPs. p21-associated cyclin D1 and cdk2 were compared by blotting anti-p21 IPs (d) Cdk2 activity is dependent upon both MAPK and PI3K signalling. Cdk2-associated histone H1 kinase activities were compared in C3.6 cells co-treated with 10% FCS plus inhibitors (UO=UO126, PD18=PD184352, PD09=PD098059, LY=LY294002) at the indicated doses for 24 h. A serum plus vehicle-treated sample was run as control (Ctrl). Total cell lysates were blotted for cdk6, p27 and activated ERK1/2. All blots and kinase assays are representative of at least three independent experiments

Figure 6 Cdk2 activation is inversely correlated with p27 association, whereas cdk6 activity is determined by expression level. (a) Association of cell cycle regulators with p27. p27 was immunodepleted from serum-starved (0.1% FCS) and exponentially growing (10% FCS) cells and levels of associated proteins examined by immunoblotting. Protein levels were also compared in total cell lysates and p27-depeleted lysates. Around 80% of p27 was depeleted from starved HB4a lysates, which contained the most p27. Lysates were also probed for p16, p21 and cyclinB1 (total cell lysate controls for b and c) and phosphoERK and phosphoAkt to show relative levels of activation under these growth conditions. (b) Associated kinase activities in serum-starved and randomly growing cells. Cdk2-, cdc2-, p21-, cdk4- cdk6- and p27-associated kinase activities were compared in serum-starved (0.1% FCS) and exponentially growing (10% FCS) cells using the substrates shown. (c) Cdk4-, cdk6- and p27-associated proteins. Anti-cdk4, cdk6 and p27 IPs were blotted for associated p27, p21 and p16. Immunodepletion was around 75, 90 and 90% for anti-cdk4, -cdk6 and -p21 IPs respectively (data not shown)

Figure 7 Cellular localization of p21 and p27 in luminal epithelial cells. (a) p21 is predominantly localized to the nucleus. Serum-starved (0.1% FCS), randomly growing (10% FCS), EGF-treated (1 nM for 4 h), and PS-341-treated (100 nM for 24 h) cells were fixed and immunostained with anti-p21 mAb. Arrows indicate nuclear staining (b) ErbB-2 overexpression promotes nuclear exclusion of p27. Serum-starved (0.1% FCS) and randomly growing (10% FCS) cells were fixed and stained with anti-p27 mAb. Two representative images are shown for each condition and similar results were obtained using another anti-p27 mAb. Images are representative of three independent experiments, and exposure times were equal for all images