|Home | About | Journals | Submit | Contact Us | Français|
Amplification and overexpression of ErbB2 strongly correlates with aggressive breast cancers. A deeper understanding of pathways downstream of ErbB2 signaling that are required for transformation of human mammary epithelial cells will identify novel strategies for therapeutic intervention in breast cancer. Using an inducible activation of ErbB2 autophosphorylation site mutants and the MCF-10A three-dimensional culture system we investigated pathways used by ErbB2 to transform epithelia. We report that ErbB2 induces cell proliferation and loss of 3D organization by redundant mechanisms whereas it disrupts apical basal polarity and inhibits apoptosis using Tyr 1201 and Tyr 1226/7 respectively. Signals downstream of Tyr 1226/7 were also sufficient to confer paclitaxel resistance. The Tyr1226/7 binds Shc, and knockdown of Shc blocked the ability of ErbB2 to inhibit apoptosis and mediate paclitaxel resistance. Tyr1226/7 is known to activate the Ras/Erk pathway, however, paclitaxel resistance did not correlate with activation of Erk or Akt suggesting the presence of a novel mechanism. Thus, our results demonstrate that targeting pathways used by ErbB2 to inhibit cell death is a better option than targeting cell proliferation pathways. Furthermore, we identify a novel function for Shc as a regulator of apoptosis and drug resistance in human mammary epithelial cells transformed by ErbB2.
ErbB2/Her2/Neu is overexpressed in 25-30% of breast cancers (Slamon et al., 1987). This overexpression correlates with a poor clinical prognosis in node positive patients (Slamon et al., 1989). ErbB2 targeted therapies such as Trastuzumab, a humanized anti-ErbB2 antibody, or Tykerb, a small molecule kinase inhibitor, are used to treat ErbB2-positive patients. Combination of anti-ErbB2 therapy with chemotherapeutic drugs such as paclitaxel or an anthracycline/cyclophosphomide regime shows synergistic response in controlling the disease (Romond et al., 2005; Slamon et al., 2001). However, only one-third of ErbB2 positive tumors respond to Trastuzumab treatment and those that do acquire resistance within three years after the initiation of treatment (Nahta and Esteva, 2007). The quick onset of resistance and the presence of de novo resistance to anti-ErbB2 therapy significantly limit the effectiveness of currently available treatments. Thus, a deeper understanding of the signaling pathways downstream of ErbB2 that regulate transformation of epithelial cells and those that block the action of cytotoxic drugs is necessary to develop additional ways to successfully treat women with ErbB2 positive breast cancers.
Others and we have shown that activation of ErbB2 transforms epithelial cells in culture by inducing an increase cell proliferation, disrupting apical-basal cell polarity and inhibiting apoptosis (Henson et al., 2006; Muthuswamy et al., 2001; Yarden and Sliwkowski, 2001; Yu et al., 1998). As a receptor tyrosine kinase, activation of ErbB2 results in the autophosphorylation of five tyrosine sites in its cytoplasmic tail (Akiyama et al., 1991). Four tyrosine residues, Y1144, Y1201, Y1226/7 and Y1253, are each sufficient to transform fibroblasts, which relates to their ability to activate Ras/MAPK signaling pathway (Dankort et al., 1997; Janes et al., 1994). The Tyrosine at position Y1028 is a negative regulator of transformation and mutation of this site augments the transforming activity of ErbB2 (Dankort et al., 1997). Although the Y1144, Y1201, Y1226/7 and Y1253 are all redundant in their ability to transform fibroblasts they do differ in their transformative potential in vivo because mice expressing Neu-Y1226/7 induce multifocal tumors with short latency (102±22 days) whereas Neu-Y1144 induce focal tumors after long latency (152±47 days) (Dankort et al., 2001b; Marone et al., 2004). It is not clear why mutant ErbB2 receptors harboring selected tyrosine residues differ in their ability to induce mammary tumorigenesis. Furthermore, it is also not clear if the pathways activated by autophosphorylation sites differ in their ability to disrupt apical-basal polarity or inhibit apoptosis. We rationalize that investigating the mechanisms by which ErbB2 transforms three-dimensional mammary acini will provide novel insights into signaling pathways required for ErbB2 to disrupt cell polarity and inhibit apoptosis.
Using a combination of an inducible ErbB2 activation system and a three-dimensional culture system, we demonstrate that Y1144, Y1201, Y1226/7 and Y1253 all redundantly mediate ErbB2 signaling to promote proliferation and disrupt organization of epithelial cells within the acinus. However, only Y1201 was as efficient as wild type to disrupt apical-basal polarity, and only Y1226/7 was able to inhibit cell death during morphogenesis and in response to treatment with the chemotherapy drug Taxol. Furthermore, we show that the signaling adaptor molecule Shc that binds to 1226/7 was required for ErbB2-induced inhibition of cell death demonstrating that while ErbB2 uses redundant signaling pathways to disrupt normal cell properties and induce transformation, it uses specific pathways to disrupt cell polarity and inhibit cell death.
To investigate the effect of inducibly activating signaling pathways downstream of each autophosphorylated tyrosine residue in human mammary epithelial cells, we generated a chimeric ErbB2 receptor that will respond to a synthetic, small molecule ligand, AP1510 (herein referred to as dimerizer), but will not respond to stimulation by EGF family of ligands (Muthuswamy et al., 1999). We have previously shown that activation of this chimeric receptor phenocopies wild type-ErbB2 in its ability to transform cells and activate downstream signaling pathways (Muthuswamy et al., 1999). This approach has significant benefits over using cells stably overexpressing constitutively active ErbB2 because the process of stable selection can allow cells to adapt to constitutive ErbB2 signaling and interfere with our ability to analyze signals immediately downstream of ErbB2 activation.
For this study, we engineered the ErbB2 chimera to create mutant version in which only one of the five known autophosphorylation sites is retained as a tyrosine and the others were mutated to phenylalanine residues. In addition, we also generated an ErbB2 chimera that lacks all five autophosphorylation residues. To keep nomenclature simple we refer to the mutants by the number that corresponds to the Tyr residue that was not mutated (Figure 1 A).
To generate populations of human mammary epithelial cells that express comparable levels of the chimera (Figure 1B), MCF 10A cells were infected with retrovirus expressing the ErbB2 chimera and the cells were sorted by flow cytometry using an antibody against the p75 extracellular domain of the chimeric receptor. Three populations of cells were collected, each expressing the chimeric receptor at different levels. The lowest level of expression was gated such that there was a 6% overlap with the non-specific signal observed in the parental MCF 10A cells that do not express the chimeric receptors and hence do not express the p75 receptor at the cell surface. Pools 2 and 3 were collected at higher levels with approximately 2.5 and 5.0 fold increase over the non-specific signal observed in parental cells. The population of cells sorted for high levels of expression (Pool 3) had constitutive activation of the receptor and did not respond to the dimerizer, making them unsuitable for these studies. Cells in Pool 2, respond well to ligand stimulation by an increase in receptor tyrosine phosphorylation levels over the levels of tyrosine phosphorylated observed in the unstimulated state. To confirm that the cells have comparable levels of receptor expression, we analyzed monolayer cultures by immunofluroresence analysis using anti-HA antibodies that recognize the tagged chimeric receptor. The populations with medium levels of expression expressed the chimeric receptors at similar levels in every cell, suggesting that the population will respond uniformly to receptor activation. We determined that this expression level and the ability to undergo dimerizer inducible phosphorylation was stable for 4 and 10 passages after sorting, and used the cell lines according to this range in our experiments. We have limited our experiments to cells within these passages.
To determine if the receptor chimeras undergo inducible phosphorylation, cells were stimulated and the protein lysates analyzed for changes in tyrosine phosphorylation of the chimeric receptor (Figure 1C). Consistent with the fact that the ErbB2 mutants possess only one of the five autophosphorylation sites, phosphotyrosine levels in the mutants were significantly lower than those observed for the wild type (wt) receptor (compare lanes 6, 8, 10 and 12 to lane 2). Interestingly, the autophosphorylation sites differed in the extent to which they were phosphorylated in response to dimerization. Y1226/7 was phosphorylated 2.5 fold more than Y1144 and Y1201and 5 fold more than Y1253. Although it is unclear why the sites are phosphorylated differentially, it is possible that Y1226/7 is phosphorylated to a greater level because it is comprised of two tyrosines immediately adjacent to each other, that it is the major phosphorylation site in ErbB2 homodimers, or that the sites are differentially sensitive to phosphatase activity. Further analysis will be required to differentiate between these possibilities.
Previous studies have demonstrated that each of the individual tyrosine mutants, but not a mutant lacking all the five autophosphorylated tyrosine sites, were capable of transforming Rat1 fibroblasts when expressed within the context of an activated form of ErbB2 (Dankort et al., 1997). We wanted to determine if the autophosphorylation site mutants within the context of our inducible chimeric ErbB2 retain the ability to transform Rat1 fibroblasts in an inducible manner. The wt and mutant chimeras were transiently transfected into Rat1 fibroblasts and their foci forming ability was assessed (Figure 2A). In the absence of receptor activation few, if any, foci were observed. Consistent with previous results, activation of wt and all of the ErbB2 mutants, with the exception of phospho-deficient mutant (YPD), induced in a significant increase in foci formation demonstrating that they were sufficient to transform fibroblasts. Despite the differences in their phoshotyrosine content, the Y1201 and Y1253 were as potent as the wt in their ability to induce foci formation in Rat1 fibroblasts. This suggests that the stochiometry of phosphorylation of these sites observed in the mutant receptor are similar to those observed within the context of a wt receptor. It is not clear why Y1226/7 has higher transforming ability than the wt receptor. It might be due to difference in levels of expression in Rat1 cells or due to the lack of the negative regulatory site Y1028. Nevertheless, these data demonstrate the inducible chimeric ErbB2 receptor phenocopy the full-length receptor in their ability to transform Rat1 fibroblast cells.
We have previously shown that ErbB2 activation disrupts the three-dimensional (3D) acini structures formed by MCF-10A cells grown on 3D matrix cultures. Unlike the fibroblast foci formation assay, transformation of 3D acini involves three qualitatively different events: induction of cell proliferation, disruption of cell polarity and inhibition of cell death. To investigate if the autophosphorylated tyrosine sites differ in their ability to regulate the events required to transform MCF-10A 3D acini, we activated ErbB2 mutants in mature, organized, 3D acini. As expected, activation of wtErbB2 induced disruption of 3D acinar organization (Fig. 2B see insert), herein referred to as a multiacinar structure. As we have reported in the past, MCF-10A cells expressing ErbB2 have a basal, dimerizer independent, receptor phosphorylation (Fig. 1C) and form multiacinar structures in 3D cultures even in the absence of small molecule ligand induced dimerization and activation (Fig. 2C). Activation of wtErbB2 resulted in 53.9% of multiacinar structures, which was 1.82 fold higher than that observed in the absence of receptor activation. All the ErbB2 mutants, except YPD, retain the ability to induce multiacinar structures in MCF-10A 3D acini. WtErbB2 and all the mutants, except YPD, were capable of increasing rates of proliferation in the three dimensional structures (Supplemental Figure 1). Therefore, as observed in the fibroblast focus-forming assay, each of the four single tyrosine phosphorylation sites on ErbB2 is capable of inducing morphological disruption of MCF-10A 3D acini.
ErbB2 induced disruption of acinar organization is accompanied by a disruption of cell polarity. Although disruption of cell polarity in MCF-10A 3D acini can be measured by monitoring localization of golgi apparatus (Aranda et al., 2006), we chose to use another the epithelial cell line, Madin Darby Canine Kidney II (MDCK II) cells, to directly analyze the effect of ErbB2 activation on disruption of apical-basal polarity. MDCK II cells are ideal for studying apical-basal cell polarity because the apical, lateral and apical-lateral borders can be effectively resolved using markers such as gp135 (apical), E-cadherin (lateral) and zona occludens 1 (ZO-1) (apical-lateral border). Our lab has previously shown that activation of ErbB2 for 24 hours results in mislocalization of tight junctions from their normal localization at the apical-lateral border in 75% of cells (Aranda et al., 2006).
We generated MDCK II cells expressing the wtErbB2 or the autophosphorylation site mutants. We selected for cells expressing comparable levels of the receptor chimera. To limit variability in phenotype due to differences in expression level of the receptor, we only examined cells with a defined intensity of anti-HA signal as determined by Image J software (see Methods section for details). In polarized MDCK cells, ZO-1 was restricted to a region of the cells that is 2.0 μm from the apical surface. Activation of ErbB2 induced mislocalization of ZO-1 to regions that are 3.0 μm or more below the apical membrane of the cell. Although wtErbB2 and all of the autophosphorylation site mutants were effective in disrupting cell polarity, we observed a quantitative difference in the extent to which they induced disruption of tight junctions. Activation of wtErbB2 induced mislocalization of ZO-1 in approximately 70% of cells whereas Y1144, Y1226/7 and Y1253, but not Y1201, mutants displayed significantly lower ZO-1 mislocalization (52 – 54%) (Figure 3). The disruption of tight junctions observed by activation of the Y1201 mutant was not significantly different from the effect observed upon activation of the wt receptor suggesting that Y1201 was as potent as the wt receptor in its ability to disrupt apical-basal polarity.
We next investigated if the ErbB2 autophosphorylation site mutants differ in their ability to inhibit cell death. Although all the mutants retained the ability to disrupt acinar structure and induce proliferation, we noticed that only activation of Y1226/7 induced filling of the luminal space in 3D acini (Supplemental Figure 2). We have previously demonstrated that the lumen formation and maintenance in MCF-10A 3D acini is regulated by death of cells in the lumen (Debnath et al., 2002). So we investigated if ErbB2 uses signals downstream of the Y1226/7 autophosphorylation site to inhibit apoptosis. Activation of wtErbB2 during MCF-10A morphogenesis in 3D blocks activation of caspase-3 in cells located in the luminal space (Debnath et al., 2002). We investigated if activation of the autophosphorylation site mutants differed in their ability to inhibit activation of caspase-3. As observed for regulation of proliferation and polarity readouts, activation of YPD had no effect in the percentage of acini that stained positive for activated caspase-3. Interestingly, while activation of Y1226/7 showed a significant decrease in acini with cells dying in the lumen, none of the other mutants showed significant inhibition of apoptosis (Figure 4A and B). We note that the modest decrease observed for Y1144, Y1201 and Y1253 mutants did not reach statistical significance in our experiments (see Methods section for more details).
Decrease in rates of apoptosis relates to the presence of filled lumens in 3D acini. We used acini stained for DAPI and monitored the presence or absence of cells in the lumen by optical sectioning approaches. Activation of wtErbB2 and Y1226/7 resulted in acinar structures that had lumens filled with cells whereas acini derived from the other mutants did not show a significant increase in percentage of acini with cells in the luminal space (Supplemental Figure 2 and Figure 4C). These results demonstrate that only signals from Y1226/7 were sufficient to inhibit cell death whereas signals from any of the other four tyrosines are sufficient to induce proliferation.
Inhibition of apoptosis is particularly important in the clinic, where it contributes to resistance to cytotoxic chemotherapies.
We tested the ability of ErbB2 activation in MCF-10A cells to inhibit death of cells treated with various cytotoxic agents. Multiple chemotheraputic drugs blocked growth of MCF-10A cells in a dose dependent manner (Table 1). Interestingly, short-term activation of ErbB2 did not protect cells against death induced by camptothecin, doxorubicin or etoposide (Supplemental Figure 3), whereas activation blocked cell death induced by paclitaxel (Figure 5A). Although previous studies have shown that cells overexpressing ErbB2 resist death induced by camptothecin, doxorubicin or etoposide (Knuefermann et al., 2003), all those experiments were performed using cells that were selected for high levels of expression of constitutively active ErbB2. It is likely that long-term activation of the receptor leads to activation of pathways that are not direct targets of the ErbB2 receptor, but are activated in cell populations selected for constitutive ErbB2 signaling. Our results using the inducible MCF-10A cells suggest that resistance to paclitaxel is mediated by signaling pathways directly downstream of ErbB2 activation.
To determine if the signaling pathways downstream of the individual tyrosine sites differ in their ability to inhibit paclitaxel-induced death, we activated ErbB2 mutants with specific autophoshorylation sites. Interestingly, only activation of Y1226/7 inhibited paclitaxel induced cell death, whereas activation of YPD, Y1144, Y1201 or Y1253 was unable to inhibit paclitaxel induced death. The inhibition of paclitaxel-induced death was observed over a range of paclitaxel concentrations from 0.05 μM to 1.0 μM (Figure 5B and data not shown) and was comparable to the effect observed in response to activation of wtErbB2 (Figure 5B) suggesting that Y1226/7 is the principal effector of ErbB2 induced signals that inhibit apoptosis. Point mutations of ErbB2 that lack one phosphorylation site, leaving the other phosphorylation site intact, Y1226/7F or Y1253F, still retained the ability to inhibit cell death (Suppl. Fig. 4) indicating that in the absence of Y1226/7, all the other sites collectively accomplish what Y1226/7 can accomplish on its own.
ErbB2 signaling is thought to activate two main pathways, the Ras/Erk and the PI3K/Akt pathway (Pupa et al., 2005). Although the Erk pathway is thought to primarily promote proliferation, there are some indications that it can function as a cell survival signal, whereas activation Akt pathway inhibits cell death (Pupa et al., 2005; Wada and Penninger, 2004). Previous studies have shown that signaling from any of the autophosphorylation site is sufficient to activate Ras/Erk signaling (Dankort et al., 2001b), consistent with this observation we find that activation of all ErbB2 mutants induced an increase in phosphorylation of Erk (Figure 5C). In addition, all the ErbB2 mutants were able to induce an increase in phosphorylation of Akt at ser 473, a surrogate for monitoring activation of the Akt pathway (Figure 5D). We note that the modest increase in Erk and Akt phosphorylation observed in YPD is consistent with the previous observation that YPD still retains the ability to interact with and activate c-Src tyrosine kinase (Kim et al., 2005), which in turn can activate downstream signaling. Although previous studies have demonstrated that activation of Akt signaling regulates inhibition of cell death pathways downstream of ErbB2 and other oncogenes (Nelson and Fry, 2001; Zhou et al., 2000), we did not observe any correlation between Akt activation and inhibition of paclitaxel induced death. It is possible the amplitude of Akt activation is not high enough for Akt to induce anti-apoptotic pathways. Consistent with this possibility we have previously shown that ErbB2 homodimers are weaker than ErbB1/ErbB2 heterodimers of EGF ligand in their ability to activate Akt (Muthuswamy et al., 1999; Zhan et al., 2006). ErbB2 induced expression of an anti-apoptotic protein Mcl-1 that has been shown to regulate cell death (Henson et al., 2006), however, we did not observe any decrease in Mcl-1 levels in response to ErbB2 activation (Supplemental Figure 5). The redundancy with which ErbB2 activates the Erk and Akt pathways, and the lack of Mcl-1 protein regulation, suggests that Y1226/7 may facilitate activation of novel pathways to inhibit cell death in response to paclitaxel.
Y1226/7 is known to bind the signaling adaptor Shc (Dankort et al., 2001a), however a recent study using an in vitro Src homology 2 (SH2) and Protein tyrosine binding (PTB) domain protein array identified Syk and Abl2, in addition to Shc, as major binding partners for the Y1226/7 phosphotyrosine (Jones et al., 2006). The same study showed that Abl2 SH2 domain also interacts with Y1144 and Y1253; the Shc SH2 domain can also interact with Y1144, Y1201, and Y1253 and the Syk SH2 domain with Y1028. The Y1028 site has been identified as a negative regulatory site, which is consistent with the observations that Syk functions as a tumor suppressor in breast cancer (Coopman et al., 2000). The sequence surrounding Y1226/7 (DNLYYWDQD) conforms best to the NPXY consensus binding site for the PTB domain of Shc (Harrison, 1996). Consistent with this possibility, far-western analysis demonstrated that the Shc-PTB domain, and not the Shc-SH2 domain, associates with Y1226/1227 (Dankort et al., 2001a). Coimmunoprecpitation analysis demonstrates that Shc does not interact with phospho Y1144, Y1201 and Y1253 whereas it interacts with Y1226/7 (Figure 5C). It is likely the data obtain from protein arrays differ from co-immunoprecipitation analysis in both stringency and sensitivity, and that Y1126/7 is the primary Shc binding site on ErbB2. Recent data identified a relationship between Shc phosphorylation and lower rates of apoptosis in Polyomavirus middle T antigen induced mammary tumors (Ursini-Siegel et al., 2008). Whether Shc plays a role in ErbB2 induced inhibition of apoptosis is not known.
We hypothesized that the recruitment of Shc to the ErbB2 receptor is required for Y1226/7 induced inhibition of apoptosis. Our results show that both wild type ErbB2 and Y1226/7, but not the other mutants, coimmunoprecipitated with Shc demonstrating that activation of ErbB2 recruits Shc through Y1226/7. To rule out the possibility that Shc phosphorylation (and hence activation of a Shc pathway) can be induced by sites other than 1226/7, we analyzed changes in Shc phosphorylation in response to activation of ErbB2 mutants. While wtErbB2 and Y1226/7 induced an increase in Shc phosphorylation, other ErbB2 mutants were not capable of activating Shc (Figure 6A) demonstrating that Shc is not activated by any other residues.
To determine if Shc is required for Y1226/7 induced inhibition of cell death we stably knocked down Shc expression in Y1226/7 cells using RNAi. We generated three shRNAs that induced a 6 to 15 fold decrease in the p46 Shc protein levels, an 8 to 35 fold decrease in the p52 Shc protein levels, and a 10 to 100 fold decrease in the p66 Shc protein levels in 10A-Y1226/7 cells (Figure 6B). Y1226/7 is thought to mediate activation of the Erk pathway by recruiting Shc to the membrane (Dankort et al., 1997). Therefore we expected that a functional Shc knockdown in these cells would decrease Erk phosphorylation in response to ErbB2 -Y1226/7 activation. Y1226/7 receptor induced phosphorylation of Erk and Akt was two-three fold weaker in Shc.RNAi cells compared to the parental cells demonstrating that knockdown of Shc interferes with signaling by ErbB2 (data not shown).
All three Shc knockdown cell populations were analyzed for the effect of Y1226/7 activation induced inhibition of paclitaxel induced cell death (Figure 6C). Downregulation of Shc inhibited the ability of Y1226/7 to block cell death demonstrating that Shc plays a critical role during Y1226/7-induced inhibition of apoptosis.
To determine is Shc is a critical effector of the anti-apoptotic signals generated by wtErbB2, we knocked down Shc in 10A-ErbB2 cells (Figure 7A) and investigated if activation of wt-ErbB2-induced inhibition of apoptosis in 3D structures requires Shc. While activation of ErbB2 inhibited luminal apoptosis in control 3D acini, activation of ErbB2 was not able to inhibit apoptosis in cells knock down for Shc expression (Figure 7B, C).
Thus, our results demonstrate that ErbB2 uses non-redundant mechanisms to inhibit cell death by activating a Shc dependent signaling pathway.
Here we use inducible ErbB2 activation system and 3D cell culture to better understand ErbB2 induced biological effects in human mammary epithelial cells. We demonstrate that ErbB2 uses redundant mechanisms to induce cell proliferation, whereas it is more selective in pathways used to disrupt cell polarity and inhibit cell death. Only activation of Y1201 was as potent as wild type ErbB2 in its ability to disrupt apical-basal polarity and only Y1226/7 is sufficient for inhibition of cell death. The ability of Y1226/7 to inhibit cell death required the signaling adaptor molecule Shc. Although Shc is known regulator of cell proliferation pathways, its role in cell death is not well understood. Thus, we have identified specific autophosphorylation residues that are sufficient to trigger cell biological effects relevant to cancer. In addition, we have defined a novel role for Shc as a regulator of cell death pathways in ErbB2 mediated transformation mammary epithelial cells.
Consistent with previous studies, our results obtained using inducible ErbB2 autophosphorylation site mutants demonstrate that phosphorylation of Y1144, Y1201, Y1226/7 or Y1253 residues is sufficient to induce cell proliferation and activate the MAP Kinase and Akt signaling pathways. These observations highlight the robustness with which ErbB2 signals to induce cell proliferation and transform mammary epithelia. It also raises caution about developing strategies to target pathways downstream of ErbB2 that regulates cell proliferation because it is more likely that it will overcome the inhibition by activating alternate pathways to drive cell proliferation.
ErbB2 was selective in its ability to disrupt polarity. Only activation of Y1201 phenocopied wild type ErbB2, however, Y1144, Y1226/7 and Y1253 are all capable of inducing a partial disruption of polarity (Figure 3). We have recently shown that ErbB2 requires an interaction with Par6/aPKC a polarity complex to disrupt cell polarity (Aranda et al., 2006). Consistent with partial to complete disruption of polarity we find that all the autophosphorylation sites retain the capacity to interact with Par6/aPKC (data not shown). Interestingly, ErbB2 also requires the Src kinase to disrupt polarity in MDCK cells (Kim et al., 2005). Src interacts with the kinase domain of ErbB2, likely to Tyr 882 within the activation loop (Kim et al., 2005). This tyrosine is intact in all of the autophosphorylation mutants used in this study and can thus explain the ability of Y1144, Y1226/7 and Y1253 mutants to disrupt polarity. However, except Y1201, activation of other tyrosine residues resulted in a partial effect suggesting that Y1201 has unique properties in addition to interacting with Src. Y1201 is known to recruit CrkII and Nck, known substrates of Src kinase and regulators of the cytoskeleton (Buday et al., 2002; Dankort and Muller, 2000; Dankort et al., 1997; Feller, 2001), raising the possibility that in addition to interacting with Src, recruitment of Src specific substrate(s) to the proximity of the receptor is necessary to induce a complete disruption of polarity. Further analysis will be necessary to test this hypothesis.
Recent studies demonstrate that pathways downstream of Y1226/7 are also sufficient to induce migration (Marone et al., 2004) and neovasculogenesis (Saucier et al., 2004). Our results, together with these observations suggest that among the signaling pathways activated by ErbB2, those that are downstream of Y1226/7 uniquely regulate biological process relevant to cell transformation and offer an opportunity for therapeutic intervention.
It is likely that understanding how Shc regulates cell death pathways will offer new therapeutic opportunities for controlling paclitaxel resistance. Our results suggest that the Shc mediated inhibition of apoptosis is independent of activation of the Ras/MAPK signaling pathways because all ErbB2 autophosphorylation site mutants retain the ability to activate Ras/MAPK signaling (Figure 5C and D) but lack the ability to inhibit cell death (Figures 4 and and5B).5B). Most known mechanisms of paclitaxel resistance depend on altering the dynamics of the microtubule polymerization (Orr et al., 2003). Shc can regulate microtubule dynamics by at least three independent mechanisms. Shc interacts with SHIP-2 (Habib et al., 1998) which has recently been shown to deter microtubule stabilization in mast cells (Leung and Bolland, 2007). It is possible that the Shc – SHIP-2 interaction leads to destabilization of microtubules. Shc has also been shown to interact directly with actin in PC12 cells (Thomas et al., 1995) and to cause reorganization of the cytoskeleton (Gu et al., 1999). Since changes in the actin cytoskeleton can increase resistance to anti-microtubule drugs (Verrills et al., 2006), it is possible that Shc's interaction with the actin cytoskeleton mediates resistance to taxol. Finally, the Mediator of ErbB2 Driven Cell Motility (MEMO), which binds to ErbB2's Y1226/7, possibly through Shc, is necessary for ErbB2 induced microtubule polymerization in T47D and SkBr3 cells (Marone et al., 2004). It is also possible that Shc is activating general anti-apoptotic pathways, such as upregulation of Bcl-2, Bfl-1 or downregulation of BH3 proteins, Bim, consistent with previous studies where overexpression of Bcl-2 has been shown to inhibit apoptosis in 3D structures (Debnath et al., 2002). Interestingly, a recent study showed that expression of a phosphorylation site mutant version of Shc (Y313F) increases cell death in a polyoma virus mT mouse model of mammary tumorigenesis, suggesting that the ability of Shc to regulate cell survival pathways is not limited to ErbB2 dependent tumors (Ursini-Siegel et al., 2008). A deeper analysis to understand how Shc signals to inhibit paclitaxel induced cell death is likely to have broad clinical benefit.
Resistance to ErbB2 directed therapies in multiple clinical settings highlights the need to develop alternative treatment paradigms for ErbB2 positive tumors. Our results suggest that inhibiting cell death pathways, and not cell proliferation pathways, downstream of ErbB2 as a potentially effective therapeutic strategy since ErbB2 uses non-redundant mechanisms to inhibit cell death. Understanding the pathways downstream of the ErbB2-Shc interaction may not only identify novel combination drug targets but also identify biomarkers for predicting response to chemotherapies.
The individual autophosphorylated tyrosine sites were mutated one tyrosine at a time to phenylalanine using the Quickchange® kit. The constructs were generated in Neu, the rat homolog of human ErbB2, intracellular domain with silent mutations abolishing the Xba restriction sites in a pCG backbone. After sequence verification the ErbB2 intracellular domain was PCR amplified, purified, digested with Xba1/Spe1 and ligated to the retroviral vector pBabe.p75.F2.HA previously described (Muthuswamy et al., 1999) at the Spe1 locus.
The primers used in to generate the tyrosine to phenylalanine mutations were as follows; 5′ Y1028 gta gac gct gaa gaa ttc ctg gtg ccc cag, 3′ Y1028 ctg ggg cac cag gaa ttc ttc agc gtc tac, 5′ Y1144 agc ccc cag ccc gaa ttt gtg aac caa tca, 3′ Y1144 tga ttg gtt cac aaa ttc ggg ctg ggg gct, 5′ Y1201 gtg gag aac cct gaa ttc tta gta ccg aga, 3′ Y1201 tct cgg tac taa gaa ttc agg gtt ctc cac, 5′ Y1226/7 ttt gac aac ctc ttc ttc tgg gac cag acc, 3′ Y1226/7 ggt ctg gtc cca gaa gaa gag gtt gtc aaa, 5′ Y1253 gag aac cct gag ttc cta ggc ctg gat, 3′ Y1253 atc cag gcc tag gaa ctc agg gtt etc.
MCF-10A cells were maintained in DMEM/F12 (Gibco), 5% Horse Serum (Hyclone), 1% penicillin/streptomycin (Gibco), 100ng/mL cholera toxin, 10 μg/mL insulin, 500 ng/ml hydromycin (Sigma) and 20 ng/ml EGF (Peprotech). Assay media, used for three dimensional studies, was made of DMEM/F12, 2% Horse Serum, 1% penicillin/streptomycin, 100 ng/mL cholera toxin, 10 μg/mL insulin and 500 ng/mL hydromycin. MDCKs and Rat1 fibroblasts were grown in Dulbecco's Modified Eagle Medium (Gibco), 10% FBS, and 1% penicillin/streptomycin.
Rat1 fibroblasts were plated at 2×10^5 cells per 6 cm dish on day 0. The next morning cells were transfected with 10 μg pBabe p75.F2.ErbB2.HA and pBabe p75.F2.8A.HA using calcium phosphate. Six hours after transfection cells were washed twice with DMEM/0.5% FBS and then incubated overnight in DMEM/10% FBS. On day2 the media was changed. By day 4 cells had become confluent and cells were stimulated by changing the media and adding 1.0 μM AP1510 (Ariad Pharmaceuticals). Media was changed every three days until d16, when cells were fixed with 4% PFA and stained with Geimsa.
MCF-10A cells expressing the chimeric receptor were dissociated by incubation at 37°C with 10mM EDTA:0.25% trypsin:PBS for 30 minutes. The cells were then collected and washed twice in growth media. 15 × 10^6 cells were resuspended in 3mL of media and incubated for 30 minutes with the p75 NGFR antibody. Cells were washed three times with fresh media, resuspended in 1.5 mL media with 15 μL of anti-mouse 488 antibody and incubated at room temperature for 1.0 hour before washing again three times in growth media. After the final wash cells were resuspened in 3.0 mL of growth media, filtered through a 3.5 μm mesh and run on the FACSVantage SE Cell Sorter. The population of positive staining cells was divided into low, medium and high expressing groups, collected and returned to culture. Medium level expressing cells were used for further experiments.
4000 cells were plated per well of an eight chamber slide coated with 70 μL Matrigel® in assay media containing 2.5% matrigel and 5ng/ml EGF. Cells were cultured for 16 days, changing media every 4 days and stimulating with 1.0 μM AP1510 on day 12. Phase images on cells were taken every four days. After imaging on day 16 slides were fixed in 4% PFA, permeabilized in 0.5% Triton and incubated with the antibodies indicated for immunoflouresence.
MDCK cells were plated on 12-well transwell filters at 0.5×10^6 cells/well and grown for four days, at which point cells were stimulated with 1uM AP1510. After 24 hours of stimulation cells were fixed in 4% PFA, permeabilized in 0.2% Triton and incubated with the antibodies indicated for immunoflouresence. Images were all taken with preset exposure times. Cells, which expressed HA at a level between 10 and 100 units of intensity according to ImageJ were analyzed for mislocalization of ZO-1.
MCF-10A cells expressing the mutants were plated at 0.6×10^6 cells per well of a six well plate in growth media. Two days after plating the media was replaced again with growth media +/-Taxol, in varying concentrations, and +/- 1.0 μM AP1510. After two days in the presence or absence of drugs, the plates were washed three times in PBS, fixed with a crystal violet solution (0.25% crystal violet in 20% methanol) for 10 minutes, washed six times in water and incubated in 700 μL lysis solution (51% water, 48% ethanol, 1% HCl) for 1 hour. To quantitate the amount of crystal violet in each sample, 25 μL of the lysate was diluted in 975 μL of water, mixed and the OD 600 was determined using the spectrophotometer. Percent survival was calculated using the untreated condition as the denominator.
The HA antibody was obtained from Covance, actin from Sigma, Ki67 from Zymed Laboratories, NFGR from Chromoprobe. The Shc, phospho-tyrosine pY20, Erk2 antibodies were all obtained from BD Transduction Laboratories. phospho Ser 473-Akt, Akt, cleaved Caspase 3, phospho-p44/42MAPK were all from Cell Signaling. Quantitation of blots was done with ImageJ blot analysis.
All statistics were performed using a standard Student's t-Test.
Supplemental Figure 1. Autophosphorylation Site Mutants Promote Proliferation in three-dimensional Epithelial Structures. MCF-10As expressing mutant ErbB2 were plated on matrigel. Structures were grown for 12 days and then stimulated for four days with 1.0 μM AP1510, fixed and stained. Immunoflourescent images of three-dimensional structures, where cells express the indicated ErbB2 mutant are seen in (A). The cell cycle marker Ki67 is visualized in red. The nuclei have been stained blue with DAPI. (B) The percentage of structures staining positive for Ki67 was quantitated in two independent experiments.
Supplemental Figure 2. Y1226/7 Mediates ErbB2 Induced Luminal Filling. MCF-10As expressing mutant ErbB2 were plated on matrigel. Structures were grown for 12 days and then stimulated for four days with 1.0 μM AP1510, fixed and the nuclei were stained with DAPI, depicted in blue. Scale bars represent 50 microns.
Supplemental Figure 3. Short-term activation of ErbB2 Does Not Inhibit Cell Death Induced by Camptothecin, Doxorubicin or Etoposide. MCF-10As expressing inducible ErbB2 were plated at 0.6 × 10^6 cells. After two days the media was changed, introducing the indicated concentrations of Camptothecin, Doxorubicin or Etoposide in the presence or absence of 1.0 μM AP1510. On day 4, cells were stained with crystal violet and then lysed. The amount of dye released into the lysate was quantitated using a spectrophotometer. Relative amounts are depicted in the graph, which represents 3 experiments.
Supplemental Figure 4. Y1226/7F Mediates ErbB2 Inhibition of Paclitaxel Induced Apoptosis. ErbB2 point mutants Y1226/7F and Y1253F were expressed in MCF-10A cells and then treated as described in Figure 5B. This graph represents percent cell survival in response to treatment with 0.5 μM paclitaxel.
Supplemental Figure 5. Activation of ErbB2 Does Not Decrease Mcl-1 Protein Levels. MCF-10As expressing mutant ErbB2 were activated with 1.0 μM AP1510 for 1, 4 or 24 hours. The membrane was first blotted for Mcl-1 and then stripped and reprobed for actin.