We describe a complex consisting of ErbB2, Src, EphrinB1 and PTPN13 that mediates EphrinB1 phosphorylation and downstream signaling in breast cancer cells. In addition, we present similar findings using multiple human cell lines suggesting that complex formation and signaling occurs in many, if not all, epithelial cells. With respect to breast cancer, ErbB2/EphrinB1 signaling may be most relevant in tumors with high ErbB2 expression and either low/absent PTPN13 expression or those harboring PTPN13 functional mutations. Our study and those of others predict that these tumors possess an aggressive phenotype and poor prognosis 
In the breast cancer cell lines studied, low/absent PTPN13 together with elevated ErbB2 expression correlate with enhanced ErbB2/EphrinB1 association as well as increased EphrinB1 and Erk1/2 phosphorylation. Interestingly, both MDA-MB231 and MDA-MB468 cells lack detectable (by western blot) ErbB2 expression yet, in the absence of PTPN13, EphrinB1 is phosphorylated. Both of these BL breast cancer cell lines demonstrate over-expression of ErbB1 
suggesting that ErbB1 may hetero-dimerize with low levels of endogenous ErbB2 (forming an ErbB1/ErbB2/EphrinB1 complex) and mediate signaling from the complex. It remains unclear why transient knock-down of PTPN13 in MDA-MB468 cells failed to increased Erk1/2 phosphorylation () though a few possibilities exist. First, while the extent of PTPN13 knock-down was not very efficient, it was enough to increase EphrinB1 phosphorylation (). This suggests that either EphrinB1 does not signal via the MAP Kinase pathway, or that the specific EphrinB1 tyrosine(s) necessary to mediate such signaling were not activated under this condition. Our data demonstrating that knock-down of EphrinB1 greatly attenuates Erk1/2 phosphorylation () argue against the former possibility and support the latter. Second, the presence of high ErbB1 expression in MDA-MB468 cells and the fact that these cells were not serum-starved suggests that ErbB1 signaling (either alone or in combination with low level endogenous ErbB2) modulates downstream pathways which include Erk1/2. Third, it is possible that the ErbB2/EphrinB1 complex is composed of additional components (in fact, we hypothesize this is true), the composition of which may differ among different cell lines and may respond differently under different contexts. Further characterization of the ErbB2/EphrinB1 complex, its association with additional transmembrane proteins (including ErbB family members), as well as intracellular binding partners and the signaling pathways they regulate are on-going and will increase our understanding of the function of this complex in breast cancer. What is clear from these experiments is that EphrinB1 associates with ErbB2 and its phosphorylation is regulated by PTPN13. The absence of PTPN13, as occurs in BL breast cancers, affects EphrinB1 phosphorylation and likely downstream signaling, which may include components of the MAP Kinase pathway.
Given its over-expression in Her2 breast cancers, we further studied the ErbB2/EphrinB1 association in additional breast cancer cell lines. T47D cells which express high levels of ErbB2 and nearly undetectable PTPN13, demonstrate increased ErbB2/EphrinB1 association, elevated EphrinB1 activation and robust Erk1/2 phosphorylation. BT474 cells which demonstrate robust ErbB2 expression and high PTPN13 expression, lack Erk1/2 phosphorylation while MCF7 cells, with low expression of both ErbB2 and PTPN13, also have undetectable phosphorylation of Erk1/2 (). These data suggest that the combination of low/absent PTPN13 and high ErbB2 expression are required for driving EphrinB1 and Erk1/2 phosphorylation.
Our studies also demonstrate that, whereas PTPN13 is the phosphatase silencing EphrinB1 activation, Src is the kinase that activates it (). Thus, signaling from the ErbB2/EphrinB1 complex is regulated by transient associations with PTPN13 and Src. The interaction of two known kinase oncogenes (ErbB2 and Src), a signaling ligand (EphrinB1) and a putative tumor suppressor phosphatase (PTPN13) likely regulates key signals not limited to Erk1/2. In cancers where protein kinase oncogene expression and/or signaling are enhanced, signals mediated from this complex may be altered. Our data in breast cancer cell lines support this hypothesis.
The finding that wildtype ErbB2 associates with EphrinB1 but does not correlate with EphrinB1 or Erk1/2 phosphorylation, suggests that in the absence of ErbB2 activation, Src either does not associate with the complex or remains in an inactive form. In fact, we show that mNeuNT (but not wildtype ErbB2) associates with activated Src, consistent with a requirement for activated ErbB2 to initiate complex signaling. These findings are highly relevant not only for Her2 breast cancers but also for epithelial cancers harboring activating ErbB2 mutations.
mNeuNT association with activated Src suggests that ErbB2 sequence alterations not only enhance complex formation but also initiate its signaling. Both mNeuNT and the human ErbB2 codon 655 polymorphism are single amino acid changes within the transmembrane domain of ErbB2 
suggesting that the transmembrane domain mediates critical interactions in disease. Interestingly, low grade in situ
lesions of the cervix are not associated with this polymorphism 
while advanced cervical cancer is strongly associated with it 
. In head and neck squamous cell carcinoma, codon 655 polymorphism is associated with malignancy 
. Our current findings suggest that ErbB2 transmembrane mutations (like neu (rat) and the human 655 polymorphism) synergize with decreased/lost PTPN13, allowing breast cancer progression via a mechanism involving increased ErbB2/EphrinB1 signaling. Our ErbB2/EphrinB1 truncation experiments also support a functional role for the transmembrane domains. In fact, a functional role for ErbB transmembrane domains has been previously described 
. Our finding that a single amino acid change within ErbB2′s transmembrane domain (mNeuNT) increases its association with EphrinB1, robustly activating complex signaling, supports these published data. The association of human ErbB2 transmembrane polymorphism (codon 655) with cancer is also consistent with this role.
Interestingly, while expression of mNeuNT is required for Erk1/2 phosphorylation in transiently transfected HEK293 cells (), in retrovirally infected HEK293 cells, Erk1/2 phosphorylation occurs in the absence of mNeuNT (). These data suggest that retroviral infection and integration mediates cellular changes not evident in transient plasmid transfections. Despite this, the stable knock-down of EphrinB1 attenuates Erk1/2 phosphorylation consistent with EphrinB1 signaling via the MAP Kinase pathway.
The activated Src antibody used in our studies recognizes tyrosine 416 when phosphorylated; phosphorylation of this residue promotes Src activity 
. Glondu-Lassis et al
showed in the mouse that PTPN13 directly de-phosphorylates this tyrosine residue, inactivating Src 
. We find no significant difference between the amount of activated Src associated with mNeuNT in the presence of wildtype or C/S PTPN13 suggesting that the cellular pool of Src associated with ErbB2 may be insensitive or inaccessible to PTPN13-mediated inactivation (, lanes 1 and 3).
The existence of this complex in breast cancer cells suggests that therapies targeting one component will likely be insufficient at blocking all cellular signaling mediated by the complex. For example, while blocking ErbB2 with trastuzumab (Herceptin) may effectively block ErbB2-mediated signals, it may not alter EphrinB1-mediated signaling from the complex. Our data demonstrate that in the absence of EphrinB1, Erk1/2 phosphorylation is greatly attenuated () suggesting not only that EphrinB1 is a good therapeutic target, but that blocking EphrinB1 together with ErbB2 may efficiently inhibit complex signaling. Our data further suggest that breast cancers with elevated ErbB2 expression and compromised PTPN13 expression and/or function would benefit the most from this type of dual targeting.
Decreased PTPN13 expression in BL breast cancers supports a tumor suppressive role for PTPN13 (). BL breast cancers do not express the estrogen or progesterone receptors nor do they over-express ErbB2 
. Thus, these patients do not benefit from targeted therapies, contributing to their poor outcome. Moreover, while only Her2 breast cancers demonstrate over-expression/amplification of Her2, the other breast cancer subtypes express Her2, albeit at normal levels. However, our data suggest that it is the combination of increased ErbB2 together with compromised PTPN13 that is necessary for EphrinB1 activation and downstream signaling. Despite this, MDA-MB231 cells demonstrate EphrinB1 activation and robust Erk1/2 phosphorylation (). The key to these data may lie in the fact that this cell line (as well as MDA-MB468) expresses high levels of ErbB1 suggesting that EphrinB1 associates with additional ErbB family members. We have some preliminary data suggesting this is indeed the case (data not shown). Thus, complex formation in BL breast cancers is likely to occur with downstream signaling amplified in the absence of PTPN13 and over-expression of ErbB1 or other ErbB family members.
This is the first study to describe a novel complex between ErbB2 and EphrinB1 that is not restricted to breast cancer cell lines but is present in many epithelial cell lines tested. Importantly, ErbB2/EphrinB1 interactions may occur in cis (within the same cell) or in trans (across neighboring cells) mediating dual or bi-directional signals, respectively. The concept of dual/bi-directional signaling in the arena of breast cancer is new and untested with potential implications for tumor growth. Further studies focused on identifying additional components of the ErbB2/EphrinB1 complex and the downstream pathways they regulate may identify additional targets for therapeutic intervention in breast cancer and other solid tumors.