The allosteric mode of activation imparts unique attributes to the activating KD. In particular, catalytic function is not required in a KD when it functions as the activating partner in a KD dimer. In fact HER3, lacks catalytic function altogether [4–6] and most likely HER3 evolved as a dedicated and specialized activating KD in this family. Consistent with this, HER3 is the only HER family member that is unable to generate a signal through homodimerization and is an obligate partner for heterodimerization with the other family members [2, 3]. In addition, the c-terminal tail of HER3 is unique in that it contains six concensus phosphotyrosine sites which bind the SH2 domain of the three regulatory subunits of PI3K [7–9]. This ability is not present in EGFR or HER2, although HER4 also contains a single PI3K-binding site, present in only one of its isoforms [10]. HER3 is also unique in the functions of its ECD. While the HER3 ECD has the ability to bind numerous ligands, the ligand-associated ECD is an aboligate heterodimer and the HER3 ECD appears unable to homodimerize [11]. Whether these unique aspects of HER3 are an evolutionary coincidence or functionally related to each other are not known. But PI3K signaling is a critical regulator of many cellular processes and its premature activation must be avoided. And since HER3 is an access point for activation of the PI3K pathway, it is plausible that nature placed a higher threshold for phosphorylation of the HER3 signaling tail, and the evolution of the HER3 ECD and KD into structures that are unable to activate through simple homodimerization functions to protect the PI3K signaling pathway from premature activation. As such, the signaling activities of the HER3 c-terminal tail depend entirely on the available repertoire and expression levels of other HER family members and ligands.

In addition to its inability to autophosphorylate and signal without partners, additional restraints appear to protect HER3 from premature signaling activity. In the absence of ligand activation, the HER3 c-terminal tail binds and covers its activation surface in trans, restraining its allosteric activation functions [6](Figure 1). In addition, the clustering of HER3, based on both ECD interactions and ICD interactions, appears to provide yet additional restraint through its sequestration away from EGFR and HER2 [6, 12]. HER4 may also participate in HER3 sequestration [6]. This may underlie the favorable prognostic role of HER4 expression observed in patients with early stage breast cancer [13–15].

In particular, HER3 plays a critical role in HER2-mediated transformation and is indispensible for continued tumor cell growth and proliferation in tumors driven by HER2 overexpression. In two different knockdown models, a loss of functional HER3 in HER2 amplified breast cancer cell lines results in a reversion of transformed growth equal to that which is seen with a loss of HER2 [25, 26]. In contrast, the loss of EGFR expression by shRNA knockdown does not affect HER2-driven tumors, confirming that HER3 is specifically essential for HER2-driven tumorigenesis [26]. Consistent with this, the majority of HER2 amplified breast tumors coexpress HER3 but the coexpression of EGFR is inconsistent [26, 27]. A critical function of HER3 in HER2-mediated transformation is the activation of PI3K/Akt signaling. Consistent with this, HER2-amplified breast cancers show widespread activation of Akt [28, 29]. While functionally important in the HER2-amplified subset of breast cancers, the role of HER3 in other types of breast cancers is much less clear. Its level of expression across all breast cancers does not appear to have a correlation with tumor biology. Numerous descriptive series have looked at the impact of HER3 expression on tumor characteristics and disease outcome in breast cancers and the results have been conflicting and inconclusive [14, 24, 30–33].

HER3 has been shown to be highly expressed in melanomas and its expression correlates with proliferation, advanced disease stage and poor prognosis [34, 35]. Its catalytic partner is unlikely to be HER2 since HER2 expression is low or absent in most melanomas [36]. High expression of HER3 in melanoma, in the absence of HER2 expression, suggests that HER3 may be functioning as an allosteric activator of EGFR or HER4 in these lesions [37]. In particular, recent evidence showing frequent mutational activation of HER4 in a subset of melanomas has highlighted HER4 as potentially important in this subset of melanomas [38]. Melanoma metastases frequently show even higher expression of HER3 compared with primary tumors and in experimental models knockdown of HER3 reduces melanoma cell migration and invasion [34]. This suggests that HER3 may play a role in cell migration and invasion. How HER3 might affect cellular motility mechanistically is yet to be determined.

Although HER family proteins have not been implicated in the pathogenesis of prostate cancer, HER3 is overexpressed in some prostate cancers. Experimental evidence suggests that HER3 is required for many of the transformed characteristics of prostate cancer cells including motility in vitro and tumorigenicity in vivo [39]. The existing data does not identify a HER family partner for HER3 in this disease. Clearly much more work is necessary to better define the function of HER3 and other HER family members in the biology of prostate cancers.

Although normal colonic tissue has little to no HER3 expression, HER3 is expressed in a significant proportion of colorectal cancers [40–44]. An intestinal specific HER3 knockout mouse model shows that a loss of HER3 results in loss of HER4 expression suggesting that in the colon, HER4 may be dependent on HER3 and that the HER3-HER4 heterodimer may have a specific function in this tissue type. When crossed to the Apc^Min mice who rapidly develop intestinal and colorectal tumors, loss of HER3 almost completely prevents colonic tumorigenesis and the small intestinal tumors that do develop in the absence of HER3 show reduced Akt signaling and increased apoptosis [45]. This study gives insight into how HER3 might function in a heterodimer with HER4 to prevent apoptosis in a subset of colorectal cancers. In light of the fact that EGFR directed therapies do show some clinical activity in colon cancers, further investigation into HER3’s role in apoptosis in these tumors is warranted.

The role of HER3 is well established in specific subsets of lung cancers but less well defined in the broader scope of lung cancers. Reports of expression or overexpression of HER3 in lung cancers vary considerably from 18–100% and at least some groups find that overexpression correlates with shorter survival times [46–50]. Brain metastases that develop in patients with non-small lung cell lung carcinomas (NSCLC) demonstrate higher levels of phosphorylated HER3 than the primary tumors further linking HER3 signaling with the biology of this disease and possibly with its ability for invasive or metastatic growth [49]. A subset of NSCLCs are driven by mutationally activated EGFR, and these cancer are sensitive to EGFR tyrosine kinase inhibitors (TKIs) [51, 52]. A substantial body of experimental evidence functionally implicates HER3 in the pathogenesis of these EGFR-driven lung cancers. In a comparison of lung cancer cell lines that are sensitive or resistant to the EGFR TKI gefitinib, the best marker of sensitivity to gefitinib is the sensitivity of HER3 signaling and HER3-dependent activation of PI3K [53]. The EGFR-HER3 interdependency is seen in tumor cell types harboring mutationally activated EGFR, but is also seen in some cancers with wild-type EGFR, revealing a role for HER3 that extends beyond the context of mutationally activated EGFR. Treatment of patients with gefitinib is highly effective in tumors with mutationally activated EGFR, although the development of acquired resistance is universal. While resistance typically develops as a result of secondary mutational events within the EGFR kinase domain rendering EGFR drug-resistant [54], rarely resistance can also develop through overactivity of the cooperating proto-oncogene c-MET and the c-MET-dependent phosphorylation of HER3, further highlighting the central role of HER3 in the pathogenesis of this disease [55].

Mounting evidence suggests a functional role for HER3 in ovarian cancers, but mechanistic insight into its precise role in this cancer subtype is currently lacking. Reports of the expression of HER3 in ovarian tumors vary widely from 3–90% [56–61]. In a recent study, the expression of HER3 does appear to correlate with decreased survival [62]. Sheng et al have identified an autocrine signaling loop involving phosphorylated HER3 in a subset of primary ovarian cancers and ovarian cancer cell lines [63]. The knockdown of HER3 inhibits proliferative growth in vitro and tumorigenic growth in vivo in some ovarian cancer cell lines. The functional importance of HER3 appears to be restricted to ovarian cancer cells with basal HER3 phosphorylation as the loss of HER3 in cells that lack HER3 phosphorylation does not have a similar affect. These data support the idea that the expression of HER3 may not be an indicator of its functional involvement in tumors, rather that it is the signaling activity of HER3, revealed by its phosphorylation, that identifies its functional role. The HER family partner that underlies the functions of HER3 in ovarian tumor remains to be defined.

The disease wherein the oncogenic role of HER proteins is best understood and most convincingly established is the HER2-amplified subtype of breast cancers. This has led to the development of several types of targeted therapies intended to inactivate HER2. These therapies have measurable clinical benefits and at least two such drugs are already in clinical practice. But their clinical activities are below what we would expect from inhibiting such a critical tumor driver. The anti-HER2 monoclonal antibody trastuzumab when administered as a single agent to patients with HER2-overexpressing metastatic breast cancer results in response rates between 11–26%, falling below expectations[68–70]. Furthermore, the majority of the responses to trastuzumab as either monotherapy or in combination with chemotherapy are transient and resistance inevitably develops. However, trastuzumab may not be a direct test of the oncogene inactivation hypothesis of cancer therapy since the mechanism of action remains unknown, and it is not a very effective inhibitor of HER2 signaling. In fact, trastuzumab does not affect HER2-HER3 signaling in HER2-amplified cancer cells very well [71–74]. On the other hand, HER family TKIs provide a much more compelling mechanistic basis for inactivation of HER2. Kinase function is essential for HER2-driven tumorigenesis [75] and inhibition of its catalytic kinase activity is a highly rational mechanistic basis for anti-tumor agents. However TKIs also show only modest and short-lived efficacy in clinical studies [76–81]. Recent work from our lab has identified the central role of HER3 in this paradox. Although this disease is driven by the overexpression of HER2, treatment of these cancer cells with HER family selective TKIs inactivates HER2-HER3 transphosphorylation for only several hours and drug therapy ultimately fails to durably suppress HER3 phosphorylation and downstream PI3K/Akt signaling [82]. This is observed with all classes of HER family TKIs including different structural classes and both reversible and irreversible inhibitors. The failure to durably inactivate HER3 signaling averts the apoptotic consequence of TKI therapy and significantly undermines their anti-tumor efficacy against this type of cancer. Mechanistic studies show that this is due to a substantial upregulation of HER2-HER3 transactivation, driven by residual HER2 activity that functions to restore PI3K/Akt signaling despite continued drug therapy. In this tumor subtype, the role of cross-talk with other RTKs such as c-MET or IGF1R or autocrine feedback loops induced by the induction of ligands has been interrogated but there is no compelling evidence to support it [83](and unpublished data). These revelations have redefined the functionally relevant driver of this disease as the HER2-HER3 heterodimer. The redefinition is an important one, particularly in the realm of drug development, since the durable inactivation of the HER2-HER3 driver now defines the bar for the development of effective therapies for this disease. A quantitative analysis of the HER2-HER3 tumor driver reveals that it is endowed with a signal buffering capacity that protects it against a nearly two-log inhibition of HER2 catalytic activity by TKIs, revealing a substantial barrier in the treatment of this disease [83]. The HER2-HER3 driver can be effectively and durably suppressed by treatments that fully inactivate HER2 catalytic function. While the complete inactivation of HER2 kinase requires concentrations of TKI that are not tolerable in vivo, such doses are tolerable and much more effective in intermittent dosing, and non-continuous treatment schedules afford one approach that is currently being explored in clinical studies [83].

The mechanisms that function to protect HER2-HER3 signaling from TKI therapy in HER2-amplified cancers are largely related to the dynamic nature of HER3 as the expression and signaling functions of HER3 are highly regulated through a multitude of mechanisms, and in fact, HER3 is likely the most highly regulated member of the HER family. In the case of HER2-amplified tumors, treatment of HER2-amplified tumor cells with TKIs leads to a compensatory increase in HER3 expression, HER3 membrane localization, and decreased HER3 dephosphorylation, resulting in significantly enhanced HER3 signaling [82]. The compensatory upregulation of HER3 signaling is specifically induced in response to the loss of downstream Akt signaling revealing the reciprocal regulation of HER3 by Akt in this tumor cell type (Figure 2). This is best shown by the observed upregulation of HER3 signaling in response to the experimental inactivation of Akt, and the observed downregulation of HER3 signaling in response to the experimental induction of Akt activity [83].

The specific mechanisms through which Akt can regulate HER3 appear to be multiplex as well as redundant, suggesting that multiple signaling events downstream of Akt, working through different efferent pathways, can regulate HER3 signaling function. Akt can manipulate the signaling functions of HER3 through mechanisms including transcriptional, translational, and post-translational regulation and control of its localization and trafficking. Using a chemical biological approach, we have interfered with each of the components of the HER3/PI3K/Akt signaling pathway including HER2, PI3K, Akt, or mTor. We find that while each manipulation induces an upregulation of HER3 signaling, the pathways involved are different, revealing a plurality of mechanisms in place that link HER3 signaling function with downstream networks (unpublished data). Data from our group and from many others shows that HER3 signaling can be induced through mechanisms that include its transcriptional upregulation, its translational upregulation, through prolongation of its protein half-life, through the inhibition of its dephosphorylation, through the promotion of its membrane localization and increased complex formation with its HER family partners, or through an induction of its ligands in an autocrine fashion. Its transcriptional upregulation is seen when HER2-amplified breast cancer cells are exposed to TKI therapy [83]. Its translational upregulation may be mediated through increased activity of the raptor complex of mTor and its substrate 4EBP1 [84]. Its protein half life may be affected through modulation of negative regulators of HER family proteins, including NRDP1 or LRIG1 [85–88]. Its membrane localization can be further promoted through increased vesicular trafficking to the membrane, mediated through the Akt regulation of nitric oxide pathway or increased membrane retention through MUC4 [89](and unpublished data). Its dephosphorylation can be inhibited through the inhibition of tyrosine phosphatases by increased oxidative stress [82] or through more specific, but yet undefined mechanisms. Its activation state can be increased through the autocrine production of its ligands [63]. HER3 levels can be also be negatively regulated via miRNAs as has been observed with miR205 whose expression decreases HER3 levels and restores sensitivity of HER2 amplified cells to HER TKIs [90].

The importance of HER3 as a cause of drug resistance has also emerged in the treatment of NSCLCs. The best marker of sensitivity to gefitinib in NSCLC cells is its ability to inactivate HER3 signaling [53]. TKI-induced HER3 inactivation even identifies TKI-sensitive NSCLCs that lack the mutational activation of EGFR, identifying its central role in the pathogenesis of this type of lung cancer. Treatment of NSCLCs driven by mutationally activated EGFR ultimately leads to the development of secondary mutations within the EGFR kinase domain, in particular the T790M mutation, which renders the kinase resistant to erlotinib or gefitinib and the emergence of TKI resistant disease [54]. A consequence of the development of EGFR resistance is that HER3 and downstream PI3K signaling become similarly resistant to TKI therapy [91]. EGFR resistance can be overcome with certain classes of irreversible TKIs, and the re-inhibition of EGFR in these resistant cells similarly leads to the re-inhibition of HER3 and PI3K signaling, further highlighting the central role of HER3 in mediating sensitivity or resistance [91]. Some NSCLCs develop resistance to TKIs without any evidence of secondary mutations in EGFR. In some of these cancers, this appears to be due to EGFR-independent HER3 signaling through cross-talk with the heterologous RTK, c-MET. In these NSCLCs, the development of TKI resistance is associated with the MET-dependent phosphorylation of HER3 [55]. This occurs due to amplification of the c-MET gene and overexpression of the MET protein [55]. Amplification of c-MET may be a pre-existing genotype that is selected for during drug therapy [92]. The direct mechanism by which MET may induce HER3 phosphorylation is not yet clear. In addition, resistance to gefitinib or erlotinib also develops in some cases of NSCLC without the development of secondary mutation in EGFR and without amplification of c-MET. While these tumors have EGFR proteins that remain sensitive to TKI therapy, HER3 and PI3K signaling appear to be resistant [55, 92]. This finding uncouples the role of EGFR and HER3, further highlighting the central role of HER3 in the pathogenesis of these lung cancers, and suggests that while EGFR can be ultimately dispensable for tumor progression, HER3 is apparently indispensable.