ERBB3

Abstract

Commentary to:

ErbB3 expression promotes tumorigenesis in pancreatic adenocarcinoma

J. Spencer Liles, J. Pablo Arnoletti, Ching-Wei D. Tzeng, J. Harrison Howard, Andrew V. Kossenkov, Peter Kulesza, Martin J. Heslin and Andrey Frolov

The accompanying article by Liles et al.1 describes the role of ERBB3 as an enhancer of pancreatic cancer cell growth and source of acquired sensitivity to the EGFR kinase inhibitor, Erlotinib (Tarceva). Beyond its potential therapeutic relevance in the context of pancreatic cancer, this study highlights the complexity of ERBB receptor signaling and cellular responses to targeted therapeutics, which are increasingly difficult to explain with established paradigms of ERBB receptor signaling. Although all four human ERBB receptors can form heterodimers, the prevailing subclassification of EGFR, on the one hand and ERBB3 and ERBB4, on the other hand, reflects the distinct nature of their ligand-dependent activation. ERBB3 (ErbB3, HER3) function has long been interpreted primarily in the context of its heterodimerization with ERBB2 (ErbB2, HER2).2 ERBB2, on the other hand, has been recognized as a universal and catalytically potent signal amplifier in all heterodimerization events across this ligand-based divide of ERBB members. Signaling through the EGF-activated EGFR/ERBB2 heterodimer has increasingly attracted attention as a key player in tumor progression in several cancers, notably in NSCLC where kinase domain mutants of ERBB2 can drive constitutive activation of both ERBB2 and EGFR.3 However, while ERBB3/EGFR heterodimerization, upon binding of EGF to EGFR or neuregulin to ERBB3, can occur and trigger PI3K signaling,4 ERBB3 has so far largely being looked upon as the primary enabler of the potent oncogenic potential of ERBB2.

Our view of ERBB3 and its relevance in tumorigenesis, however, is rapidly changing. ERBB3 plays a key role in the cellular response induced by stress and radiation;5 and ERBB3 confers and predicts resistance to the radiosensitization induced by HSP90 inhibitors,6 a potential approach to the treatment of ERBB2 overexpressing cancers. Moreover, ERBB3 is a key player in cellular resistance to EGFR and ERBB2 directed kinase inhibitor therapy. Central to this contribution, which appears more easily reconcilable with classic models, is the presence of six binding sites for the regulatory subunit p85 of phosphoinositide-3-kinase. This makes ERBB3 one of the most potent known activators of PI3K/AKT signaling in a manner that is distinct from activated EGFR homodimers in terms of both potency and mechanism.7 The contribution of ERBB3 to enhanced resistance appears to relate to the ability of cancer cells to recover phosphorylation of ERBB3 in the face of sustained inhibition of EGFR and ERBB2. One current model emphasizes the enhanced utilization of low residual activation of EGFR or ERBB2, resulting from limitations to complete inhibition by limiting toxicity and bioavailability of current inhibitors combined with extended half lives of pERBB3.8 A second model for the recovery of activated ERBB3 involves the formation of “unconventional” receptor interactions with MET/HGFR,9 which mechanistically defies all present models of extracellular domain driven and controlled receptor interactions for ERBB receptors.

In addition, the question of whether ERBB3 itself is merely a catalytically deficient and silent partner in signaling events has recently been revisited.10 Recent crystal structures show the kinase domain of ERBB3 bound to a non-hydrolyzable analog of ATP.10,13 While the structure of the kinase domain represents that of a kinase in the inactive state relative to models of allosteric cross activation that have emerged from EGFR,14 recombinant kinase domains from ERBB3 clearly show intrinsic kinase activity that may follow a noncanonical route of phosphoryl transfer.10 The extent to which this low level kinase activity (approx. 1/1,000 of the isolated EGFR kinase domain under similar test conditions) may play a more prominent role when channeled in the context of a spatially restrictive receptor complex remains to be seen. The newly described mechanism of phosphorylation by ERBB3 is insensitive to the existing kinase inhibitors that inhibit most of the phosphorylation of ERBB2 in ERBB2/ERBB3 heterodimers.10 This suggests that it is unlikely to account for the bulk of the observed and mechanistically unexplained phosphorylation of ERBB3's heterodimerization partners unless the in vitro assay conditions underestimate the potency of these inhibitors on ERBB3. The latter is possible given similar discrepancies for ERBB2 between in vitro and cell based assays (see below). However, based on these recent findings, the role of ERBB3 as a mere substrate of trans-phosphorylation certainly needs to be questioned, making existing signaling models significantly more complex.

In light of these changes in our understanding of ERBB3 contributions to signaling, the findings by Liles et al. present a very timely contribution that provides insight into the role of ERBB3 in the context of pancreatic cancer. Thus far, pancreatic cancers have shown little response to targeted treatments in a clinical setting. The 5-year survival rate for pancreatic adenocarcinomas remains below 5% and palliative chemotherapy often remains the primary form of treatment. A significant portion of pancreatic adenocarcinomas do, for example, show EGFR overexpression, but responsiveness to EGFR targeted therapies is poor. It is important to note that the poor response to therapies may not reside exclusively on a cellular and molecular level. A critical feature of pancreatic cancer appears to be its ability to exist under conditions of tight encapsulation by surrounding stroma cells and limited blood supply, thereby restricting access for therapeutics.15 This may also explain why the low molecular weight kinase inhibitor, Erlotinib, has at least a modest impact on life expectancy in pancreatic cancer while the monoclonal antibody based anti-EGFR therapy, Cetuximab, has so far failed.16 However, while keeping in mind these significant limitations to established cell culture and murine model systems, the question remains whether pancreatic cancers have unique molecular characteristics that render targeted therapy largely unsuccessful or whether pancreatic cancers share more key characteristics of resistance with other forms of solid tumors. In this context, the role of ERBB3 in enhancing the Erlotinib sensitivity of cells appears to run counter to above contributions by ERBB3 in the emergence of resistance to kinase inhibitor therapy, yet it mirrors similar paradoxical findings in other cancers, which relate to the role of ERBB2.

The work by Liles et al. starts with the earlier observation by the same group that pancreatic cancer cell lines that are initially sensitive to Erlotinib become resistant upon transient knock down of ERBB3.17 This provided the impetus to use isogenic PANC1 cells with and without ERBB3 expression to directly evaluate the contribution of ERBB3 to signaling, cell proliferation and ultimately tumor growth in a subcutaneous murine xenograft model; all of which were indeed enhanced by ERBB3. Furthermore, this study demonstrated that ERBB3 expression conferred sensitivity of EGFR expressing PANC1 cells to Erlotinib. Consistent with the anticipated enhanced contribution of ERBB3 to PI3K/AKT signaling, the enhanced Erlotinib sensitivity upon ERBB3 expression is evident primarily with respect to PI3K/AKT signaling but not MAPK signaling. Hence, the study makes a strong case for the tumor promoting capability of ERBB3 in a pancreatic cancer model system. However, this is also when the story becomes more complicated and difficult with simple paradigms of receptor interactions and signaling. While the inhibition of AKT phosphorylation by Erlotinib and the enhancement of tumor growth nicely correlate with the extent to which ERBB3 expression was retained in individual xenografts, no increase in the base activation of AKT was observed in the absence of Erlotinib, compared to ERBB3 negative controls. This suggests that the addition of ERBB3 fundamentally changed the nature of activation not necessarily the level of activation.

Difficulties in the mechanistic interpretation may be a reflection of the intrinsic complexity of ERBB receptor cross activation and prevalent assumptions that are built into dimerization models of receptor interactions, activation kinetics and inhibitor specificity. For example, the basic PANC1 model system is void of detectable ERBB3, but PANC1 cells contain significant levels of ERBB2. Despite a 40- to 3,000-fold preference for EGFR in in vitro assays,18–20 Erlotinib inhibits ERBB2/ERBB3 signaling and growth proliferation in EGFR-free model systems at the 150–720 nM range, depending on the readout.21 This represents a concentration range that is well below the upper tolerated dose of Erlotinib that is estimated to translate into a serum concentration of about 10 µM,22,23 the concentration used in most Erlotinib studies including the present study on PANC1 cells. Interestingly, ERBB2 confers a similarly paradoxical enhancement in sensitivity to Gefitinib (Iressa) in several epithelial cancer lines in a manner that correlates with additional ERBB2 expression,24–26 and clinical data indicate that EGFR overexpressing NSCLC show increased sensitivity to Gefitinib with increasing levels of ERBB2.27 Even more surprisingly, NSCLC cells show an increased sensitivity to Erlotinib in EGFR positive and ERBB2 overexpressing NSCLC that is independent of the presence or absence of the EGFR directed monoclonal, Cetuximab.21 At present, the mechanistic foundation for the observed discrepancies in apparent inhibitor specificity, the nature of the sensitivity enhancing effect of ERBB2, and the better than expected impact of Erlotinib on tumor cell growth in select model systems is not clear.

On the mechanistic end, the study by Liles et al. contains both data that are supportive of a significant role of EGFR/ERBB3 heterodimers, as well as data that are more difficult to reconcile with this particular model. The reported ability of ERBB3 to become phosphorylated by apparent crosstalk with receptors such as MetR by mechanisms that completely defy current models for extracellular domain mediated and controlled dimerization, the remaining inability of current dimerization models to explain the phosphorylation of either EGFR or ERBB2 in heterodimers with ERBB3, and studies suggesting that signaling by EGFR may in fact involve tetrameric states,28 suggest that we currently understand the nature of ERBB receptor interactions much less that prevailing clean dimerization models makes us believe. Hence, it may be prudent to base possible modes of intervention not solely on the assumption of EGFR/ERBB3 heterodimers as the main functional unit that conveys the observed tumor promoting and Erlotinib sensitivity conferring effects of ERBB3.

The data make a strong case for the potential potency of ERBB3 in pancreatic tumor progression and, as pointed out in the study, follow up studies will have to evaluate the incidence of elevated ERBB3 levels in patient samples. Given the complexity of ERBB signaling, a direct targeting of ERBB3 signaling may indeed be prudent if its role in clinical samples is confirmed. This will, of course, bring up the previously mentioned dilemma of targeted therapies for pancreatic cancers beyond murine model systems. Assuming that intrinsic and underestimated catalytic activity of ERBB3 is unlikely to be the major driving force in the observed tumor promoting properties of ERBB3, (which may have to be confirmed and would otherwise call for a new class of kinase inhibitors), and assuming that ERBB3 acts primarily as a potent recruitment agent, primarily for PI3K/AKT signaling, this leaves us at present with two readily testable options. We can either directly target its interaction partners or downstream targets (with the obvious danger of potentially incorrect assumptions on relevant interactions and pathways) or we can target ERBB3 directly in its extracellular domains by macromolecular therapeutics, such as monoclonals that are in various stages of development and evaluation. The later may be challenging in pancreatic cancer if the relative underperformance of Cetuximab compared to the at least modest response to Erlotinib was indeed related to issues of tumor penetration and not a reflection of efficacy of Erlotinib against non-EGFR targets. In any case, the work presented by Liles et al. underscores the potency of ERBB3 and its emerging role as a significant player in its own right.1 Furthermore, any mechanistic insight we can gain into the dismal response rates that current targeted therapies achieve in pancreatic cancer is welcome, and the work provided in the study by Liles et al.1 provides an interesting new piece in this puzzle, maybe especially in those aspects that do not readily reconcile with current prevailing models.

Acknowledgements

This work was supported by funding from the National Institutes of Health (R.L., CA098881).

Commentary to: