Targeting HER (ERBB) Signaling in Head and Neck Cancer: An Essential Update

Abstract

HNC (head and neck cancer) remains the 6^th most common carcinoma worldwide. The suboptimal survival and toxicities observed with conventional approaches warrant exploration of novel therapeutic strategies such as targeted therapies. Although targeting EGFR (epidermal growth factor receptor) with cetuximab demonstrated clinical promise, HER (human epidermal growth factor receptor) or ERBB (erythroblastic leukemia viral oncogene homolog) targeted therapy in HNC has overall been suboptimal to date in clinical settings. Overcoming the resistance as well as identifying new strategies therefore remains a significant challenge. In this review, we will discuss the emerging roles of HER members besides EGFR. A comprehensive “three-dimensional” view of HER signaling pathway from the importance of EGFR nuclear translocation to our maturing concept of receptors’ “spatial regulation”, as well as the interdependence and interaction among different HER members will also be addressed to complete an essential update of HER signaling in HNC.

1. Introduction:

In 2015, it is estimated that about 59,340 new cases of oral cavity, pharyngeal, and laryngeal cancers will occur, accounting for about 3.6% of new cancer cases in the United States (Siegel et al., 2015). During the same time period, ~12,290 deaths from HNC are expected (Siegel et al., 2015). SCCHN (squamous cell carcinoma of the head and neck) is the histologic type in more than 90% of these tumors (Grégoire et al., 2010). Despite advances in multimodality therapies, survival rates and functional outcomes require further improvement – this is especially true for HNC of advanced stages (Siegel et al., 2015). New strategies such as targeted therapies are thus urgently needed.

Cetuximab, a monoclonal antibody that targets EGFR signaling, was the first such agent with proven clinical activity in SCCHN (Bonner et al., 2006). However, either intrinsic or acquired resistance to cetuximab renders its clinical activity only modest (Wheeler et al., 2010), among which (details below), alterations specifically involving HER signalings such as up-regulation of EGFR receptor and its ligands, activating mutations of EGFR, signaling through other HER members such as HER3, as well as nuclear translocation of EGFR, etc. are important mechanisms (Wheeler et al., 2010). These findings suggest the plasticity of HER signaling can be “hijacked” by cancer cells to confer cetuximab resistance. It is therefore imperative to revisit HER signaling, review the newest findings regarding its family members and develop new strategies accordingly to better target HNC.

2. HER signaling, EGFR and its role in HNC:

2.1. HER signaling at a glance:

The molecular target of cetuximab, EGFR, belongs to the HER family which includes three other closely related type 1 transmembrane RTK (receptor tyrosine kinases): HER2 (ErbB2), HER3 (ErbB3) and HER4 (ErbB4) (Baselga and Swain, 2009; Yarden and Sliwkowski, 2001). By either homo- or hetero-dimerization and consequent transphosphorylation at the intracellular c-terminus, HER receptors relay extracellular growth signals intracellularly through activating a complex but tightly controlled array of signaling pathways (Citri et al., 2003), which are critical for the regulation of various aspects of cellular functions including cell proliferation, differentiation, migration, apoptosis, etc (Citri and Yarden, 2006). It is, therefore, not surprising that dysregulation of HER signaling is crucial for the initiation, maintenance and progression of many epithelial cancers (Baselga and Swain, 2009; Citri et al., 2003; Yarden and Sliwkowski, 2001), and targeting HER signaling is a rational strategy in eliminating cancer cells relying on this oncogenic signaling pathway (Ather et al., 2013; Huang et al., 2013; McDonagh et al., 2012).

Although all four members of the HER family have the same essential domains, there are variations in functional activity (Cho et al., 2003; Garrett et al., 2003; Jura et al., 2009; Ogiso et al., 2002), which provide each individual member unique properties (Table 1). Briefly, all except HER2 have known ligands; all except HER3 have considerable intrinsic tyrosine kinase activity; and HER2 is constitutively available for dimerization due to its resting “active” conformation. These characteristics make HER2 a favorable and HER3 an obligate hetero-dimerization partner (Cho et al., 2003; Garrett et al., 2003; Jura et al., 2009; Yarden and Sliwkowski, 2001). In addition, although all members activate Erk 1/2 (extracellular signal-regulated kinase) via the recruitment of Grb2 (growth factor receptor bound 2) or Shc (Src homology) adaptors (Yarden and Sliwkowski, 2001), PI3K (phosphoinositide 3-kinase) is more readily activated through HER3 and HER4 due to their ability to directly bind the p85 regulatory subunit of PI3K (Hynes and Lane, 2005; Yarden and Sliwkowski, 2001). Together, these features ensure the plasticity of HER signaling through the interdependence and interaction among different members, which results in various biological consequences.

Table 1 |.

The HER family: A comparison among individual members

	EGFR (HER1, ErbB1)	HER2 (ErbB2)	HER3 (ErbB3)	HER4 (ErbB4)
Ligand	EGF, HB-EGF, TGF-α, amphiregulin, betacellulin, epiregulin, etc.	None identified, Not required for activation	NRG1, NRG2	NRG1–4, HB-EGF, betacellulin, epiregulin
Kinase activity	Yes	Yes	None to minimal	Yes
Dimerization	Homo, hetero-	Homo, hetero-, favorable	Hetero-, mandatory	Homo, hetero-
Effectors & adaptors	Ras > PI3’K No p85 site.	Ras > PI3’K No p85 site.	PI3’K >> Ras 6x p85 sites.	Ras ~ PI3’K 1x p85 site.
Genetic changes reported in the literature and COSMIC database	Protein over-expression ~70–90%; mutations: ~ 78/3079=2.12%*	Protein over-expression variable; mutations: ~ 16/1958=0.82%*	Protein over-expression ~8–9%(Takikita et al., 2011); mutations: ~ 6/1184=0.51%*	Protein over-expression: limited studies; mutations: ~ 22/1112=1.98%*
Mutation rate of 279 sequenced SCCHN on TCGA database	13/279=4.66%	5/279=1.79%	8/279=2.87%	14/279=5.02%
FDA approved targeted Rx in HNC	Cetuximab	None yet	None yet	None yet

Rx: therapy; Ras>PI3’K: activates Ras signaling more easily than PI3’K.

^*:based on COSMIC database. The tissue labeled as “upper aerodigestive tract” was used in the calculation. These data also included non-squamous cell carcinoma such as adenoid cystic carcinoma and other cancers with non-specified histology. These data were listed in parallel to the results extracted from 279 sequenced SCCHN on the TCGA database (the study name is “Head and Neck Squamous Cell Carcinoma (TCGA, in revision)”, the same study that was recently published in Nature, 2015 Jan 29;517(7536):576–82. (Network, 2015)).

2.2. EGFR and its role in HNC:

In HNC, especially in SCCHN, EGFR seems to be the most dominant player in HER signaling among its family members (Leemans et al., 2011). Enhanced EGFR signaling, which occurs in the majority of cases by EGFR overexpression, and with lower incidence by mutation at its extracellular domain such as EGFRvIII or intracellular tyrosine kinase domain (Leemans et al., 2011), may constitutively activate four major signaling pathways including Ras/Raf/MEK/MAPK, PI3K/AKT/mTOR, PLCγ/PKC and the JAK/STAT pathways (Leemans et al., 2011; Yarden and Sliwkowski, 2001) with various amplitudes to promote unconstrained cellular proliferation and survival, enhanced invasion and angiogenesis, and evasion of apoptosis, etc., which are all the “hallmarks of cancer” (Hanahan and Weinberg, 2000, 2011).

Two major strategies are currently used to target EGFR in the preclinical and clinical settings. One is to use mAbs (monoclonal antibodies) such as cetuximab to block receptor-ligand binding and prevent EGFR dimerization (Bonner et al., 2006). Another is to use TKIs (tyrosine kinase inhibitors) such as erlotinib to abolish EGFR signaling through various mechanisms, among which competing with ATP (adenosine triphosphate) for the ATP-binding site within the intracellular tyrosine kinase domain of the EGFR is prominent (Posner et al., 1994). To date, cetuximab has proven to be effective in multiple clinical settings: for example, Bonner et al. have shown the addition of cetuximab significantly improved loco-regional control and median overall survival in patients treated with RT (radiation therapy) and cetuximab compared to RT alone (Bonner et al., 2006); and Vermorken et al. demonstrated in the phase III EXTREME trial (Vermorken et al., 2008) that for patients with R/M (recurrent/metastatic) SCCHN, cetuximab plus platinum-based chemotherapy doublet (cisplatin/5-FU or carboplatin/5-FU) improved median survival and response rate when compared to chemotherapy alone (Vermorken et al., 2008). Other EGFR mAb such as panitumumab have also demonstrated therapeutic benefit under similar situations (e.g. the SPECTRUM trial) (Vermorken et al., 2013).

Current available data have unfortunately not established TKIs as treatment options for HNC outside the clinical trial setting. However, we should not negate the value of targeting the kinase domain of EGFR as a therapeutic strategy. Rather, these data suggest that targeting EGFR alone is probably not sufficient, especially in a ligand rich environment of advanced disease, and that patient population selection is important (Harrington et al., 2013; Wilson et al., 2011). For example, although lapatinib, a dual TKI of both EGFR and HER2, was not effective either as a monotherapy in R/M SCCHN (de Souza et al., 2012), or as an adjuvant therapy for postsurgical patients with high risk (Harrington et al.), it did demonstrate promise in HPV (human papillomavirus) negative treatment naïve patients with unresectable stage III/IVA-B SCCHN (Harrington et al., 2013) – based on which, lapatinib is currently being evaluated in HPV negative SCCHN in combination with CCRT (concurrent chemoradiation therapy) (RTOG 3501). Similarly, afatinib, a TKI for pan-HER members, was recently shown in a phase II study to be able to achieve comparable disease control rate (50%) to cetuximab (56.5%) in R/M SCCHN whose disease has progressed after platinum-containing therapy (Seiwert et al., 2014). In addition, afatinib was superior to methotrexate as the second-line therapy for R/M SCCHN in the recent LUX-Head&Neck 1 phase III clinical trial (NCT01345682) (Machiels et al., 2015), which showed longer progression-free survival in patients treated with afatinib (median 2.6 months [95% CI 2.0–2.7]) than methotrexate (median 1.7 months [95% CI 1.5–2.4]), with hazard ratio 0.80 [95% CI 0.65–0.98], p=0.030) (Machiels et al., 2015).

2.3. Mechanisms of resistance to EGFR targeted therapy:

While EGFR targeted therapy especially cetuximab did achieve certain therapeutic benefit, it is overall only modest due to various intrinsic and extrinsic mechanisms leading to resistance (Burtness et al., 2013; Wheeler et al., 2010). Several review papers have already discussed the underlying reasons that have been experimentally verified (Burtness et al., 2013; Wheeler et al., 2010). Specific changes in the HER signaling pathways include up-regulation of EGFR receptor (Wheeler et al., 2008) and its ligands such as TGFα (Rajput et al., 2007), activating mutations of EGFR such as EGFRvIII (Batra et al., 1995; Sok et al., 2006), signaling through other HER members such as HER3 (Wheeler et al., 2008), persistent downstream activation through mutations of oncogenes such as HRAS (Rampias et al., 2014; Wang et al., 2014) and PIK3CA (Wang et al., 2014), as well as nuclear translocation of EGFR (Li et al., 2009), etc. These observations make investigating other members of HER signaling important, which is discussed in details below. In addition, signaling through other RTKs especially c-MET (Stabile et al., 2013) and VEGFR (Bianco et al., 2008) was also found to confer resistance. Other genetic and/or epigenetic changes that promote or maintain malignant phenotype of HNC were also considered as potential contributors to resistance to EGFR targeted therapy. These include p53 mutation/loss (discussed below), over-expression of aurora kinase (Hoellein et al., 2011) and cyclin D1 (Kalish et al., 2004), as well as epithelial-to-mesenchymal transition (Frederick et al., 2007; Holz et al., 2011).

2.4. HPV and EGFR:

A key question remaining in the era of anti-EGFR therapy is whether the HPV/p16 status alters or predicts the therapeutic effect. As a double-stranded DNA virus, HPV expresses oncogenic proteins including E5, E6, and E7. Although, in laboratory settings, E5 has been suggested to enhance EGFR signaling through the inhibition of its internalization, degradation and endosome fusion (Suprynowicz et al., 2010), it is not clear if E6 and E7 also affect EGFR signaling directly. Interestingly, in a recent study, Pogorzeiski et al. (Pogorzelski et al., 2014) found that HPV status does not impact cetuximab response in both in vitro and in vivo models of SCCHN (Pogorzelski et al., 2014). Similarly, in the retrospective analysis of the EXTREME trial (Vermorken et al., 2008), the authors found that the “survival benefits of chemotherapy plus cetuximab over chemotherapy alone are independent of tumor p16 and HPV status” (Vermorken et al., 2014). A recent retrospective analysis based on “Bonner’s trial” (Bonner et al., 2006) by Rosenthal et al. (David Ira Rosenthal et al., 2014) demonstrated that, regardless of p16 status, patients with LASCCHN (locally advanced SCCHN) benefit more from combined cetuximab and radiation compared with RT alone, although a more pronounced effect was observed in p16+ patients (David Ira Rosenthal et al., 2014). However, post-hoc analysis from both the SPECTRUM (Vermorken et al., 2013) (with panitumumab) and LUX-Head&Neck 1 (Machiels et al., 2014) (with afatinib) studies suggested more prominent effect on HPV/p16 negative patients. Further studies are therefore needed to understand whether the observed difference is due to different HER targeted therapy, patient selection, or follow up duration, etc. In addition, the ongoing clinical trial RTOG-1016 (NCT01302834) will address whether cetuximab can completely substitute cisplatin in the setting of concurrent RT for patients with HPV positive oropharyngeal cancer.

2.5. p53 status and EGFR targeted therapy:

p53 mutation is one of the earliest and most frequent genetic alterations in SCCHN with reported frequency of ~50–80% (Agrawal et al., 2011; Poeta et al., 2007). In fact, SCCHN appears to be the most common p53 mutation-carrying cancer type after ovarian cancer and lung squamous cell carcinoma (Kandoth et al., 2013). Although wild type p53 is a classical tumor suppressor, its mutation does not simply mean “loosing” the tumor suppressing function (Brosh and Rotter, 2009). Indeed, mutant p53 is also found having various “gain-of-function” properties (Brosh and Rotter, 2009). For an example, mutant p53 may activate NF-KB pathways (Weisz et al., 2007a); inactivate wild type p53 (de Vries et al., 2002), p63 (Gaiddon et al., 2001) and p73 (Di Como et al., 1999; Gaiddon et al., 2001); cooperate with RAS in transformation (Lang et al., 2004); and interact with different transcription factors (Weisz et al., 2007b), thereby ultimately help cells acquire the majority if not all of the cancer hallmarks (Hanahan and Weinberg, 2000, 2011). It is therefore not surprising to see that p53 mutation correlates with poor clinical outcome in various types of cancer including head neck cancers (Brosh and Rotter, 2009).

Since p53 mutation is found to correlate with certain treatment failure and poor prognosis (Koch et al., 1996; Poeta et al., 2007), similar question was asked if p53 status determines the sensitivity to EGFR targeted therapy. While in general, there is insufficient data to unequivocally establish the association between p53 mutation and sensitivity to EGFR targeted therapy in HNC, using acquired cetuximab and erlotinib resistant lung and SCCHN cell lines, Huang et al. observed consistently reduced level of p53 comparing to their parental cell lines (Huang et al., 2011). They further demonstrated that silencing of p53 conferred resistance to cetuximab, while restoring wild type p53 function enhanced cetuximab sensitivity through p-ERK down-regulation and apoptosis induction (Huang et al., 2011). These observations are consistent with previous studies showing that cetuximab inhibits the growth of hepatocellular cancer cells with wild type p53, but not mutated p53 (Huether et al., 2005).

However, using tumor specimens from 64 metastatic colorectal cancer patients treated with cetuximab-based chemotherapy, Oden-Gangloff et al. found an association between p53 mutations and improved clinical outcome, particularly in patients without KRAS mutations (Oden-Gangloff et al., 2009), suggesting the study of p53 mutation is likely more complicated than simply p53 loss since various mutations may have distinct functions from simple dominant-negative variants. Same principle applies to HNC. Therefore, it remains important to investigate the impact of different p53 mutations in relation to EGFR targeted therapy. Similar retrospective analysis using samples from Bonner’s trial (Bonner et al., 2006) or EXTREME trial (Vermorken et al., 2008) may offer useful insights.

3. HER3: the unpretentious member is now an essential player

3.1. The introduction of HER3:

Historically, the initial interest in HER3 was due to the fact that it is a critical activator of PI3K signaling in EGFR, HER2 or MET addicted cancers (Sergina et al., 2007), and reactivation of HER3 is a prominent mechanism by which cancers become resistant to HER inhibitors (Sergina et al., 2007). Although EGFR and HER2 can activate PI3K through the adaptor proteins Grb2 and GAB1 (Grb2-associated binding protein 1), HER3 contains six YXXM motifs that bind the p85 subunit of PI3K directly, making it the most preferred dimerization partner when signaling occurs through the PI3K pathway (Yarden and Sliwkowski, 2001). In fact, several studies have observed that, in those cancers that are sensitive to EGFR or HER2 inhibitors, HER3 serves as the major mediator of PI3K/AKT activation, and the activity of HER3 is almost completely under the control of EGFR or HER2, respectively (Engelman et al., 2005; Yakes et al., 2002). For example, in gefitinib sensitive NSCLC cells, inhibition of EGFR abrogated HER3 phosphorylation, which in turn resulted in the loss of PI3K/AKT signaling (Engelman et al., 2005). Similarly, trastuzumab treatment leads to the loss of HER3 phosphorylation and reduced activity of PI3K and AKT in those sensitive breast cancer cells (Yakes et al., 2002). However, when there is reactivation or persistent activation of HER3 signaling, either through hetero-dimerization with EGFR, HER2 or even members of other RTK families such as MET, resistance to EGFR or HER2 inhibitors is observed (Sergina et al., 2007).

3.2. What do we know about HER3 in HNC?

Today, the critical role of HER3 has been confirmed in cancers of various tissue types including HNC. For example, in an attempt to identify molecular markers that may predict response to gefitinib in SCCHN, Erjala et al. found the expression level of HER3 was associated with resistance to gefitinib (Erjala et al., 2006). In order to identify the mechanisms of resistance to cetuximab in SCCHN, Wheeler et al. established resistant clones in vitro (Wheeler et al., 2008), and found HER3 along with EGFR, HER2 and MET were highly activated in resistant cells. There were also increased levels of heterodimerization of EGFR:HER3 and HER2:HER3 (Wheeler et al., 2008) (Figure 1, activation mode a and b). When combining cetuximab with pertuzumab (which binds to HER2 and inhibits the dimerization of HER2 with other HER receptors including HER3), a potent suppression of HER3 and AKT phosphorylation was observed. By knocking down HER3 with siRNA, the authors re-sensitized resistant cells to cetuximab (Wheeler et al., 2008). Together, these findings suggest that HER3 signaling represents a critical mediator in the resistance to HER inhibitors.

Figure 1 |. The activation modes of HER3 and potential strategies of targeting:

HER3 can be activated through the following modes: a: EGFR:HER3 hetero-dimerization (HER3 can be ligand independent here); b: HER2:HER3 hetero-dimerization; c: HER4:HER3 heterodimerization; d: HER3 heterodimerization with other RTK members such as c-MET particularly in HNC; e: autocrine loop activation by NRG1. This is ligand dependent activation. HER3 will then heterodimerize with other HER or RTK members; f: activation by mutations in either the extracellular or kinase domain. This is however only putative in HNC without functional studies (Stransky et al., 2011). Also shown here are the action mechanisms of different inhibitors: HER3 mAbs such as MM-121 directly target HER3; pertuzumab blocks HER3:HER2 heterodimerization (b), while cetuximab and MM-141 block the activation modes a and d, respectively; TKIs inhibit HER3 by targeting the kinase activity of its dimerization partners. The direct targeting of NRG1 synthesis and/or degradation and HER3 activating mutations are theoretically potential strategies but are not currently available for HNC – thus are labeled with “?”.

Interestingly, while profiling the sensitivity of 690 cancer cell lines to investigational TKIs, Wilson et al. (Wilson et al., 2011) identified a subset of non-HER2 amplified cancer cells, particularly from HNC, that demonstrated striking sensitivity to lapatinib – a dual EGFR and HER2 TKI. Treatment with lapatinib, but not erlotinib, strongly suppressed phosphorylated HER3 as well as downstream AKT activation. This suggested that HER2 kinase but not EGFR kinase activity drove the downstream survival signals through HER3 in the lapatinib-sensitive cells. Subsequent experiments demonstrated elevated NRG1 (neuregulin-1) expression and these cells were found to depend on a NRG1-mediated autocrine loop driving HER3 activation after HER2:HER3 hetero-dimerization (Figure 1, activation mode e), which can be disrupted by lapatinib (Wilson et al., 2011).

Since elevated NRG1 expression and autocrine activated HER3 are strongly associated with lapatinib sensitivity, and both NRG1 and HER3 were enriched in a subset of HNC, a follow up study was performed by Shames et al. to investigate whether NRG1 and HER3 can indeed serve as actionable biomarkers in SCCHN (Shames et al., 2013). By profiling the expression of NRG1 and HER3 in more than 750 individual tumor specimens, they found that a significant subset (~ 40%) of SCCHNs expressed much higher levels of NRG1 along with activated HER3 than any other tumor types (Shames et al., 2013). The expression level of NRG1, therefore, may have the potential to define a distinguishable subset of SCCHN patients that might benefit from anti-HER3 therapy (Shames et al., 2013). Interestingly, when using dual-color chromogenic RNA in situ hybridization assay to evaluate HER3 and NRG1 mRNA location and abundance in clinically relevant tissues, Shames et al. observed a pattern shift from the paracrine expression of NRG1 and HER3 in normal tissue and well-differentiated SCCHNs towards more autocrine expression in poorly differentiated and aggressive SCCHNs, consistent with the more disordered architecture (Shames et al., 2013). This observation is reminiscent of an earlier study performed by Vermeer et al. who demonstrated that when the polarity of airway epithelial cells was disrupted, NRG1 in the apical lumen gained access to its receptors HER3 and HER4 which were normally segregated at the basolateral side, and consequently autocrine signaling was initiated (Vermeer et al., 2003) – a good example of how disrupted spatial regulation of HER receptors may lead to malignancy.

The observation of HER3 activation by NRG1 mediated autocrine signaling is important. It suggests that HER3 activation does not necessarily rely on the over-expression or mutational activation of HER3 or its dimerization partners. Therefore, through various mechanisms, cancer cells can produce NRG1 or related ligands that result in autocrine activation of HER3 signaling for survival and proliferation. Inhibition of NRG1 therefore could be a potential therapeutic strategy and autocrine production of similar receptor ligands may have predictive value in treatment.

3.3. Targeting HER3 as a therapeutic intervention:

As shown in Figure 1 and Table 2, two major strategies are utilized to target HER3: one is to apply monoclonal antibodies directed against the HER3 extracellular domain to block ligand binding, and also prevent its dimerization with other HER proteins through steric hindrance. Among the currently available mAbs, some target HER3 alone, such as MM-121 (Schoeberl et al., 2010), U3-1287 (LoRusso et al., 2013) and RG7116 (Mirschberger et al., 2013); some simultaneously target other HER members as well, such as MM-111 (McDonagh et al., 2012) and MEHD7945A (Huang et al., 2013) (Table 2). The second approach is using TKIs to target the dimerization partners of HER3 since HER3 itself lacks substantial intrinsic kinase activity. It seems that targeting the HER2 kinase domain is essential to achieve considerable effect, likely because HER2 is the preferred dimerization partner and HER2:HER3 is the strongest oncogenic driver in HER signaling (Sergina et al., 2007; Yarden and Sliwkowski, 2001). Examples in this category include lapatinib, which targets HER2 more than EGFR (Wilson et al., 2011), AZD8931, which targets HER2 and EGFR roughly equally (Tjulandin et al., 2014), and dacomitinib (Ather et al., 2013), which is a pan-HER irreversible TKI targeting EGFR, HER2 and HER4 (Table 2). In preclinical settings, all of these agents have shown promising results, and some are now in phase II or III clinical trials (Table 2).

Table 2 |.

Currently available HER3 targeting agents with their stages of development

Class	Agents	Targets	Comments	Clinical trials	Refs
Monoclonal Antibody (humanized)	MM-121	HER3	L+, d/HER2, d3	Phase I, II (SCCHN, NSCLC, breast, colon)	(Schoeberl et al.,2010)
	U3-1287	HER3	L+, d/HER2, d3	Phase I, II (NSCLC, breast, solid tumors)	(LoRusso et al.,2013)
	RG7116	HER3	L+, d/HER2, d1	Phase I (solid tumors)x	(Mirschberger et al.,2013)
	GE-huMab-HER3	HER3	L+, d/HER2, +ADCC, d3	Phase I (NSCLC)x	(Bossenmaier et al., 2012)
	AV-203	HER3	L+, d/HER2, d3	Phase I (solid tumors)	(Vincent et al., 2012)
	LMJ716	HER3	L+/−, d/HER2, d2&4	Phase I (SCCHN, breast, gastric)	(Garner et al., 2012)
	MM-111	HER3, HER2	L+, scFv	Phase I (HER2 amplified tumors)	(McDonagh et al., 2012)
	MEHD7945A	HER3, EGFR	L+/−, +ADCC, d3	Phase I, II (SCCHN, colorectal)	(Fayette et al., 2014; Huang et al., 2013)
	MM-141	HER3, IGF1R		Phase I (solid tumors)	(Fitzgerald et al., 2014)
	Pertuzumab	HER2	L+, d/HER3, d2	Phase I-III (breast, etc)	(Erjala et al., 2006; Sak et al.,2013)
TKI of HER3 hetero-dimerization partners	AZD8931	HER2 = EGFR		Phase I, II (breast, GI)	(Tjulandin et al., 2014)
	Lapatinib	HER2 > EGFR		Phase 0-III (SCCHN & various others)	(Wilson et al., 2011)
	Afatinib	HER 1, 2, 4		Phase I-III (SCCHN& various others)	(Cupissol et al.,2013)
	Dacomitinib	HER 1, 2, 4		Phase I-III (SCCHN& various others)	(Ather et al., 2013)

For monoclonal antibodies, only the humanized agents are listed here. Note that mAb can be developed to target different domains (d1-4) of HER3 ECD. The mAb can be mono- or dual/multi- specific. Most of them work for ligand dependent activation (L+); few also work for ligand independent situation (L−). A few in addition have ADCC effect (+ADCC). The majority of mAbs affect heterodimerization with HER2. L+: works best in ligand dependent environment; L−: ligand independent; d/HER2: impairs the dimerization with HER2 mainly; d1-4: designed against domain 1–4 of the extracellular part accordingly;

^X:In Europe; scFv: single chain fragment variable antibody.

However, a recent phase II trial (Fayette et al., 2014) didn’t demonstrate the superiority of MEHD7945A (a dual EGFR and HER3 mAb) over cetuximab in R/M SCCHN patients (Fayette et al., 2014), raising the question why sufficient clinical benefit is not observed based on the promising preclinical data. As pointed out by its original developers (Schaefer et al., 2011), although MEHD7945A was designed to target both EGFR and HER3, it “has a 5- to 8-fold higher affinity for HER3 than for EGFR”, and its affinity to EGFR is only about 1/3 of cetuximab (Schaefer et al., 2011) (IgG K_d: 19.9 vs. 6.03nM, their Table S1) (Schaefer et al., 2011). We therefore speculate that, in cancer cells where EGFR signaling is still predominant, MEHD7945A may not be able to demonstrate superiority over cetuximab. Different response might be observed if we combine a monospecific HER3 inhibitor such as MM-121 with cetuximab (Jiang et al., 2014), or test MEHD7945A in cetuximab resistant SCCHN patients.

Other potential strategies to target ligand dependent HER3 signaling in SCCHN may include inhibiting NRG1 expression (Tsai et al., 2003) and disrupting the autocrine loops by NRG1 neutralizing antibodies (Li et al., 2004) as shown in the studies of breast cancer. In ligand independent HER3 signaling, heterodimerization with MET is worth noting, especially considering MET overexpression is so frequent in SCCHN (Seiwert et al., 2009). In fact, Seiwert et al. showed ~80% of their examined SCCHN tumor tissue had MET over-expression (Seiwert et al., 2009). Treating the SCCHN cell lines with the MET inhibitor SU11274 partially blocked HER3/AKT signaling, which was then completely blocked by the combination with erlotinib (Seiwert et al., 2009). Figure 1 summarizes HER3 activation modes and our potential strategies to interfere in this crucial pathway.

3.4. Final words about HER3

About a decade ago, the absence of kinase activity along with the lack of detectable mutations in cancer samples were the descriptions of HER3 (Sergina et al., 2007; Yarden and Sliwkowski, 2001). This has now changed: a recent study in 2010 reported a low level of kinase activity (Shi et al., 2010); and more interestingly, Jaiswal et al. reported the identification of HER3 somatic mutation in ~11% of colon and gastric cancers, and demonstrated that mutants transformed colonic and breast epithelial cells in a ligand-independent manner (Jaiswal et al., 2013). Although HNC was not included in that particular study, a previous investigation (Stransky et al., 2011) did detect 1 out of 92 cases (~1%) harboring a point mutation (c.466G>A, p.M91I) (their supplemental table 6) (Stransky et al., 2011) at the corresponding location of domain I of HER3 ECD (Jaiswal et al., 2013). In addition, recent TCGA (The Cancer Genome Atlas) study based on 279 sequenced SCCHN identified 8 cases (2.87%) harboring HER3 mutation. It will thus be interesting to conduct relevant functional study of these mutations.

4. Revisiting HER2 and discovering HER4:

4.1. HER2: the member that cannot be neglected

Since HER2 crosstalks extensively with other HER members, and HER2:HER3 dimerization comprises the most potent oncogenic unit that drives HER signaling, it is not surprising that attention has been paid to HER2. In fact, HER2 expression in SCCHN has generated some excitement, probably due to the fact that in breast cancer, HER2 targeted therapy with trastuzumab has shown proven efficacy. Several groups have therefore investigated the HER2 expression pattern in SCCHN (Ali et al., 2010; Khademi et al., 2002; Sardari et al., 2012; Schartinger et al., 2004), hoping that HER2 overexpression may help to identify a subpopulation of SCCHN patients. The percentage of HER2 overexpression is however extremely variable in the literature, ranging from almost 0% to 70% (Ali et al., 2010; Khademi et al., 2002; Sardari et al., 2012; Schartinger et al., 2004). Therefore, whether HER2 expression correlates with tumor grade, lymph node metastasis or survival is still debatable. Several possibilities exist to explain the observed variation, including different IHC (immunohistochemistry) methods, types of antibody used, specific staining areas selected for studies (e.g. membranous vs. cytoplasmic vs. nuclear) and the criteria used to define HER2 overexpression (such as using HercepTest scoring system vs. not), etc. A standardized method is thus needed to further describe HER2 expression in SCCHN.

However, even if there is a genuine lack or low frequency of HER2 overexpression, it is still not a valid argument to deny its potential use as a molecular target in HNC. HER2 overexpression is important for ligand independent activation as observed in HER2 overexpressing breast cancers, however, it is not indispensible to have HER2 overexpression in the ligand dependent activation of HER signaling – when HER2 is only a heterodimerization partner. As we discussed earlier, this has been observed in a subset of non-HER2 amplified SCCHN that can utilize the HER2:HER3 dimerization unit to drive downstream survival signaling through the NRG1 dependent autocrine loop, which can be effectively disrupted by lapatinib (Wilson et al., 2011). Selection of HNC patients for HER2 targeted therapy should therefore seriously consider such a subset of patients utilizing ligand dependent HER signaling by overexpressing ligands such as NRG1 (Shames et al., 2013).

4.2. HER4: an emerging player

Besides HER3, HER4 can also bind to the p85 subunit and activate PI3K directly. In fact, in adult heart tissue where HER3 is absent, activation of HER4 is the only way to initiate the direct activation of the PI3K/AKT pathway by HER ligands (De Keulenaer et al., 2010). In addition, HER4 also contains docking sites for Shc, STAT5 (signal transducers and activators of transcription 5) and Grb2, thus is able to activate multiple downstream signaling pathways. These features, along with the availability of ligands and the presence of intrinsic kinase activity make HER4 also important in cancer biology.

Interestingly, it seems the existence of alternatively spliced isoforms of HER4 either in the juxta-membraneous area (JM) or cytoplasmic region (CYT) can contribute to the different functions (Nielsen et al., 2013; Paatero et al., 2013). For instance, the cytoplasmic isoform CYT2 lacks a 16-amino acid sequence that is unique to CYT1 – this sequence comprises the only direct PI3K interaction site of HER4. This indicates that only CYT1 is able to directly activate PI3K signaling. In accordance, CYT1 was found to be the main contributor of the oncogenic activities of HER4 (Paatero et al., 2013), and directing HER4 splicing towards the CYT2 isoform resulted in decreased AKT activity and cell proliferation (Nielsen et al., 2013).

There are only very few studies of HER4 in HNC. By using gene probe technology in 29 patients with laryngeal squamous cell carcinomas (LSCC), Saglam et al. showed that HER4 expression significantly discriminated between early and late stage LSCC (Saglam et al., 2007). By quantitative IHC in 67 LSCC patients, Bussu et al. demonstrated that the overexpression of HER4 and its nuclear localization were protective and associated with a better prognosis (Bussu et al., 2012). But interestingly, a recent study on 82 oral squamous cell carcinoma patients with lymph node metastasis demonstrated that the co-overexpression of HER4 and EGFR predicts poor clinical outcome (Silva et al., 2014). However, in neither of the cases was there any investigation of HER4 isoforms. Further investigations to address specific HER4 isoforms are therefore needed to clarify the function of HER4 in HNC.

In an effort to address the mechanisms of chemoresistance and relapse in NSCLC, Hegde et al. found NRG1 mediated autocrine HER4 signaling could be a potent driver of tumor growth in certain HER4 overexpressing lung cancers, and inhibition of ligand dependent HER3:HER4 signaling not only significantly increased response to chemotherapy, but also delayed relapse in several NSCLC models (Hegde et al., 2013). Since NRG1 autocrine signaling has also been reported in SCCHN (Wilson et al., 2011), and NRG1 is a ligand for HER4 as well, it will be very interesting to investigate HER4 activity in SCCHN tumors expressing high levels of NRG1 – theoretically, this will be even more relevant when resistant clones develop after HER2/HER3 blockage.

5. From EGFR nuclear translocation to the spatial regulation of HER signaling:

5.1. Nuclear translocation adds another dimension to the HER network

While membrane HER signaling, especially EGFR signaling, has been extensively studied, the process of membrane-to-nucleus translocation and its unique biological functions in cancer has just emerged as a prominent new functional mode in the past decade (Wang and Hung, 2009; Wells and Marti, 2002). The increased nuclear expression of EGFR has now been associated with poorer overall survival and prognosis in various types of cancer including SCCHN (Wang and Hung, 2009; Wells and Marti, 2002), as well as the resistance to radiotherapy, chemotherapy and HER directed therapy such as cetuximab and gefitinib (Wang and Hung, 2009; Wells and Marti, 2002).

Several models have been proposed to describe the nuclear transport of membrane HER proteins, mainly including endosome mediated nuclear translocation and retro-translocation by the endoplasmic reticulum associated trafficking machinery (Wang and Hung, 2009; Wells and Marti, 2002) (Figure 2). Using EGFR as an example, once it enters the nucleus, it can function in at least two ways (Wang and Hung, 2009; Wells and Marti, 2002): 1) as a transcription factor due to its intrinsic transactivation activity: with the help of a DNA binding transcription cofactor, such as STATs, nuclear EGFR may bind to the A/T-rich sequence in the promoters of multiple genes including cyclin D1, iNOS (inducible nitric oxide synthase) and Aurora-A, etc. to regulate their transcription (Wang and Hung, 2009); 2) since EGFR has an intracellular tyrosine kinase domain, it can directly phosphorylate PCNA (the proliferating cell nuclear antigen) and DNA-PK (DNA-dependent protein kinase) (Wang and Hung, 2009) (Figure 2). Increased nuclear EGFR activity can thus promote cell proliferation, enhance DNA repair induced by chemo- or radiotherapy, and consequently confer treatment resistance and induce cancer progression.

Figure 2 |. A comprehensive view of HER signaling and targeting strategies:

HER signaling does not exert its function only on the cell membrane. Instead, it acts in different compartments of the cells: from the plasma membrane to the cytoplasm to the nucleus as shown in panel A, and the integration of its actions in these compartments gives the final biological output. Understanding the regulation of its activity in different levels offers us versatile approaches to target HER signaling in HNC as illustrated in panel B. We may target HER signaling horizontally by blocking the dimerization and transactivation between different HER members and/or their association with other RTKs, as shown in a (panel B); or vertically which includes targeting different signaling nodes as shown in b; targeting different levels of endocytosis/trafficking pathways as shown in c; and nuclear translocation as shown in d. The color scheme of HER receptors is similar to Fig. 1: EGFR: blue; HER2: light green; HER3: red; HER4: brown; MET or other RTKs: grey. The ligand of each HER member is colored accordingly. Shown on the left side is the regulation of endocytosis. Using EGFR as the prototype (blue doublet), depending on the type of ligand, dimerization partner or other factors, after internalization, HER receptors may recycle to the cell surface to continue membranous signaling, be sorted to lysosomes for degradation, or enter the nucleus to modulate transcription or protein phosphorylation (central bottom of the figure). Shown on the right side, depending on the dimerization scheme (drawn here in couplings), HER members may activate different downstream signaling pathways. Please refer to the text for details.

In HNC, for example, Tao et al. showed that in nasopharyngeal carcinoma, the EBV encoded LMP1 (latent membrane protein 1) enhanced the nuclear translocation of EGFR, which subsequently transactivated cyclin D1 and cyclin E to promote cell proliferation (Tao et al., 2005). Radiotherapy, a common treatment modality in HNC, is now well known to be capable of triggering EGFR nuclear entry to activate DNA-PK, which then mediates repair of the double strand DNA break induced by irradiation and subsequently confers radiotherapy resistance (Dittmann et al., 2005a). Although cetuximab reduces radiation-induced nuclear translocation of EGFR and consequently inhibits the activity of DNA-PK in HNC (Dittmann et al., 2005b) – the mechanism that may have contributed to the observed benefit in cetuximab plus RT combination therapy (Bonner et al., 2006) – increased nuclear EGFR localization mediated by SFK (Src family kinase) has been associated with cetuximab resistance (Li et al., 2009). Interestingly, inhibition of SFK by dasatinib blocks the nuclear translocation of EGFR in SCCHN (Li et al., 2010). Thus, nuclear EGFR is now considered as a molecular target in cancer therapy.

Today, all four members of the HER family have been detected in the nucleus (Wang and Hung, 2009). Since the c-terminal regions of HER2 and HER4 have intrinsic transactivation activity and a tyrosine kinase domain, similar to EGFR, they may either regulate gene transcription or phosphorylate substrates in the nucleus (Wang and Hung, 2009). However, the role of nuclear translocation of HER2-4 in HNC needs to be further deciphered.

5.2. The spatial regulation of HER signaling

The shuttling of membranous receptors from the cell surface to the nucleus after endocytosis is considered only one of the many aspects of the spatial regulation of the HER family – and in fact, of the whole RTK superfamily (Casaletto and McClatchey, 2012). By using EGFR as the prototype in the majority of studies, we now know that the regulation of HER signaling is beyond the production of receptors in quantity and quality (e.g. via gene amplification, mRNA overexpression, alternative splicing or gain-of-function mutations (Jaiswal et al., 2013; Leemans et al., 2011; Stransky et al., 2011)), or ligand availability such as in the setting of autocrine stimulation (Shames et al., 2013; Wilson et al., 2011). In fact, HER signaling is also subject to tight spatial control including membrane receptor distribution and recycling, dynamic receptor localization and distinct signaling effects in different cellular compartments (Figure 2A & B) (Casaletto and McClatchey, 2012).

For example, studies have demonstrated the heterogeneous distribution of EGFR and HER2 receptors on the cell membrane (Chung et al., 2010; Nagy et al., 2010), and in areas with abundant receptors, dimers and even higher order clusters may form without ligand (Chung et al., 2010; Nagy et al., 2010), which may contribute to ligand-independent activation. Since the plasma membrane is composed of cytoskeleton meshworks, any changes in its composition, including actin filaments, lipid rafts or clathrin, etc. may potentially affect receptor distribution and consequently HER signaling. For instance, the tea extract EGCG ((-)-epigallocatechin-3-galate) has been shown to alter the organization of lipid rafts and promote internalization of EGFR (Masuda et al., 2011). As a result, EGCG can cause a marked reduction in phosphorylated EGFR levels, thereby inhibiting EGFR signaling and achieving a chemo-preventive effect in SCCHN (Masuda et al., 2011).

The polarity of receptor distribution is also important. The localization of HER receptors to either the apical or basolateral surface determines whether they can be activated by their ligands (Vermeer et al., 2003). When there is disruption of cell-cell junctions and subsequent loss of cellular polarity, HER receptors will be exposed to their ligands they are normally not exposed to, and will thus initiate aberrant signaling (Vermeer et al., 2003). As mentioned earlier, in human airway epithelial cells with normal polarity, NRG1 is exclusively secreted into the apical side, whereas HER3 and HER4 reside on the basolateral part. However, when the polarity was experimentally disrupted, NRG1 gained access to its receptors and autocrine signaling was initiated (Vermeer et al., 2003). Similarly, in the study of SCCHN, Shames, et al. observed a pattern shift towards more autocrine expression in poorly differentiated and aggressive SCCHNs that was consistent with the more disordered architecture (Shames et al., 2013).

Among all the means of spatial control, regulation through endocytosis is probably the most important and complicated process. Again, using EGFR as an example, upon activation by ligand binding, EGFR is soon internalized and then undergoes endocytosis (Figure 2). Depending on the type of ligand, its dimerization partner or other factors (Sorkin and von Zastrow, 2009), EGFR may recycle to the cell surface to continue its membranous signaling, enter the nucleus to regulate gene transcription or protein phosphorylation, or be sorted to lysosomes for degradation (Sorkin and von Zastrow, 2009) (see Figure 2). In addition, mounting evidence indicates that EGFR remains active in endosomal compartments and may activate various downstream effectors distinctly different from its plasma membrane counterparts (Sorkin and von Zastrow, 2009) (Figure 2A). Endocytic trafficking can thus control the magnitude of EGFR signaling quantitatively by altering the amount of available receptors, and qualitatively by regulating its localization specificity. In fact, “derailed endocytosis” is now considered “an emerging feature of cancer” (Mosesson et al., 2008).

Besides using different approaches that disrupt EGFR endocytic trafficking to study its regulation, the study of many endocytosis regulators that are aberrantly expressed in cancer has offered us insights into targeting this endocytic pathway. Mosesson et al. has an excellent summary of these studies in their review paper (Mosesson et al., 2008), including for example, the downregulation of VPS37A (a vacuolar protein involved in sorting) stabilized EGFR; overexpression of cortactin (cortical actin binding protein involved in actin polymerization and rearrangement) inhibited ligand-induced EGFR endocytosis, etc (Mosesson et al., 2008). Interestingly, using a very well controlled cell system expressing either wild type EGFR or the mutant EGFRvIII, Grandal et al. found that EGFRvIII has inefficient internalization and impaired sorting to lysosomes for degradation due to lack of effective ubiquitinylation (Grandal et al., 2007).

Recently, Maiti GP et al. showed that inactivation of SH3GL2, a gene involved in controlling clathrin-mediated receptor endocytosis significantly impaired EGFR endocytosis for degradation, resulting in its stabilization and overexpression in SCCHN (Maiti et al., 2013). This is reminiscent of an earlier study performed by Dasgupa et al. in NSCLC, in which forced overexpression of SH3GL2 in NSCLC cell lines led to a marked reduction of EGFR expression following its enhanced internalization and degradation (Dasgupta et al., 2013). Considering that endocytosis is also crucial in regulating other HER members and, in addition, HER members affect each other’s endocytosis and degradation (Sak et al., 2013) (e.g. HER2 overexpression decreases the degradation of HER3 (Sak et al., 2013)), approaches regulating the endocytosis pathway need to be incorporated into the strategies targeting HER signaling (Figure 2A & B).

6. A summary of targeting HER signaling in HNC and future directions:

In this article, we have comprehensively reviewed the most recent studies of HER signaling in HNC and discussed the rationales for targeting the HER family from different perspectives. As illustrated in Figure 2 panel B, HER signaling can be targeted from two directions: 1) horizontally (Figure 2B, a): by blocking the dimerization and transactivation between different HER members and/or their association with other RTKs such as MET; 2) vertically (Figure 2B, b, c & d): this implies regulating HER signaling at different cellular levels including receptor membrane distribution, internalization, endocytic sorting, receptor recycling and degradation, as well as nuclear translocation, etc. It also includes blockade of serial downstream effectors of HER signaling since they may spatially and temporally regulate HER proteins (Casaletto and McClatchey, 2012), and some of them can be autonomously and constitutively active after acquisition of activating mutations (e.g. PIK3CA) or loss of tumor suppressors (e.g. PTEN).

Future research directions in the area of HER signaling in HNC may include: optimizing HER3 based clinical trials in well-selected patient subpopulations, identifying markers that may predict response to anti-HER therapies, and further characterizing the impact of HPV status on treatment. It will also be interesting to study the somatic mutations of HER3 and HER4, investigate the existence of a HER4 autocrine loop in NRG1 overexpressed SCCHN, and develop practical approaches to target the endocytic pathway of EGFR as well as to better understand its regulation of HER2-4. In the age of personalized therapy, how to incorporate each cancer patient’s genetic profile (e.g. amplitude of HER signaling before and after treatment) and unique clinical status (e.g. performance status, co-morbidities, response and side effects) to achieve the best therapeutic effect also needs consistent input from both the bench and clinical studies.

Decades of research in EGFR have provided us with a well-characterized prototype to study other HER family members. Although we have already deciphered many mysteries to date, our mission of using the enormous amount of information generated from these research studies to benefit patient care will continue to lead us to new discoveries in this field.