The Nuclear Epidermal Growth Factor Receptor Signaling Network and its Role in Cancer

Abstract

The epidermal growth factor receptor (EGFR) is a member of the EGFR family of receptor tyrosine kinases (RTKs). EGFR activation via ligand binding results in signaling through various pathways ultimately resulting in cellular proliferation, survival, angiogenesis, invasion, and metastasis. Aberrant expression or activity of EGFR has been strongly linked to the etiology of several human epithelial cancers including but not limited to head and neck squamous cell carcinoma (HNSCC), non-small cell lung cancer (NSCLC), colorectal cancer (CRC), breast cancer, pancreatic cancer and brain cancer. Thus intense efforts have been made to inhibit the activity of EGFR by designing antibodies against the ligand binding domains (cetuximab and panitumumab) or small molecules against the tyrosine kinase domains (erlotinib, gefitinib, and lapatinib). Although targeting membrane bound EGFR has shown benefit a new and emerging role for the EGFR is now being elucidated. In this review we will summarize the current knowledge of the nuclear EGFR signaling network, including how it is trafficked to the nucleus, the functions it serves in the nucleus, and how these functions impact cancer progression, survival and response to chemotherapeutics.

INTRODUCTION

In the early 1960s Stanley Cohen isolated and characterized a protein from the salivary gland that induced eye-lid opening and tooth eruption in newborn mice(Cohen, 1962). Further experimentation with this protein showed that it could stimulate the proliferation of epithelial cells and thus was named the epidermal growth factor (EGF)(Cohen, 1965). It was not until a decade later that Graham Carpenter identified the presence of specific binding receptors for EGF on target cells, later termed the epidermal growth factor receptor (EGFR)(Carpenter et al., 1975,(Carpenter et al., 1978). In the 1980s, the EGFR was cloned and sequenced and subsequently recognized as a receptor tyrosine kinase (RTK) (Ullrich et al., 1984). The identification of the EGFR as an RTK contributed to pivotal studies advancing our understanding of RTK activation (Riedel et al., 1986; Schlessinger, 1988), and elucidation of EGFR-regulated downstream signaling pathways such as the PI3K/AKT, Ras/Raf/Mek/Erk, and PLCγ1/PKC pathways (Figure 1).

Figure 1. Classical EGFR Signaling.

Ligand binding to the EGFR activates the intrinsic tyrosine kinase activity of the EGFR and leads to the phosphorylation of specific tyrosines on the cytoplasmic tails of the receptor pair. These phospho-tyrosines serve to recruit specific effector molecules to activate several signaling pathways including the PI3K/AKT (A), RAS/MAPK (B), PLCγ/PKC (C) and STATs (D) pathways.

Since its discovery several reports have described the nuclear localization of the EGFR within several cell types suggesting a new and emerging role of EGFR in the nucleus. In this review we highlight what is currently known about nuclear EGFR and its potential role in cancer.

EGFR TRANSLOCATION TO THE NUCLEUS

Nuclear localization of EGFR was first reported in regenerating hepatocytes and in primary adrenocortical carcinomas more than two decades ago(Kamio et al., 1990; Marti et al., 1991). Nuclear expression of EGFR was further detected in other cell types and tissues, such as mouse uterus, developing mouse embryos, rat liver, placentas, thyroids and immortalized epithelial cells of ovary and kidney origins(Cao et al., 1995; Lin et al., 2001; Marti et al., 2001). High levels of EGFR were also found in the nuclei of many tumors, including those of skin, breast, bladder, cervix, adrenocortical carcinoma, thyroid and oral cavity (Kamio et al., 1990; Lin et al., 2001; Lipponen & Eskelinen, 1994; Lo et al., 2005a; Lo et al., 2005b; Marti et al., 2001; Psyrri et al., 2005). The nuclear counterpart of EGFR appears to be the full-length receptor and likely, in the phosphorylated form, as shown by a number of studies(Cao et al., 1995; Cordero et al., 2002; Dittmann et al., 2005a; Lin et al., 2001; Lo et al., 2005a; Lo et al., 2005b). Although nuclear localization was well documented, a trafficking mechanism from the cytoplasmic membrane to the nucleus was not known at this time.

Over the last decade efforts have been undertaken by a cohort of researchers to better understand how the EGFR translocates to the nucleus. With this effort, accumulating evidence suggests a novel pathway where EGFR is internalized and transported to the nucleus rather than degraded or recycled back to the cell surface. Studies using dynamin mutants that impair the formation of clathrin-coated pits at the cell membrane have shown that inhibition of receptor endocytosis blocks nuclear import of EGFR(De Angelis Campos et al., 2011; Lo et al., 2006). In order to further elucidate proteins involved in EGFR’s nuclear translocation, Kim et al. immunoprecipitated the EGFR with a mono-specific anti-EGFR antibody followed by mass spectrometry analysis(Kim et al., 2007). Efforts from this work identified the phosphoinositide kinase, PIKfyve, which synthesizes phosphatidylinositol 3,5-bisphosphate. PIKfyve has been linked to endosomal dynamics and intracellular trafficking(lkonomov et al., 2003; Ikonomov et al., 2002; Ikonomov et al., 2006; Sbrissa et al., 2002). RNA silencing of PIKfyve impaired HB-EGF stimulated EGFR nuclear translocation, as well as EGFR binding to the cyclin D1 promoter, an early identified gene target of nuclear EGFR(Lin et al., 2001), and cell cycle progression(Kim et al., 2007). Additionally, EGFR association with the early endosomal marker endosome antigen 1 in the nucleus further suggests the role of endocytosis in EGFR nuclear transport(Lo et al., 2006).

During classical cytoplasmic-nuclear transport, proteins containing nuclear localization sequences (NLS) form complexes with importin α/β that interact with nucleoporins within the nuclear pore complex (NPC) and transverse into the nucleus(Harel & Forbes, 2004; Strambio-De-Castillia et al., 2010). Lo et al. first verified that EGFR contains a functional NLS by generating an EGFR mutant where the NLS was mutated (⁶⁴⁵RRR⁶⁴⁷ to AAA). This mutant, when expressed in EGFR-null CHO cells, was deficient in EGFR nuclear entry, verifying its role as a functional NLS(Lo et al., 2005a). A much more extensive analysis of putative NLSs in the EGFR and other HER family members identified a novel tripartite NLS. This NLS (⁶⁴⁵RRRHIVRKRTLRR⁶⁵⁷) contained three distinct clusters of basic amino acids that are conserved near the juxtamembrane region of all EGFR family members and was capable of targeting cytoplasmic proteins into the nucleus; this tripartite sequence is responsible for the nuclear localization of EGFR, and is believed to associate with importin β during internalization of the receptor(Hsu & Hung, 2007; Wang et al., 2010a).

Accumulating data strongly suggests that EGFR endocytosis is absolutely necessary for its nuclear transport; however, it is unclear how EGFR moves from these vesicles to the nucleus. Recently, data from Wang et al. has shed light on a potential mechanism(Wang et al., 2010a). In this study the investigators found that cells stimulated with EGF redistributed to the Golgi and to the endoplasmic reticulum (ER). Blocking of retrograde translocation from the Golgi to the ER using the drug brefeldin A or dominant mutants of the small GTPase ADP ribosylation factor (ARF) resulted in the disassembly of the coat protein complex I (COPI) to the Golgi. This dramatically impaired EGFR transport to the ER and the nucleus. Collectively the authors conclude that nuclear transport of the EGFR is regulated by COPI-mediated vesicular trafficking from the Golgi to the ER(Wang et al., 2010a).

Although this provides a molecular mechanism of movement to the ER, it doesn’t explain how EGFR is moved into the nucleus. A significant paper in this arena identified the Sec61 translocon(Liao & Carpenter, 2007; Wang et al., 2010b), localized on the inner nuclear membrane (INM), to be critical for the transport of EGFR from the INM into the nucleus(Wang et al., 2010b)(see Figure 2). In this report, EGF stimulation of the EGFR resulted in endocytosis of the EGFR and interaction with importin β, via its tripartite NLS. EGFR is described to undergo COPI-mediated retrograde trafficking from the Golgi to the ER. The ER is contiguous with the outer nuclear membrane (ONM) whereas the INM has a distinct protein composition and associates with the underlying chromatin. The INM and ONM are physically joined at the NPC. The authors provide strong data suggesting that once at the ER the EGFR-importin β complex interfaces with the nucleoporins in the NPC to shuttle EGFR to the INM. In this report, Sec61β was found to be localized at the INM, in addition to its normal residence on the ER. Further, knockdown experiments of Sec61β reduced EGFR level in the nucleus with subsequent accumulation of EGFR in the INM. Collectively this work over the last decade has pieced together a mechanism of nuclear translocation of the EGFR from the membrane to the nucleus.

Figure 2. EGFR Trafficking from the cell surface to the nucleus.

EGF stimulation at the cell surface induces receptor dimerization and internalization to endocytic vesicles (EV). EGFR then undergoes COPI-mediated retrograde translocation through the Golgi apparatus to the endoplasmic reticulum (ER)(Wang et al., 2010a). At the ER, EGFR moves from the outer nuclear membrane (ONM) to the inner nuclear membrane (INM) via interaction between importin β (β) and the nuclear pore complex (NPC). In the INM EGFR interacts with Sec61 for removal from the membrane and release into the nucleus(Wang et al., 2010b).

NUCLEAR EGFR AND CORRELATION WITH CLINICAL OUTCOME

The relationship between nuclear EGFR and cancer prognosis has been studied in multiple solid tumor types (Table 1). In 1994, two independent reports documented nuclear EGFR status in bladder and cervical cancers using immunohistochemical (IHC) methods. In the first study, a total of 234 transitional cell bladder cancer tumor samples were examined for nuclear EGFR by IHC(Lipponen & Eskelinen, 1994). Of the 234 samples, 31 % stained positive for nuclear EGFR. Nuclear EGFR expression was significantly correlated with increased tumor grade, mitotic frequency, and cellular proliferation. In the second study, Tervahauta and colleagues reported that 37 % of cervical cancer biopsies stained positive for nuclear EGFR(Tervahauta et al., 1994). While these early studies were the first to quantitate nuclear EGFR in patient tumors, they did not examine the prognostic value of nuclear EGFR expression.

Table 1.

Authors	Year	Cancer	Number	Methods	% nEGFR
(Lo et al., 2005b)	2005	Breast	130	IHC	38 %
(Hadzisejdic et al., 2010)	2010	Breast	113	IHC	40 %
(Xia et al., 2009)	2009	Ovarian	221	IHC	28 %
				ACIS III
(Lo et al., 2005b)	2005	Oral squamous carcinoma	37	IHC	24 %
(Psyrri et al., 2005)	2005	Oropharyngeal	95 (67 for AQUA)	IHC	49 %
				AQUA
(Li et al., 2011)	2011	Gallbladder	104	IHC	24 %
(Lipponen & Eskelinen, 1994)	1994	Bladder	234	IHC	31 %
(Tervahauta et al., 1994)	1994	Cervix	97	IHC	37 %

IHC: immunohistochemistry, ACISIII: automated cellular imaging system, AQUA: automated image acquisition and analysis

In 2005, Lo et al. analyzed 130 breast carcinoma tumor samples via IHC for the levels of nuclear EGFR (Lo et al., 2005b). They found nuclear EGFR staining in 38 % of these tumor tissues, 7 % of which had high levels of expression. They further demonstrated that patients with high levels of nuclear EGFR had poor overall survival (OS) as compared to patients with undetectable nuclear EGFR staining (33.4±14.6 vs. 47.8±15.3 months; P=0.009). They also found a positive correlation between expression of nuclear EGFR and two markers for cell proliferation, Ki67 and cyclin D1. Recently, Hadžisejdić et al. validated these findings in a separate cohort of invasive ductal breast tumors, where they detected nuclear EGFR expression in 40 % of 113 tumor samples (Hadzisejdic et al., 2010). Nuclear EGFR expression was statistically correlated with tumor size (P=0.0005), lymph node involvement (P=0.0288) and Nottingham prognostic index (P=0.0011). Further, patients with high levels of nuclear EGFR demonstrated worse OS compared to those with undetectable nuclear EGFR (0.58±0.249 vs. 0.83±0.053 months; P=0.0003), in addition to having a 3.4 times greater mortality risk (hazard ratio = 3.402; P=0.0026). However, researchers did not find statistically significant correlations between nuclear EGFR, Ki67 and cyclin D1 in this cohort of breast tumors.

Xia et al. further examined the levels of nuclear EGFR in a cohort of 221 ovarian cancer patient tissues using IHC followed by analysis with the ACIS III, automated cellular imaging system (Xia et al., 2009). A high level of nuclear EGFR staining was detected in 28 % of tumor tissues, while 23 % had low levels of nuclear EGFR. Additionally, high nuclear EGFR expressing tumors were significantly correlated to have worse OS as compared to non-nuclear EGFR expressing tumors (P<0.01).

Lo et al. additionally analyzed 37 cases of oral squamous carcinomas for nuclear EGFR using standard IHC methods. They found that 24 % had nuclear EGFR in more than 5 % of the tumor cells (Lo et al., 2005b). Patients with high nuclear EGFR expression in this cohort demonstrated poor OS compared to patients with no/low levels of nuclear EGFR (39.1±24.5 vs. 63.3±15.6 months; P=0.03). In 2005, Psyrri et al. quantitated EGFR protein expression via IHC and automated image acquisition and analysis (AQUA) on a tissue microarray composed of 67 oropharyngeal carcinomas (Psyrri et al., 2005). They found nuclear EGFR staining in 49 % of these tumor tissues. Researchers determined that 88 % of patients recorded to have either a partial or no response to first line chemotherapies were more likely to express high levels of nuclear EGFR (P=0.025). Additionally, nuclear EGFR was associated with a recurrence rate of 54 % as compared to 21 % for tumors with low levels of nuclear EGFR staining (mean recurrence time 35.9 vs. 49.4 months; P=0.0277). Kaplan-Meier survival analysis demonstrated that high nuclear EGFR patients had poor 5-year disease-free survival (19 %; 24.3 months) compared to those with low nuclear EGFR expression (44 %; 35.8 months; P=0.0386).

Most recently, Li et al. investigated 104 cases of gallbladder carcinomas on tissue microarray, and demonstrated that 24 % of this cohort had nuclear EGFR (Li et al., 2011). They also found that nuclear-pEGFR^Y1086 was overexpressed in 60 % of nuclear EGFR expressing cancers. Nuclear EGFR overexpression was also significantly associated with advanced stage cancer via American Joint Committee on Cancer (AJCC) standards (P=0.01) and vascular invasion (P=0.003). At the univariate level, shorter disease-specific survival was significantly associated with both activated and total levels of nuclear EGFR (P=0.0017 and P=0.0114).

THE ROLE OF EGFR IN GENE REGULATION

One of nuclear EGFR’s main functions is to act as a transcriptional co-activator for various oncogenic genes. Since EGFR does not contain a DNA binding domain it must associate with other transcription factors in order to interact with genomic regions. EGFR has now been identified as a transcriptional co-activator for seven cancer promoting genes: cyclin D1(Lin et al., 2001), nitric oxide synthase (iNOS)(Lo et al., 2005a), B-Myb(Hanada et al., 2006), Aurora Kinase A (Aurora-A)(Hung et al., 2008), cyclooxygenase-2 (COX-2)(Lo et al., 2010), c-Myc(Jaganathan et al., 2011), and breast cancer resistant protein (BCRP)(Huang et al., 2011). In the following paragraphs we discuss the experimental methods used by various investigators to both discover and verify nuclear EGFR as a transcriptional co-activator for these identified genes (Figure 3).

Figure 3. Nuclear EGFR Signaling Network.

EGFR has been consistently detected in the nuclei of cancer cells, primary tumor specimens and highly proliferative tissues. EGFR binds to several transcription factors to regulate gene transcription of cyclin D1(Lin et al., 2001), iNOS(Lo et al., 2005a), B-Myb(Hanada et al., 2006), Aurora Kinase A(Hung et al., 2008), COX2(Lo et al., 2010), Myc(Jaganathan et al., 2011) and BCRP(Huang et al., 2011). In addition, EGFR has been shown to phosphorylate and activate both PCNA(Wang et al., 2006) and DNA-PK(Dittmann et al., 2005a) within the nucleus.

Nuclear EGFR and Cyclin D1

In 2001, Lin et al. published the first study demonstrating the new role of EGFR as a transcriptional co-activator(Lin et al., 2001). This discovery stemmed from the observation that proliferative tissues stained for EGFR via IHC often resided in the nucleus. Researchers observed this nuclear EGFR phenomenon in cells analyzed from the uterus of pregnant mice, mouse embryos, normal human mouth mucosa, and human oral and breast cancer tissue. Comparatively, cells analyzed from non-proliferating tissues had minimal levels of nuclear EGFR. Authors further validated EGFR nuclear localization in cultured A431 epidermoid carcinoma cells and MDA-MB-468 basal breast cancer cells via confocal microscopy and nuclear fractionation analysis.

The authors of this study sought to understand the biological role of EGFR inside the nucleus of proliferating cells. Using motif-analysis of EGFR sequence they identified a proline-rich region on EGFR’s C-terminal tail that mimicked the sequence and charge of transactivation domains of transcription factors. To study this, they cloned EGFR’s C-terminal proline rich region in front of the GAL4 DNA binding domain, and transfected this construct into cells along side a chloramphenicol transferase (CAT) reporter construct. This construct strongly activated CAT activity (~60 fold), while other regions of the EGFR (cytoplasmic domain, and tyrosine kinase domain) fused to GAL4 did not. These experiments demonstrated the potential for EGFR to be a transcription factor.

To further elucidate the role of EGFR as a transcription factor, authors used the cyclic amplification and selection of targets method (CASTing method) to determine potential genomic sequences that can associate with EGFR. Potential genomic sequences identified via this method were polymerase chain reaction (PCR) amplified, sub-cloned, and sequenced. Six sub-clones were sequenced, each of which contained an AT-rich minimal consensus sequence (ATRS) consisting of either TNTTT or TTTNT (N being any nucleotide). ATRS sites were further validated to specifically associate with nuclear EGFR via electromobility shift assays (EMSA) with wild-type or mutant ATRS probes. Researchers also demonstrated that wild-type ATRS luciferase reporter constructs could be activated in EGF stimulated cell lines, while mutant ATRS luciferase reporter constructs could not. Overall, this body of work strongly suggests that EGFR associates with specific genomic sequences (ATRS) that can promote transcriptional activation.

Importantly, authors demonstrate that nuclear EGFR can associate with the cyclin D1 gene, an important regulatory protein responsible for transition through the G1 checkpoint in the cell cycle(Diehl, 2002). Authors considered cyclin D1 to be a potential nuclear EGFR target gene due to the fact that its proximal promoter region contains two ATRS sites. Thus, authors made cyclin D1 promoter luciferase constructs containing either wild type ATRS sites or mutated ATRS sites; luciferase activity was only detected after transfection with the wild-type ATRS construct in cell lines stimulated with EGF to promote nuclear EGFR translocation. Chromatin immunoprecipitation (ChIP) using anti-EGFR antibodies validated EGFR association with the cyclin D1 promoter region. Overall, Lin et al. conclude that nuclear EGFR can both associate and regulate the cyclin D1 gene, validating the biological role of nuclear EGFR as a transcriptional co-activator.

The previous study by Lin et al. led to an increased interest in nuclear EGFR functions and potential nuclear proteins it could be interacting with. Since this study, two new proteins have been identified to associate with nuclear EGFR, and promote EGFR’s ability to function as a transcriptional co-activator for cyclin D1. Mucin-1 (Muc-1) was the first protein shown to associate with nuclear EGFR, and was deemed necessary for its association with the cyclin D1 promoter(Bitler et al., 2010). Through the use of siRNA directed towards MUC-1 in breast cancer cell lines MDA-MB-468 and MCFlOa, researchers demonstrated reduced levels of EGFR association with the cyclin D1 gene. Muc-1 siRNA also led to a decrease in nuclear EGFR association with RNA polymerase II. These data suggest that MUC1 regulates EGFR’s ability to become localized in the nucleus, in addition to its ability to associate with active transcriptional regions.

The second protein identified to interact with nuclear EGFR to influence cyclin D1 transcription is RNA helicase A (RHA)(Huo et al., 2010). Through nano-liquid chromatography, researchers identified various RNA helicase proteins that bind to nuclear EGFR in MDA-MB-468 cells. RHA was chosen to pursue in further study because its Drosophilia homolog is known to bind to ATRS containing sequences. Nuclear co-immunoprecipitation (IP) and immunofluorescence (IF) assays demonstrated EGFR-RHA association in the nucleus of MDA-MB-468 cells. ChIP and subsequent PCR for the cyclin D1 promoter verified that both EGFR and RHA associate with the same region on the cyclin D1 promoter. Further, co-transfection of EGFR and RHA in HeLa cells led to a dramatic increase in cyclin D1 promoter luciferase activity, as compared to EGFR or RHA transfection alone. Additionally, siRNA directed towards RHA prevented EGFR from binding to the cyclin D1 promoter. Interestingly, authors demonstrate a positive correlation between nuclear EGFR, cyclin D1, and RHA expression via IHC in 51 human breast tumor samples. Overall, authors conclude that RHA is a nuclear EGFR binding partner that is required for full activation of EGFR target genes.

Nuclear EGFR and iNOS

Four years later, iNOS was the next nuclear EGFR target gene to be discovered(Lo et al., 2005a). iNOS is the key enzyme responsible for cellular production of nitric oxide, a potent signaling molecule known to influence metastasis and angiogenesis(Lechner et al., 2005). The first goal of this study was to discover transcription factors that may form complexes with EGFR inside the nucleus. Since signal transducer and activator of transcription 3 (STAT3) is activated by plasma membrane bound EGFR, the authors asked the question whether this association could also be found in the nucleus. Through Co-IPs using nuclear fractionate from A431 cells, authors discovered that EGFR and STAT3 interact inside the nucleus. Additionally, this interaction was further validated in MDA-MB-468 breast cancer cells via immuno-gold electron microscopy.

The second goal of this study was to identify target genes that could be regulated by EGFR and STAT3 complexes. Using literature searches and databases, researches identified various target gene promoters with both ATRS sites and STAT3 binding sites. The five gene promoters found were iNOS, c-fos, cyclin D1, human telomerase reverse transcriptase (hTERT), and vascular endothelial growth factor (VEGF). To validate that nuclear EGFR and STAT3 bind to these promoters, sequential ChIP assays were performed with EGFR and STAT3 antibodies. Nuclear EGFR and STAT3 complexes bound to the iNOS promoter region, but not the cyclin D1 or c-fos promoter regions, suggesting that different nuclear EGFR complexes may regulate different genes. Through luciferase studies with constructs containing wild type or mutant ATRS and STAT3 binding sites from the iNOS promoter region, authors demonstrate that only constructs containing wild-type ATRS and STAT3 binding sites can induce high levels of luciferase; this suggests that EGFR and STAT3 form complexes to regulate iNOS gene expression.

To further verify that nuclear EGFR was responsible for an increase in iNOS gene transcription, researchers elegantly constructed EGFR overexpressing CHO cells (CHO-EGFR) and EGFR overexpressing CHO cells lacking the putative NLS necessary for EGFR nuclear import (CHO-EGFR-pNLS). CHO-EGFR cells had increased levels of nuclear EGFR, and were able to associate with STAT3 in the nucleus, while CHO-EGFR-pNLS cells were deficient in nuclear EGFR. When these cells were transfected with wild type iNOS luciferase reporter constructs, EGF stimulation strongly induced luciferase activity in CHO-EGFR cells, while luciferase activity was severely diminished in CHO-EGFR-pNLS cells. This experiment demonstrated the role of nuclear EGFR in the regulation of iNOS gene expression.

Importantly, 111 human breast carcinoma tumors previously stained for EGFR(Lo et al., 2005b) were subsequently stained for iNOS expression via IHC. A positive correlation was found between levels of nuclear EGFR and iNOS via chi-squared analysis. Additionally, they selected 15 tumors that stained positive for nuclear EGFR and iNOS, and further stained for activated STAT3 (Y705). Nuclear localized p-STAT3 positively correlated with nuclear EGFR and iNOS expression levels. Finally, iNOS expression was positively correlated with worse OS in this cohort of breast cancer patients. Overall this study strongly supports the role of nuclear EGFR and STAT3 in the promotion of iNOS gene expression, suggesting that this combination may lead to worse OS in breast cancer.

Nuclear EGFR and B-Myb

B-Myb is a proto-oncogene that plays a role in progression through the G1/S phase of the cell cycle(Joaquin & Watson, 2003). Based on previous reports demonstrating that basal type breast cancers overexpress EGFR and B-Myb(Nielsen et al., 2004; Sorlie et al., 2001), Hanada et al. looked for a potential mechanism for this association(Hanada et al 2006). Initial experiments demonstrated that B-Myb gene expression (detected via quantitative real time polymerase chain reaction (RT-PCR)) could be induced via EGF stimulation of MDA-MB-468 cells, and that this induction was independent from PI3K/AKT and MAPK activation. Since traditional EGFR signaling pathway inhibition did not affect B-Myb expression, researchers hypothesized that EGFR may be functioning as a transcriptional co-activator of B-Myb. Subsequently, researchers identified that the B-Myb promoter contains one ATRS region, and verified that nuclear EGFR does in fact associate with this region upon EGF stimulation via ChIP analysis. Interestingly, EGFR was shown to only associate with the B-Myb promoter region during G1/S phase of the cell cycle, the phase of the cell cycle when B-Myb is known to be active.

Researchers extended this study by identifying that EGFR associates with the E2F1 transcription factor on the B-Myb promoter; E2F1 is known to promote B-Myb expression during G1/S phase(Joaquin & Watson, 2003). EGFR and E2F1 were shown to associate by Co-IPs of nuclear protein isolated from EGF stimulated human keratinocytes (HaCat cells). Additionally, EGF stimulation induced E2F1 association with the B-Myb promoter region as detected via ChIP. Finally, co-overexpression of EGFR and E2F1 in CHO cells (EGFR null cells) lead to dramatic increases in B-Myb luciferase activity. Overall, these experiments demonstrate that EGFR and E2F1 associate on the B-Myb promoter to drive its expression during the G1/S phase of the cell cycle.

Nuclear EGFR and Aurora KinaseA

The overexpression of EGFR has been associated with chromosome instability in various cancers(Tomida et al., 2005), a process that leads to defects in mitosis and aneuploidy. Aurora-A is a serine/threonine kinase that associates with the centrosome during mitotic spindle formation to ensure proper spindle formation, chromatid separation, and fidelity of the spindle checkpoint(Fu et al., 2007; Katayama et al., 2003). Aurora-A is also overexpressed in various cancers, and shown to promote deregulated mitotic spindle formation(Fu et al., 2007; Katayama et al., 2003). In a 2008 study, researchers set out to determine the mechanistic link between chromosomal instability and EGFR(Hung et al., 2008). Using immunofluorescence (IF), researchers demonstrated that EGF stimulated A431 cells had increased numbers of centrosomes (stained for centrosomal proteins gamma-tubulin and Aurora-A) per cell, accompanied by a disarranged microtubule network (stained for alpha-tubulin). This was compared to non-stimulated A431 cells, which had 2 centrosomes per cell undergoing mitosis with orderly microtubule formations. Importantly, EGF stimulation of three different cell lines, A431, MDA-MB-231, and MDA-MB-468 led to increased mRNA and protein expression of Aurora-A, providing a link between EGFR activation and Aurora-A expression.

The authors hypothesized that nuclear EGFR may be responsible for increases in Aurora-A expression because EGF stimulation leads to increased nuclear localization of EGFR and not receptor degradation. Through analysis of the Aurora-A promoter region, five ATRS sites were found. Subsequently, ChIP analysis demonstrated that EGFR can bind to the Aurora-A promoter region. This was further verified via DNA pull down assays (DAPA) assays, where biotinylated Aurora-A promoter probes were shown to associate with EGFR. Further analysis of the Aurora-A promoter region demonstrated several STAT binding sites overlapping the same ATRS regions that EGFR was shown to associate with. Subsequently, STAT5 was shown to associate with these regions via ChIP and DAPA assays, while STAT3 did not. Luciferase assays using wild type and mutant ATRS constructs verified that EGFR binding regions are indeed necessary for Aurora A expression; mutated ATRS constructs had decreased luciferase activity. Additionally, siRNA directed towards STAT5 or EGFR yielded reduced luciferase activity. Overall, these data suggest nuclear EGFR and STAT5 form a complex to promote the expression of Aurora A.

Nuclear EGFR and C0X2

Prior to 2010, nuclear EGFR target genes were not identified on a genome wide scale; in fact, most researchers identified single nuclear EGFR target genes based on prior knowledge suggesting that EGFR expression was correlated with the expression of that gene. In 2010, Lo et al. performed the first genome wide search for nuclear EGFR target genes in a model of glioblastoma multiforme (GBM)(Lo et al., 2010). To identify nuclear EGFR target genes researchers established stable U87MG isogenic cell lines over-expressing either wild type EGFR or EGFR deleted of its NLS (EGFRdNLS). To identify genes that could be regulated by nuclear EGFR, total RNA was collected post EGF stimulation from each cell line and used to probe a microarray containing over 47,000 human transcripts. Genes that demonstrated differential expression levels in the U87MG-EGFRdNLS cells as compared to the U87MG-EGFR cells represented potential nuclear EGFR target genes. Through this method, 19 genes were identified to be differentially expressed in EGFRdNLS cells, one of which being the COX-2 gene. Researchers validated that EGF significantly increased both expression (via quantitative RT-PCR) and protein (via western blot) levels of COX-2 in U87MG-EGFR cells but not in U87MG-EGFRdNLS cells, suggesting that nuclear EGFR is involved in the regulation of COX-2.

Further, researchers identified that the transcription factor STAT3 is an important EGFR binding partner for the regulation of COX-2. Through ChIP studies, they demonstrated that EGFR and STAT3 could associate with the COX-2 promoter in EGFR wild type cells upon EGF stimulation. Transfection of U87MG cells with EGFR and STAT3CA (constitutively active STAT3) yielded high levels of COX-2 luciferase reporter activity when stimulated with EGF, while single transfections with either EGFR or STAT3CA yielded lower readings. Additionally, transfection of U87MG cells with EGFRdNLS and STAT3CA yielded low COX-2 luciferase readings. Taken together, these data provide strong evidence that nuclear EGFR and STAT3 form a complex that enhances COX-2 expression.

Lo et al. shed further light on the nuclear localization of a constitutively active mutant of EGFR (EGFRvIII) in GBM tumors. In accordance with their previous experiments, researchers constructed U87MG-EGFRvIII and U87MG-EGFRVIIIdNLS stable cell lines. EGFRvIII stable cells demonstrated increased COX-2 gene and protein expression, while this expression was not enhanced in the EGFRvIIIdNLS cells. EGFRvIII was also shown to enhance COX-2 luciferase activity in U87MG cells; however, dual transfection with STAT3CA only modestly increased luciferase activity above the level of EGFRvIII alone. Researchers concluded that EGFRvIII and STAT3 function to activate COX-2 expression; however, STAT3 does not seem to enhance activation of COX-2 as robustly as it does in the wild type EGFR setting. Overall, this study was the first to demonstrate a method to search for nuclear EGFR target genes on a genome wide scale, and further validated the cooperation of EGFR and STAT3 in the regulation of COX-2 gene expression.

Nuclear EGFR and c-Myc

With the growing knowledge of different EGFR gene regulatory complexes within the nucleus, it is not surprising that EGFR also complexes with other kinases to regulate gene expression. In a recent report by Jaganathan et al. a heteromeric complex of EGFR, STAT3, and c-Src kinase was identified to associate with the c-Myc promoter in a model of pancreatic cancer (Jaganathan et a1, 2011). Co-IPs from both Panc-1 and Colo-357 cells demonstrated association of EGFR, c-Src, and STAT3 in the nucleus. Interestingly, the use of an EGFR kinase inhibitor (erlotinib) or a Src family kinase inhibitor (dasatinib) had no effect on 1) nuclear EGFR localization, 2) nuclear c-Src levels, and 3) the ability for EGFR and c-Src to form a complex in the nucleus. However, STAT3 association with EGFR and c-Src was reduced upon treatment with either inhibitor, suggesting that the kinase activity of either EGFR or c-Src is needed for STAT3’s association.

Previous reports have demonstrated that simultaneous inhibition of EGFR, c-Src, and STAT3 can decrease growth of various pancreatic cancer cell lines, potentially through the inhibition of c-Myc(Jaganathan et al., 2010). The study presented above builds off this finding by demonstrating that nuclear EGFR, c-Src, and STAT3 all associate with the c-Myc promoter region via sequential ChIP analysis(Jaganathan et al., 2011). Finally, c-Myc protein expression only decreased upon siRNA or drug inhibition of at least two of the three members of the heteromeric complex; knockdown of only one member alone did not affect c-Myc expression levels. This raises the possibility that EGFR, c-Src, and STAT3 regulate the c-Myc gene. Overall, Jaganathan et al. describe a novel nuclear EGFR complex that associates with the promoter region of c-Myc.

Nuclear EGFR and BCRP

BCRP, also known as ATP-binding cassette sub-family G member 2 (ABCG2), has now been identified as yet another nuclear EGFR target gene. BCRP is an ATP binding cassette half transporter that forms homo- or heterodimers in the plasma membrane of various cell types to actively pump macromolecules out of cells(Mao & Unadkat, 2005). The expression of BCRP is increased in various cancer cells and is associated with resistance to chemotherapeutics(Mao & Unadkat, 2005; Takara et al., 2006; Usuda et al., 2007).

Recently, in an A431 epidermoid cancer model of gefitinib resistance, researchers demonstrate that resistant cells have high levels of both nuclear EGFR and BCRP protein expression(Huang et al., 2011). BCRP expression was demonstrated to be dependent on EGFR signaling as siRNA directed towards the EGFR attenuated both mRNA and protein levels of BCRP. Increased expression of BCRP was shown to be partly dependent on the function of nuclear EGFR by silencing importin b1 (necessary for EGFR nuclear import) and by using an EGFRdNLS mutant. Analysis of the BCRP promoter region yielded various ATRS presence. Through both ChIP analysis and DAPA assay, EGFR was shown to associate with numerous ATRS on the BCRP promoter. Through BCRP promoter luciferase assays, researchers demonstrate that A431 gefitinib resistant cells, which were shown to have high levels of nuclear EGFR, yielded elevated levels of luciferase activity, while gefitinib sensitive cells lacking nuclear EGFR did not. Overall, these data demonstrate that nuclear EGFR regulates the expression of BCRP in a model of gefitinib resistance.

THE ROLE OF NUCLEAR EGFR IN DNA REPLICATION AND REPAIR

Nuclear EGFR also plays essential roles in promoting both DNA replication and repair by associating with proliferating cell nuclear antigen (PCNA) (Wang et al., 2006) and DNA dependent protein kinase (DNA-PK)(Bandyopadhyay et al., 1998; Dittmann et al., 2005a). Both DNA replication and repair are essential cellular processes necessary for cancer formation and progression. In the following paragraphs we describe the studies that lead to the discovery of nuclear EGFR’s role in these processes (see Figure 3).

Nuclear EGFR and PCNA

PCNA is a DNA clamp that forms a homotrimeric ring complex around DNA serving to recruit DNA polymerase, ligase, and repair proteins during DNA replication(Stoimenov & Helleday, 2009). In a study by Wang et al., researchers identified a direct link between nuclear EGFR and stability of PCNA(Wang et al., 2006). Through electrospray ionization tandem mass spectrometry and site directed mutagenic studies, researchers identified phospho-tyrosine 211 of PCNA to be essential for its cellular stability and function; mutant PCNA at tyrosine 211 was rapidly degraded, a process that was reversed by proteasome inhibition. Researchers also identified direct association of nuclear EGFR and PCNA through Co-IPs from nuclear extract in both MDA-MB-468 cells and A431 cells. They further showed that the EGFR tyrosine kinase inhibitor AG1478 can block phosphorylation of PCNA at tyrosine 211, and thus promote PCNA degradation. These data suggest that EGFR directly phosphorylates PCNA, leading to its stabilization and ability to function as a DNA clamp. In 2005, researchers identified that nuclear EGFR was correlated with poor overall patient survival in a cohort of primary breast cancer tumor samples(Lo et al., 2005b). Using this same cohort, authors stained for PCNA phospho-tyrosine 211, which they demonstrated to be significantly correlated with both nuclear EGFR and poor OS. Overall, these data suggest that nuclear EGFR phosphorylates and regulates the activity of PCNA inside the nucleus, a process necessary for DNA replication.

Nuclear EGFR and DNA-PK

DNA-PK is a serine/threonine protein kinase that is involved in the non-homologous end joining DNA repair pathway responsible for rejoining double strand breaks in DNA(Collis et al., 2005). DNA-PK can phosphorylate various proteins needed for this processes, including ligases, polymerases, and also aid in recruiting these enzymes by acting as a scaffold(Collis et al., 2005). In 1997, Bandyopadhyay et al. demonstrated that EGFR and DNA-PK physically associate upon treatment with cetuximab (C225), causing a decrease in nuclear accumulation and activity of DNA-PK(Bandyopadhyay et al., 1998). This study was taken a step further in 2005 by Dittmann et al. who demonstrated that EGFR and DNA-PK interact in the nucleus to initiate the DNA repair pathway(Dittmann et al., 2005a). This finding was built off the observation that EGFR became rapidly phosphorylated and localized to the nucleus upon ionizing radiation treatment of A549 bronchial carcinoma cells. Ionizing radiation also led to an increase in EGFR and DNA-PK association, and an increase in DNA-PK activity. This series of events was prevented upon pre-treatment of cells with cetuximab, which blocked both EGFR nuclear transport and DNA-PK’s ability to associate with DNA binding complexes as demonstrated via EMSA assay. Additionally, γH₂AX foci formation, indicative of DNA double strand breaks, was delayed upon dual cetuximab and radiation treatment, supporting a loss of DNA-PK activity. Overall, these studies demonstrate the important regulatory role of DNA-PK by nuclear EGFR.

THE ROLE OF NUCLEAR EGFR IN RESISTANCE TO CANCER THERAPEUTICS

As cancer treatment becomes more advanced, clinicians and researchers alike have observed high levels of intrinsic and/or acquired tumor cell resistance to some of the most powerful cancer therapeutics. Nuclear EGFR has now been identified to play a role in resistance to various cancer therapies including cetuximab(Li et al., 2009), gefitinib(Huang et al., 2011), radiation(Dittmann et al., 2010; Dittmann et al., 2011; Dittmann et al., 2005a; Dittmann et al., 2008a; Dittmann et al., 2008b; Dittmann et al., 2005b; Dittmann et al., 2007; Liccardi et al., 2011; Wanner et al., 2008), and cisplatin(Dittmann et al., 2005a; Hsu et al., 2009; Liccardi et al., 2011). In the following paragraphs we highlight these studies, and the lessons learned from them in order to provide the best possible treatment regimes for different tumor types.

Nuclear EGFR and resistance to cetuximab

In attempt to understand mechanisms of resistance to cetuximab, Wheeler et al. created cetuximab resistant cell lines by chronically exposing the NSCLC cell line H226 to increasing concentrations of cetuximab over a 6-month time period(Wheeler et al., 2008). Cetuximab resistant clones demonstrated increased expression of all HER family members, in addition to Src family kinase (SFK) activity(Wheeler et al., 2008; Wheeler et al., 2009). In 2009, Li et al. elaborated on this study by demonstrating that cetuximab resistant clones also harbored high levels of nuclear EGFR(Li et al., 2009). They further discovered that SFKs played an essential role in the nuclear localization of EGFR, as the pan SFK inhibitor dasatinib decreased nuclear EGFR accumulation, and increased plasma-membrane bound EGFR. Importantly, dasatinib treatment re-sensitized cetuximab resistant clones to cetuximab, suggesting that removal of nuclear EGFR and its re-allocation onto the plasma membrane enabled cetuximab to become effective. To verify that nuclear EGFR was playing a role in cetuximab resistance, authors constructed stable cell lines with high levels of nuclear EGFR through the fusion of EGFR to an extra NLS (cells were designated as EGFR-NLS). Importantly, EGFR-NLS cells demonstrated increased resistance to cetuximab both in vitro and in vivo. Overall, this study highlights the role of nuclear EGFR in cetuximab resistance and provides rationale for treating cetuximab resistant tumors with cetuximab and dasatinib concurrently.

Nuclear EGFR and resistance to gefitinib

Over the past decade various studies have noted perinuclear and nuclear localization of the EGFR in gefitinib resistant cancer cells(Kwak et al., 2005; Nishimura et al., 2008). To explore potential mechanisms of nuclear EGFR in gefitinib resistance, Huang et al. used wild type EGFR overexpressing A431 cancer cells that were treated with increasing amounts of gefitinib over a two-month time span to induce acquired resistance(Huang et al., 2011). Gefitinib resistant A431 cells had increased levels of nuclear EGFR. Subsequently researchers identified that AKT phosphorylates EGFR directly at serine 229 in A431 gefitinib resistant cells, inducing the nuclear localization of EGFR; inhibition of AKT with the pharmacological inhibitor API-2, or through siRNA directed towards AKT1/2/3 reduced nuclear EGFR localization. As discussed previously, nuclear EGFR was shown to regulate BCRP gene transcription in this same study. Additionally, gefitinib has been shown to be a substrate of BCRP(Nakamura et al., 2005; Shi et al., 2009; Shi et al., 2007). These data led authors to hypothesize that nuclear EGFR enhancement of BCRP expression could potentially increase gefitinib extrusion from cells leading to gefitinib resistance. Thus, this study provides rationale for circumventing gefitinib resistance through the targeting of both AKT and BCRP/ABCG2 transporters.

Nuclear EGFR and resistance to radiation therapy

It is now well accepted that radiation therapy induces EGFR nuclear translocation(Dittmann et al., 2010; Dittmann et al., 2011; Dittmann et al., 2005a; Dittmann et al., 2008a; Dittmann et al., 2008b; Dittmann et al., 2005b; Dittmann et al., 2007; Liccardi et al., 2011; Wanner et al., 2008). Radiation therapy is used because of its ability to create DNA damage, and induce apoptosis in a variety of different cancer types. Nuclear EGFR’s ability to associate and stabilize DNA-PK in the nucleus upon radiation treatment suggests that it plays a role in radiation resistance(Dittmann et al., 2005a). This same finding was demonstrated via knockdown of protein kinase C ε (PKCε) with siRNA, which also decreased nuclear EGFR accumulation and DNA-PK activation after radiation treatment(Wanner et al., 2008). Both studies demonstrated increased residual DNA damage upon dual radiation treatment with either cetuximab or siPKCε, suggesting that the removal of nuclear EGFR sensitizes cells to radiation.

Recently, nuclear EGFR was discovered to be associated with various chromatin-binding proteins necessary for modulating chromatin access during the DNA repair process post radiation(Dittmann et al., 2011). Through a time course of Co-IPs with γH₂AX post radiation, researchers show an increased association with EGFR, ataxia telangiectasia mutated (ATM), KAP1, promyelocytic leukemia protein (PML), and marker of chromatin relaxation histone H3 acetyl-K9. Thus, radiation induces nuclear EGFR association with various proteins that play a role in both DNA relaxation and repair. This experiment was bolstered through siRNA studies directed towards EGFR, which led to a decrease in the activation of ATM and TIP60 acetylase. Finally, inhibition of nuclear EGFR through the use of a specific EGFR-NLS peptide led to increased residual γH₂AX and decreased clonogenic survival of A549 cells upon radiation treatment. Overall these data suggest that nuclear EGFR enhances resistance to radiation through the modulation of chromatin access.

Nuclear EGFR and resistance to chemotherapy

Cisplatin treatment, a common chemotherapy used for first line treatment for various malignancies, has also been shown to induce the nuclear translocation of EGFR(Dittmann et al., 2005a). In a report by Hsu et al. researchers build on this original observation through the use stable MCF-7 cells expressing either wild type EGFR, EGFR lacking its NLS (EGFRdNLS), or EGFRdNLS restored with an NLS sequence from the large T antigen of the SV40 virus(Hsu et al., 2009). MCF-7 cells overexpressing wild type EGFR were resistant to cisplatin treatment as measured by proliferation assay and γH₂AX foci formation, while EGFRdNLS cells were much more sensitive to treatment. Additionally, EGFRdNLS restored with the SV40 NLS sequence demonstrated increased resistance to cisplatin (similar to that of wild type EGFR cells), supporting the role of nuclear EGFR in resistance to cisplatin. In 2011, a study by Liccardi et al. further supported this finding by using an isogenic NIH3T3 murine fibroblast model. Cells overexpressing EGFR or EGFRVIII demonstrated increased association with DNA-PKs and resistance to cisplatin, while EGFRdNLS and EGFRL858R (constitutively active EGFR tyrosine kinase) overexpressing mutant cell lines did not(Liccardi et al., 2011). Overall, these studies support the role of nuclear EGFR in resistance to the chemotherapy cisplatin, potentially through its ability to associate with DNA-PKs.

Closing Remarks

The epidermal growth factor receptor represents one of the most well studied molecules in all of biology. From its early identification and cloning, to the discovery of its role in cancer, it has been at the forefront of our understanding of receptor tyrosine kinases and cell signals that mediate homeostasis, but when overexpressed facilitate tumorigenesis. Since its first identification of its role as a transcription factor in 2001(Lin et al., 2001), EGFR has led the way in discovery of various other nuclear localized RTKs, including HER2(Wang et al., 2004), HER3(Offterdinger et al., 2002), HER4(Ni et al., 2001), IGFlR(Sehat et al., 2010), cMET(Matteucci et al., 2009) and others.

However, it must be stated that much work is needed to continue the upward trend of our understanding of nuclear EGFR and beyond. Nuclear EGFR has been shown to be expressed in normal tissues, where cells are highly proliferative, suggesting that nuclear EGFR plays a role in normal biology. Although nuclear EGFR has been strongly correlated with poor OS in several human cancers, associated with resistance to chemotherapeutic agents, as well as involved in the regulation of genes needed for cell proliferation, no evidence to date has definitively defined nuclear EGFR as a true oncogene. Research over the next decade will hopefully pursue intense studies to test this hypothesis. Data and results stemming from these findings will most likely implicate nuclear EGFR and various other nuclear RTKs in the etiology of cancer, and thus provide strong rationale for clinically targeting this mode of signaling in the future.