Breast Cancer Expressing the Activated HER 2 / neu Is Sensitive to Gefitinib In vitro and In vivo and Acquires Resistance through a Novel Point Mutation in the HER 2 / neu

The HER2/neu oncogene is an important diagnostic and prognostic factor and therapeutic target in breast and other cancers. We developed and characterized a breast cancer cell line (Bam1a) that overexpresses the activated HER2/neu and ErbB-3 and has a gene expression profile consistent with the ErbB-2 genetic signature. We evaluated the effects of the epidermal growth factor receptor (EGFR)/HER2 inhibitor, gefitinib, on this breast tumor line in vitro and in vivo . We characterized the effects of gefitinib on EGFR, HER2, and ErbB-3 phosphorylation by Western blot and determined the effects on downstream signaling through growth, survival, and stress pathways and the effect on proliferation, cell cycle, and apoptosis. Gefitinib treatment diminished phosphorylation of the ErbB-3 > EGFR > HER2/neu and signal transducers and activators of transcriptions in a dose-dependent fashion. Downstream mitogenic signaling through mitogen-activated protein (MAP)/extracellular signal regulated kinase kinase, p44/42 MAP kinase (MAPK) and stress signaling through c-Jun-NH2-kinase (JNK) 1 and c-Jun was impaired (1 Mmol/L, 4–24 h), leading to cytostasis and cell cycle arrest within 24 h by decreased cyclin D1, cyclin B1, and pRb and increased p27. Proliferation and colony formation were inhibited at 0.5 and 1 Mmol/L, respectively, and correlated with altered gene expression profiles. Diminished survival signaling through Akt, induction of bim, loss of connexin43, and decreased production of vascular endothelial growth factor-D preceded caspase-3 and poly(ADP)ribose polymerase (PARP) cleavage and apoptosis (>50% 2 Mmol/L, 48 h). Oral administration of gefitinib was able to prevent the outgrowth of Bam1a tumor cells from palpable lesions, shrink established tumors, eliminate HER2 and HER3 phosphorylation, and decrease MAPK and Akt signaling in vivo. A variant of the Bam1a cell line, IR-5, with acquired ability to grow in 5 Mmol/L gefitinib was developed and characterized. IR-5 bears a novel point mutation in the HER2/neu that corresponds to a L726I in the ATP-binding pocket and correlates with a log decrease in sensitivity to gefitinib, increased heterodimerization with EGFR and HER3, and impaired down-regulation. Gene expression profiling of IR-5 showed increased expression of EMP-1, NOTCH-1, FLT-1, PDGFB , and several other genes that may contribute to the resistant phenotype and sustain signaling through MAPK and Akt. This model will be useful in understanding the differences between intrinsic drug sensitivity and acquired resistance in the context of therapeutic strategies that target oncogene addicted diseases. [Cancer Res 2007; 67(14):6825–43]


Introduction
Receptor tyrosine kinases (RTK) play important roles in regulating normal cell growth by initiating specific intracellular signaling pathways in response to binding of extracellular growth factors.A wide variety of cellular functions are modulated by the four members of the ErbB [or epidermal growth factor receptor (EGFR)] family, including regulation of mitogenesis, cell death, angiogenesis, and cell differentiation (1).The oncogenic potential of the ErbB family members has been correlated to overexpression or alterations in a variety of human cancers, including breast, ovarian, non-small cell lung, glioblastoma, prostate, pancreas, head and neck, and other cancers (2).Thus, targeting the signaling activity of the receptor has emerged as an attractive approach for treatment and prevention of RTK-driven malignancies (3) with well-appreciated complexity [reviewed by Hynes and Lane (4)].
Signal transduction from the oncogenic HER2/neu leads to neoplastic transformation, initiation, cellular immortalization, and tumor progression.This is in part due to dysregulated signaling through the HER2 kinase domain to the growth and survival pathways.In models of spontaneous tumorigenesis (5), overexpressed or mutated p185 leads toward the formation of homodimers or heterodimers with other EGFRs.As these dimers transduce positive growth signals in a ligand-independent way (6), they are involved in the initiation and progression of neoplastic transformation (7).Expression of the activated neu oncogene in transgenic mice has been associated with both the synchronous (single step) and the stochastic (ref.8; multistep) transformation of mammary epithelium.Sequence analyses revealed that activation of neu occurs through a single amino acid change in the transmembrane portion of the protein (7).This single point mutation replaces the valine residue at position 664 in the transmembrane domain of p185 with glutamic acid, favors p185 homodimerization and heterodimerization, and transforms the Her-2/neu proto-oncogene into a dominant transforming oncogene (7).
Monoclonal antibodies specific for the EGFR (9) and HER2 (10) receptors have been developed with successful clinical outcomes (10).More recently, several small-molecule inhibitors targeting the tyrosine kinase domains of specific RTKs have been developed as therapeutic agents to treat a variety of cancers and include classes of quinazolines that act as reversible or irreversible small-molecule competitive substrate (ATP) inhibitors (11).
In the case of the ErbB family, the preclinical efficacy of EGFRselective, small-molecule tyrosine kinase inhibitors [TKI; i.e., ZD1839 (gefitinib, Iressa; AstraZeneca) and OSI-774] in EGFRdependent tumor models has been well characterized (12,13) ZD1839 (gefitinib) is an orally available, active, selective EGFR TKI that blocks signal transduction pathways implicated in proliferation and survival of cancer cells and other host-dependent processes promoting cancer growth (13)(14)(15).In tyrosine kinase activity assays, EGFR inhibition occurs at IC 50 doses of 0.023 to 0.079 Amol/L.Inhibition of c-erbB-2 (IC 50 , 3 Amol/L) and KDR/vascular endothelial growth factor (VEGF) receptor 2 (3.7-33Amol/L) also occurs but at doses 100-fold higher than EGFR inhibition.In vitro, the effect of gefitinib on human breast, ovarian, and colon cell lines expressing various amounts of EGFR and/or HER2 has been described as mainly cytostatic with increasing apoptotic activity at the higher doses (14,15).Supra-additive antiproliferative effects were observed with a broad range of cytotoxics (12,16) and radiation (17) in vitro and in vivo.The efficacy of gefitinib in HER2-overexpressing human breast cancer cell lines has been described and seems to be contingent upon the expression levels of both HER2 and EGFR as well as the degree to which these receptors are coupled to the growth [mitogen-activated protein kinase (MAPK)] and survival (Akt) pathways (14,15).
In recent studies, a subset of EGFR mutations in lung cancer patients was shown to correlate with clinical responsiveness to gefitinib therapy (18).Subsequently, several HER2 and EGFR mutations and polymorphisms have been identified in a variety of human tumors that influence patient prognosis and sensitivity to gefitinib (19) and may alter sensitivity to other RTKIs.
We have already shown the effectiveness of gefitinib on the phosphorylation and signaling of the oncogenically activated (rat) HER2/neu in the context of a salivary gland adenocarcinoma (20).Suppression of HER2 signaling by gefitinib induced profound cytostasis by silencing growth signaling through MAPK.Nevertheless, cells displayed intrinsic resistance to gefitinib-induced apoptosis because signaling from the Akt pathway was intact to sustain survival and inhibition of fas was required for apoptosis.
To test the efficacy of gefitinib in breast cancer expressing the mutated HER2/neu oncogene, we isolated and characterized a breast cancer cell line and the mechanisms of gefitinib with respect to HER2 signal transduction through the growth and survival pathways as well as proliferation, anchorage-independent growth, kinetics of cell cycle progression, cell death through apoptosis, and tumor growth and signal transduction in vivo.Further, we determined the ability of HER2-overexpressing breast cancer to develop resistance to this agent and characterized mechanisms responsible for decreased sensitivity to gefitinib.We determined that a novel point mutation in the ATP-binding pocket HER2/neu receptor was responsible for the resistant phenotype and led to additional genetic alterations that are likely to contribute to acquired gefitinib resistance.

Materials and Methods
Routine methodology was used as we have previously described (20).The specific experimental details are provided in Supplementary Methods.

Results and Discussion
Characterization of the Bam1a cell line.The Bam1a cell line was derived from a mammary gland tumor that developed in a BALB-NeuT female mouse.Mammary gland tumorigenesis by the activated rat HER2 oncogene in female BALB-NeuT transgenic mice has been described (21).We characterized the tumorigenic potential of the Bam1a cell line and compared it with the morphology of the parental.Bam1a cells grown in immunocompetent BALB/c were evaluated using H&E and immunohistochemistry for expression of the HER2 receptor.Histologic features of the mammary tumor that was used to establish the Bam1a cell line are shown in Fig. 1A.
When grown from single-cell suspension in the fat pad of female BALB/c mice, tumors that formed were morphologically similar to the parental and developed well-circumscribed and encapsulated nodules, made of smaller acini, separated by fine stroma (Fig. 1A).Acini have prominent, open lumens so that focally the confluence of acini creates a cribriform pattern and often are filled with eosinophilic secretions.The stroma is, overall, more developed than in the parental tumor, creating a more pronounced division of the acini and prominent comedo-type necrosis is present.Immunohistochemistry reveals strong and diffuse positivity for HER2 that is similar to that of the parental with a similar (f2+) membrane and cytoplasmic intensity.
Bam1a cells in vitro are uniform and cuboidal in appearance and grow as monolayers to a high saturation density and are morphologically similar to several human breast cancer cell lines (22).When evaluated in monolayer cultures in situ, HER2 receptors are diffusely distributed throughout the cytoplasm and as aggregates adjacent to the plasma membrane.Established cultures express uniformly high levels of HER2 on their cell surface.
We further characterized the Bam1a cell line using microarray analysis.Lobular carcinomas from BALB-NeuT transgenic animals have been reported to have a gene expression profile that resembles the genetic signature of human ErbB-2 breast cancers of the basal subtype (23,24) that correlates with aggressive disease and poor prognosis.The Bam1a cell line has a similar expression pattern as determined by whole-genome analysis.Highly expressed genes are reflected in Supplementary Table S1.When compared with Universal mouse RNA on the Agilent whole-mouse genome chip containing 44,000 genes, 4,823 gene sequences were identified as being overexpressed (>2-fold) in Bam1a when compared with the universal mouse RNA prep.We defined the Bam1a transcriptome of 1,285 (1,045 unique) known genes (Supplementary Table S1) and compared it with the published profiles of mouse mammary tumor virus (MMTV)-neu mammary tumors from two different strains (23,25) and genes characteristically expressed in human ErbB-2 breast and basal subtype cancers (24,26).For comparative purposes, Supplementary Table S2 lists the genes commonly overexpressed in Bam1a and MMTV-neu tumors and cell lines as reported by Astolfi (ref. 23; Supplementary Table S2A) and genes overexpressed in Neu mammary tumors relative to agematched glands as reported by Landis (ref. 25; Supplementary Table S2B).The table also includes several genes from the Bam1a profile that are present in the ''intrinsic'' gene list published by Sorlie et al. (24) that was used to define histologic subtypes of human breast cancers (Supplementary Table S2C) and Bam1a genes associated with the human ErbB-2 amplicon as described by Bertucci et al. (ref. 26; Supplementary Table S2D).These data support the use of the 44K mouse CGH Agilent gene chip and universal mouse RNA as a screen for defining the genetic signature of HER2/neu-expressing mouse mammary tumor cell lines and shows the similarities among Bam1a, independent Neu-expressing mammary tumors and cell lines, and human ErbB-2 breast cancers.Additional breast cancer and HER2 relevant genes that are overexpressed in Bam1a cells and diverse human breast cancer cell lines (22) include ERBB-3, STARD10, ADAMTS8, FGF1, FGF9, BTC, AREG, EPGN, CD44, CYP2J6, ITGB4, LAMA4, and STFA1.
We determined the effect of gefitinib on this basal gene expression pattern and identified 1,976 genes with a consistent (P > 0.05) change in expression level that was >2-fold.Based on known functions, we reduced this list to 504 (416 unique) genes with gefitinib-induced alterations in expression; 384 with decreased expression and 120 with increased expression and is provided as Supplementary Table S3.Table 1 lists unique genes with greatest level of modulation by gefitinib.In this gefitinib-sensitive cell line, genes regulating cell cycle and associated processes, HER2 tumor Figure 1.A (i-iv), histochemical analysis of lobular carcinoma used to derive the Bam1a cell line (i, iii ) and the outgrowth of Bam1a tumor cells in vivo (ii, iv ).Histochemical analyses of tumors using H&E (i and ii ).Immunohistochemical analyses of the HER2 receptor (iii and iv ) were done on serial paraffin sections as described in Supplementary Methods.Histologically, the parental tumor was found to be a large, single proliferation, with vaguely multilobular architecture, each composed of smaller acini, separated by fine stroma, coalescing in varying degrees, and focally forming tightly packed, solid nodules.Acini were mostly devoid of lumens, or have fine ones, and outlines are ill defined, with lobules expanding irregularly adjacent adipose tissue.By immunohistochemistry, tumor cells showed (iii ) strong and diffuse expression of HER-2, in a membranous pattern, whereas the other histologic components were negative.Bam1a cell line grown in vivo (ii, iv ) in syngenic hosts were morphologically similar to the parental.HER-2 staining is strong and diffuse in a membranous pattern.Magnification, Â40.B, effect(s) of gefitinib on ErbB receptor phosphorylation.Bam1a cells were grown to 80% confluence and subsequently treated by changing growth medium with fresh medium containing increasing levels of gefitinib and cultured for 4 or 24 h.Cells were harvested and lysates were extracted and processed as described in Supplementary Methods.Whole-cell lysates (20 Ag/lane) were resolved in 4% to 20% SDS-PAGE.Blots were probed with the indicated phosphospecific antibodies and pan actin was used as a loading control.Doses of gefitinib (Amol/L concentration) are indicated above each lane.C, effect of gefitinib on MAPK signaling.Bam1a cells were grown to 80% confluence and subsequently treated by changing growth medium with fresh medium containing increasing levels of gefitinib and cultured for 4 or 24 h.Cells were harvested and lysates were extracted and processed as described in Supplementary Methods.Whole-cell lysates (20 Ag/lane) were resolved in 4% to 20% SDS-PAGE.Blots were probed with the indicated phosphospecific antibodies and then reprobed with antibodies against total proteins.These data are representative of at least four independent preparations of whole-cell lysates.The trends are highly reproducible and consistent.Immunofluorescence photomicrographs: In parallel, Bam1a cells were grown on glass coverslips, treated with a medium change containing 0 Amol/L (i-iii ) or 1 Amol/L (iv-vi ) gefitinib for 4 h, fixed with methanol, and stained for HER2 in green (i and iii ), 4 ¶,6-diamidino-2-phenylindole in blue (ii and v ) or phospho-MAPK in red (iii and vi ).Photographs were taken with the Â100 objective under oil immersion.D, effect of gefitinib on SAPK signaling.Bam1a cells were grown to 80% confluence and subsequently treated by changing growth medium with fresh medium containing increasing levels of gefitinib and cultured for 4 or 24 h.Cells were harvested and lysates were extracted and processed as described in Supplementary Methods.Whole-cell lysates (20 Ag/lane) were resolved in 4% to 20% SDS-PAGE.Blots were probed with the indicated phosphospecific antibodies and then reprobed with antibodies against total proteins.Doses of gefitinib (Amol/L concentration) are indicated above each lane.Immunofluorescence photomicrographs: In parallel, Bam1a cells were grown on glass coverslips, treated with a medium change containing 0 Amol/L (i ) or 1 Amol/L (ii ) gefitinib for 4 h, fixed with methanol, and stained for phospho-SAPK in red.Photos were taken with the Â100 objective under oil immersion.biology and progression are the primary targets for downregulation.In particular, KI67, CYCLINB1, CHEK1, MMP1A, KIF11, CKAP2, AREG, SKP2, COL1A1, FGF18, MAP2K6, STFA1, ETV4, S100A14, LGALS9, BIRC5, and CELSR2.Genes that were upregulated by gefitinib included oncogenes, and genes involved in matrix remodeling, drug metabolism, antiproliferative genes, heat shock, and DNA damage.These included ADAMTS15, CTSB, HYAL3, HSPB1, OSMR, OSM, NOTCH1, BGAL1L, BTG2, BMF, DHRS6, GADD45B, PDGFRA, TRP53INP1, FLT1, NOX4, FMO2, TIMP3, WNT5B, KIT, LTF, FGFR2, CEBPD, CYP2F2, C3, SULT1A1, and SLPI.
Several have been confirmed by independent microarray Western blot.Gene signatures are becoming increasingly important in designing treatment strategies that use agents that target specific signal transduction pathways (27).In the case of gefitinib, several studies have identified sensitive and resistant profiles that correlate with responsiveness (28,29).These profiles will be discussed when we compare Bam1a with its resistant variant.
Effect of gefitinib on HER2, ErbB-3, and EGFR receptor phosphorylation and downstream signaling.Although EGFR is the preferred target for gefitinib, several groups have shown the effectiveness of this agent against a variety of HER2-overexpressing human cell lines, including lung (30) and breast (31).We first determined the concentration of gefitinib necessary to eliminate phosphorylation of the HER2/neu, EGFR, and ErbB-3 in Bam1a cells.
As shown in Fig. 1B, treatment with gefitinib for 4 or 24 h had a dramatic effect on phosphorylation of all ErbB family members.Distinct down-regulation of receptor autophosphorylation sites is evident with 1 Amol/L IR in whole-cell lysates at 4 h of treatment.When treated for 24 h with increasing doses of gefitinib, autophosphorylation of the activated HER2 signaling domain decreased in a dose-dependent fashion.Phosphorylation of Tyr-1248 and Tyr-1221 of the HER2 was reduced by f50% between the doses of 1 to 2 Amol/L, as one would expect based on the IC 50 (Fig. 1B).EGFR is expressed at a lower level than the HER2 oncogene in these cells and is, as expected, more sensitive to gefitinib (Fig. 1B).Phosphorylation on EGFR Tyr-1173 and Tyr-992 was diminished by f50% by 1 Amol/L but was not completely eliminated.
It has been well documented that in tissues where ErbB2 is mutated or overexpressed, it serves as the dominant signaling receptor due to its promiscuous heterodimerization and impaired endocytosis (32).The residual phosphorylation that is detected on the EGFR may represent phosphorylation that persists in HER2/ EGFR heterodimers.Phosphorylation on Tyr-1289 of HER3, the kinase inactive family member was most sensitive probably due to its dependence on heterodimerization and transphosphorylation.At higher doses of gefitinib, total ErbB-3 receptor levels were clearly up-regulated, suggesting enhanced receptor stabilization in the presence of gefitinib.This was even more pronounced after 24 h.Tyrosine phosphorylation on HER2 Tyr-877, the Src phosphorylation site was least sensitive to gefitinib-induced suppression, consistent with retention of Tyr-416 phosphorylation on Src (the active state; Fig. 3A) at the same concentrations of gefitinib.Cross-talk from another signaling pathway that is insensitive to gefitinib suppression may modulate these interactions.Although significant decrease in phosphorylation was evident with 1 Amol/L IR on various tyrosine residues in all three receptors, 6 Amol/L IR was required for near-complete elimination of phosphotyrosines.After 24 h, phosphorylation recovered slightly at the lowest doses *Average fold difference is the ratio (n = 3, P value cutoff >0.05) of Bam1a RNA from gefitinib-treated cells (1 Amol/L, 24 h) to Bam1a RNA from sham-treated cells.RNA from gefitinib-treated cells was labeled in red (cy5); RNA from sham-treated cells was labeled in green (cy3).Three independent RNA preparations per condition were analyzed on three separate chips and averaged; a confidence cutoff of >0.95 was used for the raw values of red/green before normalization.The fold change (**treated/**control) = normalized ratio value(cy5/cy3), fold change cutoff >2 fold (up-regulate), or <1/2 (down-regulate).The P value represents the consistency between the three replicate samples.
of gefitinib (1-2 Amol/L) but dropped sharply as total HER2 and EGFR levels both decreased coincident with their ubiquitination and proteasomal degradation (not shown).Downstream of the ErbBs, mitogenic signaling through MAP/ extracellular signal regulated kinase (ERK) kinase 1/2 (MEK1/2) and p44/42 MAPK was greatly diminished.Phosphorylation of MEK1/2 and MAPK was completely eliminated between 1 and 2 Amol/L within 4 h in whole-cell lysates and cytoplasm and remained absent over 24 h (Fig. 1C) despite the retention of phosphotyrosines in the autophosphorylation sites of the ErbBs (Fig. 1B).We also detected a 2-fold increase in cell surface HER2/neu expression in gefitinib-treated cells by flow cytometry (not shown) and membrane accentuation by immunofluorescence (Fig. 1C).This is similar to our finding in salivary gland carcinoma (20).
We next evaluated the effect of gefitinib on stress signaling through the stress-activated protein kinase (SAPK) pathway and observed a rapid activation of SAPK and JNK in the presence of gefitinib within 4 h.Using 1 Amol/L gefitinib for 4 h, we observed that active JNK and c-Jun (not shown) were eliminated from the cytoplasm and translocated to the nucleus (Fig. 1D).After 24 h, active JNK (p46) was no longer observed but activation of SAPK (p54) persisted.Signaling to the serum response factor c-Jun, which is downstream of both MAPK and JNK, was greatly impaired.Failure to activate c-Jun would eliminate transcription of genes modulated by serum and growth factor signaling.
Gefitinib induces cytostasis through multiple mechanisms.Consistent with the observed suppression of mitogenic signaling, gefitinib effectively inhibited proliferation of Bam1a cells in a time-and dose-dependent manner.Within 48 h, a 50% reduction in the rate of proliferation was observed in the presence of 250 to 500 nmol/L gefitinib (Fig. 2A).Inhibition of anchorage-independent growth required 1 Amol/L gefitinib (Fig. 2B).Cell cycle analysis showed the onset of cytostasis within 24 h of treatment with gefitinib (Fig. 2C).Most studies have characterized the effects of gefitinib on cell cycle as G 0 arrest (14).Similar to our observations in salivary gland carcinoma overexpressing HER2/neu (20), the effect of gefitinib on Bam1a cell cycle kinetics is primarily due to inhibition of S-phase entry and execution.At 1 Amol/L, when maximum inhibition is achieved within 24 h, there is retention of cells in G 0 (49-71%), a sharp decrease in the S-phase population (23-2%) but no change in the G 2 M fraction (28-28%).This cell cycle profile is consistent with the effects of gefitinib on cell cycle regulators controlling the various phases of the cell cycle as shown in Fig. 2D.We have obtained qualitatively and mechanistically similar results using the irreversible inhibitor There is a sharp decline in S-phase activity between the doses of 0.25, 0.5, and 1.0 Amol/L gefitinib (solid columns ; 22%, 9%, and 1.4%, respectively).Cell cycle histograms depict cell cycle distribution after 24 h in the absence (left ) or presence (right ) of 1 Amol/L gefitinib.The increase in G 0 -G 1 from 49% to 71% coincides with a proportional decrease in S phase from 23% to 1.4%, whereas G 2 -M activity is unaffected (i.e., 28-28%).D, Western blot analysis of whole-cell lysates for cell cycle control proteins in Bam1a cells after 4 or 24 h in the presence of increasing concentrations (Amol/L) of gefitinib as indicated above each lane.

Cancer Research
Cancer Res 2007; 67: (14).July 15, 2007 PD168393 (11) with respect to HER2/neu phosphorylation, signal transduction, proliferation, and cell cycle (data not shown).We suspect that similar small-molecule inhibitors with specificities for the ATP-binding pocket of the EGFR and/or ErbB-2 (i.e., erlotinib, lapatinib) will have efficacy against this intrinsically sensitive breast cancer cell line.
Cell cycle arrest was multifactorial and somewhat complex.To determine mechanisms involved in cytostasis, we evaluated the key cell cycle regulators controlling each phase and checkpoint of the cell cycle (Fig. 2D).We observed a significant decrease in the levels of cell cycle regulators cyclin D1, phospho Ser795 Rb, phospho Y15 cdc2 within 4 h of treatment.By 24 h, complete inhibition of these regulators and cyclin B1 is achieved with 1 Amol/L gefitinib, the dose and time at which MAPK and Akt signaling is silenced (Figs.1C and 3A) and cytostasis is achieved (Fig. 2A and C).Changes in these cell cycle regulators explain the distribution of cells in the different phases of the cell cycle.Cyclin D1 levels drop significantly due to ubiquitin-mediated degradation (data not shown) and increased levels p27 reinforce inhibition of cyclin D1 and progression of the G 0 -G 1 phase.Furthermore, loss of cyclin D1-dependent activities (by cdk4/6) and suppression cdk2 by increased levels of p27 would result in hypophosphorylation of Rb and impaired cell cycle progression through G 0 .Downstream of these events, hypophosphorylated Rb binds to E2F and impairs DNA synthesis and S-phase progression.As a result, the S-G 2 transition does not occur.The hypophosphorylation of cdc2 leads to cyclin B1 ubiquitination and degradation (Fig. 2D) to impair the G 2 -M transition.Cell cycle histograms depict cell cycle distribution after 24 h in the absence (Fig. 2D, left) or presence (Fig. 2D, right) of 1 Amol/L gefitinib.Cytostasis is characterized by (a) a decrease in the percentage of cells in S phase (23-1.4%),(b) accumulation of cells in G 0 (49-71%), and (c) retention of cells G 2 -M (i.e., 28-28%).Therefore, G 0 arrest and impaired G-S phase transition was due to decreased levels of cyclin D1 and increased levels of p27 and hypophosphorylation of Rb causing a retention of cells in G 0 .However, accumulation of additional cells in G 0 (true G 0 arrest) did not occur due to impaired transitioning through the G 2 -M checkpoint, due to decreased cyclin B1 levels.
We also have overwhelming support from our microarray data that cell cycle regulators are significantly modulated at the transcriptional level.CYCLIN B1 is a primary target of downregulation and is decreased >30-fold by a 24 h treatment with 1 Amol/L gefitinib relative to cells receiving control medium (Table 1; Supplementary Table S3).Other cyclins, including A2, B2, E1, and E2, and the gene encoding KI67 were also transcriptionally suppressed.Because this treatment is effective in reducing phosphorylation of HER2/neu, MAPK signaling, proliferation, and cell cycle, one would expect alterations in gene expression levels to have a major contribution to the growth and cell cycle suppression induced by gefitinib and vice versa.
Effects of gefitinib on survival signaling and apoptosis.Concentrations of gefitinib >1 Amol/L failed to generate additional therapeutic benefit with respect to proliferation and cell cycle arrest but higher concentrations were needed to effectively impair anchorage-independent growth and survival.Inactivation of survival signaling by gefitinib is usually associated with inhibition of HER2/EGFR and HER2/HER3 activity (30,31) and phosphatidylinositol 3-kinase leading to apoptosis through the intrinsic (mitochondrial) pathway (33).
The activity of the serine/threonine protein kinase Akt/PKB is modulated by various growth and survival factors.Akt promotes cell survival through two distinct pathways: inhibition of proapoptotic (death) signals and activation of IKK-a signaling to p65 nuclear factor-nh (NF-nh).In Fig. 3A, within 4 h of treatment with gefitinib, we observed reduction of Akt phosphorylation.At this early time point, the doses (4-6 Amol/L) required to diminish Akt phosphorylation were higher than those needed to inhibit HER2/neu (1-2 Amol/L) or MAPK (1 Amol/L).However, within 24 h, complete inhibition of Akt phosphorylation was achieved with 1 Amol/L.Inhibition of phosphorylation of signal transducers and activators of transcription 1 (STAT1) and STAT3 was similar to that of the EGFR and probably represents crosstalk between these receptors through the MAPK pathway.In some instances, STATs can serve an autocrine function to induce cytokine expression and rescue cells undergoing growth factor deprivation or EGFR inhibition (34).Phosphorylation of NF-nh p65 was not adversely affected by gefitinib and perhaps slightly enhanced.
Impaired growth and survival signaling was tightly coupled to alterations in proapoptotic BH3-only mitochondrial proteins bim and bad.In the context of HER2 overexpression, bim is constitutively suppressed through a MAPK-dependent mechanism (35).Suppression of EGFR/HER2 signaling by gefitinib leads to induction of bim, a rheostat for sensing growth factor deprivation.We previously reported the induction of bim by gefitinib in salivary gland carcinoma (20).The early induction of bim by gefitinib in Bam1a cells at 4 h (Fig. 3B) could be mediated through inhibition of MAPK as is seen with MEK inhibitors.Bim has also been implicated in anoikis and was recently shown to be sensitive to modulation by ERK and MEK inhibitors (35).Transcript levels for BTG2 and BMF were also increased by gefitinib (Table 1) whereas survivin transcript levels were decreased.Diminished phosphorylation of bad on serine residue 112, which is under MAPK control, provides additional support for a dominant role of the mitogenic pathway in regulation of the bcl-2 family.As a result of the increased mitochondrial stress generated by gefitinib exposure, the onset of apoptosis through the caspase cleavage pathway is detectable within 24 h over a range of doses (Fig. 3B).Within 24 h of treatment (1 Amol/L gefitinib), 33% of cells undergo apoptosis, as determined by Annexin binding.Apoptosis increased in a timeand dose-dependent manner coincident with caspase-3 and PARP cleavage observed within 24 h at 2 Amol/L gefitinib in whole-cell lysates (Fig. 3B).
The ability of gefitinib to efficiently induce apoptosis in HER2/ neu-overexpressing breast cancer is in contrast to our observations in salivary gland carcinoma where induction of bim and hypophosphorylation of bad by gefitinib did not result in caspase cleavage and apoptosis.Salivary gland carcinoma exhibited an intrinsic resistance to gefitinib.The primary difference between these two activated HER2/neu-overexpressing models is in survival signaling through the Akt pathway.Our data indicate that in Bam1a breast cancer cells, which express high levels of HER3, unlike salivary gland carcinoma, HER2 signaling is tightly coupled to the Akt survival pathway.Several studies have shown that coexpression of HER2 and HER3 improves tumor cell sensitivity to gefitinib and other RTKIs by coupling it to the phosphatidylinositol 3-kinase pathway to facilitate down regulation of the survival signaling pathway (30).In cell lines where gefitinib impedes the Akt/survival signaling pathway (33) apoptosis occurs through the intrinsic pathway.
Novel targets of gefitinib in breast cancer.We also observed that Bam1a cells express hyperphosphorylated form of the gap junction protein connexin43 (cx43) in culture.Hyperphosphorylated cx43 impairs GJIC and cell-cell coupling and this form of the protein has been shown to be up-regulated in breast hyperplasias and carcinomas and neoformed capillaries (36).Upon treatment with gefitinib, there is a marked reduction in the phosphorylated form of cx43 within 4 h (Fig. 3B).By 24 h, cx43 protein is no longer detected.Regulation of cx43 by gefitinib probably represents several distinct mechanisms.Direct mechanisms of gefitinib that are mediated through tyrosine kinase and MAPK activities can alter cx43 phosphorylation, transcription, and degradation.Indirect mechanisms that can modulate cx43 may occur through the effects of gefitinib on the dynamics of cell cycle, cell growth, adhesion, and cytoskeletal remodeling.Our microarray analyses did not reveal a change in cx43 expression but did indicate that two other gap junction genes (GJB4 and GJC1) were sensitive to modulation at the transcriptional level along with multiple adhesion molecules and cytoskeletal kinases that affect connexin trafficking, stability, and assembly.
The c-fos-inducible lymphangiogenic factor, VEGF-D (37), is also dramatically reduced at 1 Amol/L gefitinib within 24 h.Loss of activated c-Jun in the cytoplasm and nucleus of Bam1a treated with gefitinib may contribute to decreased expression of VEGF-D.Many groups have reported the antiangiogenic effects of gefitinib as decreases in VEGF (13,38) and/or interleukin-8 (39), but there have been no reports of gefitinib effects on expression of VEGF-D.These two novel targets of gefitinib, cx43 and VEGF-D, may have important implications in the clinical management of breast cancer.Disruption of cx43 inhibits breast cancer diapedesis (40) and disruption of VEGF-D impairs lymphangiogenesis, providing a novel mechanism for abolishing tumor cell extravasation and metastasis.
Effectiveness of gefitinib in vivo.To determine our ability to effectively target HER2 growth and survival signaling in vivo, we challenged mice with Bam1a tumor cells and tested the ability of gefitinib to suppress tumor outgrowth and shrink established tumors.We originally observed that 50 mg/kg gefitinib was able to suppress the outgrowth of small palpable Bam1a nodules when administered daily for 5 days a week.This dose was effective for 3 weeks, at which time nodules began to increase in size in the 50 mg/kg treatment group, subsequent treatment of these animals with 100 mg/kg was able to impede tumor growth (not shown).Animals bearing established Bam1a tumors were effectively treated with 100 mg/kg gefitinib as this dose administered 5 days a week for 4 weeks eliminated tumor burden (Fig. 3C).Finally, we established that we could effectively target HER2 in naturally occurring mammary LCIS tumor-bearing female transgenic mice and detected reduced levels of phosphorylated HER2, HER3, MAPK, and Akt phosphorylation in tumor biopsies taken from animals treated for 5 consecutive days with 100 mg/kg gefitinib (Fig. 3D).These findings establish that we can indeed target HER2/neu signal transduction pathways in naturally occurring tumors in vivo.
These preliminary findings using the Bam1a cell line show the effectiveness of gefitinib in targeting the signaling pathways downstream of the HER2/neu oncogene and mechanisms of action.The acquisition of resistance in this intrinsically sensitive cell line could represent the development of mutations in the target receptor that reduce drug binding, biochemical uncoupling of receptor kinase activity from specific growth and survival signaling pathways, and/or alterations in gene expression that minimize the role of RTKs in tumor cell survival.The issues of intrinsic and acquired resistance or hypersensitivity to RTK inhibitors have been addressed by several investigations.Recently, several somatic and acquired mutations in the EGFR (41,42) and HER2 (43) have been identified, which alter responsiveness to gefitinib and other RTKIs (44,45).These mutations dictate levels of intrinsic sensitivity or resistance at the level of the target receptor.In some instances, the outgrowth of resistant subpopulations in previously sensitive tumors has shown the presence of additional receptor mutations that dramatically reduce the IC 50 (44).The intrinsic resistance in tumors lacking receptor mutations occurs through other biochemical mechanisms that uncouple receptor phosphorylation and catalytic activity from downstream signaling to growth and survival targets and usually relate to constitutively active survival signaling intermediates or loss of negative regulatory factors.
Characterizing the mechanism(s) responsible for acquired resistance also generates a valuable tool for dissecting and mapping interactions and the potential for the development of autocrine or paracrine pathways that emerge as a consequence of chronic pathway suppression or attenuation.This is supported by the observation that the mechanisms responsible for the acquired resistance to the ErbB2 TKI, lapatinib, in breast cancer cells, reflects a shift in the sole dependence of cell survival on the ErbB-2 to a codependence on the ErbB-2 and estrogen receptor via activation of factors that enhance the transcriptional activity of the estrogen receptor (46).
In the case of gefitinib resistance, EGFR receptor mutations that alter the ATP-binding cleft differentially stabilize or destabilize gefitinib binding and competition for ATP, leading to altered hydrolysis and catalytic rate of the kinase (47).The L858A mutation in the activation loop increases kinase activity and sensitivity to gefitinib (41).In subpopulations, patients with the L858A mutation that originally respond to gefitinib, outgrowth of resistant tumors harbor a second mutation at codon 790.This T790M mutation in the hinge region decreases sensitivity to gefitinib f100-fold (48).Several somatic exon 20 mutations in HER2 have also been identified in lung adenocarcinomas.Insertion of YVMA at codon 776 in the HER2 kinase domain confers resistance to gefitinib and a gain of function (49).
Acquired resistance to gefitinib in Bam1a cells.We developed a variant of the Bam1a cell line from a soft agar colony that grew in the presence of gefitinib.Cells that were recovered from this colony were continuously exposed to medium supplemented with gefitinib.Concentrations of gefitinib were gradually increased over time.The clone that we developed, designated IR-5, grows in 5 Amol/L gefitinib.Compared with the parental cell line, IR-5 cells display a disorganized growth pattern in culture and a lack of contact inhibition.Cells express lower levels of HER2/neu on their cell surface compared with the parental cell line (Fig. 4A).To characterize the mechanisms associated with the acquired resistance to gefitinib, we first compared the effect of gefitinib on HER2 receptor phosphorylation and heterodimerization in the parental Bam1a cell line and the resistant IR-5 cell line.As shown in Fig. 4B, HER2 phosphorylation is eliminated in Bam1a cells treated with 5 Amol/L gefitinib and preserved in IR-5 cells up to 15 Amol/L gefitinib.Decreased phosphorylation of HER2 and EGFR correlated with an increase receptor migration in SDS-PAGE in whole-cell lysates and HER2 immunoprecipitates.Phosphorylation of ErbB-3 is not detected in Bam1a treated with gefitinib.In control Bam1a cells, ErbB-3 that is present in HER2/ErbB-3 heterodimers is weakly phosphorylated and gefitinib completely eliminates this phosphorylation.In IR-5, phosphorylation of ErbB-3 is detected at all concentrations of gefitinib.HER2/ErbB-3 heterodimers are more abundant in IR-5 cells compared with Bam1a cells.Phosphorylation of ErbB-3 in HER2/neu heterodimers is slightly reduced in IR-5 treated with 15 Amol/L gefitinib and correlates with the reduction in HER2/neu phosphorylation observed at this dose.Similarly, we observe increased migration of EGFR in HER2 heterodimers in IR-5 treated with 15 Amol/L gefitinib, suggesting decreased EGFR phosphorylation.In Bam1a cells, EGFR and HER2/EGFR heterodimers show increased migration upon treatment with gefitinib.
Others have shown that HER2 kinase domain mutations show increased kinase activity over the wild-type receptor and increased association with and phosphorylation of EGFR and ErbB-3 (49).These mutant HER2 also form gefitinib-resistant ErbB-3 and EGFR heterodimers and confer a malignant phenotype.Similar to these studies, we found that EGFR homodimers in IR-5 cells were also less sensitive to inhibition by gefitinib than in Bam1a cells (Fig. 4C).Phosphotyrosine is not detected in EGFR homodimers from Bam1a treated with gefitinib and total EGFR is dramatically reduced and coimmunoprecipitates with hsp60, indicative of receptor downregulation.The HER2 that coimmunoprecipitates with EGFR in these cells is also down-regulated.In IR-5 cells, EGFR homodimers are less sensitive to gefitinib-induced down-regulation than the parental cell line.We observe a slight decrease in EGFR phosphorylation with 5 Amol/L gefitinib but this does not result in increased EGFR migration or association with hsp60.Treatment with 15 Amol/L was effective in decreasing EGFR phosphorylation and increasing receptor migration and coupling to hsp60.ErbB-2 that coimmunoprecipitates with EGFR are also down-regulated.These data show a log-shift in the HER2/neu sensitivity to gefitinib in IR-5 cells and resembles resistance that occurs through receptor mutations.Acquired resistance to imatinib is associated with T334I mutation of c-Abl.Acquired resistance to gefitinib is associated with T790M mutation of the EGFR (48).Pharmacophore modeling has shown that the sites of these mutations align.It is proposed that T790M in the hinge region of the ATP-binding pocket causes ligand to escape (47).
Gefitinib-induced mutation of the HER2/neu.We sequenced the HER2/neu from the parental and IR-5 cell lines to determine whether similar mutations had emerged that could explain the observed resistance.We detected a single nucleotide point mutation of C to A (Note: G to T on the antisense strand shown in Fig. 4D) that changes the conserved leucine at codon 726 to an isoleucine relative to the human erbB-2 (Fig. 4D).This sequence is conserved between human and rat and aligns with a homologous region in exon 18 of the human EGFR.These regions encode a portion of the ATP-binding pocket that is highly conserved among the human EGFR and erbB-2 and the orthologous rat HER2/neu transgene (Fig. 4D).No additional mutations in or around the transmembrane region or the kinase domain were detected.When aligned with the EGFR, this residue is within hydrophobic region II that represents the hydrophobic slot of the ATP-binding pocket.It is possible that this mutation is sufficient to alter the competition between ATP and gefitinib in favor of ATP, leading to hydrolysis, receptor autophosphorylation, and kinase activity.
Several mutations have been detected in this region of the EGFR (50).We denote two in particular that correspond to enhanced sensitivity (i.e., G719S) and resistance (i.e., E709G) to gefitinib in the clinic and in vitro (50,51).Thus, our novel mutation in the HER2/ neu corresponds to a site in the EGFR that is frequently mutated in C, Western blot analysis of EGFR immunoprecipitated lysates from Bam1a (right ) or IR-5 (left ) treated as described above and probed for the indicated antigens.D, identification of a HER2/neu point mutation in IR-5 cells.Genomic DNA from Bam1a (top ) or IR-5 (bottom ) was PCR amplified and sequenced using ABI PRISM 3100.Electropherograms compare nucleotide sequences of the antisense strand of the rat HER2/neu kinase domain from the parental Bam1a cell line (top ) and the gefitinib-resistant IR-5 variant (bottom ).A single point mutation is detected (C/G to A/T) encoding a leucine to isoleucine mutation at the conserved codon 726 of the human ErbB-2 orthologue.Homologous sequences from the human EGFR and human ErbB-2 (identical in the rat HER2/neu orthologue) are aligned.Dashes, mismatches; *, point mutations at specific codons that have been reported to correlate with responsiveness to gefitinib (50,51).R, resistance; S, sensitizing.
human cancers with differential sensitivity and resistance to gefitinib.We are not aware of any reports of a similar mutation in this region of ErbB-2 that correlate with intrinsic or acquired sensitivity or resistance to specific inhibitors.We suspect that decreased sensitivity of the EGFR and ErbB-3 to gefitinib (in IR-5 cells Fig. 4B and C) is mediated through the mutated HER2/neu and not the result of mutations in either of these receptors.This is supported by the fact that mutant HER2 with resistance to EGFR RTKIs harbor a gain of function that endows them with the capacity to heterodimerize and transphosphorylate kinase-dead EGFR and ErbB-3 (49).
Altered signaling activity of the mutated HER2/neu and/or selection in the presence of gefitinib may have lead to additional genetic changes that are commonly seen in HER2-expressing cell lines with intrinsic resistance to gefitinib.To characterize the mutated receptor and resistant phenotype, we compared the gene expression profiles of the parental Bam1a cell line and IR-5 under various treatment conditions and the effects of gefitinib on signal transduction pathways, cell cycle, and apoptosis in IR-5 cells.
We used the 44K mouse CGH array to generate an expression profile of the IR-5 cell line and compared it with the gene expression profile of the sensitive, parental cell line, Bam1a.Using a 2-fold cutoff limit and P > 0.05, we identified 1,642 divergent genes between the two cell lines.From this, we generated a list of 475 (430 unique) known genes that were differentially expressed by 2-fold or greater.This comprehensive gene list is provided as Supplementary Table S4 and a selection of these genes is given in Table 2. Several striking differences in the gene expression profiles of these two cell lines include the overrepresentation of Notch signaling-related genes in IR-5 cells as well as the presence of several genes that have recently been linked to clinical resistance to gefitinib.Notch pathway genes that are increased in IR-5 include NOTCH-1 (6-fold), NOTCH-3 (2.3-fold), JAG-1 (3.5-fold), JAG-2 (5.6fold), MTAP1B (3.8-fold), MMP7 (7.7-fold), CBL (2.9-fold), NFjb1 (2.1-fold), CD44 (2.1-fold), and b-CATENIN (2.8-fold).Overexpression of activated Notch-1 and Notch-3 has been shown to induce mammary tumor formation in mice (52).Additional up-regulated genes that may contribute to the resistant phenotype include FLT-1, PDGFB, MAF, CAV1, and EFNA1.In breast cancer models with acquired resistance to the ErbB-2-specific kinase inhibitor, lapatinib, resistant tumor cells amplify signaling through the estrogen receptor pathway, leading to the up-regulation of Foxo3a and caveolin (46).Although we also detect a 3.5-fold increase in caveolin in gefitinib-resistant IR-5 cells, a functional role for this protein in our estrogen receptor-negative cell line is not predicted.Thus, in the context of HER-2-overexpressing estrogen receptornegative breast cancer, acquired resistance to gefitinib amplifies genes involved in the Notch signaling pathway.
By examining the genes that are down-regulated in IR-5 and abundantly expressed in the parental cell line, it is possible to detect a trend in the genes that correlate with intrinsic sensitivity to gefitinib.For instance, elevated expression of the EGFR, phosphatases (DUSP18, PPP2R2B); drug-metabolizing enzymes (ABCG2, EPHX2, CYP2J6, and NOX4) and apoptosis and cell cycle regulators (TRAF1, STK17B, P15, and GSPT2) in Bam1a cells may facilitate the antiproliferative and proapototic activity of gefitinib.Indeed, gene profiling of NSCLC cell lines with mutant and wildtype EGFR was used to identify genes that could globally predict gefitinib sensitivity or resistance (28).We compared our list of differentially expressed genes to this data set and found concordance between several genes, attesting to the robustness of these specific markers, as they predict drug sensitivity in diverse tumor targets (lung and breast) of differing etiologies and primary oncogene dependence (EGFR and HER2).The three genes that were commonly up-regulated in IR-5 and gefitinib resistant lung cancers included MAP1B, CDH2, and GAS5, and two genes that were preferentially expressed in Bam1a and gefitinib-sensitive lung cancers included TMEM30B and ALDH1A3.These similarities suggest that resistance developed by exposure to gefitinib amplifies genes involved in intrinsic resistance to gefitinib.
In the case of acquired resistance, one biomarker, EMP-1, has been identified in vivo and validated as a clinical marker of de novo and acquired resistance to gefitinib (29).Serial passage and continued exposure of a HER2-overexpressing adenocarcinoma to gefitinib in vivo resulted in the emergence of a gefitinib-resistant variant with a >10-fold increase in expression of EMP-1.Clinical samples from gefitinib-treated patients correlated with this observation.We found that relative to the parental Bam1a cell line, IR-5 cells with acquired in vitro resistance to gefitinib have a 6.6-fold increase EMP-1 and a 2.1-fold increase in EMP-2.Our data support a role for EMP-1 expression in the acquisition of resistance to gefitinib in a HER-2-overexpressing adenocarcinoma.
Reviewing trends in gene expression data, we observed that several genes that were differentially expressed between Bam1a and IR-5 were also targets that were sensitive to modulation by gefitinib in Bam1a cells.We found that 57 genes (11.2%) that were modulated by gefitinib in Bam1a cells were also differentially expressed between Bam1a and IR-5.In this comparison (Supplementary Table S5), 33 of the gefitinib-induced changes in Bam1a corresponded to the difference between Bam1a and IR-5, whereas 24 were inconsistent.It is interesting to conceive that this subset of gefitinib-responsive genes may have an active role in driving and maintaining resistance to gefitinib.
Because resistance to gefitinib in the IR-5 cell line is not complete with respect to the phosphorylation of the HER2/neu, we used high doses of gefitinib to characterize mechanism(s) of responsiveness at the transcriptional and signal transduction levels to determine the similarities and differences between Bam1a and IR-5.Treatment of IR-5 cells with 10 Amol/L gefitinib induced change in expression levels of 338 genes, 200 known genes, and 190 unique when compared with basal gene expression of cells cultured in 5 Amol/L gefitinib.This complete list is provided as Supplementary Table S6 and includes 14 up-regulated genes and 184 down-regulated genes.We found that several genes that were sensitive to high-dose gefitinib in IR-5 cells were also sensitive when an irreversible EGFR inhibitor was tested against gefitinib-resistant, EGFR mutant cell lines (53).These genes included AREG, CENPA, CCNB1, DEPDC1A, FOSL1, HMMR, KNTC2, NEK2, NUSAP1, and SHCBP1 and are deemed to be critical in antiproliferative and antisurvival response in gefitinib-resistant cells bearing mutant HER family members.Although gefitinib-resistant cells can still respond to EGFR/HER-2targeted therapies, the extent of the response is blunted due to compromised signaling activities of the mutant receptors per se and genetic changes that provide cells with compensatory mechanisms and survival pathways similar to those associated with intrinsic drug resistance.The extent to which these cell lines still depend on the HER2/EGFR for growth and survival will determine the utility of targeting the HER2 axis to control these diseases and understanding the mechanism(s) of resistance will indicate treatment modalities that could be successful.
Relative to the parental Bam1a cell line, the number of gefitinibresponsive genes in IR-5 cells and the level of modulation is  To evaluate the effect of gefitinib in the culture medium of IR-5 cells, we determined the gene expression pattern of IR-5 cells after gefitinib washout.We observed 382 genetic changes consisting of 176 known and 168 unique genes.To our surprise, the majority, 154 genes, were down-regulated and only 22 genes were found to be increased.These data suggest that the presence of gefitinib is *Average fold difference compares the ratios (n = 3, P value cutoff > 0.05) of genes expressed by Bam1a versus IR-5 cells using the data obtained from Bam1a RNA normalized to Universal Mouse RNA versus IR-5 RNA normalized to Universal Mouse RNA.RNA from the cell lines was labeled in red (cy5), Universal RNA is labeled in green (cy3).Three independent RNA preparations per condition were analyzed on three separate chips and averaged; a confidence cutoff of >0.95 was used for the raw values of red/green before normalization.The fold change (**Bam1a/**Universal Mouse) = normalized ratio value(cy5/cy3).

Cancer Research
Cancer Res 2007; 67: (14).July 15, 2007 required for maintaining the expression of several genes.If the presence of gefitinib in the culture medium was responsible for the suppression of specific genes, we would expect that elimination of gefitinib would restore gene expression.This was the case for TOPIIA (3.25), LPD (2.31), MMP10 (3.65), MMP1A (2.32), SERPINA11 (4.14), and USP26 (2.89).Two of these genes, MMP1A and LPD, are also sensitive to suppression by high-dose gefitinib.The phenotypic consequences of these genetic changes was further evaluated by characterizing the effects of high-dose gefitinib treatment on IR-5 compared with the response pattern profile of the parental cell line.
Effects of gefitinib on IR-5 cells.In IR-5 cells, treatment with 12 Amol/L gefitinib for 24 h was required to diminish phosphorylation of HER2 (Fig. 5A).At this dose, ErbB-3 phosphorylation and EGFR levels were also reduced.Interestingly, c-Cbl levels were increased in IR-5 cell lysates compared with Bam1a.This correlates with the increased (2.9-fold) gene expression levels in IR-5 cells (Table 2).However, phosphorylation of cbl was dramatically reduced in IR-5, even in the absence of gefitinib.Dephosphorylated cbl abrogates CIN85 binding and receptor internalization, leading to enhanced receptor stabilization.These data may indicate altered trafficking of the mutated HER2 in IR-5 cells.Mutant EGFR has been shown to preferentially associate with underphosphorylated cbl to impede internalization (54).Although we are able to achieve considerable down-regulation of receptor phosphorylation with 15 Amol/L of gefitinib in IR-5 cells, MAPK and STAT3 phosphorylation persist (Fig. 5B).Cross-talk between pathways or establishment of an autocrine loop (34) capable of signaling through these coupled effectors are likely to be responsible for this observation.This in contrast to the profile generated by Bam1a at 2 Amol/L where residual phosphorylation on HER2 is still detected and the phosphorylation of MAPK is eliminated.On the other hand, in IR-5 cells, decreased phosphorylation of c-Jun and S6 still correspond to the dose-response for HER2 phosphorylation, suggesting that these signaling pathways remain tightly coupled to HER2 activity.Preserved signaling through the MAPK pathways suggests decreased dependence on the HER2 for growth signals and the amplification/development of another source coupled to the MAPK pathway.
We observe differential sensitivity of several downstream targets of HER2 in response to the gefitinib-induced decrease in HER2 phosphorylation.For example, VEGF-D modulation by gefitinib is no longer observed; this may be secondary to retention of MAPK signaling (Fig. 5B).Similarly, changes in protein levels of the cell cycle regulators p Ser795 Rb, cyclin D1, cyclin B1, and p27 were less dramatic relative to the parental cell line (Fig. 5C).CYCLIN B1 transcript levels were reduced 2.2-fold in IR-5 treated for 24 h with 10 Amol/L gefitinib.This reduction in cyclin B1 is 11-fold lower than that achieved by 1 Amol/L gefitinib in Bam1a cells.Cell cycle analysis shows that 9 Amol/L IR is required to impair cell cycle kinetics of IR-5 cells (Fig. 5D) when compared with the cytostasis achieved with 1 Amol/L IR in Bam1a (Fig. 2C).Consistent with retention of MAPK signaling and cell cycle regulators, a complete elimination of the S-phase population and G 2 block was not observed in IR-5 cells.This cytostatic response in IR-5 cells was incomplete and transient, as an increase in the S-phase population was observed at all doses of gefitinib within 48 h (not shown).We also evaluated the integrity/fidelity of signaling to the survival pathway through Akt and apoptosis.Akt phosphorylation was modestly reduced at 15 Amol/L gefitinib at 24 h in IR-5 cells compared with complete elimination in Bam1a with 2 Amol/L (Fig. 5B).The observed induction of bim expression at this dose (15 Amol/L gefitinib; Fig. 5C) in IR-5 cells suggests a connection between bim expression, HER2/neu phosphorylation, and the Akt pathway.
Acquired resistance to gefitinib in HER2-overexpressing breast, prostate, and gastric cancers has been attributed to the upregulation of growth factor receptors [i.e., insulin-like growth factor-IR (55) and EGFR (56)] to create compensatory signaling pathways that couple to MAPK and drive cell growth.Under this paradigm, it is reasonable to suggest that the coupling of Notch signaling pathways to MAPK and Akt (57) is a potential mechanism contributing to the resistance observed in IR-5 cells.
Elimination of gefitinib from the culture medium also restored cx43 expression in whole-cell lysates extracted from IR-5 cells, but this expression level is substantially lower than that observed in the parental Bam1a cells, suggesting altered regulation of cx43 in IR-5 cells.Microarray analysis suggested that transcript levels of GJA-1 were not sensitive to changes in gefitinib concentration 0 or 10 Amol/L gefitinib compared with 5 Amol/L gefitinib.Finally, induction of apoptosis in IR-5 cells required treatment with >15 Amol/L gefitinib for at least 48 h compared (Fig. 5D) with 24-h incubation of Bam1a cells with 1 Amol/L gefitinib (Fig. 3B).The equivalent level of apoptosis that is achieved with 1 to 2 Amol/L gefitinib between 24 and 48 h in Bam1a cells requires 15 Amol/L gefitinib for 48 to 72 h in IR-5 cells.
These data show that mammary tumor cells overexpressing the activated rat HER2-neu have a high level of intrinsic sensitivity to gefitinib via inhibition of receptor signaling through the MAPK and Akt pathways in vitro and in vivo.Continuous exposure of these tumor cells results in the acquisition of gefitinib and the ability to grow in the presence of 5 Amol/L gefitinib.Acquired resistance to gefitinib in Bam1a cells is associated with a novel mutation in the ATP-binding pocket of the HER2 that alters its sensitivity to gefitinib.Resistance is characterized by a decreased fidelity in the signaling pathways from HER2 to MAPK and Akt and is phenotypically similar to aberrant signal transduction that has been observed in cells with intrinsic resistance to gefitinib.The upregulation of constitutively active survival factors also contribute to the resistant phenotype observed in IR-5 cells.Genes that are differentially expressed and regulated by gefitinib in Bam1a and IR-5 cells have been associated with intrinsic and acquired resistance to gefitinib and the amplification of genes involved in Notch signaling.These data suggest that acquired resistance to gefitinib can be treated by strategies that target genes/pathways used to achieve HER2 independence.
The relationship between the receptor mutation and resistant phenotype will be tested by cloning and transfection of the wildtype and resistant receptors in a genetic background that has not been selected for resistance.Aspects of acquired and intrinsic resistance are being investigated by using specific chemical inhibitors and small interfering RNA that target NOTCH-1, PDGFB, FLT-1, and several other genes to validate their functional role(s) in mediating resistance to gefitinib.

Figure 2 .
Figure 2. Effect of gefitinib on proliferation (A ) and anchorage-independent growth (B) in Bam1a cells.Bam1a cells were monodispersed and seeded in the presence of increasing concentrations of gefitinib or diluent (DMSO).Metabolic activity of quadruplicate wells was evaluated at 24 h intervals by the addition of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT ).Points, mean for each respective concentration of gefitinib at each time point; bars, SE.Cytostasis occurs between 250 and 500 nmol/L of gefitinib.Similarly, Bam1a were monodispersed and suspended in soft agar containing varying amounts of gefitinib and subcultured for 14 d.Box, mean for the total number of colonies per well; bars, SE; solid symbols, individual data (n = 3).C and D, cell cycle distribution of Bam1a treated for 24 h with gefitinib.Percentage of cells in G 0 -G 1 (shaded columns ); S phase (solid columns); and G 2 -M (striped columns ).There is a sharp decline in S-phase activity between the doses of 0.25, 0.5, and 1.0 Amol/L gefitinib (solid columns ; 22%, 9%, and 1.4%, respectively).Cell cycle histograms depict cell cycle distribution after 24 h in the absence (left ) or presence (right ) of 1 Amol/L gefitinib.The increase in G 0 -G 1 from 49% to 71% coincides with a proportional decrease in S phase from 23% to 1.4%, whereas G 2 -M activity is unaffected (i.e., 28-28%).D, Western blot analysis of whole-cell lysates for cell cycle control proteins in Bam1a cells after 4 or 24 h in the presence of increasing concentrations (Amol/L) of gefitinib as indicated above each lane.

Figure 3 .
Figure3.A and B, effect of gefitinib on survival signaling and apoptosis.Bam1a cells were grown to 80% confluence and subsequently treated by changing growth medium with fresh medium containing increasing levels of gefitinib and cultured for 4 or 24 h.Cells were harvested and lysates extracted and processed as described in Supplementary Methods.Whole-cell lysates (20 Ag/lane) were resolved in 4% to 20% SDS-PAGE.Blots were probed with the indicated phosphospecific or cleavage-specific antibodies and then reprobed with antibodies against total proteins.Doses of gefitinib (Amol/L concentration) are indicated above each lane.B, gefitinib suppresses cx43 and VEGF-D, alters bcl-2 family members, and induces proteolytic cleavage leading to apoptosis in Bam1a cells.Cell treatment, lysate preparation, and Western blot were as described (above).For apoptosis assays, cells were treated for 24 to 48 h with various doses of gefitinib and evaluated by Annexin staining.Columns, percentage Annexin V-positive cells detected by flow cytometry after 24 h (open columns ) or 48 h (solid columns ).C and D, the effect of gefitinib on Bam1a tumor growth and signal transduction in vivo .Tumor-bearing mice were treated with 100 mg/kg gefitinib by oral gavage for 5 consecutive days per week for 4 wks beginning on day 25 after tumor cell injection (50,000 cells) when palpable tumor lesions reached 5 mm.Tumor growth was monitored weekly and measured with calipers in two perpendicular diameters.Boxes, mean size range of five animals per group; solid symbols, individual tumor area measurements; whiskers, SE.D, Western blot analysis of total tumor lysates.Female Balb-NeuT transgenic mice bearing multiple macroscopic lobular carcinomas were treated for 5 consecutive days with 100 mg/kg gefitinib (IR) or diluent (0) by oral gavage and sacrificed 2 h after the last dose.Tumor lysates (50 Ag) were resolved in SDS-PAGE, transferred to polyvinylidene difluoride membranes, and probed for the indicated phosphospecific antigens.

Figure 4 .
Figure 4. A, comparison of parental (Bam1a, right ) and gefitinib resistant (IR-5, left) cells by photomicroscopy of cell growth and morphology in vitro (top ) and by flow cytometric analysis of cell surface HER2/neu expression (bottom ).Phase-contrast photomicrographs (Â20 objective) were taken of confluent cultures of Bam1a (right) and IR-5 (left) cells.Histograms of Bam1a (right ) or IR-5 (left ) stained for an epitope in the extracellular domain of the rat HER2/neu.Filled histograms, specific antibody-stained cells detected with phycoerythrin-labeled secondary; clear overlay, cells stained with nonimmune IgG and phycoerythrin-conjugated secondary.The mean channel fluorescent values for Bam1a and IR-5 were 1,185 and 432, respectively.B and C, effect of gefitinib on ErbB receptor phosphorylation and heterodimerization in Bam1a and IR-5 cells.Bam1a and IR-5 cells were treated for 4 h by replacing medium with the indicated concentrations of gefitinib.Whole-cell lysates (25 Ag, right ) or ErbB-2 immunoprecipitates (IP ; left ) were resolved in 4% to 12% SDS-PAGE.Western blots were probed for the indicated antigens.C, Western blot analysis of EGFR immunoprecipitated lysates from Bam1a (right ) or IR-5 (left ) treated as described above and probed for the indicated antigens.D, identification of a HER2/neu point mutation in IR-5 cells.Genomic DNA from Bam1a (top ) or IR-5 (bottom ) was PCR amplified and sequenced using ABI PRISM 3100.Electropherograms compare nucleotide sequences of the antisense strand of the rat HER2/neu kinase domain from the parental Bam1a cell line (top ) and the gefitinib-resistant IR-5 variant (bottom ).A single point mutation is detected (C/G to A/T) encoding a leucine to isoleucine mutation at the conserved codon 726 of the human ErbB-2 orthologue.Homologous sequences from the human EGFR and human ErbB-2 (identical in the rat HER2/neu orthologue) are aligned.Dashes, mismatches; *, point mutations at specific codons that have been reported to correlate with responsiveness to gefitinib(50,51).R, resistance; S, sensitizing.

Figure 5 .
Figure 5. A, sensitivity of ErbB receptor phosphorylation to increasing doses of gefitinib in IR-5.Western blot analysis of whole-cell lysates from IR-5 or Bam1a cells treated for 24 h with the indicated concentrations of gefitinib after medium change.Lysates (25 Ag/well) were resolved in 4% to 20% SDS-PAGE and membranes were probed for the indicated antigens.B, effect of gefitinib on mitogenic and survival signaling in IR-5 cells.Western blot analysis of whole-cell lysates from IR-5 or Bam1a cells treated for 24 h with the indicated concentrations of gefitinib after medium change.Lysates (25 Ag/well) were resolved in 4% to 20% SDS-PAGE and membranes were probed for the indicated antigens.C, effect of gefitinib on cell cycle and apoptosis proteins in IR-5 cells.Western blot analysis (top ) of whole-cell lysates from IR-5 or Bam1a cells treated for 24 h with the indicated concentrations of gefitinib after medium change.Lysates (25 Ag/well) were resolved in 4% to 20% SDS-PAGE and membranes were probed for the indicated antigens.D, cell cycle distribution of IR-5 treated for 24 h with gefitinib (top ).Percentage of cells in G 0 -G 1 (shaded columns ), S phase (solid columns ), and G 2 -M (striped columns ).Induction of apoptosis in IR-5 cells treated for 48 and 72 h with gefitinib (bottom ).Cells were treated for 48 to 72 h with various doses of gefitinib and evaluated for apoptosis by Annexin staining.Percentage Annexin V-positive cells detected by flow cytometry after 48 h (open columns ) or 72 h (solid columns ).

Table 1 .
Gefitinib-induced changes in Bam1a gene expression