DNA Vaccination Controls Her-2⁺ Tumors that Are Refractory to Targeted Therapies

Her-2/neu⁺ tumor cells refractory to antibody or receptor tyrosine kinase inhibitors are emerging in treated patients. To investigate if drug resistant tumors can be controlled by active vaccination, gefitinib and antibody sensitivity of four neu⁺ BALB/c mouse mammary tumor lines were compared. Significant differences in cell proliferation and Akt phosphorylation were observed. Treatment-induced drug resistance was associated with increased chromosomal aberrations as shown by spectral karyotyping analysis, suggesting changes beyond neu signaling pathways. When mice were immunized with pneuTM encoding the extracellular and transmembrane domains of neu, antibody and T-cell responses were induced, and both drug-sensitive and drug-resistant tumor cells were rejected. In T-cell–depleted mice, drug-sensitive tumors were still rejected by vaccination, but drug-refractory tumors survived in some mice, indicating their resistance to anti-neu antibodies. To further test if T cells alone can mediate tumor rejection, mice were immunized with pcytneu encoding full-length cytoplasmic neu that is rapidly degraded by the proteasome to activate CD8 T cells without inducing antibody response. All test tumors were rejected in pcytneu-immunized mice, regardless of their sensitivity to gefitinib or antibody. Therefore, cytotoxic T lymphocytes activated by the complete repertoire of neu epitopes were effective against all test tumors. These results warrant Her-2 vaccination whether tumor cells are sensitive or resistant to Her-2–targeted drugs or antibody therapy. [Cancer Res 2008;68(18):7502–11]

ErbB-2/Her-2/neu, a member of the ErbB receptor tyrosine kinase family, is weakly to moderately expressed in normal adult tissues. Dysregulated signal transduction from overexpressed or mutated Her-2 leads to cellular immortalization, neoplastic transformation, and tumor progression (1). Overexpression of Her-2/neu in 20% to 30% of human breast cancers is correlated with more aggressive disease and reduced survival (2–5). Upon dimer formation with other ErbB family members, transphosphorylation of tyrosine residues in the COOH terminus provides docking sites for signal transduction and adaptor molecules, leading to cell proliferation, migration, adhesion, and transformation (6–9). The activated signaling involves phospholipase C, mitogen-activated protein kinase, c-src, and the phosphatidylionsitol-3 kinase-Akt pathways.

Trastuzumab (Herceptin), a humanized murine monoclonal antibody (mAb; 4D5), which binds to Her-2 at the juxtamembrane region of domain IV, is used to treat metastatic Her-2⁺ breast cancer or in adjuvant settings for patients with less advanced disease (10). When tested as a monotherapy, trastuzumab showed 15% response rate with 9.1 months median duration (11). The efficacy of trastuzumab can be increased by combining with paclitaxel with 59% response rate and 10.5 months duration (12). Several mechanisms may account for trastuzumab activity, such as down-modulation of Her-2, disruption of downstream signaling (13, 14), or induction of antibody-dependent cell-mediated cytotoxicity (ADCC). Cardiac dysfunction is the most significant toxicity, particularly when combined with selected chemotherapy. A common failure site in patients treated with trastuzumab is the brain. In a retrospective review, 25% of metastatic breast cancer patients with prolonged survival from trastuzumab treatment developed brain metastases (15, 16). Taken together, these results indicate the need for intervention before the stage of metastatic disease using multiple treatment modalities.

Tyrosine kinase inhibitors lapatinib and gefitinib are candidate therapeutics for Her-2⁺ breast cancer. Lapatinib is a dual receptor tyrosine kinase inhibitor (RTKI) with selective inhibitory activity for epidermal growth factor receptor (EGFR) and Her-2 (17–19). Another RTKI, gefitinib, selective for EGFR, exerts inhibitory activity on Her-2 signaling at higher concentrations (20, 21). In treating Her-2⁺ tumors, mAb, RTKI, and Her-2 vaccine may have overlapping or complementary activities.

Because Her-2/neu⁺ tumor cells refractory to antibody or RTKI are emerging in treated patients, we tested whether Her-2 DNA vaccination would be effective against drug-sensitive versus drug-resistant tumors. We assembled a panel of rat neu–expressing tumors that are differentially sensitive to anti-neu mAb and RTKI. The neu oncogene, identified in ethylnitrosourea-induced rat neuroglioblastomas, is a homologue of human ErbB-2 (22). Expression of constitutively activated neu in transgenic mice was associated with spontaneous tumorigenesis (23). Neu⁺ tumor cell lines with varying levels of drug sensitivity were established from NeuT spontaneous tumors or by transfection of hormone-induced tumors to mimic breast cancer cells with varying levels of responsiveness to Her-2–targeted therapies.

Mice. All animal procedures were conducted in accordance with accredited institution guidelines and the US Public Health Service Policy on Humane Care and Use of Laboratory Animals.³

BALB/c (ages 6–8 wk) female mice were purchased from Charles River Laboratory. BALB NeuT (NeuT) mice expressing a transforming neu under the control of mouse mammary tumor virus promoter were provided by Dr. Guido Forni (University of Torino; ref. 24). Female NeuT mice developed spontaneous mammary tumors around ages 17 wk. Male NeuT mice developed salivary gland tumors when they were ∼7 mo old. Heterozygous NeuT mice were maintained by mating with BALB/c mice and transgene-positive mice were identified by PCR.

Cell lines and reagents. All tissue culture reagents were purchased from Invitrogen unless otherwise specified. Cell lines were cultured in 5% CO₂, and maintained in vitro in DMEM supplemented with 10% heat-inactivated fetal bovine serum (Sigma), 10% NCTC 109 medium, 2 mmol/L l-glutamine, 0.1 mmol/L MEM nonessential amino acids, 100 units/mL penicillin, and 100 μg/mL streptomycin.

TUBO (24) was cloned from a spontaneous mammary tumor in a BALB NeuT (NeuT; ref. 25) mouse. TUBO grew progressively in wild-type BALB/c mice and gave rise to tumors that were histologically similar to autochthonous tumors in BALB NeuT females. Bam1a cell was established in soft agar from another BALB NeuT spontaneous mammary tumor, then maintained as a cell line in monolayer culture. Bam IR-5 variant was derived from Bam1a by culturing in increasing concentrations of gefitinib until stable growth was achieved in the presence of 5 μmol/L gefitinib (26). Gefitinib [Iressa; ZD1839; 4-(3-chloro-4-fluoroanilino)-7-methoxy-6- (3-morpholinopropoxy) quinazoline; Zeneca Pharmaceuticals] is a RTKI.

Antigen-presenting cells (APC) 3T3/KB and 3T3/NKB were generated as previously described (27). Briefly, BALB/c NIH 3T3 fibroblasts were transfected with K^d and B7.1 (KB), or with K^d, B7.1, and neu (NKB). Stable clones were selected and maintained in medium supplemented with 0.8 mg/mL G418 and 7.5 μg/mL of puromycin (3T3/KB) or 0.8 mg/mL G418 and 0.8 mg/mL of zeocin (3T3/NKB).

D2F2 was derived from a mouse mammary tumor that arose in a BALB/c hyperplastic alveolar nodule line, D2 (28). D2F2 cells were cotransfected with pRSV/neo and pCMV/neu, which encodes wild-type rat neu to establish D2F2/neu (29). Transfected cells were maintained in medium supplemented with 0.8 mg/mL of G418 (Geneticin; Invitrogen).

DNA immunization. pcDNA/neuTM encoding the extracellular and transmembrane domains of rat neu was previously described (24).

pCMV/cytneu (pcytneu) was constructed by deleting the endoplasmic reticulum (ER) signal sequence from pCMV/neu with a PCR strategy (30). The first 684 bp of the protein-coding region excluding the ER signal sequence was amplified using the high-fidelity DNA polymerase Pfu (Stratagene). The upper primer α, 5′-GCGGGGGAGCTCCGCCACCATGGGCACCCCAAGTGTGTAC-3′, is homologous to the Kozak consensus ribosome-binding site (Kozak, 1986), the initiation codon ATG and 15 bp immediately downstream from the ER signal sequence but excludes the 72-bp signal sequence itself. The lower primer β, 5′-GTGGAGGCAGGCCAGGCAGTCAGAATGC-3′, contains a naturally occurring BsmI site. This PCR product was digested with SacI and BsmI and used to replace the corresponding region in pCMV/neu to generate the plasmid pCMV/cytneu (pcytneu). The recombinant cytneu is designed to direct the synthesis of a cytoplasmic protein.

pEFBos/granulocyte macrophage colony-stimulating factor (pGM-CSF) encoding murine GM-CSF was provided by Dr. N. Nishisaka at Osaka University, Osaka, Japan. pCMV is the control empty vector. Mice were injected in the quadriceps muscle with plasmid DNA as previously described (30). I.m. DNA injection was followed immediately by square wave electroporation over the injection site using a BTX830 (BTX Harvard Apparatus) as we previously described (29). A tweezer electrode was used to deliver 8 pulses at 100 V for 25 ms per pulse.

T-cell depletion. To deplete CD4 or CD8 T cells, mice received i.p. GK1.5 or 2.43 mAb (American Type Culture Collection), respectively, in the form of ascites fluid. Mice were treated once or twice before tumor challenge and then once to twice per week until completion of the experiment. T-cell depletion was verified by fluorescence-activated cell sorting analysis using peripheral blood leukocytes (PBL).

Tumor challenge. Mice were challenged s.c. with 2.5 × 10⁵ (TUBO or D2F2/neu) or 5.0 × 10⁵ (Bam1a or Bam IR-5) cells in the flank. Tumor growth was monitored by weekly palpation, and mice were sacrificed when any one dimension of the tumor reached 20 mm. Differences in tumor incidence were analyzed by the Log-rank test.

Measurement of anti-neu antibody by flow cytometry. To measure anti-neu antibody, 3T3/NKB cells were incubated with serially diluted immune mouse sera. Briefly, phycoerythrin (PE)-conjugated goat-anti-mouse antibody directed to the γ-chain of mouse IgG (Jackson ImmunoResearch) was used to detect bound antibody by flow cytometry. Normal mouse serum or isotype-matched mAb was the negative control. A standard curve for neu binding was generated using c-erbB2/c-neu mAb (Ab4, clone 7.16.4; Calbiochem). Antibody concentrations in the test sera were calculated by linear regression analysis based on the standard curve as we previously reported (31). Flow cytometric analysis was performed with a FACSCalibur (Becton Dickinson). Differences in antibody concentration were analyzed by the Student's t test.

Measurement of IFN-γ–secreting T cells by ELISPOT assay. Neu reactive T cells were measured by ELISPOT assay (29). PBL or spleen cells were suspended in RPMI 1640 supplemented with 10% FCS, 2 mmol/L l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. A total of 4 × 10⁵ spleen cells were added to each well of a 96-well HTS IP plate (Millipore), which was precoated with 2.5 μg/mL rat anti-mouse IFN-γ (clone R4-6A2; BD PharMingen). 3T3/NKB cells were added as APC over the spleen cells. 3T3/KB cells were used as control. The ratio of spleen cells to APC was 10:1. After 48 h of incubation, cells were removed and 2 μg/mL biotinylated rat anti-mouse-IFN-γ (clone XMG 1.2; PharMingen) was added. Plates were incubated for another 12 h at 4°C, then washed to remove unbound antibody. Bound antibody was detected by incubating the plates with 0.9 μg/mL avidin-horseradish peroxidase (PharMingen) for 2 h at room temperature. The substrate 3-amino-9-ethylcarbazole (AEC; PharMingen) in 0.1 mol/L acetic acid and 0.003% hydrogen peroxide was added, and the plate was incubated for 3 to 5 min. AEC solution was discarded, and the plates were washed six times with water. The visualized cytokine spots were enumerated with the ImmunoSpot analyzer (CTL), and the results were expressed as the number of cytokine producing cells per 10⁶ cells. Data were analyzed using the Student's t test.

Proliferation assay. Cell proliferation was measured indirectly by mitochondria metabolic activity using a modified 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrasodium bromide (MTT) assay (32). Briefly, quadruplicate wells of cells were plated in flat-bottomed 96-well plates at 10,000 cells per well. Approximately 16 h after plating, when cells reached 40% to 50% of confluence, dilutions of anti-neu mAb 7.16.4 or gefitinib were added. Replicate plates were terminated at 24, 48, 72, 96, and 120 h posttreatment. At each time point, 10 to 20 μL of 5 mg/mL MTT in PBS were added and incubated for 4 h at 37°C before the stop reagent (Isopropanol with 0.04N HCl) was added and the absorbance measured at 600 to 650 nm. Differences between treatments were analyzed by Student's t test.

Measurement of cell surface neu and MHC class 1 K^d expression by flow cytometry. Cells were stained with anti-neu mAb 7.16.4 or anti-K^d mAb SF1.1, followed by PE-conjugated affinity pure F(ab')₂ fragment goat anti-mouse IgG Fc (Jackson ImmunoResearch) at 1:200 on ice for 25 min. After staining, samples were washed twice and analyzed by flow cytometry using a FACSCalibur (Becton Dickinson) flow cytometer. Between 10,000 and 20,000 events were collected for each sample, and the data were analyzed using flowjo software package.

Western blot analysis. Whole cell lysates were prepared as previously described after cells were incubated with indicated concentrations of gefitinib or anti-neu mAb or immune sera (26). Equal amounts of protein were loaded onto a 10% SDS gel and transferred to Immobilon-P membranes. After blocking, primary antibodies to phospho-Akt (Cell Signaling) were added to detect the target proteins. The membrane was washed and labeled for 90 min at room temperature with Peroxidase-conjugated AffiniPure Goat anti-mouse or goat anti-rabbit secondary Abs (Jackson ImmunoResearch Laboratories). The membranes were developed with enhanced SuperSignal West Pico Chemiluminescent Substrate (Pierce Biotechnology, Inc.) and imaged with Kodak-MR film. For normalization, blots were stripped and reprobed for β-actin.

Spectral karyotyping analysis. Cells were treated with colcemid for 2 h, and chromosomal slides were prepared using the standard protocols including hypotonic treatment, fixation, and air drying as we described (33). After pepsin treatment and fixation with formaldehyde followed by dehydration, the chromosomal slides were denatured and hybridized with denatured painting probes (SkyPaint) for 48 h at 37°C. After slide washing, the chromosomes were counter stained with 4′,6-diamidino-2-phenylindole (DAPI) and mounted with antifade (34). Thirty to 50 mitotic figures from each cell line were randomly selected for spectral karyotyping (SKY) images. The selection criterion were high-quality hybridization signals and minimal chromosomal overlapping. The presence of chromosomal abnormalities was not a selection criterion. This procedure was used for both the spectral and DAPI images.

Chromosomes were karyotyped by the color and the size of each chromosome using software developed by Applied Spectral Imaging, Inc. A unique single color was assigned to individual chromosomes. The presence of mixed colors along a chromosomal arm indicated a translocation (indicated by arrows). Clonal chromosomal aberrations (CCA) referred to identical or closely related chromosomal aberrations occurring at least twice in 20 to 50 mitotic figures (35). When CCAs dominated a cell population, the tumor was considered more stable and homogenous. The levels of non-CCAs (NCCA) were defined by occurrence at a frequency of <4% among 50 to 100 mitotic figures and correlated with overall genetic instability, indicating a nonlinear pattern of cancer initiation and progression (35).

Characterization of neu⁺ tumor cell lines. The efficacy of neu targeted therapy was tested with neu-expressing BALB/c mammary tumor lines TUBO, Bam1a, Bam IR-5, and D2F2/neu. TUBO (24) was established from a NeuT spontaneous mammary tumor (25). Bam1a was derived independently from another NeuT spontaneous mammary tumor by soft agar culture (26). Bam IR-5 was derived from Bam1a by continuous culture in increasing concentrations of gefitinib to become resistant to 5 μmol/L of gefitinib. D2F2/neu was generated by cotransfecting BALB/c mammary tumor cell D2F2 (28) with pCMV/neu and pRSV/neo (29). All test cells, except D2F2, expressed neu as measured by flow cytometry using anti-neu mAb 7.16.4 (Fig. 1A,, top row, open histogram). In addition, all test cells expressed MHC class I, K^d, (Fig. 1A , bottom row, open histogram). These cells may represent Her-2/neu–positive breast cancer cells eligible for Her-2/neu–targeted therapy.

Inhibition of tumor cell proliferation by RTKI and anti-neu mAb. The effect of RTKI and anti-neu mAb on cell proliferation was measured by MTT assay after 24 to 120 hours of incubation with the test agent (Fig. 1B). Reduction of TUBO cell proliferation was observed at 1 μmol/L or higher concentrations of gefitinib. Reduction in Bam1a cell proliferation was observed at 0.25 μmol/L gefitinib (Fig. 1B). Although both cell lines were derived from spontaneous NeuT mammary tumors, Bam1a cells were more sensitive to RTKI. Bam IR-5 and D2F2/neu cells continued to proliferate even in the presence of 4 μmol/L gefitinib, showing their resistance to RTKI, whether acquired or intrinsic.

To test the direct effect of anti-neu mAb, tumor cells were incubated with 0.025 to 2.5 μg/mL of anti-neu mAb. TUBO and Bam1a cell proliferation was significantly inhibited by 0.25 μg/mL or higher concentrations of mAb (P < 0.05) with greater inhibition observed in Bam1a cells. Inhibitory activity was substantially reduced in Bam IR-5 because 2.5 μg/mL mAb was required to induce growth inhibition. D2F2/neu cells were not affected, even at 2.5 μg/mL. Therefore, cells that are resistant to RTKI are resistant or less sensitive to anti-neu mAb.

Effect of targeted therapy on neu expression and neu-mediated signaling. Inhibition of tumor cell proliferation by RTKI or anti-neu Ab may be mediated via a common or different pathway and may indicate how tumor cells would respond to Her-2/neu vaccination. The effects on neu expression and Akt phosphorylation were analyzed. Cells were incubated with 4 μmol/L of gefitinib for 2 to 24 hours before expression of neu on the cell surface was measured by flow cytometry. Figure 2A shows representative results from 8 hours of culture. Treatment with gefitinib increased neu expression on TUBO and Bam1a cells, with a very modest effect on Bam IR-5 and no effect on D2F2/neu cells. This was consistent with our earlier observation by Western blotting showing sustained total Her-2/neu level in gefitinib-treated Bam1a cells despite a marked reduction in phosphorylated Her-2/neu (26). Therefore, inhibition of TUBO and Bam1a cell proliferation by RTKI was not due to loss of neu expression.

Incubation with 2.5 μg/mL anti-neu mAb, on the other hand, down-modulated neu expression on TUBO, Bam1a, and Bam IR-5 cells, and this may contribute, in part, to reduced tumor growth (Fig. 1B). Consistent with this finding is our observation by Western blot analysis that antibody binding induces degradation of activated Her-2/neu in Bam1a and BamIR-5 cells (data not shown).

Because Akt phosphorylation was critical in neu-mediated signaling, it was measured by Western blotting. Figure 2B shows representative results from 2-hour incubation with gefitinib or anti-neu mAb. In TUBO and Bam1a cells, Akt phosphorylation was reduced by gefitinib treatment, with modest effect on Bam IR-5 and no effect on D2F2/neu cells (26). Therefore, gefitinib dampened neu-mediated signaling in TUBO and Bam1a cells, correlating with the growth inhibitory effect of the drug. Reduction of phospho-Akt was also induced by anti-neu mAb in TUBO and Bam1a cells, consistent with the growth inhibitory activity (Fig. 1B). A modest effect was induced in Bam IR-5, consistent with the modest growth inhibition by mAb (Fig. 1B). D2F2/neu cells that were resistant to RTKI and mAb showed little change in Akt phosphorylation after treatment by either agent, indicating the use of independent signaling pathways for their growth and survival.

SKY analysis of neu-expressing cell lines. The differential sensitivity of TUBO, Bam1a, Bam IR5, and D2F2/neu cells to gefitinib and antibody may represent intrinsic or induced changes of Her-2/neu⁺ tumor cells during therapy. Changes beyond neu-mediated signaling may render the tumors refractory to other therapies. Because genomic instability has been correlated with tumor progression and disposition to drug resistance, neu-expressing tumor cells were subjected to SKY analysis (Fig. 3). As chromosomal instability escalates, the cell population forms various combinations of NCCAs/CCAs (36). Table 1 shows chromosomal aberrations in neu-expressing tumor lines, including CCAs, NCCAs, and chromosome fragments.

Although both TUBO and Bam1a were derived from NeuT spontaneous mammary tumors, significant disparities exist in their karyotypes. CCAs with chromosomal translocations t(15;17), t(4;10), and t(4;13) were found only in TUBO cells. Similarly, NCCAs and chromosome fragments were found in TUBO, not Bam1a cells, showing more severe chromosomal aberrations in TUBO cells, correlating with their reduced drug sensitivity (Fig. 1).

Bam IR-5 cells, which are resistant to gefitinib and anti-neu mAb showed one CCA, 15% NCCAs, and many abnormal centromeres (data not shown). Bam IR-5 cells also exhibited more trisomies of chromosome 6, 13, and 16 and monosomies of chromosome 18 and 19 relative to the parental Bam1a (Fig. 3). It is not clear if these aberrations occurred after gefitinib treatment or if cells with pre-existing aberrations survived the drug selection. Regardless, continuous treatment with RTKI resulted in a cell population with drug resistance and more profound chromosomal aberrations.

When comparing D2F2/neu with D2F2 cells, they shared one CCA at t(11;16) and chromosome fragments X, 13, 3, and 1 (Table 1), verifying the common origin of these cells. But the transfected D2F2/neu cells exhibited a more stable karyotype. Therefore, the process of generating the transfected cell line selected for a less aberrant clone, which expressed neu at a high level. Despite a more stable karyotype, D2F2/neu remained refractory to Her-2–targeted drugs and mAb.

These four cell lines, all expressing neu, showed divergent sensitivity to RTKI and anti-neu antibody as a result of genetic aberrations and different usage of signaling pathways. In human cancers, Her-2–positive cells with heterogeneous karyotypes may coexist. Upon treatment with Her-2/neu–targeting agents, the selection pressure would eliminate sensitive cells, whereas those with greater chromosomal aberrations and drug resistance may expand.

Inhibition of tumor growth by neu DNA vaccination. To test if Her-2/neu vaccination can inhibit tumor growth regardless of their chromosomal aberrations and resistance to neu-targeted therapy, BALB/c mice were electrovaccinated twice with pGM-CSF and pneuTM encoding the extracellular and transmembrane domains of neu (Fig. 4). At 2 weeks after the second immunization, an average of 19.2 ± 9.5 μg/mL anti-neu antibody was induced (Fig. 4A). T-cell response was measured by IFN-γ ELISPOT assay using PBL. After incubation with 3T3/NKB cells that were engineered to express neu, K^d, and B7.1 (CD80), 150 ± 50/10⁶ PBL IFN-γ–secreting cells were detected in immunized mice, significantly higher than that in control mice receiving pCMV vector (Fig. 4B).

To determine if anti-neu immunity from DNA vaccination protected mice from tumor growth, immunized mice were inoculated s.c. with TUBO, Bam1a, Bam IR-5, or D2F2/neu. In mice that received control vectors, 100% of the inoculated mice developed progressively growing tumors, and mice were sacrificed when tumors reached 20 × 20 mm (Fig. 4C). In pneuTM-vaccinated mice, complete protection was achieved against every challenging tumor: TUBO, Bam1a, Bam IR-5, or D2F2/neu. Therefore, neu DNA vaccination protected mice from neu-expressing tumors, regardless of their chromosomal aberration or resistance to neu-targeted therapy.

To distinguish the effect of anti-neu Ab on different tumor cells, CD4⁺ and CD8⁺ T cells in some immunized mice were depleted with specific mAb before tumor cell inoculation and then at weekly intervals throughout the experiment. Complete protection against TUBO and Bam1a tumors was achieved even after T-cell depletion, showing the activity of vaccine-induced anti-neu antibody against TUBO and Bam1a cells. Protection against Bam IR5 and D2F2/neu, however, was compromised when T cells were depleted. In 3/7 Bam IR-5 and 4/6 D2F2/neu-inoculated mice, tumors grew progressively. Therefore, T-cell immunity may not be required in controlling drug-sensitive tumors but are indispensible for controlling drug-resistant tumors, although anti-neu Ab may contribute to tumor rejection also by direct suppression of Her-2 signaling or facilitation of ADCC.

Bam IR-5 and D2F2/neu tumors that grew in vaccinated, T-cell–depleted mice were excised (Fig. 4D), plated in monolayer culture, and their expression of neu was measured. Bam IR-5 tumor outgrowth expressed comparable levels of neu as cells maintained in culture. But D2F2/neu cells completely lost neu expression after growing in mice. These results indicate that neu-mediated signaling is still critical to the survival of Bam IR-5 cells, thus the sustained expression, yet dispensible in D2F2/neu cells, which may use alternate signaling pathways for survival.

Effect of pcytneu DNA vaccine on tumor growth. To test if cytotoxic T lymphocytes (CTL) alone could control neu⁺ tumors, pcytneu vaccine was generated to encode cytoplasmic neu with a truncated ER signal peptide. Similar to pcytE2 that we previously reported (37), pcytneu was released into the cytoplasm upon synthesis and rapidly degraded by the proteasome (data not shown). Processing of proteins through this pathway would result in the complete repertoire of neu peptides for MHC class I presentation, but mature protein was not expressed on the cell surface for antibody recognition.

BALB/c mice were vaccinated twice with pcytneu and pGM-CSF (Fig. 5). Neu Abs were not detected in any of the mice as expected (data not shown). Neu-reactive T cells from 6 immunized mice were measured individually, and an average of 154 ± 60/10⁶ splenocytes of IFN-γ–producing cells were detected (Fig. 5A). Therefore, vaccination with pcytneu activated T cells without inducing an antibody response. To test if T cells alone can protect mice from neu⁺ tumor, immunized mice were challenged s.c. with TUBO, Bam1a, Bam IR5, or D2F2/neu cells, and tumor growth was measured weekly (Fig. 5B). All four tumors were rejected by pcytneu-vaccinated mice, showing control of neu-expressing tumors by CTL, regardless of their drug sensitivity, and that vaccine-induced anti-neu antibody was not required for complete tumor rejection.

A panel of neu-expressing BALB/c mammary tumor cell lines, TUBO, Bam1a, Bam IR-5, and D2F2/neu have been established to represent three types of Her-2⁺ human breast cancers. In a portion of Her-2⁺ breast cancers, Trastuzumab monotherapy inhibits tumor growth (11). TUBO and Bam1a cells, which are highly sensitive to anti-neu antibody and RTKI, would represent these responsive cancer cells. In Bam IR-5 cells, neu L726I mutation induced by gefitinib corresponds to Her-2 L726F mutation in human breast cancer cells that are resistant to another RTKI, lapatinib (38). Therefore, Bam IR-5 represents treatment-induced drug-resistant tumor cells. The majority of Her-2–expressing human tumors have intrinsic resistance to Trastuzumab because only 11% to 23% of metastatic breast cancer patients respond to antibody monotherapy (11, 39). D2F2/neu, which are refractory to Her-2–targeted therapeutics without prior selection, is a representative of these resistant cells. When developing Her-2–based immunotherapy or vaccines, all three types of Her-2/neu⁺ tumors should be tested to fully assess treatment efficacy. The propensity to develop cerebral metastases after Trastuzumab treatment suggests therapy-induced genetic alterations or the selection of aggressive tumor cells (15, 16), warranting vaccines that can initiate or amplify Her-2 immunity to control drug-resistant tumor cells.

Although TUBO and Bam1a are both derived from spontaneous NeuT mammary tumors and they express comparable levels of neu, more chromosomal aberrations are found in TUBO cells, including the translocation of chromosome 4, which has been reported in other neu-induced tumors (40). It is not clear whether the disparities between TUBO and Bam1a cells indicate heterogeneity of neu-induced tumors or changes induced during tissue culture selection. However, increased chromosomal aberrations in TUBO cells correlate with their reduced sensitivity to gefitinib when compared with Bam1a cells, suggesting the use of alternative survival and growth signaling pathways in TUBO cells.

Bam IR-5 cells established by prolonged culture in gefitinib have developed CCAs, NCCAs, trisomies, and monosomies, which were not detected in the parent Bam1a cells. The point mutation of leucine 726 to isoleucine 726 in the ATP binding pocket of the neu tyrosine kinase domain may change the binding of gefitinib to Her-2 and mediate their resistance to drug and antibody treatment, although other genetic changes such as those revealed by SKY analysis may also contribute to the resistance.

Neu-transfected D2F2/neu cells are resistant to drug and antibody treatment both in vitro and in vivo, showing their independence of neu in survival and proliferation. The common origin of D2F2 and D2F2/neu cells is evidenced by the shared chromosome 11 and 16 translocations, which occur in some hormone-induced tumors (41). It is of interest that expression of neu resulted in a less aberrant SKY profile, suggesting the transfection and expansion of a cell with less aberration.

Whether gefitinib resistance is a result of drug selection as in Bam IR-5 or intrinsic resistance as in D2F2/neu, these cells are also resistant to antibody treatment, indicating shared escape mechanisms. But, all test cells are rejected by pcytneu vaccination, which induces CTL without antibody response. We reported previously using pcytE2-vaccinated mice that depletion of CD8, but not CD4 T cells, significantly reduced antitumor immunity, showing the induction of CTL, which mediated tumor rejection (37). Therefore, a strong CTL response to the complete repertoire of neu peptides abolishes tumor growth regardless of their sensitivity to neu-targeted therapy. When a strong humoral response is induced without effector T cells, as in T-cell–depleted, pneuTM-immunized mice, only drug-sensitive cells were completely rejected, with partial protection against drug-resistant cells. In patients whose tumors are refractory to drug and antibody therapy, induction of comprehensive immunity by active vaccination will be critical to their long term protection.

Treatment with gefitnib increased neu expression. Because neu activation is required to induce receptor endocytosis and degradation, inactivation of the kinase domain by gefitinib binding may trigger a compensatory increase in Her-2/neu surface expression. Binding of Her-2⁺ breast cancer cells with Trastuzumab has been associated with decreased Her-2 surface expression (42) with increased internalization and degradation of Her-2 (43). Priming of endogenous Her-2 immunity and increased tumor cell lysis by CTL has also been associated with Trastuzumab treatment of breast cancer (44, 45). In Bam1a and TUBO cells, treatment with anti-neu mAb reduced neu expression, Akt phosphorylation, and cell proliferation. Anti-neu immunity may also be enhanced by anti-neu antibody binding to tumor cells.

It is of note that even D2F2/neu cells that are resistant to antibody in vitro are partially inhibited by vaccine-induced antibodies in vivo, suggesting effector mechanisms such as ADCC in tumor growth control. D2F2/neu cells that survived immune attack in vivo completely lost neu expression, indicating the selection of anitgen-negative tumor cells in vaccinated hosts, whereas Bam IR-5 cells maintained neu expression in immunized hosts, showing their continued dependence on neu for tumor outgrowth.

Regarding the specificity and safety of Her-2/neu vaccination, mice immunized with Her-2/neu DNA could not reject D2F2 cells, which did not express Her-2/neu (data not shown), showing specific recognition of Her-2/neu by DNA vaccination (Fig. 4). Little to no binding of human EGFR (Her-1) transiently expressed on 3T3 cells was detected with Her-2–immune sera (data not shown). Immunized mice do not show any apparent abnormality for at least 1 year except for their anti–Her-2/neu immunity (data not shown), further showing the specificity of Her-2/neu DNA vaccination. The complete blood count of untreated and DNA-vaccinated Her-2 Tg mice did not show obvious differences (data not shown). Therefore, the induction of Her-2/neu specific immunity without noticeable toxicity supports the efficacy of Her-2/neu DNA vaccines.

Taken together, we have shown the benefit of DNA vaccination in the management of diverse tumors that exhibit differential sensitivity or resistance to Her-2–targeted therapies. Anti-neu antibody can suppress tumor growth either directly or through mechanisms such as ADCC and may amplify T-cell immunity. A strong Her-2/neu–specific T-cell response continues to be the goal of Her-2 vaccination as drug-sensitive and drug-resistant tumors are both controlled by T cells.

No potential conflicts of interest were disclosed.

Grant support: NIH CA 76340, CA125680 (W.Z. Wei), and GM 58905-7 (predoctoral training of P.J. Whittington).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank David Shim and Andi Cani for their technical assistance.