The A2L2 cell line, which expresses high levels of rat HER2/neu, was derived from the mouse tumor cell line 66.3 in our laboratory 29. The A2L2 cell line has expressed high levels of HER2/neu for more than 5 years and consistently induces tumors in BALB/c mice when injected into a mammary fat pad or intravenously (i.v.). The line was maintained in Eagle's minimal essential medium containing 5% fetal bovine serum, sodium pyruvate, nonessential amino acids, L-glutamine, and vitamins (Gibco, Carlsbad, CA, USA). The monolayer cultures were subdivided at approximately 75% confluence by treatment for 1 to 3 min with 0.25% trypsin and 0.02% ethylenediaminetetraacetic acid (EDTA) at 37°C.

The SINCP plasmid, which contains Sindbis virus genes, was provided by Dr John Polo (Chiron, Emeryville, CA, USA). SINCP is nearly identical to the ELVIS plasmid we previously tested 29 except for an internal ribosome entry site on the 5' side of the insertion point for an antigen gene. The gene for rat neu was excised from pSV2-neu and inserted into SINCP by standard recombinant DNA techniques 29. The correct insertion of the complete neu gene was confirmed by sequence analysis. The E1,E2a-deleted adenovirus (Adeno) has been described in detail 3738. The gene for rat neu was excised from pSV2-neu and incorporated into Adeno as previously described 38.

To evaluate the level of immunoglobulin G (IgG) in the serum of vaccinated mice, we incubated A2L2 cells for 1 hour at 37°C with either immune serum or control serum diluted 1:100 in PBS, pH 7.2. Fluorescein-isothiocyanate-labeled (FITC-labeled) goat antimouse IgG (Pharmingen, San Diego, CA, USA) diluted 1:1000 in PBS was added to the cells, and incubation continued for 1 hour at 37°C. The cells were washed three times in PBS and analyzed by flow cytometry using an EPICS Profile Analyzer (Coulter, Hialeah, FL, USA). As a positive control, A2L2 cells were stained with a 1:1000 dilution of a polyclonal rabbit antibody against p185 (sc-284; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and an FITC-labeled goat anti-rabbit IgG.

Female BALB/c mice 6 to 8 weeks old were obtained from the National Cancer Institute-Frederick Cancer Research Facility (Frederick, MD, USA). Mice were acclimated for at least 1 week before use. Our small-animal facility is approved by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC).

SINCP-neu was prepared in milligram quantities by Aldevron (Fargo, ND, USA). Intramuscular (i.m.) injections of 0.1 ml containing 100 μg of plasmid DNA formulated with 0.25% bupivacaine (Sigma Chemical, St Louis, MO, USA) were administered to the quadriceps using a 24-gauge needle. Adeno-neu and adenovirus lacking an inserted gene (Adeno-empty) were suspended in 100 μl of 0.85% saline and injected into the quadriceps using a 24-gauge needle. Blood was collected at intervals from the tail vein, and the serum was separated by centrifugation after incubation at 37°C for 1 hour and overnight refrigeration.

We used a mammary fat pad tumor prevention model to assess the effect of vaccination on solid tumor development. A2L2 cells from cultures that had reached 75% confluence were harvested by treatment with 0.25% trypsin and 0.02% EDTA for 1 to 3 min at 37°C. The cells were washed once in serum-containing culture medium and once in PBS. Mice were anesthetized by inhalation of isoflurane using an apparatus developed by veterinarians at the University of Texas MD Anderson Cancer Center 39. The fur was shaved over the lateral thorax, and a 5-mm-long incision was made to reveal mammary fat pad number 2 40. A 0.1-ml sample containing 2.5 × 10⁴ A2L2 cells in normal saline was injected into the fat pad. The incision was closed with a wound clip. After 7 days, wound clips that had not already fallen off were removed. The mice were then observed daily, and their tumors were measured in perpendicular directions using a caliper. Mice with tumors at least 10 mm in the greater dimension were killed according to our approved Institutional Animal Care and Use protocol. At the end of the experiment, all mice were killed by CO₂ inhalation, and all tumors were excised and weighed. Examination of the surface of the lungs during necropsy did not reveal tumor nodules.

Although a solid tumor develops after injection of A2L2 cells into the mammary fat pad, it is not highly metastatic. The tumor induces moribundity in mice before an appreciable number of lung metastases develop. For this reason, we used an experimental lung metastasis model rather than a spontaneous lung metastasis model to assess the effect of vaccination on lung metastasis. A 0.1-ml sample containing 2.5 × 10⁴ A2L2 cells in normal saline was injected into the tail vein of each immunized mouse. Thirty days later, the mice were killed by CO₂ inhalation, and surface lung metastases in each animal were counted.

To determine the number of interferon-γ-producing (IFNγ-producing) cells in the spleens of vaccinated mice, we used an IFNγ enzyme-linked immunospot (ELISPOT) technique (kit no. 552569; Pharmingen) and used reagents from Pharmingen whenever possible. Immune spleens were dissected from vaccinated mice and prepared exactly as previously described for tetramer analysis 41. Wells containing only immune spleen cells served as negative controls, and spleen cells from vaccinated mice cultured with 5 μg/ml concanavalin A (Con A) (Sigma-Aldrich, St Louis, MO, USA) overnight served as a positive control. The finished plates were air-dried overnight at room temperature in the dark and sent to ZellNet Consulting (New York, NY, USA 42), where the spots were counted automatically using an ImmunoSpot Series I analyzer (BD Bioscience, San Diego, CA, USA). If confluence (overlapping spots) was present in wells, the number of spots in a non-confluent area of that well was determined, and the following equation was used to estimate the total number of spots in each well with confluence: total spot number = spot count+2(spot count × % confluence/ [100% – %confluence]).

Student's t test was performed using Prism 4.0 GraphPad software (San Diego, CA, USA).

To determine the effectiveness of vaccination on solid tumor growth, two groups of 13 mice each were vaccinated once i.m. in the quadriceps with 100 μg of SINCP-neu or SINCP-βgal. Two weeks later, the mice were challenged with A2L2 cells injected into the mammary fat pad. Thirty-five days after the challenge, the mice were killed and if a solid tumor was present, its mass was determined. All of the mice vaccinated with SINCP-βgal developed tumors. In the group vaccinated with SINCP-neu, only six mice developed tumors, and the mean mass of these six tumors was significantly less than the mean mass of the tumors in the SINCP-βgal group (0.63 vs 1.02 g; P = 0.0186). When we calculated the mean mass for all 13 mice in the SINCP-neu group and compared it with the value of the SINCP-βgal group, the difference was significant (0.25 vs 1.02 g; P = 0.0001) (Fig. 1a).

Two groups of 10 mice were each vaccinated i.m. in the quadriceps once with 1 × 10¹⁰ particles of Adeno-neu or Adeno-empty. Six weeks later, the mice were challenged with A2L2 cells injected into the mammary fat pad. Thirty-five days after the challenge, the mice were killed and if a solid tumor was present, its mass was determined. All of the mice vaccinated with Adeno-empty developed tumors, whereas in the group vaccinated with Adeno-neu only 1 of 10 mice developed a tumor (mean mass 1.75 vs 0.2 g; P = 0.0001) (Fig. 1b).

Our findings show that reduced solid tumor growth is attributable to the induction of cellular immunity. This immunity is antigen specific, because it was absent in mice vaccinated with SINCP-βgal or Adeno-empty, both of which lack the neu gene.

To determine the effectiveness of vaccination on experimental metastasis of A2L2 cells, two groups of five mice each were vaccinated i.m. with 100 μg of SINCP-neu or SINCP-βgal three times at 2-week intervals. Two weeks after the third vaccination, the mice were challenged with A2L2 cells injected into the tail vein. Twenty-one days after the challenge, the mice were killed, the lungs removed, and the number of tumor nodules on the surface of the lungs counted with the naked eye. All of the mice in the SINCP-βgal group had more than 25 surface lung nodules, whereas in the SINCP-neu group all of the mice had between 1 and 10 nodules (mean number of nodules 105 vs 15; P = 0.0079) (Fig. 2a).

Two groups of nine mice each were vaccinated once with Adeno-neu or Adeno-empty and 6 weeks later challenged with A2L2 cells injected into the tail vein. Thirty-five days after the challenge, the mice were killed, the lungs removed, and the number of surface lung nodules counted with the naked eye. All but one mouse in the Adeno-empty group had lung nodules, whereas in the Adeno-neu group five of nine mice had lung nodules (mean number of nodules 15 vs 1; P = 0.0078) (Fig. 2b).

Our results show that inhibition of experimental metastasis is attributable to the induction of cellular immunity. This immunity is antigen specific because it was absent in mice vaccinated with SINCP-βgal or Adeno-empty, both of which lack the neu gene.

To determine if vaccination induced a humoral immune response to p185, immediately before tumor challenge 0.1 ml of blood was collected from the tail vein of each of the mice treated as described for the experimental metastasis model. The sera for each group were pooled and analyzed for the presence of p185-specific IgG by flow cytometry using the A2L2 cell line as described previously 29 and in Materials and methods. Serum from mice vaccinated with SINCP-neu had a stronger IgG response to A2L2 cells than that from mice vaccinated with SINCP-βgal (Fig. 3a). Likewise, serum from mice vaccinated with Adeno-neu exhibited a stronger IgG response to A2L2 cells than that from mice vaccinated with Adeno-empty (Fig. 3b). These results show that the gene vaccines induce humoral immunity as well as the cellular immunity described above.

Because SINCP-neu and Adeno-neu were effective as preventive vaccines delivered before tumor cell challenge, we evaluated both vaccines as therapeutic vaccines delivered after tumor cell challenge. Vaccination conditions were the same as those described for the experimental metastasis model. In both the therapeutic mammary fat pad tumor model and the therapeutic experimental metastasis model, neither vaccine prolonged the overall survival rate of mice when delivered after tumor cell challenge (data not shown). Even when administered as early as 2 days after tumor challenge, neither vaccine was effective.

To determine whether the therapeutic effectiveness of the vaccines could be increased by sequential administration, we evaluated a prime–boost protocol in which the mice were primed with a single injection of SINCP 2 days after tumor cell challenge with A2L2 cells and boosted with injections of Adeno 9 and 16 days after the challenge. SINCP was always the prime and Adeno was always the boost, because numerous studies have reported that the most effective protocol is a plasmid followed by a virus 4344454647. All combinations of SINCP-neu or SINCP-βgal used as the prime and Adeno-neu and Adeno-empty used as the boost were tested in both the therapeutic mammary fat pad tumor model and the therapeutic experimental metastasis model.

In the therapeutic mammary fat pad tumor model, none of the combinations significantly increased the overall survival rate (data not shown). In the therapeutic experimental metastasis model, only one prime–boost combination was effective: the overall survival rate was significantly higher when SINCP-neu was the prime and Adeno-neu was the boost (P = 0.0004) (Fig. 4a). A comparison of mice vaccinated with SINCP-βgal as the prime showed that the overall survival rate was slightly higher when Adeno-neu was the boost than when Adeno-empty was the boost, but that this increase was not as high when neu was in both the prime and the boost (Fig. 4b). Thus, priming with SINCP-neu is essential for the effect, depends on the presence of the neu gene, and cannot be replaced by non-specific stimulation with SINCP-βgal.

We evaluated the effectiveness of prime–boost vaccination in an in vitro assay. Naive mice were primed by vaccination with SINCP-neu and boosted by vaccination twice with Adeno-neu, or they were primed by vaccination with SINCP-βgal and boosted by vaccination twice with Adeno-empty 48. The presence of the neu gene in both the prime and the boost resulted in a strong IgG response to the A2L2 cells (Fig. 5a). The IgG response from vaccination with SINCP-βgal followed by Adeno-empty (Fig. 5a) was nearly identical to the IgG level in serum collected from the mice before vaccination (Fig. 5b). These observations indicated that prime–boost vaccination produces strong humoral immunity.

To determine whether prime–boost vaccination produced neu-specific T lymphocytes, we used an IFNγ ELISPOT assay to compare spleen cells of mice vaccinated with SINCP-neu as the prime and Adeno-neu as the boost with spleen cells of mice vaccinated with SINCP-βgal as the prime and Adeno-empty as the boost. Spleen cells from vaccinated mice were co-cultured overnight with A2L2 cells, and the next day the number of cells producing IFNγ was determined. Vaccination with SINCP-neu followed by Adeno-neu resulted in a mean of more than 150 IFNγ-producing cells per million spleen cells at all ratios of effector (spleen cell) to stimulator (A2L2 cells) cells, whereas vaccination with SINCP-βgal followed by Adeno-empty resulted in a mean of fewer than 20 IFNγ-producing cells per million spleen cells (Fig. 6). As a positive control and possibly as a measure, of a maximum response, spleen cells were stimulated with 5 μg/ml Con A, which resulted in a mean that was comparable to the means obtained with neu in both the prime and the boost. That the presence of neu in both the prime and the boost resulted in IFNγ production after co-culture with A2L2 cells and that this effect was not seen with vaccination with SINCP-βgal followed by Adeno-empty clearly demonstrated that prime–boost vaccination resulted in antigen-specific induction of cellular immunity.