A chaperone protein-enriched tumor cell lysate vaccine generates protective humoral immunity in a mouse breast cancer model

Tumors express specific patterns of mutated or overexpressed antigens that may serve as potential targets for activated effector immune cells (1, 2). Therefore, vaccines that deliver multiple tumor-derived antigens are being increasingly investigated, as they may stimulate polyvalent T-cell and B-cell responses against the tumor cells from which the vaccine has been generated. We have developed one such autologous tumor-derived vaccine termed chaperone-rich cell lysate (CRCL) and have shown that CRCL vaccine generates anticancer protection through T-cell-mediated immunity in several tumor models (3, 4). We have now extended our work by investigating whether tumor-derived CRCL can also promote tumor-specific responses through adaptive humoral immunity. Because CRCL is a complex of several chaperone proteins (3), it has the possibility to deliver both MHC class I and II epitopes and generate tumor-specific helper T cells, CTLs, and antibodies that may target tumor cells.

The CRCL vaccine was derived from HER2/neu-positive tumor (TUBO) and was evaluated as a therapeutic vaccine in an established tumor setting. HER2/neu overexpression occurs in 20% to 30% of breast cancer cases and is a marker for poor prognosis (5). HER2/neu is a self-protein, but preexisting host immunity to HER2/neu antigen has been shown in breast cancer patients (6). Active immunization strategies that generate optimal antibody and T-cell-specific responses are being used in treating HER2/neu cancers (7). The induction of a humoral response is particularly important due to the overexpression of HER2/neu protein on the surface of the cancer cells. Thus it is not surprising that Herceptin, a humanized monoclonal antibody, has been approved for clinical use (8).

We report here that the administration of CRCL derived from TUBO tumors delayed growth of preestablished tumors. Analysis of the B-cell-specific response showed the presence of anti-HER2/neu antibody in the serum of TUBO CRCL vaccinated animals. We have investigated the mechanisms of antibody-mediated tumor destruction in response to TUBO CRCL vaccine and have shown that multiple immune effectors such as CD4⁺ T lymphocytes and B cells are required. We have thus documented that CRCL, in addition to its potential to stimulate efficient antitumor T-cell responses, is also capable of generating potent humoral responses that can lead to delay in growth of preestablished tumors.

Female wild-type (WT) or severe combined immunodeficient BALB/c (H2^d) mice (6-8 weeks old) were obtained from the National Cancer Institute. B-cell-deficient BALB/c (H2^d) mice (9) were purchased from Taconic Farms. Animal experiments were conducted according to the guidelines of the University of Arizona Institutional Animal Care and Use Committee.

TUBO is a cloned cell line derived from BALB/c mice transgenic for the transforming rat HER2/neu oncogene (BALB-NeuT; ref. 10). Dr. Guido Forni (University of Turin) and Dr. Carla de Giovanni (University of Bologna) generously provided the TUBO cells. Rat p185 neu is a xenogeneic protein in normal mice (11) and does not induce anti-rat p185 neu antibodies or any detectable CTL when implanted in WT BALB/c mice (10). TUBO cells were cultured at 37°C, 5% CO₂ in DMEM (Life Technologies) supplemented with 25 mmol/L HEPES, pyridoxine-HCl (Life Technologies) with 20% heat-inactivated fetal bovine serum (Gemini). 12B1 is a murine leukemia cell line (kindly provided by Dr. W. Chen, Cleveland Clinic) in BALB/c background (4) and was used as an irrelevant antigen control. The cells were cultured in RPMI 1640 (Life Technologies) supplemented with 10% fetal bovine serum.

CRCL was prepared from tumors that were implanted s.c. in BALB/c mice as described previously (3, 4). In brief, tumor tissues were homogenized in detergent-containing buffers, and the high-speed supernatants underwent forward solution isoelectric focusing in a Rotofor device (Bio-Rad) at 15 W constant power. Fractions were harvested and analyzed for chaperone protein content. Fractions of interests were pooled and prepared as vaccines by dialysis, detergent removal, and centrifugal concentration. The level of endotoxin in CRCL was lower than that in media control (<0.01 EU/μg) as determined by Limulus amebocyte lysate assay kit (Cambrex Bio Science).

WT or B-cell-deficient BALB/c mice were injected with 1 × 10⁵ viable TUBO cells s.c. in the right groin. The mice were vaccinated six times on the opposite groin with 20 μg tumor-derived CRCL on days 0, 2, 4, 7, 10, and 14 after palpable tumor development (tumor diameter, 2-3 mm). Mice injected with saline and 12B1 CRCL (BCR-ABL positive) served as control groups. Tumor volume was calculated every week and mice that reached 4,000 mm³ in tumor volume were euthanized. The data were analyzed with GraphPad Prism software (GraphPad Software) and the log-rank statistics were used to test differences between groups (12).

HER2/neu-specific ELISA was done with sera collected from tumor-bearing vaccinated mice. Mice were bled from the lateral tail vein before tumor implantation (day 0), during vaccination (after third vaccination, day 16), and 1 week (day 30) and 3 weeks (day 44) after the last immunization and stored at -80°C. They were analyzed for antibodies to extracellular domain of human HER2/neu (ECD^HER2) by ELISA using 96-well microtiter plates coated with 1 μg/mL ECD^HER2 protein (13–15). Soluble human ECD^HER2 was purified from BHK/erbB2 cell culture supernatants (generously provided by Dr. James D. Marks, University of California-San Francisco) using affinity chromatography with anti-HER2/neu IgG3 immobilized on CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech). The ECD^HER2-coated plates were blocked with bovine serum albumin (BSA; Sigma) and incubated overnight at 4°C with serum dilutions. Alkaline phosphatase–labeled rat anti-mouse IgG1 and IgG2a (Zymed) or alkaline phosphatase–labeled goat anti-mouse IgG3 (Southern Biotechnology Associates) were used to detect bound IgG and absorbance values were read at 410 nm after color development for 2 h with p-nitrophenyl phosphate disodium (Sigma).

Sera from mice immunized with PBS or TUBO CRCL were collected a week after the administration of six doses. A total of 2 × 10⁵ TUBO cells were stained by an indirect immunofluorescence method. Tumor cells were incubated with naive or immune sera at 1:80 serum dilution followed by fluorescein (Alexa Fluor 488)–conjugated goat anti-mouse IgG (Molecular Probes). The cells were evaluated using a FACScan (Becton Dickinson) and the data were analyzed with Cell Quest Software (Becton Dickinson).

Severe combined immunodeficient BALB/c mice were randomized and distributed into six mice per group. Sera from mice vaccinated with PBS, 12B1 CRCL, or TUBO CRCL were pooled 1 week after the last dose of vaccine and stored at -80°C. Mice received i.v. injection of 200 μL pooled sera on day -1 and were then challenged with 10⁵ tumor cells a day later (13). A group of mice that did not receive any serum were used as controls for the transfer groups.

Down-regulation of tumor antigen molecules, including HER2/neu molecules, from the cell surface of TUBO cells were assessed by confocal microscopy after incubation with sera collected a week after administering six doses of vaccine. TUBO cells were incubated with sera of various dilutions and permeabilized to detect cytoplasmic expression of tumor antigens. Alexa Fluor 488–conjugated IgG (Molecular Probes) was used to assess the expression of membrane-bound and cytoplasmic tumor antigens. Internalization of fluorescence was measured on a confocal microscope (LFM310; Zeiss; 488 nm argon laser). Green fluorescence was detected after excitation at 488 nm and images were recorded as TIF files and processed (LSM Image Examiner; Zeiss) to subtract background and enhance lower and middle intensity fluorescence (10).

Antibodies contained in the sera of TUBO CRCL vaccinated mice were evaluated for tumor cell lyses by antibody-dependent cell-mediated cytotoxicity (ADCC) assay. Tumor cells were plated at 1 × 10⁴ per well in 96-multiwell plates and allowed to attach overnight at 37°C. Cultures were incubated for 2 h on ice with sera from experimental and control mice at different dilutions. After washing, spleen cells from naive BALB/c mice were added at 50:1 effector-to-target ratio and plates were incubated at 37°C overnight. Nonadherent cells were carefully washed out and 20 μL WST-1 solution (Roche Diagnostics) was added to each well. The absorbance at 450 nm was read after 4-h incubation.

Natural killer cells were depleted with anti-asialo GM1 (25 μL/mouse, 1/8 diluted with PBS; Wako Chemical). Rat anti-mouse CD4⁺ (GK1.5) and rat anti-mouse CD8⁺ (2.43.7) hybridomas (American Type Culture Collection) were sent to Taconic Farms for the production of mouse ascites. Mice were depleted of CD4⁺ or CD8⁺ cells by i.p. injection with 50 μL GK1.5 or 2.43.7 in PBS. Immune cells were depleted on -3 and -1 days before s.c. injection with 10⁶ TUBO cells. This was followed by two more depletions 7 and 21 days after tumor implantation to maintain the depletion of specific cell types. Mice were vaccinated six times after palpable tumor development as described previously.

All statistical analyses were done using the Student's t test to evaluate significance between groups, except that the log-rank statistics was used to test differences between groups. For all cases, results were regarded significant if P values were <0.05.

CRCL was generated from TUBO tumors and Western blot analysis established the presence of heat shock proteins 110, 90, 70, and 60 and glucose-regulated protein 94, confirming our previous reports (3, 4, 12, 16, 17). To evaluate the efficacy of CRCL vaccine in this mouse breast cancer model, groups of 12 BALB/c mice were injected with 10⁵ TUBO cells s.c. and were monitored for tumor growth. Mice were randomized and only animals with palpable tumors were vaccinated. The results indicate that vaccination with TUBO CRCL vaccination significantly delayed tumor growth when compared with nonimmunized mice or mice immunized with irrelevant 12B1 CRCL (Fig. 1). By day 44, nonimmunized and 12B1 CRCL-immunized mice exhibit an average tumor volume of 4,351 ± 248 and 4,122 ± 338 mm³ respectively, whereas the tumor volume of mice immunized with TUBO CRCL was 2,635 ± 305 mm³ (P < 0.01). This delay in tumor volumes confirmed the antitumor efficacy of TUBO CRCL against preestablished murine tumors.

It has been shown that antibody plays a critical role in HER2/neu-positive tumor immunity (18–22). Because TUBO cells overexpress rat p185 neu, we predicted that CRCL derived from TUBO tumors might chaperone HER2/neu antigen. We therefore measured anti-HER2/neu antibody titers in tumor-bearing mice vaccinated with TUBO CRCL. Sera from PBS and 12B1 CRCL-immunized animals were used as controls for the specificity of the antigen. A purified extracellular domain from human HER2/neu was used to test HER2/neu-specific isotypes (14, 15). It has been shown previously that antibodies against rat p185 neu can cross-react to the ECD^HER2 of human origin (13, 14), because rat neu protein is 89% homologous to human HER2/neu protein (23). Increase in total IgG1 response to ECD^HER2 was evident in the HER2/neu-positive CRCL-immunized group (Fig. 2) when compared with sera from preimmune mice (day 0; P < 0.01). These absorbance values were six times higher a week after the last dose (day 30) and this difference was sustained for at least an additional 2 weeks (day 44). In comparison, analysis of saline and 12B1 CRCL-immunized sera did not show the same specificity (P > 0.05). Analysis of IgG2a and IgG3 levels showed substantial increase compared with the controls (P < 0.05) but lower in contrast to IgG1 production.

The importance of B cells in delaying the onset of mammary carcinoma cells has been documented in HER2/neu mouse models (15, 24). To examine the role of B cells in tumor growth delay, we used J_H knockout mice lacking B cells and antibodies (Ig^-/-) and compared them with tumor growth in WT BALB/c mice. The mice were injected s.c. on day 0 with TUBO cells and mice with measurable tumors were vaccinated with tumor-derived CRCL. In Ig^-/- tumor-bearing mice, immunization with TUBO CRCL provided no significant protection compared with saline controls. By day 32, they had an average tumor volume of 3,032 ± 373 and 2,439 ± 212 mm³, respectively (P > 0.05; Fig. 3A and B). Tumor growth kinetics in Ig^-/- mice immunized with TUBO CRCL was significantly faster compared with TUBO CRCL-immunized WT BALB/c mice (P < 0.05), indicating that B-cell response generated by TUBO CRCL vaccination played a role against the growth of TUBO tumors.

Next, we addressed whether antitumor activity can be transferred through immune sera. CB.17 severe combined immunodeficient mice were injected i.v. with 100 μL pooled sera from TUBO CRCL-immunized mice. Following transfer of sera, mice were challenged the following day with TUBO cells. As shown in Fig. 3C, no apparent antitumor activity was observed with sera from control mice receiving PBS or irrelevant 12B1 CRCL (P > 0.05). Conversely, mice receiving sera from TUBO CRCL-immunized animals showed a significant tumor growth delay (P < 0.05) from day 24 onwards (380 ± 78 versus 702 ± 112 and 710 ± 92 mm³). This provides additional evidence that TUBO CRCL-associated antibodies have an effect on delaying tumor growth.

Various mechanisms have been proposed to account for the antitumor activity of antibodies. We have shown substantial IgG1, IgG2a, and IgG3 production against ECD^HER2 from mice immunized with TUBO CRCL. These isotypes are known to mediate ADCC analogous to the antitumor activity of trastuzumab (Herceptin) in vivo (25). Therefore, to understand whether TUBO CRCL-induced antibodies can cause tumor cell lyses, we did ADCC assays using mouse splenocytes as effectors. Figure 4 shows that at 1:5 dilution of the sera ∼30% cytotoxicity was detected from TUBO CRCL-vaccinated mice when compared with the 5% killing with control sera (P < 0.05). This effect was specific as we were able to abrogate lysis of tumor cells with increasing dilutions of the sera obtained by TUBO CRCL immunization.

It has been reported that antibody binding to the HER2/neu antigen can cross-link HER2/neu molecules on the cell surface and trigger endocytosis (18). Reduced HER2/neu at the cell surface allows less HER2/neu heterodimer formation, potentially resulting in reduction in growth factor–induced signaling and proliferation. Based on the results depicted in Fig. 2, we reasoned that the antibodies contained in the sera of TUBO CRCL-immunized mice may bind to and dampen the expression of tumor antigens at the tumor cell surface. Flow cytometry analysis indicated a positive staining only in the cells incubated with anti-TUBO CRCL serum (Fig. 5A), where the average mean fluorescence intensity of experimental sera was 121.3 ± 21.9 when compared with 31.7 ± 5.2 of control sera. This indicates that the sera of TUBO CRCL-immunized mice contain antibodies (IgG) that can specifically bind to TUBO cells, some of which may presumably be anti-HER2/neu antibodies.

Next, we analyzed by confocal microscopy whether binding of these antibodies can induce the internalization of the IgG/anti-IgG-fluorochrome complexes from the tumor cell surface. TUBO cells were incubated with 1:10 dilution of immune sera for 3 h at 4°C and 37°C, respectively. Figure 5B shows that at 4°C sera from mice immunized with TUBO CRCL bound to the TUBO cell membrane as shown by the green fluorescence stain on the surface of the tumor cells. When the cells were exposed to 37°C, the bound complexes were then internalized and confined to the cytoplasm of the tumor cells as shown by the presence of large cytoplasmic dots. No such effect was observed with TUBO cells incubated with sera from PBS or 12B1 CRCL-vaccinated mice, showing that this response was TUBO CRCL specific. Although it is possible that anti-HER2/neu antibodies present in the sera may contribute significantly to this process, we cannot exclude that TUBO cells may also express antigens analogous to HER2/neu that are also susceptible to internalization by the binding of antibody to the surface of the cells.

We have shown previously in 12B1 BCR-ABL⁺ leukemia model that the protective effect of CRCL vaccination was completely abrogated in the absence of CD8⁺ and CD4⁺ T cells (12). Therefore, to assess the contribution of specific T cells in the TUBO model, we depleted either CD8⁺ or CD4⁺ T cells before and after CRCL vaccine administration. CD8⁺ T did not play a role in antitumor effect, as both the TUBO CRCL-vaccinated groups (one with intact T cells and the other with depleted CD8⁺ T cells) revealed no significant difference in tumor growth delay (Fig. 6A). By day 49, the average tumor volumes of the undepleted group (2,139 ± 178 mm³) and CD8-depleted group (2,061 ± 330 mm³) were similar (P > 0.05). In contrast, protective immunity could not be elicited without CD4⁺ T cells as all mice in depleted group (3,329 ± 429 mm³) grew tumors significantly faster (P < 0.05) than the immunized mice with intact CD4⁺ (2,139 ± 178 mm³; Fig. 6B).

Polyvalent autologous cancer vaccines circumvent issues associated with peptide-specific immunization by including a multitude of tumor-specific antigens in one vaccine preparation. Such vaccines may trigger the activation of effector polyclonal immune responses, which may overcome the emergence of escape tumor cell variants. We have developed one such vaccine (CRCL) and have now reported that, in addition to tumor-specific T-cell responses (3, 4, 16), CRCL can also provide protective immunity by the induction of humoral immunity. Several reports have indicated that single chaperone protein vaccines can induce antigen-specific antibodies (19, 26); however, it is not known whether humoral responses directly account for the antitumor effect.

Herein, we document that CRCL vaccine isolated from mammary carcinoma delays growth of preestablished tumors and that tumor-specific immunity involves antibody production, B cells, and CD4⁺ T lymphocytes. Additionally, transfer of serum to severe combined immunodeficient mice implied that protection against TUBO is mediated by humoral immunity. These findings are consistent with other reports highlighting the importance of B cells in tumor growth control following immunization with antibody-cytokine fusion proteins (13, 15) or DNA plasmids encoding both extracellular domain and intracellular domain of HER2/neu (10, 20).

In the TUBO model, the role of the humoral response seems to be particularly prominent because of the dual role of rat p185 neu, which is both a target tumor antigen and a membrane-exposed receptor regulating cell growth (21). The extracellular domain of HER2/neu contains several B-cell epitopes and it has been shown that active immunization with a combination of these epitopes is effective in preventing mammary tumors (22, 27). Because CRCL was isolated from TUBO cells (expressing rat p185 neu), we evaluated the presence of anti-HER2/neu antibodies in the serum of TUBO CRCL-immunized animals. Anti-HER2/neu IgG1, IgG2a, and IgG3 isotypes analyzed in tumor-bearing vaccinated mice increased 1 week after the first dose of vaccine and persisted for at least 3 weeks after the last dose, suggesting that a long-term immunity can be elicited. Subclass of the antibody response has distinct biological function and can affect antigen specificity, as FcγR-mediated effector responses such as ADCC is primarily mediated by IgG1 and IgG3 isotypes (28). The latter have also been reported to be associated with T helper 1 immune response (29), whereas IgG1 production is usually linked to T helper 2 response (13). Thus, generation of wider isotype specificity in response to CRCL immunization suggests that polyclonal expansion of antigen-reactive B cells may be possible in this model.

Several B-cell epitopes reside in the extracellular domain of the HER2/neu antigen (27). Although further confirmation is required, HER2/neu antigen chaperoned within the CRCL preparation may deliver several of these immunogenic B-cell epitopes. Similar to peptide vaccines, endogenous and persisting antibody titers generated by CRCL would be more valuable compared with repeated administration of trastuzumab (30). Indeed, passive administration of antibodies may be cleared from the circulation, which may lessen their therapeutic potential (31). Furthermore, compared with vaccines that target single or limited antigens, antibodies generated by TUBO CRCL immunization may target other unidentified antigens that are naturally expressed in vivo on the surface of TUBO cells.

Depletion of natural killer cells before tumor implantation and during vaccine administration had no consequence in tumor delay (data not shown), suggesting that CRCL-induced antibodies may interact with Fc receptors on other effectors such as macrophages and neutrophils (32). A recent report highlights interaction of trastuzumab with specific Fc receptor subtypes and that FcγR is a dominant component of the activity of trastuzumab and similar monoclonal antibodies (25). Therefore, the role of these effectors in eliciting ADCC in the response to CRCL vaccination remains to be investigated.

CD4⁺ T cells are important in antitumor immunity as they can produce cytokines supporting an evolving B-cell immunity. In the TUBO model, tumor delay was achieved in the absence of CD8⁺ T cells, but CD4⁺ T cells were required. Other investigators have also established that CTLs are difficult to detect in this tumor model and that tumor-reactive CD4⁺ T lymphocytes are crucial for the induction of antitumor immunity (10, 33, 34).

Our data also indicate that B cells play a major role in CRCL-induced protection in mice. It is known that, in addition to the production of antibodies, B cells also execute multiple functions in the immune response (35, 36). These cells may indeed act as antigen-presenting cells and present antigens to CD4⁺ T cells and modulate their function by promoting up-regulation of costimulatory signals required for their expansion and differentiation (36).

One significant aspect of this work lies in the observation that tumor-derived CRCL delays tumor growth by the induction of a humoral immunity in an established tumor model. The work presented here is part of an ongoing effort to better understand the immune responses elicited by the tumor-derived multichaperone protein vaccine and its potential for further clinical application. The above data also advance our knowledge on the utility of CRCL vaccine and provide a foundation that could lead to improved therapy for cancers responding to antibodies, such as HER2/neu-expressing tumors.