Background
Modified tumor cells can induce tumor-specific immunity and, in certain models, activate both adaptive and innate immune responses [
1]. However, in some mouse models and the vast majority of human cancers, the tumor vaccines currently employed have not been successful. This may be attributed to a failure of adequate stimulation of appropriate components of immunity and/or tolerance to tumor antigens [
2]. In general, tumor vaccination strategies have focused on enhancing a cytotoxic T cell (CTL) response. Activation of both T-helper cells and CTLs is achieved primarily through cross-presentation of tumor antigens by professional antigen presenting cells (APCs) [
2]. Antigens from apoptotic cells have been reported to be a preferred vehicle for activating tumor immunity, rather than tolerance, through cross-presentation by APCs [
1,
3]. Direct antigen presentation by tumor cells could potentially activate T cells provided the tumor cells can deliver an MHC-restricted antigen-specific signal together with appropriate costimulatory signals [
4‐
6]. However, the role of tumor cells as APCs has not been well defined. Nevertheless, MHC class I mediated direct priming of CTLs has been observed in an engineered tumor model which is dependent on the density of MHC/peptide complexes and the expression of B7 costimulatory molecules on tumor cells [
7]. Moreover, transfection of MHC class II negative tumors with MHC class II and B7-1 genes produces a cellular vaccine capable of eliciting immunity [
8]. MHC class II positive tumor cells are also effective APCs in vivo and can present novel endogenous antigenic peptides not presented by host APCs [
5]. Furthermore, transfection of tumors with class II transactivator (CIITA) elicits MHC class II expression and can restore the ability of certain tumor cells to present antigen and induce immunity [
9,
10]. Although cross-presentation is the major mechanism generating immunity [
2,
3], the above studies on tumors as APC suggest that, at least in certain tumors, direct antigen presentation could provide an alternative or additional pathway in tumor immunity. An important issue is whether direct presentation can be enhanced in vivo and become a quantitatively significant component of tumor immunity.
Tumor escape has been attributed to selection of tumor cells with mutations in genes involved in both the initiation and effector phases of immunity [
11]. Recent evidence also suggests that tumor cells may exploit epigenetic silencing of immune genes to escape immune destruction [
12]. Epigenetic repression of immune genes in tumors was first suggested for MHC class II genes and CD40, which are infrequently mutated although often deficient in tumors [
13]. A number of other immune genes, including MHC class I, components of the class I peptide presentation pathway (TAP1, TAP2, LMP2, LMP7 and Tapasin), B7-1/2, NKG2D ligands and certain tumor antigens, are also silenced by chromatin in multiple tumor types [
12,
14].
Covalent modifications of chromatin are well established regulators of gene expression and an array of epigenetic alterations, including acetylation and methylation, have been shown to target histones [
15]. Histone acetylation induced by histone deacetylase inhibitors (HDACi) frequently increases accessibility of transcription factors to promoter sites and results in enhanced gene transcription, although some genes are inhibited by HDACi. The HDACi trichostatin A (TSA) regulates the expression of ~5% of the genome [
16]. HDACi treatments have been used in clinical trials on the basis of their ability to induce differentiation and apoptosis of tumor cells [
17]. HDACi-treated tumor cells have also been successfully employed in a murine tumor vaccine model [
4]. Reversal of gene silencing by HDACi treatment can convert a plasma cell tumor to an APC, in vitro, through upregulation of MHC class II and costimulatory molecules [
4,
18]. Moreover, HDACi treated melanoma tumor cells can mediate direct antigen presentation via MHC class I, which stimulates IFN-γ-producing T cells in vitro [
14]. The present study was constructed to determine the mechanisms involved in immunity generation by TSA-treated melanoma vaccines and whether this epigenetically modified tumor cell vaccine is effective in a treatment model.
Materials and methods
Cells, mice and reagents
Mouse B16 melanoma, EL4 thymoma cell lines and hybridoma GK1.5, 2.43 and 53-6.7 cells (ATCC, Manassas, VA) were maintained in culture as specified. Six to eight week old female C57BL/6 (B6) (NCI, Bethesda, MD), MHC class I deficient B6.129-β2mtm1 [class I-/-], class II deficient B6.129-H2dlAb1-Ea (class II-/-) (Jackson Laboratory, Bar Harbor, ME), class I and class II double deficient B6.129-Aββtm1-β2mtm1 (class I-/--II-/-) (Taconic, Germantown, NY) and ovalbumin-specific I-Ab-restricted TCR transgenic (OT-II) (Protul Shrikant, RPCI, Buffalo, NY) mice were maintained in the Department of Laboratory Animal Resources at RPCI. Principles of laboratory animal care (NIH publication 85-23, revised 1986) were followed and all work was carried out under RPCI IACUC approval. TSA (Wako Biochemical, Richmond, VA), H2O2 (Sigma, St. Louis, MO) and mouse IL-2 (R&D System, Minneapolis, MN) were diluted in ethanol, water and PBS containing 0.5% BSA, respectively.
Flow cytometry
R-Phycoerythrin conjugated anti-mouse I-A
b, H-2D
b, CD4, CD40, CD80, CD86 and Pan-NK (DX5), FITC conjugated anti-mouse I-A
b and CD3, PE-Cy5.5 conjugated anti-mouse CD8 and CD11c mAb, isotype controls matched to each antibody (Pharmingen, San Diego, CA) and FITC conjugated annexinV (Caltag, Burlingame, CA) were used in flow cytometry experiments as previously described [
4]. The gated cell populations included both apoptotic and non-apoptotic cells. In all preparations utilized here, the adherent cells studied were >95% viable by trypan blue exclusion.
Tumorigenesis assays
TSA-treated or untreated B16 cells (1 × 105 trypan blue negative adherent cells) were injected s.c. in the ventral trunk of the mice. This tumor challenge dose is 10 fold higher than the minimum number of untreated cells required to generate palpable tumor in 100% of mice within 3 weeks after injection. Treated cells were assayed prior to injection to ensure consistency between experiments in MHC and costimulatory molecule expression as well as apoptosis. Tumors were measured every 3 days and mice were euthanized when tumor diameter reached 1 cm. After 40 days, all tumor-free mice were re-challenged s.c. with untreated B16 (1 × 105) cells in the opposite site and observed for another 60 days. In addition to palpable tumor measurement, the absence of evidence of tumor in immune mice was confirmed by visual inspection. At the end of the study period (100 days after vaccination), a mouse from each group of tumor-free mice was euthanized and the tumor inoculation sites and regional lymph nodes dissected and examined for evidence of tumor (no visible tumor nodule or other evidence of tissue aberration) and compared with a palpable tumor-bearing mouse (visible tumor nodules in lymph nodes). Some tumor-free B6 mice were re-challenged with EL4 (1 × 104) cells simultaneously as a control for tumor specificity. For the treatment model, B6 mice bearing palpable (~0.5 mm) B16 tumor, 5 days after s.c implantation in the trunk, were vaccinated with TSA- treated and irradiated (2000 Gy) B16 cells in the opposite side. Groups of tumor-bearing control mice received irradiated B16 cells or were left untreated. Mice that became tumor-free were re-challenged 42 days latter with wild type B16 in the trunk and observed for another 40 days.
Cytotoxicity assays
To determine tumor immunity, tumor-free mice, 30 days after vaccination with TSA-treated B16 cells, were re-challenged with untreated B16 cells and observed for an additional 15 days. Spleens isolated from three immune or control mice were disrupted and splenocytes were purified over nylon wool (Polyscience, Warrington, PA). Non-adherent lymphocytes (5 × 10
5) were re-stimulated with TSA-treated (500 nM for 48 h) and irradiated (200 Gy) tumor cells (2.5 × 10
5) in RPMI-1640 with IL-2 (10 U/ml). After 4 days, T-cell-enriched (>90%) viable cells were isolated after centrifugation through Ficoll-Paque and the level of anti-B16 cytotoxicity was determined utilizing a standard 4.5 h
51Cr-release assay [
4].
ELISpot assays
The mouse IFN-γ ELISpot kit (BD Bioscience, San Diego, CA) was used to determine antigen specific IFN-γ-secreting cells in spleens of immune mice as described [
19]. Briefly, RBC-depleted splenocytes (1 × 10
6 cells/well) isolated from immune and control mice were incubated in triplicate wells with B16 tumor cell lysate (~8 × 10
4 cell lysate/well) or 5 μM mgp100
25–33 peptide (Invitrogen, Grand Island, NY) for 24 h. IFN-γ spots were developed with AEC substrate and counted using a Zeiss Imaging system. B6 mice bearing palpable tumor 15 days after inoculation of untreated B16 (1 × 10
5 cells) and naïve mice were used as controls. Splenocytes (2 × 10
5 cells/well) from naïve mice treated with PMA (40 ng/ml) and Ionomycin (1 μM) served as positive control. For in vitro antigen presentation assays, popliteal lymph node T cells (2 × 10
5), isolated from ovalbumin-primed OT-II mice and purified magnetically (Pan T cell isolation kit, Miltenyi, Auburn, CA), were assayed in triplicate with TSA-treated (500 nM for 48 h) or untreated B16 cells (1 × 10
5) for 24 h. Splenocytes isolated from OT-II mice were used as control APCs. Tumor cells and splenocytes were pulsed with ova-peptide
322–338 or the control E333A ova-peptide
322–338 (10 μM) and irradiated (2000 Gy and 30 Gy respectively) before use in the ELISpot assays to measure IFN-γ-secreting T cells.
In vivo depletion studies
Anti-CD4 (GK1.5) and CD8 (2.43) hybridomas were grown i.p. in SCID mice and ascites fluid was partially purified by ammonium sulfate precipitation. To deplete CD4+ or CD8+ T cells, B6 mice were injected i.p. with CD4 or CD8 mAb (200 μg in 200 μl PBS/mouse), given at day -2, 0, +2, +4, +8 and +14. Rabbit anti-asialoGM1 γ-globulin (Wako), reconstituted in water and diluted in PBS, was used for NK cell depletion using the same schedule. Control mice received purified rat IgG (Chemicon International, Temecula, CA). Control and depleted mice were vaccinated with TSA-treated B16 cells after the 2nd dose of antibody injection and observed for 30 days. Depletion was assessed on the day of vaccination and a week later by flow cytometric analysis of splenocytes for CD3+CD4+, CD3+CD8+ T cells and DX5+ NK cells.
Bone marrow chimera generation
Bone marrow (BM) cells were harvested from femur and tibia of groups of donor (B6 and
class I-/--II-/-) mice by flushing the bones with RPMI-1640. RBCs were lysed using Tris-buffered ammonium chloride and single cell suspensions were obtained by passing BM cells through a cell strainer [
20]. T cells were depleted from BM preparations by incubation with CD8 (53-6.7) and CD4 (GK 1.5) mAb (1:10 dilution) for 60 min at 4°C, followed by lysis with low-tox-M rabbit complement (Cederlane, Ontario, Canada) for 90 min at 37°C in 5% CO
2 incubator [
21]. In preliminary experiments, BM cells after treatment with hybridoma supernatants (53-6.7 or GK1.5) were analyzed by flow cytometry using CD3, CD4 (L3T4) and CD8 (5H10) mAb to confirm the depletion of CD3
+CD4
+ or CD3
+CD8
+ T cells. One day before BM infusion, recipient B6 mice were injected i.p. with a single dose anti-asialoGM1 (20 μl in 200 μl sterile PBS) antibody to prevent NK mediated rejection of BM. The recipient mice were exposed to 11 Gy total body irradiation (TBI) administered in two treatments 3 h apart from a
137Cs-radiation source on the day of BM infusion. A total of 5 × 10
6 BM cells from donor mice were injected into the tail vein of each recipient. Chimeras were provided with water containing 2 mg/ml neomycin sulfate for 4 weeks after irradiation. Splenocytes, isolated from irradiated mice 2 days after TBI, were analyzed by flow cytometry to confirm the elimination of CD11c
+ APCs and CD3
+ T cells and >99% elimination was observed. To obtain a dendritic cell (DC) enriched APC population, low-density splenocytes were isolated by collagenase digestion [
22]. At 4–8 weeks after BM transplant, chimerism was evaluated using PBLs and splenocytes isolated from chimeric mice.
Isolation of apoptotic tumor cells
Apoptotic B16 cells were isolated using annexinV microbead kit, MS separation column and magnetic separator (Miltenyi, CA) according to the manufacturer's protocol. Briefly, TSA or H2O2-treated B16 cells were labeled with annexinV microbead for 15 min at 6–12°C and passed through a separation column in a magnetic field. AnnexinV positive (an+) apoptotic cells were retained in the column while annexinV negative (an-) non-apoptotic cells ran through. Apoptotic cells were eluted after removal of the column from the magnetic field.
Discussion
This study demonstrates that vaccination with TSA-treated melanoma cells expressing both MHC class II and costimulatory molecules can generate tumor-specific durable immunity and that the vaccine is effective against an existing B16 tumor in a treatment protocol. The induction of tumor immunity described here requires CD8+, CD4+ T and NK cells. Experiments performed in MHC class I knockouts and CD8+ T cell antibody-depleted mice substantially reduced the anti-tumor effects of the vaccine thereby supporting the essential role of CD8+ T cells in this model. Additionally, the partial abrogation of the immune responses in class II knockout mice, the lack of immunity in MHC double knockout mice and the tumorigenesis data in CD4+ T cell antibody-depleted animals suggest that CD4+ T cells are also essential in the generation of immunity. The data from T cell deficient or depleted mice together with the findings of cytotoxic and IFN-γ-secreting lymphocytes in immune mice in association with long term immunity demonstrate that a single dose of the TSA-treated tumor cell vaccine is capable of activating naïve T cells to generate cytotoxic effector and memory T cells. Our results also showed that depletion of NK cells completely abrogated immunity generation by TSA-treated B16 cells suggesting that NK cells, together with T cells, play an important role in initiation of immunity following epigenetic vaccination.
The association of immunity with enhanced class II and costimulatory molecule expression and in vitro activation of antigen-specific CD4
+ T cells by TSA-treated melanoma cells is consistent with the hypothesis of direct antigen presentation by tumor cells in vivo. That direct presentation could be a component of tumor immunity is suggested by: 1) the reported demonstration of direct presentation in viral and other tumor model systems [
7,
29]; 2) the effectiveness of tumor vaccines established by transfecting tumor cells with MHC and costimulatory genes [
30]; and 3) the conversion of tumor cells to APCs by epigenetic agents discussed above and previously reported [
14,
18]. The substantial effectiveness of the TSA-modified tumor cell vaccine shown here, together with the delayed tumor generation data in MHC deficient chimeras, suggest a short-lived in vivo effect from direct antigen presentation. Our view is that cross-presentation is likely the major mechanism of T cell activation in most tumors (as outlined by others and reviewed in [
2]) but that, in certain tumors and conditions, a component of direct presentation may exist. The important issue is whether clinically significant direct presentation can be established by appropriate treatments designed to convert a sizable portion of the tumor cells in the tumor bed to APCs. Since tumors may have a billion or more cells when first detected [
31], converting even a small percentage of cells could potentially have significant effects. Furthermore, we would expect that Treg cells and other immunosuppressive mechanisms might also repress direct presentation. It remains to be determined whether combinations of the epigenetic vaccine with other modalities, such as adjuvants or anti-Treg antibody, can enhance the efficacy of direct antigen presentation by tumors. Current evidence suggests that sustained antigen presentation is important in promoting immunity and that the duration of antigen presentation is a major factor in activation of memory T cells [
32,
33]. The short-term effects seen in vivo in the direct presentation model could reflect the lack of sustained antigen presentation. Further experiments are required that include multiple injections, adjuvants with sustained antigen release, further adjustments of amount and concentration of the apoptotic population and other procedures employed for DC vaccines directed at enhancing and maintaining MHC/peptide complex and maintaining the adjuvant environment and activation state of the tumor [
33,
34].
Since HDACi affect the expression of many genes [
12], we consider it likely that additional mechanisms other than direct presentation may contribute to epigenetically induced tumor immunity. In this regard, TSA treatment can enhance the expression of NKG2D ligands [
35] on tumor cells, which may activate NK cell mediated immune responses and synergize with CD28 in inducing proliferation and IFN-γ production by CD8
+ T cells [
36]. However, whether TSA induction of NKG2D ligands is mediated by chromatin or stress effects on tumor cells is uncertain. In addition, although our data and other reports suggest that NK cells can mediate immunity against tumor cells expressing NKG2D ligand [
37] or MHC class I [
38], the role of NK cells in generating a long-lasting memory immune response after epigenetic vaccination is uncertain. One possibility suggested by the strict requirement for NK cells is that NK cells interact with tumor cells treated with TSA (and expressing CD40) as described for DC [
39,
40] and enhance antigen presentation [
33].
Apoptotic tumor cells could potentially contribute to immunity generation through several pathways. Previous studies suggested that phagocytosis of apoptotic cells by DCs in certain conditions can induce tolerance [
41], while other reports demonstrate that cross-presentation of antigen from apoptotic tumor cells can generate tumor specific immune responses [
42]. The experiments performed with TSA-treated apoptotic B16 preparations demonstrated that apoptotic tumor cells are capable of eliciting long-term immunity. The induction of immunity by apoptotic tumor cells has been related to their ability to induce inflammation and to mature DC, in part owing to the presence of heat shock proteins in apoptotic cell preparations, especially those containing necrotic cells [
1,
43,
44]. Since apoptotic cells were defined by annexinV binding in our study, the presence of an
+ necrotic cells is likely in TSA-treated apoptotic preparations. It has been reported that apoptotic tumor cells produced by treatment with TSA have different expression patterns of MHC class II and costimulatory molecules compared to apoptotic cells elicited by L-phenylalanine mustard, H
2O
2 and γ-irradiation [
28]. Moreover the data presented here shows that apoptotic tumor cell preparations produced by TSA, but not H
2O
2, treatment generated long-term immunity. The failure of the H
2O
2-treated apoptotic B16 cells to enhance immunity demonstrates a selective effect of TSA-treated apoptotic cells and suggests that apoptosis inducing agents should be carefully evaluated for differential effects in anti-cancer therapy. In this regard, recent evidence suggests that apoptotic cell death due to caspase-3 activation results in the release of bioactive lipid molecules, which promote DC maturation and pro-inflammatory responses [
45]. Since TSA may induce apoptosis in certain tumor cells through a caspase-3 and mitochondria dependent pathway [
46], it is possible that TSA-treated apoptotic B16 cells release bioactive lipids, which promote innate, perhaps TLR and NK mediated, immunity as well as adaptive responses.
The reported enhancement of DC vaccine potency by the addition of apoptosis promoting agents suggests a role for apoptotic cells in synergizing with live cells in eliciting immunity [
47]. Our results showed that addition of TSA-treated non-apoptotic cells to apoptotic (50% an
- + 50% an
+) B16 inocula significantly improved the vaccine efficacy. Therefore, in addition to the conversion of tumor cells to APCs, the combination of replicating tumor cells producing high levels of antigen with an apoptotic adjuvant-like effect could explain the superiority of the high dose (500 nM) TSA-treated B16 vaccines compared with lower doses of TSA and the H
2O
2-treated cells. Also, since TSA alters chromatin and indirectly inhibits DNA methylation [
48], it could potentially activate repressed endogenous tumor antigens, such as MAGE [
49] and HMW-MAA [
50] that may serve as epigenetically induced tumor antigens. Finally, HDACi modulates DNA damage (ATM/ATR) [
35] and MAPK pathways [
51] and could therefore serve as an intracellular adjuvant, which have been shown to enhance DC vaccine potential [
34].
The data presented here suggests that an autologus tumor cell vaccine generated by HDACi in vitro is effective in both preventative and treatment models. Further work, currently in progress, will focus on obtaining more effective vaccines using different HDACi preparations, combination with adjuvant or anti-Treg as well as multiple vaccinations. We would envision future clinical protocols using this vaccine, most likely in combination with surgery and/or other strategies, to treat recurrence and metastasis of tumors. Clearly there are challenges and potential problems in adapting a cell-based vaccine of this type to the treatment of human cancer. If clinically effective vaccines could be developed, they would have the advantage of being histocompatible with the host and applicable without prior knowledge of the tumor antigens involved. These studies also suggest that current therapies with epigenetic agents could affect host immunity to the tumor and attention should be given to this aspect in ongoing and future clinical protocols using these agents systemically.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
A. Nazmul H. Khan – had the primary responsibility for designing and conducting all the experiments, data analysis and initial drafting of the manuscript; William J. Magner – was involved in study design, drafting and review of the manuscript; Thomas B. Tomasi – conceived the study, contributed to the design and subsequent drafts of the manuscript. All authors read and approved the final manuscript.