Review For reprint orders, please contact:
[email protected]
Immunotherapy of chronic myeloid leukemia: present state and future prospects In spite of the considerable successes that have been achieved in the treatment of chronic myeloid leukemia (CML), cure for the disease can only be obtained by the present means in a rather small minority of patients. During the past decade, considerable progress has been made in the understanding of the immunology of CML, which has raised hopes that this disease may be curable by supplementing the current targeted chemotherapy with immunotherapeutic approaches. More than ten small-scale clinical trials have been carried out with experimental vaccines predominantly based on the p210bcr–abl fusion protein. Their results suggested beneficial effects in some patients. Recent data obtained in human patients as well as in animal models indicate that the p210bcr–abl protein does not carry the immunodominant epitope(s). These observations, combined with the recognition of an ever increasing number of other immunogenic proteins in CML cells, strongly support the concept that gene-modified, cell-based vaccines containing the full spectrum of tumor antigens might be the most effective immunotherapeutic approach. Recently created mathematical models have provided important leads for the timing of the combination of targeted drug therapy with vaccine administration. A strategy of how targeted drug therapy might be combined with vaccination is outlined. KEYWORDS: antigenic make-up n cellular vaccine n combination therapy n chronic myeloid leukemia n immunity n peptide vaccine n tumor cell
Chronic myeloid leukemia (CML) is a malignant myeloproliferative disease of the pluripotent hematopoietic stem cell and is characterized by extensive proliferation and differentiation of myeloid cells. It is a relatively rare malignancy, with an incidence being approximately 1–2/100,000. The course of the disease can be divided into three stages: the chronic phase, the accelerated phase and blastic crisis. The latter stage proceeds in a similar manner to acute leukemia and has very bad prognosis. CML was the first malignancy recognized to be associated with a unique chromosomal change, in particular with the Philadelphia chromosome (Ph+). This structure is a consequence of a reciprocal translocation between chromosomes 9 and 22: t(9;22)(q34;q11). The traslocation results in the formation of the bcr–abl fusion gene. The product of this gene, the cytoplasmic chimeric BCR–ABL protein, displays an abnormally high tyrosine-kinase activity, which is coded for by the abl gene. The overproduced enzyme activates a number of cellular signal transduction pathways, which results in malignant transformation. It is generally accepted that the product of the fusion gene is responsible for both the transformation and the maintenance of the transformation state. The fusion between the bcr and abl
genes can occur at different sites. As a consequence, different forms of BCR–ABL protein develop, markedly differing in their size. The one most frequently detected (in more than 90% of cases) is the p210 fusion protein, designated as p210bcr–abl. However, even this protein exists in two configurations, e13a2 (i.e., b2a2) and e14a2 (i.e., b3a2), which mutually differ in the location of the breakpoint either between exons 13 and 14 or between exons 14 and 15 of the bcr gene. In both configurations, the fusion is associated with the generation of a new amino acid sequence. The peptides spanning the breakpoints between BCR and ABL proteins are specific to CML cells and thus represent unique tumor antigens. The only method used to cure CML today is transplantation of allogenic hematopoietic stem cells. However, such transplantation is limited to only a minority of patients and is associated with relatively high morbidity and mortality. Fortunately, dramatic progress in the therapy of CML has been achieved in recent years. The drugs used to date, namely busulfan, hydroxyurea and IFN-a, are being rapidly replaced by molecularly targeted drugs. The first of these disease-tailored drugs has been imatinib-mesylate (IM), now known under the commercial name of Glivec (or Gleevec®, Novartis Pharmaceuticals
10.2217/IMT.10.2 © 2010 Future Medicine Ltd
Immunotherapy (2010) 2(2), 227–241
Vladimír Vonka Department of Experimental Virology, Institutute of Hematology & Blood Transfusion, U nemocnice 1, 128 20 Prague 2, Czech Republic Tel.: +420 221 977 383 Fax: +420 221 977 392
[email protected]
ISSN 1750-743X
227
Review
Vonka
Corporation, NJ, USA). It blocks the ATPbinding pocket on the tyrosine kinase, thus preventing the activation of this enzyme. In addition to Abl kinase it also inhibits c-Kit tyrosine kinase, and the platelet-derived tyrosine kinase but does not affect other closely related protein kinases [1] . IM is well tolerated and in a great majority of patients it induces long-lasting hematological and cytogenetic remissions in the chronic phase of the disease. According to a recent update, a 5-year survival of IM-treated CML patients has been achieved in 89% of cases [2] . However, IM is much less effective in the accelerated or blastic phase of the disease [3] . Moreover, in a number of patients, resistance to the drug develops owing to either the mutations in the catalytic domain of the kinase or to amplification of the bcr–abl gene [4–6] . The problem is being partially solved by the gradual introduction of a new generation of targeted drugs, for example, dasatinib [7] or nilotinib [8] , both of which appear to be more potent against unmutated BCR–ABL than IM. In general, the results obtained with targeted drugs are very good indeed. However, these drugs cannot cure the disease, most likely owing to their failure to hit the nonproliferating cancer stem cells. After interrupting the therapy, the disease relapses sooner or later. In addition, there are well-based concerns that these substances might produce concurrent immunosuppressive effects. Conversely, immunostimulatory activities of these drugs have also been reported (refer to the recent review [9]). A significant number of investigators believe that the problem of curing CML might be solved by immunotherapeutic approaches employed either alone or, more likely, in combination with other therapeutic modalities. These efforts are strongly supported by the advances in understanding cellular and molecular processes that govern immune regulation.
Why immunotherapy of CML should be possible There are four reasons for considering CML a good candidate for immunotherapy. First, it develops slowly, which provides relatively long time periods for immunotherapeutic inter vention. Second, leukemic cells circulate in the blood and in the lymphatic system, which makes them easily accessible to cells of the immune system. Third, CML cells carry a specific and welldefined tumor antigen. As already mentioned, the junctions spanning BCR–ABL-derived sequences are unique moieties. They represent an ideal target for immunotherapy, as they are not 228
Immunotherapy (2010) 2(2)
present in normal cells. Moreover, both the ABL and BCR proteins are overexpressed in leukemic cells compared with their levels in normal cells. In addition, the reciprocal ABL–BCR fusion protein with its novel epitope may be a fitting target for specific immune reactions [10] . The fourth reason is that highly effective therapy of CML by other means is also available, which enables a reduction of the tumor mass prior to the initiation of immunotherapy. It should be noted that the research into tumor biology carried out during the past 10 years or so has proved beyond reasonable doubt that, in the development and activation of antitumor immunity, the cancer is not a passive player but it actively militates against it by producing immunsupressive and proangiogenic factors and by mobilizing certain other cells of the immune system with similar activities (refer to the recent reviews [11–13]). It follows that the smaller the cancer mass is, the weaker these effects are. The reduction of the tumor mass by the aforementioned targeted drugs creates very good conditions for applying immunotherapeutic approaches. Although the efforts aiming at immunotherapy have recently received significant new stimuli, there are a number of older observations that provide more or less indirect evidence for the existence and efficacy of immune reactions in CML. For example, these include the observation that in the peripheral blood of most healthy individuals bcr–abl transcripts can be detected, provided that sufficient amounts of leukocytes (>107) are tested [14,15] . Some researchers interpret the fact that these subjects remain healthy as evidence of the efficiency of their immune system. However, other explanations may be offered, for example, in these subjects the translocation did not take place at the stem cell level but in more differentiated cells, which could not be transformed into malignant cells. Some further indirect evidence for the role of the immune system in the pathogenicity of CML is the observation that certain HLA haplotypes are associated with a decreased risk of the disease [16] , while some other haplotypes seem to increase its risk [17,18] . These observations suggested the existence of a natural immunosurveillance against the antigens of the tumor cells. In addition, the efficacy of CML treatment with IFN-a is apparently mainly caused by its immunomodulatory activities [19–22] , although its antiproliferative effects should not be neglected. A clear demonstration of the importance of immune reactions in CML pathogenesis is the high efficiency of allogenic transplantations. T cells present in the transplant are capable of future science group
Immunotherapy of chronic myeloid leukemia
liquidating the tumor cells, most probably owing to their reactivity with the tissue restricted minor histocopatibiliy antigens exclusively expressed in hematopoetic cells [23,24] . This type of treatment is not risk-free. The antileukemia effects can be followed by graft-versus-host disease, with serious complications that may result in the patient’s death. A major breakthrough in the study of CML immunology has been the demonstration of the immunogenicity of the peptides derived from the fusion zone of the p210bcr–abl protein presented in the context of MHC class I and II, a precondition for an immune response to be provoked [25–34] . Futhermore, T lymphocytes specifically activated not only against the fusion zone of the p210bcr–abl protein but also against the extrajunctional epitopes of the fusion protein [35] and against a number of other leukemic cell proteins have been detected in the blood of CML patients (see the proceeding sections). This convincingly indicates that under natural conditions some, but apparently not fully effective, immune reactions against leukemic cells do develop. One can hypothesize that boosting this immunity with vaccination might result in a durable remission or even cure the disease.
Review
Trends in immunotherapy of CML In addition to the aforementioned procedures used in the treatment of CML – in other words, IFN-a administration and allogenic transplantation, which can be labelled as immuno therapeutic, more direct approaches have been employed in the past several years. These include the use of various immunostimulators, the implantation of ex vivo-generated cytotoxic T lymphocytes and, especially, tumor-specific vaccines. A total of 13 minor clinical trials have been reported to date. They are listed in chrono logical order in Table 1. Most of the preparations tested were directed against the p210bcr–abl protein and most of these comprised mixtures of peptides spanning the fusion point. Cellular vaccines based on autologous dendritic cells have also been used. Since leukemia cells have been shown to produce some proteins known to be expressed only in fetal cells, or to overproduce proteins expressed in only small amounts in normal cells (see after), attempts have also been made to develop and test vaccines directed against these proteins. They included peptides derived from the Wilms’ tumor (WT)-1 protein, proteinase-3 (Pr-3) and the heat shock
Table 1. Vaccine trials in patients with chronic myeloid leukemia. Vaccine
Adjuvant
Clinical response
Immunological response
BCR–ABL Autologous dendritic cell-based vaccine Autologous dendritic cell-based vaccine Proteinase-3-derived peptide BCR–ABL-derived peptide mixture BCR–ABL-derived peptide mixture Leukocyte-derived Hsp70–peptide complex BCR–ABL-derived peptides
QS-21 –
No No
Yes Yes
[94]
QS-21
No
Yes
[95]
GM-CSF
Yes
Yes
[96]
QS-21
Yes
Yes
[97]
QS-21+ Molgramostin –
Yes
Yes
[98]
Yes
Yes
[99]
PADRE+ GM-CS –
Yes
Yes
[100]
Yes
Yes
[101]
Montanide ISA 51 GM-CSF – Montanide ISA 51+ GM-CSF –
Yes
Yes
[102]
Yes Yes
Yes Yes
[104]
Yes
Yes
[36]
Autologous dendritic cell-based vaccine Pr-3 and WT-1-derived peptides BCR–ABL-derived peptides BCR–ABL-derived peptides Inactivated, genetically modified (GM-CSFexpressing) K562 cells
Ref. [93]
[103]
GM-CSF: Granulocyte–macrophage colony-stimulating factor; Hsp: Heat shock protein; IFA: Incomplete Freund adjuvant; PADRE: Pan DR epitope; Pr: Proteinase; QS: Quillaka saponaria; WT: Wilms’ tumor.
future science group
www.futuremedicine.com
229
Review
Vonka
protein (Hsp)–70-peptide complex. In one trial, a vaccine based on the p210bcr–abl-positive human cell line was used. For this purpose, K562 cells, originally isolated from a patient with CML in blastic crisis, were employed. Prior to their use as an allogenic vaccine, the cells were genetically modified by introducing a transcriptionally active gene for granulocyte–macrophage colony-stimulating factor (GM-CSF) and were inactivated by radiation. In several of the studies listed in Table 1, patients with acute leukemia and patients with the myelodysplastic syndrome were also included. In general, the results were favorable; the vaccines were tolerated well and both immunological and clinical responses varying from a reduction in the bcr–abl transcript level to complete molecular remission were reported in a portion of patients in nearly all of the studies, including that in which gene-modified K562 cells were used. Unfortunately, detailed results of this trial are not available. They are only briefly mentioned in the authors’ recent review [36] . Therefore, it is difficult to comment on this highly interesting study. However, the results of the recent clinical trials must be evaluated with caution. The patient groups were rather small (3–20 patients), none of the undertakings was carried out as a double-blind study and, for obvious reasons, the survival end point could not be determined. Furthermore, in all of them, other standard therapeutic modalities were also employed (before the vaccination or, more frequently, concurrently with it), and thus it is difficult to reliably differentiate between the effects of this therapy and of the anti-CML vaccination. Another weakness is that in some of the studies no clear correlation between the clinical and immunological responses could be demonstrated. Owing to the use of different vaccines, different immunization schedules and of different methodologies for examining immune responses, meta-analysis of the data from these trials is extremely difficult. Although the data strongly suggest that beneficial effects really did occur in a subset of the patients, a longer follow-up is needed. In spite of this criticism, the remarkable efforts undertaken with the introduction of the first anti‑CML vaccines should be highly appreciated.
The BCR–ABL protein does not carry the immunodominant epitopes Until very recently, a great majority of the efforts aiming at immunotherapy of CML were directed at the unique epitope carried by the p210bcr–abl protein. Two studies performed by Brossart’s 230
Immunotherapy (2010) 2(2)
team in Tübingen, Germany, have brought an important twist in the tale and have dramatically changed the prevailing concept. Using dendritic cells transfected by RNA extracted from K562 cells and blasts from CML and acute myeloid leukemia patients, and employing CTL assays, they convincingly demonstrated that p210bcr–abl-carried antigens were not the dominant immunogens of CML cells [37] . However, in their subsequent paper, they demonstrated that the expression of the p210bcr–abl protein is a conditio sine qua non for the expression of a number of tumor antigens and for immunogenicity of the tumor cells [38] , thus indicating the crucial role of p210bcr–abl-associated activities in creating the tumor cell antigenic make-up. Similar results were nearly simultaneously reported by another group [39] . Accordingly, treatment with IM, at variance with IFN-a treatment, results in an impaired CTL response against CML cells, as had been observed earlier [21] . If p210 bcr–abl-derived antigens are not the dominant immunogens, which other cellular proteins are their carriers? Although such antigens have not yet been identified, it has been demonstrated that a number of self-proteins are aberrantly overexpressed in leukemic cells relative to normal hematopoietic cells. It is likely that their present number is not final. The antigens that attract increased attention are listed in Table 2 . The table also shows whether T lymphocytes are specifically reactive with these antigens as demonstrated directly in CML patients or were induced ex vivo. With a possible exception of the ABL-BCR protein, whose function is unknown; and which may just be a ‘fellow traveler’ of the p210bcr-abl protein, it is highly probable that the aberrant expression of most of the other proteins listed is of significance in the pathogenesis of CML. This suggestion is based on their known or suspected functions or on their overproduction in other human ca ncers. Some of those findings and their interpretation follow. Thus, the aforementioned WT-1, which is a zinc finger transcription factor, was originally considered to be a tumor suppressor, but it apparently acts as an oncoprotein under certain circumstances. Aberrantly expressed in CML cells are the primary granule proteins (PGP; involved in granule formation), namely Pr-3 (also previously mentioned), human neutrophil elastase (HNE) and cathepsin G (Cat-G). Pr-3 (also termed myeloblastin), which, together with WT-1 is presently in the focus of interest, belongs to the myeloid serine proteinase family stored in primary azurophilic granula and is future science group
Immunotherapy of chronic myeloid leukemia
maximally expressed at the promyelocytic stage. It is upregulated in the accelerated phase and blast crisis. It has been demonstrated to cleave p21waf1, an important cell regulatory protein apparently playing a key role in hematopoietic quiescence. This could be a mechanism contributing to cellular transformation [40] . The possible role of HNE in leukemogenesis is not yet clearly understood. It has been suggested that its overproduction can change the growth environment by inhibiting growth factors, such as granulocyte colony-stimulating factor (G-CSF) or GM-CSF, which gives selective advantage to the bcr–abl-transformed cells [41] . Cat-G is a serine protease released upon degradation of neutrophils. In other systems it is suspected to modulate interactions between tumor and stromal cells, favoring tumor growth, and has been implicated in playing an important role in tumor invasion and metastasis [42] . Bcl-2 and BclxL are among the members of the Bcl-2 family of proteins with a strong antiapoptotic effect. Their overexpression has been documented in a number of cancers including various hematological malignancies [43] . G250 is a carbonic anhydrase IX, a membraneous protein that is known to regulate cell proliferation as a reaction to hypoxic stimuli. It is strongly suspected of playing a role in oncogenesis and tumor progression [44] . Increased amounts of the human telomerase reverse transcriptase (hTERT), which is present in CML cells, have been detected in a great majority of human tumors and it is already considered a broadly utilizable immunologic target [45] . The same holds true for Hsp70, which has already been used as an experimental vaccine (Table 1) . Mitosis-associated phosphoprotein 11 (MAPP11), which acts as a chaperone, is one of the phosphoproteins associated with cell division; most probably it plays a regulatory role in mitosis. Its overexpression in some human tumors has been well documented and its role in their pathogenesis has been suggested [46] . Helicase antigen (HAGE), preferentially expressed in melanoma (PRAME), Aurora (Aur) A kinase and CML28 belong to the family of cancer/testis antigens. HAGE is overexpressed in many human cancers [47] . Its function is not well known; however, since it carries the DAED motive, which is shared by ATP-dependent RNA helicases, it is considered a member of this family and is assumed to be involved in RNA metabolism. The bcr–abl transcript levels correlate with HAGE expression. PRAME functions as a co-repressor of retinoic acid in solid tumors, its role in hematological future science group
Review
Table 2. Chronic myeloid leukemia-associated antigens. Antigen ABL–BCR WT-1 Pr-3 Cathepsin-G HNE Bcl-2 G250 hTERT Hsp70 MPP11 HAGE PRAME Au-A CML28 CML66 Adipophilin RHAMM Survivin NM23-H2 BCR
T-cell response in CML patients Yes Yes Yes Yes Yes Not tested Not tested Yes Yes Not tested Yes Yes Yes Yes Yes Not tested Yes Not tested Yes Yes
Ref. [10,34] [35,103,105,106,108] [35,36,96,109,111,112] [109] [109] [38] [107] [35] [99] [107] [110] [112–114] [115] [116] [117] [38] [107,118] [38,119,120] [121] [35]
CML-specific T cells were present in the patients or were detected after induction ex vivo. Not tested: the author did not find any information on the immunogenicity of the respective proteins in CML patients; however, the immmunogenicity of Bcl-2 , BclxL, G250, adipophilin and survivin was demonstrated in patients with other cancers [122–126]. CML: Chronic myeloid leukemia; HAGE: Helicase antigen; HNE: Human neutrophil elastase; Hsp: Heat shock protein; hTERT: Human telomerase reverse transcriptase; MPP: Mitosis-associated phosphoprotein; PRAME: Antigen preferentially expressed in melanoma; RHAMM: Receptor for hyaluronic acid mediated cell motility; WT: Wilms’ tumor.
malignancies is not yet known but it is suspected to block myeloid differentiation [48] . Owing to its presence in some cancers, it was already included among the tumor-associated antigens some time ago [49] . Aur A is one of the serine–threonine kinases regulating cell division and has been implicated in tumorigenesis. Its overexpression in human cancers is associated with poor prognosis [50] . CML28, expressed in both hematological and nonhematological cancers but not in normal tissues (except the testes), is probably involved in RNA processing [51] . CML66 is a differentiation antigen strongly expressed in human tumors, including leukemias, but its function is not yet known. It is suspected of playing a role in tumor-cell proliferation. It has recently been demonstrated that the silencing of the CML66 gene can suppress tumor growth [52] . Adipophilin is involved in lipid storage but can be found in other cells, including tumor cells [53] . The receptor of hyaluronic acid-mediated cell motility (RHAMM) is overexpressed in many tumors. It participates in control of mitotic activity and is involved in cell motility, wound healing and invasion. Survivin, which inhibits apoptosis by blocking www.futuremedicine.com
231
Review
Vonka
caspases 3 and 7 [54] , is expressed in embryonic but absent in terminally differentiated cells. It is overproduced in many cancers, including hematological malignancies. In CML, its expression increases with the progression of the disease. Downregulation of survivin results in inhibition of cell growth [55] . Nonmetastasis protein 23 (NM23-H2) is a member of the nucleoside diphosphate kinase family. It is involved in cell proliferention, differentiation and in some other cell processes, and apparently activates the cMyc oncogene [56] . The outcome of NM23-H2 overexpression, observed in many malignancies, seems to be different in different cancers; in some it increases, in others it is associated with a decrease of their aggressiveness. It may also be of interest that in CML patients, but not in normal subjects, specific immune reactions against BCR-derived peptides located outside the fusion zone were demonstrated [35] . It remains to be determined whether this is associated with the delocation of the BCR–ABL protein from the nucleus to the cytoplasm, which might result in the different processing or whether other mechanisms are involved. Three observations may be of special interest. First, the aforementioned proteins have been found expressed in the CML cells of significant but varying proportions of patients (25–90%) [57] ; however, none of them was detected in all the patients. Second, in many cases, the extent of their production depended on the stage of the disease. Third, wherever tested, specifically activated T cells were detected, indicating that the respective antigens are properly processed and expressed on the CML cell surface in context with HLA molecules. It is not understood why these immune mechanisms do not operate effectively in vivo. Should they be fully functional, they would prevent the development of the disease. It has been speculated that this might be due to immune editing [34] , which causes a gradual escape of CML cells from T-cell recognition, or to immune suppressor molecules secreted within the tumor microenvironment. It is also possible that the interaction between the programmed death (PD)-1 signaling molecule, highly expressed in CD8 + T cells from CML patients, and CML cells expressing PD-ligand (PD-L)-1 contributes to the functional exhaustion of cytotoxic T lymphocytes, as recently suggested [58] . These authors have shown that specifically activated T cells were not able to function as fully differentiated effectors ex vivo. However, these data suggest that immune reactions against any of these proteins play a role in 232
Immunotherapy (2010) 2(2)
the course of the disease and that any or, preferentially, all of them could be included in CML therapeutic vaccines. One can expect that in the near future the exploitation of the rapidly growing databases created in the fields of genomics and proteomics will enable a more effective use of in silico tools for predicting T-cell epitopes in cancer patients and that the outcome of these investigations will be utilized for monitoring immune reactions, for identifying potential vaccine candidates and, in the final run, for optimizing the vaccine protocols [59] . To our knowledge, such efforts in CML have not yet been reported. In spite of considerable successes, there are still many pitfalls with the present in silico tools, resulting in frequent inconsistencies and discrepancies. The present difficulties are apparently associated not only with the in silico methods being used but also with the antigens tested, the alleles examined and the validation strategies being employed. They make predictions in the case of self-antigens not very reliable [60] . Before reliable data in CML can be obtained, it will be necessary to reduce the limitations of the present in silico tools.
Studies in animal models & some conclusions drawn therefrom The first vaccine studies shown in Table 1 indicated that the experimental vaccines used were well tolerated and apparently produced a moderate beneficial effect. Similar to some other investigators, we are convinced that prior to developing and optimizing vaccines for human use, it will be necessary to obtain more information on the immunology of bcr–abl-transformed cells in animal models. Some pioneering work in this respect has been performed by Katsanis and his colleagues at the University of Arizona, AZ, USA. When immunizing with a nona-peptide covering the fusion point of the hybrid protein [61] , they succeeded in inducing immunity against challenge with 12B1 cells, a mouse (Balb/C) bcr–ablgene-transformed cell line producing leukemia and subcutaneous tumors in syngeneic animals. In another series of experiments, they demonstrated that the chaperone-rich cell extracts were effective vaccines against cancer induced in mice by bcr–abl-transformed syngeneic cells [62] . Several other groups have recently reported highly interesting and stimulating immuno logical studies with bcr–abl-transformed cells in murine systems [63–65] . We began immunization experiments with mouse bcr–abl-transformed cells several years ago. First, we undertook to test and compare the future science group
Immunotherapy of chronic myeloid leukemia
potency of several genetic vaccine types, namely DNA vaccines, recombinant viral and protein vaccines, vaccines based on virus-like particles and cell-based vaccines, in immunization-challenge experiments. In most cases, our experimental vaccines were based on the 25 amino acid-long fusion zone of the p210bcr-abl protein, having the e14a2 configuration – in other words, with lysin in the center linked with 12 amino acids derived from the C-terminal of the BCR protein fragment and 12 amino acids representing the N-terminal of the ABL protein fragment. For our experiments we used, in addition to the 12B1 cells developed by McLaughlin et al. [66] and kindly provided by Katsanis, also the B210 cells developed by Daley and Baltimore [67] and kindly provided by Daley (Whitehead Institute of Biochemical Research, Cambridge, MA, USA). Both cell lines produced comparable amounts of p210bcr–abl protein with the e14a2 configuration and both induced a disease resembling acute leukemia; however, they differed in several important aspects [68,69] . The 12B1 are approximately 100-times more oncogenic than the B210 cells after intravenous administration and, in addition, they induce solid tumors after subcutaneous administration. These tumors exhibit a high propensity to metastasize to other organs, primarily to the spleen, liver and bone marrow. Owing to their capability of inducing measurable solid tumors, we used 12B1 cells in most immunization experiments. The first experiments brought a series of disappointments. Although we were able to induce protection against challenge with the highly aggressive 12B1 cells by a DNA vaccine that carried the whole fusion gene, the other DNA vaccines carrying only the fragment coding for the fusion zone failed to do so, in spite of the fusion gene fragment having been linked to other genes whose products are known to significantly enhance the development of anti-tumor immunity [70] . This indicated that the immunizing epitopes were located outside the selected fusion zone. Our attempts to induce immunity by the recombinant vaccinia virus expressing the 25-amino-acid fragment of p210bcr–abl or by the pentaeicosapeptide recombined with the detoxified CyA protein of B. pertussis were also unsuccessful [Nĕmečková S, Lučanský V et al., Unpublished Data] . Furthermore, vaccination with the chimeric polyoma virus-like particles carrying inside the particles a 171-amino-acid long sequence of the hybrid p210bcr–abl protein, which includes, in addition to the 25-amino-acid fusion zone, the adjacent regions of the BCR and ABL proteins, future science group
Review
was also a complete failure [71] . All these results indicated that, in the murine system used, the new epitope created by the fusion of the two proteins was not capable of inducing protection against challenge with syngeneic bcr–abl-transformed cells. Of course, it is possible that the failure of the vaccination was associated with the mouse strain used and that the same vaccines might be efficient in other mouse strains or in HLA transgenic mice. We were more successful with the cellular vaccines based on gene-modified bcr–abl-transformed cells. Already, the first results indicated that it was relatively easy to induce protection against challenge with B210 cells by immunization with irradiated homologous cells or 12B1 cells or with live B210 cells administered intraperitoneally, or with sublethal doses of these cells administered intravenously. On the other hand, it was much more difficult to induce protection against 12B1 cells. None of the animals immunized with irradiated B210 and only approximately half of those immunized with irradiated 12B1 cells were protected [Sobotková E, Vonka V et al., Unpublished Data] . The next step was to use the cell-based vaccines therapeutically. For this purpose, we prepared a set of gene-modified cells expressing IL-2 or GM-CSF or IL-12. The first generation of these vaccines was derived from B210 cells [72] . The expression of the immuno stimulatory cytokines was associated, in all three examples, with the loss of oncogenic potential. Therefore, these constructs were used as live therapeutic vaccines in mice subcutaneously inoculated with approximately ten tumorinducing dose 50 (TID50 ) of the highly aggressive 12B1 cells [73] . Four repeated doses of any of the gene-modified cells administered during the incubation period delayed tumor development (p
