Cancer-testis antigens in haematological malignancies
Seah H Lim, MD, PhD, FRCP, FRCPath., Texas Oncology, 1000 Coulter Drive, Amarillo, Texas 79106, USA. E-mail: email@example.com
Immunotherapy is an attractive therapeutic option for patients with haematological malignancies. Until recently, the progress in the development of tumour vaccines for haematological malignancies had been slow due to the lack of suitable targets. Cancer-testis (CT) antigens are potentially suitable molecules for tumour vaccines of haematological malignancies because of their high immunogenicity in vivo and their relatively restricted normal tissue distribution. This review evaluates the properties and potential functions of CT antigens. We discuss the expression of CT antigens in patient with haematological malignancies and provide evidence in support of their immunogenicity in vivo in these patients. We also address the role of ‘epigenetic’ regulation of CT antigens in haematological malignancies and how hypomethylating agents could induce the expression of some of these antigens in tumour cells to overcome the problem of heterogeneity of expression of the antigen within individual tumour specimens. Data implicating the interaction of the promoter genes of some of these CT antigens with the MeCP2 protein also suggest the potential role of the histone deacetylase inhibitors in inducing antigen expression in tumour cells. Finally, we discuss the direction of future research in advancing the development of tumour vaccines for haematological malignancies.
Immunotherapy is an ideal approach for patients with haematological malignancies. Experience of patients undergoing allogeneic stem cell transplants, especially non-myeloablative transplants, suggest that the tumour cells from most haematological malignancies are susceptible to the cytotoxic effects of T lymphocytes. Therefore, provided suitable antigens are targeted in the optimal clinical setting, immunotherapy should be an effective therapeutic approach in patients with haematological malignancies. In addition to the likelihood of being less toxic and more specific than chemotherapy, immunotherapy may also provide a different mode of tumour cytotoxicity that may synergize with that induced by chemotherapeutic agents. It also offers the prospect of inducing immune memory that may prevent disease relapse.
Despite these potential advantages, advances in clinical immunotherapy have been slow because of the paucity of suitable targets for this purpose. Ideal tumour antigens are protein molecules that are specifically and stably expressed by the tumour cells but absent from normal tissues and crucial for the survival of the tumour cells. Unfortunately, except for some viral-associated tumour antigens, most of the characterised molecules are auto-antigens and have not fulfilled these criteria. Furthermore, auto-antigens are generally weak antigens, because of thymic negative selection to eliminate auto-reactive T cells, unless the auto-antigens are selectively or preferentially expressed only in immune sanctuary sites, such as the testis.
Cancer-testis (CT) antigens are a group of normal testicular proteins aberrantly expressed in cancer cells (Old & Chen, 1998). Although initially thought to be testicular-specific, further sensitive studies involving real time polymerase chain reaction have found ‘leaky’ expression of many of these antigens in some normal tissues, albeit at a much lower level when compared with that in normal testis and in tumour cells. CT antigens have been divided into those that are encoded on the X chromosome (CT-X antigens) and those that are not (non-X CT antigens) (Simpson et al, 2005). Whereas approximately 10% of the genes on the X-chromosome have been shown to belong to the CT-X families, the genes for the non-X CTs are believed to be distributed throughout the genome and, furthermore, do not usually form gene families or reside within genomic repeats (Simpson et al, 2005). CT antigens seem also to be expressed at different stages of germ-cell differentiation; CT-X genes tend to be expressed in the spermatogonia (which are proliferating germ cells) whereas non-X CT genes are generally expressed in later stages of germ cell differentiation, such as in spermatocytes (Tureci et al, 1998a; Jungbluth et al, 2000; Grizzi et al, 2003; Tapparel et al, 2003; Xu et al, 2004; Simpson et al, 2005).
The earlier CT antigens, such as MAGE 1 (Van Pel et al, 1995) BAGE and GAGE 1 (Boel et al, 1995; De Backer et al, 1999), were discovered through autologous typing and the application of the DNA-cloning methodology for defining the targets of T-cell recognition in patients with solid malignancies, especially patients with melanoma. This approach, however, requires the establishment of stable T-cell lines/clones. Using more universally applicable approaches, such as serology screening of expression cDNA library (SEREX) (Old & Chen, 1998) and bio-informatics, there are increasingly more CT antigens being identified.
Because of their wide distribution in a number of tumours and their primary expression in normal testis, CT antigens represent excellent targets for immunotherapy. The testis is considered an immune privileged site for two reasons: the apparent lack of human leucocyte antigen (HLA) class I expression on the surface of germ cells (Fiszer & Kurpisz, 1998) and most notably the presence of the blood–testis barrier that inhibits contact between differentiating germ-line and immune cells (Westbrook et al, 2004). As a result, thymic negative selection for CT reactive T cells is less likely to occur, ensuring the preservation within the immune repertoire of high affinity and frequency of these CT-reactive precursor T cells that may be necessary for successful tumour immunotherapy. However, many of the CT antigens identified so far have been expressed in spermatogonia or spermacytes located in the non-protected side of the blood–testis barrier, the basal compartment of the seminiferous epithelium (Westbrook et al, 2004). Therefore, tumour immunotherapy targeting CT antigens could potentially induce autoimmune orchitis and sterility, although this is probably only a theoretical problem as the fertility of most cancer patients are already compromised following chemotherapy. This risk will be indeed lowered by targeting specific CT antigens that are expressed only on the protected side of the barrier (Tung & Teuscher, 1995; Filippini et al, 2001).
Despite the growing interest in CT antigens, their biological functions in both the germline and tumours have remained poorly understood (Scanlan et al, 2002). A central question is whether their expression contributes to tumorigenesis or whether they are a functionally irrelevant byproduct of the process of cellular transformation, possibly due to global chromatin changes. Analysis of the MAGE family of CT antigens suggests that they could have a fundamental role in human tumorigenesis. More than 25 genes, which are characterised by the presence of a large central region, termed the MAGE homology domain (MHD), form the extensive MAGE family of CT antigens (Chomez et al, 2001). Moreover, multi-cellular organisms, such as Drosophila melanogaster and Aspergillus nidulans, carry the MHD although this domain is absent in Caenorhabditis elegans and in unicellular organisms (Simpson et al, 2005). Even though the MHD does not seem to contain any important regions of homology with other known proteins, this domain is believed to be a significant site of protein–protein interaction, as analysis of a number of type II MAGE proteins (non-CT) suggested (Simpson et al, 2005; Taniura et al, 2005). Therefore, it is possible that their products influence a range of cellular processes, including signalling, transcription, translation and chromosomal recombination.
CT antigen expression in haematological malignancies
Although a long list of CT antigens has been reported to be expressed in various tumour cells, these studies have primarily concentrated on solid tumours. However, in the last few years increasingly more CT antigens have been identified in haematological malignancies (Lim et al, 2001; Wang et al, 2003, 2004a; Zhang et al, 2003). These antigens include Sperm protein 17/CT22 (Lim et al, 2001), MAGE-1 (Van Baren et al, 1999), NY-ESO-1/CT6 (Dhodapkar et al, 2003), SEMG 1 (Zhang et al, 2003), SCP 1 (Lim et al, 1999), MAGE-C1/CT7 (Jungbluth et al, 2005), SSX/CT5 (Tarte et al, 2002), SPAN-Xb (Wang et al, 2003), PASD1 (Guinn et al, 2005) and HAGE (Adams et al, 2002). (Table I).
Probably a reflection of the poor prognosis of the disease and the increased efforts to identify therapeutic approaches outside the realm of standard chemotherapy, most of these studies have been concentrated in multiple myeloma (MM). Overall, these CT antigens are generally expressed in higher frequencies in MM as the malignant disease progresses (Van Baren et al, 1999) and their expressions have been correlated to poor prognosis (Van Rhee et al, 2005). Expression patterns of these antigens are also heterogenous, even within individual tumour specimens. The expression occurs not only at the mRNA level but also the protein level. Results from various studies have indicated that, by far, MAGE-A and MAGE-C1 are expressed at the highest frequencies (80–100%) in MM, especially those with stage III disease (Jungbluth et al, 2005). NY-ESO-1 is expressed in up to 60% of patients (Van Rhee et al, 2005) while the expression frequencies of the other CT antigens, such as BAGE (Van Baren et al, 1999), GAGE 1 (Van Baren et al, 1999), SCP 1 (Lim et al, 1999), Sp17 (Lim et al, 2001), SLLP1 (Wang et al, 2004a), SPAN-Xb (Wang et al, 2003) and SEMG 1 (Zhang et al, 2003) are lower (typically 10–40%).
Cancer-testis antigens have also been investigated in myeloid malignancies. HAGE (Adams et al, 2002), SPAN-Xb (Wang et al, 2003) and SEMG 1 (Zhang et al, 2003) have been found in up to 60% of patients with chronic myeloid leukaemia (CML). Unlike CT antigen expression in MM, HAGE expression in CML did not appear to be significantly different as the disease progressed (Adams et al, 2002). Other CT antigens are expressed at lower frequencies and they include PRAME (Paydas et al, 2006) and SLLP 1 (Wang et al, 2004a). CT antigens have also been detected in acute myeloid leukaemia. These include SPAN-Xb (Wang et al, 2003), SLLP1 (Wang et al, 2004a), PRAME (Paydas et al, 2005) and PASD1 (Guinn et al, 2005).
Cancer-testis antigens are not only expressed by tumour cells in haematological malignancies, they also frequently induce B cell immunity in these tumour-bearing patients. High titre immunoglobulin (Ig)G antibodies, implying the cognitive involvement of specific CD4 T cells, against SPAN-Xb (Wang et al, 2003), SLLP 1 (Wang et al, 2004a), PASD1 (Guinn et al, 2005), SEMG 1 (Zhang et al, 2003) and NY-ESO-1 (Van Rhee et al, 2005) have all been demonstrated in patients with haematological malignancies expressing the respective CT antigens. These results, therefore, suggest the in vivo immunogenicity of these antigens in the autologous host although the reason why tumour immunosurveillance fails and tumour develops remains speculative. It is possible that, in health or when a patient has early stage disease, tumour immune surveillance establishes an in vivo effector:target ratio that prevents expansion of the tumour load. This intricate balance between immune surveillance and tumour cells, however, is offset during a brief period of immunosuppression, such as that induced by a viral infection or medication. Once this balance is tilted in favour of tumour development, the tumour cells start to proliferate, leading to a decrease in the in vivo effector:target ratio to allow tumour escape from immune surveillance and subsequent disease presentation or progression.
Cancer-testis antigens are also immunogenic to cytotoxic T lymphocytes (CTLs). CT antigen-specific CTLs directed at Sp17 (Chiriva-Internati et al, 2002a) and NY-ESO-1 (Van Rhee et al, 2005) have been successfully generated from the peripheral blood mononuclear cells from patients with haematological malignancies. More importantly, these CTLs were able to kill autologous tumour cells expressing the respective CT antigens. These results, therefore, provide the basis for offering CT antigen-based tumour vaccines to the patients. Morover, CTLs directed at Sp17 have also been successfully generated from the peripheral blood of healthy donors (Chiriva-Internati et al, 2001, 2002b). These CTLs were able to kill HLA-matched allogeneic myeloma cells. Therefore, it may be possible to enhance an anti-myeloma effect without inducing graft-versus-host disease following allogeneic stem cell transplant using adoptive transfer of these CTLs generated from the donors.
The immunogenicity of these CT antigens in cancer-bearing patients is an important facet supporting the suitability of these antigens for immunotherapy of patients with haematological malignancies. Sp17 is so far the only CT antigen that has been applied to clinical immunotherapy in haematological malignancies. Sp17-pulsed dendritic cells were administered to a patient with relapsed MM after an allogeneic stem cell transplant (Dadabayev et al, 2005). The in vivo immunity of Sp17 was demonstrated by the detection of IgG antibodies against Sp17 in this patient following the immunisation. Although the patient responded to the immunisation with more than 90% reduction in the serum paraprotein level, the patient developed a graft-versus-host disease, probably induced by the low dose interleukin (IL)-2 given to the patient after the immunisation. It was, therefore, difficult to assess if the tumour response observed in this patient was a result of the specific anti-Sp17 immunity, the more global effect of graft-versus-host disease, or a combination of the two. The study, nevertheless, confirmed the immunogenicity of the protein in cancer-bearing patients.
Regulation of CT antigen expression in haematological malignancies
One of the potential major obstacles to the development of successful tumour vaccine is the heterogeneity of expression of the targeted tumour antigens, even within individual tumour specimens, as has been reported for various CT antigens in haematological malignancies. Furthermore, unless these CT antigens are vital for tumorigenesis, specific CT antigen targeting of tumours with heterogenous CT antigen expression will be expected to result in tumour escape by antigen-negative variant tumour cells. Therefore, much interest has been concentrated on studying the mechanisms regulating CT gene expression in order that these mechanisms could be exploited to induce/upregulate antigen expression in tumour cells to increase the applicability and efficacy of a potential tumour vaccine.
Through studies involving the MAGE family of CT antigens in solid tumours, it appears that DNA demethylation may play a key element in the regulation of CT gene expression. It is now well established that molecular events such as global DNA hypomethylation, gene-specific hypomethylation and regional hypermethylation occur during tumorigenesis (Simpson et al, 2005). Hypermethylation of tumour-suppressor genes in tumours is often associated with different hypomethylation mechanisms, such as repetitive DNA hypomethylation, global genome hypomethylation and hypomethylation of the promoter genes (Kaneda et al, 2004). It is, however, noteworthy that in some tumour cells, e.g. colon cancer cells, although DNA is universally hypomethylated, the expression of CT antigens is not common, suggesting that the induction of CT-X gene expression is certainly not dependent solely on promoter demethylation (Goelz et al, 1985). However, the fact that all CT-X genes studied to date have methylated CpG islands in normal somatic tissues underlines the importance of DNA demethylation (De Smet et al, 1996, 1999; Coral et al, 2002; Zhang et al, 2007). Due to both its effect on chromatin structure and to transcription-factor binding, methylation of CpG islands within gene promoters is responsible for gene silencing (Baylin & Herman, 2000; Simpson et al, 2005). The importance of DNA demethylation could also be highlighted by the fact that every CT-X gene studied so far has been activated during spermatogenesis, one of the two only phases in the human life circle when ‘epigenetic reprogramming’, which involves alterations in DNA methylation and chromatin restructuring, occurs (Kimmins & Sassone-Corsi, 2005; Simpson et al, 2005). Finally, several studies provided experimental evidence that demethylation of CT-X gene promoters can induce antigen expression in cells that do not normally produce them (Weber et al, 1994; De Smet et al, 1996, 1999; Coral et al, 2002; Gure et al, 2002; Simpson et al, 2005).
In haematological malignancies, the expression of Sp17 (Wang et al, 2004b), SPAN-Xb (Wang et al, 2006a) and SEMG 1 (Zhang et al, 2007) have all been shown to correlate with promoter demethylation, suggesting the primary role of DNA methylation in the regulation of expression of these genes. Treatment of myeloma cell lines and fresh myeloma cells that did not express these antigens with the hypomethylating agent, 5-azacytidine, resulted in the hypomethylation of specific CpG dinucleotides within the promoters of these genes and subsequent gene and protein expression. These results suggest the opportunity to treat patients with the hypomethylating agents to enhance antigen expression prior to tumour vaccine to enhance the efficacy of the vaccine. Obviously, further work is needed to determine whether there is any differential sensitivity of the hypomethylating agents on tumour cells compared with normal cells to avoid increasing any non-specific toxicity of the vaccines. This approach would be particularly attractive if the vaccine could be applied to myelodysplastic syndrome, so that not only is the cytotoxicity by specific CTLs enhanced, the underlying disease could also benefit directly from the hypomethylating agents.
Methyl-CpG-binding proteins (MBDs) are an essential part of the epigenetic machinery. They constitute a link between DNA methylation and histone modification in processes leading to stable repression of gene expression (Fuks et al, 2003). On the one hand, they bind both to methylated DNA and to DNA methyltransferases; on the other hand, they also bind directly to histone deacetylases (Ng et al, 1999). Hence, MBDs can be viewed as a molecular bridge between two key epigenetic events. MeCP2 is one of the best characterised MBD, which can bind as little as a single methylated CG nucleotide, although higher number of methylated residues may increase the binding (Lewis et al, 1992); it associates with the Sin3A histone deacetylase complex and recruits the HDAC1 and HDAC2 to the promoter chromatin (Wade, 2001). Histone deacetylation suppresses gene expression by removing the acetyl groups from histones, which leads to compact and transcriptionally inactive chromatin (Jones et al, 1998; Ng & Bird, 2000). Using antibodies directed at the MeCP2 protein in chromatin immunoprecipitation, it has also been demonstrated that MeCP2 binding to the promoter sequence of SPAN-Xb (Wang et al, 2006b) and SEMG 1 (Zhang et al, 2007) was associated with repression of these genes. Treatment of tumour cells with 5-azacytidine resulted in the dissociation of the MeCP2 from the promoters and subsequent expression of the SPAN-Xb and SEMG 1 genes and proteins. These results further implicate the role of DNA methylation, in conjunction with the MeCP2 protein, in the regulation of expression of some CT genes. This conclusion, however, needs to be confirmed in MeCP2 knock-out experiments. If verified, it may imply that these CT antigens may also be upregulated by histone deacetylase inhibitors.
Obviously, it remains to be determined whether or not upregulation of expression of these antigens in tumour cells results in improved cytotoxicity of the tumour cells by specific CTLs and whether or not there is a differential dose-response among the different normal tissues in their sensitivity to the hypomethylating effect of these agents. However, as the susceptibility of a target cell expressing an antigen to CTL-mediated lysis depends not only on the gene being expressed and translated but also on factors such as whether or not the protein is processed and presented appropriately in the context of major histocompatibility complex (MHC) molecules, and whether the number of copies of the MHC-peptide complex present on the surface of the target cells is sufficient to mediate efficient effector-target cell interaction, the mere upregulation of the protein on some normal tissues may not necessarily predict for tissue injury by the cytotoxic T-lymphocytes.
Secondary regulatory mechanisms of CT gene expression
In some tumour cells, hypomethylation alone is not sufficient for the induction of CT gene expression, e.g. colon cancer cells, for example, are universally hypomethylated in their genome (Goelz et al, 1985), even though CT antigen gene expression is rare in this tumour type. Such observation suggests the presence of secondary regulatory mechanisms controlling the expression of CT genes. Studies in SPAN-Xb (Wang et al, 2006a) and SEMG 1 (Zhang et al, 2007) suggest that specific cytokines play a secondary regulatory role in the control of these CT genes, at least in MM cells. Treatment of myeloma cells with IL-7 and granulocyte-macrophage colony-stimulating factor upregulates SPAN-Xb expression, but only in the presence of a demethylated promoter gene. In the absence of a demethylated SPAN-Xb promoter, these cytokines were unable to affect the antigen expression in these tumour cells. A similar scenario occurs for SEMG 1 expression in myeloma cells. The CT antigen could be upregulated by IL-4 and IL-6 but only if the myeloma cells were already expressing the SEMG 1 gene (SEMG1). IL-4 and IL-6 were ineffective in myeloma cells that did not expression SEMG1 but became effective only after the myeloma cells had been treated with 5-azacytidine. Obviously, it is possible that other cytokines may be operative in other tumour types since the presence of specific cytokine receptors on the tumour cells is likely a prerequisite for the cytokines to mediate any intracellular molecular events needed for the upregulation of these CT antigens.
One reason for the lack of successes in clinical tumour immunotherapy is the heterogeneity of tumour antigen expression within individual tumour specimens. Therefore, successful targeting of one specific tumour antigen may be followed by the emergence of antigen-negative variant tumour cells. Furthermore, clinical successes may be hampered by the relatively high tumour burden, when compared with the antigen-specific effector cells, generated by the tumour vaccine in vivo. Future work will probably focus on addressing these two problems by identification of other novel CT antigens that could be used in polyvalent vaccines to produce high effector:target ratios of multiple specificity in vivo to reduce the chance for tumour escape by the emergence of antigen-negative variant tumour cells. This may be achieved through using the yeast two-hybrid system to identify the ligands for these CT antigens, as most intracellular proteins are expressed in conjunction with its interacting proteins. The yeast two-hybrid system has been used successfully to identify the binding partners for MAGE-A1 and MAGE-A4. If a protein is testicular-specific, it is possible that its binding partner protein may also be testicular-specific. Similarly, if a CT antigen has distinct function within the tumour cells that express the antigen, then it is also likely that its partner protein will also be expressed within the tumour cells. Therefore, identification of the binding partner proteins of known CT antigens, using the yeast two-hybrid system with the known CT antigens being the bait in the system, may provide another way to identify other novel CT antigens. Work is presently ongoing in our laboratory to exploit this approach for the identification of other novel CT antigens. In addition, to improve antigen expression and the efficacy of a tumour vaccine, more efforts will be focused on identifying cytokines, hypomethylating agents and histone deacetylase inhibitors that could be used differentially to induce antigen expression in the tumour cells without affecting normal cells to reduce the risk for non-specific toxicity of the vaccines.
Vaccine-based therapeutic interventions are ideally designed to suit the unique characteristics of haematological malignancies. The ease of tumour accessibility, the ability of current treatments to achieve a minimal residual state that could be accurately monitored, and the ability of myeloid cells to differentiate in vitro to functional dendritic cells combine to facilitate any intent to generate successful immonotherapeutic strategies (Borrello & Sotomayor, 2002). Therefore, provided a suitable tumour target/s is/are chosen, immunotherapy for haematological malignancies should provide the lead for further development of tumour vaccines.
This study was supported by grants from the National Institute of Health/National Cancer Institute (RO1 CA088434 and RO1 CA116283).