Overexpressed oncogenic tumor-self antigens.

Overexpressed tumor-self antigens represent the largest group of candidate vaccine targets. Those exhibiting a role in oncogenesis may be some of the least studied but perhaps most promising. This review considers this subset of self antigens by highlighting vaccine efforts for some of the better known members and focusing on TPD52, a new promising vaccine target. We shed light on the importance of both preclinical and clinical vaccine studies demonstrating that tolerance and autoimmunity (presumed to preclude this class of antigens from vaccine development) can be overcome and do not present the obstacle that might have been expected. The potential of this class of antigens for broad application is considered, possibly in the context of low tumor burden or adjuvant therapy, as is the need to understand mechanisms of tolerance that are relatively understudied.

tumor protein D52

murine TPD52

human TPD52

tumor associated antigen

T regulatory cell

oligodeoxynucleotide

Transgenic adenocarcinoma of the mouse prostate

cytotoxic T lymphocyte-associated antigen 4

Wilms tumor-1

vascular endothelial growth factor receptor 2

Anaplastic lymphoma kinase

human epithelial growth factor receptor 2

androgen receptor

cytotoxic T lymphocyte

human leukocyte antigen

Introduction

Current statistics reveal an increase in cancer incidence with very little decrease in mortality. It is estimated that about 1 660 290 new cancer cases were diagnosed in the USA in 2013, with 580 350 (35%) deaths. Improvement of treatments to decrease mortality may be met through immune-based therapies. The employment of the immune system to treat malignant tumors, commonly referred to as tumor immunotherapy, encompasses two general categories: passive and active. Passive immunotherapy largely involves the administration of specific antibodies, cytokines, or cells (most commonly tumor antigen-specific T cells generated ex vivo, known as adoptive cell therapy). Passive administration of T cells or of monoclonal antibodies against the T cells inhibitory receptor CTLA-4, which is referred to as immune checkpoint blockade, has recently gained international acclaim as the breakthrough of the year (2013), though not without immune related adverse effects. Active immunotherapy can be most easily defined by vaccination. While passive immunotherapies often engage the immune system independent of the knowledge of defined tumor antigens (an exception being some forms of adoptive cell therapy), vaccines elicit specific immune responses by targeting defined tumor antigens.

With the exception of vaccines to prevent cervical cancer by targeting select serotypes of human papillomaviruses, most if not all vaccine clinical trials have focused on therapy. In this approach the vaccine is administered when tumors are present in the patient, with the primary goal of shrinking the tumor mass. However, the majority of clinical trials to treat established solid tumors by vaccination have yielded disappointing results. This lack of success is underscored by the existence of only one FDA approved cell-based therapeutic vaccine Sipuleucel T (Provenge®), approved in 2010. However this is not to say that cancer vaccination research is without merit. On the contrary, much work is still being done to advance and improve cancer vaccines.

The idea of preventive cancer vaccination has recently gained new attention, and may represent the next significant advance in the field of immune-based cancer treatments. The logic behind preventive vaccination is sound and supported by centuries of success against infectious agents. In this case, why hasn't vaccination been applied more broadly to cancer prevention? It is reasonable to speculate that cancer vaccines have been applied as an option for compassionate administration as a last effort for patients who have failed conventional therapy. This is essentially the approved application for Sipuleucel T, which is not curative but extends life for some late stage prostate cancer patients. Advances in early cancer detection have opened the door to develop vaccines to prevent primary and recurrent tumors rather than shrink existing tumor masses.

Whether the vaccine is therapeutic or preventive, it is clear that future success will depend on the character of the tumor antigen targeted by the vaccine. The current collective of tumor antigens ranges from specific to associated, non-self to self, and comprises hundreds of vaccine candidates (far too many to discuss in a single review article). In this light, an emerging group of cancer vaccine target antigens, defined primarily by their overexpression in tumor cells compared with low but detectable levels in normal cells, and a role in oncogenesis, represents the focus of this review. Other related topics such as comparisons of vaccine strategies, or the use of adjuvants from different clinical studies, have been recently reviewed elsewhere.

Tumor Antigen Classifications

Tumor antigens recognized by the immune system have been most often defined based on the nature of the antigen (protein, mucin) and its expression pattern, a practice that gave rise to numerous categories and complexities. Examples include, viral proteins specifically associated with virus-induced malignancies (approximately 12% of all cancers), and oncogene and tumor suppressor gene products or their mutant variants. Oncofetal antigens and melanoma associated antigens represent those with restricted expression in non-malignant tissues, whereas cancer testis antigens are only expressed, as the name suggests, in the tumor and in testes, with testes being protected by mechanisms of immune privilege. Overexpressed antigens are a large and diverse group that includes any protein found at increased levels in tumors compared with normal healthy cells and tissues.

Recently, Coulie and colleagues classified tumor antigens as either high tumor specificity or low tumor specificity, each with two sub-categories; mutation (most tumors) or tumor-specific expression (many tumors) being associated with high tumor specificity, and tissue-specific expression (melanomas) or overexpression (some tumors) being associated with low tumor specifity. The overall premise of this classification is whether or not the immune system (T cells in particular) recognizes the tumor antigen.

illustrates a simplified classification based on the immunologic character of the tumor antigen relative to its potential immunogenicity. Arguably tumor-specific antigens represent the obvious choice for vaccine targets, with no or limited expression in normal cells, inferring the potential for a stronger immune response without tolerance. Some examples of tumor-specific antigens include viral antigens (non-self), cancer testis antigens, and melanoma associated antigens (altered-self, with respect to limited expression in normal cells). Antigens from each of these groups are in the advanced stages of vaccine development. Tumor associated antigens (variable expression in normal cells) may represent a more obscure class of antigens, but ironically also represent the largest class of candidate vaccine target antigens. It is presumed that tumor associated antigens will induce a weaker immune response insufficient for tumor rejection, which may be why many have been understudied as vaccine targets. A subclass of overexpressed tumor associated antigens involved in the generation of oncogenesis and critical for maintaining the oncogenic phenotype may represent some of the most promising, widely applicable vaccine targets.

Figure 1.

Immunologic character of tumor antigens. Simplified tumor antigen classification based on the immunologic character of the tumor antigen relative to its potential immunogenicity. Depicted are two main classes of antigens represented by tumor-specific and -associated antigens defined by normal cell expression, and four sub-classes ranging from stronger to weaker immunity. Non-self (e.g., viral proteins) and altered-self (e.g., mutant protein or restricted expression) antigens are proposed to elicit strong immune responses, and self antigens, whether involved in oncogenesis (tumor-dependent) or not (tumor-independent), would elicit weaker immune responses.

Overexpressed Oncogenic Tumor-Self Antigens

A National Cancer Institute sponsored project to prioritize cancer vaccine target antigens for translational research revealed that aberrantly expressed self-proteins represent the largest number of candidate antigens for vaccine development (nearly half of those categorized). The study used preweighted objective criteria for prioritizing candidate tumor vaccine target antigens and selected 75 antigens for comparison and ranking (from hundreds considered). Nine criteria were used to rank the antigens. Therapeutic function was considered as the most important followed by immunogenicity, specificity, and oncogenicity. The relative weights of the remaining five criteria were orders of magnitude less than those applied to the top four. Cellular localization of expression (internal or surface) carried the least weight of importance, but circulating antigen was determined to be not preferable. Therapeutic function data resulting from clinical trials were heavily considered in selecting the 75 antigens that were ranked. Because the goal of the study was to accelerate translational (clinical) research, this bias was logical and acknowledged. However, some newer and potentially promising antigens were not ranked, largely because relevant studies were still in the pre-clinical phase. To summarize the conclusions, an ideal vaccine target antigen would be immunogenic, eliciting a response that eliminates tumor cells leaving normal cells unharmed (immunogenicity, therapeutic function, and specificity) and play a role in inducing or maintaining the oncogenic phenotype making the antigen indispensable to the tumor.

When considering the specificity criterion and focusing only on those antigens exhibiting aberrant expression (varying degrees of normal cell expression) 87% (65/75) of the ranked antigens can be classified as having some normal cell expression. If post-translational modifications (e.g., the mucin, MUC1), and tissue specific (e.g., gp100), mixed (e.g., ALK), stromal (e.g., VEGFR2), and oncofetal (e.g., WT1) expression are excluded from the aberrant expression class, only overexpressed antigens remain, representing nearly 40% of the ranked antigens (28/75). If overexpressed antigens that do not play a role in oncogenesis are excluded from the list of 75, there are 9 overexpressed tumor-self antigens that are immunogenic and oncogenic (two criteria considered to be heavily weighted and important for an ideal tumor vaccine antigen). In the following sections, we will briefly highlight several of the ranked overexpressed oncogenic tumor-self antigens, and end by focusing on perhaps the newest, widely shared overexpressed oncogenic tumor-self antigen, tumor protein D52 (TPD52).

Limited tumor expression

Androgen Receptor

The androgen receptor (AR) is a steroid receptor involved in prostate development, and a target for hormone deprivation therapy for advanced metastatic prostate cancer. Its function ascribes the AR with a role in prostate oncogenesis. AR is expressed in normal tissues with the prostate being a major site, and is widely overexpressed in prostate cancer, and in a subset of breast cancers. Pre-existing AR-specific antibodies and T cells have been detected in prostate cancer patients, demonstrating immunogenicity and the potential for targeting the AR by vaccination. The fine specificity of the AR-specific CTL responses in prostate cancer patients was defined by recognition of two HLA-A2-restricted peptides, AR805 and AR811, with the latter capable of eliciting specific CTLs following immunization of A2/DR1 transgenic mice. An AR-based DNA vaccine targeting the AR ligand-binding domain (LBD) effectively induced CTLs responses in HHDII-DR1 (HLA-A2+ and HLA-DR1+) transgenic mice which were capable of lysing HLA-A2+ human prostate cancer cells. In vivo prostate tumor regression was also observed in rats immunized with the AR-LBD DNA vaccine, supporting the potential for targeting the AR with vaccination. These studies demonstrate that AR is immunogenic in patients and AR-based vaccines are capable of in vivo tumor rejection in pre-clinical animal models.

Shared tumor expression

We defined shared tumor expression as overexpression in multiple different cancers, but not as widely shared or potentially universal in tumor overexpression as that reported for human telomerase reverse transcriptase (hTERT) and survivin (discussed in the following section). The following are three examples of shared overexpressed oncogenic tumor-self antigens that are currently being evaluated in clinical trials as vaccine targets. Her-2/neu and p53 are perhaps the most commonly recognized and extensively studied TAAs of this category.

Her-2/neu

The transmembrane tyrosine kinase Her-2/neu is one of the most studied cancer proteins and therapeutic targets (nearly 1000 reviews). Her-2/neu is overexpressed in multiple cancers including lung, prostate and most notably breast cancer. To date the majority of clinical vaccine trials have been conducted in patients with breast cancer. Early evidence supporting Her-2/neu as a vaccine target came from the demonstration that specific CTLs exist in the peripheral blood of breast cancer patients. Clinical trials evaluating vaccines targeting the intracellular and extracellular domains of Her-2/neu demonstrated the generation of specific T cells without the induction of autoimmunity and no significant toxicity, supporting the use of vaccines in the adjuvant setting.

p53

The important role played by p53 in cancers of multiple types has been recognized for decades. CTLs have been generated against mutated and non-mutated epitopes of p53 without inducing autoimmunity and pre-existing p53-specific CTLs have been demonstrated in cancer patients. Clinical trials are underway evaluating therapeutic vaccines targeting p53 in ovarian cancer, colorectal cancer, and non-small cell lung cancer. Overall, p53-specific immune responses have been observed in patients, but significant reductions in tumor burdens have not been demonstrated. It has been suggested that multiple epitope vaccines, Treg elimination, and/or CTLA-4 blockade should be assessed in combination with p53 vaccines to increase their clinical efficacy.

EphA2

EphA2 is a cell surface-expressed receptor tyrosine kinase and is one of 14 members of the Eph family (named for erythropoietin-producing hepatocellular carcinoma cell lines). This family and its ligands play key roles in normal development and in tumorigenesis, where EphA2 is overexpressed in multiple cancers including brain malignancies, which have been the focus of most vaccine trials. The vaccine potential of targeting EphA2 in gliomas was initially demonstrated by generating HLA-A2-restricted CTLs from the peripheral blood of HLA-A2+ normal donors and glioma patients using a single synthetic peptide (TLADFDPRV), and by vaccination of HHD mice with the same EphA2 peptide, demonstrating that tolerance could be broken by vaccination without inducing autoimmunity. Subsequent studies in murine models of colon, liver and brain cancer further support the potential of EphA2 as a vaccine target. Initial clinical trials testing EphA2 vaccines against adult and pediatric brain cancers have shown them to be safe, immunogenic and promising.

Widely shared (universal) tumor expression

There are but a few tumor associated antigens that appear to be universal for most if not all cancers, and these have shown promise as candidates for vaccine development. The two most studied universal tumor associated antigens are hTERT and the anti-apoptotic protein survivin. Neither of these antigens are specific for tumor cells, but both are over-active and/or aberrantly expressed in tumor cells compared with normal or non-malignant cells, and play a role in prolonging survival of tumor cells by preventing natural death mechanisms associated with proliferation such as telomere shortening and apoptosis. These attributes classify hTERT and survivin as shared overexpressed oncogenic tumor-self antigens.

hTERT

The activity of hTERT is a rate limiting step in the proliferative capacity of advanced cancers and represents a prototypical and universal cancer antigen and marker. Variations in hTERT function in normal self-renewing tissues and tumors exist that can be taken advantage of for vaccine development. hTERT was first characterized as a widely shared tumor associated antigen by demonstrating the induction of specific CTLs against more than 85% of human cancers, and by the identification of an HLA-A2 peptide vaccine candidate, 1540 (ILAKFLHWL). Circulating 1540-specific CTLs were detected in nearly 91% of HLA-A2+ patients with 6 different cancers supporting vaccine development. Additional CTL-specific hTERT-peptides include peptides restricted by additional HLAs. A 16 amino acid hTERT peptide (GV1001) comprised of HLA class II and class I epitopes capable of inducing both CD4+ and CD8+ T cells when administered is also under investigation as a vaccine. Several clinical trials assessing the efficacy of hTERT vaccines in patients with multiple cancers have been conducted or are underway, and although variations were reported in terms of tumor reductions, specific T cells were commonly induced in the majority of patients, with no adverse toxicities.

Survivin

Survivin is an inhibitor of apoptosis and acts upstream of the main effector proteases of apoptosis, caspase 3 and 7, and is active in a cell-cycle regulated manner during the G2/S phase, as a safeguard against activated cell death during successive rounds of cell division. Like hTERT, survivin is expressed at low levels in normal, differentiated adult tissues but is overexpressed in cancers originating from a variety of tissues including lung, colon, breast, pancreas, and prostate cancer and several hematopoietic cancers. The immunogenic potential of survivin was initially demonstrated by the induction of specific HLA-A2-restricted CD8+ cytotoxic T cell responses by dendritic cells that had processed and presented recombinant survivin protein. CTLs specific for survivin were also detected in the peripheral blood of cancer patients supporting its potential as a vaccine target. A survivin peptide vaccine administered with IFNα demonstrated efficacy and benefit for patients with advanced pancreatic cancer, although contrasting clinical benefits were observed for similar vaccine approaches in in advanced melanoma patients.

Tumor protein D52

A third example of a new overexpressed oncogenic tumor self antigen with widely shared tumor expression is tumor protein D52 (TPD52) and like p53, survivin and hTERT is an intracellular protein. Early clinical support for the promise of TPD52 as a vaccine target antigen came from the identification of circulating serum antibodies with specificity for TPD52 in patients with breast cancer.

Expression of TPD52 in cancer

TPD52 is an overexpressed tumor self-protein actively involved in transformation, leading to increased proliferation and metastasis. TPD52 overexpression has been demonstrated in several human malignancies including breast, prostate, and ovarian carcinomas. Expression microarray and other analyses predict TPD52 overexpression in many other cancers including multiple myeloma, Burkitt's lymphoma, pancreatic cancer, testicular germ cell tumors, and melanomas, as well as multiple other adult and pediatric cancers.

Murine ortholog of TPD52

The murine ortholog of TPD52 (mD52) parallels normal tissue expression patterns and known functions of human TPD52 (hD52), with 86% amino acid identity. mD52 induced anchorage independent growth and spontaneous lung metastasis when overexpressed in normal, non-tumorigenic cells. Reduction of hD52 expression via RNAi resulted in increased apoptosis in human breast cancer cells, and hD52 overexpression was associated with decreased overall survival in human breast cancer patients. These studies demonstrate that TPD52 overexpression is important for initiating and maintaining an oncogenic and metastatic phenotype and may be important for tumor cell survival. TPD52 is naturally expressed and involved in tumor formation and metastasis in human cells (hD52) and in mouse cells (mD52). This makes mD52 a unique and powerful overexpressed tumor-self antigen for study as a cancer vaccine target in murine models of cancer.

TPD52 as a vaccine target

The first demonstration that tumor protective immunity could be induced against TPD52 involved a recombinant protein-based mD52 vaccine that induced protection against tumor challenge when administered with CpG-ODN as a molecular adjuvant. mD52 protein administered without CpG-ODN failed to elicit an immune response, indicating that the TLR agonist was necessary to break tolerance. Subcutaneous injection of mD52 protein with CpG-ODN required concomitant CD4+CD25+ T regulatory (Treg) cell depletion to improve tumor protection. DNA-based vaccine approaches using the TRAMP model of prostate cancer demonstrated that mD52 DNA vaccination induced an immune response that prevented tumors with increased efficacy when administered with GM-CSF and induced long-term immunologic memory. When mD52 DNA vaccination was compared head-to-head with hD52 DNA vaccination, the partial xeno-antigen (hD52) was more effective at protecting against tumor challenge, however both strategies induced durable responses that rejected secondary tumor challenge months later. The T cell cytokine secretion patterns for all the TPD52 vaccine studies demonstrated that a TH-1-type cellular immune response was responsible for tumor rejection and that a complete response may be hindered by a potentially unique subset of CD8+ IL-10+ regulatory T cells. An overlapping peptide-based mD52 vaccine, evaluated independently, demonstrated efficacy in a murine breast cancer model. Important facts have been revealed by preclinical TPD52 vaccine studies to date (summarized in ). First, the successful use of the basic vaccine formulation demonstrated that a tumor self-protein can be immunogenic when delivered as a simple protein, peptides or plasmid DNA. Second, TPD52 vaccines prevent tumor formation without inducing autoimmunity, even when classical CD4+ CD25+ Treg cells were depleted. These studies suggest that TPD52-specific T cells are present and not completely eliminated by central tolerance, and that peripheral tolerance is involved in obstructing complete tumor rejection to include suppression by an as yet undefined but potentially unique subset of CD8+ Treg cells. An additional note-worthy observation from our preclinical vaccine studies is that DNA-based vaccines (most notably xenogeneic hD52 DNA) appear to be more potent and effective suggesting that TLR-9 plays a role as a molecular adjuvant. This is supported by the requirement for the inclusion of CpG ODN with recombinant protein to induce protective immunity.

Table 1.

TPD52 vaccines in murine models of sarcoma and prostate cancer

Vaccine Approach	Route Administered	Primary Tumor Rejection	Prevents Recurrence	Mouse Strain/Model
mD52-DNA	i.m.	14–50% protection (av. < 30%)	YES	C57BL-6/TRAMP
hD52-DNA	i.m.	70% protection	YES	C57BL-6/TRAMP
mD52-DNA +GM-CSF	s.c.	70% protection	YES	C57BL-6/TRAMP
mD52-Protein/ODN/Alum	i.m.	No protection	NO	C57BL-6/TRAMP
mD52-Protein+ODN/IFA	s.c.	No protection	NO	C57BL-6/TRAMP
mD52-Protein+ODN/IFA, CD4 Treg depletion	s.c.	70% protection	YES	C57BL-6/TRAMP
hD52-Protein+ODN/IFA	s.c.	No protection	NO	C57BL-6/TRAMP
hD52-DNA (2)/mD52 protein + ODN/IFA (2)	i.m./ s.c.	80% protection	YES	C57BL-6/TRAMP
mD52-DNA, CD4 Treg depletion	i.m.	50%	YES	C57BL-6/TRAMP
Protein/ODN/Alum	i.m.	50%	YES	Balbc/mKSA
Protein/ODN/IFA	s.c.	No protection, possibly due to dominant role of Treg cells	NO	Balbc/mKSA
Protein/ODN/IFA, deplete Treg cells	s.c.	70% protection	YES	Balbc/mKSA

mD52, murine TPD52; hD52, human TPD52; ODN, oligodeoxynucleotide; IFA, incomplete Freund's adjuvant; Treg, T regulatory cell; GM-CSF, granulocyte monocyte colony stimulating factor; i.m., intramuscular; s.c., subcutaneous; TRAMP, transgenic adenocarcinoma of the mouse prostate; see refs. .

As a first step to human studies and eventual clinical trials, we generated CTLs specific for hD52 from the peripheral blood of an HLA-A2+ male normal donor by in vitro stimulation with a synthetic peptide (QLFHSFSV; modified at P2 and P9 to increase affinity for HLA-A2) derived from the amino acid sequence of hD52 using established protocols. These CTLs only lysed HLA-A2+ prostate cancer cell lines that naturally overexpressed hD52 (). This initial experiment further supports the potential use of TPD52 as a vaccine target in humans.

Figure 2.

Generation of CTLs from normal PBLs with D52 peptide. The data show hD52-specific HLA-A2-restricted killing of human prostate cell lines. hD52 expression in the human prostate cell lines was determined by 30 cycle RT-PCR using hD52-specific primers and GAPDH as an internal reference. (A) Human normal prostate cell line 568 NPTX and tumor cell line LnCap are hD52-low expressors. (B) Killing of human prostate cancer cell lines determined using a standard lysis assay at an E:T of 5:1. Targets: human prostate cancer and normal-derived cells from HLA-A2+ patient 568 (normal = NPTX, tumor = CP1TX and CP2TX) were generated as previously described, human prostate cancer cell lines PC-3 (HLA-A2-) and LnCap (HLA-A2+) are commercially available. Methods: The method used to generate hD52 peptide-specific CTLs by IVS with hD52 peptide Q(A/L)FSHSFS(I/V) has been previously explained in detail.

Concluding Remarks

Employing the immune system to fight cancer has long been viewed as promising, as is evidenced by the extensive list of publications attesting to the hard work of many. This promise is now being realized by FDA approval of two of the newest treatments for prostate cancer and melanoma, Sipuleucel T and CTLA-4 blockade. Both of these immunotherapies rely on eliciting T cell-mediated immunity which requires antigen recognition and specificity. Sipuleucel T is a therapeutic vaccine, and CTLA-4 (T cell checkpoint) blockade is being explored in combination with vaccination. Most vaccine trials focus on patients with advanced cancer (therapeutic vaccines) and have largely produced disappointing results. Historically vaccines have been successful only when applied to prevent disease. Perhaps immunization to prevent primary cancer, recurrence and/or metastasis as opposed to eliminating already existing primary and metastatic tumors is the next step in cancer vaccine development.

Regardless of the application, cancer vaccines are only as good as the antigens they target. In this context it is logical to argue that the antigen content of the vaccine formulation is more important than the delivery vehicle. A review of the literature will reveal that much of the past vaccine development work has focused on generating new delivery vehicles and formulations to make a few well-studied antigens more potent in murine models of cancer and clinical trials. This focus on a minority of antigens is also true for the study of oncogenes in general. The recent prioritization of vaccine antigens for translational studies supports the notion that it is time to find new, better antigens on which to focus vaccine efforts.

What constitutes an ideal cancer vaccine antigen? Many would argue, and logically so, that completely foreign (e.g., viral) non-self proteins would be the most immunogenic (Fig. 1) and therefore effective. Much of the early preclinical experimental literature and the FDA approval of preventive cervical cancer vaccines would be in agreement. Unfortunately most clinically significant cancers are not (or at least have yet) to be associated with viruses. This truth led to the TAA discovery movement and the development of various technologies for its accomplishment. This produced many candidate vaccine antigens that with further effort, could yield ideal targets for the next generation of cancer vaccines, among them being the overexpressed tumor-self antigens. This large group of understudied vaccine targets represents (as the name implies) self proteins with limited or low but detectable expression in many normal healthy cells and tissues and overexpression in multiple to most cancers. A subset of this group comprises those with the added desirable feature of being involved in oncogenesis, making their expression indispensible to the tumor. Preclinical murine models of cancer have been critical for development of this group of antigens as vaccine targets, and for demonstrating that these self proteins are immunogenic and thus capable of eliciting immunity that will kill tumors and not healthy cells (no autoimmunity). Perhaps the newest promising overexpressed oncogenic tumor-self antigen is the oncoprotein TPD52.