Tumor Antigens Recognized by T Cells
The first human tumor antigen
defined by T-cell recognition was identified by Thierry Boon and colleagues at the Ludwig Institute almost 20 years ago. A T-cell clone was generated that recognized an autologous melanoma cell line. The cDNA library of the tumor line was transfected into antigen-presenting cells (APCs) expressing the restricting allele and screened using the tumor-reactive T-cell clone. The cDNA of target cells sensitized to lysis by the antigen-specific T cell was recovered and sequenced. By this method, the first human T cell-defined tumor antigen was found to be MAGE-A1.
5,
6 It was later discovered that in addition to the MAGE-A1 antigen, a number of MAGE-like antigens, including GAGE, BAGE, LAGE, NY-ESO-1, and SSX-1, were also recognized by tumor-reactive T cells. Based on their restricted tissue expression, that is, tumor cells and germinal cells such as testis, fetal ovary, and placenta, these antigens were grouped together into the family of “cancer-testis” or CT antigens.
7 There are now known to be over 80 CT antigens, several of which are immunogenic to T cells and are expressed in a wide variety of solid and liquid tumors including lung cancer, colorectal cancer, breast cancer, ovarian cancer, and leukemia.
Using a similar approach, the melanoma-associated antigens, tyrosinase, gp100, and MART-1, were also identified as T-cell target antigens.
8,
9,
10 In this case, their association with pigmentation pathways
found in normal melanocytes represented the first examples of nonmutated “differentiation” antigens serving as immunologic targets for human T cells.
11 Since then, other “differentiation” antigens for prostate cancer (PSMA, Kallikrein), colorectal cancer (CEA), and breast cancer (NY-BR-1, mammoglobin) have been identified. Along with differentiation antigens, other nonmutated self-proteins that are found to be over-expressed in tumor cells include Her-2-neu (breast cancer), adipophilin (renal cancer), mesothelin (ovarian cancer, pancreatic cancer), and more “universally” expressed antigens that confer a survival advantage to tumor cells, such as, survivin, telomerase, and WT-1.
Mutations in genes associated with tumorigenesis, such as those responsible for cell cycling (the cyclin-dependent kinase, CDK4), delivery of mitogenic signals (B-RAF), and apoptosis (CASP-8), represent attractive targets for immunotherapy because of the decreased likelihood for antigen-loss tumor variants developing. Unfortunately, most of these mutations are not highly prevalent among most tumors and appear to exhibit low immunogenicity.
T cells recognize peptide fragments of target antigens presented by self-MHC molecules on the surface of tumor or APCs such as dendritic cells (DCs), B cells, and monocytes. Those peptide fragments recognized by T cells in the context of the MHC complex are the result of internal processing by proteasomes (class I-restricted epitopes) or lyso/phagosome-associated enzymes (class II-restricted epitopes) followed by binding to their respective MHC complexes and surface presentation. The identification of epitopes recognized by tumor-reactive T cells has been a focus of considerable research since such epitope peptides can be used to elicit antigen-specific T cells and track T-cell responses and are amenable to clinical use as readily available GMP grade reagents to sort and collect antigen-specific T cells. For class I-restricted epitopes designed to elicit CD8 T-cell responses, these epitopes are generally nine to ten amino acids in length; for class II-restricted epitopes eliciting CD4 T-cell responses, the epitopes can be 14 to over 20 amino acids long since class II alleles are less stringent and can accommodate overhanging flanking regions.
In cases where a tumor-reactive antigen-specific T-cell clone has been isolated, such a clone can be used to probe overlapping peptides to identify the minimal epitope sequence. For common alleles, the target epitope may be deduced using algorithms that predict the sequence on the basis of consensus motifs based on known binding preferences as well as the predilection of the proteasome for certain cleavage sites. On occasion, splice variants, excision of intervening sequences, and even sequence reversal during antigen processing can lead to unexpected surface presentation of the cognate epitope.
12 Often, the definitive sequence can only be deduced by eluting peptides from surface MHC and subjecting the mix to mass spectrometry.
However, when the tumor-associated epitope is nonmutant, as is frequently the case, the naturally occurring peptide ligand may not engender robust and sustained antitumor CTL responses.
13,
14 This is a result of immune tolerance mechanisms that suppress or eliminate high-avidity autoreactive T cells.
15 What remains is a low
frequency of tumor-specific T cells or T cells that bear low-avidity T-cell receptors for the cognate tumor antigen.
16,
17,
18,
19 One method to activate and mobilize these rare and low-avidity tumor-specific T cells uses superagonist altered peptide ligands (APL)
20,
21 that deviate from the native peptide sequence by one or more amino acids, to allow for enhanced binding to the restricting MHC molecule
20,
21 or favorable interaction with the T-cell receptor (TCR) of a given tumor-specific T-cell subset. While, superagonist APLs have been identified that generate tumor-reactive T cells and have even been used to elicit desired immune responses in clinical studies,
22,
23 a comprehensive method for identifying superagonist ligands remains to be developed. Furthermore, the use of APLs must address the potential drawbacks, including cross-reactivity, of the induced T cell not only to the wild-type epitope but also to undesirable autoimmune targets and the possibility that APLinduced responses respond with lower avidity to endogenous target antigens. One advantage of adoptive therapy in this respect is the ability to choose
ex vivo from among the population of APLinduced T cells that satisfy these criteria by screening
against those that recognize normal tissue targets and selecting for those that recognize endogenously expressed tumor target antigens with high avidity.
Endogenous Specificity: Generating Antigen-Specific T Cells from Existing Repertoire
The in vitro isolation and expansion of antigen-specific T cells for adoptive therapy are in essence a recapitulation of the in vivo events of priming and expansion and involve at least four components: An effector population (source of T cells), stimulator cell, TCR ligand, and T-cell growth factor (i.e., lymphokines such as IL-2) (
Table 33-1).
Naturally Occurring Antigen-Presenting Cells as Stimulator Cells
Although DCs, under specific conditions, can be tolerogenic, they are generally considered “professional” APCs with specialized stimulatory features, for example, their capacity to up-regulate expression of costimulatory ligands, and Th1-type cytokines such as IL-12. They represent a robust in vitro stimulator population. In one embodiment, an enriched population of DCs can be generated by treatment of adherent mononuclear cells with GM-CSF and IL-4. Following maturation with an immunomodulatory cocktail or conditioned media, DCs can then be loaded with peptide, transfected with RNA or expression plasmid or transduced with recombinant viral vectors expressing the target antigen of interest. When cocultivated with autologous T cells, and a source of cytokines such as IL-2, antigen-specific T cells can be expanded in vitro for downstream applications. Other approaches for generating human dendritic cells include the use of cytokines such as IL-15
24,
25, toll receptor (TLR) agonists, or CD40 to enhance DC function and the addition of FLT3 ligand to expand DCs in vitro or in vivo prior to peripheral blood mononuclear cell harvest.
22,
26
A more readily available source of autologous APCs has also been developed using CD40 ligand to activate and facilitate long-term growth of human B cells.
27 These CD40-activated B cells express high levels of costimulatory molecules and when transfected with the target antigen of interest or epitope peptide serve as robust APCs in vivo and in vitro. Furthermore, CD40 activation of malignant B cells can render them effective APCs and provide a direct means of stimulating leukemia-reactive CD4 and CD8 T cells.
28
Whatever the source, when stimulator cells are cocultivated with autologous T cells, in vitro expansion of antigen-specific cells
requires a
γ-chain receptor cytokine, such as IL-2, and iterative cycles of in vitro restimulation. In contrast to more nonspecific strategies for expanding effectors such as LAK and TIL where high doses of IL-2 (upwards of 6,000 U/mL) are required, low doses of IL-2 (10 to 50 U/mL) are generally sufficient to induce antigen-driven expansion due to the up-regulation of high-affinity IL-2R
α coreceptor following TCR engagement. Other
γ-chain receptor cytokines, such as IL-7 and IL-15, can also be used to augment expansion and have been shown to augment a population of memory effector cells. The addition of IL-21 is unique in its capacity to generate antigen-specific T cells displaying elevated levels of CD28 and, in some cases, CD62L, with effector function, and may represent a central memory-like helper-independent CD8
+ T cells with features of arrested differentiation and enhanced replicative potential.
29
Artificial Antigen-Presenting Cells as Stimulator Cells
The use of artificial APCs addresses some of the obstacles to using autologous mononuclear cells by providing a convenient ex vivo source of stimulator cells, a product that is more likely to be uniform in its physical and functional properties and the flexibility of expressing desired antigen-specific and/or costimulatory receptors.
Mouse fibroblast (3T3) lines and insect cells (
D. melanogaster) can be engineered to express the appropriate HLA (usually the prevalent HLA-A2 allele that presents many of the known tumor-associated antigenic epitopes) as well as costimulatory receptors, B7.1, ICAM-1, and LFA-3, found to be necessary for optimal CD8 T-cell stimulation.
30 Pulsed with the desired peptide, these APCs could be used to generate human antigen-specific CTL responses in vitro. Using HLA-A2+ insect cells expressing B7.1 and ICAM, Mitchell et al.
31 were able to enrich for and expand tyrosinase-specific CTL from peripheral blood of patients with metastatic melanoma to more than 10
9 cells in the presence of IL-2 and IL-7 for use in a clinical trial of adoptive therapy.
The human NK-susceptible chronic myeloid leukemia (CML) cell line, K562, has also been explored for use as an artificial APC. In the absence of MHC expression, it can be used as a “blank” slate for decorating with the desired TCR ligand and costimulatory molecules. June et al. stably transduced Fc receptors to allow for display of anti-CD3 and anti-CD28 antibodies. When 4-1BBL (CD137L) was coexpressed, optimal, nonspecific in vitro expansion of T cells was achieved. By transfecting K562 with the relevant HLA allele and pulsing with the desired epitope peptide, tumor-associated antigen-specific CD8
+ T cells could be reliably generated in vitro.
32 Engineering aAPCs to express
γ-chain receptor cytokines, such as IL-21, led to further enhancement in the qualitative and quantitative expansion of antigen-specific CTL.
33
Acellular products used as artificial APCs include magnetic beads, liposomes, and exosomes. Magnetic beads covalently linked to anti-CD3 and anti-CD28 provide a means to rapidly expand a population of polyclonal T cells and have been used in a number of clinical trials: anti-CD3-/anti-CD28-activated donor lymphocyte infusions (DLI) have led to increased responses posttransplant for CML, non-Hodgkin’s lymphoma (NHL), and myeloma
34,
35,
36 compared to untreated DLI. For generating antigen-specific T cells, class I and class II Ig dimers can be attached to the beads to permit exogenous loading of peptides for in vitro stimulation and, as a vaccine reagent, for in vivo expansion of adoptively transferred CTL.
37
Ex Vivo Selection and Expansion
By whatever means a population of antigen-specific T cells is generated, the question facing immunotherapy investigators is whether the population will require further expansion or in vitro selection before infusion. For some antigens, by virtue of a preexisting high endogenous frequency
38 or an elevated frequency that might be predicted in patients due to a concomitant serologic response,
39 application of the above methods to generate antigen-specific T cells in culture may be sufficient to produce a population with adequate antitumor activity. Further expansion may involve the use of ligands or agonist antibodies to TCR (CD3) and/or costimulatory receptors (CD28, 4-1BB, etc.) coupled with irradiated feeder cells and cytokines. In some cases, the population of T cells can be expanded from 100- to 1,000-fold over 2 to 3 weeks.
40,
41,
42
If selection is desirable, then reagents to nondestructively identify antigen-specific T cells, for example, peptide-MHC multimers, can be used together with a separation method (i.e., flow cytometric cell sorting). Since the natural ligand for the TCR, the peptide-MHC complex, cannot be used singly as a staining reagent for antigen-specific T cells because of its high dissociation rate, multimerizing pMHC complex to a fluorophore-conjugated molecule in a method first pioneered by the Davis lab permits a robust and sensitive means for not only detecting antigen-specific T cells (in a nondestructive manner) but also for isolating tumor-reactive T cells for downstream analysis or adoptive therapy.
43,
44 Later developments in this field include the novel use of Ig fragments to multimerize pMHC complexes for ex vivo detection,
45 (and as described above, for in vitro as well as in vivo stimulation) and the creation of class II tetramers to identify and select for antigen-specific CD4 T cells.
46 Global approaches to the selection of antigen-specific T cells exploit functional antigen-driven properties such as cytokine production and surface marker up-regulation. A bispecific antibody binding to a constitutively expressed T-cell surface marker (e.g., CD45) linked to an antibody that captures a secreted cytokine, such as IFN-
γ, permits selection of viable antigen-activated IFN-
γ+T cells when a second detection antibody to IFN-
γ is used.
47 Alternatively, among surface markers that are up-regulated during antigen recognition (CD25, CD69, 4-1BB/CD137), CD137 surface up-regulation correlates most strongly with TCR ligation and can be used to identify and sort for antigen-specific CD8 T cells.
48