Identification of a major histocompatibility complex class I-restricted T-cell epitope in the tumour-associated antigen, 5T4

Abstract

5T4 is a surface glycoprotein expressed on placental trophoblasts and also on a wide range of human carcinomas. Its highly restricted expression on normal tissues and broad distribution on many carcinomas make 5T4 a promising target for cancer immunotherapy. In the current study, we set out to investigate whether a 5T4-specific cytotoxic T lymphocyte (CTL) repertoire exists in healthy individuals. CD4-depleted peripheral blood mononuclear cells (PBMCs) from blood donors were screened using an ex vivo interferon-γ (IFN-γ) enzyme-linked immunospot (ELISPOT) assay. A panel of overlapping peptides, spanning the full length of the 5T4 protein, was used as a source of antigen. In the process of screening, one out of 30 blood donors demonstrated a positive ex vivo IFN-γ ELISPOT response to a single 5T4 peptide. A polyclonal T-cell line was derived from this donor by culturing PBMCs with autologous peptide-pulsed dendritic cells (DCs). The resulting polyclonal T-cell line and clones were tested in a ⁵¹Cr-release assay and by ELISPOT and were shown to be peptide specific. Furthermore, antigen-presenting cells (APCs), infected with a viral vector expressing 5T4, were able to stimulate IFN-γ production by the peptide-specific T-cell clones. A minimal CD8 epitope, PLADLSPFA, has been identified and found to be restricted through human leucocyte antigen (HLA) Cw7. Subsequently, we have demonstrated that HLA-Cw7-positive colorectal cancer patients vaccinated with a recombinant vaccinia viral vector encoding 5T4 (TroVax) are capable of mounting a strong IFN-γ ELISPOT response to this novel CTL epitope. These findings have potential application in cancer immunotherapy in terms of subunit vaccine design and the monitoring of immune responses induced in patients by 5T4-based therapies.

Introduction

Over the last decade, new tumour-associated antigens (TAA) and their T-cell epitopes have been sought as potential targets for cancer immunotherapy. A number of TAAs have been studied extensively and their ability to induce tumour-specific immune responses has been characterized. Tumour antigens have been classified into a number of categories based on their origins; these include foreign viral antigens, self-tissue specific and differentiation antigens, overexpressed or mutated self-antigens, and proteins that are expressed during fetal development but absent on normal adult tissues.^1,2 5T4 falls into this last category.

Human 5T4 is a 72-kDa oncofoetal glycoprotein expressed on a variety of human tumours, including carcinoma of the breast, gastro-intestinal tract, lung, colon, cervix and ovary.^3–6 However, with the exception of the placenta, 5T4 is rarely expressed on normal tissues. The role of 5T4 in carcinogenesis has still to be elucidated, but there is evidence that 5T4 expression in cancer cells is associated with poor prognosis in colorectal and gastric cancer.⁷ Additionally, tumour cells transfected with 5T4 cDNA display increased motility, suggesting a role in the metastatic process.⁸

Given that 5T4 is expressed on a wide range of human carcinomas, it is a potential target for cytotoxic T-lymphocyte (CTL)-mediated killing if an efficacious immune response can be elicited in cancer patients. Animal studies have demonstrated that it is possible to break tolerance to murine 5T4 (m5T4) following vaccination of mice with a recombinant modified vaccinia Ankara (MVA) virus expressing m5T4.⁹ In addition, BALB/c mice vaccinated with MVA expressing human 5T4 (h5T4) were protected from tumour growth after subsequent challenge with a murine tumour cell line stably transfected with h5T4.¹⁰ Furthermore, in a therapeutic setting, two injections of mice with MVA.h5T4 resulted in a significant reduction in tumour burden.¹⁰ MVA expressing h5T4 (TroVax) has been developed as a cancer vaccine and tested in a phase I/II clinical trial in patients with colorectal cancer. The vaccine was shown to be safe and well tolerated, and induced 5T4-specific immune responses in the majority of patients (R. Harrop et al., submitted). TroVax is currently being tested in two phase II clinical trials in colorectal cancer patients in whom 5T4-specific cellular and humoral immune responses are being monitored. In the current study, the existence of a 5T4-specific CD8 T-cell repertoire in healthy individuals was investigated. CD4-depleted peripheral blood mononuclear cells (PBMCs) from 30 healthy donors were screened for their responsiveness to a 5T4 peptide library spanning the entire length of the 5T4 protein. T cells from a single donor showed a positive interferon-γ (IFN-γ) response to one of the 5T4 peptides. 5T4-specific CD8 T cells detected in this individual were expanded in vitro and used as an investigative tool to identify the first 5T4-specific CD8 T-cell epitope and its human leucocyte antigen (HLA) restriction element. Furthermore, CTLs specific for this epitope were induced in HLA-Cw7-positive colorectal cancer patients following vaccination with TroVax. Thus, this novel 5T4 epitope has application for immunomonitoring and as a component of 5T4-based immunotherapy approaches.

Materials and methods

PBMCs and cell lines

Blood samples from 30 healthy individuals were obtained from the National Blood Transfusion Centre (Bristol, UK). Blood samples were also obtained from colorectal cancer patients enrolled in an ongoing phase 2 clinical trial in which they were immunized up to six times with 5 × 10⁸ plaque-forming units (PFU) of TroVax. The trial was approved by the South Manchester Local Research Ethics Committee prior to commencement of enrolment, and informed consent was obtained from each patient prior to enrolment. Blood samples were obtained prior to TroVax administration and 2 weeks after each vaccination. PBMCs from both healthy donors and immunized cancer patients were isolated from heparinized blood by standard density-gradient centrifugation through Histopaque-1077 columns (Sigma-Aldrich Ltd, Dorset, UK). Where required, dendritic cells (DCs) were generated from the adherent fraction of PBMCs following 7 days of culture in RPMI containing 10% fetal calf serum (FCS) supplemented with interleukin-4 (IL-4) (1000 U/ml) and granulocyte–macrophage colony-stimulating factor (GM-CSF) (50 ng/ml) (Totam Biologicals Ltd, Northampton, UK). The non-adherent fraction of PBMCs isolated from healthy donors was depleted of CD4 T cells by incubation with M450-CD4 Dynabeads (Dynal, Oslo, Norway); these were used as a responder cell population in functional T-cell assays. Autologous Epstein–Barr virus (EBV)-transformed B-cell blasts carrying the B95.8 virus strain (LCL) were established and maintained as a source of stimulation and/or target cells.

Synthetic 5T4 peptides

A set of peptides [95% pure by high-performance liquid chromatography (HPLC)], spanning the entire length of the 5T4 protein, was synthesized by Mimotopes, Ltd (Melbourne, Australia), using a solid-phase method based on standard Fmoc chemistry. The peptides, 10 amino acids long and offset by two amino acids, were solubilized in dimethyl sulphoxide (DMSO) at a concentration of 5 mg/ml.

Recombinant MVA viruses

The highly attenuated vaccinia virus vector, MVA, was engineered to express 5T4 (termed TroVax) or β-galactosidase (MVA–LacZ). These recombinant viruses were used throughout this study to express 5T4 or the β-galactosidase control antigen, endogenously in target cells. Details of the recombinant viral vector construction have been described previously.¹⁰

Generation of T-cell lines and clones

A DC preparation matured for 24 hr with recombinant tumour necrosis factor-α (rTNF-α) (25 ng/ml) (Totam Biologicals Ltd) was loaded with the target peptide (50 µg/ml) and β₂-microglobulin (2·5 µg/ml) for 2 hr at 37°. After unbound peptide was removed by washing, DCs were cocultured with CD4-depleted PBMCs in the presence of interleukin-7 (IL-7) (5 ng/ml) (Totam Biologicals Ltd) at a PBMC:DC ratio of 20 : 1. One week later, responding cells were stimulated further by adding freshly prepared peptide-loaded DCs, and interleukin-2 (IL-2) was added to the culture to a final concentration of 100 U/ml (Totam Biologicals Ltd). After four weekly rounds of stimulations, the resulting polyclonal T-cell line was assessed for antigen specificity by a standard ⁵¹Cr-release assay and cloned, by limiting dilution, at three cells/microwell, on gamma-irradiated autologous peptide-loaded LCL (10⁴/well) and gamma-irradiated allogeneic PBMC feeders (10⁵/well) in IL-2-supplemented medium.¹¹ Microcultures were expanded and used subsequently to identify a minimal 5T4 CD8 epitope and its HLA restriction.

IFN-γ enzyme-linked immunospot assay

Ninety-six-well Immobilon plates (Millipore, Bedford, MA, USA) were precoated with an anti-hIFN-γ monoclonal antibody (mAb), 1-DIK (MabTech, Stockholm, Sweden), at a concentration of 15 µg/ml. Ex vivo enzyme-linked immunospot (ELISPOT) assays were performed using 5 × 10⁵ CD4-depleted PBMCs per well from healthy donors, or 2 × 10⁵ total PBMCs per well from colorectal cancer patients. Cells were incubated in the presence of single or pooled (10 adjacent peptides per pool) peptides at a final concentration of 5 µg/ml per peptide. A peptide pool containing major histocompatibility complex (MHC) class I-restricted epitopes from human cytomegalovirus, EBV and influenza virus (CEF) was used as a positive control (MabTech). T-cell lines and clones which had undergone several rounds of in vitro stimulation were used at 10³ cells/well and cocultured with autologous LCL which had been preloaded with peptide or infected with TroVax. Plates were incubated overnight at 37°, the cells discarded the following day, and the plates were processed according to the manufacturer's instructions. Individual cytokine-producing cells were detected as dark spots and counted using an ELISPOT plate reader (AID, Strassberg, Germany).

Cytotoxicity assay

The lytic activity of polyclonal T-cell lines and clones was tested by a standard ⁵¹Cr-release assay. Briefly, autologous LCL were loaded with the peptide (5 µg/ml) for 1 hr or infected with TroVax and control MVA overnight, at a multiplicity of infection (MOI) of 10, and labelled with 100 µCi of ⁵¹Cr for 1 hr at 37°. After three washes, ⁵¹Cr-labelled target cells (5 × 10³/well) were incubated with various numbers of effector cells for 5 hr at 37°. The percentage of cytotoxicity was calculated as percentage specific lysis, as follows:

Results

Ex vivo 5T4-specific immune response in healthy individuals

To investigate whether a 5T4-specific T-cell repertoire exists in healthy individuals, randomly selected normal blood donors were screened by IFN-γ ELISPOT assay using a set of overlapping 5T4 peptides spanning the entire protein sequence. Freshly isolated PBMCs were enriched for CD8 T cells by negative selection and cultured for 24 hr with 5T4 peptide pools in ELISPOT plate wells.

Every donor was tested against every single 5T4 peptide pool. In the course of screening, one out of 30 donors (designated OB8) demonstrated the presence of 5T4-specific CD8 T cells. Only one pool out of the 20 peptide pools tested, pool 8, induced substantial IFN-γ secretion compared with the cells that were not exposed to any antigen (Fig. 1). None of the other 5T4 peptide pools elicited IFN-γ secretion, either in donor OB8 or in any other donor (data not shown). Subsequently, the positive response to peptide pool 8 was dissected by coculturing OB8 T cells with each constituent peptide (peptides 8.1–8.10). Peptide 8.7 (PLADLSPFAF, amino acids 153–162) induced a specific response of similar magnitude to that induced by pool 8 (Fig. 1). The precursor frequency of p8.7-specific CD8 T cells in the blood of OB8 donor was estimated to be one in 10 000.

Figure 1.

Ex vivo enzyme-linked immunospot (ELISPOT) assay of OB8 CD4-depleted peripheral blood mononuclear cells (PBMCs) with 5T4 peptides of pool 8. The cells were incubated for 24 hr with the peptide pool 8, previously defined as positive, and with the individual peptides constituting peptide pool 8 at a concentration of 2 µg/ml in triplicate wells. Peptide 8.7 induced interferon-γ (IFN-γ) production of the same magnitude as the pool. This experiment was repeated three times, with similar results obtained on each occasion. No Ag cont., no antigen control; SFC, spot forming cells.

In vitro expansion of 5T4-specific T cells

Having identified a 10-mer peptide 8.7 (designated p8.7) as a potential natural epitope within the 5T4 protein, we set out to define the minimal epitope within the peptide. To address this, a polyclonal T-cell line, specific for 5T4, was established. Autologous monocyte-derived DCs were loaded with p8.7 and cocultured with OB8 CD8 T cells. After four weekly stimulations with p8.7, the resulting polyclonal T-cell population was tested in a standard ⁵¹Cr-release assay for the presence of 5T4-specific T cells. Two LCL lines were used as target cells: the autologous (OB8) LCL; and an allogeneic LCL (CmC), which was partially HLA mismatched (Table 1). To verify that CTLs generated with p8.7-loaded DCs could also recognize 5T4 protein expressed endogenously in the target cells, OB8 and CmC LCL cells were infected with either TroVax or the control virus, MVA–LacZ, 24 hr prior to the cytotoxicity test. OB8 CTLs recognized both autologous and CmC target cells expressing 5T4 protein, whereas target cells infected with the control vector, MVA–LacZ, showed low background killing (Fig. 2a). This experiment proved that p8.7 is indeed a naturally processed 5T4 epitope. This finding was corroborated by the ELISPOT assay (Fig. 2b), which demonstrated that IFN-γ secretion by the polyclonal T-cell line was induced by exogenously loaded p8.7 and by 5T4 protein endogenously processed by the target cells following infection with TroVax.

Table 1.

Interferon-γ (FN-γ) secretion induced in OB8 T cells following the presentation of p8.7A peptide by LCL cells of various human leucocyte antigen (HLA) types. The HLA type of donor OB8 and OB8-matched HLA alleles of other donors are highlighted in bold.

LCL	HLA type			IFN-γ secretion
OB8	A 1.01, 2.02	B 8.01, 44.02	Cw 5.01, 7.01	Yes
CmC	A 2.01, 11.01	B 8.01, 44.02	Cw 5.01, 7.01	Yes
IM17	A 1.01, 2.01	B 13.01, 44.02	Cw 5.01, 6.02	No
IM108	A 3.01, 11.01	B 8.01, 35.01	Cw 4.01, 7.01	Yes
IM119	A 2.01, 68.02	B 15.11, 40.02	Cw 3.05	No
GS	A 2.01, 29.01	B 7.02, 44.03	Cw 7.02, 16.01	Yes

Figure 2.

An in vitro-generated OB8 polyclonal T-cell line was tested for the presence of 5T4-reactive T cells in functional assays. Cytotoxic T lymphocytes (CTLs), expanded using autologous dendritic cells (DCs) pulsed with p8.7, demonstrated killing of autologous and human leucocyte antigen (HLA)-matched target LCL, expressing 5T4 protein, in a ⁵¹Cr-release assay at a effector:target (E:T) ratio of 20 : 1 (a). These data are representative of three experiments. The CTLs also produced interferon-γ (IFN-γ) in response to autologous LCL exogenously loaded with p8.7 and expressing 5T4 from TroVax, when tested by enzyme-linked immunospot (ELISPOT) assay (b). APC, antigen-presenting cells; SFC, spot forming cells.

Identification of the minimal 5T4 epitope

Using a limiting-dilution method, T-cell clones were generated from the polyclonal 5T4-specific T-cell line by employing autologous γ-irradiated LCL loaded with p8.7. The clones were expanded and tested for specific reactivity by a ⁵¹Cr-release assay against autologous and HLA-matched LCLs infected with TroVax and MVA–LacZ as target cells. The clones found to be 5T4 specific exhibited a relatively low level of cytotoxicity, comparable with that detected in a polyclonal T-cell line. 5T4-specific clones were subsequently used as an investigative tool to further characterize responses to the p8.7 peptide. To identify a minimal epitope within p8.7, peptides were synthesized that were one or two amino acids shorter than the original 10-mer peptide. OB8 LCL were pulsed with each peptide and tested for recognition by the p8.7-specific T-cell clone using a ⁵¹Cr-release assay (Fig. 3a). Target cells loaded with peptide p8.7 were killed by the T-cell clone, but to a lower level than cells loaded with a 9-mer peptide (PLADLSPFA), designated p8.7A. Three other variants of p8.7 – p8.7B, p8.7C and p8.7D – induced minimal killing of target cells. This allowed us to conclude that the 9-mer peptide, PLADLSPFA, is the minimal epitope within p8.7.

Figure 3.

A generated OB8 T-cell clone was tested by ⁵¹Cr-release assay against target cells loaded with p8.7 and its variants [effector:target (E:T) = 5 : 1]. T-cell clones recognized original 10-mer peptide p8.7 and a 9-mer variant, p8.7A, but not other variants of p8.7 (a). Titration of peptides p8.7 (PLADLSPFAF) and p8.7A (PLADLSPFA) showed that the lowest concentration at which target cells were recognized was 100 nm (b).

The original 10-mer peptide, p8.7, and defined minimal epitope, p8.7A, were tested over a range of concentrations in a peptide-sensitization assay. CTLs were tested on targets pre-exposed to peptides at concentrations between 10 and 1000 nm. Peptide titration demonstrated that the lowest peptide concentration capable of sensitizing targets for killing was 100 nm (Fig. 3b).

Identification of the epitope HLA restriction element

To identify the HLA restriction element of peptide 8.7A, a panel of LCLs, partially matched with donor OB8 LCL at a number of MHC class I alleles (Table 1), were used as antigen-presenting cells (APCs) in an ELISPOT assay.

The LCLs IM17 and IM119, when loaded with p8.7A did not induce IFN-γ production by OB8 T cells, ruling out the possibility that the epitope was restricted through HLA-A1, -2 -B44 or -Cw5 alleles. However, a response was induced when IM108 LCL were used as APCs, suggesting that the epitope was presented through either HLA-B8 or -Cw7.

To further define the HLA restriction element, a blood sample was collected 1 year later from donor OB8, in order to produce an additional source of 5T4-specific effector cells. After three rounds of in vitro stimulation with autologous p8.7A-loaded DCs, we were able to generate 5T4-specific T-cell lines and clones from the donor's PBMCs. One T-cell clone, which strongly recognized the p8.7A peptide in a ⁵¹Cr-release assay, was used as a source of effector cells in functional assays to define HLA restriction for the p8.7A epitope. An LCL line (GS) was selected, which was HLA-matched with HLA-Cw7 but with no other class I elements present in donor OB8. These cells were used as target cells for a CTL assay and as APCs for ELISPOT assays. As shown in Fig. 4(a), CTLs caused 70% lysis of autologous LCL loaded with peptide p8.7A. As demonstrated previously, there was no recognition of IM17 targets pulsed with the peptide. However, GS LCLs loaded with the p8.7A peptide were killed by OB8 CTLs. The results of the cytotoxicity test correlated with the ELISPOT data shown in Fig. 4(b). In addition to autologous LCL, GS and IM17 LCLs were pulsed with p8.7A, cocultured with OB8 CTLs and IFN-γ secretion was measured 24 hr later. Again, GS LCL induced a positive response, as did autologous peptide-loaded LCL. The fact that GS LCL are matched with OB8 LCL by only one HLA class I allele, HLA-Cw7, allowed us to conclude that HLA-Cw7 is a restriction element for 5T4 peptide p8.7A.

Figure 4.

The human leucocyte antigen (HLA) restriction element of p8.7A was identified by a ⁵¹Cr-release assay (a) and enzyme-linked immunospot (ELISPOT) (b) using a 5T4-specific T-cell clone as effector cells and partially HLA-matched LCLs as targets and antigen-presenting cells (APCs), respectively. The T-cell clone recognized autologous LCL and GS LCL matched with OB8 by the Cw7 allele. These data are representative of three experiments. No Ag cont., no antigen control; SFC, spot forming cells.

Induction of p8.7-specific T cells in colorectal cancer patients vaccinated with TroVax

The novel 5T4 CTL epitope identified using PBMCs from healthy donors has subsequently been used as a tool for monitoring immunological responses in an ongoing phase II clinical trial. PBMCs recovered from colorectal cancer patients either prior to, or following, vaccination with TroVax, were tested in an ex vivo ELISPOT assay with p8.7. No p8.7-specific CTLs were detected in PBMCs taken from 16 HLA-Cw7-positive patients prior to TroVax vaccination. However, following vaccination with TroVax, four of these patients gave positive IFN-γ ELISPOT responses to p8.7. One patient had a detectable response after two vaccinations, which increased further with additional TroVax injections, such that after five injections, one in 2114 PBMCs were specific for p8.7 (Table 2). Such increases in antigen-specific precursor frequency were not evident in the response to the control pool of recall antigens (CEF). Three other patients responded to TroVax vaccination by mounting a CD8 response to 5T4 antigen, with peptide p8.7-specific T-cell frequencies ranging from one in 24 390 to one in 2096 PBMCs (data not shown).

Table 2.

Precursor frequencies specific for 5T4 peptide 7.8, and for a peptide pool containing major histocompatibility complex (MHC) class I-restricted epitopes from human cytomegalovirus, Epstein–Barr virus (EBV) and influenza virus (CEF), from peripheral blood mononuclear cells (PBMCs) of an HLA-Cw7 cancer patient, before and after vaccination with TroVax

	Precursor frequency
Number of vaccinations	Peptide 7.8	CEF peptide pool
0	≤1/200 000	1/1029
2	1/12 500	1/826
4	1/4049	1/826
5	1/2114	1/735

Discussion

It has been over a decade since the 5T4 protein was identified as a TAA.^3–8 The utility of 5T4 as a target for cancer immunotherapy was first demonstrated in a preclinical model, where MVA expressing murine 5T4 was shown to induce an immune response, in mice, against m5T4. Such immune responses were capable of protecting mice against challenge with syngeneic tumours expressing m5T4.⁹ It is generally thought that CTLs play a crucial role in tumour surveillance and protection. However, to date, no CTL epitopes within the 5T4 antigen have been identified. In the course of this study, we set out to investigate whether a 5T4-specific T-cell repertoire exists in healthy individuals by randomly screening PBMCs by IFN-γ ELISPOT using a panel of overlapping 5T4 peptides. As a result, one out of 30 healthy individuals demonstrated an ex vivo IFN-γ ELISPOT response to one of the 5T4 10-mer peptides. The precursor frequency of 5T4-specific CD8 T cells was estimated to be one in 10 000, which is relatively high for a self-antigen. However, it is not unusual that T cells, specific for self-antigens, can be detected in healthy individuals. For example, CTLs specific for the melanoma antigens tyrosinase and Melan-A/Mart-1 were detected in healthy donors, although this was only possible after in vitro stimulation with the cognate antigen.^12,13 In contrast, CTL precursors specific for Melan-A/Mart-1 were detected in unstimulated PBMCs recovered from patients with melanoma. The frequency of these Melan-A/Mart-1-specific CTLs, quantified by tetramer staining, was comparable to the 5T4 CTL frequency in donor OB8.¹⁴ CTLs to another TAA, self-antigen cyclin B1, were found in the unstimulated blood of a breast cancer patient, and their frequency, estimated by ELISPOT, was greater than one in 1000, which is 10-fold higher than that observed for 5T4 p8.7.¹⁵ Based on these data, we hypothesize that 5T4-specific T cells detected in the unstimulated blood of donor OB8 had been primed in vivo. Potentially, this could occur when a 5T4-positive tumour arose spontaneously and was successfully cleared by the induced 5T4-specific immune response prior to being diagnosed. An assumption that tolerance to 5T4 can be broken in vivo was supported by the fact that the mice immunized with MVA encoding murine 5T4 mounted a 5T4-specific antibody response.⁹ Furthermore, it has been corroborated in ongoing phase II clinical trials by the detection of strong 5T4-specific CD8 T-cell responses in colorectal cancer patients following vaccination with TroVax (data not shown; R. Harrop, N. Connolly, I. Redchenko et al., manuscript in preparation). Interestingly, when an additional blood sample from donor OB8 was collected 1 year later, no p8.7- or p8.7A-specific T cells could be detected ex vivo in unstimulated PBMCs by ELISPOT assay. However, following three rounds of in vitro stimulation with autologous p8.7A-loaded DCs, we were able to generate a 5T4-specific polyclonal T-cell line and T-cell clones from the donor's PBMCs. If our hypothesis is true, and donor OB8 had mounted an immune response to 5T4 in vivo during the year between blood samples, 5T4-specific effector memory T cells could have been eliminated through an activation-induced cell death (AICD) mechanism. However, there is a possibility that a subset of 5T4-specific central memory T cells lacking immediate effector function¹⁶ remained in the circulation, and it is this population that were expanded by in vitro stimulation.

A minimal 5T4 9-mer CTL epitope was identified by performing a killing assay of target cells loaded with the shorter variants of the original 10-mer peptide. Peptide titration analysis performed with the minimal epitope peptide demonstrated that in order to sensitize target cells for recognition by CTLs, they had to be pulsed with a relatively high concentration of the peptide. The lowest peptide concentration inducing killing of the target cells was 100 nm, indicating that in vitro-reactivated CTLs were of low affinity. This finding was not wholly unexpected as a high concentration of peptide (50 µg/ml) was used to stimulate the T cells. It has also been reported previously that the use of high peptide concentrations for CTL generation resulted in T-cell clones of differential affinity for the target cells.¹⁷ One can also suggest that because the 5T4 protein is a self-antigen, T cells that display a T-cell receptor (TCR) with a high avidity for 5T4 were tolerized or eliminated in the thymus, leaving only T cells with low avidity, thus excluding autoreactive T cells from circulating and destroying normal tissues.

ELISPOT and CTL assays performed against a panel of APCs and target cells partially matched with donor OB8 by one or more HLA alleles allowed identification of the epitope HLA restriction to the Cw7 allele. The HLA-Cw7 allele has not been characterized as extensively as others with regards to its primary anchor motifs. However, the occurrence of alanine in position 3 of the HLA-Cw7 allele has been documented,¹⁸ supporting our findings.

A polyclonal T-cell line, generated by in vitro stimulation of PBMCs with DCs loaded with the 5T4-derived peptide, has proved not only to be specific for the peptide, but also able to recognize the full-length 5T4 protein endogenously expressed in the target cells. Unfortunately, we have not been able to demonstrate the killing of tumour cells, owing to the lack of a tumour cell line that expresses 5T4 in the context of HLA-Cw7. Another limiting factor was the number of T-cell effectors generated from donor OB8. The OB8 5T4-specific T-cell clones were exhausted in the course of the study, and a resupply of PBMCs from donor OB8 was not possible. We were unable to use the blood samples from the cancer patients, owing to changes in clinical trial standards and restrictions placed by the Ethics Committee, which do not allow us to derive a T-cell line from the patients' PBMCs. However, both the polyclonal T-cell line and individual T-cell clones recognized autologous and 5T4-positive HLA-matched LCLs, thus providing a strong indication that they should also be capable of killing other 5T4-positive Cw7-positive tumour cell lines.

Interestingly, Cw7-positive GS LCL was recognized by a specific T-cell clone much less efficiently than autologous LCL targets. The difference in recognition level is probably caused by differences in the level of Cw7 expression on the LCLs. Alternatively, this can be accounted for by the fact that OB8 and GS LCLs express different subtypes of Cw7 allele: Cw7.01 for OB8 and Cw7.02 for GS. HLA-Cw7.02 and Cw7.01 differ by two amino acids at positions 66 and 99, which may contribute to the A, B, C and D binding pockets.¹⁹ Therefore, GS LCL are likely to bind p7.8A less efficiently than autologous LCL, thus leading to lower levels of recognition by effector cells.

The detection of p8.7-specific responses in colorectal cancer patients following vaccination with TroVax is encouraging, for several reasons. The induction of such responses in HLA-Cw7-positive patients corroborates our finding that Cw7 is indeed the restriction element for this epitope. Furthermore, it demonstrates that p8.7 is a natural CD8 T-cell epitope, which is processed and presented following infection of antigen-presenting cells by TroVax.

The study reported here demonstrates, for the first time, the presence of a CD8 T-cell repertoire specific for the TAA 5T4 in an apparently healthy donor. This represents the first CD8 T-cell epitope identified within the 5T4 antigen and is restricted through HLA-Cw7, which is expressed by Caucasian, Black and Oriental populations at average frequencies of 23%, 21% and 15%, respectively.¹⁸ CTLs specific for this epitope were induced in HLA-Cw7-positive colorectal cancer patients following vaccination with TroVax. These findings, apart from being of significant scientific value, have potential application in cancer immunotherapy of HLA Cw7-positive patients in terms of subunit vaccine design and the monitoring of immune responses induced by 5T4-based therapies.

Abbreviations

AICD

activation-induced cell death

APC

antigen-presenting cell

CTL

cytotoxic T lymphocyte

DC

dendritic cell

DMSO

dimethyl sulphoxide

EBV

Epstein–Barr virus

ELISPOT

enzyme-linked immunospot

FCS

fetal calf serum

GM-CSF

granulocyte–macrophage colony-stimulating factor

HPLC

high-performance liquid chromatography

h5T4

human 5T4

HLA

human leucocyte antigen

IFN-γ

interferon-γ

IL-2

interleukin 2

IL-4

interleukin-4

IL-7

interleukin 7

mAb

monoclonal antibody

MHC

major histocompatibility complex

MVA

modified vaccinia Ankara

MOI

multiplicity of infection

m5T4

murine 5T4

PBMC

peripheral blood mononuclear cell

PFU

plaque-forming unit

rTNF-α

recombinant tumour necrosis factor-α

TAA

tumour-associated antigen

TCR

T-cell receptor

TroVax

a recombinant vaccinia viral vector encoding 5T4