Specific Cytolytic T-Cell Responses to Human CEA from Patients Immunized with Recombinant Avipox-CEA Vaccine

It has now been demonstrated that several human tumor-associated antigens can be recognized by human CTLs in the context of MHC-peptide complex on the surface of human tumor cells (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14). A Phase I clinical trial involving the use of a recombinant vaccinia virus expressing CEA² (rV-CEA) in patients with metastatic carcinoma (gastrointestinal, lung, or breast)has been completed (15). The only toxicity observed was that seen with the conventional smallpox vaccine. Cytotoxic T-cell lines specific for a human CEA epitope designated CAP-1 (a 9-mer immunodominant class I HLA-A2 CEA peptide) have been generated from patients immunized with the rV-CEA vaccine (16).

Poxvirus family members, such as vaccinia and avipox, have been used as poxvirus-based vaccines against viral pathogens (17). Replication of avipox viruses, which include fowlpox and canarypox, is highly restricted in avian species (18). It has been reported that fowlpox and canarypox virus-rabies recombinants express the rabies glycoprotein in non-avian tissue culture cells, and the level of expression is sufficient to induce rabies-specific neutralizing antibodies and to protect against a lethal rabies virus challenge (19, 20). Potency tests in mice have shown that a canarypox virus vector is highly efficient in expressing rabies glycoprotein (20). Additional data indicated that the avipox vectors are effective as immunizing agents in non-avian species using other viral immunogens (21, 22).

Preclinical evaluation of an ALVAC-human cytomegalovirus gB vaccine in mice indicated that it can induce both humoral and cell-mediated immune responses to human cytomegalovirus in hosts both with and without immunity to vaccinia (23). Furthermore, results from preclinical studies in nonhuman primates indicate that prior exposure to ALVAC recombinants should not preclude subsequent vaccination with novel ALVAC recombinants (24). Those findings were supported by Hodge et al. (25),who demonstrated that mice immunized with rV-CEA and then ALVAC-CEA elicit CEA-specific T-cell responses to levels greater than those found with the use of either vector alone. Phase I clinical trials with ALVAC-rabies glycoprotein recombinant and HIV-I glycoprotein 160 recombinant demonstrated that the experimental vaccines were well tolerated and induced humoral and cellular immune responses in humans (26, 27).

A Phase I clinical trial in advanced carcinoma patients using a replication-defective avipox vector containing the human CEAgene (ALVAC-CEA) as vaccine has been completed (28). The vaccine was administered i.m. three times at 28-day intervals. The vaccine was well tolerated at all dose levels(2.5 × 10⁵, 2.5 ×10⁶, and 2.5 × 10⁷pfu) with no significant toxicity (28). No objective antitumor responses were observed during the clinical trial in advanced carcinoma patients with measurable disease. The generation and characterization of CEA-specific cytotoxic T-cell lines from patients immunized with ALVAC-CEA is reported here. These results demonstrate that an ALVAC-CEA recombinant expressing CEA can induce specific anti-CEA CTLs capable of lysing, in a MHC-restricted manner, allogeneic and autologous tumor cells, thereby demonstrating the potential of the ALVAC vector in human anticancer vaccine development.

Colorectal carcinoma cell lines SW480 (HLA-A2, 24), SW1463(HLA-A1, 2) and LS174T (HLA-A2, -) were purchased from American Type Culture Collection (Manassas, VA). The cultures were free of Mycoplasma and were maintained in complete medium (DMEM;Life Technologies, Inc., Grand Island, NY) supplemented with 10% fetal bovine serum, 2 nm glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Inc.). The C1R cell line is a human plasma leukemia cell line that does not express endogenous HLA-A or -B antigens (29). C1R-A2 cells are C1R cells that express a transfected genomic clone of HLA-A2.1 (30); these cells were obtained from Dr. William E. Biddison (National Institute of Neurological Disorder and Stroke, NIH,Bethesda, MD). C1R-A2 cultures were free of Mycoplasma and were maintained in RPMI 1640 complete medium (Life Technologies, Inc.).

The peptides CAP-1 (CEA amino acid positions 571–579; YLSGANLNL;Ref. 16) and NCA-1 (YRPGENLNL) were generated on a peptide synthesizer (model 432A; Applied Biosystem, Inc., Foster City, CA); and products were dissolved in aqueous solution, sterile filtered, and frozen at −70°C at a concentration of 2 mg/ml. The purity of the peptides was >90%, as analyzed by high-performance liquid chromatography.

PBMCs were obtained from heparinized blood of patients with metastatic colon carcinoma who enrolled in a Phase I trial using ALVAC-CEA (28). All experiments involving patient materials were conducted according to NIH guidelines, and written,informed consent was obtained from all individuals. The protocol described by Tsang et al. (16) was used to generate the T-cell line. PBMCs were obtained prior to and after ALVAC-CEA injections given i.m. on days 1, 29, and 58 at 2.5 ×10⁵ pfu (level I), 2.5 ×10⁶ pfu (level II) or 2.5 ×10⁷ pfu (level III). PBMCs from the patient were separated using lymphocyte separation medium gradient (Organon Teknika,Durham, NC) as described previously (31). Washed PBMCs were resuspended in complete medium (RPMI 1640; Life Technologies,Inc.) supplemented with 10% pooled human AB serum (Valley Biomedical,Winchester, VA), 2 mm glutamine, 100 units/ml penicillin, and 100 μg/ml of streptomycin (Life Technologies, Inc.). Cells (2 × 10⁵) in complete medium in a volume of 100 μl were put into each well of a 96-well flat-bottomed assay plate (Corning Costar Corp., Cambridge, MA). CAP-1 peptide was added to cultures at a final concentration of 25 μg/ml. Cultures were incubated for 5 days at 37°C in a humidified atmosphere containing 5% CO₂. After the peptide-containing medium was removed, the cultures were supplemented with recombinant human IL-2(provided by the National Cancer Institute-Frederick Cancer Research and Development Center; 10 units/ml) for 11 days. Medium containing IL-2 was replenished every 3 days. The 5-day incubation with peptide and 11-day treatment with IL-2 constituted one IVS cycle. Primary cultures were restimulated with CAP-1 peptide (25 μg/ml) on day 16 to begin the next IVS cycle. Irradiated (4000 rads) autologous PBMCs(5 × 10⁵) were added in 50 μl of complete medium as APCs.

Target cells were labeled with 50 μCi of ¹¹¹In oxyquinoline (Medi-Physics, Inc.,Arlington, IL) for 15 min at room temperature. Target cells (0.5 × 10⁴) in 100 μl of complete medium were added to each of 96 wells in U-bottomed assay plates (Corning Costar Corp.). The labeled target cells were incubated with peptides at various concentrations for 60 min at 37°C in 5% CO₂before adding effector cells. Effector cells were suspended in 100 μl of complete medium supplemented with 10% pooled human AB serum and were added to target cells. The plates were then incubated at 37°C in 5% CO₂ for 4 or 16 h. Supernatant was harvested for gamma counting with the use of harvester frames (Skatron,Inc., Sterling, VA). Determinations were carried out in triplicate, and SDs were calculated. All experiments were carried out three times. Specific lysis was calculated with the following formula:

Spontaneous release was determined from wells to which 100 μl of complete medium were added. Total releasable radioactivity was obtained after targets were treated with 2.5% Triton X-100.

The method for single-color flow cytometric analysis has been described (32). Briefly, cells were washed three times with cold Ca²⁺ and Mg²⁺-free DPBS,and then stained for 1 h with a MAb against HLA-A2 (A2, 69,131HA-1; One Lambda, Inc., Canoga Park, CA) using 10 μl of the 1×working dilution/10⁶ cells. MOPC-104E(Cappel/Organon Teknika Corp., West Chester, PA) was used as isotype control. The cells were then washed three times and incubated with 1:100 dilution of FITC-labeled goat antimouse IgG (Kirkegaard & Perry Labs, Gaithersburg, MD). Anti-CEA MAb COL-1 was used as 100 μl of culture supernatant. The cells were then washed three times with cold DPBS and incubated for 1 h more in the presence of 1:100 dilution(volume of 100 μl PBS containing 1% BSA) of FITC-conjugated goat antimouse immunoglobulin (Kirkegaard & Perry Labs). The cells were again washed three times with DPBS and resuspended in DPBS at a concentration of 1 × 10⁶ cells/ml. The cells were immediately analyzed using a Becton Dickinson FACScan equipped with a blue laser with an excitation of 15 nW at 488 nm. Data were gathered from 10,000 live cells, stored, and used to generate results.

The procedure for dual-color flow cytometric analysis was similar to that for single-color analysis, except for the following: the antibodies used were anti-CD4 FITC/anti-CD8 PE conjugate, anti-CD2 FITC/anti-CD54 (Intercellular adhesion molecule-1) PE, anti-CD45 FITC/anti-CD49d PE, anti-CD11a (LFA-1) FITC/anti-CD58 (LFA-3) PE,anti-CD3 FITC/anti-CD62L PE, and anti-IgG1 FITC/anti-IgG2a PE (isotype controls). All of the antibodies listed above were purchased from Becton Dickinson. Staining was done simultaneously for 1 h, after which cells were washed three times, resuspended as above, and immediately analyzed using a Becton Dickinson FACSort equipped with a blue laser (excitation, 15 nW at 488 nm) and the CellQuest program.

Total cellular RNA was isolated from 5 ×10⁶ T cells using the standard guanidine isothiocyanate/acid phenol method (33). First-strand cDNA was then synthesized from 1 μg of total RNA and RT Superscript II(Life Technologies, Inc., Gaithersburg, MD).

Vβ subfamily-specific primers were used as described by Nishimura et al. (34). The previously fluorescent runoff Cβ-specific primer (5′-X-CACAGCGACCTCGGGTGGG-3′;Ref. 35) was synthesized by the manufacturer(Perkin-Elmer, ABI Division, Custom Oligo Synthesis Service, Foster City, CA).

Aliquots of the cDNA synthesis reaction (corresponding to 200 ng of total RNA) were amplified in 50-μl reactions with 1 of the 25 Vβoligonucleotides and the Cβ oligonucleotide. The final concentration of each primer was 20 pm/reaction. 0.2 mm of deoxynucleoside triphosphate and 1.5 mmMgCl₂ were in the Taq polymerase buffer (PE Applied Biosystems, Branchburg, NJ). The PCR cycle was as follows:denaturation at 94°C for 1 min, primer annealing at 60°C for 1 min,extension at 72°C for 4 min, and a final, 10-min polymerization at 72°C. A blank control without cDNA was included for each of the 24 Vβ-Cβ PCR reactions.

The labeled PCR products were loaded on a 6% acrylamide sequencing gel with the PCR reaction diluted 1:10 prior to loading. Samples were then run on an ABI 373 sequencer to determine their size and fluorescence intensity (35). The intensities of the peaks present in the labeled PCR products were measured at the end of the electrophoresis run, and the different peaks present in all Vβsubfamilies were added together. The relative percentages of each Vβsubfamily were then calculated and represented as histograms. This analysis was performed using the GeneScan collection and analysis software.

T-cell supernatants incubated with peptide-pulsed APCs for 24 h in IL-2-free medium at a responder:stimulator ratio of 1:3(10⁶ to 3 × 10⁶cells/ml) were screened for the secretion of IFN-γ, tumor necrosis factor-α, and IL-4 using an ELISA kit (Genzyme, Cambridge, MA). The results were expressed in pg/ml.

RNA from CAP-1-peptide-stimulated and unstimulated cells were analyzed by multiprobe RNase protection assay. Defined riboprobes for human cytokines were purchased from PharMingen (San Diego, CA). Assays were performed as described previously (36). Radioactivity contained in bands on dried polyacrylamide gels was quantified with a Storm System PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The net cpm for a given band was calculated by the following formula (cpm of cytokine gene − cpm of background) and was expressed as a percentage of the housekeeping gene transcript L32.

Statistical analysis of differences between pre- and postvaccination precursor frequencies was done using a paired t test. Statistical analysis of differences between mean values was performed with a two-tailed t test.

A Phase I clinical trial was carried out using the replication-deficient avipox vector (ALVAC) containing the CEA gene (ALVAC-CEA) in patients with metastatic tumors expressing CEA. Three dose levels (2.5 ×10⁵, 2.5 × 10⁶, and 2.5 × 10⁷) were used. At each dose level,patients received three injections of vaccine i.m. at 28-day intervals via the Bioject system. In an attempt to establish CEA-specific CTL lines from patients, PBMCs were obtained prior to vaccination, as well as 1 month after the third vaccination. Of the 20 patients enrolled in the study, eight were shown to possess the HLA class I A2 allele. Because an HLA class I A2 binding peptide, designated CAP-1, has been identified (16), the PBMCs of these HLA-A2-positive patients were chosen for further study.

PBMCs obtained from these patients pre- and postvaccination were incubated with the CAP-1 peptide and IL-2 using the protocol noted in “Materials and Methods.” As described, each five-day incubation with CAP-1 was followed by an 11-day incubation with IL-2, which constitutes an IVS cycle. Using prevaccination PBMCs, T-cell cultures could be obtained from five of eight patients, but none of these cultures could be expanded past IVS-5. These cultures were assayed for CTL activity versus both C1R-A2 cells alone and those pulsed with CAP-1 peptide. CAP-1-specific lysis was not observed in any of the cultures from the prevaccination samples (Table 1).

Using postvaccination PBMCs, T-cell lines could be established from all eight HLA-A2-positive patients vaccinated with ALVAC-CEA. The earliest IVS cycle in which a sufficient number of T cells could be obtained for cytotoxic assays was IVS-4. Cultures were first assayed for CTL activity using C1R-A2 cells, with and without the CAP-1 peptide, as targets. As seen in Table 1, CAP-1-specific lysis was obtained from cultures from seven of eight patients. Cultures generated from one patient, from the lowest dose level, were the only ones to show no specific lysis.

The cultures from those patients that showed specificity for lysis of cells pulsed with the CAP-1 peptide were then assayed for their ability to lyse human carcinoma cells endogenously expressing CEA. Three carcinoma cell lines were used as targets: SW1463 and SW480, both of which express HLA-A2 and CEA; and LS174, which expresses CEA but little, if any, HLA-A2 (Table 6). Cell lines derived from the seven patients that showed CAP-1-specific lysis all showed lysis to both the allogeneic SW1463 and SW480 carcinoma cell lines but not to the LS174 line (Table 2). The difference in percentage of specific lysis between the A2⁺lines and A2⁻ LS174 line was statistically significant (P < 0.05), as determined by the two-tailed t test.

T-cell lines derived from PBMCs of six vaccinated patients were also analyzed for their CD8⁺ and CD4⁺ phenotype. As seen in Table 3, line 8 (from patient 8) was ∼97%CD4⁻/CD8⁺. Lines 11, 14,16, and 18 were predominantly CD4⁻/CD8⁺, with a component of CD4⁺/CD8⁺double-positive cells; neither of these four lines showed a CD4⁺/CD8⁻ component. Line 15, on the other hand, was ∼60%CD4⁺/CD8⁻ and 37%CD4⁺/CD8⁺.

Because this was a Phase I study in patients with advanced carcinoma,tumor specimens from vaccinated patients were not readily available. For one patient (no. 11), however, a biopsy of a metastatic gastric carcinoma was available, and a culture of carcinoma cells from this biopsy was established. The T-cell line (no. 11) from this patient was thus chosen for further study. At IVS-8, this T-cell line was analyzed by flow cytometry for the expression of various T-cell markers, including the presence of various homing-associated and adhesion molecules. As seen in Table 4,>98% of the cells in the population were CD4⁻/CD8⁺ or CD4⁺/CD8⁺; 92% of the cells were CD2⁺/CD54⁺; 98%of the cells were CD45⁺/CD49d⁺ and CD11a⁺/CD58⁺; 58% of the cells were CD62L⁻/CD3⁺;and only 41% of the cells were CD62L⁺/CD3⁺. When the CTL line from patient 11 was stimulated with C1R-A2 APC and the CAP-1 peptide, it was shown to produce 1024 pg/ml of IFN-γ. When CAP-1 peptide was omitted, or control peptides PSA-3 and NCA were used,IFN-γ levels were below the 16 pg/ml detection level. No detectable levels (≤30 pg/ml) of IL-4 were produced in the presence of CAP-1. RNase protection experiments for cytokine expression were also carried out using CEA-specific CTLs that had been stimulated with CAP-1 peptide. The RNA expression levels of five different cytokines are shown in Fig. 1,A and are normalized to the L32 housekeeping gene in Fig. 1 B. Levels of IFN-γ and IL-5 RNA expression are shown to increase after stimulation with CAP-1, whereas only low levels of IL-2,IL-15, and IL-4 expression are observed before and after stimulation.

CAP-1-specific lysis was also observed over a wide range of peptide concentration (Table 5). A demonstration of the MHC-A2-restricted nature of this lysis is shown by antibody-blocking experiments (Table 6).

Studies were undertaken to determine whether the CTL line derived from this patient could lyse autologous tumor obtained at biopsy. Flow cytometry analyses demonstrated that the autologous tumor expressed both class I HLA-A2 and CEA (Table 7) at levels similar to those of the established allogeneic carcinoma cell lines SW480 and SW1463. The specificity was shown by lysis of C1R-A2 cells only with the addition of the CAP-1 peptide. The CTLs demonstrated lysis of the autologous tumor at levels similar to those seen with CEA-expressing and HLA-A2-expressing allogeneic tumor cells. The CEA-negative C1R-A2 cells and the HLA-A2-negative (but CEA-positive) LS174T tumor cells served as controls (Table 7).

The TCR Vβ gene usage of T-cell lines from patients 11, 16, and 18 was determined at IVS-6 by Vβ-Cβ PCR amplification using a fluorescent Cβ primer and was analyzed in an automated ABI sequencer. After the intensities of the peaks present in the labeled PCR products were added together, the relative percentages of each Vβ subfamily were calculated and represented as histograms (Fig. 2). TCR Vβ usage was defined in T-cell lines from patients 11 and 16, with 17 of 25 and 16 of 25 subfamilies present, respectively. A relatively restricted TCR Vβ usage was noted in the T-cell line from patient 18, with 14 of 25 subfamilies present(Fig. 3); however,>90% of the total T-cell line population from patient 18 was represented by Vβ 5, 6, 7, 8, 13, 14, and 21. Although TCR Vβ 16,18, 19, and 20 were not detected in any of the T-cell lines, TCR Vβ5, 7, 8, 13, 14, and 21 were detected at a higher frequency in all three T-cell lines.

The present study was undertaken to investigate the effectiveness of the ALVAC-CEA vaccine in eliciting T-cell immune responses against CEA in patients with advanced, CEA-expressing carcinoma. The results obtained from this investigation could provide valuable information on the safety of the ALVAC vaccine vector and the utility of ALVAC-CEA as a vaccine for immunotherapy of CEA-positive cancers. These studies demonstrate that CTL responses to CEA can indeed be elicited by vaccination with ALVAC-CEA. CEA-specific CTLs could be established from the postvaccination PBMCs of seven of eight patients. No CTL activity was detected from T-cell cultures established from prevaccination PBMCs. Cytotoxic activity of these established T-cell lines was shown against both CAP-1-pulsed C1R-A2 cells and allogeneic,CEA-expressing HLA-A2-positive tumor cells. CTL activity was seen when C1R-A2 cells were pulsed with CAP-1 peptide at a concentration as low as 1.6 μg/ml. Moreover, the cytolysis by these CTLs was shown to be MHC class I restricted. It has been reported that CTL activity and CEA-specific lymphoproliferative T-cell responses can be generated using rV-CEA vaccine in Phase I clinical trials (16, 37). This study demonstrated the ability to generate human CEA-specific CTLs from PBMCs of patients vaccinated with a replication-defective ALVAC-CEA. Preclinical data from murine studies demonstrated that vaccination with ALVAC-CEA can elicit CEA-specific T-cell responses,and the most potent responses were observed when ALVAC-CEA vaccination followed primary vaccination with rV-CEA (25).

It has been reported that the frequency of CAP-1-specific CTL precursors was higher in the postvaccination PBMCs than in the prevaccination PBMCs in a rV-CEA Phase I clinical trial (16). Similarly, in seven of nine patients, the precursor frequencies of CAP-1-specific CTLs were higher in postvaccination PBMCs (28). The increase in CTL precursor frequency to CAP-1 may be interpreted as the ability of ALVAC-CEA to elicit CEA-specific T-cell responses. The studies reported here demonstrate CTL response to only one CEA peptide (i.e., CAP-1). However, it has been shown (38) that patients can elicit T-cell responses to other A2 CEA peptides after vaccination with rV-CEA. Moreover, human T-cell responses to other CEA peptides (A3 and A24 alleles) have been demonstrated in vitro(39, 40).

The Vβ gene usage of T-cell lines from each of three patients (nos. 11, 16, and 18) at IVS-6 was analyzed using 25 Vβ oligonucleotides. A relatively restricted TCR Vβ usage was noted in T-cell line 18 but not in lines 11 and 16. TCR Vβ 5, 7, 8, 13, 14, and 21 were detected at a higher frequency in all three T-cell lines. Vβ 5, 6, 7, 8, 13,14, and 21 represented most of the population of T-cell line 18. The CDR3 analysis of all three T-cell lines suggests that these Vβsubfamilies consist of relatively polyclonal patterns. This result is in contrast to our previous report that the V8T cell line (41) established from a patient vaccinated with rV-CEA showed oligoclonal patterns in Vβ gene usage. The polyclonal pattern may be attributable to the fact that the Vβ subfamily analysis was performed using T-cell lines, not clones, at a relatively low IVS(IVS-6). Vβ subfamily analysis of the V8T cell line was performed at IVS-10 and IVS-20. It is conceivable that oligoclonal patterns in Vβsubfamilies may be observed in the higher IVS in T-cell lines 11, 16,and 18. Furthermore, these results suggest that the TCR repertoire determined in T-cell lines established from patients vaccinated with ALVAC-CEA was diverse. Similar results were obtained from T-cell lines established from patients vaccinated with rV-CEA (41). Diversity in the TCR Vβ usage has been reported in CTLs specific to other tumor-associated antigens (42, 43, 44).

Expression of homing-associated adhesion molecules and cytotoxic activity to autologous tumor cells was also investigated to ascertain the effectiveness of the ALVAC-CEA vaccine in eliciting specific T-cell responses. T-cell line 11 was used for the detailed analysis. The expression of cell adhesion molecules on T-cell lines has been implicated in CTL function. Molecules such as LFA-1 (CD11a) and CD49d have been shown to be involved in lymphocyte homing in vivo(45). In this study, CD11a and CD49d were expressed in 98% of the population of T-cell line 11. The expression of these molecules may be important for CTL function and for the development of potential adoptive transfer immunotherapy protocols. Furthermore,T-cell line 11 has a cytokine profile of a CD8 Tc1 effector cell. Analysis of the cytolytic activity of T-cell line 11 indicated that CTLs generated from a patient vaccinated with ALVAC-CEA using CAP-1 peptide can kill not only CAP-1-pulsed C1R-A2 cells but also HLA-A2-and CEA-expressing autologous and allogeneic tumor cells.

Phase I clinical trials using CEA as immunogen have demonstrated that CEA is immunogenic in humans (16, 28, 46, 47). Analysis of patient sera for antivaccinia antibodies before and after each rV-CEA vaccination indicated an increase in antivaccinia antibodies after the first vaccination (16). For this reason, CEA-specific immune response to subsequent vaccination with rV-CEA was most likely limited because of inhibition of virus replication. Preclinical studies in mice have shown that prior exposure to vaccinia virus did not diminish the development of human cytomegalovirus gB-specific immune responses after a single dose of ALVAC-gB injected i.p. or s.c. (23). New strategies involving diversified prime and boost protocols with rV-CEA, followed by boosting with ALVAC-CEA, may be more superior in the induction of CEA-specific immune responses (25). DCs are potent APCs that have been shown to stimulate both memory and naive T-cell responses in vitro(48). CEA-specific CTLs can be generated in vitro using peptide-pulsed DCs (47, 49). In addition,DCs infected with a poxvirus encoding MART-1/Melan A have been shown to sensitize T-cells in vitro (50). The use of DCs infected with ALVAC-CEA, perhaps in combination with costimulatory molecules and cytokines, may augment the CEA-specific immune responses and can be used to develop future immunization protocols.

The studies reported here demonstrate for the first time that a replication-defective avipox recombinant can be used to vaccinate cancer patients and elicit T-cell responses specific for a given tumor-associated antigen and epitope (CAP-1), which are capable of lysing human tumor cells expressing that antigen.