α fetoprotein (AFP)-derived peptide epitopes can be recognized by human T cells in the context of MHC class I. We determined the identity of AFP-derived peptides, presented in the context of HLA-A*0201, that could be recognized by the human (h) T cell repertoire. We screened 74 peptides and identified 3 new AFP epitopes, hAFP137–145, hAFP158–166, and hAFP325–334, in addition to the previously reported hAFP542–550. Each possesses two anchor residues and stabilized HLA-A*0201 on T2 cells in a concentration-dependent class I binding assay. The peptides were stable for 2–4 h in an off-kinetics assay. Each peptide induced peptide-specific T cells in vitro from several normal HLA-A*0201 donors. Importantly, these hAFP peptide-specific T cells also were capable of recognizing HLA-A*0201+/AFP+ tumor cells in both cytotoxicity assays and IFN-γ enzyme-linked immunospot assays. The immunogenicity of each peptide was tested in vivo with HLA-A*0201/Kb-transgenic mice. After immunization with each peptide emulsified in CFA, draining lymph node cells produced IFN-γ on recognition of cells stably transfected with hAFP. Furthermore, AFP peptide-specific T cells could be identified in the spleens of mice immunized with dendritic cells transduced with an AFP-expressing adenovirus (AdVhAFP). Three of four AFP peptides could be identified by mass spectrometric analysis of surface peptides from an HLA-A*0201 human hepatocellular carcinoma (HCC) cell line. Thus, compelling immunological and physiochemical evidence is presented that at least four hAFP-derived epitopes are naturally processed and presented in the context of class I, are immunogenic, and represent potential targets for hepatocellular carcinoma immunotherapy.

Hepatocellular carcinoma (HCC)3 has almost the lowest reported overall 5-year survival rate of all malignancies in the United States. The global annual incidence of this disease is 1.2 million. Current treatment options are largely surgical and of potential benefit to only a small percentage of patients. Because of the global pandemic of hepatitis B and C infections, the incidence of HCC is rapidly rising in both Asian and Western countries (1), and this trend is expected to continue for the next 50 years because of the long latency between infection and development of HCC. Therefore, novel treatment strategies are urgently needed for this disease.

Approximately 80% of HCC reactivate α fetoprotein (AFP) expression (2). The murine and human T cell repertoires can recognize AFP-derived peptide epitopes in the context of MHC class I (3, 4). Despite being exposed to high plasma levels of this oncofetal protein during embryonic development, clearly some AFP-specific T cells have not been deleted during the ontogeny of the immune system. Activation of these T cell clones can be achieved by presenting these epitopes in an immunostimulatory context, such as by professional Ag-presenting dendritic cells (DC). DC, when transduced with a recombinant adenovirus (AdV) vector encoding AFP, will process and present peptide epitopes in the context of MHC and will induce AFP-specific protection in mice (4).

We previously reported that hAFP542–550 (GVALQTMKQ) is naturally processed and presented in the context of HLA-A*0201, is recognized by the human T cell repertoire, and can be used to generate AFP-specific CTL in human T cell cultures and in HLA-A*0201/Kb-transgenic mice (3). Herein, we report a complete analysis of 74 AFP-derived peptides and identify 3 additional HLA-A*0201-restricted immunodominant epitopes as well as 10 potentially subdominant epitopes. These peptides have been characterized by both immunological and physicochemical methods. Our finding suggests that the expression of AFP by HCC cells can be exploited for immunotherapy strategies.

The University of Wisconsin (Madison, WI) Genetics Computer Group program “find patterns” was used to screen the hAFP sequences (GenBank accession numbers: J00077, J00076, and V01514) and identify 9- and 10-mer peptides, as described previously (3). The general HLA-A2.1 motif used was I, L, M, or V in P2 and V, I, or L in P9 (5).

HLA-A*0201 donors and cell lines were screened with the BB7.2 (HLA-A2) Ab and subtyped by PCR analysis by the University of California Los Angeles (UCLA) Tissue Typing Laboratory. HepG2, Hep3B, K562, B95-8, BB7.2, and W6/32 cell lines were obtained from American Type Culture Collection (Manassas, VA). The amount of AFP produced by HepG2 (606 ng/ml/106 cells/24 h) and Hep3B (100ng/ml/106 cells/24 h) was determined by the UCLA Clinical Labs. The M202 human melanoma cell line has been described previously (6). T2 cells were provided by Peter Cresswell (Yale University School of Medicine, New Haven, CT). JY cells (HLA-A*0201 homozygous) were provided by Martin Kast (Loyola University Cancer Center, Chicago, IL). Jurkat/A2Kb cells were provided by Linda Sherman (Scripps Research Institute, La Jolla, CA). EBV-transformed lymphoblastoid cell lines (LCL) were generated by incubating PBMC with supernatant from B95-8 cells. Cell lines were cultured in RPMI 1640 (Life Technologies, Gaithersburg, MD) or IMDM (JY cells; Life Technologies) with 10% FBS (Omega Scientific, Tarzana, CA) and PSF (penicillin/streptomycin/fungizone; Life Technologies). Anti-β2-microglobulin and the anti-CD4, CD14, CD19, and CD56 NA/LE Abs were obtained from BD PharMingen (San Diego, CA), anti-pan class I Ab was prepared from concentrated supernatant of W6/32 hybridoma, and CD4-FITC, CD8-PE, and CD16-PE Abs were obtained from Caltag (Burlingame, CA).

The stable transfectant cell lines LCL/hAFP, M202/hAFP, JurkatA2/Kb/hAFP (Jurkat/AFP), and JurkatA2/Kb/MART1 (Jurkat/MART) have been described previously (3). IL-2 was provided by Hoffman-LaRoche (Nutley, NJ), IL-7 was obtained from Biosource (Camarillo, CA), and keyhole limpet hemocyanin (KLH) and β2-microglobulin were obtained from Sigma (St. Louis, MO).

The AdVhAFP contains the 1.9-kb hAFP cDNA and is driven by the CMV promoter/enhancer in a pAC-CMVpLpA AdV type 5 backbone (3). The empty AdV vector, AdVRR5, has been described previously (7) and served as a control.

Peptides were synthesized by Chiron (Victoria, Australia) and at the UCLA Peptide Synthesis Facility (Dr. Joseph Reeve, Jr., Director) with standard F-moc technology.

T2 binding assay

Each peptide was tested for concentration-dependent binding to T2 cells in an HLA-A*0201 stabilization assay (8, 9). T2 (TAP deficient) cells, that had been incubated at room temperature the previous night to increase cell surface MHC class I molecule expression, were then incubated overnight with each peptide over a range of peptide concentrations from 0.1–100 μM. Stability of HLA-A*0201 was assayed by flow cytometry after staining the cells with anti-HLA-A2 Ab (BB7.2) and goat anti-mouse-FITC. The HLA-A*0201 strongly binding Flu matrix peptide (aa 58–66; Flu) was used as a positive control.

The off-kinetics assay was performed as described (3, 10). HLA-A*0201 LCL were stripped with a mild pH 3.2 citrate-phosphate acid buffer. Each peptide was immediately pulsed onto cells at 200 μM for 1 h in the presence of β2-microglobulin at 3μg/ml at room temperature. Excess peptide was washed off and the cells were incubated at 37°C for 0, 2, 4, and 6 h. Cells were washed at the end of each time point and stained for cell surface HLA-A2 expression and then analyzed by flow cytometry. The peptide-MHC class I complex was considered stable if the mean fluorescence intensity increased at least 1.5-fold from cells that were stripped but not pulsed with peptide.

Peptide-specific CTL were generated as described previously (3, 11, 12) to various peptides. Briefly, normal donor HLA-A*0201 PBMC were pulsed with peptide, washed, and cultured on day 0 with IL-7 (10–25 ng/ml) and KLH (5 μg/ml) in RPMI 1640/10% autologous serum at 3 × 106 cells/1.5 ml/well. Cells were restimulated weekly by removing the nonadherent cells from the culture and adding them to fresh, autologous, peptide-pulsed, washed, and irradiated PBMC at a 1:1 ratio. IL-2 was added twice weekly at 10 U/ml. After 3–4 wk of culture, the cultures were tested for cytotoxicity and/or cytokine release.

DC (prepared as described from PBMC incubated with GM-CSF and IL-4; Refs. 13 and 14) were transduced with AdVhAFP or AdVMART1 at a multiplicity of infection of 1000 for 2 h. Transduced DC were washed, irradiated, and plated at 1–2 × 105 cells/ml to serve as stimulators for CTL generation. Autologous nonadherent cells were depleted of CD4, CD14, CD19, and CD56+ cells by magnetic bead depletion (Dynal, Lake Success, NY) to prepare CD8+-enriched responder cells (population generally 80% CD8+, not shown). The CD8+ cells were plated with the transduced DC at 2 × 106 cells/ml in 5% autologous medium plus IL-7 at 10–25 ng/ml. Cultures were supplemented with IL-2 at 10 U/ml every 3–4 days. The CD8+ CTL were restimulated weekly with fresh, autologous AdV-transduced DC at a ratio of 1 DC to 10–20 CD8+ CTL. Most cultures were phenotyped for CD4+ and CD8+ cells on a weekly basis.

Chromium release assays were performed as described previously after 6 days of in vitro restimulation of primed splenocytes (3). Briefly, target cells were chromated (and T2 cells were peptide pulsed at 50μg/ml) for 1.5–2 h and plated. Target cells were washed three times, diluted to 5 × 104 (or 1 × 105 for murine assay) cells/ml, and plated with CTL. To control for nonspecific lysis, a 10–50 fold excess of unchromated K562 were added to target populations before adding to CTL (in human assay). After 4–5 h, supernatants were harvested and counted in a γ counter. Triplicate wells were averaged and the percentage of specific lysis was calculated as follows: (sample − spontaneous release)/(maximum release − spontaneous release).

To determine the frequency of Ag-specific cytokine-producing cells, the ELISPOT technique was used as described previously (3, 15, 16). T cell restimulation was performed with 2–5 × 106 CTL incubated with 1 × 105 autologous LCL or JY cells pulsed with specific or nonspecific peptides or with tumor cell lines. Restimulator cells without CTL served as a negative control. The capture Ab (BD PharMingen)-coated plates (Millipore, Bedford, MA) were incubated with restimulated cells (in duplicate at three dilutions) at 37°C for 24–48 h. The colored spots, representing cytokine-producing cells, were counted under a dissecting microscope.

HLA-A*0201/Kb-transgenic female mice (created by Dr. Linda Sherman, Scripps Research Institute) were originally purchased from Harlan Sprague-Dawley (Indianapolis, IN), and currently are bred by the animal facility of the Dept. of Radiation Oncology at UCLA and handled in accordance with the animal care policy of the UCLA. For peptide immunizations, mice received 100 μg of AFP or control peptide emulsified 1:1 in CFA (Difco, Detroit, MI) s.c..

DC were differentiated from murine bone marrow progenitor cells following the Inaba method (17) with modifications (3, 18). In vitro cultured DC were transduced in RPMI 1640/2% FCS at an multiplicity of infection of 100. Transduction was conducted for 2 h at 37°C, the DC then were washed and resuspended at 5 × 105 DC per 0.2 ml of PBS per animal for injection. In all cases, viability exceeded 95%.

Two weeks after priming (by AdV/DC or peptides), splenocytes (3 × 106/well) were activated ex vivo with irradiated, mitomycin C-treated Jurkat/hAFP or Jurkat/MART (5 × 105/well) in 2 ml RPMI 1640/10% FBS and 50 U/ml IL-2 in 24-well plates for 6 days.

Groups of HLA-A*0201/Kb mice were primed by AdV-transduced or peptide-pulsed DC immunizations, and 2 wk later, splenocytes (4 × 106 cells/ml) were activated ex vivo with an optimal concentration of peptide in complete medium (RPMI/10%FBS) for 48 h, or 5 × 106 cells/ml were cultured in T-25 flasks in an equal volume with irradiated Jurkat-MART or -AFP cells (1 × 105 cells/ml) plus 50 U/ml IL-2. After restimulation for 48 h, the ELISPOT assay was performed as described. The frequency of Ag-specific cells was determined from the difference between the number of spots seen with and without Ag during restimulation.

To elute peptides, HepG2 and Hep3B cells were washed three times with PBS before being incubated with 5 ml of citrate-phosphate buffer at pH 3.2 (19) for 1 min. The suspension was centrifuged (800 × g for 5 min) and a total of 500 ml of cell-free supernatant was collected for each cell line. The materials were lyophilized to dryness and stored at −20°C.

Lyophilized materials were redissolved in 30 ml of water/acetonitrile/triflouroacetic acid (W/A/TFA, 95/5/0.1 all by vol.). This solution was pumped onto an analytical reverse-phase HPLC column (C18 Betasil, 250 mm × 2 mm, 5-μm particle size, 100-Å pore size; Keystone Scientific, Bellefonte, PA) equilibrated in W/A/TFA, at a flow rate of 0.2 ml/min. The column was eluted by using an increasing linear gradient of 0.1% TFA in acetonitrile (time/% acetonitrile = 0/5, 5/5, 55/100, 60/100). Column eluate absorbance was monitored at 215 and 280 nm, and 1-min fractions were collected. The retention times of the synthetic peptides with amino acid sequences corresponding to the immunostimulatory peptides were obtained by using the same separation gradient on a separate column.

A Voyager-RP (PerSeptive Biosystems, Framingham, MA) matrix-assisted laser desorption ionization/time-of-flight (MALDI-TOF) instrument was used to acquire the mass spectra. The instrument uses a stainless-steel target on which the samples are deposited and dried. All spectra were externally calibrated with insulin, resulting in mass accuracy typically within ± 0.1%. Lyophilized HPLC fractions were resuspended in 10 μl of 70% acetonitrile with 0.1% TFA. A total of 1μl of this material was spotted along with 1 μl of the matrix α-cyano-4-hydroxycinnamic acid (Sigma, 10 mg/ml in 70% ACN/0.1% TFA). Spectra were obtained by scanning from m/z 500-7000.

We identified potential HLA-A*0201-restricted hAFP peptide epitopes by using a computer-based analysis of the human AFP amino acid sequence. A total of 74 9- and 10-mer peptides, the sequences of which conformed to the well characterized HLA-A*0201 binding motif (3, 20, 21, 22), were synthesized. Several immunological strategies were used to determine whether any of these peptides were immunogenic and naturally processed and presented in the context of HLA-A*0201. Fourteen of these AFP peptides were reproducibly positive in one or more of these assays and this subset of candidate epitopes is presented in Table I (and for comparison, 10 consistently negative hAFP peptides, and positive controls Flu M158–66 and MART-127–35). Their structural characteristics also are shown in this table and include length, number of anchor residues in P2 and P9/10 positions, concentration required for detectable binding to TAP-deficient T2 cells, and duration of binding to acid-stripped JY cells in an off-kinetics assay.

Table I.

Properties of peptides

AAaLengthbSequencecAnchordBindingeOff-KineticsfTg Mouse AdVhAFP/DCgAdVAFP RatiohHuman Peptide CTLiAFP+ TumorjAdVhAFP/huDCkTg Mouse pep/CFAl
Positive peptides            
MKWVESIFL 0.5 ++ 2.0 Yes Yes Yes Yes 
137 PLFQVPEPV 2.5 +++ 2.5 Yes Yes Yes Yes 
158 FMNKFIYEI 0.5 +++ 3.2 Yes Yes Yes Yes 
178 ILLWAARYD 100 >6 1.4 No No Yes Yes? 
218 LLNQHACAV 0.5 n.s. 1.9 Yes No No ND 
235 FQAITVTKL >100 >6 ++ 2.35 Yes Yes ND ND 
306 10 TTLERGQCII >100 <2 1.7 ND ND ND ND 
325 10 GLSPNLNRFL 10 +++ 4.4 Yes Yes Yes Yes 
485 CIRHEMTPV 25 ++ 2.1 ND ND ND ND 
492 PVNPGVGQC 100 ++ 2.67 Yes ND Yes 
507 10 NRRPCFSSLV >100 <2 1.7 ND ND ND ND 
542 GVALQTMKQ >100 +++ 3.0 Yes Yes Yes Yes 
547 10 TMKQEFLINL >100 1.5 Yes Yes? Yes? Yes? 
555 NLVKQKPQI >100 ++ 2.35 Yes Yes? Yes? Yes 
Negative peptides            
121 RHNCFLAHK >50 <2 0.7 ND ND ND ND 
211 KELRESSLL >100 1.0 No ND ND ND 
242 KLSQKFTKV 50 n.s. 1.1 ND ND ND ND 
257 KLVLDVAHV 0.5 n.s. 1.0 Yes No ND 
287 YICSQQDTL 50 1.3 No No ND ND 
294 TLSNKITEC 50 n.s. 1.2 No ND ND ND 
404 YIQESQALA 10 0.75 No ND No? ND 
441 10 QLTSSELMAI 50 <2 1.0 No ND ND ND 
549 KQEFLINLV >100 <2 Yes No No No 
599 LISKTRAAL 25 0.8 ND ND No ND 
Control peptides            
58 GILGFVFTL (FLUM1) 0.25 ND ND Yes ND ND ND 
27 AAGIGILTV (MART-1) 25 Yes No No No 
AAaLengthbSequencecAnchordBindingeOff-KineticsfTg Mouse AdVhAFP/DCgAdVAFP RatiohHuman Peptide CTLiAFP+ TumorjAdVhAFP/huDCkTg Mouse pep/CFAl
Positive peptides            
MKWVESIFL 0.5 ++ 2.0 Yes Yes Yes Yes 
137 PLFQVPEPV 2.5 +++ 2.5 Yes Yes Yes Yes 
158 FMNKFIYEI 0.5 +++ 3.2 Yes Yes Yes Yes 
178 ILLWAARYD 100 >6 1.4 No No Yes Yes? 
218 LLNQHACAV 0.5 n.s. 1.9 Yes No No ND 
235 FQAITVTKL >100 >6 ++ 2.35 Yes Yes ND ND 
306 10 TTLERGQCII >100 <2 1.7 ND ND ND ND 
325 10 GLSPNLNRFL 10 +++ 4.4 Yes Yes Yes Yes 
485 CIRHEMTPV 25 ++ 2.1 ND ND ND ND 
492 PVNPGVGQC 100 ++ 2.67 Yes ND Yes 
507 10 NRRPCFSSLV >100 <2 1.7 ND ND ND ND 
542 GVALQTMKQ >100 +++ 3.0 Yes Yes Yes Yes 
547 10 TMKQEFLINL >100 1.5 Yes Yes? Yes? Yes? 
555 NLVKQKPQI >100 ++ 2.35 Yes Yes? Yes? Yes 
Negative peptides            
121 RHNCFLAHK >50 <2 0.7 ND ND ND ND 
211 KELRESSLL >100 1.0 No ND ND ND 
242 KLSQKFTKV 50 n.s. 1.1 ND ND ND ND 
257 KLVLDVAHV 0.5 n.s. 1.0 Yes No ND 
287 YICSQQDTL 50 1.3 No No ND ND 
294 TLSNKITEC 50 n.s. 1.2 No ND ND ND 
404 YIQESQALA 10 0.75 No ND No? ND 
441 10 QLTSSELMAI 50 <2 1.0 No ND ND ND 
549 KQEFLINLV >100 <2 Yes No No No 
599 LISKTRAAL 25 0.8 ND ND No ND 
Control peptides            
58 GILGFVFTL (FLUM1) 0.25 ND ND Yes ND ND ND 
27 AAGIGILTV (MART-1) 25 Yes No No No 
a

Starting amino acid.

b

Length of peptide in amino acids.

c

Peptide sequence.

d

Number of anchor residues for HLA-A2.1.

e

Concentration of detectable binding to T2 cells in μM.

f

Length of time bound to JY cells. n.s., Not stable; <2, detected at time 0 only; 2, 4, 6, >6, length detected.

g

Recognition of peptide by AdVhAFP/DC-immunized HLA-A2.1/Kb-transgenic (tg) mouse splenocytes by IFN-γ ELISPOT; +++, highly reproducible and high frequency; ++, reproducible and frequency ratio 2.0 or greater; +, reproducible and somewhat weaker response; 0, no reproducible response.

h

Ratio of AFP peptide/control peptide IFN-γ spots from column 7.

i

Peptide-specific human CTL generated.

j

Recognition of AFP-expressing tumor target by peptide-specific human CTL.

k

Recognition of peptide by AdVhAFP/DC-stimulated human T cells by IFN-γ ELISPOT.

l

Recognition of AFP-expressing target by peptide/CFA-immunized tg mouse splenocytes by IFN-γ ELISPOT. ?, Weak response, not always reproducible.

HLA-A*0201/Kb-transgenic mice proved to be a convenient and sensitive means to screen these 74 potential peptide epitopes. DC were prepared from bone marrow progenitors by differentiation in GM-CSF and IL-4 and were transduced with a recombinant AdV vector that encoded hAFP (AdVhAFP). Splenocytes from HLA-A*0201/Kb mice immunized with AdVhAFP/DC recognized AFP-transfected targets in IFN-γ ELISPOT and cytotoxicity assays (Fig. 1, A and B, and data not shown). Over a dozen of the 74 AFP peptides also could be recognized by these primed splenocytes. Three peptides (hAFP137–145 (PLFQVPEPV), hAFP158–166 (FMNKFIYEI), and hAFP325–334 (GLSPNLNRFL)) and the previously reported hAFP542–550 (GVALQTMKQ; Ref. 3) consistently elicited cytokine production and cytotoxicity, and therefore are “immunodominant” (23). The remaining peptides had either weak or less reproducible responses but were positive in more than one type of assay and therefore are potential “subdominant” epitopes. These experiments indicated that the DC from these HLA-A*0201/Kb mice were capable of processing and presenting peptide epitopes in the context of HLA-A*0201, which elicited responses from the available murine T cell repertoire.

FIGURE 1.

AFP-specific CTL activity and cytokine release from AdVhAFP/DC or AdVMART1/DC immunization in A*0201/Kb-transgenic mice. A and B, Two weeks after immunization, splenocytes from mice were restimulated with Jurkat cells stably transfected with hAFP (Jurkat/AFP) or with MART-1 (Jurkat/MART). A, Frequency of AFP-specific vs MART-1-specific IFN-γ-release was determined by ELISPOT, p < 0.02, averages of three independent experiments. B, Cytotoxicity was assayed againstJurkat/AFP and Jurkat/MART in a 5-h 51Cr release assay. AdVhAFP/DC-immunized vs AdVMART1-immunized splenocyte recognition of Jurkat/AFP: E:T 1:1, p < 0.01; 11:1, p < 0.07; 33:1, p < 0.3; 100:1, p < 0.2. AdVMART1/DC recognition of Jurkat/MART vs Jurkat/AFP: E:T 1:1, p < 0.4; 11:1, p < 0.4; 33:1, p < 0.3; 100:1, p < 0.3. C, Two weeks after immunization, splenocytes from mice were restimulated with 30 μg of hAFP peptide, MART-127–35, or no peptide for 48 h before assaying cytokine secretion in an ELISPOT assay. Lower frequency of spots was observed in splenocytes from AdVMART-1/DC-immunized mice to the same hAFP peptides in the same experiments (not shown). Three similar experiments averaged is shown AFP137, AFP158, AFP325, and AFP542 vs MART-127 recognition p < 0.01, 0.01, 0.01, and 0.002, respectively. Replicate ELISPOT assays confirmed recognition of each AFP-derived peptide: AFP137, 8/8 experiments; AFP158, 4/4 experiments; AFP325, 6/6 experiments; AFP542, 18/18 experiments.

FIGURE 1.

AFP-specific CTL activity and cytokine release from AdVhAFP/DC or AdVMART1/DC immunization in A*0201/Kb-transgenic mice. A and B, Two weeks after immunization, splenocytes from mice were restimulated with Jurkat cells stably transfected with hAFP (Jurkat/AFP) or with MART-1 (Jurkat/MART). A, Frequency of AFP-specific vs MART-1-specific IFN-γ-release was determined by ELISPOT, p < 0.02, averages of three independent experiments. B, Cytotoxicity was assayed againstJurkat/AFP and Jurkat/MART in a 5-h 51Cr release assay. AdVhAFP/DC-immunized vs AdVMART1-immunized splenocyte recognition of Jurkat/AFP: E:T 1:1, p < 0.01; 11:1, p < 0.07; 33:1, p < 0.3; 100:1, p < 0.2. AdVMART1/DC recognition of Jurkat/MART vs Jurkat/AFP: E:T 1:1, p < 0.4; 11:1, p < 0.4; 33:1, p < 0.3; 100:1, p < 0.3. C, Two weeks after immunization, splenocytes from mice were restimulated with 30 μg of hAFP peptide, MART-127–35, or no peptide for 48 h before assaying cytokine secretion in an ELISPOT assay. Lower frequency of spots was observed in splenocytes from AdVMART-1/DC-immunized mice to the same hAFP peptides in the same experiments (not shown). Three similar experiments averaged is shown AFP137, AFP158, AFP325, and AFP542 vs MART-127 recognition p < 0.01, 0.01, 0.01, and 0.002, respectively. Replicate ELISPOT assays confirmed recognition of each AFP-derived peptide: AFP137, 8/8 experiments; AFP158, 4/4 experiments; AFP325, 6/6 experiments; AFP542, 18/18 experiments.

Close modal

These four putative hAFP epitopes then were subjected to additional immunological and physicochemical studies. These studies included: (1) repetitive peptide stimulation of naive HLA-A*0201 human T cell cultures, which would demonstrate peptide immunogenicity in the context of the human T cell repertoire and the ability of peptide-specific T cells to recognize AFP-transfected targets; (2) generation of a human T cell response in vitro by AdVhAFP transduced DC, which provides information about the spectrum of human T cell specificities to DC-processed AFP epitopes; (3) peptide/CFA immunization of HLA-A*0201/Kb mice, which is essentially an in vivo counterpart to peptide/human T cell cultures; and (4) HPLC and mass spectrometric identification of AFP peptides eluted from an HLA-A*0201 human HCC line. All of these lines of evidence, summarized in Table I, are formally presented for the four immunodominant epitopes we have identified.

HLA-A*0201/Kb mice were immunized with AdVhAFP/DC, and their splenocytes were restimulated in vitro with each AFP peptide separately and assayed for both cytotoxicity and IFN-γ production by ELISPOT assays. AdVhAFP/DC immunization induced AFP peptide-specific responses compared with a control A*0201-restricted melanoma Ag MART-127–35 peptide (Fig. 1 C). Conversely, AdVMART1/DC-immunized mice specifically recognized the immunodominant MART-127–35 peptide but not AFP-pulsed targets (not shown). Naive splenocytes showed no cytotoxicity and had few IFN-γ-producing cells (not shown). Responses to the previously identified peptide hAFP542–550 were reproducibly stronger than responses to the three new epitopes.

Synthetic peptides then were used to stimulate human T cells in vitro. Bulk T cell cultures were generated from PBMC pulsed with each AFP-derived peptide (supplemented with KLH, IL-7, and IL-2) and were tested between weeks 3 and 7 of expansion for the ability to recognize both peptide-pulsed and AFP-expressing targets (Fig. 2). These cultures expanded peptide-specific T cells (Fig. 2,A), as evidenced by the ability to secrete IFN-γ on recognition of specific peptide-pulsed JY cells and not control MART-127–35-pulsed JY in the ELISPOT assay. The AFP peptide-specific bulk T cells also recognized both AFP-stably transfected and AdVhAFP-transduced HLA-A*0201 melanoma cells (M202) compared with unmodified or empty AdVRR5-transduced parental cells (Fig. 2,B) as shown by an increased frequency of IFN-γ-producing AFP-specific T cells. In cytotoxicity assays, these bulk AFP peptide-stimulated T cells were able to lyse the AdVhAFP-transduced melanoma cells (Fig. 2,C). To assess the ability to recognize the HLA-A*0201+, naturally AFP-expressing HCC cell line HepG2 (compared with the HLA-A2-/AFP+ HCC line Hep3B), both cytotoxicity (Fig. 2,C) and ELISPOT assays (Fig. 2 D) were preformed. Levels of cytotoxicity were lower than for AFP-transduced melanoma cells, but in general, frequencies of IFN-γ-producing T cells were similar for the HCC and transfected/transduced melanoma cells.

FIGURE 2.

Peptide and AFP-specificity of hAFP peptide lymphocyte cultures. A, PBMC were stimulated weekly with different hAFP peptides and assayed (after 4–5 wk of expansion) for the frequency of IFN-γ-producing T cells after restimulation with peptide-pulsed JY cells to confirm peptide-specificity. AFP137, AFP158, AFP325, and AFP542 recognition of AFP vs MART-1, p < 0.04, 0.05, 0.02, and 0.01, respectively. B, hAFP peptide-stimulated bulk T cell cultures were assayed for frequency of AFP-specific IFN-γ-secreting cells by ELISPOT after restimulation with AFP-stable transfectant (M202/hAFPhyg), AdVhAFP-transduced (M202AdVhAFP), control AdVRR5-transduced (M202RR5), or untransfected parental HLA-A*0201+M202 melanoma cells. AFP542 recognition of M202hAFP vs parental cells, p < 0.01; M202AdVhAFP vs M202AdVRR5, p < 0.003. These trends for all four of the peptides were reproducible in three additional experiments. C, Cytotoxicity of hAFP peptide cell cultures against HepG2 (AFP+/A*0201+) and Hep3B cells (AFP+/A*0201) or M202AdVhAFP- and M202AdVRR5-transduced cells. AFP137, AFP158, AFP325, AFP542, and MART-127 T cell recognition of HepG2 vs Hep3B targets at E:T = 50:1; p < 0.008, 0.001, 0.04, 0.003, and 0.57, respectively. AFP137, AFP158, AFP325, AFP542, and MART-127 T cell recognition of or M202AdVhAFP vs M202AdVRR5 targets at E:T = 50:1; p < 0.17, 0.08, 0.10, 0.06, and 0.73, respectively. D, Frequency of HepG2 and Hep3B-reactive IFN-γ-producing cells in peptide-stimulated cultures AFP137, AFP158, AFP325, AFP542, and MART-127 T cell recognition of HepG2 vs Hep3B cells: p < 0.01, 0.0001, 0.04, 0.006, and 0.34, respectively. Overall, AFP peptide-stimulated T cell cultures recognized AFP-expressing, HLA-A*0201+ targets in eight IFN-γ ELISPOT assays and five cytotoxicity assays.

FIGURE 2.

Peptide and AFP-specificity of hAFP peptide lymphocyte cultures. A, PBMC were stimulated weekly with different hAFP peptides and assayed (after 4–5 wk of expansion) for the frequency of IFN-γ-producing T cells after restimulation with peptide-pulsed JY cells to confirm peptide-specificity. AFP137, AFP158, AFP325, and AFP542 recognition of AFP vs MART-1, p < 0.04, 0.05, 0.02, and 0.01, respectively. B, hAFP peptide-stimulated bulk T cell cultures were assayed for frequency of AFP-specific IFN-γ-secreting cells by ELISPOT after restimulation with AFP-stable transfectant (M202/hAFPhyg), AdVhAFP-transduced (M202AdVhAFP), control AdVRR5-transduced (M202RR5), or untransfected parental HLA-A*0201+M202 melanoma cells. AFP542 recognition of M202hAFP vs parental cells, p < 0.01; M202AdVhAFP vs M202AdVRR5, p < 0.003. These trends for all four of the peptides were reproducible in three additional experiments. C, Cytotoxicity of hAFP peptide cell cultures against HepG2 (AFP+/A*0201+) and Hep3B cells (AFP+/A*0201) or M202AdVhAFP- and M202AdVRR5-transduced cells. AFP137, AFP158, AFP325, AFP542, and MART-127 T cell recognition of HepG2 vs Hep3B targets at E:T = 50:1; p < 0.008, 0.001, 0.04, 0.003, and 0.57, respectively. AFP137, AFP158, AFP325, AFP542, and MART-127 T cell recognition of or M202AdVhAFP vs M202AdVRR5 targets at E:T = 50:1; p < 0.17, 0.08, 0.10, 0.06, and 0.73, respectively. D, Frequency of HepG2 and Hep3B-reactive IFN-γ-producing cells in peptide-stimulated cultures AFP137, AFP158, AFP325, AFP542, and MART-127 T cell recognition of HepG2 vs Hep3B cells: p < 0.01, 0.0001, 0.04, 0.006, and 0.34, respectively. Overall, AFP peptide-stimulated T cell cultures recognized AFP-expressing, HLA-A*0201+ targets in eight IFN-γ ELISPOT assays and five cytotoxicity assays.

Close modal

Because AdVhAFP/DC in vitro stimulated human T cells specifically recognized hAFP-transfected targets in both CTL and ELISPOT assays (3), we determined whether these four AFP peptides were specifically recognized by the AdVhAFP/DC stimulated T cells. After 7–21 days of culture, CD8-enriched T cells stimulated weekly with AdVhAFP/DC were tested for both cytotoxicity (Fig. 3,A) and the frequency of hAFP peptide-specific IFN-γ cytokine-producing cells (Fig. 3 B). AdVhAFP/DC T cell cultures were cytotoxic for JY cells pulsed with each of the four AFP peptides, indicating that CTL to these peptides could be expanded from peripheral blood of normal donors. After restimulation with autologous peptide-pulsed LCL or JY cells, these bulk cultures also contained a much higher frequency of IFN-γ-secreting cells specific for AFP peptides compared with MART-127–35, indicating that, in addition to hAFP542–550, these three peptides also are naturally processed and presented by AdVhAFP-transduced DC.

FIGURE 3.

Cytotoxicity and cytokine release by AdVhAFP-transduced human DC/T cell cultures. A, CD8+-enriched lymphocyte cultures were stimulated with AdVhAFP/DC for 2 wk and assessed for peptide-specific cytotoxicity against AFP peptides or MART-127–35-pulsed JY cell targets. AFP137, AFP158, AFP325, and AFP542 vs MART-127 T cell recognition p < 0.001, 0.02, 0.004, and 0.005, respectively, at E:T = 50:1; p < 0.01, 0.01, 0.07, and 0.0001, respectively, at E:T = 17:1; and p < 0.03, 0.04, 0.2, and 0.1, respectively, at E:T = 5:1. B, AdVhAFP/DC T cell cultures were assayed for frequency of IFN-γ-secreting cells by ELISPOT at 21 days of culture after restimulation with peptide-pulsed autologous LCL cells. AFP137, AFP158, AFP325, and AFP542 recognition vs MART-1, p < 0.0007, 0.001, 0.08, and 0.009, respectively. Overall, these AdVhAFP/DC-stimulated T cell cultures were tested for reactivity to these four AFP peptides and controls in six IFN-γ ELISPOT assays and four cytotoxicity assays with similar results. C, AdVhAFP/DC T cell cultures were assayed for frequency of IFN-γ-, TNF-α-, IL-4-, and IL-10-secreting cells by ELISPOT at 14 days of culture after restimulation with the human HCC lines HepG2 (HLA-A*0201+/AFP+) and Hep3B (HLA-A*0201/AFP+). AFP-specific T cells secreting IFN-γ to HepG2 vs Hep3B cells, p < 0.02; T cells secreting TNF-α to HepG2 vs Hep3B cells, p < 0.002; to IL-4, p < 0.2; and to IL-10, p < 0.09. IL-10 synthesis by HepG2 or Hep3B cells alone vs by HCC cells with AFP-specific T cells, p < 0.6 and 0.8, respectively.

FIGURE 3.

Cytotoxicity and cytokine release by AdVhAFP-transduced human DC/T cell cultures. A, CD8+-enriched lymphocyte cultures were stimulated with AdVhAFP/DC for 2 wk and assessed for peptide-specific cytotoxicity against AFP peptides or MART-127–35-pulsed JY cell targets. AFP137, AFP158, AFP325, and AFP542 vs MART-127 T cell recognition p < 0.001, 0.02, 0.004, and 0.005, respectively, at E:T = 50:1; p < 0.01, 0.01, 0.07, and 0.0001, respectively, at E:T = 17:1; and p < 0.03, 0.04, 0.2, and 0.1, respectively, at E:T = 5:1. B, AdVhAFP/DC T cell cultures were assayed for frequency of IFN-γ-secreting cells by ELISPOT at 21 days of culture after restimulation with peptide-pulsed autologous LCL cells. AFP137, AFP158, AFP325, and AFP542 recognition vs MART-1, p < 0.0007, 0.001, 0.08, and 0.009, respectively. Overall, these AdVhAFP/DC-stimulated T cell cultures were tested for reactivity to these four AFP peptides and controls in six IFN-γ ELISPOT assays and four cytotoxicity assays with similar results. C, AdVhAFP/DC T cell cultures were assayed for frequency of IFN-γ-, TNF-α-, IL-4-, and IL-10-secreting cells by ELISPOT at 14 days of culture after restimulation with the human HCC lines HepG2 (HLA-A*0201+/AFP+) and Hep3B (HLA-A*0201/AFP+). AFP-specific T cells secreting IFN-γ to HepG2 vs Hep3B cells, p < 0.02; T cells secreting TNF-α to HepG2 vs Hep3B cells, p < 0.002; to IL-4, p < 0.2; and to IL-10, p < 0.09. IL-10 synthesis by HepG2 or Hep3B cells alone vs by HCC cells with AFP-specific T cells, p < 0.6 and 0.8, respectively.

Close modal

The AdVhAFP/DC-stimulated T cell cultures also had a low frequency of cytokine-producing cells that recognized the A*0201+/AFP+ HCC line HepG2 but not the A*0201/AFP+ HCC line Hep3B, as shown in the ELISPOT in Fig. 3 C. T lymphocytes synthesizing the Th1 cytokines IFN-γ and TNF-α were detected, whereas the Th2 cytokine IL-4 was not detected. IL-10 also was detected when the HCC lines were plated without T cells, indicating that production of this cytokine was tumor cell derived.

Cytotoxicity against hAFP137–145- and hAFP325–334-pulsed targets was generally more difficult to detect than against hAFP158–166- and hAFP542–550-pulsed targets by chromium release assay. Similarly, a lower frequency of IFN-γ-producing cells were obtained for these two peptides.

To confirm the in vivo immunogenicity of these four hAFP peptides, HLA-A*0201/Kb mice were immunized with each AFP peptide pulsed onto syngeneic DC. IFN-γ-specific ELISPOT assays were performed with splenocytes restimulated in vitro with either the immunizing AFP peptide (or MART-1 peptide, Fig. 4,A) or with Jurkat/AFP or Jurkat/MART transfected cell lines (Fig. 4,B). Immunization with each hAFP peptide and subsequent restimulation with either peptide or Jurkat/AFP induced large numbers of AFP-specific IFN-γ-producing cells. Lymphocytes from PBS-injected mice showed neither cytotoxicity nor IFN-γ production regardless of restimulation (Fig. 4 and data not shown). Mice immunized with MART-127–35 peptide produced MART-1-specific responses but no AFP peptide responses.

FIGURE 4.

AFP-specific cytokine release from hAFP peptide-pulsed DC-immunized transgenic mice. Increased frequency of (A) AFP peptide-specific and (B) AFP Ag-specific IFN-γ production was detected in splenocytes from A*0201/Kb-transgenic mice immunized with each hAFP peptide (or MART-127) on DC. Two weeks after immunization, splenocytes from mice were restimulated with either the immunizing peptide (A) or J-AFP or J-MART (B) for 48 h before assaying cytokine secretion in an ELISPOT assay. Two similar experiments averaged is shown. Splenocyte recognition of immunizing AFP137, AFP158, AFP325, and AFP542 vs MART-127, p < 0.005, 0.0001, 0.001, and 0.001, respectively. Splenocyte recognition of Jurkat/AFP vs Jurkat/MART p < 0.004, 0.003, 0.0002, and 0.008, respectively. MART-127 immunized splenocyte recognition of MART-127 vs AFP542, p < 0.001; Jurkat/MART vs Jurkat/AFP, p < 0.002.

FIGURE 4.

AFP-specific cytokine release from hAFP peptide-pulsed DC-immunized transgenic mice. Increased frequency of (A) AFP peptide-specific and (B) AFP Ag-specific IFN-γ production was detected in splenocytes from A*0201/Kb-transgenic mice immunized with each hAFP peptide (or MART-127) on DC. Two weeks after immunization, splenocytes from mice were restimulated with either the immunizing peptide (A) or J-AFP or J-MART (B) for 48 h before assaying cytokine secretion in an ELISPOT assay. Two similar experiments averaged is shown. Splenocyte recognition of immunizing AFP137, AFP158, AFP325, and AFP542 vs MART-127, p < 0.005, 0.0001, 0.001, and 0.001, respectively. Splenocyte recognition of Jurkat/AFP vs Jurkat/MART p < 0.004, 0.003, 0.0002, and 0.008, respectively. MART-127 immunized splenocyte recognition of MART-127 vs AFP542, p < 0.001; Jurkat/MART vs Jurkat/AFP, p < 0.002.

Close modal

We then sought physicochemical evidence for their presence on the surface of a human HCC line. We performed MALDI-TOF mass spectrometry to analyze HPLC-fractionated peptides acid-eluted from the AFP-producing HCC cell lines HepG2 (HLA-A2+) and Hep3B (HLA-A2). We identified potential mass candidates for three of the four peptides in the HPLC-fractionated peptide pool eluted from the HLA-A*0201+ HepG2 cells but not from the HLA-A*0201 Hep3B cells.

The MALDI-TOF analysis of the HPLC fractions established that almost all fractions contained up to 20 different peptides in the mass range from 700 to 1500 Da, although frequently with a few dominating signals. Out of this complex mixture, we identified peaks with m/z values corresponding to the calculated monoisotropic protonated molecules ((M+H)+) of hAFP542–550, hAFP158–166, and hAFP325–334 in the peptide pool eluted from HepG2 cells (Table II and Fig. 5). A peptide of m/z 975.6 was identified in HPLC fraction 21 from the HepG2 peptide pool (Fig. 5,A). The calculated (M+H)+ of hAFP542–550 was 975.5, and the retention time of the synthetic peptide with amino acids corresponding to hAFP542–550 was 21.2 min. Furthermore, no signal at m/z of 975.5 ± 1 was observed in samples with matrix alone (data not shown) and in HPLC fractions 18–22 from the Hep3B elution (fraction 21 shown in Fig. 5,A). Similarly, peaks with m/z corresponding to the calculated (M+H)+ of hAFP158–166 and hAFP325–334 also were found in the appropriate fraction (28) derived from HepG2 predicted from the behavior of the standard peptides (Table II and Fig. 5, B and D). These peaks were absent in fractions 26–32 in the peptide pool eluted from Hep3B (fraction 28 shown in Fig. 5). A peak at 1152.2 m/z was observed in fraction 28, suggesting the presence of the sodium adduct of hAFP325–334 (Fig. 5 D).

Table II.

Summary of the MALDI-TOF mass analysis

PeptideaHPLC Retention Time of Synthetic Peptide (min)bCalculated (M+H)+cObserved (M+H)+ in HepG2dObserved (M+H)+ in Hep3Be(M+H)+ Identified in HPLC FractionfImmunologically Reactive Fractions (IFNγ ELISPOT)g
HepG2Hep3B
hAFP542–550 21.2 975.5 975.6 None 21 20, 21 
hAFP158–166 28.9 1204.6 1204.9 None 28 27, 28, 29 
hAFP137–145 28.1 1025.6 None None – 27, 28, 29 
hAFP325–334 27.7 1130.6 1130.1 None 28 27, 28, 29 
PeptideaHPLC Retention Time of Synthetic Peptide (min)bCalculated (M+H)+cObserved (M+H)+ in HepG2dObserved (M+H)+ in Hep3Be(M+H)+ Identified in HPLC FractionfImmunologically Reactive Fractions (IFNγ ELISPOT)g
HepG2Hep3B
hAFP542–550 21.2 975.5 975.6 None 21 20, 21 
hAFP158–166 28.9 1204.6 1204.9 None 28 27, 28, 29 
hAFP137–145 28.1 1025.6 None None – 27, 28, 29 
hAFP325–334 27.7 1130.6 1130.1 None 28 27, 28, 29 
a

Peptide identification.

b

Typical HPLC retention time of control synthetic peptide.

c

Expected mass/charge measurement.

d

Observed mass/charge measurement in acid-eluted peptides from HepG2.

e

Observed mass/charge measurement in acid-eluted peptides from Hep3B.

f

Fraction (min) of observed hAFP peptide mass/charge measurement in acid-eluted peptides from HepG2.

g

Fractions containing peptides capable of restimulating AdVhAFP/DC primed splenocytes by IFN-γ ELISPOT.

FIGURE 5.

Mass spectra of HPLC-fractionated peptide pools acid eluted from HepG2 and Hep3B cells showing the m/z regions ((M+H)+ ± 200) of (A) hAFP542–550, (B) hAFP158–166, (C) hAFP137–145, and (D) hAFP325–334. The spectra were normalized to approximately the same level of background noise.

FIGURE 5.

Mass spectra of HPLC-fractionated peptide pools acid eluted from HepG2 and Hep3B cells showing the m/z regions ((M+H)+ ± 200) of (A) hAFP542–550, (B) hAFP158–166, (C) hAFP137–145, and (D) hAFP325–334. The spectra were normalized to approximately the same level of background noise.

Close modal

In the three peptides that were identified, the peaks were observed in repeated scanning of the spotted samples. A board peak at m/z 1020.9 was observed in fraction 29 from the HepG2 peptide pool (Fig. 5 C). Because this peak was beyond the margin of error tolerated by this physicochemical analysis, we concluded that we could not document the presence of hAFP137–145 on the surface of HepG2 cells.

To confirm the presence of AFP peptides in these fractions immunologically, 1 μl of each HPLC fraction from either HepG2 or Hep3B cells was used to restimulate AdVhAFP/DC-immunized murine splenocytes in an ELISPOT assay (Table II). A total of 200–250 spots/106 cells were observed from fractions containing AFP immunodominant peptides, whereas 100–130 spots/106 cells were observed from the other fractions, and a maximum of 50 spots/106 cells were observed from Hep3B fractions. This further supports the mass spectrometry identification of these AFP peptides.

The oncofetal protein AFP is transcriptionally derepressed in HCC. We have demonstrated previously that neither the murine nor human T cell repertoire are tolerant of AFP-reactive T cells despite having been exposed to high serum concentrations of this protein during embryogenesis. Rather, the immune system appears to be clonally ignorant of AFP unless it is presented in an immunostimulatory context. We report herein a comprehensive screening of the human AFP protein sequence for immunogenic epitopes and specifically characterize three additional HLA-A*0201-restricted peptide epitopes recognized by AFP-specific T cells, the 9-mer hAFP137–145 (PLFQVPEPV), and hAFP158–166 (FMNKFIYEI) and the 10-mer hAFP325–334 (GLSPNLNRFL). These results were obtained by using immunological and physicochemical approaches. DC genetically engineered to express hAFP were able to stimulate AFP peptide-specific T cell responses in CD8-enriched human lymphocyte cultures. Similarly, immunized HLA-A*0201/Kb mice also recognized AFP peptide-pulsed cells in cytokine release assays. We also showed that AFP peptide-stimulated human and HLA-A*0201/Kb mouse T cell responses recognized hAFP-engineered targets and, to a lesser extent, naturally AFP-expressing human HCC cells. Finally, mass spectrometry was used to identify at least three AFP epitopes from complex mixtures of peptides eluted from HLA-A*0201+ HCC cells. Thus, we provide multiple lines of evidence that each of these four AFP peptides is immunogenic and naturally processed and presented in the context of HLA-A*0201.

The human AFP protein has 65% amino acid homology to murine AFP. Therefore, the T cells in these transgenic mice recognize hAFP as a xenoantigen in the context of the human HLA-A*0201. For the four hAFP immunodominant peptides, there are 3, 2, 4, and 1 amino acid differences, respectively, between the human and murine 137, 158, 325, and 542 peptide sequences. Although studies by others have demonstrated that A*0201 mice respond to largely the same set of HLA-A2.1-binding peptides as humans (24, 25), we performed the human T cell culture assays to support the human immunogenicity of each peptide. Moreover, we have demonstrated previously that mAFP is immunogenic in normal inbred mice if presented in the immunogenic context of engineered DC (3).

Each of these three newly reported peptides has two anchor residues and strong binding affinity to HLA-A*0201. Molecular modeling of hAFP137–145 (PLFQVPEPV), hAFP158–166 (FMNKFIYEI), and hAFP325–334 (GLSPNLNRFL) in the HLA-A2 binding groove show that the peptides adopt the general backbone conformation seen in the x-ray structures of HLA-A2 (26, 27): the NH2 and COOH termini of peptides are accommodated by conserved pockets at both ends of the binding groove, whereas the backbone in the central portion of the peptides bulges upward toward the TCR. It should be noted that the proline residues at positions 6 and 8 in hAFP137–145 (PLFQVPEPV) impose a conformation of the peptide that is more rigid than the other peptides in this study. As expected, the side chains of the anchor residues at positions 2 and 9 (10 in hAFP325–334) of the peptides are buried in the deep polymorphic pockets of HLA-A2 (pockets B and F, respectively). Side chains in the central part of the peptide, both hydrophilic and hydrophobic, tend to project upward toward the TCR (K4, F5, Y7, and E8 in hAFP158–166 (FMNKFIYEI); Q4, V5, and E7 in hAFP137–145 (PLFQVPEPV); N5, L6, N7, R8, and F9 in hAFP325–334 (GLSPNLNRFL)). These contrast with hAFP542–550, which lacks a hydrophobic side chain for pocket F.

It proved to be important to screen candidate peptides with fewer than two anchor residues, as 8 of 14 immunogenic peptides had one anchor residue. Although strong binding, or “on-rate” (10 μM or less) was a characteristic of 3 of 4 immunodominant peptides, the fourth immunodominant peptide (hAFP542–550) and 8 of 14 immunogenic peptides overall showed weak (>50 μM) binding. Of the physical characteristics assessed, off-kinetics proved to correlate best with immunogenicity with 9 of 14 immunogenic peptides having off-kinetics greater than 2 h and 5 of 14 greater than 4 h. Of the five peptides with 6 h or longer kinetics, all were immunogenic. A sufficiently long half-life is likely to facilitate the stimulation of many TCRs on a given T cell, which is thought to be important for strong T cell activation (28, 29). Thus, it is possible that additional immunogenic epitopes exist from Ags for which candidate determinants were screened based on only two anchor residues or strong binding affinity (30, 31, 32).

The three new immunodominant and the 10 potentially subdominant peptides described in this report may serve as important tools for direct peptide-based immunotherapy as well as analysis of the diversity of the response to whole AFP Ag-based immunotherapy. We have initiated a phase I/II clinical immunotherapy trial in which A*0201/AFP-positive patients with unresectable HCC are immunized with all four of these AFP immunodominant peptides emulsified in IFA. Our early observations are that these patients are capable of generating peripheral T cell responses to all four peptides as measured by tetramer and ELISPOT assays (data not shown). Together, this data confirms that AFP can serve as a tumor-specific Ag in HCC and may be a suitable target for T cell-mediated immunotherapy strategies.

1

This work was supported by National Institutes of Health/National Cancer Institute Grants ROI CA 77623, RO1 CA 79976, T32 CA75956, and K12 CA 79605, the Monkarsh Fund, Naify Fund, and the Stacy and Evelyn Kesselman Research Fund (all to J.S.E.), Tumor Immunology Training Grant CA 09120 (to C.M.V.), and a Jonsson Cancer Center Fellowship (to W.S.M.).

3

Abbreviations used in this paper: HCC, hepatocellular carcinoma; AFP, α fetoprotein; h, human; DC, dendritic cells; AdV, adenovirus; ELISPOT, enzyme-linked immunospot; LCL, lymphoblastoid cell lines; KLH, keyhole limpet hemocyanin; TFA, trifluoroacetic acid; MALDI-TOF, matrix-assisted laser desorption ionization/time-of-flight.

1
Schafer, D. F., M. F. Sorrell.
1999
. Hepatocellular carcinoma.
Lancet
353
:
1253
2
Kirkwood, J. M., M. T. Lotze, J. M. Yasko.
1994
.
Current Cancer Therapeutics
Current Medicine, Philadelphia.
3
Butterfield, L. H., A. Koh, W. Meng, C. M. Vollmer, A. Ribas, V. B. Dissette, E. Lee, J. A. Glaspy, W. H. McBride, J. S. Economou.
1999
. Generation of human T cell responses to an HLA-A2.1-restricted peptide epitope from α fetoprotein.
Cancer Res.
59
:
3134
4
Vollmer, C. M., Jr, F. C. Eilber, L. H. Butterfield, A. Ribas, V. B. Dissette, A. Koh, L. D. Montejo, M. C. Lee, K. J. Andrews, W. H. McBride, J. A. Glaspy, J. S. Economou.
1999
. α-fetoprotein-specific genetic immunotherapy for hepatocellular carcinoma.
Cancer Res.
59
:
3064
5
Celis, E., J. Fikes, P. Wentworth, J. Sidney, S. Southwood, A. Maewal, M. F. Del Guercio, A. Sette, B. Livingston.
1994
. Identification of potential CTL epitopes of tumor-associated antigen MAGE-1 for five common HLA-A alleles.
Mol. Immunol.
31
:
1423
6
Butterfield, L. H., T. C. Stoll, R. Lau, J. S. Economou.
1997
. Cloning and analysis of MART-1/Melan-A human melanoma antigen promoter regions.
Gene
191
:
129
7
Toloza, E. M., K. Hunt, S. Swisher, W. McBride, R. Lau, S. Pang, K. Rhoades, T. Drake, A. Belldegrun, J. Glaspy, J. S. Economou.
1996
. In vivo cancer gene therapy with a recombinant interleukin-2 adenovirus vector.
Cancer Gene Ther.
3
:
11
8
Stuber, G., G. H. Leder, W. T. Storkus, M. T. Lotze, S. Modrow, L. Szekely, H. Wolf, E. Klein, K. Karre, G. Klein.
1994
. Identification of wild-type and mutant p53 peptides binding to HLA-A2 assessed by a peptide loading-deficient cell line assay and a novel major histocompatibility complex class I peptide binding assay.
Eur. J. Immunol.
24
:
765
9
Nijman, H. W., J. G. Houbiers, M. P. Vierboom, S. H. van der Burg, J. W. Drijfhout, D. A. J, P. Kenemans, C. J. Melief, W. M. Kast.
1993
. Identification of peptide sequences that potentially trigger HLA-A2.1-restricted cytotoxic T lymphocytes.
Eur. J. Immunol.
23
:
1215
10
van der Burg, S. H., E. Ras, J. W. Drijfhout, W. E. Benckhuijsen, A. J. Bremers, C. J. Melief, W. M. Kast.
1995
. An HLA class I peptide-binding assay based on competition for binding to class I molecules on intact human B cells: identification of conserved HIV-1 polymerase peptides binding to HLA-A*0301.
Hum. Immunol.
44
:
189
11
Plebanski, M., C. E. Allsopp, M. Aidoo, H. Reyburn, A. V. Hill.
1995
. Induction of peptide-specific primary cytotoxic T lymphocyte responses from human peripheral blood.
Eur. J. Immunol.
25
:
1783
12
Butterfield, L. H., S. M. Jilani, N. G. Chakraborty, L. A. Bui, A. Ribas, V. B. Dissette, R. Lau, S. C. Gamradt, J. A. Glaspy, W. H. McBride, B. Mukherji, J. S. Economou.
1998
. Generation of melanoma-specific cytotoxic T lymphocytes by dendritic cells transduced with a MART-1 adenovirus.
J. Immunol.
161
:
5607
13
Arthur, J. F., L. H. Butterfield, M. D. Roth, L. A. Bui, S. M. Kiertscher, R. Lau, S. Dubinett, J. Glaspy, W. H. McBride, J. S. Economou.
1997
. A comparison of gene transfer methods in human dendritic cells.
Cancer Gene Ther.
4
:
17
14
Romani, N., S. Gruner, D. Brang, E. Kampgen, A. Lenz, B. Trockenbacher, G. Konwalinka, P. O. Fritsch, R. M. Steinman, G. Schuler.
1994
. Proliferating dendritic cell progenitors in human blood.
J. Exp. Med.
180
:
83
15
Mayer, S., C. Scheibenbogen, K. H. Lee, W. Keilholz, S. Stevanovic, H. G. Rammensee, U. Keilholz.
1996
. A sensitive proliferation assay to determine the specific T cell response against HLA-A2.1-binding peptides.
J. Immunol. Methods
197
:
131
16
Herr, W., J. Schneider, A. W. Lohse, K. H. Meyer zum Buschenfelde, T. Wolfel.
1996
. Detection and quantification of blood-derived CD8+ T lymphocytes secreting tumor necrosis factor alpha in response to HLA-A2.1-binding melanoma and viral peptide antigens.
J. Immunol. Methods
191
:
131
17
Inaba, K., M. Inaba, N. Romani, H. Aya, M. Deguchi, S. Ikehara, S. Muramatsu, R. M. Steinman.
1992
. Generation of large numbers of dendritic cells from mouse bone marrow cultures supplemented with granulocyte/macrophage colony-stimulating factor.
J. Exp. Med.
176
:
1693
18
Ribas, A., L. H. Butterfield, W. H. McBride, S. M. Jilani, L. A. Bui, C. M. Vollmer, R. Lau, V. B. Dissette, B. Hu, A. Y. Chen, J. A. Glaspy, J. S. Economou.
1997
. Genetic immunization for the melanoma antigen MART-1/Melan-A using recombinant adenovirus-transduced murine dendritic cells.
Cancer Res.
57
:
2865
19
Storkus, W. J., H. J. d. Zeh, R. D. Salter, M. T. Lotze.
1993
. Identification of T-cell epitopes: rapid isolation of class I-presented peptides from viable cells by mild acid elution.
J. Immunother.
14
:
94
20
Falk, K., O. Rotzschke.
1993
. Consensus motifs and peptide ligands of MHC class I molecules.
Semin. Immunol.
5
:
81
21
Rammensee, H. G., K. Falk, O. Rotzschke.
1993
. MHC molecules as peptide receptors.
Curr. Opin. Immunol.
5
:
35
22
Ruppert, J., J. Sidney, E. Celis, R. T. Kubo, H. M. Grey, A. Sette.
1993
. Prominent role of secondary anchor residues in peptide binding to HLA-A2.1 molecules.
Cell
74
:
929
23
Sercarz, E. E., P. V. Lehmann, A. Ametani, G. Benichou, A. Miller, K. Moudgil.
1993
. Dominance and crypticity of T cell antigenic determinants.
Annu. Rev. Immunol.
11
:
729
24
Wentworth, P. A., A. Vitiello, J. Sidney, E. Keogh, R. W. Chesnut, H. Grey, A. Sette.
1996
. Differences and similarities in the A2.1-restricted cytotoxic T cell repertoire in humans and human leukocyte antigen-transgenic mice.
Eur. J. Immunol.
26
:
97
25
Vitiello, A., D. Marchesini, J. Furze, L. A. Sherman, R. W. Chesnut.
1991
. Analysis of the HLA-restricted influenza-specific cytotoxic T lymphocyte response in transgenic mice carrying a chimeric human-mouse class I major histocompatibility complex.
J. Exp. Med.
173
:
1007
26
Fremont, D. H., M. Matsumura, E. A. Stura, P. A. Peterson, I. A. Wilson.
1992
. Crystal structures of two viral peptides in complex with murine MHC class I H-2Kb.
Science
257
:
919
27
Madden, D. R., J. C. Gorga, J. L. Strominger, D. C. Wiley.
1992
. The three-dimensional structure of HLA-B27 at 2.1 A resolution suggests a general mechanism for tight peptide binding to MHC.
Cell
70
:
1035
28
Viola, A., A. Lanzavecchia.
1996
. T cell activation determined by T cell receptor number and tunable thresholds.
Science
273
:
104
29
Valitutti, S., S. Müller, M. Cella, E. Padovan, A. Lanzavecchia.
1995
. Serial triggering of many T-cell receptors by a few peptide-MHC complexes.
Nature
375
:
148
30
Nukaya, I., M. Yasumoto, T. Iwasaki, M. Ideno, A. Sette, E. Celis, K. Takesako, I. Kato.
1999
. Identification of HLA-A24 epitope peptides of carcinoembryonic antigen which induce tumor-reactive cytotoxic T lymphocyte.
Int. J. Cancer
80
:
92
31
Peoples, G. E., B. W. Anderson, B. Fisk, A. P. Kudelka, J. T. Wharton, C. G. Ioannides.
1998
. Ovarian cancer-associated lymphocyte recognition of folate binding protein peptides.
Ann. Surg. Oncol.
5
:
743
32
Jackson, M., B. Smith, D. J. Bevitt, M. Steward, G. L. Toms, M. F. Bassendine, A. G. Diamond.
1999
. Comparison of cytotoxic T-lymphocyte responses to hepatitis C virus core protein in uninfected and infected individuals.
J. Med. Virol.
58
:
239