AFP-specific CD4⁺ Helper T-cell Responses in Healthy Donors and HCC Patients

Summary

Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide, and is often diagnosed at an advanced stage. We have investigated α-fetoprotein (AFP) as a tumor-associated antigen for HCC. We identified major histocompatibility complex class I-restricted peptide epitopes derived from AFP and studied CD8⁺ T-cell responses in vivo and in vitro in ongoing immunotherapy studies. Helper T cells are of critical importance in shaping the immune response; therefore, we investigated the frequency and function of AFP-specific CD4⁺ T cells in the general population and among HCC patients. CD4⁺ T-cell responses were assessed by direct ex vivo multicytokine enzyme-linked immunospot assay and by measurement of cytokine levels using a multicytokine assay. Our analysis indicates that healthy donors have very low frequencies of AFP-specific CD4⁺ T-cell responses, which are of T_H1 type, detectable ex vivo. In contrast, these T cells were either reduced or eliminated in HCC patients at advanced stages of disease. To better activate these cells, we compared the stimulatory capacity of both AFP protein-fed and AdVhAFP-engineered dendritic cells (DC). Healthy donors have CD4⁺ T-cell responses, which were activated in response to AFP protein-fed DC whereas HCC patients do not demonstrate significant responses to AFP protein. AdVhAFP-transduced DC were capable of activating higher frequency T_H1 CD4⁺ responses to AFP in both healthy donors and AFP-positive HCC patients. Importantly, CD4⁺ T-cell cytokine expression profiles were skewed towards interleukin-2 and interferon-γ production when activated by adenovirally engineered DC, which has therapeutic implications for vaccination efforts.

Hepatocellular carcinoma (HCC) has a very poor prognosis. Small tumors are potentially curable with ablative or surgical approaches, including liver transplantation. ^1–3 However, the majority of cases are detected at advanced stages, which, even if adequately treated locally, relapse systemically, and HCC recurs in 75% to 100% of patients at 5 years.⁴ There are no effective systemic therapies for this disease.^5,6 This leads to a 9% 5-year survival rate after diagnosis in the United States, the second lowest survival rate for any type of cancer.⁷

α-fetoprotein (AFP) is the most common serum protein during embryonic development, and is a physiologic counterpart of adult serum albumin. Suppression of AFP synthesis occurs shortly after birth. However, AFP mRNA can be detected in human liver at low, but consistent levels^8,9 and AFP expression can be also increased after liver injury.¹⁰ In addition, serum AFP is detected in 10% to 51% of subjects with viral hepatitis, in 25% with liver cirrhosis and 18.8% with benign diffuse liver disease.^11–16 Lastly, 90% to 95% of embryonic cancers (embryonic carcinomas, yolk sac tumors, hepatoblastoma) and 50% to 80% HCC show AFP reexpression during tumor development at levels from 10 ng/mL to 1 mg/mL.^10,17–19 Thus, all individuals have been exposed to high levels of this antigen during fetal development and those with liver diseases and ultimately HCC are exposed to this secreted antigen, often at high concentrations.

Although the exact function of AFP remains speculative, it seems to be involved in cell differentiation, growth regulation, and carcinogenesis. There are putative AFP cell surface receptors on cancer cell lines (including HCC), and also on normal hepatocytes, placental and immune cells.^20–22 Several studies have shown that peripheral blood mononuclear cells (PBMC) with impaired AFP receptor expression show decreased proliferative responses and reduced CD4⁺ T-cell mitogen responses.²³ HCC cell lines also show inhibition of proliferation if AFP receptors are blocked by specific antibodies.²⁴ In addition, the effect of AFP on immune cells is controversial with reports finding both stimulatory and inhibitory effects.^25–36

We have demonstrated that human and murine CD8⁺ T cells can recognize peptide epitopes derived from AFP.^37–44 We have also tested major histocompatibility complex (MHC) class I peptide-based vaccines in AFP+HCC subjects. Using MHC tetramer and interferon-γ (IFN-γ) enzyme-linked immunospot (ELISPOT) assays, we found that immunization with immunodominant peptides in Montanide adjuvant or pulsed onto dendritic cells (DC) was capable of increasing the frequency and function of AFP-specific CD8⁺ T cells in the peripheral blood^45,46 (Butterfield et al, submitted, 2006). These data indicate that these cells have not been deleted in patients despite high levels of circulating antigen.

AFP-specific helper T cells have also been detected in HCC patients.^33,47,48 The frequencies of spontaneous AFP-specific T cells are higher in cirrhotic patients without HCC and in those with lower serum AFP level.⁴⁸ Similar results were obtained in a study investigating responses to an AFP-derived HLA-DR13–restricted peptide.⁴⁷ These 3 published reports have focused on IFN-γ production by these helper T cells. In our ongoing efforts to target AFP as a tumor rejection antigen for immunotherapy, we wished to determine the nature of CD4⁺ helper T-cell immunity to AFP, given the importance of CD4⁺ cells in shaping the quality of the cellular immune responses, and allowing full function and proliferative response of CD8⁺ CTL.^49–51 Our current study has investigated the baseline frequency and function of spontaneous AFP-specific CD4⁺ T cells in the general population and among HCC patients. We have investigated potential sexual dimorphism, and the influence of different modes of antigen presentation by DC for amplification of AFP-specific helper responses. We find that healthy donors have detectable preexisting AFP-specific CD4⁺ T-cell responses whether AFP was presented by protein-pulsed autologous DC or by adenovirally engineered DC. In HCC patients, AFP-specific helper responses were only detected when DC were engineered to express AFP, indicating that genetically modified DC have a significantly improved ability to present this antigen in a stimulatory fashion to CD4⁺ cells. Importantly, we found that the cytokine profile of the AFP-specific CD4⁺ T cells depended on the mode of antigen presentation.

MATERIALS AND METHODS

Patient Samples, Cells and Cell Lines

PBMC were obtained from 26 healthy donors (15 males and 11 females) from the Central Blood Bank (Pittsburgh, PA; University of Pittsburgh IRB no. 04-001) and 6 male HCC patients previously enrolled in an AFP peptide-pulsed DC vaccination clinical trial⁴⁵ (UCLA IRB no. 00-01-026, FDA BB IND no. 9395).

PBMC were separated by Ficoll-Plaque Plus (Amersham Biosciences, Uppsala, Sweden) and stored in 90% human AB serum (Omega Scientific, Inc, Tarzana, CA)/10% dimethyl sulfoxide in liquid nitrogen. All cell culture was performed in culture media containing AFP-free human AB serum (Omega Scientific, Inc, Tarzana, CA) or serum-free X-vivo media (Life Technologies, Inc).

Viruses and Proteins

Recombinant adenoviruses AdVhAFP and Ad-VlacZ are Ad type 5 E1a/E1b-deleted first generation adenoviruses previously described.^52,53 AdVhAFP encodes a synthetic form of AFP matching the Genbank reference sequence NM_001134, driven by the CMV promoter. Adenoviral vectors were amplified on 293 cells (ATCC, Manassas, VA) and purifications were performed according to the manufacturer’s instructions (Adeno-X, BD Biosciences, San Diego, CA). Soluble cord blood AFP (CALBIOCHEM, San Diego, CA) was resuspended in phosphate-buffered saline (PBS)/0.1% bovine serum albumin (BSA) at 1 mg/mL.

Cell Isolation

DC

DC were obtained from loosely adherent mononuclear cells after 7 days of culture with interleukin-4 (IL-4) (500 U/mL, Schering-Plough, Kenilworth, NJ) and granulocyte macrophage-colony stimulating factor (GMCSF) (800 U/mL, Immunex, Seattle, WA). Adenoviral transduction of DC was performed in RPMI 1640/2% human AB serum with AdV vectors at multiplicity of infection=1000:1 pfu/DC (moi=1000) on day 7 of DC culture by incubation at 37°C for 2 hours.^38,54 This moi routinely results in >90% transduction efficiency. Cells were washed with excess media and resuspended at 1 × 10⁶ cells/mL. AFP protein pulsing of DC was performed in serum-free IMDM media (Life Technologies, Inc) with soluble serum AFP at 10 µg/mL on day 7 of DC culture at room temperature for 2 hours. Cells were washed and resuspended at 1 × 10⁶ cells/mL.

CD4⁺ T-lymphocytes

CD4⁺ cells were isolated by first, removal of adherent cells by plastic adherence, second, removal of CD8⁺ T cells by positive magnetic bead isolation, and finally by negative magnetic cell sorting using a CD4⁺ T Cell Isolation Kit which specifically removes any CD8⁺, CD14⁺, CD16⁺, CD19⁺, CD36⁺, CD56⁺, CD123⁺, TCR-γδ⁺, and CD235a⁺ cells (Miltenyi Biotec).

ELISPOT

The ELISPOT assay was performed according to Herr et al⁵⁵ with minor modifications. Ninety-six-well plates with nitrocellulose membranes (Millipore, Bedford, MA) were coated with primary antibodies (IL-2; IFN-γ, TNF-α, IL-5, IL-10) (BD Biosciences Pharmingen) in PBS at 4 µg/mL and incubated overnight at 4°C. Plates were washed with PBS, and then blocked with PBS/1% BSA for an hour at 37°C. Plates were washed with PBS and cells were plated in 200 µL of serum-free X-Vivo-10 media (Life Technologies, Inc). Negative controls included DC alone (1 × 10⁵ per well), DC without antigen with CD4⁺ T cells (2 × 10⁵, 1 × 10⁵, and 5 × 10⁴ per well), DC transduced with control vector (an empty AdV vector or with AdVlacZ) plus CD4⁺ T cells (as above), CD4⁺ T cells alone (10⁵). CD4⁺ T cell stimulated with 10 to 100µg/mL of PHA (Sigma) served as a positive control. Experimental conditions included DC transduced with AdVhAFP, or pulsed with soluble serum AFP plus CD4⁺ T cells. Each condition was plated in duplicate (each of 3 conditions, 6 wells total). Cells were incubated 24 hours (IL-2, IFN-γ, and TNF-α) or 48 hours (IL-5, IL-10). Cell-free supernatants from individual wells were frozen for subsequent Luminex assay analysis. Plates were washed with PBS and with PBS/0.05% Tween-20. Corresponding secondary antibody in PBS/0.05% Tween-20/1% BSA was added and incubated overnight at 4°C. Plates were washed 2 times with PBS, and then with PBS/0.05% Tween-20 at room temperature. Avidin-horseradish peroxidase (Vector Laboratories) was added at 1:2000 and incubated in the dark 2 hours. Plates were washed with PBS/Tween-20 and developed with AEC buffer [3-amino-9-ethylcarbozole (Sigma) in formamide/0.05M NaOAc buffer, pH5.0] with H₂O₂. The reaction was stopped in tap water. Spots were counted with an ImmunoSpot Analyzer Series 3A (Cellular Technology Ltd, Cleveland, OH).

Luminex

The Luminex assay was performed by the Luminex Core Facility of University of Pittsburgh (A. Lokshin, Director), the limit of detection depended on cytokine, and ranged between 3 and 40 pg/mL. The LabMAP technology (Luminex) combines the principle of a sandwich immunoassay with the fluorescent-bead–based technology allowing individual and multiplex analysis of up to 100 different analytes in a single microtiter well. The LabMAP media supernatant assays for IL-2, IFN-γ, TNFα, IL-5, IL-10, IL-4, GM-CSF were performed in 96-well microplate format according to the protocol by Biosource International (Camarillo, CA). A filter-bottom, 96-well microplate (Millipore, Billerica, MA) was blocked for 10 minutes with PBS/BSA. To generate a standard curve, 5-fold dilutions of appropriate standards were prepared in serum diluent. Standards and samples were pipetted at 50 µL per well in duplicate and mixed with 50 µL of the bead mixture. The microplate was incubated for 1 hour at room temperature on a microtiter shaker. Wells were then washed thrice with washing buffer using a vacuum manifold. Phycoerythrin-conjugated secondary antibody was added to the appropriate wells and the wells were incubated for 45 minutes in the dark with constant shaking. Wells were washed twice, assay buffer was added to each well, and samples were analyzed using the Bio-Plex suspension array system (Bio-Rad Laboratories, Hercules, CA). Analysis of experimental data was performed using 5-parametric-curve fitting, all as described in.⁵⁶

Statistical Analysis

Statistical analyses assessing the relationship between cytokine levels in a treated group and those in a control group were performed using a stratified Wilcoxon rank sum test. Two-sided P values were computed under the null hypothesis, that within each stratum, the ranked data are random, and do not depend on group. For analyses of individual cytokines, strata consisted of the 3 dilutions. T_H1 response was also assessed in an analysis that combined IL-2, IFN-γ, and TNF-α data; in this case, there were 9 strata: 3 dilutions for each of the 3 cytokines. The Wilcoxon test relies on the assumption that the strata are independent. Spearman test was performed to examine relationships between clinical data and the level of CD4⁺ immune responses in HCC patients. Two-sided P values less or equal to 0.05 were accepted as significant.

RESULTS

To optimally target AFP-expressing HCC for immunotherapy, CD4⁺ and CD8⁺ T cells should be activated for T_H1 and cytotoxic responses. Because of the fetal expression of AFP and its reactivation in disease states, the ability to develop potent AFP-specific CD4⁺ T-cell responses to this complex self-antigen has been investigated.

CD4⁺ T-cell Responses to AFP Protein in Healthy Donors

To establish a baseline for AFP CD4⁺ immunity, we wished to determine whether AFP-specific CD4⁺ T cells could be detected in the peripheral blood of healthy donors and, if so, to characterize the frequencies and cytokine profile of these cells. We initially examined this by stimulating cells overnight with autologous immature DC-fed AFP protein, which would approximate the expected mode of presentation when serum AFP is present. AFP-specific CD4⁺ T-cell responses were determined by direct ex vivo multicytokine ELISPOT assay and by measurement of cytokine levels in cell-free supernatants from the ELISPOT assay using Luminex array technology.

Eight healthy donors of 26 (30%) studied showed statistically significant preexisting CD4⁺ T-cell responses to soluble AFP protein, by at least 1 cytokine by ELISPOT assay (Table 1). Frequencies of helper T-cell responses were very low, at 2/1,000,000 to 47/1,000,000 CD4⁺ T cells (with a limit of detection of 1/1,000,000 cells). All 8 of the healthy donors had a T_H1 type of response: donor 11 had significant responses for IL-2, IFN-γ and TNF-α; donor 24 had a significant IL-2 response, and donors 02, 03, 13, 14, and 30 demonstrated significant TNF-α responses. Donor 18 had a T_H1 (TNF-α) and IL-10 response. By combining the trends in positivity in single cytokines, the pooled T_H1 cytokine data identified 3 additional positive donors (donors 6, 15, and 32) for a total of 11. Together, we observed that 10 of the 11 AFP-responsive healthy donors had purely T_H1 antigen-specific responses by ELISPOT assay. We considered that pregnancy might have an impact on responses, but did not have such data from female donors. Regardless, among the responders, 5 were females and 6 were males, indicating a lack of sexual dimorphism. Figure 1 shows 2 examples of cytokine profiles detected in responding and nonresponding healthy donors. Figure 1A shows a cytokine profile in an AFP protein “negative” donor (no. 08), and Figure 1B shows the cytokine profile in an AFP protein “positive” donor (no. 11).

TABLE 1.

Distribution of CD4⁺ T-cell Responses to Soluble AFP Protein-fed DC in Healthy Donors

Donor ID*	Sex†	Combined TH1 Cytokine‡	IL-2§	IFN-γ§	TNF-α§	IL-5§	IL-10§
00	Female	0.433	0.452	1.000	1.000	1.000	1.000
02	Female	0.027	0.328	0.978	0.009	1.000	1.000
03	Male	0.001	0.327	1.000	0.006	1.000	1.000
04	Male	0.561	0.231	0.995	0.817	1.000	1.000
05	Male	0.585	0.920	1.000	0.094	1.000	0.658
06	Male	<0.001	0.058	0.141	0.160	1.000	0.987
07	Male	0.788	0.879	0.965	0.982	1.000	1.000
08	Male	0.295	0.257	1.000	1.000	1.000	1.000
09	Male	0.743	0.997	1.000	0.326	1.000	1.000
10	Male	0.779	1.000	1.000	0.830	0.419	0.076
11	Male	<0.001	0.010	0.006	0.010	0.514	0.947
13	Male	0.092	0.866	0.993	0.007	1.000	0.518
14	Female	0.979	0.197	0.335	0.007	1.000	1.000
15	Male	0.007	0.150	1.000	0.051	1.000	0.992
17	Male	0.159	0.969	1.000	0.163	1.000	0.489
18	Male	0.014	0.581	1.000	0.007	1.000	0.032
19	Male	1.000	0.561	1.000	0.750	1.000	1.000
21	Female	0.669	1.000	1.000	0.673	0.164	1.000
22	Female	0.363	1.000	1.000	0.265	1.000	0.997
23	Female	0.438	0.491	1.000	0.995	1.000	1.000
24	Female	<0.001	0.027	0.054	0.122	1.000	1.000
28	Female	0.093	0.488	1.000	0.295	1.000	1.000
29	Female	0.865	1.000	1.000	1.000	1.000	1.000
30	Female	0.034	0.733	1.000	0.021	1.000	1.000
32	Female	0.039	0.253	1.000	0.254	1.000	0.251
33	Male	0.391	0.708	1.000	0.275	1.000	1.000

Results are presented as P values for combined T_H1 cytokine responses and for individual cytokine responses. Significant P values (P<0.05) are shown in bold font. Underlined font is used to show full T_H1 cytokine profile (IL-2/IFN-γ/TNF-α) responders. Italic font is used to show samples with mixed T_H1/IL-10 regulatory CD4 T-cell responses.

^*Identification number of healthy donor.

^†Sex of donor.

^‡Combined T_H1 cytokine is statistically significant T_H1 type responses calculated by combining trends in positivity responses for IL-2, IFN-γ and TNF-α.

^§Significance of responses of individual cytokines.

FIGURE 1.

Examples of CD4⁺ T-cell responses to AFP protein presented by autologous DC by direct ex vivo multicytokine ELISPOT assay. A, Cytokine profile in AFP protein nonresponsive donor 08. The difference between AFP protein-fed DC versus DC only was not significant with P = 0.29 for combined positivity in T_H1 cytokine responses, and P = 1.000 for TNF-α. B, Cytokine profile in AFP protein reactive donor 11. Results are expressed in average spot distribution between wells. "DC+CD4" is DC without antigen+CD4⁺ T cells; "DC (AFP)+CD4" is AFP protein-fed DC+CD4⁺ T cells.

To expand the number of cytokines examined, ELISPOT supernatants were tested by Luminex methodology for the 5 cytokines tested by ELISPOT as well as IL-4 and GM-CSF. Unfortunately, the Luminex array results for AFP protein-specific helper T-cell responses were below the level of detection and were not significant. Therefore, there was no evidence for additional cytokines produced (other than to PHA positive mitogen control, not shown).

CD4⁺ T-cell Responses to AdVhAFP and Their Comparison With Responses to AFP Protein in Healthy Donors

We next wished to determine whether a more potent and immunogenic method of antigen presentation would allow detection of AFP-specific CD4⁺ T cells. We have previously shown that AdV transduction is an efficient method of engineering DC to express a transgene for at least 10 days.^54,57 Because AdV-transduced DC also process and present AdV epitopes,⁵² which would be recall antigens for most subjects, responses to AdVhAFP-transduced DC (AdVhAFP/DC) were compared with responses to AdVlacZ-transduced DC. AdVlacZ uses the same backbone as AdVhAFP and should process and present the same AdV epitopes as AdVhAFP/DC. All analyses were performed by 2 assays: ELISPOT and Luminex array, in parallel.

Fifteen healthy donors (57%) had statistically significant responses to at least 1 individual cytokine and when T_H1 cytokines were combined, 22 healthy donors (84%) demonstrated significant CD4⁺ T-cell responses to AdVhAFP/DC by ELISPOT assay (Table 2), from 54 to 2050 positive cells/10⁶. Twenty donors had exclusively T_H1 responses and 2 donors (donors 18 and 24) showed mixed T_H1/T_H2/IL-10 regulatory CD4⁺ helper responses. We were able to detect AFP-specific CD4⁺ responses in a much higher percentage of donors by presenting the antigen via AdVhAFP/DC, compared with AFP protein-fed DC. Donor 18, who had a mixed T_H1/regulatory response to AFP protein (Table 1), reproduced this type of response with AdVhAFP/DC, adding significant production of IL-2 and IFN-γ. Surprisingly, 1 sample (donor 02) positive for an AFP-specific response to AFP protein was negative for response to AdVhAFP. All other donors who had detectable AFP protein responses were also positive with AdVhAFP/DC. Two examples are shown in Figure 2 (donors 32 and 09).

TABLE 2.

Distribution of CD4⁺ T-cell Responses to AdVhAFP in Healthy Donors

Donor ID*	Sex†	Combined TH1 Responses‡	IL-2§	IFN-γ§	TNF-α§	IL-5§	IL-10§
00	Female	0.877	0.245	0.466	1.000	0.972	0.927
02	Female	1.000	1.000	0.133	1.000	1.000	1.000
03	Male	0.015	0.158	0.157	0.435	0.717	1.000
04	Male	0.002	0.049	0.057	0.544	1.000	0.961
05	Male	0.047	0.681	0.062	0.569	1.000	0.677
06	Male	<0.001	0.012	0.033	0.049	1.000	1.000
07	Male	<0.001	0.053	0.060	0.057	1.000	1.000
08	Male	0.280	0.591	1.000	0.527	1.000	1.000
09	Male	<0.001	0.051	0.062	0.065	1.000	1.000
10	Male	<0.001	0.195	0.009	0.012	0.187	0.259
11	Male	<0.001	0.008	0.032	0.007	1.000	1.000
13	Male	<0.001	0.009	0.012	0.009	0.507	0.970
14	Female	<0.001	0.009	0.007	0.039	1.000	0.058
15	Male	<0.001	0.090	0.010	0.010	0.981	0.091
17	Male	<0.001	0.057	0.158	0.012	1.000	0.111
18	Male	<0.001	0.008	0.027	0.008	1.000	0.041
19	Male	<0.001	0.013	0.066	0.008	1.000	1.000
21	Female	0.345	1.000	1.000	0.313	0.326	1.000
22	Female	0.012	0.241	0.507	0.277	1.000	1.000
23	Female	<0.001	0.005	0.332	0.013	1.000	1.000
24	Female	0.001	0.146	0.292	0.011	0.047	0.008
28	Female	<0.001	0.009	0.049	0.035	1.000	1.000
29	Female	0.002	0.056	0.665	0.051	1.000	0.984
30	Female	<0.001	0.015	1.000	0.005	1.000	1.000
32	Female	0.007	0.245	0.518	0.251	0.997	0.274
33	Male	<0.001	0.009	0.006	0.012	1.000	1.000

Results are presented as P values for combined T_H1 cytokine responses and for individual cytokine responses. Significant P values (P<0.05) are shown in bold font. Underlined font is used to show full T_H1 cytokine profile (IL-2/IFN-γ/TNF-α) responders. Italic font is used to show samples with mixed T_H1/IL-10 regulatory CD4 T-cell responses.

^*Identification number of healthy donor.

^†Sex of donor.

^‡Combined T_H1 cytokine is statistically significant T_H1 type responses calculated by combining trends in positivity responses for IL-2, IFN-γ and TNF-α.

^§Significance of responses of individual cytokines.

FIGURE 2.

Example of differences in cytokine profiles between AFP protein/DC-specific and AdVhAFP/DC-specific CD4⁺ T-cell responses in healthy donors. A, Cytokine profile in donor 32, responsive for AdVhAFP/DC and for AFP protein/DC. B, Cytokine profile in donor 09, nonresponsive for AFP protein/DC but positive to AdVhAFP/DC. Results are expressed in average spot distribution between wells. DC+CD4 is DC without antigen+CD4⁺ T cells; DC (AFP)+CD4 is AFP protein-fed DC+CD4⁺ T cells; "DC (AdVlacZ)+CD4" is AdVlacZ engineered DC+CD4⁺ T cells; "DC (AdVhAFP)+CD4" is AdVhAFP-engineered DC+CD4⁺ T cells.

The cytokine profiles from AdVhAFP/DC-stimulated cells were significantly different from AFP protein/DC-stimulated cells, with more cells producing IL-2 and IFN-γ and lower frequencies of cells exclusively producing TNF-α. In fact, significant responses to TNF-α alone were reduced from 5/8 (AFP protein) to 1/15 (donor 17, AdVhAFP) among responders. The significant frequencies of IFN-γ producing cells strongly increased from 1/8 (donor 11) (AFP protein) to 8/15 (AdVhAFP). A similar tendency was observed for IL-2, from 2/8 (AFP protein) to 11/15 (AdVhAFP). Spot sizes and their intensity in the ELISPOT assay were also strongly increased when antigen was presented by AdVhAFP/DC (data not shown).

Using a Luminex assay, many samples had insufficient levels of cytokine to be in range. However, a majority of the analyzable samples showed similar results for AdVhAFP-specific responses in healthy donors between Luminex and ELISPOT assays for pooled T_H1 cytokine data (18/19 concordant, Table 3).

TABLE 3.

Comparison of ELISPOT and Luminex Assays Results for CD4⁺ T-cell Responses to AdVhAFP in Healthy Donors

		Combined TH1 Cytokine‡
Donor ID*	Sex†	ELISPOT	Luminex
00	Female	0.877	1.00
04	Male	0.002	0.044
05	Male	0.047	0.352
10	Male	0.001	0.008
11	Male	0.001	0.010
13	Male	0.001	0.004
14	Female	0.001	0.014
15	Male	0.001	0.010
17	Male	0.001	0.004
18	Male	0.001	0.026
19	Male	0.001	0.008
21	Female	0.345	0.282
22	Female	0.012	0.006
23	Female	0.001	0.006
24	Female	0.001	0.008
29	Female	0.002	0.044
30	Female	0.001	0.008
32	Female	0.007	0.026
33	Male	0.001	0.010

Results are presented as P values for combined T_H1 cytokine responses and for individual cytokine responses. Combined T_H1 cytokine—statistically significant T_H1 type responses calculated by combining trends in positivity responses for IL-2, IFN-γ and TNF-α. Significant P values (P < 0.05) are shown in bold font.

^*Identification number of healthy donor.

^†Sex of donor.

^‡Combined T_H1 cytokine is statistically significant T_H1 type responses calculated by combining trends in positivity responses for IL-2, IFN-γ and TNF-α.

CD4⁺ T-cell Responses to Soluble AFP in HCC Patients

We next wished to assess the state of CD4⁺ immunity in subjects with AFP-positive HCC. The characteristics of these patients are shown in Table 4. They were all male, most had stage IV disease, they had different HCC risk factors and a wide range of serum AFP concentrations. They are listed in order of serum AFP levels. We tested the same modes of antigen presentation (protein-fed and AdVhAFP-transduced DC) and the same ex vivo multicytokine ELISPOT and Luminex assays. Unfortunately, there were insufficient cells to study IL-5 and IL-10 cytokine responses by ELISPOT.

TABLE 4.

Clinical HCC Patient’s Data and Distribution of AFP – /AdVhAFP-specific CD4⁺ T-cell Responses

ID*	Age†	Sex‡	Race§	Risk Factors‖	Stage¶	AFP Level (ng/ml)#	AFP-specific Responses**	AdVhAFP-specific Responses††
B12	52	M	White	Unknown	III	74.3	0.095	0.013
A3	69	M	White	Alcohol	IVa	3080	0.293	0.981
B11	60	M	White	Porphypia	IVa	4340	0.461	0.031
A4	62	M	Asian	HCV	IVb	10,800	0.987	0.033
B1	67	M	White	HBV	IVb	60,800	0.083	0.090
B10	59	M	White	HCV	IVa	463,040	0.327	1.000

Results are presented as P values for combined T_H1 cytokine responses.

Combined T_H1 cytokine—statistically significant T_H1 type responses calculated by combining trends in positivity responses for IL-2, IFN-γ, and TNF-α. Significant P values (P<0.05) are shown in bold font.

^*HCC patient identification.

^†Age of patient.

^‡M—male sex.

^§Race of patient.

^‖Risk factor identified for HCC, HCV, hepatitis C virus; HBV, hepatitis B virus.

^¶Stage of HCC disease.

^#AFP level—level AFP in the blood in time of leukapheresis.

^**AFP-specific responses: CD4⁺ T-cell responses to AFP protein-fed DC.

^††AdVhAFP-specific responses: CD4⁺ T-cell responses to AFP activated by AdVhAFP-engineered DC.

We did not detect any statistically significant CD4⁺ T-cell responses to AFP protein/DC by single or by combination of T_H1 cytokines among these HCC patients. A summary of the cytokine expression profiles in HCC obtained by ex vivo ELISPOT assay to AFP protein-fed DC is shown in Figure 3. Three of the 6 HCC patients appeared to have a trend toward AFP-specific CD4⁺ responses (B12, A3, B1) but these were not significant. Patients A3 and B1 demonstrated slight but not significant TNF-α increases. PHA stimulation of CD4⁺ T cells stimulated high frequencies of IL-2, IFN-γ, and TNF-α producing T cells, demonstrating that the cells were capable of strong, multicytokine responses. The only exception was patient B10 [who had the highest AFP level (463,040 ng/mL)] with viable cells but very low PHA responses.

FIGURE 3.

Summary of T_H1 cytokine profiles of CD4⁺ helper cells activated by AFP-fed DC versus empty DC among HCC patients. Patients are listed in order of increasing serum AFP level. Results are expressed as average spot counts. DC+CD4 is DC without antigen+CD4⁺ T cells; DC (AFP)+CD4 is AFP protein-fed DC+CD4⁺ T cells.

CD4⁺ T-cell Responses to AdVhAFP and Clinical Data in HCC Patients

When HCC patient cells were stimulated by AFP in the form of AdVhAFP/DC, the responses were much stronger, as we observed with the healthy donors. However, single cytokine responses were still not statistically significant (Fig. 4). By combined T_H1 cytokine analysis, 3 HCC patients (50%) had significant responses to AdVhAFP/DC (B12, B11, A3). There has been a report that high level AFP (10 µg/mL) is inhibitory to DC function, ³⁶ and half of the HCC patients in our study (A4, B1, B10) were above this level. Notably, most of the pooled T_H1 cytokine responders had lower serum AFP levels than nonresponders, and also earlier stages of disease (Table 4), in agreement with Um et al.³⁶ In addition, patient B12 (stage III, lowest serum AFP) had a detectable [but not statistically significant (P=0.095)] increase in combined T_H1 cytokine (IL-2, IFN-γ, and TNF-α) response to AFP protein/DC and significant responses to AdVhAFP engineered DC (Figs. 3, 4; Table 4).

FIGURE 4.

Summary of T_H1 cytokine profiles of CD4⁺ helper cells activated by AdVhAFP-engineered DC versus DC transduced with AdVlacZ among HCC patients (B12, A3, B11, A4, B1, and B10). Patients are listed in order of increasing serum AFP level. Results are expressed as average spot counts. DC (AdVlacZ)+CD4 is AdVlacZ engineered DC+CD4⁺ T cells; DC (AdVhAFP)+CD4 is AdVhAFP-engineered DC+CD4⁺ T cells.

We found that the frequency of AdVhAFP/DC activated CD4⁺ helper T cells expressing IL-2 and IFN-γ seemed to be related to level of serum AFP level (Fig. 4); however, Spearman test indicated no significant correlation between the level of serum AFP and the expression of IL-2 (P=0.10), IFN-γ (P=0.24) (or TNF-α). If this nonsignificant trend in IL-2 and IFN-γ (not TNFα) with high AFP level were found significant in a study with a larger sample size, it would suggest that serum AFP concentration may affect CD4⁺ immunity in HCC patients. There were no trends or significant differences in responses to AdV antigens (AdVlacZ) or AFP protein/DC and serum AFP level.

DISCUSSION

In the present study, we have investigated spontaneous CD4⁺ T-cell responses to the oncofetal antigen AFP presented by either soluble protein-fed immature DC or AdVhAFP-engineered DC in healthy donors and HCC patients. The majority of reports of AFP-specific T-cell responses have focused on CD8⁺ T-cell responses to MHC class I peptides.^{33,38,39,41–46,58} It is clear, however, that activation of fully functional CTL requires helper T cells.⁵¹ “Helped” CTL are capable of undergoing secondary expansion and acquiring full effector function. ⁵⁰ Here, using autologous DC presenting 2 forms of the antigen, we have detected low levels of preexisting T_H1 CD4⁺ T-cell responses by direct ex vivo multicytokine ELISPOT assays in healthy donors (to AFP protein-fed DC) and higher frequency of T_H1 helper responses (to AdVhAFP-transduced DC) in both HCC patients and healthy donors, by both ELISPOT and Luminex cytokine assays.

Our analysis indicates that 30% to 57% of healthy donors (8/26-15/26) have significant AFP-specific CD4⁺ helper cell responses, but that HCC patients at advanced stages of disease no longer have significant responses. This may be due to exhaustion or activation-induced cell death from chronic antigen exposure. It is also possible that these cells are present in HCC patients at a frequency not detectable ex vivo, and that brief in vitro stimulation would reveal that these cells are not completely eliminated. This is under investigation. It is also possible that there are AFP-specific helper cells producing cytokines other than the 7 most commonly described cytokines tested here.

We have also shown that the mode of antigen presentation has an important effect on the ability to detect these T-cell responses. AdVhAFP-engineered DC are clearly more potent APC than DC pulsed with soluble AFP protein. It has been shown that AdV-transduced DC become more mature owing to down-regulation of CD14, up-regulation of CD83, CD86, and HLA-DR, and also decreased production of IL-10, and increased expression of IL-12p70.⁵⁹ AdV-transduced DC also present antigen for as long as 10 days⁵⁷; however, in the context of a 24 to 48-hour ex vivo cytokine assay, this is unlikely to be a critical difference. It has also been shown that DC transduced with type 5 AdV vectors produce type I IFN (IFN-α), which can also drive DC maturation.^59,60 Here, we find that, for healthy donors, the frequencies of responding CD4⁺ T cells were 10 to 100 times higher with AdVhAFP/DC-stimulated helper T cells and the amount of expressed cytokines (measured by Luminex array) was often 2 to 10 times higher (data not shown).

Interestingly, the cytokine profiles elicited by these 2 modes of antigen presentation were different. TNF-α was the most commonly expressed cytokine in CD4⁺ T cells activated by soluble AFP, whereas in AdVhAFP-induced T cells, IL-2, IFN-γ, and TNFα were more equivalently expressed. Only one donor had a significant IFN-γ response when activation was via soluble AFP, and 2 had significant IL-2 responses (one of which was also IFN-γ +), whereas 5 were TNF-α responses exclusively and 1 was both TNF-α and IL-10 (Table 1). Therefore, if we had only investigated IFN-γ, we would have concluded that 1/26 donors had preexisting levels of AFP-specific CD4⁺ T cells. These data are in sharp contract to those obtained with stimulation of cells with AdVhAFP/DC. Table 2 shows that only 1 donor had exclusively TNF-α responses, whereas 13 donors responded with 2 to 3 different T_H1 cytokines. This indicates that the more potent APC mode stimulated a greater frequency and functional breadth in responding cells. The weaker mode of presentation was designed to mimic the mode naturally occurring in vivo when immature DC take up soluble AFP during fetal life and in the presence of tumor. This seems to result in low level activation of cells with a limited range of function (primarily TNF-α secretion). The AdVhAFP/DC were more potent CD4⁺ T-cell activators, and have also been recently found to activate a broad range of CD8⁺ T-cell antigenic specificities.⁴⁴

The ability of CD4⁺ T cells to produce IFN-γ can positively impact antitumor immunity by inducing expression of the immunoprotoesome in APC, as well as TAP transporter proteins and MHC class I molecules, also making tumors more easily recognized by CD8⁺ T cells.⁶¹ IL-2 acts in several ways, potentially promoting CD8⁺ T-cell proliferation and death. It is thought that IL-2 produced by helper cells promotes activation and expansion of CTL with low CD8⁺ precursor frequencies and low affinity of peptide-MHC interactions.⁶² Importantly, IL-2 is an early signal for establishing long-term T-cell expansion and ability to differentiate into effector cells,⁶³ and anergy of T cells can be reversed by exogenous IL-2.⁶⁴

B-cell responses to AFP are generally not detected, which may point to deficits in CD4⁺ T-cell responses. However, it has been shown that AFP-specific CD4⁺ T-cell responses (to an HLA class II DR-restricted AFP-derived epitope) were only found in HCC and cirrhotic patients (n=40 and n=13, respectively) but not in any healthy donors (n=7).⁴⁷ We have found that frequencies of AFP-specific responses to soluble AFP protein detected by ELISPOT assay are very low in healthy donors (in agreement with⁴⁷). Importantly, given that we are detecting all possible CD4⁺ helper cells responding to the full length AFP protein, the frequencies of cells responding to the single AFP-derived epitope would be lower.

There are several potential scenarios in which healthy individuals might be exposed to AFP transiently, allowing for expansion of AFP-specific CD4⁺ T cells. First, AFP expression has been identified in normal liver cells, localized in areas of liver blood vessels and sinusoids.^8,9,65 Second, reactivation of AFP synthesis in hepatocytes occurs reversibly after liver injuries¹⁰ and during acute viral hepatitis A, B, and C infections. ^11,12,14,15 Third, it has been shown that some immune cells can develop into memory cells owing to exposure to AFP during early stages of embryonic development.⁶⁶ Fourth, women are exposed to AFP during their pregnancies. It may be important in future studies to test healthy donor cells and serum for evidence of exposure to hepatitis viruses to determine the potential role of exposure to antigen from viral infection in helper cell activation.

Our data indicate lower frequencies of AFP-specific CD4⁺ T cells in some of the HCC patients with the highest levels of serum AFP. The trend toward this inverse correlation was found for IL-2 and IFN-γ producing cells, although it did not reach significance (P = 0.10 and 0.24, respectively). Earlier studies have found that high levels of AFP (greater than 100 ng/mL to 10 µg/mL) suppress MHC class II expression on monocyte-derived DCs and on CD4⁺ T cells.^26,34,36 AFP-exposed DC exhibited lower cell viability, inhibition of cell stimulatory capacity, decreased ability to produce cytokines, and increased susceptibility to apoptosis.³⁶ TNF-α did not show any significant relationship with either serum AFP level or other cytokine production, which could be complicated by binding of this cytokine by AFP protein.^67–69

There have been 2 clinical trials conducted in which advanced stage HCC patients were immunized with MHC class I-restricted AFP peptides with the goal ofactivating CD8⁺ T cells.^45,46 Although the majority of patients were successfully immunized, there were no objective clinical responses. Several other immunotherapy trials have been tested in HCC. Most recently, Gao et al⁷⁰ have tested autologous tumor lysate-pulsed DC in 30 postoperative HCC patients. The survival rate at 18 months was improved in the DC-vaccine treated group (vs. the chemotherapy group) and the hepatic recurrence rate in the DC group was 13% (vs. 54% in the chemotherapy group, P < 0.05),⁷⁰ which suggests that immunotherapy may clinically impact HCC. The complex mixture of protein antigens in tumor lysate would be expected to broadly stimulate multiple CD8⁺ and CD4⁺ T cells, which might have played a role in the improved clinical outcomes. We have not observed any AFP vaccine-related toxicities, but as we test improved vaccine strategies, improved T-cell responses to this self-antigen could result in some level of autoimmune toxicity. In previous murine studies, mice immunized with murine AFP were examined pathologically, and there was no evidence of toxicity.⁴³

An important advantage of using the adenovirally engineered DC as the mode of antigen presentation maybe to program a T_H1 type of adaptive response to tumor associated antigens. However, the exact role of virus-specific adenoviral responses in promotion of tumor antigen responses has yet to be identified. We have detected these antiadenovirus cytotoxic responses in addition to tumor antigen responses from AdV-transduced DC in vitro.⁵² Notably, in murine model studies, systemic AdV delivery resulted in neutralizing anti-AdV antibody responses, and multiple AdV-engineered DC immunizations also resulted in AdV antibody stimulation.⁷¹ However, transgene-specific T-cell responses induced by AdV/DC were not reduced by the presence of anti-AdV neutralizing antibodies. Potentially higher avidity, T_H-1 skewed recall T-cell responses to the foreign adenoviral epitopes may also play a critical role in shaping these CD4⁺ responses; this is an area of active investigation.