Paradoxical enhancement of CD8 T cell-dependent anti-tumor protection despite reduced CD8 T cell responses with addition of a TLR9 agonist to a tumor vaccine


  • Dev Karan,

    1. VA Medical Center, Iowa City, IA
    2. Department of Urology, University of Iowa, Iowa City, IA
    3. The Prostate Cancer Research Program in the Holden Comprehensive Cancer Center, University of Iowa, Iowa City, IA
    Search for more papers by this author
  • Arthur M. Krieg,

    1. Coley Pharmaceutical Group, Inc., Wellesley, MA
    Search for more papers by this author
  • David M. Lubaroff

    Corresponding author
    1. VA Medical Center, Iowa City, IA
    2. Department of Urology, University of Iowa, Iowa City, IA
    3. The Prostate Cancer Research Program in the Holden Comprehensive Cancer Center, University of Iowa, Iowa City, IA
    4. Department of Microbiology, University of Iowa, Iowa City, IA
    5. Interdisciplinary Program in Immunology, University of Iowa, Iowa City, IA
    • Department of Urology, 375 Newton Road, 3210 MERF, Iowa City, IA 52242-2600, USA
    Search for more papers by this author


Generation of antigen-specific CD8+ T cell responses is considered optimal for an effective immunotherapy against cancer. In this study, we provide a proof of principle that in vitro observed diminished CD8+ T cell response provided a strong in vivo tumor protection. Immunization with an adenovirus vaccine containing ovalbumin (OVA) gene (Ad5-OVA) strongly induces antigen-specific CD8+ T cell responses measured in vitro using various immunological assays. However, in an attempt to augment the antigenic CD8+ T cell response, coinjection of a TLR9 agonist CpG ODN with the viral vaccine unexpectedly reduced the CD8+ T cell responses measured in vitro but provided a remarkably enhanced tumor protection compared to the CD8+ T cell response generated by Ad5-OVA vaccine alone. Interestingly, despite reduced ex vivo/in vitro CD8+ T cell responses following Ad5-OVA+CpG immunization, immunodepletion studies revealed that the augmented anti-tumor immunity was primarily dependent on CD8+ T cells. The magnitude and effector function of anti-OVA CD8+ T cells remain low following primary and secondary antigenic challenge, presenting a dichotomy between in vitro CD8 T cell responses and in vivo anti-tumor immunity. To examine the impact of CpG ODN, we observed that presence of CpG suppresses the CD8+ T cell proliferation both in vitro and in vivo. These data demonstrate that coadministration of adenovirus vaccine with a TLR9 agonist can generate potentially effective tumor-reactive CD8+ T cells in vivo. In addition, the results indicate that widely used standard immune parameters may not predict the vaccine efficacy containing a TLR9 agonist as adjuvant. © 2007 Wiley-Liss, Inc.

Therapeutic vaccinations are designed to stimulate an excess accumulation of antigen-specific T cells with effector functions that result in the destruction of antigen-bearing tumors. Since healthy individuals harbor a repertoire of auto-reactive T cells, use of cancer vaccines with tumor cell lines or tumor-specific antigens has been successful in breaking immune tolerance to these self-antigens.1, 2 There are multiple choices for the delivery of these antigens, such as RNA, DNA, whole protein, peptides, viral vectors or dendritic cells (DC). Immunizations using DCs loaded with the antigen strongly influence the effector T cell function with induced anti-tumor immunity.3, 4 Use of recombinant adenovirus as a genetic delivery system containing a gene of interest has been an attractive approach to elicit the immune response because of their ability to infect a broad range of cell types and a high propensity to transport the tumor antigens.5, 6 It has been demonstrated that use of adenovirus as a vehicle for antigen delivery as well as vaccine carrier strongly induces cytotoxic T cell response.7, 8, 9 They are further capable of modulating DC maturation by an increase in the expression of MHC antigens and costimulatory molecules.10

Current strategies to further improve the effectiveness of a vaccine formulation against the targeted tumor cells include combining adenovirus vaccines with an adjuvant capable of enhancing antigen presentation to DCs. A variety of adjuvants have been evaluated in immunization protocols but none have been successful in mediating the regression of the disease in advanced stages. Discovery of Toll-like receptor 9 (TLR9) ligands (commonly known as CpG ODN), containing nonmethylated CpG motifs have provided a family of novel and relatively safe Th1-promoting adjuvants.11, 12 Use of CpG as a monotherapy or an adjuvant has successfully demonstrated an induction of innate or adaptive T cell immunity where CpG induced IL-12 secretion by antigen presenting cells (APCs) which in turn activate NK cells to secrete IFN-γ, hence generation of an effective immune response depends on the presentation of antigen via APCs. Therefore, activation of APCs via CpG ODN appeared to be a prerequisite to stimulate functional antigen-reactive T cells.13, 14, 15

Various models have successfully demonstrated that addition of CpG ODN with the antigen during primary immunization cause an induced T cell function including tumor clearance and CD8 T cell proliferation.13, 16, 17 Several investigators observed that, in addition to its immune stimulatory effect, CpG ODN can also reduce the magnitude of CD8+ T cells and IFN-γ secretion.18, 19 However, these studies did not examine the impact on in vivo immune responses against the tumor challenge. Recently, we examined the sensitivity of such in vivo responses and demonstrated that despite diminished CD8+ T cell responses measured in vitro, there was an augmented anti-tumor response in vivo following immunization with Ad5-PSA and CpG ODN.20 Here in this study, we transform such observations in details to understand the basic parameters of clinical importance in a mouse tumor model using ovalbumin (OVA) as a model antigen. The ability of a combined approach of immunization (Ad5-OVA+CpG) was examined at multiple time points to analyze the magnitude and effector function of CD8+ T cells following primary (prophylactic) as well as secondary antigenic challenge (post-tumor analysis). The results conclude that clinical trials should be carefully designed when using the vaccine strategy in combination with TLR9 agonist as adjuvant.

Material and methods

Mouse strain and tumor lines

Six- to eight- weeks-old C57BL/6 (H2b) male mice were purchased from National Cancer Institute and were maintained in filtered cages. OT-I mice were kindly provided by Dr. Timothy Ratliff (University of Iowa). OVA expressing E.G.7 tumor cells and non ova-expressing EL4, obtained from ATCC, were used to measure in vitro anti-OVA CTL activity and in vivo tumor outgrowth. Cells were maintained in regular culture medium RPMI supplemented with 10% FBS, 1% glutamine, 1% sodium pyruvate (GIBCO, Invitrogen, CA) and 0.05 mg/ml of gentamicin (Mediatech, VA).

Vaccine, CpG ODN and vaccination

Adenovirus OVA (Ad5-OVA) vaccine was obtained from the University of Iowa's Gene Transfer Vector Core. Nonmethylated CpG ODN 1826 (5′ TCCATGACGTTCCTGACGTT) with a phosphorothioate-modified backbone for nuclease resistance was provided by Coley Pharmaceutical Group (Wellesley, MA) and had no detectable endotoxin. Mice were anesthetized using ketamine and shaved on the back. Ad5-OVA vaccine (108 pfu) ± CpG (50 μg) was injected subcutaneously on the flank and 2 weeks later mice were sacrificed to obtain the spleen and/or lymph nodes. For each group, experiments were repeated multiple times unless otherwise mentioned specifically. All the experiments were performed in accordance with guidelines and regulations approved by the University of Iowa Institutional Animal Care and Use committee.

In vitro cytotoxic assay

OVA antigen-specific lytic activity was measured using a standard method of 51Cr release assay. Spleen cells were seeded at a density of 1 × 107 cells/well for each group in the presence of cytokine IL-2 (10 U/ml) in 24-well plates and were stimulated once with mitomycin C treated E.G.7 cells (4 × 105). Following a 5-day coculture at 37°C and 5% CO2, the effectors were harvested by Fico/Lite-LM (Atlanta Biologicals, GA) separation. Each of the target cells, E.G.7 or EL4, were labeled with 100 μCi of Na251CrO4 (Amersham Biosciences, NJ) for 1 hr at 37°C. Labeled cells were washed twice with complete medium and 5 × 104 radiolabeled target cells were mixed with graded number of effectors per well in a V-shaped 96-well plate. Following a 4 hr incubation at 37°C, 100 μl aliquots of the supernatant were removed from each well and counted in COBRA™ II, Auto-gamma counter (Packard Instrument Company, IL). The antigen-specific percent lytic activity was calculated as previously.20

In vivo tumor study

OVA-expressing E.G.7 tumor cells in log-phase growth were thoroughly washed with D-PBS (Dulbecco's phosphate buffer saline, GIBCO, Invitrogen, CA) and 100 μl of the cell suspensions containing 1 × 107 viable cells were injected subcutaneously into C57BL/6 mice. Tumor outgrowth, determined by tumor size as a function of time, was measured twice a week and tumor volume was calculated as [(D1: smaller diameter)2 × (D2: larger diameter) × 0.5326] described by Shariat et al.21 Survival of the tumor bearing mice was also determined. Mice were sacrificed for humane reasons if a single tumor was greater than 25 mm in any dimension or if the mice appeared ill from the tumor burden. Each experimental group consisted of 5 mice and was repeated at least twice.

Ex vivo analysis of antigen-specific CD8+ T cells

Following immunization, spleen cell suspensions were processed to analyze the number of IFN-γ producing cells using 3 different techniques: intracellular cytokine staining (ICS), enzyme linked immunospot (ELISpot), Enzyme linked immunosorbent assay (ELISA) as well as analyzing antigen-specific CD8+ T cells by tetramer assay.

Spleens from the immunized or control groups of mice were removed, red blood cells lysed using ACK buffer and the cell suspensions passed through a 70 μm cell strainer (BD Falcon). To quantify the antigen-specific CD8+ T cells for IFN-γ by ICS, effectors were plated at a density of 5 × 106 cells/well in round bottom 96-well plates, and were stimulated with MHC Class I OVA-peptide (SIINFEKL; 1 μg/106 cells) for 4 hr. Nonstimulated effectors incubated simultaneously were used as controls for normalization to calculate the antigen-specific CD8+ T cell response. Within 30 min of incubation at 37°C, brefeldin A (SIGMA) was added to the wells to inhibit the secretion of IFN-γ. After the incubation, cells were treated with Fc Block (anti-CD16/CD32) for 10 min and then processed for ICS staining using Cy-5-conjugated anti-CD3, FITC-conjugated anti-CD8, and PE-conjugated IFN-γ antibodies (BioLegend). Cells were fixed and permeabilized with Cytofix/Cytoperm Plus Kit (Pharmingen) according to manufacturer's instructions. Flow cytometric analysis was performed collecting 1 × 105 events and data were analyzed with FlowJow 6.4.2 (Tree Star, Stanford) software.

Simultaneously, the number of IFN-γ producing cells was examined using the ELISpot assay. Splenocytes were seeded at a density of 5 × 106 cells/well in UNIFILTER® 96-well ELISpot plates (Whatman) previously coated with mouse anti-IFN-γ antibody (3 μg/ml; BioLegend), and were stimulated with MHC Class I peptide as mentioned previously. Following overnight incubation (15–18 hr) at 37°C, the cells were washed 3 times with PBS containing 0.05% Tween-20. Plates were incubated for 3 to 4 hr at room temperature with biotinylated mouse secondary antibody for IFN-γ (BioLegend). After 3 washes as before, the plates were further incubated for 2 hr with anti-rabbit IgG HRPO (SIGMA) and then processed for staining using an AEC kit according to manufacturer's instructions (Vector Laboratories). The number of spot- forming cells were analyzed using ImmunoSpot software (Cellular Technology).

To analyze the total number of antigen-specific CD8+ T cells ex vivo, 2–5 × 106 splenocytes or lymph nodes from the immunized mice as explained earlier, were stained for MHC I peptide-specific (SIINFEKL) tetramer (NCI), and flow analysis was performed collecting 1 × 105 events.

Enzyme linked immunosorbent assay

During the coculture set up for in vitro CTL assays, 100 μl of supernatant from each well was collected following 48 hr of coculture and was pooled for each sample to be analyzed for IFN-γ secretion. ELISA for IFN-γ was performed using Quantikine™ murine ELISA kit (R&D System, MN) following manufacturer's instruction.

Immune cell depletion and tumor growth

Mice were depleted in vivo of various immune cell populations using 2 injections of 100 μg of monoclonal antibodies to CD8 (2.43), CD4 (GK1.5), NK1.1 (PK136) or for both CD4 and CD8 on day 10 and 12 after immunization. On day 14, one mouse from each group was sacrificed to verify the cell depletion by flow cytometry for CD8+ (depletion 97.23%) and CD4+ (depletion 93.07%) T cells, while 51Cr-release assay was performed for NK cells against NK sensitive YAC-1 cells. All other groups of immunized as well as immune cell depleted mice were challenged with E.G.7 tumor cells. The tumor outgrowth was followed as previously described.

In vitro and in vivo T cell proliferation

Spleens were obtained from naïve OT-I mice, which are TCR transgenic for OVA peptide (257–264) in the context of H-2Kb (SIINFEKL). Single cell suspensions were prepared as before and the splenocytes labeled using Vybrant™ CFDA SE Cell Tracer Kit (Molecular Probes, OR). CFSE labeled OT-I splenocytes were seeded at a density of 1 × 107 cells/well in a 24-well plate in presence of rh-IL2 (10 U/ml) and Ad5-OVA±CpG (1 × 107 pfu ± 50 μg CpG) while adenovirus Ad5-LacZ+CpG was used as a control. These cultures were incubated at 37°C, 5% CO2 in the incubator, and harvested at different time points. For in vivo analysis of CD8 T cell proliferation, 1 × 107 CFSE labelled OT-I splenocytes were adoptively transferred via i.v. injection into naïve C57BL/6 mice. Onthe next day, these recipients were immunized subcutaneously on the flank with Ad5-OVA±CpG or Ad5-LacZ+CpG vaccine, and lymph nodes were removed at different time points. In both cases, cells were stained with PE-tetramer specific for OVA-peptide and fixed with 2% paraformaldehyde to analyze the antigen-specific CD8+ T cell proliferation by collecting 10,000 events of CSFE labelled OT-I cells.

Statistical analysis

Levels of significance and comparisons between the treatments were determined using Student's t-test.


Revelation of a dichotomy between In vitro and in vivo conditions

The immunostimulatory effect of CpG ODN as a strong Th1 adjuvant has been demonstrated in various immunotherapeutic studies. To augment the adaptive T cell response using CpG ODN, we recently demonstrated that coinjection of CpG ODN with an adenovirus vaccine for PSA (Ad5-PSA) dramatically reduced CD8+ T cell response measured in vitro while enhancing the in vivo anti-tumor activity against the antigen-bearing tumor cells.20 Here, we investigated a similar phenomenon in a different mouse model using a well defined antigen (OVA) to explore the clinical importance of a tumor vaccine and CpG ODN combination. C57BL/6 mice were immunized either with Ad5-OVA vaccine or with a mixture of Ad5-OVA and CpG ODN. The control groups of mice were injected with Ad5-LacZ+CpG. Two weeks later spleens were removed from the immunized mice and the functional activity of the developed anti-OVA CD8+ T cell response was measured using 51Cr-release assay against the OVA expressing E.G.7 cells. Similar to our results with Ad5-PSA+CpG, immunization with Ad5-OVA+CpG also produced a lower intensity of OVA-specific lytic activity (Fig. 1a). Splenocytes from the control groups of mice (Ad5-LacZ+CpG) lacked the lytic activity. CTL activity of the splenocytes from all the groups against EL4 (control) target tumor cells was in the range of 0–10% (data not shown).

Figure 1.

Dichotomy between in vitro and in vivo analysis. (a) Representative plot of anti-OVA specific CTL activity against E.G.7 target tumor cells in a 51Cr release assay. Splenocytes obtained from the group of mice immunized with Ad5-OVA+CpG showed reduced target cell killing in vitro compared to Ad5-OVA immunization. These experiments were repeated multiple times. (b) Measurement of anti-tumor responses developed following immunization with Ad5-OVA vaccine with or without CpG ODN. C57BL/6 mice were challenged with 1 × 107 cells of OVA-expressing E.G.7 or control EL4 on day 14 after immunization. Control groups of mice immunized with Ad5-LacZ+CpG, Ad5-OVA or Ad5-OVA+CpG inoculated with EL4 tumor cells showed progressively growing tumor. (c) Percent (%) survival in the same group of mice immunized above. Irrespective of low amount of anti-OVA CD8+ T cell response measured in vitro because of a single injection of Ad5-OVA+CpG, note the extension in survival period with 40–60% mice remain tumor free until 8 weeks. Data from 3 independent experiments (5 mice/group) for E.G.7 tumor challenge with similar results are pooled together. All deaths were due to tumor burden. Error bar represents the SE of the mean.

On the basis of our previous observations, the strength of the generated in vivo immune response was measured against a highly aggressive immunogenic tumor cell line (E.G.7) expressing the OVA to test the ability of the immunized mice to destroy the solid tumors. Ad5-OVA±CpG immunized mice were challenged (on day 14) with injections of 1 × 107 E.G.7 tumor cells or control (EL4) cells. As expected, in correlation with CD8+ T cells response, the group of mice immunized with Ad5-OVA vaccine displayed better tumor protection (Fig. 1b) with longer survival time (Fig. 1c) compared to the control group of mice. However, in contrast to the reduced anti-OVA CTL activity following Ad5-OVA+CpG immunization, these mice demonstrated a significant decrease in tumor outgrowth with an extended survival period. All other groups of mice had large, progressively growing tumors (Figs. 1b and 1c). As clearly viewed in Figure 1b, in vivo tumor outgrowth in the mice immunized with Ad5-OVA vaccine or Ad5-OVA+CpG was remarkably different from the control groups (Ad5-LacZ+CpG) while no difference in tumor outgrowth was observed between the group of mice challenged with non OVA-expressing EL4 tumor cells. Immunization with Ad5-OVA vaccine delayed the tumor growth and enhanced the longevity of the mice up to 5 weeks compared to control groups (3 weeks). Coadministration of CpG adjuvant with Ad5-OVA improved the life expectancy, and 40–60% of the mice remain tumor free. These data suggest that the addition of the TLR9 agonist CpG ODN 1826 augmented antigen-specific immune responses that suppressed in vivo tumor growth, supporting and validating the previous observations for a dichotomy between in vitro measured CTL response and in vivo tumor protection.20

Effectiveness of CD8+ T cell responses

After the establishment of such a dichotomy, we explored the impact of immunization on the presence of IFN-γ, a key inflammatory cytokine produced by effector T cells, NK and NKT cells and considered an indicator of T cell activation. Splenocytes were obtained 14 days after immunization with Ad5-OVA±CpG and single cell suspensions were prepared as previously described. Analysis of IFN-γ cytokine production was performed using various immuno-assays. The results shown in Figures 2a and 2b (ICS) demonstrated that Ad5-OVA vaccine induced a robust increase in IFN-γ secretion by the antigen-specific CD8+ T cells (p = 0.0015). Similar to the reduction of anti-OVA CTL activity, inclusion of CpG with the vaccine reduced the number of CD8+ T cells secreting IFN-γ (p = 0.042). No significant effect on IFN-γ secreting CD8+ T cells was observed on the splenocytes from control groups of mice immunized with Ad5-LacZ+CpG. The presence of IFN-γ secreting cells was also examined by ELISpot assay using unfractionated splenocytes from the immunized group of mice. As clearly viewed in Figure 2c, immunization with Ad5-OVA strongly induced the immune cells to secrete IFN-γ while co-injection of Ad5-OVA+CpG dramatically reduced the number of IFN-γ secreting cells. No effect on IFN-γ was observed in the splenocytes from control groups. Final confirmation for the level of IFN-γ secretion was examined by ELISA assay (Fig. 2d) and correlated with the results obtained from ICS and ELISpot.

Figure 2.

Analysis of vaccine-induced CpG modulated CD8+ T cell response in splenocytes following immunization. (a) Representative of a dot plot expressing anti-OVA immune response by ICS for IFN-γ secreting anti-OVA CD8+T cells after OVA-peptide (SIINFEKL) stimulation; (b) Analysis for the number of IFN-γ secreting OVA-specific CD8+ T cells normalized against unstimulated controls during 4 hr incubation in ICS assay; (c) ELISpot assay for OVA-specific IFN-γ from the unfractionated splenocytes; (d) Analysis of IFN-γ cytokine secretion by ELISA from the co-culture supernatant obtained following 48 hr incubation of splenocytes with OVA-expressing E.G.7 cells. Asterisks represents the level of significant (p < 0.0001) in a paired t-test. In all 3 assays, the number of IFN-γ secreting CD8+ T cells as well as the amount of IFN-γ secretion are significantly less in the group of cells obtained from Ad5-OVA+CpG immunized mice compared to Ad5-OVA alone. The values in figures (b) and (d) are plotted from 4 independent experiments. Error bars represents the SE of the mean.

To further document a decrease in the induction of antigen-specific CD8+ T cells following immunization with Ad5-OVA vaccine plus CpG, we used MHC-I OVA peptide (SIINFEKL) specific tetramers. In agreement with the results shown in Figure 2, there was a significant increase in peptide specific CD8+ T cells (p = 0.005) following Ad5-OVA immunization as compared to control (Ad5-LacZ+CpG) while co-injection of CpG ODN with the Ad5-OVA vaccine causes a remarkable diminution (p = 0.007) in the frequency of antigen-specific CD8+ T cells (Figs. 3a and 3b; upper panel).

Figure 3.

Ex vivo analysis of OVA-specific CD8+ T cells on day 14 in lymphoid organs following immunization. (a) Representative dot plots (Spleen: upper panel; Lymph nodes: lower panel) using MHC class I tetramers for SIINFEKL-peptide; (b) Graphical presentation of the pooled values from 3 independent experiments. There is a significant decrease in total number of OVA-specific CD8+ T cells in the group of mice immunized with Ad5-OVA+CpG compared with Ad5-OVA alone (p values are based on paired t-test).

A theoretically possible explanation for the observed differences was that CpG ODN could have resulted in the migration of antigen-specific CD8+ T cells from the spleen to lymph nodes. To test this hypothesis, mice were immunized as earlier mentioned, and 2 weeks later cell suspensions from the draining lymph nodes were analyzed for tetramer-specific CD8+ T cells. These observations showed that inclusion of CpG ODN with the Ad5-OVA did not cause an alteration in trafficking of the lymphocytes to lymph nodes (Figs. 3a and 3b; lower panel).

Effector cell population mediating CTL function

With the observations that immunization with a mixture of tumor vaccine and CpG ODN suppresses the in vitro CD8+ T cell responses but enhance the ability of mice to protect against the tumor challenge in vivo, it became critical to investigate whether CpG ODN has caused the activation of lymphocyte population(s) other than CD8+ T cells. To delineate the functional activity of the immune cells in vivo, we performed immunodepletion studies using monoclonal antibodies for CD8 (clone 2.43), CD4 (clone GK1.5) or NK cells (clone PK136). Antibody injections were given on day 10 and 12; on day 14 tumor cells (E.G.7 or EL4) were injected in these mice. Tumor outgrowth and survival period were compared between the mice depleted for various immune cells. The results indicated that CD8+ T cells were the primary effector cell population associated with the augmented anti-tumor immunity, since depletion of CD8+ T cells in Ad5-OVA+CpG immunized mice abrogated the anti-tumor effect (Figs. 4a and 4b). It also appears that NK1.1 cells played a minor role in developed anti-tumor immunity, since 50% of NK-depleted mice died because of tumor burden compared to 20% in Ad5-OVA+CpG immunized group (control), whereas depletion of CD4 T cells did not contribute, as the tumor outgrowth was minimal with 100% survival rate during the course of experiment (Fig. 4b). Similar results for tumor outgrowth were obtained in immunodepleted mice immunized with Ad5-OVA alone (data not shown).

Figure 4.

In vivo tumor growth (a) and survival (b) in the mice depleted of immune cell populations after immunization. CD4, CD8 or NK1.1 immune cells were depleted with their corresponding monoclonal antibodies (MAbs). Cell depletion was verified (as described in methods) on the day of tumor challenge with E.G.7 (1 × 107) cells and tumor growth was monitored. Immune cell depletion was maintained during the course of experiment by the injection of MAbs 2 times/week. Mice immunized with vaccine+CpG and depleted for CD8+ or both for CD4+ and CD8+ T cells succumbed to death almost simultaneously as the control (Ad5-LacZ+CpG) group of mice. A death rate of 50% was also observed in absence of NK1.1 cells compared to a 20% rate of death of mice immunized with Ad5-OVA+CpG, while 100% of the mice depleted for CD4+ T cells survived during the course of the experiment. All deaths were due to tumor burden.

Kinetic study for CD8+ T cell response following primary and secondary antigenic challenge

Until now we presented the data on the CD8+ T cell response at one time point (14 day), while CpG coadministration with vaccine developed an extended tumor protection. Since development of an effector CD8 T cell response is a very dynamic process, it is possible that effectors predominate during the early phase and memory cells develop at later time points. Therefore, addition of CpG ODN with the vaccine may have influenced the peak of effector CD8+ T cell response. To examine such a possibility, we performed a time course study starting day 3, 5, 7 and later to measure the functional activity and the number of anti-OVA CD8+ T cells using in vitro CTL and ex vivo peptide-specific tetramer assay. On day 3 and 5, CD8+ T cell responses were not detectable in vitro. On day 7 or 8, we could detect the immune response, however, to a lesser degree compare to day 10 or later. Significantly lower proportions of anti-OVA specific cytolytic activity (Fig. 5a) as well as total number of MHC-I tetramer specific CD8+ T cells (Fig. 5b) were obtained at all time points in mice that received a mixture of Ad5-OVA+CpG compared to the mice immunized with Ad5-OVA alone. The spleen cells from control group of mice (Ad5-LacZ+CpG) lacked the effective killing of target cells. The lytic activity of the effectors from Ad5-OVA±CpG immunized was <10% against EL4 (control) target tumor cells (data not presented in the graphs).

Figure 5.

Dynamics of CD8+ T cell responses. (a) In vitro CTL assay as described in Figure 1a; (b) Percentage of SIINFEKL-specific CD8+ T cells in a tetramer assay following primary challenge (vaccine immunization). In both experiments the magnitude of the anti-OVA CD8+ T cell response remains low in presence of CpG ODN. (c, d) Immunized mice were inoculated subcutaneously with E.G.7 tumor cells on day 14 and the spleens were obtained subsequently at various time points. Even in post tumor analysis (secondary antigenic challenge) the number of anti-OVA CD8+ T cells secreting IFN-γ (ICS: Fig. 5c) and SIINFEKL-specific CD8+ T cells in a tetramer assay (Fig. 5d) remains low in the group of mice that received Ad5-OVA+CpG. These experiments were repeated twice with similar results.

Interestingly enough, despite ex vivo/in vitro observed lower antigen-specific CD8+ T cell response, CD8+ T cells were primarily accountable for enhanced tumor protection in the group of mice immunized with Ad5-OVA+CpG. It was still possible that immunized mice (primary antigenic challenge) followed by subsequent tumor implantation (secondary antigenic challenge) could develop an increased CD8+ T cell response. To address this question, mice were immunized with Ad5-OVA±CpG along with appropriate controls. On day 14, these mice were inoculated subcutaneously with E.G.7 tumor cells, and CD8+ T cell responses (peptide-specific tetramer and ICS for IFN-γ) were analyzed at selected time points. These results revealed that even after secondary antigenic challenge, there was no increase in anti-OVA CD8+ T cell response in the group of mice immunized with Ad5-OVA+CpG (Figs. 5c and 5d). Paradoxically, in all the possible combinations, effectiveness of anti-OVA CD8+ T cell response was always lower when measured in vitro whereas the mice generated an augmented in vivo tumor protection in CD8+ T cell dependent manner following a combined approach of immunization.

Impact of CpG ODN on CD8+ T cells In vitro and in vivo

To gain an insight into the CD8+ T cell response to CpG, we investigated the rate of proliferation of CFSE labeled OT-1 splenocytes incubated with Ad5-OVA vaccine alone or in combination with CpG ODN for in vitro analysis or adoptively transferred CFSE labeled OT-I cells in mice followed 1 day later by Ad5-OVA±CpG immunization. On the day of harvesting the cells from in vitro culture or from the recipient's lymph nodes were stained with OVA-specific tetramer to examine the anti-OVA specific CD8+ T cell expansion. In agreement with the reduced CD8+ T cell responses measured in vitro/ex vivo (Figs. 1a and 2), anti-OVA CD8+ T cell proliferation was remarkably inhibited in the presence of CpG ODN both in vitro and in vivo (Fig. 6).

Figure 6.

TLR9 agonist suppresses the antigen-specific CD8+ T cell proliferation. (Upper panel): Proliferation of CFSE-labelled OT-I splenocytes incubated in vitro with Ad5-OVA vaccine alone (solid line) or in presence of CpG (shaded area) and stained for SIINFEKL-specific tetramers on the day of harvest. (Lower panel): CFSE-labelled OT-I splenocytes were adoptively transferred into syngeneic C57BL/6 mice. One day later recipients were immunized with Ad5-OVA vaccine (108 pfu) with or without CpG (50 μg) while control group of mice were injected with Ad5-LacZ+CpG (broken line). Lymph nodes were obtained at different time points after immunization to analyze the anti-OVA specific CD8+ T cells proliferation by collecting the CFSE-labelled cell events. Read-outs are the representatives of 3 independent experiments.


Various therapeutic approaches are being developed and tested to improve the treatment for cancer patients. Along with the vaccines, use of adjuvants is well recommended to further enhance the immune response both qualitatively as well as quantitatively. In this context, CpG ODNs are considered potent stimulators of innate and adaptive immune responses and are widely used in pre-clinical and clinical studies.13, 15, 22, 23 Here, we demonstrated that immunization with Ad5-OVA alone strongly induced the generation of antigen-specific CD8+ T cells, while in an attempt to further augment the T cell immune response, addition of CpG ODN to the Ad5-OVA vaccine suppresses the anti-OVA CD8+ T cell responses. When comparing the strength of generated immune responses following immunization with Ad5-OVA±CpG, it appeared that despite reduced anti-OVA CD8+ T cell response because of coinjection of CpG, mice developed an augmented anti-tumor immunity. These observations extend our previous study in a number of ways that (i) the observation is not unique to the PSA vaccine, but occurs with an Ad5-OVA vaccine in an identical manner; (ii) there is a reduction in total number of antigen-specific CD8+ T cells, not only T cells involved in CTL function and IFN-γ production; (iii) using cell depletion experiments, identifies the cells responsible for in vivo tumor destruction as CD8+ T cells; (iv) demonstrate that the reduced in vitro activity was observed from early to very late after immunization; (v) demonstrate by tetramer staining that mice immunized with Ad5-OVA + CpG had little to no expansion and contraction of antigen-specific CD8+ T cells as did mice immunized with Ad5-OVA alone; (vi) the reduction in anti-OVA CD8+ T cells is independent of the presence of OVA-expressing tumors; and (vii) that the reduction in anti-OVA CD8+ T cells is a function of reduced cell proliferation.

Various studies have shown that generation of powerful CTL response, trafficking of antigen-reactive T cells in circulation or into the lymph nodes, migration of antigen-reactive T cells to the site of infection, are certain possible mechanisms documented to eliminate the tumor development.24 While examining the trafficking of effector cell population to other lymphoid organs because of CpG ODN, it was evident that the decrease in antigen-specific CD8+ T cell response from the spleen is not due to the migration of anti-OVA CD8+ T cells to lymph nodes following immunization in presence of CpG. Similarly, a lower number of antigenic CD8+ T cells were observed in the blood from the Ad5-OVA+CpG immunized mice (unpublished data).

CpG ODN has the potential to activate a variety of different immune cells including NK or B cells that can contribute to anti-tumor activity.25 In experiments performed to determine the necessity of antibodies against OVA for the enhanced anti-tumor activity, we were not able to detect the level of IgA, IgM, IgG2a, b and c antibodies in serum samples from the group of mice immunized with Ad5-OVA+CpG (unpublished data). Such observations exclude the involvement of antibody mediated T cell cytotoxicity to the effector T cell function in the present scenario. Analyzing the participation of T cells in tumor clearance, immunodepletion studies showed that enhanced antigen-specific tumor clearance was predominantly CD8+ T cell-dependent whereas CD4+ T cell depleted mice were fully protected against the tumor challenge. However, a minor contribution of NK cells in tumor clearance was observed.

Although a correlation between in vitro immune response and in vivo tumor protection was not evident, it was possible that the magnitude of the CD8+ T cell response could be influenced by the vaccination in presence of CpG because of very dynamic nature of T cells.26, 27, 28 An initial burst of primary response determines the larger size of adaptive T cell response at the later stage and may provide improved anti-tumor activity.29 Alternatively, CpG ODN can also drive the CD4+ T cell response during the primary antigenic challenge to provide a base to augment the CD8+ T cell response during the secondary antigenic challenge (post tumor inoculation). In both cases, neither a higher antigenic CD4+ T cell response following primary immunization nor an enhanced anti-OVA CD8+ T cell response following tumor (secondary) challenge was observed.

In clinical settings, one of the approaches to immunize patients is to transfer the autologous tumor reactive T cells that produce higher IFN-γ and possess higher cytotoxic activity against the antigen-bearing tumor cells in vitro.30 However, due to the low success rate of immunotherapy, there has been a growing interest in exploring the number and nature of effector CD8+ T cells in vitro and in vivo. Some studies revealed that following immunization the patients that showed regression of tumors actually have low number of tumor reactive T cells compared to the patients having higher number of reactive T cells without any tumor regression.31, 32 It is also reported that fully functional effecter CD8+ T cells generated in vitro were less effective in vivo compared to the naïve early effector CD8+ T cells.33, 34 Retrospective analyses in melanoma patients revealed that generated CD8+ T cells for enhanced IFN-γ and cytolysis induced the clinical response only in a small minority of the patients.35, 36 In the present study, we clearly demonstrated a phenomenon that theoretically so called fully functional CD8+ T cells with higher CTL and IFN-γ production (in vitro) generated because of Ad5-OVA immunization paradoxically possess less in vivo anti-tumor immunity compared to the CD8+ T cells with reduced in vitro CTL activity and IFN-γ production following Ad5-OVA+CpG immunization. These data demonstrate that higher frequencies of tumor-specific IFN-γ-secreting CD8+ T cells do not necessarily ensure an augmented tumor protection, and should not be considered as a surrogate marker for vaccine efficacy.

DCs are the key regulators of immune effector function and can promote or suppress anti-tumor T cell responses depending upon the cytokine milieu and the signaling cascades that are activated. DC-based immunotherapeutic studies using tumor antigen to induce protective immune responses against the antigen-expressing tumor cells have been well documented.4, 37, 38 Both of the reagents used in the present study (adenovirus vaccine and CpG ODN) are separately proficient at enhancing DC maturation and stimulatory effects.10, 23, 39 To the best of our knowledge, the presence of CpG ODN in subcutaneous immunization protocols such as ours uniformly has been reported to promote the T cell expansion.16, 40, 41 Similarly, the presence of CpG ODN in intravenous or intraperitoneal immunization settings has been shown to induce CD8 T cell response.17, 42 In clinical studies, use of CpG ODN generates a strong peptide-specific CD8+ T cell response.12, 22, 43 To gain insight into the impact of CpG on CD8+ T cells, we observed that CpG ODN suppresses the rate of proliferation of antigen-specific CD8+ T cells both in vitro and in vivo. These observations at least partly explained the lower number of IFN-γ secreting CD8+ T cells, since production of IFN-γ depends on the rate of T cell proliferation.28, 44 A decrease in antigen-specific CD8+ T cells has also been reported using a mixture of adenovirus containing OVA or carcinoembryonic antigen (CEA) and CpG ODN.18, 19 In recent studies a critical role was reported for the route of the CpG injection in determining whether CpG would have immune suppressive effects (mediated via Treg induction) or immune stimulatory effects.18, 45

In contrast to adenoviral vaccine, co-injection of CpG ODN with plasmid vectors has been associated with increase CTL response.46, 47, 48 A few reports in prime-boost protocols described the use of recombinant vaccinia virus following antigenic peptide and CpG ODN to promote the CTL responses.49, 50 It will be interesting to understand if the dichotomy (suppressed CD8+ T cell responses of enhanced anti-tumor immunity) is unique to adenovirus in combination with CpG ODN since adenovirus is a DNA virus and may activate TLR9 on its own or other TLR ligands that interact with TLR9 pathway. DCs and macrophages are known to be able to cross-present virus like particles. It may be possible that virus like particles (as adenovirus in this case) facilitate cross-presentation by APCs, which are being simultaneously activated by CpG. This may lead to the presentation of antigen by activated APCs to induce an optimal CD8+ T cell response.51, 52, 53 Therefore, powerful induction of protective anti-tumor response in combinations of adenovirus with CpG may be due to altered processing capabilities of DCs and macrophages. Nonetheless, use of such a combined approach of immunization may permit the extension of this protocol in association with solid tumors of defined antigen to potentially provide a unique strategy for the generation of more effective in vivo adaptive T cell immunity, despite their lower in number.

Conclusively, this study highlights several unique observations with clinical importance that: (i) the combination of vaccine and CpG improves both survival and tumor growth kinetics; (ii) standard techniques for monitoring the immunological effects of vaccination failed to predict a benefit for the combination approach; (iii) coinjection of CpG diminishes the adaptive CD8+ T cell response by suppressing the anti-OVA CD8+ T cell expansion; (iv) challenges the assumption that a bigger T cell response is better in vivo. It is likely that the immunization mixture (Ad5-OVA+CpG) produced a microenvironment to select the CD8+ T cells that are advantageous in pathological conditions. Unnecessarily higher number of CD8+ T cells can cause a damage to host tissue system or can contribute towards the autoimmune disease, and a higher amount of IFN-γ may exert a pro-apoptotic effect on CD8+ T cell expansion.54, 55 It is apparent that developing an immunization protocol to selectively induce antigen-specific CD8+ T cell responses and enhanced anti-tumor immunity could prove a novel strategy in the field of immunotherapy where labor intensive, cost effective and potentially risky protocols are used to generate in vitro effector CD8+ T cells. We continue to explore the nature of the CD8+ T cells optimally required to generate effector function in vivo following this combined approach of immunization. It is possible that other soluble factors secreted by various immune cells play a critical role to provide microenvironment in favor of CD8 T cells. It remains a mystery though how a small proportion of antigen-specific CD8+ T cells can have higher in vivo effector function than a larger proportion.


The authors thank Ms. Kristina Greiner for editing the manuscript.