Regulatory T-cell depletion synergizes with gp96-mediated cellular responses and antitumor activity

Abstract

Despite its potent immunostimulatory properties, vaccination with autologous tumor-derived gp96 has relatively modest antitumor effect in a range of clinical trials. Based on our previous study showing a gp96-mediated immune balance between CTL and Tregs, here we investigated possible synergy between gp96 vaccine and systemic Treg depletion on induction of antitumor T-cell immunity and the mechanisms accounting for synergistic efficacy. In gp96–peptide complex immunized BALB/c mice, anti-CD25 mAb treatment significantly increased IFN-γ-producing CD8⁺ and CD4⁺ T cells by about 1–2-fold in spleen and 40–50% in lymph node. A significantly higher number of peptide-specific CTL were observed under anti-CD25 mAb treatment compared with no treatment. Moreover, Treg depletion synergistically improved the anticancer activity of tumor-derived gp96 vaccine in the poorly immunogenic and highly tumorigenic B16 melanoma model in C57BL/6 J mice. While gp96 immunization alone led to the modest enhancement of CTL activities in spleen, the combination with Treg depletion dramatically increased tumor-specific CTL responses. In addition, the combination resulted in a significant increase of CD8⁺ T-cell infiltration in tumor, which correlated with an enhanced inhibition of tumor growth. Our results provide evidence that targeting Tregs may provide a more efficient strategy to potentiate gp96-mediated T-cell responses and enhance the antitumor efficiency of gp96-based therapeutic vaccine.

Introduction

Heat-shock protein (HSP) gp96 is a member of HSP90 family, residing within the lumen of the endoplasmic reticulum (ER). It has been implicated as an essential ER player for chaperoning peptides generated by proteasomal degradation within the cell and then transferring peptides to MHC class I molecules [1–3]. In rodent models, the autologous gp96–peptide complexes isolated from tumor tissues have been shown to elicit potent activation of innate and adaptive immunity and generate antitumor responses in both poorly immunogenic (e.g. B16) and immunogenic (e.g. Meth A) tumor models, acting as powerful preventive and therapeutic vaccines [4, 5]. The mechanisms of immunoregulatory and adjuvant functions of gp96 are supposed to be its presentation of bound peptides to MHC class I and class II molecules, leading to the activation of specific CD8⁺ and CD4⁺ T cells [6–9], and/or its activation of macrophages and dendritic cells through interaction with a subset of Toll-like receptors (TLRs) and induction of proinflammatory cytokines [10–13], although alternative molecular mechanisms of gp96-mediated immunoresponses have also been suggested [14, 15]. Owing to its remarkable immunogenicity, autologous gp96–peptide complexes as therapeutic vaccines have been initiated in clinical trial for the treatment of a range of tumors and disease stages, including Phase I and II trials in glioma, melanoma, etc., and two Phase III studies in melanoma and renal cell carcinoma [16–20]. These studies showed that treated patients demonstrated significant T-cell immune responses and a near-significant improvement in the survival of patients with earlier-stage disease. Despite its potent immunostimulatory properties, the antitumor effect of gp96 has been relatively modest in these clinical trials, indicating that poorly characterized immunosuppressive mechanisms may enter to play in patients with established tumors [16, 17, 21–24]. Given the apparent dependence of gp96 on CTL-mediated antitumor effects, it is critical to further explore the mechanisms that affect T-cell function during gp96 immunization. This will allow physicians to identify the patient group that will gain the most benefit from the immunotherapy and to enhance gp96-based therapeutic efficacy by combination with immunomodulatory agents that either promote CTL responses and/or abrogate the suppressive effect of tumor immune evasion.

Regulatory T cells (Tregs), characterized by high expression of IL-2R α chain (CD25) and intracellular expression of the forkhead/winged helix transcription factor Foxp3, play an important role in the maintenance of self-tolerance, as well as immune tolerance to tumors and pathogens because of their vigorous inhibitory action on effector T cells, NK cells, B cells and other immune cells [25]. Many publications have reported that there are a higher proportion of Treg cells in the tumor-infiltrating lymphocytes (TIL), which suppress effector T cells and nature killer T (NKT) cells by direct cell–cell contact or the production of soluble immunosuppressive factors such as IL-10 and TGF-β in the tumor microenvironment [26]. Therefore, targeting Tregs has the potential to impact multiple pathways responsible for dampening immune responses and enhance the efficacy of tumor immunotherapy. Elimination of Tregs using anti-CD25 monoclonal antibody has been reported to elicit potent immune responses and lead to tumor eradication [26, 27]. However, it has also been shown that Treg depletion with anti-CD25 antibodies failed to enhance specific T-cell activation and tumor rejection in both melanoma mouse model and clinical trials, suggesting that timing and dosing as well as Treg specificity may affect the therapeutic efficiency of Treg inactivation [28].

Several studies have reported that high dose of gp96 immunization may down-regulate inflammatory and cellular responses and decrease its antitumor effect, suggesting that myeloid suppressor cells or Tregs may restrict gp96-induced activation of CD4⁺ and CD8⁺ T cells [29–31]. Enhanced suppressive function of Treg by gp96-mediated activation of TLR4 has been demonstrated in a transgenic mouse model [32]. Moreover, our previous study showed a gp96-mediated immune balance between CTL and Tregs, indicating that activated Treg may abrogate gp96-induced T-cell responses [33]. Based on the observations above, the aims of this study are to determine whether gp96 immunization with systemic Treg depletion induces potent CD8⁺ T-cell responses and has additive or synergistic inhibition of the growth of established tumor. We first monitored the activation of peptide-specific T cells under Tregs inactivation in gp96–peptide complexes immunized BALB/c mice. We then investigated the effect of combination treatment with gp96 tumor vaccine and anti-CD25 antibody on antitumor immunity and the mechanisms accounting for synergistic efficacy in the B16 melanoma mouse model. The results indicate that the success of gp96-based immunotherapy depends on the balance between effector T responses and suppressive immune mechanisms. Targeting Tregs may provide an efficient strategy to potentiate gp96-mediated T-cell responses by the elimination of tolerogenic or immune suppressor influences.

Materials and methods

Immunization of mice and splenocytes isolation

HBcAg_87–95 K^d-restricted CD8⁺ T-cell epitope SYVNTNMGL and control peptide HBcAg_18–27 FLPSDFFPSV were chemically synthesized by Jier Biological (Shanghai, China). The purity (>95%) and molecular weight of the peptides were determined by high-performance liquid chromatography and mass spectrometry. HSP gp96 was obtained from healthy murine livers as described previously [34, 35]. HSP gp96–peptide complexes were generated by incubating gp96 protein and the peptide at 50°C for 10 min, followed by 30 min at room temperature. Female BALB/c mice (6–8 weeks old) were immunized s.c. with 50 μg of HBcAg_87–95 peptide bound to 20 μg of gp96 at 1, 3 and 4 week, respectively. One group of mice was injected intravenously with 30 μg of anti-CD25 mAb (PC61 clone, BioLegend, San Diego, CA) every 2 weeks for twice, starting 3 days prior to gp96 immunization. Control mice received non-specific rat IgG1. All groups contained at least five mice. Mice were killed 2 days after the third immunization. Splenocytes were isolated as previously described [33].

IFN-γ ELISPOT assay

The enzyme-linked immunosorbent spot (ELISPOT) assay for enumeration of antigen-specific, IFN-γ secreting cells was performed according to manufacturer’s instruction. In brief, 96-well PVDF plates (BD-Pharmingen, San Diego, CA) were precoated with the coating Ab overnight at 4°C and blocked for 1 h at 37°C. A total number of 10⁶ splenocytes and 20 μg of HBcAg_87–95 peptide were added to each well. PMA/ionomycin was employed as positive control and the control peptide HBcAg_18–27 as negative control. Each test condition was performed in triplicate. The plate was incubated at 37°C for 20 h. The number of IFN-γ spots was counted by the ELISPOT Reader (Biosys, Germany).

Flow cytometric analysis

The indicated populations of T cells (1–3 × 10⁶ splenocytes per sample) were collected and blocked with 5% BSA at 4°C for 20 min. Then, various fluorochrome-conjugated Abs against interested surface markers were added to cell pellets, incubated for 30 min on ice and washed 3 times before analysis. PerCP-Cy5.5-conjugated anti-mouse CD3, FITC-conjugated anti-mouse CD4 and CD8, PE-conjugated anti-mouse CD25 and CD4, APC-conjugated anti-mouse CD8, Foxp3 and IFN-γ were purchased from eBioscience. Foxp3 and IFN-γ intracellular cytokine staining were performed using the Cytofix/Cytoperm kits (eBioscience and BD-Pharmingen). Samples were analyzed using FACSCalibur and CellQuest software (BD Biosciences) after staining.

Pentamer staining

Purified splenocytes (1 × 10⁷ cells/well) were stimulated with 20 μg/ml of HBcAg_87–95 peptide at 6-well plates for 5 days. Mitomycin-C-treated P815 cells served as feeder cells. PE-labeled H-2 K^d Pentamer complexes loaded with the peptide HBcAg_87–95 were purchased from ProImmune (Oxford, United Kingdom). Cultured splenocytes cells were stained with PE-labeled H-2 K^d Pentamer, PerCP-Cy5.5-conjugated anti-mouse CD3 and APC-conjugated anti-mouse CD8, respectively, and detected with four-color flow cytometric analyses.

Tumor model

Autologous gp96 as the therapeutic tumor vaccine was purified from B16.F10 melanoma tumors tissues as described [34, 35]. Female C57BL/6 J mice (6–7 weeks old) were challenged subcutaneously with 3 × 10⁴ B16.F10 melanoma cells/mouse and monitored for tumor progression. When the tumor size reached 50 mm³, mice received 4 subcutaneous B16 tumor-derived gp96 or normal liver-derived gp96 (as control) vaccinations (20 μg/mouse) at a 3-day interval, with or without a single dose of 30 μg anti-CD25 mAb or 0.4 mg of cyclophosphamide (Sigma-Aldrich, USA) per mouse, respectively. Control animals received a single dose of 30 μg control Ab or PBS. Tumor diameter was measured at regular intervals.

Tumor-infiltrating lymphocytes (TIL) isolation

Tumor-infiltrating lymphocytes were isolated as previously reported [36]. In brief, tumor specimens were minced into small pieces and incubated for 2 h in RPMI 1640 containing 0.05% collagenase type IV and 0.002% DNase type I at 37°C. The single-cell suspension was passed through a sterile wire mesh grid to remove undigested debris, and cell pellets were suspended in 80% Percoll, overlaid with 40% Percoll and centrifuged at 2000 g for 30 min. Cells at the interface were collected, washed and used for flow cytometric analysis.

Immunohistochemistry

Tumor samples were fixed in 4% methanal followed by paraffin embedding. Consecutive sections 5 μm in thickness were deparaffinized with xylene and stained with mouse anti-mouse CD8 mAb followed by secondary biotinylated horse anti-mouse Ab. After staining with Vectastain ABC-HRP (Vectastain ABC kit; Vector Laboratories) according to the manufacturer’s protocol and with AEC substrate, the sections were counterstained with hematoxylin staining solution. For the enumeration of cells, high-powered fields (hpf, × 400, Olympus BX51) were used for counting positive cells. Only brightly colored cells were counted. For each tumor, means were calculated for the positive cell numbers in 3 microscopic fields distributed over three sections by the two independent observers.

Cytotoxicity assay

CFSE-based cytotoxicity assay was performed as described previously [37]. In short, B16.F10 melanoma cells were labeled with 2 μM CFSE as the target cells, and then the cells were seeded into a 96-well U-bottom microtiter plate. CTLs were added at different effector/target ratios: 5:1, 10:1 and 20:1. Plates were then incubated in a humidified atmosphere of 5% CO₂ for 4 h. The cells were harvested, and apoptosis assay was performed by PI staining using Vybrant^® Apoptosis Assay Kit (Invitrogen, USA).

Statistical analysis

Student’s t-test was used for comparison between groups. A P value <0.05 is considered a significant difference. Pearson correlation analysis was performed between Treg and tumor weight. The combination index (CI) method according to Chou and Talalay [38] was used to analyze the synergistic effect. The antitumor effect of gp96 vaccine and anti-CD25 mAb was analyzed using the software CalcuSyn (Biosoft, Ferguson, MO), which applies the median-effect equation of Chou and the CI equation of Chou and Talalay. Calculated CIs were used to determine the presence of synergism (<1.0), additive effect (equal to 1.0) or antagonism (>1.0).

Results

Treg depletion enhances gp96-induced antigen-specific T-cell activation

We first tested the ability of the anti-CD25 mAb (PC61) for Treg depletion. Female BALB/c mice were injected with 30 μg of anti-CD25 mAb, and 3 days later, the extent of Treg depletion was determined by flow cytometry. The antibody injection dose was based on the titration experiment showing that injection with 30 μg of anti-CD25 mAb was enough to eliminate most Tregs in BALB/c mice. As can be seen in Fig. 1a, 3 days after injection of anti-CD25 mAb, more than 97% of CD4⁺CD25⁺Foxp3⁺ Tregs were effectively eliminated in the spleen. Treg depletion with anti-CD25 mAb did not significantly influence the expression of Foxp3. Depletion of Tregs in spleen and peripheral blood of mice could last for at least 2 weeks after anti-CD25 mAb treatment.

Fig. 1.

Anti-CD25 mAb treatment increases T-cell responses in mice immunized with gp96–peptide complexes. a Anti-CD25 mAb treatment reduces the number of CD4⁺CD25⁺Foxp3⁺ Treg cells. BALB/c mice were injected with either 30 μg of anti-CD25 mAb or control IgG1 mAb, and 3 days later, splenocytes were stained with CD3, CD4, CD25 and intracellular Foxp3. CD25 and Foxp3 expression of CD3⁺CD4⁺ cells are shown by flow cytometric analysis. b–f BALB/c mice were immunized three times with gp96–peptide complexes or peptide (PBS) as control, and an additional gp96-immunized group received two intravenous injections of anti-CD25 mAb at 2-week interval, starting 3 days prior to gp96 immunization (five mice per group). Two days after the third immunization, flow cytometric analysis was performed to identify CD4⁺CD25⁺Foxp3⁺ Treg (b), IFN-γ-producing CD8⁺ (c) and CD4⁺ T cells (d) in spleen and lymph node, and CD8⁺CD3⁺ cells in spleen were further analyzed for MHC class-1/K^d Pentamer (e). In the meantime, splenocytes at 1 × 10⁶ cells/well were stimulated with HBcAg_87–95 peptide and analyzed by IFN-γ ELISPOT assay (f). Data show mean ± SD of five mice. Student’s t-test was used to determine P-values. *P < 0.05 and **P < 0.01. Data are representative of three independent experiments

To understand how Tregs might influence gp96-mediated T-cell responses, we examined the effect of Treg inactivation on antigen-specific T-cell activities. HBcAg_87–95 peptide is a well-characterized CTL epitope that can bind with gp96 effectively [33]. A three-dose immunization schedule with gp96-HBcAg_87–95–peptide complexes was performed at 1, 3 and 4 week, together with anti-CD25 mAb or control Ab treatment at 1 and 3 weeks, respectively. Splenoctyes and lymph node cells from each treatment group were analyzed by flow cytometry for the presence of Treg, IFN-γ-producing CD8⁺ and CD4⁺ T cells and peptide-specific CTL. Compared with the PBS control (no immunization), gp96 immunization increased Treg populations in spleen and lymph node by 20.6 and 23.5%, respectively (both P < 0.05), which is consistent with our previous study [33]. In gp96-immunized mice, anti-CD25 mAb treatment significantly decreased the frequencies of Treg by about one-half in both spleen and lymph node 12 days after treatment (Fig. 1b). With regard to T-cell responses, gp96-immunized mice treated with anti-CD25 mAb exhibited significant increase in IFN-γ-producing CD8⁺ (Fig. 1c) and CD4⁺ T (Fig. 1d) cells by 222.1 and 132.7% in spleen, and 46.6 and 58.3% in lymph node, respectively, compared with no anti-CD25 treatment (all P < 0.01). A significant higher number of peptide-specific CTL (9.90 ± 0.95 versus 4.75 ± 0.75, P < 0.01) by pentamer analysis were observed under anti-CD25 mAb treatment compared with no treatment (Fig. 1e). Similar results were obtained in the ELISPOT assay (Fig. 1f). These data indicate that Treg depletion significantly enhances gp96-mediated CTL response in BALB/c mice.

Increase of Treg in the spleen of B16-F10 tumor-bearing mice

The well-characterized B16 melanoma tumor was chosen as a model in the present study because this poorly immunogenic and highly tumorigenic tumor is infiltrated with Tregs, and mice carrying established tumors are notoriously difficult to cure [39, 40]. Although previous studies reported accumulation of CD4⁺CD25⁺ Tregs inside the B16 tumor, no compelling investigations are available on Treg expansion in lymphoid organs in the mouse system. We studied the presence of Treg population in the spleen of mice bearing B16-established tumors. Female C57BL/6 J mice (6–7 weeks old) received subcutaneous injection of B16.F10 melanoma cells (2 × 10⁴ cells/mouse), and tumor growth was monitored after 2 weeks. A significant difference in the percentage of Treg was observed in the spleen between tumor-bearing and tumor-free mice (7.21 ± 0.34 vs. 4.37 ± 0.91, P < 0.01) (Fig. 2a). Interestingly, Pearson analysis showed that there was a positive correlation between Tregs and tumor weight in tumor-bearing mice (r = 0.94, P < 0.05) (Fig. 2b). The results suggest that Treg may have important impact on B16 melanoma growth.

Fig. 2.

Treg increases in spleen of B16-F10 tumor-bearing mice. Female C57BL/6 J mice received subcutaneous injection of B16.F10 melanoma cells (2 × 10⁴ cells/mouse), and tumor growth was monitored. a Flow cytometric analysis was performed to identify CD4⁺CD25⁺Foxp3⁺ Treg in spleen of tumor-bearing mice (n = 10) and tumor-free mice (n = 10). b Treg frequency in spleen positively correlates with tumor weight in tumor-bearing mice (n = 10). The correlation between Treg frequency and tumor weight was analyzed using Pearson correlation analysis. **P < 0.01. Data show mean ± SD from a representative experiment that was repeated two times with similar results

Anti-CD25 synergizes with gp96 tumor vaccine to suppress B16 tumor growth

We next tested autologous gp96 therapeutic tumor vaccine combined with two different Treg-depleting drugs, anti-CD25 mAb and low dose of cyclophosphamide to treat big B16 melanoma tumors. Female C57BL/6 J mice (6–7 weeks old) received B16.F10 melanoma cells (3 × 10⁴ cells/mouse) on day 0. When the subcutaneous tumors reached 50 mm³, the mice were immunized with gp96 tumor vaccine at a 3-day interval, along with a single-dose treatment with anti-CD25 mAb or cyclophosphamide. Tumor diameter was measured every other day. Mice were killed when they became moribund, and tumor weight was measured. Consistent with reported findings [4, 5], compared with no treatment (PBS), tumor growth was slowed in mice treated with B16 tumor-derived gp96 but not in those treated with normal liver-derived gp96 (Fig. 3a), demonstrating the tumor-specific immunogenicity of B16-gp96. Anti-CD25 mAb treatment alone did not affect tumor growth. In contrast, mice receiving the combination treatment of gp96 vaccine and anti-CD25 had significantly reduced tumor burden compared with either treatment alone (P < 0.01) (Fig. 3a, b). Anti-CD25 treatment led to decrease in tumor weight by 51.9% in gp96-immunized mice. The synergistic effect on the tumor suppression of gp96 vaccine and anti-CD25 was evaluated. All the CI values were less than 1.0 at 8, 10 and 12 days after treatment, which was an indicative of synergism in tumor inhibition between gp96 vaccine and anti-CD25 treatment (Table 1). Co-treatment with gp96 vaccine and anti-CD25 significantly enhanced the survival of tumor challenged mice through 50 days of observation compared with either treatment alone (both P < 0.01) (Fig. 3c). These results clearly show that gp96 tumor vaccine and anti-CD25 treatment had a synergistic effect on B16 tumor inhibition.

Fig. 3.

Anti-CD25 mAb synergistically enhances the antitumor efficacy of gp96 tumor vaccine. Female C57BL/6 J mice were injected subcutaneously with B16.F10 melanoma cells (3 × 10⁴ cells/mouse). When the subcutaneous tumors reached 50 mm³, mice were immunized 4 times with tumor-derived gp96 or liver-derived gp96 vaccine at a 3-day interval or treated with a single dose of anti-CD25 mAb (Anti-CD25) or PBS (PBS) as control on day 0. Some mice given the vaccine were also treated with a single dose of anti-CD25 mAb (B16-gp96+Anti-CD25) or control IgG1 (B16-gp96+IgG1 and Liver gp96+IgG1), or cyclophosphamide (B16-gp96+cy). a Tumor diameter was measured every other day for total of 12 days. Each curve represents the tumor volume in a single mouse. b Tumor weight was measured when mice were killed on day 12. Data show mean ± SD of seven mice. c Kaplan–Meier plot of mouse survival under different treatment. Each treatment group contained 10 mice. Mice were monitored for 50 days for survival. Treatment groups were compared with the log-rank test. **P < 0.01. These experiments were repeated at least twice with comparable results

Table 1.

Combination index values of tumor growth inhibition for co-treatment with anti-CD25 Ab and gp96 tumor vaccine in C57BL/6 J mice

	Days after treatment
	2	4	6	8	10	12
Combination index valuea	>1	>1	>1	0.943	0.971	0.867

^aCombination index values shown are from representative experiments repeated 2 times with similar results

Elimination of Treg enhances B16-gp96-induced tumor-specific CTL responses and CD8⁺ T-cell infiltration in tumor

To understand the effect of Treg depletion on CTL induction by autologous gp96 vaccine derived from B16 tumors in tumor-bearing mice, we first analyzed tumor-specific CTL in the spleen by intracellular IFN-γ staining, ELISPOT and cytotoxic assays. Both anti-CD25 and cyclophosphamide treatments decreased the frequencies of Treg in gp96-immunized mice (Fig. 4a). However, compared with control IgG1, anti-CD25 treatment decreased Treg populations by 60.7% (P < 0.01), much higher than the decrease achieved using cyclophosphamide (only 28.8%). With regard to CTL, mice receiving combination treatment with gp96 vaccine and anti-CD25 exhibited a significant increase in IFN-γ-secreting CD8⁺ T cells by about 2.5-fold compared with gp96 vaccine alone (P < 0.01) (Fig. 4b). Similar results were obtained in the ELISPOT assay (Fig. 4c). We also examined the lytic function of gp96-induced CTL using B16 cells as target cells (Fig. 4d). Immunization with gp96 effectively elicited CTL with high tumor specificity and high cytotoxicity, while only low-cytotoxic T-cell immunity against B16 tumor could be detected in PBS- or anti-CD25-treated mice, indicating that the tumor-bearing mice are immunotolerant to B16 tumor. Moreover, a significant difference in antitumor cytotoxic T-cell response was observed between mice treated with and without anti-CD25 in gp96-immunized mice (P < 0.01). Similar trend was observed for cyclophosphamide treatment, but the difference did not reach statistical significance, which suggests the ability of anti-CD25 mAb to enhance gp96-mediated tumor-specific CTL responses is more prominent than cyclophosphamide.

Fig. 4.

Anti-CD25 treatment enhances tumor-specific T-cell responses induced by B16-gp96 vaccine. Mice were treated as described in Fig. 3, and splenocytes were isolated for detection on day 12. a and b, Splenocytes of different treatment groups were stained with CD3, CD4, CD25 and intracellular Foxp3 (a) or with CD3, CD8 and intracellular IFN-γ (b) and detected by flow cytometric analysis. c Detection of murine tumor-specific T cells by IFN-γ ELISPOT assay. Splenocytes at 1 × 10⁶ cells/well were stimulated with 20 μg of B16 melanoma cell lysates or BSA as negative control for background evaluation. d Cytotoxicity assay. B16 cells were labeled with CFSE as the target cells and seeded into a 96-well U-bottom microtiter plate. Splenocytes were added at different effector/target ratios: 5:1, 10:1 and 20:1. After 4 h, PI staining was then assessed in the CFSE labeled target cell population to quantify apoptosis by flow cytometric analysis. Data show mean ± SD of seven mice. Student’s t-test was used to determine P values. *P < 0.05 and **P < 0.01. Data are representative of two independent experiments

To investigate whether Treg frequency may be reduced in tumor tissues of anti-CD25 or cyclophosphamide-treated mice, TILs were isolated from tumor tissues and analyzed by flow cytometric analysis. Anti-CD25 mAb or cyclophosphamide treatment led to a significant decrease of Treg compared with no treatment in gp96-immunized mice (P < 0.01 or 0.05) (Fig. 5a). In the mean time, CD8⁺ T cells in tumor were detected by flow cytometric analysis and immunohistochemistry staining. As can be seen in Fig. 5b, a dramatic increased number of tumor-infiltrating CD8⁺ T cells were observed between gp96-immunized mice treated with and without anti-CD25 (50.84 ± 3.88 vs. 31.90 ± 1.30, P < 0.01) and immunized mice treated with or without cyclophosphamide (39.06 ± 2.60 vs. 31.90 ± 1.30, P < 0.05). The CD8/Treg ratio was also higher in anti-CD25 and cyclophosphamide-treated mice (Fig. 5c). A similar result was obtained by immunohistochemical assay. As shown in Fig. 5d, anti-CD25 or cyclophosphamide treatment in gp96-immunized mice led to a significant increased infiltration of CD8⁺ TIL compared with no treatment (P < 0.01 or 0.05). However, no obvious increase in CD8⁺ T by anti-CD25 treatment was observed in non-immunized mice, although Treg was substantially reduced in these mice. This may explain why anti-CD25 mAb treatment alone is not sufficient to affect B16 melanoma growth. Together, the results confirmed the increase of CTL inside tumor tissues of gp96-immunized mice under Treg depletion and suggested that increased CTL in tumor is associated with enhanced suppression of B16 melanoma growth.

Fig. 5.

Anti-CD25 treatment decreases Treg and increases CD8⁺ T-cell infiltration in tumor. Mice were treated with B16-gp96 as described in Fig. 3 and killed for TIL analyses on day 12. a and b Tumor tissues were digested and the single-cell suspensions were subjected to discontinuous Percoll gradient centrifugation. The separated lymphocytes were analyzed by flow cytometry to identify CD4⁺CD25⁺Foxp3⁺ Treg (a) and CD8⁺CD3⁺ cells (b). c Ratio of CD8⁺ cells to Treg cells was calculated. d Consecutive paraffin sections of tumors were stained with anti-CD8 mAb and counterstained with hematoxylin. i, ii, iii, iv and v represent mice treated with PBS, Anti-CD25, gp96+IgG1, gp96+cy and gp96+Anti-CD25, respectively. Arrow indicates CD8⁺ T cells. Data show mean ± SD of seven mice. *P < 0.05 and **P < 0.01. Data are representative of two independent experiments

Discussion

Since it was found in the chemically induced BALB/c sarcomas as a tumor rejection antigen more than 20 years ago [4, 41], the autologous gp96 tumor vaccine purified from tumors has been extensively studied in both animal models and human clinical trials. However, its effectiveness in immunotherapy seems to be limited probably due to immune evasion of tumor [21, 23]. Here, we demonstrated that the blockade of Treg by a monoclonal antibody significantly increased gp96–peptide complex mediated peptide-specific CTL responses in BALB/c mice and synergistically enhanced gp96 tumor vaccine-induced antitumor immunity in the B16 melanoma model, supporting our hypothesis that the balance between regulatory and responder T cells mediated by gp96 may determine the outcome of gp96-based immunotherapy. Removing negative regulators of the immune system could tip the balance toward enhanced gp96-mediated CTL responses. Our current work may therefore help to design a more efficient combination immunotherapy for this personalized cancer vaccine in clinical trials.

Treg cells suppress the activation and effector functions of many cell types, including CD4⁺ and CD8⁺ T cells, nature killer T (NKT) cells and dendritic cells via contact-dependent and suppressive cytokine (e.g. IL-10 and TGF-β)-dependent mechanisms [42, 43]. Tregs play crucial roles in preventing autoimmune diseases and maintaining self-tolerance. However, in latent or chronic diseases, such as cancer, Treg cell responses may interfere with cancer-specific T-cell responses that play a critical role for tumor rejection. In a number of tumor models, Treg suppression/depletion by CD25-specific antibody, low-dose cyclophosphamide or anti-glucocorticoid-induced tumor necrosis factor receptor family-related gene (GITR) antibody has been reported to break tumor immune evasion and enhance T-cell or NKT-cell-mediated antitumor immunity [44–48]. Moreover, anti-TGF-β has been shown to synergistically enhance tumor vaccine-mediated CTL activity in a subcutaneous TC1 lung cancer model [49]. Since targeting Treg has been demonstrated to act as a promising strategy for breaking tumor immune evasion and inducing tumor regression, clinical trials of using anti-CD25 antibody have been initiated for enhancement of immune response to tumor antigen vaccination [50, 51]. However, several studies have shown that the efficiency of Treg depletion and other immunosuppressive cells and factors all could determine the effectiveness of Treg targeting therapy in cancer immunotherapy [28, 52]. Furthermore, several lines of evidence have demonstrated that Tregs can not only function as suppressor cells but also as helper T cells [53, 54]. In a recent phase I/II clinical trial in patients with metastatic melanoma, Treg depletion by anti-CD25 mAb did not enhance the efficacy of a dendritic cell vaccine probably due to its simultaneous suppression of CD25(+) effector cells [51]. These studies indicate that the immunotherapeutic effectiveness of Treg depletion may depend on the type of immune responses, the microenvironment in which the response is generated, and the timing and dosing of Treg-depleting treatment, etc. As for gp96-based tumor vaccine, until now, only a few studies have addressed this issue. The antitumor effect of tumor secreting gp96-Ig was found to be enhanced by treatment with low-dose cyclophosphamide that decreased the number of Tregs [55]. Our previous study further showed that gp96-mediated Treg not only inhibit overall T-cell responses but also exert much more suppression on functional CTL [33]. To start shedding light on the possible advantages of combination with Treg depletion, we observed that after anti-CD25 treatment, gp96–peptide complexes induced T-cell activity was significantly enhanced in BABL/c mice (Fig. 1). The effect of Treg depletion on antitumor activity of autologous gp96 vaccine was then determined in the established B16 melanoma model of C57BL/6 mice. A synergistic effect on B16 tumor inhibition was observed under co-treatment with gp96 vaccine and anti-CD25 mAb (Fig. 3). Whereas anti-CD25 treatment alone was not sufficient to markedly affect tumor growth, which is consistent with previous studies [28, 52]. Moreover, following experiments demonstrated that anti-CD25 treatment can effectively suppress Treg both systematically and intratumorally and thus increase CTL activity in gp96-immunized mice. By contrast, cyclophosphamide treatment could not significantly enhance gp96-mediated antitumor activity probably due to its limited suppression of Treg (Fig. 4a). Interestingly, no obvious increase of CD8⁺ TIL by anti-CD25 treatment was observed in non-immunized mice (Fig. 5b, d), although Treg was substantially reduced in these mice. This may explain why anti-CD25 mAb treatment alone did not significantly affect B16 melanoma growth.

Clinical studies have established that autologous gp96 vaccine is effective against a range of tumors in earlier-stage disease [16–20, 22, 24]. However, its therapeutic efficiency seems to be limited in these clinical trials. It has been suggested that enhanced expression of receptors that either down-regulate T-cell activity, such as PD-L1/B7-H1, or mediate T-cell death, such as tumor necrosis factor-induced apoptosis inducing ligand (TRAIL) or the Fas ligand (FasL), may compromise the gp96-mediated CTL responses [21, 23]. Our results provide direct evidence that combination of gp96 tumor vaccine and Treg depletion has synergistic effects of inhibition of tumor. Meanwhile, there are already drugs in clinical trials targeting Treg [50, 51]. Therefore, combining gp96 vaccine with Treg inactivation is a promising strategy for breaking tumor immune evasion and deserves further evaluation for the treatment of cancer.

Another intriguing question arises from our observations that the frequency of Tregs increased in the spleen of B16 tumor-bearing mice, which could contribute to the maintenance of immune tolerance to tumor (Fig. 2). Immune tolerance is a primary barrier to the development of effective vaccines to eradicate established tumor. Schreiber, et al. [56] have shown that gp96-Ig immunization by intraperitoneal injection of gp96-Ig-expressing cells EG7 induced immune responses against established EG7 lymphoma by decreasing CD11b⁺Gr-1⁺ cells and Foxp3⁺ cells and thus subjugating tumor-induced suppression of CTL expansion. Similar results were observed in our recent study showing reduced Tregs under immunization with HBV surface and core antigens combined formulation along with gp96 (10 μg/mouse for 3 times) in HBV transgenic mice [57]. Differently from these data, we did not see the same trend of gp96-induced Treg decrease in B16 melanoma model. There is no significant difference in Treg numbers both in spleen and in tumor tissue between gp96-immunized mice (20 μg/mouse for 4 times) and non-immunized mice (Figs. 4a, 5a). Indeed, as can be seen in Fig. 1b, gp96–peptide complexes immunization could result in detrimental bystander effects on T-cell stimulation by the activation of Treg in tumor-naïve mice. Our previous results and other studies involving gp96 as adjuvant have also indicated that high dose of gp96 immunization (100 μg/mouse) can down-regulate inflammatory events and hamper antitumor immunity by activation of Treg or myeloid suppressor cells [29, 30, 33]. Based on these studies, together with our current results, we speculate that the total immunization doses and the spatial summation of gp96 may affect Treg numbers and its activation. This different behavior of gp96-based vaccine could be attributed to different methods of immunization, including different gp96–antigen complex formulations, immunization procedures and doses, as well as various conditions including different tumor-induced immune responses, pre-existing immune responses and tumor microenvironments. Gp96 might act, therefore, as a mediator in the balance between Treg and CTLs, depending upon different amounts of gp96 and the immune contexts. More studies are needed to understand the immunomodulatory role of gp96 in various tumor types and stages, which may help to optimize the efficiency of gp96-based vaccines against established tumor. Conceivably, the immunotherapeutic benefit of Treg depletion is more likely achieved in an immune context where the gp96-mediated balance is toward Treg. Clearly, investigation on Treg activation/inactivation during gp96 immunization is important to indentify the tumor settings that are more likely to respond to co-treatment with Treg depletion.

In conclusion, this study has significant implications and provided the preclinical rationale in combination of Treg depletion with gp96 based tumor vaccines for the treatment of established tumors. By careful examination of gp96-mediated balance of Treg and CTLs, it is possible to develop a combination of immunological therapies, allowing to fine-tune gp96 tumor vaccine based on a consideration of tumor type and clinical stage, or even on individual basis.