Central role of IFNgamma-indoleamine 2,3-dioxygenase axis in regulation of interleukin-12-mediated antitumor immunity

Abstract

Sustained intratumoral delivery of interleukin-12 (IL-12) and granulocyte macrophage colony-stimulating factor induces tumor regression via restoration of tumor-resident CD8+ T-effector/memory cell cytotoxicity and subsequent repriming of a secondary CD8+ T-effector cell response in tumor-draining lymph nodes (TDLN). However, treatment-induced T-effector activity is transient and is accompanied with a CD4+ CD25+ Foxp3+ T-suppressor cell rebound. Molecular and cellular changes in posttherapy tumor microenvironment and TDLN were monitored to elucidate the mechanism of counterregulation. Real-time PCR analysis revealed a 5-fold enhancement of indoleamine 2,3-dioxygenase (IDO) expression in the tumor and the TDLN after treatment. IDO induction required IFNgamma and persisted for up to 7 days. Administration of the IDO inhibitor D-1-methyl tryptophan concurrent with treatment resulted in a dramatic enhancement of tumor regression. Enhanced efficacy was associated with a diminished T-suppressor cell rebound, revealing a link between IDO activity and posttherapy regulation. Further analysis established that abrogation of the regulatory counterresponse resulted in a 10-fold increase in the intratumoral CD8+ T-cell to CD4+ Foxp3+ T-cell ratio. The ratio of proliferating CD8+ T-effector to CD4+ Foxp3+ T-suppressor cells was prognostic for efficacy of tumor suppression in individual mice. IFNgamma-dependent IDO induction and T-suppressor cell expansion were primarily driven by IL-12. These findings show a critical role for IDO in the regulation of IL-12-mediated antitumor immune responses.

Figure 1. Long-term CD8+ T-effector cell activity in post-therapy mice

Panel A. CD8 depletion in the surgical metastasis model. The schematic outlines the surgical metastasis model and the different CD8+ T-cell depletion timelines. Tumor bearing mice (200 mm³) with established lung metastases were treated with a single intratumoral injection of IL-12/GM-CSF microspheres on day 0. Treated primary tumors were then resected on day 7. Mice were sacrificed on day 28, lungs were removed, processed into single cell suspensions and plated in culture. Tumor colonies growing in plates were quantified on day 42. Panel B. Results of the clonogenic metastasis assay. Lung tumor burden from mice in different groups are shown. In the untreated group tumors were resected upon reaching 200 mm³ in volume without treatment. Error bars = SE, n = 5–6 mice per group. Assumption of homogeneity of variance among groups was tested by ANOVA and satisfied (p < 0.05).

Figure 2. Effector and suppressor activity in the tumors and the TDLN of post-therapy mice

Panel A. Quantitative analysis of CD8 and Foxp3 mRNA expression. Total RNA was isolated from the tumors and the TDLN of mice on day 0 (pre-treatment), at 6 hours post-treatment and on days 1, 3, 5 and 7. The mRNAs for CD8 and Foxp3 were then quantified using real-time PCR analysis. The fold-change in mRNA level as normalized to day 0 is shown. Error bars = SE, n = 4 per group. For CD8 the differences between day 0 and days 5 or 7 were significant in tumors (p ≤ 0.024), and between day 0 and days 3 or 7 in the TDLN (p ≤ 0.009). For Foxp3, the differences between day 0 and days 3 or 7 were significant in the TDLN (p ≤ 0.012) and between day 0 and day 7 in the tumor (p = 0.033). Panel B. Quantitative analysis of IFNγ and IDO mRNA expression. Fold-change in mRNA levels was normalized to day 0 (pre-therapy). For IDO the differences between day 0 and days 3 or 7 were significant in the TDLN (p ≤ 0.005), and between day 0 and 1 in the tumor (p = 0.02). For IFNγ the differences between day 0 and all other days were significant in the TDLN (p ≤ 0.04). In the tumor the differences between day 0 and days 1 or 3 were significant (p ≤ 0.05). Error bars = SE, n = 4 per time point. Panel C. IDO induction in BALB/c versus GKO mice. Fold-changes in mRNA levels one day after treatment are shown. The differences between wild-type and GKO mice were significant in both the tumor and the TDLN (p = 0.004 and 0.043, respectively). Error bars = SE, n = 4 per group. The data are representative of 2 separate studies.

Figure 3. Effect of D-1MT co-administration on antitumor efficacy of IL-12/GM-CSF microsphere therapy

Tumors were induced and allowed to grow to ~40mm³ in size. They were then treated with a single intratumoral injection of IL-12/GM-CSF microspheres. D-1MT was administered once intratumorally on day 0 (2mg/ml in 0.05 ml with microspheres) and orally in drinking water (2 mg/ml) between days -2 and 14. Control mice received a single intratumoral injection of blank microspheres. Each line represents a single mouse. Control and D-1MT groups had 6 mice each whereas IL-12/GM-CSF alone and the combination groups had 11 mice each. Tumors regressed completely in 5 of 11 mice in the combination treatment group. The data shown are representative of 3 separate experiments. Log-Rank analysis of survival curves demonstrated that the difference between IL-12/GM-CSF-alone and IL-12/GM-CSF + D1-MT groups was significant (p = 0.022, Supplemental Figure 1).

Figure 4. Effect of D-1MT co-administration on post-treatment T-cell kinetics in mice receiving IL- 12/GM-CSF therapy

Panel A. CD4+ Foxp3+ T-suppressor cell expansion in post-therapy mice. Mice bearing tumors were treated either with IL-12/GM-CSF alone (control) or IL-12/GM-CSF + D-1MT (D-1MT). Tumors and TDLN were harvested on days 3, 7 and 10 and T-suppressor cells were quantified. The differences between control and D-1MT groups were significant on days 3 and 7 in the TDLN (p ≤ 0.031) and on day 3 in the tumor (p = 0.036). The dot-plots represent typical results (CD45+ CD3+ cells were gated on and analyzed for CD4 and Foxp3 expression) on day 7. Panel B. Analysis of proliferating CD8+ T-cells. The above samples were also analyzed for proliferating CD8+ T-cell populations using the BrdU pulse-labeling strategy. CD45+ CD3+ cells were gated on and analyzed for CD8 and BrdU staining. The differences between the control and D-1MT samples for day 7 TDLN (p = 0.006) and for day 10 tumors (0.026) were significant. Error bars = SE, n = 4–5 mice per group. This experiment was repeated twice with similar results.

Figure 5. Ratio of proliferating CD8+ T-cells to CD4+ Foxp3+ T-suppressor cells as a prognostic marker of therapeutic efficacy

Panel A. Kinetics of CD8+ T-cell to T-suppressor cell ratio in mice treated with IL-12/GM-CSF alone (control) or IL-12/GM-CSF + D-1MT (D1-MT). TDLN and tumors were harvested from mice on indicated days after treatment and the ratio of CD8+ BrdU+ T-cells to CD4+Foxp3+ T-suppressor cells was determined. Day 0 data are from untreated mice. The differences between day 0 and all other time points were significant (p ≤ 0.013) in the TDLN. The differences between day 0 and days 3 or 7 were significant (p ≤ 0.046) in the tumor. Error bars = SE, n = 4–5 mice per group. Panel B. Correlation of CD8+ BrdU+ to CD4+ Foxp3+ T-cell ratio with tumor suppression. Mice were treated with IL-12/GM-CSF alone (control) or with cytokines plus D-1MT. Tumors were harvested on day 7 after treatment, weighed and the ratio of CD8+ BrdU+ to CD4+ Foxp3+ T-cells was determined for each tumor. The ratio was then plotted against tumor weight for each individual. Correlation coefficients (r) for control and D-1MT-treated groups were 0.637 and 0.881, respectively (n = 8 and 9).