IFN-γ is required for cytotoxic T cell-dependent cancer genome immunoediting

Abstract

Genetic evolution that occurs during cancer progression enables tumour heterogeneity, thereby fostering tumour adaptation, therapeutic resistance and metastatic potential. Immune responses are known to select (immunoedit) tumour cells displaying immunoevasive properties. Here we address the role of IFN-γ in mediating the immunoediting process. We observe that, in several mouse tumour models such as HA-expressing 4T1 mammary carcinoma cells, OVA-expressing EG7 lymphoma cells and CMS5 MCA-induced fibrosarcoma cells naturally expressing mutated extracellular signal-regulated kinase (ERK) antigen, the action of antigen-specific cytotoxic T cell (CTL) in vivo results in the emergence of resistant cancer cell clones only in the presence of IFN-γ within the tumour microenvironment. Moreover, we show that exposure of tumours to IFN-γ-producing antigen-specific CTLs in vivo results in copy-number alterations (CNAs) associated with DNA damage response and modulation of DNA editing/repair gene expression. These results suggest that enhanced genetic instability might be one of the mechanisms by which CTLs and IFN-γ immunoedits tumours, altering their immune resistance as a result of genetic evolution.

Figure 1. 4T1-HAc cells respond to IFN-γ.

(a) HA (left panel) and IFN-γRα chain (right panel) expression on 4T1-HAc (thin line) and 4T1-HAγRDN (thick line) cells were analysed by flow cytometry. Staining of 4T1-HAc and 4T-HAγRDN cells with isotype control mAb was indistinguishable (the level indicated by the dotted line). HA expression level on parental 4T1-HA cells was comparable to that on 4T1-HAγRDN and 4T1-HAc cells. (b) MHC class I expression on 4T1-HAc and 4T-HAγRDN cells was analysed by flow cytometry after 24 h culture with (thick lines), or without (thin line), IFN-γ. Staining of both cell populations with isotype control mAb was indistinguishable after the culture with or without IFN-γ (the level indicated by the dotted line). MHC class I expression level of parental 4T1-HA cells was comparable to that of 4T1-HAγRDN and 4T1-HAc cells and was similarly augmented by IFN-γ as for 4T1-HAc cells. (c) After incubation with or without HA peptide in the presence or absence of IFN-γ for 24 h, 4T1, 4T1-HAc, and 4T1-HAγRDN cells were co-cultured with HA-specific WT CTL for 24 h, then IFN-γ levels in the cell-free culture supernatants were determined by ELISA. Data are shown as mean±s.d. of three independently cultured cells. *P<0.05 as compared with the supernatant harvested from the culture of the same cells that were pre-incubated without IFN-γ by unpaired, two-tailed Student's t-test. (d) 4T1-HAc and 4T1-HAγRDN cells were inoculated into the same RAG^−/− and WT mice, and 10 days later RAG^−/− mouse was treated with HA-specific WT CTL. Five days after ACT, 4T1-HAc and 4T1-HAγRDN cells were isolated from the growing tumour mass. 4T1-HAc cells grown in RAG^−/− mouse treated with HA-specific IFN-γ^−/− ACT-treated (at day 10) were also collected at day 15. Phosphorylation of STAT1 and STAT3 in tumour cells was analysed by western blotting. Similar results were obtained in four experiments (a,b) and three experiments (c,d).

Figure 2. Integrated HA gene loss in 4T1-HA cells responding to IFN-γ and CTL in vivo.

(a) mRNA was prepared from WT splenocytes, 4T1 cells cultured in vitro, and representative HA-positive and HA-negative 4T1-HA cells isolated after the growth under the indicated conditions, then the indicated segments of HA gene were amplified by RT–PCR. RT–PCR was performed on every 4T1-HA cells independently, and the indicated number of PCR products were mixed and loaded in the respective groups. (b) Genome DNA and mRNA were prepared from 4T1-HA cells grown in the indicated mice for 25–35 days, and the indicated segments of the HA gene were amplified by RT–PCR and genomic PCR. (c) Genome DNA was prepared from three 4T1-HA cells grown independently in vitro with or without IFN-γ or in IL-12-treated RAG^−/− mice for 30 days, and the indicated segments of the HA gene were independently amplified by genomic PCR. PCR products were mixed and loaded in the respective groups. Similar results were obtained in two independent experiments in all presented experiments (a–c).

Figure 3. Genomic alteration in 4T1-HA cells responding to IFN-γ and CTL in vivo.

Genomic DNA were prepared from 4T1-HAc and 4T1-HAγRDN cells that were isolated from the growing tumour mass in the indicated mice (>99% purity) (a) or 4T1-HAc cells isolated from the growing tumour mass in pfp/IFN-γ^−/− mice 25 days after the ACT with HA-specific IFN-γ^−/− or pfp^−/− CTL or in IL-12-treated RAG^−/− mice at 30 days or from 4T1-HA cells cultured with IFN-γ for 30 days in Fig. 2b,c (>99% purity) (b). Then, CNAs were examined by a-CGH used for s.c. inoculation as the reference sample. (c) 4T1-HA cells and 4T1-HAS1DN cells were inoculated into the same RAG^−/− mice that were treated with ACT on day 0. Genomic DNAs were obtained from both tumour cells prepared from the growing tumour masses 30 days after the tumour inoculation. Then, CNAs were examined by a-CGH. In a-CGH analysis, the tumour cells used for the s.c. inoculation are the reference sample. The positions showing significant CNA are indicated by the lines and arrows.

Figure 4. Antigen gene expression in EG7.1 cells and CMS5a1 cells after the exposure to tumour-specific CTL in vivo.

(a,b) EG7.1 cells were inoculated into the indicated mice, and some mice were treated with ACT of OVA-specific OT-1 CTLs on day 0. mRNA and genome DNA were prepared from EG7.1 cells grown 25–35 days. The indicated segment of the OVA gene, that contains H-2K^b-restricted CTL epitope targeted by OT-1, was amplified by RT–PCR (a) or genomic PCR (b). (c) mRNA was prepared from CMS5a1 cells isolated from the growing tumour mass in the indicated mice. ERK gene was amplified by RT–PCR, then, PCR products were digested by Sfcl restriction enzyme that selectively cleaves mutated ERK, but not wild type ERK2. Similar results were obtained in two independent experiments in all presented experiments (a–c).

Figure 5. Genetic alteration in EG7.1 cells and CMS5a1 cells after the exposure to tumour-specific CTL in vivo.

(a) Genomic DNAs were prepared from EG7.1 cells isolated from the growing tumour mass in the indicated mice (Fig. 4a,b) and EG7.1γRDN cells isolated from the growing tumour mass in OT-1-treated WT mice 25 days after tumour inoculation (Supplementary Fig. 8d). Then, CNAs were examined by a-CGH employing tumour cells used for s.c. inoculation as the reference sample. (b) Genomic DNAs were prepared from CMS5a1 cells isolated from the growing tumour mass in the indicated mice as in Fig. 4c and Supplementary Fig. 9a–c. Then, CNAs were examined by a-CGH employing tumour cells used for s.c. inoculation as the reference sample. The positions showing significant CNA are indicated by the lines and arrows.

Figure 6. Altered expression of genes implicated in DNA repair and maintenance in tumour cells exposed to CTL-mediated immunoediting.

(a) Expression of Apobec 1 and 3 genes in 4T1-HA and 4T1-HAγRDN cells growing in ACT-treated RAG^−/− mice or CMS5a1 and CMS5a1γRDN cells growing in WT or ACT-treated WT mice. The gene expression was normalized to β-actin levels, and the relative expression compared with the respective cells grown in RAG^−/− mice (in upper panels) or respective in vitro cultured cells (in lower panels). Results are indicated as the average±s.d. of the results obtained from the experiments using the numbers of tumour cells indicated in parentheses. *P<0.05 compared with IFN-γR DN expressing respective cells; **P<0.005 compared with CMS5a1 cells grown in WT mice. Both are analysed by unpaired, two-tailed Student's t-test. Similar results were obtained in two independent experiments. (b) mRNA was prepared from freshly isolated 4T1-HA and 4T1-HAS1DN cells grown in the same ACT-treated RAG^−/− mice (n=3 each) or CMS5a1 cells grown in RAG^−/− or ACT-treated RAG^−/− mice (n=3 each). Expression of DNA repair genes was examined by quantitative RT–PCR array. (c) mRNA was prepared from freshly isolated 4T1-HAc and 4T1-HAγRDN cells growing in vitro, in RAG^−/−, WT HA-specific CTL-treated RAG^−/−, and WT mice. Double strand DNA repairing protein kinase ataxia-telangiectasia and Rad3 related, Atr, and protein kinase ataxia-telangiectasia mutated, Atm, gene expression was examined by quantitative RT–PCR. The gene expression was normalized to Gapdh levels, and the relative expression compared with the mean value of the in vitro growing tumour samples is presented. Results are indicated as the average±s.d. of the results obtained from the experiments using the numbers of tumour cells indicated in parentheses. *P<0.05 compared with cells in vitro; **P<0.005 compared with cells in vitro; ^#P<0.05 compared with cells in RAG^−/−; ^##P<0.005 compared with cells in RAG^−/−. All are analysed by unpaired, two-tailed Student's t-test.

Figure 7. CNAs induced in CMS5a1 cells in RAG^−/− mice treated with ATR inhibitor and WT ACT.

(a,b) CMS5a1 cells were inoculated into RAG^−/− mice, and some mice were treated with CD8 T cells prepared from DL of CMS5a1-bearing WT mice that were treated with anti-CD137 mAb as indicated by the black arrows on day 0 and 5. These ACT-treated mice were also treated with anti-CD137 mAb to activate CTL on day 0, 5 and 9 as indicated by the grey arrows. Some mice were treated with ATR inhibitor, VE822, on day 5, 7 and 9 as indicated by the black arrows. Tumour growth was measured and tumour cells were isolated 25 days after tumour inoculation (a). Genomic DNA and mRNA were prepared from CMS5a1 cells isolated from the tumour mass on day 25. Then, CNAs were examined by a-CGH employing tumour cells used for s.c. inoculation as the reference sample (b). The positions showing significant CNA are indicated by the lines and arrows. mRNA of ERK gene was amplified by RT–PCR, then, PCR products were digested by Sfcl restriction enzyme that selectively cleaves mutated ERK, but not wild type ERK2 (c). Concerning tumour growth and HA expression at RNA level, similar results were obtained in two independent experiments.

Figure 8. CNAs induced in IFN-γ-producing 4T1 tumours in RAG^−/− mice treated with WT or IFN-γ^−/− ACT.

5 × 10⁵ of 4T1-HA cells producing high amount of IFN-γ (4T1-HAIFNγTf) were inoculated into RAG^−/− mice. As indicated by arrow, when palpable tumours developed after 10 days, mice were received T cells (5 × 10⁷ per mice) obtained from draining lymph node of WT or IFN-γ^−/− mice that were inoculated with 4T1-HAIFNγTf cells 7 days before the sacrifice. Tumour cells were isolated from tumour mass 30 days after ACT (a), and genomic DNAs and mRNAs were prepared. CNAs were examined by a-CGH employing tumour cells used for s.c. inoculation as the reference sample (b). The positions showing significant CNA are indicated by the lines and arrows. The indicated portion of the HA gene were amplified by RT–PCR and genomic PCR (c). Concerning tumour growth and HA expression at RNA level, similar results were obtained in two independent experiments.