Syngeneic animal models of tobacco-associated oral cancer reveal the activity of in situ anti-CTLA-4

Abstract

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide. Tobacco use is the main risk factor for HNSCC, and tobacco-associated HNSCCs have poor prognosis and response to available treatments. Recently approved anti-PD-1 immune checkpoint inhibitors showed limited activity (≤20%) in HNSCC, highlighting the need to identify new therapeutic options. For this, mouse models that accurately mimic the complexity of the HNSCC mutational landscape and tumor immune environment are urgently needed. Here, we report a mouse HNSCC model system that recapitulates the human tobacco-related HNSCC mutanome, in which tumors grow when implanted in the tongue of immunocompetent mice. These HNSCC lesions have similar immune infiltration and response rates to anti-PD-1 (≤20%) immunotherapy as human HNSCCs. Remarkably, we find that >70% of HNSCC lesions respond to intratumoral anti-CTLA-4. This syngeneic HNSCC mouse model provides a platform to accelerate the development of immunotherapeutic options for HNSCC.

Fig. 1. Development of a novel syngeneic mouse model for oral squamous cell carcinoma.

a Experimental scheme of 4NQO syngeneic model. C57Bl/6 mice were given 4NQO (50 μg/mL) in the drinking water for 16 weeks and then regular water until week 22. Cells were isolated from the lesions, cultured, and then implanted into the tongue of wild-type C57Bl/6 mice. The Scheme was drawn by Yagi and Allevato. b Mutational signatures associated with tobacco smoking. The somatic mutational profiles of the four lesions from mice exposed to 4NQO were correlated to known mutational signatures in human cancer (Pearson correlation > 0.93)^,. Top, Signature 4 extracted from cancers associated with tobacco smoking, this signature was found only in cancer types in which tobacco smoking increases risk and mainly in those derived from epithelia directly exposed to tobacco smoke; middle, the pattern of a mutational signature of lesions from mice exposed to 4NQO, compilation of all four samples analyzed; bottom, the pattern of a mutational signature of lesions from mice exposed to DMBA. The similarity between signature tobacco smoking associated HNSCC and signature 4NQO is 93.9%; and the similarity between signature tobacco smoking associated HNSCC and signature DMBA is only 39.7%. c Percentage of somatic substitutions located in translated or untranslated in tobacco smoking associated HNSCC patients (left), 4NQO derived lesions (middle) and DMBA derived lesions (right). d Graphical matrix representation of the individual mutations in four syngeneic cell lines (4MOSCs) isolated from lesions from mice exposed to 4NQO. Listed are the alterations most frequently observed in human HNSCC and the corresponding percentage of mutations. Mutations (red), or no mutations (blue) are listed in rows and four different cell lines are in column. e Mutational plot of TP53 mutations in 243 HPV-negative tumor samples from TCGA (top) and of four syngeneic cell lines (4MOSCs) (bottom). Frequency of mutation is depicted by height of lollipop, blue circles represent mutations unique to human or mouse, and red circles depict mutations in common between human and mouse HNSCCs.

Fig. 2. Characterization of 4NQO-induced murine oral squamous cell model.

a Left panel, C57Bl/6 mice were implanted with 1 × 10⁶ of either 4MOSC1 or 4MOSC2 cells into the tongue. Tongue lesions when the tumor volume reached ~100 mm³. Middle panel, representative H&E staining of histological tissue sections from mouse tongues with 4MOSC1 or 4MOSC2 tumors. Right, representative pictures of tumors stained to show expression of cytokeratin 5 (CK5, green) and DAPI (blue) (n = 3 mice per group). b Top panel, representative H&E stain of a nonmetastatic cervical lymph node from mice with 4MOSC1 tumors. Bottom panel, representative H&E stain of a metastatic cervical lymph node from mice with 4MOSC2 tumors. Metastatic growth of 4MOSC2 cells into the lymph node is depicted with a dotted line in the bottom area (n = 5 mice per group). c Representative tumor tissue sections stained for LYVE-1 by immunohistochemistry in 4MOSC1 or 4MOSC2 tumors (n = 3 mice per group). d Absolute number of immune cells infiltrating 4MOSC1 or 4MOSC2 tumors. Shown is the average number of live cells infiltrating per mm³ of tumor (n = 3 mice per group). e Immunofluorescent staining of CK5 and CD8 to show squamous cell character of the lesion and CD8 infiltration in mice with 4MOSC1 or 4MOSC2 tumors, respectively (n = 3 mice per group) (CK5, green; CD8, red; DAPI, blue). 4MOSC1 or 4MOSC2 tumors were isolated from mice and mechanically and enzymatically digested. Single-cell suspension was then stained with CD45, Nk1.1, CD3, CD8, CD44, PD-L1, PD-1, CTLA-4, LAG-3, and TIM-3 fluorescent labeled antibodies and analyzed by flow cytometry. Shown are representative flow cytometry plots of f the frequency of tumor cells (CD45 negative) expressing PD-L1 and g the frequency of CD8⁺/CD44⁺ cells expressing inhibitory receptors PD-1, CTLA-4, LAG-3, and TIM-3 in individual tumors (n = 4 mice per group). Contour plots of lymphocytes from tumor (green), and corresponding cervical lymph nodes (blue), and blood (red) are overlaid and the frequencies of tumor CD8⁺/CD44⁺ T cells expressing each inhibitory receptor are shown (n = 4 mice per group).

Fig. 3. Variable responses to anti-PD-1 in mice with 4MOSC1 tumors.

a C57Bl/6 mice were implanted with 1 × 10⁶ of 4MOSC1 cells into the tongue. After tumors reached ~30 mm³, mice were treated IP with 10 mg/kg of isotype control or anti-PD-1 (n = 10 per group). Individual growth curves of 4MOSC1 tumor-bearing mice are shown. b A Kaplan–Meier curve showing the survival of mice from a. The death of animals occurred either naturally, when tumor compromised the animal welfare, or when tumor volume reached 100 mm³ (n = 10 mice per group; Log-Rank/Mantel–Cox test.). c Absolute number of live CD45⁺CD3⁺CD8⁺ T cells infiltrating 4MOSC1 tumors with or without anti-PD-1 treatment. Shown is the average of the number of live CD8 T cells infiltrating per mm³ of tumor (n = 4 mice per group; two-sided Student’s t test; data are represented as mean ± SEM). d Immunofluorescent staining of CD8 highlights an increase in CD8 T-cell recruitment with anti-PD-1 treatment. Shown is the average CD8 positivity was by three regions of interest (ROI) per mouse (n = 3 mice per group; two-sided Student’s t test; data are represented as mean ± SEM). e Dependency of anti-PD-1 on CD8 T cells. C57Bl/6 mice were treated with CD8 T cell depleting antibody daily for 3 days before tumor implantation and then once a week after. Mice were then implanted with 1 × 10⁶ of 4MOSC1 cells into the tongue. After tumors reached ~30 mm³, mice were treated IP with 10 mg/kg isotype control or 10 mg/kg anti-PD-1 (n = 5 per group). Individual growth curves of 4MOSC1 tumor-bearing mice are shown. f Immunofluorescence staining of CK5 and Ki-67 in cervical lymph nodes of control or CD8-depleted 4MOSC1-bearing mice. Metastatic lesions in the lymph nodes showed abundant Ki-67⁺ proliferating tumor cells (n = 5 mice per group). g C57Bl/6 mice were implanted with 1 × 10⁶ of 4MOSC2 cells into the tongue. After tumors reached ~30 mm³, mice were treated IP with 10 mg/kg isotype control or 10 mg/kg anti-PD-1 (n = 5 per group). Individual growth curves of 4MOSC2 tumor-bearing mice are shown.

Fig. 4. Efficacy of intratumoral delivery of immune oncology agents.

a Left panel, C57Bl/6 mice were implanted with 1 × 10⁶ of 4MOSC1 cells into the tongue. After tumors reached ~30 mm³, mice were either treated IP or by intratumoral (IT) delivery of PBS, IP with 10 mg/kg or IT with 5 mg/kg anti-PD-1. Shown is the average volume of each tumor (n = 4 mice per group; two-sided Student’s t test; data are represented as mean ± SEM). Right panel, representative pictures of tongues from mice in a with tumors depicted with a dotted line. b Distribution of anti-PD-1 antibody in mice with 4MOSC1 tumors using IP or IT delivery of the treatment. Staining for anti-hamster IgG showed the localization of anti-PD-1 antibody in the tongue, lymph nodes, and spleen of treated mice (n = 4 mice per group). c RNA from each tumor was isolated and comprehensive immune profiling was analyzed using the NanoString nCounter PanCancer Mouse Immune Profiling gene expression platform. The advanced analysis module of the nSolver software was used to analyze genes associated with listed immune cells and given a score. Shown is the Z-score of each cell profile score (n = 3 mice per group). d Absolute number of live CD45⁺CD3⁺CD8⁺ T cells infiltrating 4MOSC1 tumors with or without anti-PD-1 or anti-CTLA-4 treatment. Shown is the average of the number of live CD8 T cells infiltrating per mm³ of tumor (n = 3 mice per group; two-sided Student’s t test; data are represented as mean ± SEM). e Frequency of live CD45⁺CD3⁺CD4⁺ FoxP3⁺ Tregs infiltrating 4MOSC1 tumors with or without anti-PD-1 or anti-CTLA-4 treatment. Left panel, a representative flow cytometry plot from one mouse showing the frequency of Tregs (CD4⁺FoxP3⁺) out of CD4⁺ cells is shown. Right panel, the frequency of Tregs out of CD4⁺ cells was quantified following treatment with anti-PD-1 or anti-CTLA-4 (n = 5 mice per group; two-sided Student’s t test; data are represented as mean ± SEM).

Fig. 5. Mice with 4MOSC1 tumors show nearly complete response to anti-CTLA-4.

a C57Bl/6 mice were implanted with 1 × 10⁶ of 4MOSC1 cells into the tongue. After the tumors reached ~30 mm³, mice were treated 10 mg/kg of isotype control or anti-CTLA-4 for IP administration (left), and 5 mg/kg of isotype control or anti-CTLA-4 for IT administration (right). Individual growth curves of 4MOSC1 tumor-bearing mice plotting primary tumor growth are shown (n = 10 mice per group). b Shown is the immunofluorescent staining of the distribution of anti-CTLA-4 antibody for mice with 4MOSC1 tumors using IP or IT delivery of the treatment. Staining for anti-hamster IgG (red) showed the localization of anti-CTLA-4 antibody in the tongue, lymph nodes, and spleen of treated mice. DAPI staining for nuclei is shown in blue (n = 4 mice per group). c Immunofluorescent staining of CD8 highlights an increase in CD8 T-cell recruitment with anti-CTLA-4 treatment. Quantification of CD8 T cells with or without anti-CTLA-4 treatment was done by immunofluorescent staining of tumor (CK5) in the tongue. Shown is the average CD8 positivity by three regions of interest (ROI) per mouse, quantified by Qupath software for each condition (n = 3 mice per group two-sided Student’s t test; data are represented as mean ± SEM). Antigen specific T-cell cytotoxic assay. C57Bl/6 mice were implanted with 1 × 10⁶ of 4MOSC1 cells into the tongue, and when they reached ~30 mm³, mice were treated IT with isotype control, anti-PD-1, or anti-CTLA-4 every other day for three treatments total. CD8 T cells from each group were isolated and cocultured with preplated 4MOSC1 cells. DMSO (10%) was used to treat 4MOSC1 as a positive control, and DRAQ7 was added in the culture medium to mark dead cells (red). d Real time live-imaging experiments were performed using the 880 confocal fast scan (Zeiss), and representative images of tumor cell killing (CD8 T cells from anti-CTLA-4 group) are shown at the indicated times. e Quantification of dead cancer cells at the end of experiment. (n = 3 mice per group; two-sided Student’s t test; data are represented as mean ± SEM).