Comprehensive Testing of Chemotherapy and Immune Checkpoint Blockade in Preclinical Cancer Models Identifies Additive Combinations

Comprehensive Testing of Chemotherapy and Immune Checkpoint Blockade in Preclinical Cancer Models Identifies Additive Combinations

Front. Immunol., 11 May 2022 | https://doi.org/10.3389/fimmu.2022.872295
Comprehensive Testing of Chemotherapy and Immune Checkpoint Blockade in Preclinical Cancer Models Identifies Additive Combinations
1National Centre for Asbestos Related Diseases, University of Western Australia, Perth, WA, Australia
2School of Biomedical Sciences, University of Western Australia, Crawley, WA, Australia
3Institute for Respiratory Health, Perth, WA, Australia
4JJP Biologics, Warsaw, Poland
5Telethon Kids Institute, Perth, WA, Australia
6Medical School, University of Western Australia, Crawley, WA, Australia
Antibodies that target immune checkpoints such as cytotoxic T lymphocyte antigen 4 (CTLA‐4) and the programmed cell death protein 1/ligand 1 (PD-1/PD-L1) are now a treatment option for multiple cancer types. However, as a monotherapy, objective responses only occur in a minority of patients. Chemotherapy is widely used in combination with immune checkpoint blockade (ICB). Although a variety of isolated immunostimulatory effects have been reported for several classes of chemotherapeutics, it is unclear which chemotherapeutics provide the most benefit when combined with ICB. We investigated 10 chemotherapies from the main canonical classes dosed at the clinically relevant maximum tolerated dose in combination with anti‐CTLA-4/anti-PD-L1 ICB. We screened these chemo-immunotherapy combinations in two murine mesothelioma models from two different genetic backgrounds, and identified chemotherapies that produced additive, neutral or antagonistic effects when combined with ICB. Using flow cytometry and bulk RNAseq, we characterized the tumor immune milieu in additive chemo-immunotherapy combinations. 5-fluorouracil (5-FU) or cisplatin were additive when combined with ICB while vinorelbine and etoposide provided no additional benefit when combined with ICB. The combination of 5-FU with ICB augmented an inflammatory tumor microenvironment with markedly increased CD8+ T cell activation and upregulation of IFNγ, TNFα and IL-1β signaling. The effective anti‐tumor immune response of 5-FU chemo-immunotherapy was dependent on CD8+ T cells but was unaffected when TNFα or IL-1β cytokine signaling pathways were blocked. Our study identified additive and non-additive chemotherapy/ICB combinations and suggests a possible role for increased inflammation in the tumor microenvironment as a basis for effective combination therapy.
Introduction
Drugs that block immune checkpoint receptors such as CTLA-4, PD-1 or PD-L1 have revolutionised cancer treatment, with durable anti-tumor responses observed in a subset of cancer patients ( 1 , 2 ). However, the majority of patients treated with immune checkpoint blockade (ICB) demonstrate little or no benefit. Conventional chemotherapy remains standard treatment for many cancers. In addition to cytotoxic effects on cancer cells, many chemotherapeutics are immunostimulatory, capable of; inducing immunogenic cell death ( 3 ), increasing antigen cross-presentation ( 4 ), increasing immune cell infiltration ( 5 ), depleting immunosuppressive cells ( 6 , 7 ), and altering expression of immune checkpoint ligands ( 8 , 9 ). As these characteristics have been linked to ICB efficacy, some chemotherapeutics could potentially enhance anti-tumor immune responses when combined with ICB and therefore combination therapy warrants further investigation. Combination ICB and chemotherapy has shown efficacy in several cancer types. In fact, of the many different drug classes that have been combined with ICB, classical cancer chemotherapy remains one of the most successful ( 10 ). Particularly in thoracic cancers, chemotherapy/ICB combinations have shown efficacy, with FDA approval in non-small cell lung cancer ( 11 ) and small cell lung cancer ( 12 ), and with promising results in malignant pleural mesothelioma ( 13 ).
Although the effects of individual chemotherapeutics on discrete components of the immune system have been extensively described, a systematic analysis of how different chemotherapies combine with ICB in vivo is lacking, and the molecular mechanisms underlying additive chemo-immunotherapy combinations remains unknown. In this study, we systematically interrogated the therapeutic interaction between ICB and different canonical classes of cancer chemotherapeutics, given at maximum tolerated dose (MTD), in two preclinical cancer models, and mapped the molecular and cellular profiles of additive combinations, with the aim of prioritizing combinations to take forward into clinical trials.
Materials and Methods
Mice
Female BALB/c and C57BL/6 mice (RRID: IMSR_ARC:BC, RRID: IMSR_ARC:B6) were bred and maintained at the Animal Resource Centre (Murdoch, WA, Australia) or Harry Perkins Institute of Medical Research (Murdoch, WA, Australia). All mice used were between 8-10 weeks of age and were maintained under standard specific pathogen free housing conditions at the Harry Perkins Bioresources North Facility (Nedlands, WA, Australia). All experiments were conducted in accordance with the code of conduct of the National Health and Medical Research Council (NHMRC) of Australia, and under the approval of the Harry Perkins Institute of Medical Research Animal Ethics Committee (protocols AE029, AE100, AE179).
Cell Lines
Murine mesothelioma cell lines AB1 (CBA, Cat# CBA-0144, RRID: CVCL_4403), AB1-HA (CBA, Cat# CBA-1374, RRID: CVCL_G361) and AE17 (CBA, Cat#CBA-0156, RRID: CVCL_4408) were derived as previously described ( 14 , 15 ). Cell lines were maintained in RPMI 1640 (ThermoFisher Scientific, Scoresby VIC, Australia) supplemented with 20 mM HEPES, 0.05 mM 2-mercaptoethanol, 100 units/mL penicillin (CSL, Melbourne VIC, Australia), 50 μg/mL gentamicin (David Bull Labs, Kewdale VIC, Australia), 10% Newborn Calf Serum (NCS; ThermoFisher Scientific, Scoresby VIC, Australia) and 50 mg/mL of geneticin for AB1-HA only (G418; Life Technologies). Cells were cultured for a minimum of 4 passages after thawing before inoculation into mice. Cell lines were validated yearly by flow cytometry for MHC-I molecules H2‐Kb (consistent with C57BL/6) and H2‐Kd (consistent with BALB/c), and for fibroblast markers E-cadherin, epithelial cell adhesion molecule, and platelet-derived growth factor receptor α (negative). All cell lines were tested for Mycoplasma spp., every 3 months by PCR and found to be negative.
Tumor Cell Inoculation
Cells were harvested when they reached 80% confluence. The right-hand flanks of mice were inoculated subcutaneously with 5 x 105 tumor cells suspended in 100 µL of PBS. Mice were randomized prior to treatment, when tumors were palpable. Tumor dimensions (length and width) were measured with digital calipers by an investigator blinded for treatment allocation and tumor growth was represented as area (mm2).
Chemotherapy, ICB and Antibody Treatments
Chemotherapy and ICB were administered on the same day, initiating treatment when tumors were approximately 20-25 mm2 in size. Chemotherapies were provided by Sir Charles Gardiner Pharmacy (Nedlands, WA, Australia) and was administered at the predetermined MTD as previously reported ( 16 ), except 5-FU which was administered at 75 mg/kg because MTD 5-FU with ICB caused severe toxicity ( Table S1 ). Anti-CTLA-4 (clone 9H10, JJP Biologics) was dosed once at 100 μg/mouse and anti‐PD‐L1 (clone MIH5, JJP Biologics) was dosed 3 times with 2-day intervals at 100 μg/mouse ( 17 ). For depletion experiments, anti-CD4 (clone GK1.5, BioXcell), anti-CD8 (clone YTS 169, BioXcell) or anti-IL1β (clone B122, BioXcell) antibodies were administered 3 times with 3-day intervals at 100 μg/mouse with the first dose commencing 3 days before chemo‐immunotherapy. Anti-TNFα (clone XT3.11, BioXcell) was administered using the above schedule but at 2 mg/mouse. All treatments were diluted in sterile 0.9 % sodium chloride and administered intraperitoneally (i.p.) or intravenously (i.v.) as described in Table S1 .
Preparation of Single Cell Suspensions
Spleen and draining lymph nodes (DLNs) were digested with 1 mg/mL type IV collagenase (Worthington Biochemical) and 1 μg/mL DNase (Sigma Aldrich) in RPMI-1640 supplemented with 2% NCS and 20 mM HEPES for 25 minutes at room temperature. Red blood cells were lysed with Pharm Lyse (BD Biosciences). All cell suspensions were resuspended in EDTA‐BSS-NCS. Absolute numbers of leucocytes in DLNs were obtained using the Z2 Coulter Counter Analyzer (Beckman Coulter).
Tumors were processed using the Miltenyi Biotec mouse tumor dissociation kit, as per manufacturer’s protocol. Briefly, tumors were cut into 2-4 mm pieces and added to GentleMACs C tubes with 2.35 mL RPMI media supplemented with 10% NCS. Prioprietary enzyme mix was added, and samples mechanically digested using the GentleMACS Octo Dissociator 37C_M_TDK_2 protocol.
Flow Cytometry
Three flow cytometry panels outlined in Table S2 were used to characterize lymphoid and myeloid cell subsets. CD16/32 Fc block (eBioscience) and Zombie UV™ (Biolegend) viability dye were diluted in PBS and added to samples prior to surface antigen staining. All antibodies for surface staining were diluted in PBS + 2% NCS. Cells were permeabilized using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience). Cells were washed with Permeabilization Buffer (eBioscience) and subjected to intracellular staining. Single stain and fluorescence minus-one (FMO) controls were also performed. To measure granzyme B (GzmB) and IFNγ, samples were incubated in Brefeldin A (Biolegend) for 4 hours at 37°C before antibody staining. Data was acquired using a BD LSRFortessa™ SORP with 50,000 live events collected per sample where possible. All flow cytometry analyses were completed using FlowJo™ Software version 10 (BD Biosciences). Summary of antibody concentrations and gating strategies are outlined ( Table S2 ; Figure S1 ).
Flow Cytometry Data Analysis
FCS files were subjected to automatic quality control of signal acquisition and dynamic range by the flowAI (v1.8) package using default parameters. Bad events (defined by negative outliers) were excluded, and manual gating was performed as outlined in Figure S1 . For clustering analysis on the lymphoid cells, each sample was downsampled to 5,000 CD45+ cells using the DownSample (v3.1.0) package. All samples from all groups were then concatenated. The UMAP (v2.2) and Phenograph (v1.3) packages using default parameters (k = 30) were performed in FlowJo using the concatenated FCS file. Clusters were manually grouped into the final subsets described in Figure 3 . Clusters were combined based on similar location on the UMAP plot and similar expression of key markers. Clusters that were CD45+ but had no expression of other phenotypic markers in the panel were colored grey and excluded from the analysis.
Tumor Preparation for Bulk RNAseq
Whole tumors were harvested and stored in RNAlater (Life Technologies) at -80°C. RNA was extracted from frozen tumors using the RNeasy Plus Mini Kit and Tissue Ruptor (QIAGEN). RNA quality was confirmed on the Bioanalyzer (Agilent Technologies). Library preparation and sequencing (100-base pair single-end on an Illumina HiSeq platform) were performed by the Australian Genome Research Facility (Melbourne, VIC, Australia).
Bulk RNAseq Analysis
Raw FASTQ files were aligned to the GRCm38/mm10 reference genome using Kallisto ( 18 ). Transcripts with low counts were removed and two count matrices were compiled using Tximport ( 19 ). The DESeq2 package ( 20 ) was used to identify differentially expressed genes (DEGs) between the following comparisons: PBS vs ICB, 5-FU, cisplatin, 5-FU+ICB or cisplatin+ICB; ICB alone vs 5FU+ICB or cisplatin+ICB, 5-FU alone vs 5-FU+ICB and cisplatin alone vs cisplatin+ICB. P values were adjusted for multiple comparisons using the Benjamini-Hochberg (B-H) method. A p value < 0.05 and a Log2 fold change cut-off of 0.5 were used to select DEGs. A full list of DEGs between each comparison can be found in Supplementary File 1 .
Pathway analysis on up-and down-regulated DEGs between each comparison were performed using Enrichr ( 21 ). Over-representation of pathways from KEGG Mouse 2019 and Reactome 2016 databases were mapped using DEGs as input. The enrichment of upregulated ligands from the LINCS L1000 connectivity map were also analyzed using DEGs in Enrichr. Upstream regulator analysis was performed with DEGs and associated log fold changes as input, using the Ingenuity Systems program ( 22 ). Default settings were used and activation Z‐scores were used to determine the activation state of each upstream regulator. Those with activation Z-scores of ≥ 2 were considered ‘activated’ while activation Z-scores of ≤ -2 were considered ‘inhibited’. Upstream regulators included cytokines, transcription regulators, complexes, enzymes and kinases. For these analyses, p values were adjusted for multiple comparisons using the Benjamini-Hochberg method and p < 0.05 was considered significant.
Count data was scaled up to library size using Tximport ( 19 ) resulting in scaled transcripts per million (TPM) normalized count matrices. Heatmaps with unsupervised hierarchical clustering of the top 200 variable DEGs, determined by standard deviation were performed using the pheatmap package in R (v3.6). Gene set enrichment analysis (GSEA) was completed on the normalized gene expression data using 50 MSigDB hallmark gene sets on the Broad Institute software ( 23 ). Gene sets enriched with a FDR > 0.25 were considered significant. A total of 1000 permutations were performed, and all other default parameters were used. CIBERSORTx was used to identify immune cell populations in normalised RNAseq data as previously described ( 17 ).
Statistical Analysis
Data are presented as mean ± SD. For flow cytometry experiments, statistical analyses were performed using Mann-Whitney U tests with multiple comparisons to compare between monotherapies and combination chemo/immunotherapy-treated samples using GraphPad Prism v8. Survival data were analyzed using Log-rank (Mantel-Cox) test in GraphPad Prism v8. To compare combination treatments to monotherapy controls, hazard ratios (HR) were calculated using logrank analysis of survival curves to determine agonistic or antagonistic effects. To further define additive interactions, as described before ( 24 ), HR was calculated for each treatment group compared to PBS or best monotherapy treated controls. Additive effects were defined as HR(combination) < [HR(combination) - HR(mono 1) - HR(mono 2) + 1]. Results of these analyses are displayed in Table S3 .
Results
5-FU and Cisplatin Generate Robust Anti-Tumor Responses When Combined With ICB
The addition of ICB with chemotherapy regimens are increasingly being trialed in the clinic to improve patient outcomes ( 25 ). However, the impact of individual chemotherapies on ICB efficacy remains unclear. To assess anti-tumor responses of chemotherapy when combined with ICB in vivo, we screened 10 chemotherapeutics from different canonical classes in combination with anti‐CTLA‐4/anti‐PD-L1 antibodies in two murine mesothelioma models ( Figures 1A, B ). As there is a difference in the therapeutic response to the different chemotherapies ( Figures S2, 3 ), we compared survival of the combination therapy with survival of the best monotherapy (either chemotherapy or ICB alone) and plotted each as a hazard ratio (HR). ICB alone induced complete tumor regression in 0-30% of AB1 tumor bearing animals but not in the AE17 model. When combined with ICB, all tested chemotherapies had varying effects on anti-tumor efficacy across the two models ( Figures 1C, D , S2, 3 , Table S3 ). Gemcitabine, irinotecan, doxorubicin and bleomycin provided no benefit when combined with ICB in either AB1 (HR = 1.14, 0.979, 0.845, 0.845, 0.692 respectively) or AE17 (HR = 0.929, 1.14, 1.07, 0.759 respectively). ICB provided no further benefit when added to cyclophosphamide in AB1 (HR = 0.895), but the combination significantly improved median survival in AE17 (HR = 0.244). The combination of vinorelbine or etoposide with ICB was antagonistic in AB1 (HR = 3.65, 4.08) but had no effect in AE17 (HR = 1.96, 0.391). The reverse was the case for pemetrexed (HR = 0.692 in AB1, 2.261 in AE17).
FIGURE 1
Figure 1 Different combinations of chemotherapy and immune checkpoint blockade (ICB) demonstrate additive and antagonistic responses. (A, B) Treatment schedule for mice inoculated with AB1 (A) or AE17 (B) mesothelioma cell lines. (C, D) Hazard ratio (HR) analysis of survival plots comparing combination chemotherapy and ICB (anti-CTLA‐4/anti‐PD-L1) to the best performing monotherapy in AB1 (C) and AE17 (D). HR is defined as the risk of a negative (death) outcome occurring in one group at the next instance of time, compared to another group at the same time. A lower ratio i.e., less than 1 indicates a higher rate of survival in the chemo-immunotherapy combination compared to monotherapy. (E) Survival curves of 5-FU chemo-immunotherapy combinations in AB1 (left; n = 8-10 per group, two pooled experiments) and AE17 (right; n = 5 per group, one experiment). (F) Survival curves of cisplatin chemo-immunotherapy combinations in AB1 (left; n = 5 per group, one experiment) and AE17 (right; n = 13-15 per group, two pooled experiments). Mantel-Cox survival test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
The combination of 5-FU with ICB (5-FU+ICB) resulted in robust anti-tumor responses in the AB1 model (HR = 0.101), with significant increase in median survival compared to both 5-FU (p = 0.0001) and ICB (p = 0.005) monotherapy. Complete tumor regression occurred in >80% of 5-FU+ICB treated animals, compared to 0-20% or 20-30% complete responders in 5-FU or ICB monotherapy, respectively ( Figure 1E ). 5‐FU+ICB was also additive in AE17 (HR = 0.308), with an increase in median survival compared to the monotherapies (5-FU: p = 0.199, ICB: p = 0.0142). Cisplatin and ICB (cisplatin+ICB) were additive in both AB1 (HR = 0.610) and AE17 (HR = 0.286) ( Figure 1F ). In AE17, cisplatin+ICB significantly increased median survival compared to cisplatin (p = 0.0025) and ICB (p

Images Powered by Shutterstock