Multiomic analysis for optimization of combined focal and immunotherapy protocols in murine pancreatic cancer

Abstract

Background: Although combination immunotherapies incorporating local and systemic components have shown promising results in treating solid tumors, varied tumor microenvironments (TMEs) can impact immunotherapeutic efficacy. Method: We designed and evaluated treatment strategies for breast and pancreatic cancer combining magnetic resonance-guided focused ultrasound (MRgFUS) ablation and antibody therapies. With a combination of single-cell sequencing, spectral flow cytometry, and histological analyses, we profiled an immune-suppressed KPC (Kras^+/LSL-G12D; Trp53^+/LSL-R172H; Pdx1-Cre) pancreatic adenocarcinoma (MT4) model and a dense epithelial neu deletion (NDL) HER2⁺ mammary adenocarcinoma model with a greater fraction of lymphocytes, natural killer cells and activated dendritic cells. We then performed gene ontology analysis, spectral and digital cytometry to assess the immune response to combination immunotherapies and correlation with survival studies. Result: Based on gene ontology analysis, adding ablation to immunotherapy enriched immune cell migration pathways in the pancreatic cancer model and extensively enriched wound healing pathways in the breast cancer model. With CIBERSORTx digital cytometry, aCD40 + aPD-1 immunotherapy combinations enhanced dendritic cell activation in both models. In the MT4 TME, adding the combination of aCD40 antibody and checkpoint inhibitors (aPD-1 and aCTLA-4) with ablation was synergistic, increasing activated natural killer cells and T cells in distant tumors. Furthermore, ablation with immunotherapy upregulated critical Ly6c myeloid remodeling phenotypes that enhance T-cell effector function and increased granzyme and protease encoding genes by as much as 100-fold. Ablation combined with immunotherapy then extended survival in the MT4 model to a greater extent than immunotherapy alone. Conclusion: In summary, TME profiling informed a successful multicomponent treatment protocol incorporating ablation and facilitated differentiation of TMEs in which ablation is most effective.

Keywords: Combination immunotherapy; Digital cytometry; Focused ultrasound; Sequencing; Spectral cytometry.

Figure 1

Schematic of immune effects generated by the combination of ablation and immunotherapy. Overview of the study. Single-cell RNA sequencing (scRNAseq) and spectral cytometry profiled treatment-naïve tumors to guide combination treatment strategies while next generation sequencing based analyses, such as gene ontology analysis and digital cytometry of treatment combinations, were compared with spectral cytometry characterizations. Tumors were sorted for immune cells prior to scRNAseq and spectral cytometry.

Figure 2

Analysis of the naïve tumor microenvironment of MT4 pancreatic and NDL breast cancer models reveals distinct TMEs and immunological signatures. A-D) Hematoxylin and eosin staining before (A-B) and after (C-D) ablation in MT4 (A, C) and NDL (B, D) models. A) MT4 pancreatic ductal adenocarcinoma has decreased cellularity compared to B) NDL mammary adenocarcinoma, which is comparatively well vascularized (white arrows) with scattered leucocytes (black dots). E-K) Results of single-cell RNA sequencing, including Uniform Manifold Approximation and Projection (UMAP) plots for the MT4 (E) and NDL (F) tumors. B cells (dark pink) were CD19⁺, CD79⁺ and Ly6d⁺. T cells were CD3e⁺, with CD8⁺ (dark blue) and CD4⁺ (light blue) T cell subsets defined by CD8a and CD4, respectively. NK cells (lavender) were Klrb1b⁺ and Klrb1c⁺ (NK1.1⁺). Eosinophils (yellow) were Siglec-F⁺. Neutrophils (light orange) were Ly6g⁺. Monocytes (light pink) were Ly6c⁺, Ccr2⁺, Mrc1⁺, and Ccl9⁺. Macrophages (green) were Itgam⁺ and Adgre1⁺. Dendritic cells (purple) were Itgax⁺, H2-Ab1⁺, Fcgr1⁺, Ly6g^-, Siglecf^-, Klrb1c^-. Granulocytes (turquoise) were Tmem189⁺, Sap30⁺, and Idha⁺. G) Quantitative summary of immune cells within each tumor model. The MT4 model has a smaller faction of CD8⁺ T cells, macrophages and NK cells compared to the NDL model. H-K) Gene expression distribution and levels across UMAP cell clusters in the MT4 (H, J) and NDL (I, K) models. In the MT4 model, overall cellular Myd88 expression levels are higher, and CD40 expression is higher in dendritic cell (DC) clusters compared to the NDL model. The MT4 model also has a greater fraction of both CD4⁺ and CD8⁺ T cells expressing PD-1, CD8⁺ T cells expressing CTLA4, and DCs expressing CD40. Scale bars represent 300 µm.

Figure 3

Ablation or agonist CD40 (aCD40) treatment combined with aPD-1 has reduced, but immune-targeted, effects on gene expression in MT4 pancreatic tumors as compared with highly-differentiated NDL breast tumors. Two-component treatment protocols of aPD-1 + ablation (n=4, treated tumor, (A-aPD-1-T)) or aCD40 + aPD-1 (n=3 or 4, aCD40-aPD-1) were delivered as a one-time treatment to MT4 (B-E) or NDL (F-I) tumor-bearing mice and compared with a no treatment cohort (n=4). A) Protocol and processing methodology. B-E) Changes in gene expression and ontologies in the MT4 model after (B, D) aPD-1 combined with ablation (A-aPD-1-T) or (C, E) aPD-1 combined with aCD40 (aCD40-aPD-1) treatment. Gene ontologies increased by A-PD-1-T: leukocyte migration and chemotaxis; by aCD40-aPD-1: leukocyte and receptor activation. In the NDL model, aPD-1 was combined with (F, H) ablation or (G, I) aCD40. In the NDL model, gene ontologies increased by A-PD-1-T: wound healing; by aCD40-aPD-1: leukocyte migration, cell adhesion and chemotaxis. Differential expression based on comparison to no treatment control is displayed for an adjusted p value < 0.05 and a fold change > 2, and the changes were subsequently used for gene ontology analysis.

Figure 4

Comparing digital cytometry results for two-component treatment with aCD40 + aPD-1 or ablation + aPD-1 in the MT4 pancreatic and NDL breast cancer models. aCD40 + aPD-1 increases leukocytes and activated dendritic cell numbers. Digital cytometry was applied to bulk RNA sequencing data acquired in the MT4 and NDL models under the protocol in Figure 3A. A-C) MT4 model. D-F) NDL model. A, D) CIBERSORTx absolute score. B, E) Fold change from the no treatment control (NTC) cohort, plotted between ablation + aPD-1 in the directly-ablated tumor (A-aPD-1-T) and aCD40 + aPD-1 (aCD40-aPD-1). C, F) Fold change from the NTC cohort, plotted between ablation + aPD-1 in the distant tumor (A-aPD-1-C) and aCD40 + aPD-1 (aCD40-aPD-1). Note that log ratios are based on CIBERSORTx absolute scores. RNAseq experiments were performed with n = 4 replicates with a negative binomial test and Bonferroni correction for p values. Expression of genes in the grey region of E-F was zero in the NTC. Abbreviations: Mast cells (MCs), Polymorphonuclear leukocytes (PMN).

Figure 5

Applying spectral cytometry to phenotype individual immune cells following treatment combinations of aCD40 and checkpoint inhibitors in the MT4 tumor model revealed mobilized monocytes and increased T-cell and NK-cell effector phenotypes. MT4 tumor-bearing mice were treated based on the protocol in Figure 3A, comparing a one-time injection of aCD40 to an injection of aCD40 combined with the checkpoint inhibitors aPD-1 and aCTLA-4 (denoted CP4), or aPD-1 and aCTLA-4 alone, using spectral cytometry at 72 hrs. A) Master pseudocolor UMAP and its annotated version (750,000 total events) for the no treatment control (NTC) (n = 5 tumors) (n = 5 tumors), aCD40 (n = 4 tumors), and CP4 treatments (n = 3 tumors) (250,000 events for each treatment evenly distributed among tumor replicates). Combinations of markers used to label each subset are described in the caption for Figure S12A. B) NTC MT4 scRNA-seq plot annotated using a similar number of the same parameters as used in spectral cytometry. C) Pseudocolor UMAP subplots (NTC and CP4 separately) and Ly6C overlay (NTC and CP4 combined), each derived from the master UMAP (250,000 events for each treatment subplot, 500,000 for Ly6C overlay). The colorbar represents Ly6C expression, where red is high and blue is low to zero.D) Major immune subsets as a percentage of non-granulocyte leukocytes (live, CD45⁺Siglec-F^-Ly6G^-) or lineage- leukocytes (live, CD45⁺Siglec-F^-Ly6G^-CD64^-). E) Representative pseudocolor dot plots of monocyte populations in response to NTC, aPD-1 + aCTLA-4, aCD40, and CP4 treatment (22,831 events each) with Ly6C - BV605 median fluorescence intensity of each subset. The two monocyte populations were distinguished by I-A/I-E presence (inflammatory monocytes were I-A/I-E^- and differentiating monocytes were I-A/I-E⁺) F) T cells (39,220 events each) and G) NK cells (7,971 events each) with respective subsets as a percentage of total T cells or NK cells. The two monocyte populations were differentiated by I-A/I-E presence (inflammatory monocytes were I-A/I-E^- and differentiating monocytes were I-A/I-E⁺). Data in D, E, F, and G are presented as mean ± SD. Statistical analyses were performed using one-way ANOVA with Tukey's multiple comparisons test. ns = non-significant, * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

Figure 6

Digital cytometry analysis in the MT4 model demonstrates the enhanced immune activation resulting from a four-component treatment combining ablation with CP4 (aCD40 + aPD-1 + aCTLA-4). A) Treatment protocol. Mice were treated with two doses of checkpoint inhibition priming prior to an application of checkpoint inhibitors with aCD40 and ablation each added in a subset of mice (n=4 each group) and compared to no treatment control (NTC) mice (n=4). Bulk RNA sequencing was performed 72 hrs after ablation. B-E) Volcano plots showing gene expression response to treatment combinations. B) Ablation + aPD-1 in the distant tumor (A-aPD-1-C) altered expression of 50 genes. C) CP4 altered expression of 285 genes. D-E) Ablation + CP4 resulted in D) 1379 differentially expressed genes in the treated (A-CP4-T) tumor and E) 475 differentially expressed genes in the distant (A-CP4-C) tumor. F) Ablation + CP4 upregulated genes in key immune pathways such as the adaptive immune (GO:0002819), innate immune (GO:0045088) and toll-like receptor (TLR) (GO:0002224) pathways and downregulated the Kras cancer gene in both the treated and contralateral tumors to a greater degree than systemic CP4 treatment alone. G-J) Digital cytometry was applied to bulk RNA sequencing data. Fold change from the NTC is plotted between ablation + CP4 in the ablated tumor (A-CP4-T) versus G) CP4, H) ablation + CP4 in the distant tumor (A-CP4-C), I) ablation-only in the treated tumor (A-T), and J) ablation-only in the distant tumor (A-C). Ablation + CP4 stimulated immune cell changes in both the treated and distant tumor sites, increasing CD4⁺ T cells and dendritic cell and NK-cell activation.

Figure 7

Combination of ablation with CP4 in the MT4 tumor model generates a systemic anti-tumor effect. Tumor growth from the protocols shown in Figure 6A (n=4 for treatments and n=3 for the no treatment control (NTC) cohort). A) Survival for NTC, Ablation alone, Ablation + aPD-1, Ablation + aCTLA-4, Ablation + aCD40, CP4 alone, Ablation + CP4 cohorts. B-G) Tumor growth for B) NTC, C) CP4, D) Ablation in the treated tumor, E) Ablation in the distant tumor, F-G) Ablation + CP4 in the treated (F) and distant (G) tumor. Tumor volume plots are provided in Figure S16. H) Cox-hazard analysis comparing cells to survival outlined the importance of activated dendritic cells. I) Pearson correlation analysis indicated high correlations between 1) survival and dendritic cell (DC) activation, 2) NK cells and CD4⁺ T cells and 3) plasma cells (PCs) and dendritic cell activation.