Organotypic tumor slice cultures provide a versatile platform for immuno-oncology and drug discovery

Abstract

Organotypic tumor slices represent a physiologically-relevant culture system for studying the tumor microenvironment. Systematic characterization of the tumor slice culture system will enable its effective application for translational research. Here, using flow cytometry-based immunophenotyping, we performed a comprehensive characterization of the immune cell composition in organotypic tumor slices prepared from four syngeneic mouse tumor models and a human liver tumor. We found that the immune cell compositions of organotypic tumor slices prepared on the same day as the tumor cores were harvested are similar. Differences were primarily observed in the lymphocyte population of a clinical hepatocellular carcinoma case. Viable populations of immune cells persisted in the tumor slices for 7 days. Despite some changes in the immune cell populations, we showed the utility of mouse tumor slices for assessing responses to immune-modulatory agents. Further, we demonstrated the ability to use patient-derived xenograft tumor slices for assessing responses to targeted and cytotoxic drugs. Overall, tumor slices provide a broadly useful platform for studying the tumor microenvironment and evaluating the preclinical efficacy of cancer therapeutics.

Keywords: Tumor slices; cancer; ex vivo models; immuno-oncology; syngeneic model.

Figure 1.

Organotypic tumor slices can be made from a broad range of tumor tissues and used to test multiple hypotheses. (a) A schematic showing the experimental workflows in which organotypic tumor slice cultures can be used. (b) A summary of different tumor slice cultures used in this study and the experiments performed.

Figure 2.

Immunophenotyping of organotypic tumor slices reveals a complex immune population. (a) The gating strategy used for identifying the immune cell composition in tumor samples is represented using polychromatic dot plots of cells from a mouse tumor derived from subcutaneously injected syngeneic breast cancer cells (Py8119). Cells were analyzed on the day of tumor harvest (Day 0). Viable cells gated in the top left were depleted of doublets based on scatter plot characteristic and separated into CD45⁺ and CD45⁻ populations. Lymphocytes were gated as % CD45⁺ cells – T cells: CD3⁺ B220⁻; B cells: B220⁺ CD3⁻. T cell subsets were gated as % CD3⁺ cells – CD4⁺ CD8⁻ TILs and CD8⁺ CD4⁻ TILs. Myeloid cells were gated as % CD45⁺ cells – TAM: CD11b⁺ F4-80⁺/CD11b⁺ CD68⁺/CD163⁺ CD66b⁻; TANs: CD11b⁺ F4-80⁻ Ly6G⁺/CD11b⁺ CD66b⁺; TA-Monocytes: CD11b⁺ F4-80⁻ Ly6G⁻ Ly6C⁺ and TADCs: CD11b⁺ F4-80⁻ Ly6G⁻ Ly6C⁻ CD11c⁺. See Supplementary Figure S1 for additional information. (b) The immune landscape of both the tumor core and organotypic tumor slices on Day 0 from various mouse syngeneic tumor models and a clinical case of HCC. N = 2–6 cores and 2–12 slices from ≥2 independent experiments for all syngeneic tumors. N = 3 cores and 3 slices for HCC clinical case. Error bars represent SEM.

Figure 3.

Major immune cell populations in organotypic tumor slice cultures persist for 7 days. Immunophenotying of tumor slices from mouse tumors derived from subcutaneously injected syngeneic (a) pancreatic (Panc02) and (b) breast cancer cells (Py8119) was performed daily or every other day from Day 0 to Day 7. Plots showing (Top) percentage of viable cells and CD45⁺ leukocytes; (Bottom) Cell count of each of the immune cell populations in a slice. N = 4–10 slices per timepoint from 2 independent experiments. Error bars represent SEM.

Figure 4.

Stimulation of pancreatic tumor slices with IFNγ alters the immune cell population and activation state. (a) Tumors from Panc02 cells injected into syngeneic mice were harvested, slices prepared, and IFNγ was added. After 60 hr, immunophenotyping was performed. Control slices were cultured for 60 hr without IFNγ. (b) Plots showing (Left) overall viability and ratio of viable leukocytes to non-leukocytes; (Right) % of CD4⁺ and CD8⁺ TILs. Plots showing the proportion of cells positive for stimulatory or inhibitory markers on (c) TAMs and (d) TADCs. N = 3 tumor slices per treatment. Error bars represent SEM. * denotes p < .05, ** denotes p < .01, *** denotes p < .001, and **** denotes p < .0001. (e) Plot of changes in the signal intensity of proteins associated with growth factor signaling and apoptosis n Panc02 tumor slices after treatment with IFNγ as measured by RPPA. Data are normalized to the intensity for each protein in the control slices. Data are shown as mean ± SEM (N = 4). * denotes p < .05, ** denotes p < .01, *** denotes p < .001, and **** denotes p < .0001, ANOVA/Dunnet.

Figure 5.

PD-L1 blockade expands and activates TIL population in HCC and CT26 tumor slices. (a–b) Slices prepared from clinical HCC samples were incubated with anti-human PD-L1 antibody or culture media for 48 hr . Immunophenotyping was performed at the end of the assay. (a) Plots showing % of CD3⁺ (left) and CD8⁺ (right) TILs. (b) Plots showing the proportion of cells positive for a stimulatory marker CD27 and its median intensity on CD8⁺ TILs. n = 6-9- tumor slices per treatment from two independent experiments of 2 clinical HCC cases. (c-e). Slices prepared from CT26 colon cancer model were treated with anti-PD-L1 antibody, anti-CTLA-4 antibody or their combination for 48hr followed by FACS-based imunophenotyping. Plots showing (c) % of viable CD45⁺ leukocytes and (d) the levels of PD-L1 on leukocytes. (e) Plots displaying (Left) % of CD3⁺ TILs, (Middle) the levels of activation (CD44) and (Right) inhibitory (PD-1) markers on the surface of CD3⁺ TILs. n = 4–8 tumor slices per treatment. Error bars represent SEM. * denotes p < .05, ** denotes p < .01, *** denotes p < .001, and **** denotes p < .0001.

Figure 6.

Tumor tissue slices prepared from PDX models respond to various cytotoxic and targeted therapeutics. (a) Ex vivo measurement of tissue viability in colon PDX tumor slices. Left, representative IVIS images of viability signal measured in staurosporine (500 nM)-treated slices (N = 2) compared with DMSO-treated control slices (N = 3) after 96 hours. Right, Plot of total flux signal (mean ± SEM) for each treatment condition. p < .01, t-test. (b) Graph of changes in the signal intensity of proteins associated with apoptosis and growth factor signaling in colon PDX slices after treatment with staurosporine (500 nM) as measured by RPPA. Data are normalized to the intensity for each protein in the control slices. Data are shown as mean ± SEM (N ≥ 3). *** denotes p < .001, ** denotes p < .01, *denotes p < .05, t-test (corrected for multiple comparison, Holm-Sidak method). (c) Plot showing the response of breast PDX tumor slices to 18 kinase inhibitors and chemotherapeutic drugs. Data are shown as mean ± SEM (N = 3). * denotes p < .05, ** denotes p < .01, *** denotes p < .005, **** denotes p < .001, ANOVA/Dunnet. (d) Plot showing the response of colon PDX tumor slices to 15 kinase inhibitors and chemotherapeutic drugs. Data are shown as mean ± SEM (N ≥ 2). * denotes p < .05, ** denotes p < .01, ANOVA. (e) Comparison of drug responses in colon PDX tumor slices and matched PDX organoids. Drugs with greater sensitivity in tumor slices compared with organoids are shown in red circles; drugs with greater sensitivity in organoids compared with slices are shown in blue circles. Black circles are drugs with a consistent effect on both systems. Each replicate is plotted separately.