Three-dimensional colon cancer organoids model the response to CEA-CD3 T-cell engagers

Abstract

Rationale: The CEA-CD3 T cell bispecific antibody cibisatamab (CEA-TCB) is currently undergoing clinical trials. Here we study its performance against three-dimensional tumor organoids in cocultures with T cells as compared to a higher affinity CEACAM5-CD3 (CEACAM5-TCB) bispecific antibody using time-lapse confocal microscopy. Methods: Pre-labelled spheroids derived from colon cancer cell lines and primary organoids derived from four colorectal cancer surgical specimens, which expressed different graded levels of CEA, were exposed in cocultures to T lymphocytes. Cocultures were treated with CEA-CD3 T-cell engagers and were followed by live confocal microscopy. Caspase 3 activation detected in real-time was used as an indicator of tumor cell death. Co-cultures were also set up with autologous tumor-associated fibroblasts to test the co-stimulatory effect of a fibroblast activated protein (FAP)- targeted 4-1BBL bispecific antibody fusion protein currently undergoing clinical trials. Results: Tumor-cell killing of 3D colon carcinoma cultures was dependent on the levels of surface CEA expression, in such a way that the lower affinity agent (CEA-TCB) did not mediate killing by human preactivated T cells below a certain CEA expression threshold, while the high affinity construct (CEACAM5-TCB) remained active on the low CEA expressing organoids. Modelling heterogeneity in the levels of CEA expression by coculturing CEA high and low organoids showed measurable but weak bystander killing. Cocultures of tumor organoids, autologous fibroblasts and T cells allowed to observe a costimulatory effect of anti-FAP-4-1BBL both to release IFNγ and to attain more efficacious tumor cell killing. Conclusion: Three-dimensional tumor cocultures with T cells using live confocal microscopy provide suitable models to test the requirements for colon-cancer redirected killing as elicited by CEA-targeted T-cell engagers undergoing clinical trials and treatment allow combinations to be tested in a relevant preclinical system.

Keywords: Colon organoids; T-cell engager; colon cancer; live confocal microscopy.

Conflict of interest statement

Competing Interests: IM reports receiving commercial research grants from BMS, Bioncotech, Alligator, Pfizer, Leadartis and Roche; has received speakers bureau honoraria from MSD; and is a consultant or advisory board member for BMS, Roche, Genmab, F-Star, Bioncotech, Bayer, Numab, Pieris, Alligator, and Merck Serono. PU, VK, MC, CK and MB are employed by Roche and declare stock ownership and patents with Roche. All other authors declare no competing interests.

© The author(s).

Figures

Figure 1

Modelling cytotoxicity of CD3-CEA bispecific T-cell engagers of different affinity on 3D tumor spheroids. (A) Representative flow cytometry histograms of CEA surface expression on LS174T cells. (B) Representative images of cocultures of LS174T spheroids (Red) with T cells (Blue) and the CEA-TCB, CEACAM5-TCB T-cell engagers and an irrelevant mouse IgG1 control (Ctrl). Active caspase 3 is shown by a fluorescent probe (Green) that marks apoptotic tumor cells (Bars 50 μm). (C) Quantification of apoptotic cells in images as in B and videos as Video S1. (D) Representative flow cytometry histogram of CEA surface expression on HT29 cells. (E) Representative images of cocultures of HT29 spheroids (Red) with T cells (Blue) and the CEA-TCB, CEACAM5-TCB T-cell engagers and an irrelevant mouse IgG control (Ctrl). Active caspase 3 is shown by a fluorescent probe (Green) that indicates apoptotic tumor cells (Bars 50, 20 μm). (F) Quantification of apoptotic cells in images as in E and videos as Video S1. Representative experiments are shown out of at least two rendering similar results. Quantifications in C and F were performed analyzing four to seven spheroids per condition. Means±SD are shown for quantitative data. * p<0.05; U Mann Whitney tests were performed for statistical comparisons.

Figure 2

Modelling T-cell engager-induced cytotoxicity on primary colon cancer organoids. Organoids (Red) generated from surgical specimens from four different colon cancer patients were cocultured with T cells (Blue) and the CEA-TCB, CEACAM5-TCB T-cell engagers and an irrelevant mouse IgG control (Ctrl). Active caspase 3 is shown by a fluorescent probe (Green) that indicates apoptotic tumor cells. Panels A, D, G and J show representative histograms of CEA surface expression of such four organoids as measured by flow cytometry. B, C, H and K show representative confocal microscopy snapshots of time-lapse videos of cocultures of the four different organoids treated with bispecific T cell engagers. Panels C, F, I and L show quantitative data of caspase 3 activation as in Video S2 at different time points. Data come from at least two experimental replicates rendering similar results. Quantifications were performed analyzing four to eight organoids per condition. Means±SD are shown for quantitative data. * p<0.05; U Mann Whitney tests were performed for statistical comparisons.

Figure 3

Limited bystander killing mediated by T-cell engagers. (A) Representative immunohistochemistry of CEA expression in a case of colorectal cancer showing wide heterogeneity in CEA intensity in different regions. (B) Cocultures of T cells with the organoid 53436 expressing medium levels of CEA were treated with CEA-TCB, CEACAM5-TCB or irrelevant control (Ctrl) and recovered after 24 hours. Relative numbers of tumor cells in the cocultures and surface CEA levels were then assessed then by flow cytometry. Correlation of surface CEA expression with the percentage of tumor cells in culture as a surrogate marker for T cell-mediated killing. (C,D) Cocultures of HT29 (CEA low spheroids, pink) and LS174T (CEA high spheroids, Red) were setup with T cells (Blue) and a caspase 3 active fluorescent detection probe (Green) to follow tumor cell killing over time. In another well the same cocultures were placed using HT29 cells only. Both conditions were treated with CEA-TCB T-cell engagers and followed overtime by time-lapse confocal microscopy. (C) Representative snapshots showing such cocultures. (D) Quantification of videos as in C when compared to cocultures of HT29 spheroids alone performed in parallel. (E,F) A similar experiment to the one using spheroids in C and D was performed using CEAHigh organoids (Patient 47389 (Purple) and CEALow organoids (Patient 47550 (Red)). (E) Representative snapshots showing such cocultures over time. (F) Quantification of videos as in C in comparison with cocultures of 47550 organoids co-cultured with T cells alone performed in parallel. Means±SD are shown for quantitative data.

Figure 4

Bystander killing mediated by CEA targeted T-cell engagers is increased when augmenting the proportion of CEAHI tumor cells and is dependent on INFγ. (A) Cocultures of HT29 (CEA low spheroids, pink) and LS174T (CEA high spheroids, Red) were setup with T cells (Blue) and a caspase 3 active fluorescent detection probe (Green) at increasing LS174T-spheroid to HT29-spheroid ratios. All conditions were treated with CEA-TCB T-cell engagers and followed overtime by time-lapse confocal microscopy. Representative snapshots showing such cocultures and a quantification of the videos at the 16 hours' time point are shown. (B) Cocultures of HT29 (CEALow spheroids, pink) and LS174T (CEAHigh spheroids, Red) (50:50 ratio) were setup with T cells and treated with CEA-TCB (1 μg/mL) and anti FAS-L (10 μg/mL), anti IFNγ (4 μg/mL) or an irrelevant IgG control (10 μg/mL). Representative snapshots showing such cocultures at the end of the videos (14 hours) and a quantification of caspase 3 activity induction overtime are shown. Means±SD are shown for quantitative data. Statistical comparisons were performed with U Mann Whitney tests.

Figure 5

Co-cultures of tumor organoids with autologous FAP+ fibroblasts do not impair CEA redirected killing by T cells. (A) Representative images of a coculture of tumor cell organoids (Red) and tumor-associated fibroblasts (Green) from the patient 47389 with allogeneic T cells. A 3D reconstruction and a single plane of the same two organoids are shown to better visualize the interactions of tumor cells with fibroblasts. (B) Series of Z-stack planes of an organoid/fibroblast/T-cell co-culture as in A to show close contact of fibroblasts with organoids. (C) Time-lapse sequence of a triple Organoid/fibroblast/T-cell coculture treated with CEA-TCB antibodies and showing tumor cell cytotoxicity by activated caspase 3. (D) LS174T spheroids were set up in cocultures with T cells and CEA-TCB antibody or IgG1 irrelevant control antibody (Ctrl) and different amounts of fibroblasts (0,5x103, 1.5x104, 3x104 fibroblasts) were added to the culture. Numbers of recovered tumor cells assessed by flow cytometry after 24 hours of coculture are shown. (E) Representative flow cytometry histogram of fibroblast activated protein (FAP) expression on fibroblast in cultures from patients 47389 and 47550. (F) CD137 induction on CD8 T cells in cocultures as in D as measured by flow cytometry.

Figure 6

Costimulation of CEA T-cell engagers by a FAP targeted 4-1BB ligand. (A,B,C) Spheroids from LS174T cell line (A) or organoids from patients 47315 (B) and 47389 (C) were set up in cocultures with tumor associated fibroblasts and T cells and to be treated with irrelevant mIgG1 antibody, CEA-TCB plus untargeted control 4-1BBL (DP47-4-1BBL) or FAP targeted 4-1-BBL (FAP-4-1BBL). 72 hours after the coculture setup, IFNγ in the supernatants was measured by ELISA. Results come from two pooled experiments with triplicate wells per condition. (D and F) Representative microscopy images of cocultures of tumor organoids 47389 (D) and 52436 (F) (Red), autologous fibroblasts (Green) and T cells (blue) treated with CEA-TCB plus DP47-4-1BBL or FAP-4-1BBL following 48 hours of coculture. Induction of apoptosis in tumor cells is shown by the caspase 3 activity probe also in green. The green channel of the images is shown apart to better identify caspase 3 positive cells and differentiate them from green fibroblasts. (E and G) Quantitative assessment of caspase 3 activity induction on tumor organoids overtime as in Video S5 showing fibroblast/organoid/T-cell co-cultures treated with CEATCB plus UT or FAP 4-1BBL using 47389 CEAHigh organoids (E) and 53436 CEALow organoids (F). D to G show representative data from three and two experiments rendering similar results. Means±SD are shown for quantitative data. U Mann-Whitney tests were used to assess for individual significant differences between the relevant conditions.