3D In-vitro Modeling of CAR-T Cell Infiltration in HER2+ Breast Tumors

Introduction:: The lack of immune-mediated cancer cell destruction is a dominant mechanism of immunotherapy failure. HER2+ breast cancer cells overexpress human epidermal growth factor receptor 2, which is an ideal target of CAR-T therapy. However, the role of HER2+ breast tumor cells in modulating CAR-T cell infiltration remains unclear. Chemokines in the tumor microenvironment such as CXCL9 and CCL5, and tumor-T cell interactions have been reported to influence CAR-T cell infiltration in solid tumors, but few studies have examined the trajectories and moving patterns of CAR-T cells in HER2+ breast tumor. In this study, we developed an in-vitro 3D collagen droplet co-culture system to simulate CAR-T cell infiltration in the tumor microenvironment and track T cell motility and interactions with tumor cells in real-time under different micro physiological environments. We employed CAR-T cells overexpressing the SNAP tag (SNAP-CAR T cells), where the contacts between CAR-T cells and HER2+ breast tumor cells can be initiated by adding benzyl guanin-conjugated Herceptin (BG-Herceptin).

Materials and Methods:: HER2+ breast cancer cell lines BT474 and HCC1954, labeled with H2BRFP, were cultured in a 4mg/ml Type I collagen matrix and seeded at 6000 cells per 3ul collagen drop in a 96-well plate. SNAP CAR-T cells, provided by Dr. Jason Lohmueller lab were labeled with green CellTracker and seeded in the surrounding media (Fig 1 A-B). Time-lapse images were captured every 10 minutes for 24 hours with a ZEISS confocal fluorescent microscope. The number of infiltrating CAR-T cells were quantified with the TrackMate plugin in ImageJ, and cell motility was analyzed in R. Metrics such as total displacement, cumulative distance, cell migration speed, mean squared displacement (MSD), track density, fractal dimension and maximum step width were calculated. One-way ANOVA with Bonferroni’s correction was implemented for statistical analysis.

Results, Conclusions, and Discussions:: Results and Discussions: Quantification of infiltrating cell numbers showed that BT474 and HCC1954 cells recruited SNAP-CAR T cells at similar baseline levels. Upon adding BG-Herceptin, contacts between HER2+ breast tumors and SNAP CAR-T cells were activated. CAR-T cell infiltration increased significantly in the BT474 collagen droplets but remained unaffected in HCC1954 droplets. To investigate how tumor cell-derived chemokines affect CAR-T cell motility, we added TAK779, a CXCR3 and CCR5 inhibitor, to the 3D co-culture system. TAK779 significantly reduced the CAR-T cell infiltration in BT474 droplets to baseline levels, but not in HCC1954 droplets (Fig 1 C-D).

Cell motility patterns were further analyzed. In the BT474 droplet, the maximum step width of CAR-T cells was increased from 42.1um to 57.9um with BG-Herceptin, and the median track density increased from 2.8 to 25 within a 200um radius. With TAK779, the CAR-T cell maximum step width (39.4um) and track density (7) returned to baseline levels or lower (Fig 1 E). However, similar trends of CAR-T cell motility in the HCC1954 3D collagen droplets are not shown (Fig 1 F). While both BT474 and HCC1954 cells secreted CXCR3 and CCR5 ligands, HCC1954 secreted more chemokine types that potentially saturated the CAR-T cell infiltration and compensated for TAK779 inhibition.

Conclusions: Our findings suggest that CAR-T cell infiltration towards the solid tumor depends not only on the tumor cell secreted factors but also on tumor-CAR-T cell interactions initiated by BG-Herceptin, highlighting the importance of immunogenic cell death in CAR-T cell recruitment. Furthermore, this study demonstrated the effectiveness of simulating CAR-T cell infiltration in solid tumors using a 3D in-vitro collagen droplet model, and the capability of the quantification of T cell motility analysis with time-lapse confocal imaging. Our low-cost, real-time, versatile and quantitative model has potential applications in immune infiltration studies for solid tumors. Ongoing research is examining the cause of the tumor cell line-dependent T motility changes and the effects of the other immune cells in the immunosuppressive tumor microenvironment on CAR-T cell infiltration, such as tumor-associated macrophages and fibroblasts.





Acknowledgements (Optional): : The authors thanks Dr. Youngbin Cho, Dr. Elisa Ruffo, Michael Kvorjak, Victor So and Avani Bharat Parikh for providing SNAP CAR-T cells and the CAR-T cell maintaining protocol.


References (Optional): :