(M1030-05-26) The Tumor Microenvironment in Three-Dimensions: Immune Cells, Cytokines, and Tumor Invasion

Purpose: Cancer comprises a complex and heterogeneous amalgam of cell types, many of which are non-cancerous. Malignant progression of cancer within its three-dimensional (3D) organization is facilitated by cellular interactions with a multitude of microenvironmental components. Our group has developed a 3D ex-vivo culture platform which allows for control of media delivery and waste metabolite removal via perfusion through the soft granular support medium, Liquid Like Solids (LLS), which mimic a capillary bed by creating an open porous network between the particles. We aimed to develop methods to make spatiotemporal measurements, in-situ, of cytokine concentrations and immune cell motion in the tumor microenvironment as well as investigate the dynamics of tumor invasion in 3D by using fast scanning laser confocal microscopy.

Methods: Microexplants on the order of 200-400 µm in characteristic diameter were created from a resected glioblastoma tumor. From a blood draw, PBMC’s were purified and labelled with CellTrace CFSE and placed into co-culture in LLS using supplemented organoid media. Making use of the mechanical stability of LLS, 3D printed arrays or random dispersions of ELISA beads can be made throughout the experiment and then leveraged to make spatiotemporal measurements of cytokine gradients, in-situ. The cytokine on-rates and off-rates were used alongside measurements of each bead’s fluorescence intensity at specific times and positions in relation to the tumor to fit spatiotemporal reaction-diffusion models that allowed for extraction of the tumor’s production rate, concentrations at the margin, and quantification of the entire immuno-regulatory microenvironment. Additionally, confocal imaging at 5-minute intervals provided high resolution data with the temporal fidelity to monitor tumor evolution-dynamics by tracking the 3D motion, proliferation, action, death, and elimination of both tumor cells and immune cells. Functionalization of LLS microgel particles with Collagen-1 allowed for cellular adhesion to the particles thus facilitating adhesion-dependent opportunistic invasion by cancer cells to available spaces in the surrounding areas in a manner independent of proteolytic activity. Artificial Intelligence (AI) algorithms, originally developed for astrophysics data processing, were used to perform the cell tracking.

Results: Gleaned from the reaction-diffusion model of cytokine concentration, a value of 2 ng/ml of IL8 was found at the tumor margin after 10 hours and a cellular IL8 production rate was quantified to be 2 molecules per cell per second. Super diffusive behavior was seen in the invasive fronts emanating into the proximal interstitial space from the micro-tumor. It was revealed via simulation with off-lattice agent-based simulation that invasion paths were restricted mechanically by the interstitial spaces between the LLS particles, thus accounting for the super-diffusive behavior. The glioblastoma cells were seen to utilize the Collagen-1 functionalized LLS particles to explore their surroundings via anchorage-dependent migration along with geometrical cues which guided their invasion along accessible paths in 3D space - tortuosity measurements revealed monotonic increases in perimeter tortuosity reaching values above 3 within 48 hours.  The COL-LLS allows invasion without proteolytic degradation of the environment. Analysis of the mean-squared displacement of the tracked CD8+ immune cells yielded an average migration speed of greater than 2.8 µm/min. Investigation of the tracking data further revealed both chemotaxis and chemokinetic behavior. The average rate of CD8+ T-cell killing was found to be around 3 cancer cells per hour, which monotonically decreased after 12 hours to about 1 cancer cell per hour.

Conclusion: Successful quantification of spatiotemporal cytokine concentrations, T-cell motion and activity, and tumor invasion dynamics was performed with this in-vitro immune-oncology platform via in-situ fast scanning fluorescence microscopy.

Acknowledgments: WGS is an equity shareholder in Aurita Bioscience. Funding for DTN was provided from an NSF Graduate Student Fellowship. DP is funded through a research Grant with Alcon Laboratories. WGS and MS are Faculty at the University of Florida.

A) Biopsied tissue sample and blood draw from properly consented donor. B) The tissue sample is mechanically disaggregated into microexplants on the order of 1µl each. C) Microexplant samples are collected in the BioPelle, a custom micromanipulator mounted on a microscope, for dispensing into sample wells. D) Immune cells are separated from the blood draw, sorted for CD8+ and CD4+ T cells and stained with CFSE prior to dispersing in LLS and being dispensed into culture wells. E) Microexplants are placed into a 3D dispersion of immune cells in LLS in each culture well and a liquid cover is added atop the LLS layer. F) Maximum intensity projection image of glioblastoma microexplant (cyan) in a field of immune cells (green) within the LLS microgel bed (red).



A) Glioblastoma microexplant over the course of 48 hours in inert polyacrylamide LLS (top) and Collagen-1 coated polyacrylamide LLS (bottom). Where Collagen-1 context is provided, invasion outward into the surrounding regions is seen while the microexplant without Collagen context remains adherent to self. B) Outlines of the microexplant cultured in Collagen-1 coated LLS over the course of 48 hours. C) Microexplant outline progression over duration of the experiment (48 hours). D) Tortuosity index over time showing monotonic increases in LLS with COL functionalization.



(top) ELISA beads measure the local concentration of tumor secreted IL8. (bottom) Spatiotemporal models can measure tumor production rates of cytokines in real-time.