Computationally guided design of a metastasis-on-a-chip platform for quantitative evaluation of chemotactic cues in developmental cancers

Technical Summary: Computationally Guided Design of a Metastasis-on-a-Chip Platform

Problem Statement
Metastatic dissemination remains the primary cause of death in children with solid tumors, yet the earliest stages of this process are poorly understood. Traditional migration assays (e.g., Boyden chambers, Transwells) fail to reproduce the dynamic, spatially organized biochemical gradients that tumor cells encounter in vivo. These conventional methods offer limited control over the temporal and spatial organization of chemotactic cues, often resulting in poorly defined gradients that dissipate over time. Furthermore, they struggle to distinguish directed chemotaxis from increased random motility (chemokinesis). While microfluidic "organ-on-a-chip" technologies offer a solution, their design typically relies on empirical trial-and-error, consuming significant time and biological resources before achieving biologically relevant conditions. Additionally, the concentration of growth factors loaded into a device does not necessarily reflect the concentration experienced by cells due to diffusion attenuation across complex geometries.

Methodology
The authors developed a "computationally guided" metastasis-on-a-chip (MET-on-a-chip) platform based on the concept of the Minimally Functional Unit (MFU). This engineering philosophy dictates that only the biological components strictly necessary to answer a specific scientific question should be included, avoiding unnecessary complexity.

• Device Design and Fabrication: The platform consists of two independent culture chambers connected via intermediate reservoirs and an array of 125 parallel, confined microchannels (15 µm wide, 1.38 mm long). This geometry mimics the physical constraints of microvascular dissemination. The device was fabricated using two-layer SU-8 photolithography and PDMS soft lithography.
• Computational Modeling (COMSOL): Before fabrication or biological experimentation, finite-element simulations using COMSOL Multiphysics were employed to predict molecular transport. The models accounted for diffusion, first-order degradation, and the specific geometry of the device. These simulations were used to:
  • Predict the establishment of concentration gradients for VEGF-A165 and VEGF-C.
  • Determine the minimum loading concentrations required to generate biologically relevant exposure at specific locations.
  • Identify the optimal positioning of tumor cells relative to the predicted gradients.
• Experimental Validation:
  • Diffusion Validation: 70-kDa FITC-dextran was used to visualize and quantify macromolecular diffusion across the device, confirming the formation of stable spatial gradients.
  • Release Kinetics: ELISA assays quantified the release of VEGF-A165 and VEGF-C from collagen hydrogels to verify loading conditions.
  • Biological Assay: Human neuroblastoma (SK-N-BE(2)), Ewing sarcoma (SK-N-MC), and osteosarcoma (U2OS) cells were embedded in collagen-fibrin hydrogels. Migration through the microchannel array was quantified under control conditions and in the presence of VEGF-A165 (vascular cue) or VEGF-C (lymphatic cue).

Key Results

1. Computational Guidance: Simulations revealed that the concentration of growth factors experienced by cells is orders of magnitude lower than the loading concentration due to diffusion through reservoirs and microchannels. Crucially, the models dictated that tumor cells must be positioned in the reservoir adjacent to the microchannel array (Reservoir 1) rather than in the main culture chamber to receive biologically relevant exposure.
2. Gradient Formation: Experimental validation with FITC-dextran confirmed the formation of stable, long-range concentration gradients across the device, aligning with computational predictions.
3. Tumor-Specific Responses:
   • Ewing Sarcoma and Osteosarcoma: Both cell types exhibited a significant increase in migration through the confined microchannels in response to VEGF-C. VEGF-A165 did not induce a statistically significant migratory response in these lines.
   • Neuroblastoma: SK-N-BE(2) cells showed minimal migration under all conditions, indicating a lack of responsiveness to isolated VEGF-A165 or VEGF-C gradients.
4. Receptor Expression: Transcriptomic analysis of public datasets confirmed that VEGF receptors (KDR/VEGFR2 and FLT4/VEGFR3) are expressed across these pediatric tumor types, validating the biological relevance of the study, though bulk expression did not predict the functional migratory response.

Significance and Claims
The paper claims to establish a rational workflow for designing microphysiological systems, shifting the paradigm from empirical optimization to predictive engineering. By integrating computational modeling before experimentation, the authors demonstrate that:

• Assay Design: Device geometry, growth factor loading, and cell positioning can be rationally defined to ensure cells are exposed to physiologically relevant concentrations, a factor often overlooked in microfluidic assays.
• Mechanistic Insight: The platform successfully isolated the direct chemotactic effects of VEGF ligands on tumor cells, independent of endothelial or stromal interactions. The results suggest that while VEGF-A165 is a known driver of angiogenesis, it may not act as a direct chemoattractant for these specific pediatric tumor cells under confined conditions. Conversely, VEGF-C acts as a potent direct chemoattractant for Ewing sarcoma and osteosarcoma cells.
• Modularity: The MFU approach allows for the progressive addition of biological complexity (e.g., endothelial barriers, immune cells) only when required to answer specific questions, thereby improving reproducibility and mechanistic interpretation.

The authors conclude that this computationally guided strategy provides a robust framework for developing minimally functional systems capable of dissecting individual steps of the metastatic cascade, potentially serving as a preclinical tool for evaluating anti-metastatic therapies in a personalized setting.