Development of hiPSC-derived type 2 diabetes mellitus microphysiological system

Introduction:: The pathological development of obesity-induced type 2 diabetes mellitus (T2DM) involves complex heterotypic interactions between tissues and cells. A molecular cascade of a pro-inflammatory environment leads to dysregulated lipolysis in adipose tissue, resulting in ectopic accumulation of lipids in multiple tissues and the subsequent development of insulin resistance and overt T2DM. While animal models provide a platform to study these complex inter-tissue interactions, these models deviate in metabolic features from humans. Thus, using human induced pluripotent stem cells (hiPSC) to generate a relevant combination of cells to build an isogenic in vitro multi-tissue/cellular microphysiological system (MPS) platform can provide an invaluable tool to study the disease. In this study, we developed multi-tissue MPS comprising hiPSC-derived adipocytes, hepatocytes, macrophages, and β cells to model the progression of T2DM with obesity (Figure 1A).



Individual MPS have been developed previously.^1-4 However, major roadblocks to the integration of multiple-tissue MPSs and to recapitulate multi-tissue interactions are three-fold: the design of a common medium that maintains the functionality of the five hiPSC-derived cell types; functional scaling of tissues; and, the design of a multi-tissue MPS. We combined a heuristic differential evolution algorithm to identify a common medium that allowed integration of tissues into a single MPS device and interrogate functional coupling. We show different tissue functional scaling factors in the MPS that can be used to model the pathological progression of obesity induced T2DM. These scaling factors were then used to refine the design of our multi-tissue MPS device and begin testing its function.


Materials and Methods:: Common medium: We identified an RPMI-based human plasma-like media (HPLM)⁵ as the base medium for the five cell types. However, factors required for the maintenance of each tissue needed optimization. This included macrophage stimulating factor (MCSF; 1-50 ng/mL), thyroid hormone T3 (1-1000 nM), and hepatocyte growth factor (HGF; 1-20 ng/mL). To explore the parametric space in an efficient manner, we implemented a coupling of heuristic differential evolution (DE) algorithm and experiments. DE allowed a systematic method of preparing “mutants” or variation in our media formulation based on an existing “population” or a pool of formulation candidates and allowed the propagation of better media variants based on scoring via an objective function (Figure 1C). In our study, the objective function is a percentage drift of the functionality of each of the individual tissues in a mutant media variant when compared to the gold standard culturing media for that tissue.
Device Fabrication: Multi-layered photolithography was used to create master-molds for the device that were used for replica molding of PDMS that was then bonded to glass slides (Figure 1A).

Tissue Scaling: We pursued a strategy based on a multi-functional scaling whereby we match the profile of hormones and signaling molecules observed in vivo to the multi-tissue MPS system. This required optimization of cell numbers, flow rates, diffusion profiles of biomolecules, and residence time of each signaling molecule in the MPS tissues. Initial guesses of these parameters were used in finite element models, followed by experimental temporal measurements of signaling molecules (e.g. insulin).

Results, Conclusions, and Discussions:: Initial attempts of culturing hiPSC-derived tissues in the commercial medium of the other cell-types showed loss of functionality and sub-optimal viability (e.g., Figure 1B). Thus, we employed DE to define a common medium as described in the Methods section. We observed that the base HPLM medium alone induced inflammation inactivation of the M0-macrophages as measured by a cytokine panel, but improved the insulin responsiveness of adipocytes. Convergence was observed towards a more suitable common medium for the 5 cells (Figure 1C,D), whereby we determined media variants maintaining tissue functionality across the cell-types either comparable or better than HPLM, or media optimized for each cell type. For example, macrophages and hepatocytes maintain phenotypic stability, and the islet β cells have a higher glucose-stimulated-insulin-secretion. Similarly, adipocytes demonstrated better insulin response (via glucose and fatty acid uptake) versus medium alone (Figure 1D). The common medium allowed characterization of the hiPSC-derived cells in the multi-tissue MPS (Figure 1A,E). Form factors of the cell chambers in the MPS were based on published work.^1-4 Diffusion barriers between the cell chamber and media channels (Figure 1A) can be altered based on finite element model prediction to mirror the diffusion of biomolecules in vivo. Basal levels of functional outputs were measured in the MPS such as the albumin secreted by the hepatocytes and C-peptide (a surrogate of insulin measurement) by the β-clusters (Figure 1F). These characterizations allowed an estimation of circulating hormones and signaling molecules in the fluidic circuit as a function of cell numbers, flow rates, and diffusion profiles of these biomolecules in each tissue, and form the basis of multi-functional tissue scaling. For example, we showed the development of insulin resistance in the hepatocytes (as measured by hepatic glucose production upon insulin challenge) with a fat-to-liver ratio of 6:1 in the MPS (Figure 1G). In conclusion, we developed a novel human-relevant multi-tissue MPS platform for studying obesity-induced T2DM, paving the way for the platform to be utilized for formulating better preventive, diagnostic, and treatment strategies for obesity-induced T2DM.


Acknowledgements (Optional): :

References (Optional): :

1. Loskill et al. Lab Chip 2017.


2. Lee-Montiel et al. Front Pharmacol 2021.


3. Qi et al. Small 2021.


4. Goswami et al. Lab Chip 2022.


5. Cantor et al. Cell 2017.