Long-term cell culture in microfluidic devices is an essential prerequisite for "on a chip" biological and physiological based studies. We investigated how medium delivery, from continuous to periodic perfusion, affects long-term cell cultures in a microfluidic platform. Computational simulations suggested that different delivery strategies result in different temporal profiles of accumulation and washing out of endogenous (EnF) and exogenous (ExF) factors, respectively. Thus, cultures exposed to the same overall amount of medium with different temporal profiles were analysed in terms of homogeneity, cell morphology and phenotype. Murine and human cell lines (C2C12 and HFF) and mouse embryonic stem cells (mESC) were cultured in microfluidic channels. An ad hoc experimental setup was developed to perform continuous and periodic medium delivery into the chip, tuning the flow rate, the perfusion time, and the interval of perfusion while using the same amount of medium volume. Periodic medium delivery with a short perfusion pulse ensured cell homogeneity compared to standard cell culture. Conversely, a continuous flow resulted in cell heterogeneity, with abnormal morphology and vesiculation. Only dramatic and unfeasible increasing of perfused medium volume in the continuous configuration could rescue normal cell behaviour. Consistent results were obtained for C2C12 and HFF. In order to extend these results to highly sensitive cells, mESC were cultured for 6 days in the microfluidic channels. Our analysis demonstrates that a periodic medium delivery with fast pulses (with a frequency of 4 times per day) resulted in a homogeneous cell culture in terms of cell viability, colony morphology and maintenance of pluripotency markers. According to experimental observations, the computational model provided a rational description of the perfusion strategies and of how they deeply shape the cell microenvironment in microfluidic cell cultures. These results provide new insight to define optimal strategies for homogeneous and robust long-term cell culture in microfluidic systems, an essential prerequisite for lab on chip cell-based applications.

Our goal was to develop a computational model describing the spatiotemporal evolution of cell heterogeneity within a three-dimensional porous scaffold during cell growth in a perfusion bioreactor. The scaffold was assumed formed by an ensemble of independent parallel cylindrical channels with a defined diameter distribution. The total flow rate partitioning in each channel depends on the effective diameter, which is reduced by the cell growth on the channel wall. The mass balance for one metabolite and the cell volume balance were solved. For each channel diameter, the model simulation provide the spatiotemporal evolution of velocity, shear stress, metabolite concentration, and cell volume growth. In particular, all of these outcomes can be analyzed as a function of channel diameter providing an evaluation of cell property heterogeneity. The model describes that the cell growth can be substantially different in each channel diameter. For instance, in the small diameter channel, cell growth is limited by metabolite mass transport, whereas in the larger diameter channel, shear stress inhibits cell growth. This mathematical model could be an important tool for a priori estimation of the time variation of the cell volume fraction and its degree of heterogeneity as a function of operational parameters and scaffold pore size distribution.

Human tissue in vitro models on-chip are highly desirable to dissect the complexity of a physio-pathological in vivo response because of their advantages compared to traditional static culture systems in terms of high control of microenvironmental conditions, including accurate perturbations and high temporal resolution analyses of medium outflow. Human adipose tissue (hAT) is a key player in metabolic disorders, such as Type 2 Diabetes Mellitus (T2DM). It is involved in the overall energy homeostasis not only as passive energy storage but also as an important metabolic regulator. Here, we aim at developing a large scale microfluidic platform for generating high temporal resolution of glucose uptake profiles, and consequently insulin sensitivity, under physio-pathological stimulations in ex vivo adipose tissues from nondiabetic and T2DM individuals. A multiscale mathematical model that integrates fluid dynamics and an intracellular insulin signaling pathway description was used for assisting microfluidic design in order to maximize measurement accuracy of tissue metabolic activity in response to perturbations. An automated microfluidic injection system was included on-chip for performing precise dynamic biochemical stimulations. The temporal evolution of culture conditions could be monitored for days, before and after perturbation, measuring glucose concentration in the outflow with high temporal resolution. As a proof of concept for detection of insulin resistance, we measured insulin-dependent glucose uptake by hAT from nondiabetic and T2DM subjects, mimicking the postprandial response. The system presented thus represents an important tool in dissecting the role of single tissues, such as hAT, in the complex interwoven picture of metabolic diseases.