Optimization of Conditions for Endothelial Seeding of Microfluidic Devices with Long Branching Networks and Small Channel Dimensions

Many hematologic diseases may have interactions with proteins and endothelium within small microvascular environments that are currently hard to understand. These effects are hard to produce in vitro, and relevant studies performed in vivo are difficult to image and evaluate at high resolutions. In order to better understand these phenomena, in vitro models with optically clear material should be developed that act as platforms for higher power imaging and analysis. Microfluidic devices made out of polydimethylsiloxane (PDMS) bonded to glass provide this platform and are capable of generating channels of long length (>2.5 cm from inlet to outlet) and of microvascular size (<30 microns). However, current methods of seeding endothelial cells into devices of such dimensions have proven to be very difficult to perform.

One goal of this thesis is to develop a new and reproducible method of seeding endothelial cells into microfluidic devices of many branching networks and small channel dimensions. To examine conditions that lead to optimal seeding, we fabricated a microfluidic device that contains five sets of binary diverging junctions; starting from an inlet of 1000x30 micron dimensions, the flow path splits into a total of 32 small channels of 20x30 micron dimensions. The total distance from inlet to outlet is 3 cm and each channel length is 0.4 cm. We found that traditional flow-in seeding methods are not very effective for long channel networks in which the smallest size is 20x30 microns or less, with poor cell attachment and clogging at junctions being among the biggest hindrances to successful seeding.

Taking all of these factors into account, we developed a reproducible seeding method by dynamically adjusting flow-rates during infusion and using high cell concentrations. This enabled successfully seeding and the formation of confluent layers of endothelial cells lining the inner surfaces of the channels. By utilizing high cell concentrations (3.5-7.5 million cells/mL) of young endothelial cells, we were able to reduce flow rates to allow for attachment and increase flow rates to shear away clogs at junctions and within the smallest channels. After sufficient initial coverage was generated throughout the device, the device was incubated for thirty minutes to allow for secure cell attachment and any excess aggregates within the device were then sheared away at high flow rates. This process was repeated again after a couple days to clear out any remaining aggregates. Development of the device for 3-4 days in an incubator with constant media flow after seeding helped the endothelial cells to grow into a confluent monolayer within the channel walls of the device.

A second goal is to perform initial studies of flow of sickle cells within endothelialized channels. To test the function of the endothelium in the microfluidic device, samples of recessive HBSS sickle cell blood were allowed to flow into the channels. The flow paths taken by the cells were heavily influenced by the shape, size, and height of endothelial protrusions from the channel walls. In some of the wider channels, the effects of endothelial protrusions were so great that multiple small, enclosed flow paths formed within individual channels, resulting in the creation of flow networks within these individual channels. Finally, after extended flow periods, aggregations formed in a scattered fashion across the endothelium of larger channels that served as high-resistance buildups for further flow into the device.