Methods

Design Specifications/Constraints for Testing

The automated bioreactor design was intended to meet the following major performance specifications and constraints:

Fluidic Performance - The bioreactor must uniformly deliver and aspirate set amounts of media, around 500 uL per well for 24 well plates and 1 mL for 6 well plates, accurately, precisely, and most importantly with performance comparable to research-grade serological or micropipettes.

In Vitro Performance - Furthermore, the bioreactor should be able to keep the cells healthy (viable) by continuously delivering and aspirating media from the well plate every day for at least 7-14 days and the media must be kept at 10-15 degrees celsius to prevent its spoilage at room temperature.

Methodology

Raspberry Pi driven touchscreen GUI used for testing fluidic performance of the bioreactor setup.

For testing the fluidic performance of the 3D printed 24 and 6 well plate insert manifolds we utilized the Raspberry Pi connected to the 12 V DC peristaltic pump and programmed it to run for 1500 steps with a step delay of 0.001 seconds to evenly distribute colored deionized (DI) water into the well plates. The above-mentioned steps and delay was chosen based on the average delivery of 500 ul per well and 1 ml per well for the 24 and 6 well plates respectively.
We first primed the pumps to ensure that fluid was able to reach the insert and then performed three consecutive runs on separate well plates of the same format. The plates filled with DI water were then aspirated one well at a time using micropipettes and the fluid dispensed by the system was measured in a weighing boat placed atop a standard laboratory balance which had a least count of 0.01 grams.

The weights were converted into volumetric data based on the assumption that water has a density of 1 g/ml and the volumes were recorded into a spreadsheet (Google Sheets and Microsoft Excel) as per the row and column designation of the wells in the plate. The volumes were normalized and converted into percentages based on the theoretical amount of water that should have been dispensed by the system. The percentage distribution data points were inputted into R Studio to generate heatmaps using the heatmaply function. The maximum percentage graphs were then created by calculating the percent change from the maximum and minimum values. The error bars were made by calculating the percent changes from the median to the maximum and minimum.
Finally, the bell-curve distribution, which shows the frequency of occurrence and the volume in milliliters, was made by populating the data points in Excel then finding the mean, standard deviation, and the 99.7% maximum and minimum values in the curve. Iterative values for the curve were then computed by finding the interval count and interval values to achieve the bell-shaped data points. With the data points, the bell-curve distribution was created.

For the in vitro performance evaluation (bio-verification) of the bioreactor, we first 3D printed a fresh 6 well plate insert manifold in clear PETG and sanitized the insert by immersing the print in 70 % ethanol for 10-15 minutes followed by UV sterilization in the biosafety cabinet for 15 minutes.
To ensure overall sterility of the system we also autoclaved the silicon tubing and used 3D printed caps/plugs to prevent the entry of microbes, debris and dust before attaching them to the pumps and the insert in the incubator (all of which were in the laminar flow hood).
In a manner similar to the fluidic testing, we first primed the pumps to ensure that the media was able to reach the insert and then set the insert on top of a sterile 6 well plate populated by retinal pigment epithelium (RPE) cells at low density. We connected the vacuum and delivery tubes to the insert’s OUT and IN channels and set the Pi to deliver and aspirate media every 6 hours.

RPE cells seeded at low density (10x Brightfield Micrograph)

As RPE cells have melanin pigment in them we can easily assess the cell confluency (surface area of well covered by cells) and viability (ratio of live cells to total cells) by visualizing the cells via phase contrast and brightfield inverted microscopy.
The cells were imaged using these two methods at day 1 of testing and the cell confluency and viability was computed as percentages. The system was left to function independently for 7 days as we monitored for any errors or fluid delivery/aspiration issues. After 7 days of automated cell feeding, we took out the well plate, and placed a well plate lid under the insert to prevent media from dripping in the incubator. The cells were reimaged and the cell confluency/viability percentages were calculated to assess the in vitro performance of the bioreactor.

Editor: Devansh Agarwal

Page updated

Report abuse