There are two types of cells in the production of monoclonal antibodies (mAb), hybridoma and protein expression cells. After isolating the specific mAb, we can transfect its sequence into unique cell lines that can express recombinant proteins in large amounts, such as Chinese hamster ovary (CHO) cells. [1][2]
However, many laboratories must invest time and effort in culturing cells and refreshing culture medium in person. Here, we designed a bioreactor for CHO cell perfusion culturing to replace the staff-consuming fed-batch culture method, hoping to lower the threshold of large-scale antibody manufacturing.
Lab-made bioreactor systems are usually placed in cell incubators messily. [3][4] We suggested using a chassis and outer shell to house all the functional parts of our bioreactor, attaining cell culturing, efficiency promoting, and quality monitoring in one device.
| Part | Function |
|---|---|
| Culture Bottle | Cell container for suspension culture. With a motor and impeller, it can mix cells and nutrients homogenously. |
| Culture Medium Bottle | Contain fresh and waste culture medium. |
| Antibody Isolation Buffer Bottle | Contain buffers with characteristics suitable for antibodies’ purification, which is for countercurrent exchange, such as serum-free buffer. [2] |
| Electro Balance | Monitor the weight change of the culture bottle for further control. |
| Peristaltic Pump | The power source of the whole bioreactor system. |
| Slot | For microfluidic chip and biosensor insertion and renewal. |
| Biomimetic Microfluidic Chip | |
| SAM Biosensor |
The cell retention device is a critical part of the bioreactor system. To avoid cells being pulled away along with waste medium during protein harvesting, we decided to make a microfluidic chip to separate the antibodies, CHO cells, debris, and medium.
The most common microfluidic designs for cell retention are spiral and tangential flow filtration (TFF), both with benefits and drawbacks. [4] Spiral is high-throughput and low cost, but the cell retention rate is slightly lower. On the other hand, TFF has nearly 100% of the cell retention rate, but the microchannel may fouling blockage after a long time. [3][4][5][6][11]
To avoid those problems and boost efficiency, we combined spiral and TFF. A spiral in the front can separate most cells out by Dean vortices inside the microchannel [4][6][11]; the rest of the medium will pass through the downstream TFF, catch the smaller cell debris and isolate the antibodies. It is crucial to find a method for dead cell debris isolation. If the dead cells are retained in the cell culture bottle, they may release proteolytic enzymes, affecting cell productivity and antibody quality. [4][11]
The TFF microfluidic designs nowadays have a common problem: the channel and the filter pore might be blocked by dirt and cell debris after prolonged use. [4][5][20] We tried to optimize TFF by looking deep into nature and borrowing ideas from filter-feeding animals, such as giant oceanic manta rays (Mobula birostris).
Manta rays are the largest type of ray in the world, but they feed on the smallest creatures like copepods or crustacean larvae. The unique wing-like microstructures of their gills will make solid particles “ricochet” between micro wings and won’t get stocked, giving a highly-efficient solid-fluid separation. The manta ray-inspired “micro wing” design in our TFF may avoid fouling and increase the antibody isolation rate. [7][8][9]
Furthermore, the countercurrent exchange (CCE) design inspired by fish gills may help to increase the antibody collecting rate. [10][21] These bio-inspired designs may boost not only the bioreactor’s efficiency but also the public’s attention to endangered creatures like manta rays.
TFF has surprisingly few multiphysics simulation references compared to spiral microfluidics [3][4][6][11], so we focus on establishing models for biomimetic TFF to check whether the solid particles will ricochet on the micro wings.
Ansys Fluent is used here to establish the model. The flow was set as laminar, and the inlet velocity was 0.32 m/s at the beginning. There were two types of geometry, TFF with and without CCE.
| Model & Parameter | Setting | Reference |
|---|---|---|
| Inhomogeneous Model | Eulerian | |
| Granular Viscosity Model | Gidaspow | |
| Granular Bulk Viscosity Model | Lun-et-al | |
| Viscous Model | Laminar | |
| Diameter Distribution Method | Rosin-Rammler | |
| Drag Law | Syamlal-OBrien | |
| Gravity | -9.81 m/s2 (Z-axis) | |
| Inlet Velocity | 0.32 m/s (X-axis) | [3][4][11] |
| Particle Density | 1050 - 1080 kg/m3 | [13][14] |
| Particle Minimum Diameter | 1e-5 m | [3][4][11][12][13] |
| Particle Maximum Diameter | 2e-5 m | [3][4][11][12][13] |
| Particle Mean Diameter | 1.5e-5 m | [3][4][11][12][13] |
Initially, we used TFF with a CCE channel to make the model. The result showed that a rough combination between TFF and CCE would affect the regular operation of the microfluidic, although ricocheting motion still occurs. Next, we made a TFF geometry without CCE for the model, ensuring the biomimetic TFF is functional. The fluid streamline result showed the ricocheting motion of the solid particles.
Here’s our speculation for how to solve the problem in TFF with CCE: Since most CCE applications in microfluidic are designed for gas exchange, the reverse direction flows are immiscible or segmented by thin films. Adding an artificial selectively permeable membrane might block the crossing flow and relieve the interference between two channels of our biomimetic microfluidic design. [10][22][23]
We have opted for using Polydimethylsiloxane (PDMS) as the material of the microfluidic chip due to its high transparency, deformability, inexpensiveness, and being widely considered bio-compatible. [15][16]
We originally planned to make the microfluidic chip with soft lithography technology and contacted Prof. Tung (see Integrated Human Practice), a microfluidic expert from Academia Sinica, to assist us in the fabrication process. However, due to the severe affection of the pandemic, the fabrication schedule was hugely delayed. Reacting to the situation, we focus on 3D CAD model drawing and multiphysics model establishment.
When it comes to cell culture, there’s always a wondering: could we precisely monitor cell growth and protein production with an easy method? Shake flasks and roller bottles are simple for cell culture, but it is challenging to monitor critical parameters, and little control is possible. On the other hand, benchtop bioreactors can do bigger-scale cultivation and more precise monitoring than flasks, but they need extensive retrofitting. [24]
Here, we designed a self-assembled monolayer (SAM) biosensor for antibody quality monitoring, which can change the binding rate of antibodies into electrochemical signals. [18][19] With Prof. Wang’s help (see Human Practice), we learned how to prepare SAM biosensors. In the SAM preparation process, the antigen will be bound to gold nanoparticles on the PET substrate in the end, and the substrate will get the ability to catch specific antibodies. [18][19][25][26]
We reserved space on our bioreactor outer shell for slotting the sensor downstream of the microfluidic chip. After the microfluidic chip isolates the antibodies, we can capture a droplet of the sample and use the biosensor to check the antibody concentration and affinity. [17][18][19][25][26] Due to the pandemic, we didn’t have the opportunity to make our biosensor and test it, but we still visited National Chung Hsing University and practiced the SAM preparation process.
