Team:TU Darmstadt/Tech
Tech
Approaching a large scale production
As we were thinking about possible future dimensions of our modular Virus-like particle (MVP) platform for
diverse applications, we had to consider the challenges that lie beyond proof-of-concept. In
particular, industrial large scale production. Many aspects thereof have already been described in literature.
[1]
[2]
We set out to tackle one of the most important steps, the cultivation of bacteria, by designing an automated
bioreactor. The dual expression vector designed by us,
which enables us to ensure the integrity and modularity of
self-assembling VLPs, needs to be implemented in an automated bioreactor which is setup according to the vector's specific
requirements. We wanted to achieve this with the following process so that VLPs can be reliably manufactured,
regardless of the desired application.
Figure 1 : Scheme of the automatic bioreactor system for modular Virus-like particle production.
The first step of the automated production of tailored VLPs is to define the amount of modification, dependent on the size of the desired cargo. This, we planned to realize with a web application, in which one can easily adjust the modification degree of one's VLPs. Then, the constant measurement of the optical density (OD600) by a self-designed and 3D-printed SensorBrick determines the perfect moment at which the induction of the growing bacterial culture should occur. When this moment is reached, induction is automatically performed by a self-printed syringe pump system. As shown in the Modification-Ratio assay, the amount of modification can be controlled by the inducer concentrations. After cultivation, the purification of the assembled VLPs should be performed quickly. Therefore, harvesting of the culture has to be fast. We considered purification methods which could be implemented in a large scale production process.
Do it yourself
When we started designing the components and parts of our bioreactor (Fig. 1), we focused on standardization, low budget costs, and reproducibility for other iGEM teams. OD600-sensors are basic parts which are implemented in bioreactor controlling systems and were already designed by other iGEM Teams like Aachen 2015, Graz 2017 and Singapore 2018. But, as continuous reading of OD600 still seemed like a challenging problem, we wanted to try a different approach. We decided to use a different syringe-pump system in which the sample solution is stationary while measuring, since we considered this, as well as the intensity of the chosen lamp, as the main problem for a correct determination of the optical density. Most parts that we used can be 3D-printed or belong to standard lab supplies. Therefore, we consider it easy to reproduce our design with the uploaded CAD files. Only the measurement chamber itself is fabricated out of transparent polycarbonate bricks, due to the high dependence on transmittance. Follow this web page below and click on the specific part you are interested in to open and read more.
Keeping the work simple! That‘s the main idea and aspect of our web application. With a web app and a small computer called RaspberryPi we started working on automatizing our bioreactor. With the web app, you have the possibility to control our 3D printed syringe pumps either manually or automatically and use them for inducing the growing cell culture. By simply choosing the amount of modification on the web app interface, the app can evaluate the needed amount of inducers. The induction is triggered when the matching OD600 is reached and the app initiates the adding of the predefined inducer amount.
The code for the web app is realized by an Angular framework and communicates over a REST interface with the backend, that is written in Rust. The advantage: No app installation on the smartphone is necessary. Just start the RasberryPi, connect it to your smartphone, open a browser, type in the address of the RaspberryPi and start the VLP production.
The code of the web app is still adaptable for comparable setups.
The syringe volume can be configured in the app and saved in a
database, so the user can adjust this easily.
For further information about the programming see GitLab.
Animation 1: Working principle of the web application.
In order to automatically induce at the right moment, continuous measurement of the optical density (OD600) of the cell culture is required. We have developed and produced a low-cost OD600-sensor we called SensorBrick. This SensorBrick is connected in between the bioreactor and the syringe pump. The SensorBrick is compatible with any setup of a bioreactor and a syringe pump system and can be used by any other iGEM teams, that are interested in automating processes.
You want to have your own OD600-sensor? Read in the following how we build ours and download the files.
The SensorBrick consists of polystyrole bars, which were cut into cuboids of a length of 3 cm. The sides of the cuboid have a diameter of 1.2 cm. A drill with a 4 mm drill head was used to create lengthwise holes in the center of our cuboids. One side is connected to our reactor, the other side is connected to one of our syringe pumps. To connect the different parts of our setup we are using small silicone tubes and standardized Luer Lock adapters for fluid control. Two additional holes with a diameter of 1.2 mm were drilled into the opposite sides of our cuboid. These holes must be centered to ensure that they meet the lengthwise hole exactly in the center. It is essential that both holes lay on a straight line, so that the light from the first optic plastic fiber cables (connected to the LED), which are fixed to the holes by epoxy adhesive, passes through the larger hole and is collected by the second plastic fiber cable (connected to the sensor). This hole will later be the measuring chamber.
The flow-through can then be tested by placing the sensor on one side of the optical fiber cables and the LED source on the other, and fixing them with heat shrinking tubes. Since the measurement is interfered by any light source, the SensorBrick must be isolated. Thus we covered the whole SensorBrick in light protective sheets. Additionally, we printed a case to stabilize the composite structure. CAD files to print the case can be found above. All parts are designed to fit with M2,5 threaded rods.
By controlling the syringes, we succeeded in transferring medium and cell culture from the reactor into the SensorBrick. Light out of the optic plastic fiber cables is directed through the sample. Afterwards, a second fiber collects the light and guides it to an optical spectral sensor. We use a sensor which measures light at 6 different wavelengths (450, 500, 550, 570, 600 and 650 nm). As we wanted to measure cell density at a specific wavelength of 600 nm, the sensor is perfectly suited for our application, but also for other applications that require the listed wavelengths. Once analyzed, the fluid is collected inside the syringe and discarded as GMO waste to avoid any contamination.
Before operating the bioreactor, the sensor must be calibrated by measuring the OD600 of a cell culture simultaneously with our SensorBrick and in a photometer. We measured the OD600 every 15 minutes and plotted the graphs in Fig. 2 and 3.
Figure 2: OD600 measurement with our SensorBrick.
Figure 3: OD600 measurement with a photometer.
To determine a regression, the values of the SensorBrick are plotted against the OD of the photometer. The Pearson coefficient of 0.98 proofs the correlation of both measured optical densities is suitable. The function of the regression in Fig. 3, can now be used to calculate the exact OD600 from the data measured with our SensorBrick.
Figure 4: Calibration curve for the OD600 measurement with the SensorBrick.
Now that the SensorBrick is calibrated, the value of the SensorBrick can be inserted into the given function, so that an absolute value of the OD can be obtained. To verify our calibration we ran another two experiments and compared the data to the photometer. We grew E. coli (Fig. 5) and then transformed E. coli with the dual expression vector (Fig. 6).
Figure 5: OD600growth-curve of E. coli measured by a photometer and our SensorBrick.
Figure 6: OD600growth-curve of transformed E. coli measured by a photometer and our SensorBrick.
In the following we will compare the calibrated OD600 values of the SensorBrick with the measured data from the photometer. In Fig. 5 we see that the OD600 of the SensorBrick is lower. This might be due to us having used the dual expression vector
pTeTW3con2, expressing superfolder green fluorescent protein (sfGFP), for
calibrating the SensorBrick. As a result, the expressed fluorescence increased the intensity of
light in the range of 600 nm,
so the calculated values of OD600 appear lower than they actually are. The deviation of the
calculated values to the photometer can
be explained by the increasing amount of fluorescence during the time of expression.
Fig. 6 shows that a calibrated SenorBrick determines approximate values above an
OD600 of 0.3. We learned that it is
necessary to calibrate the SensorBrick with the used bacteria strain individually. It can not be
universally calibrated, especially when
fluorescent proteins are present. In order to achieve this, the sensor would need to persists of a
monochromator.
Therefore, the OD600 measurement by the SensorBrick can only be seen as qualitative
values. This still suits
the purpose of determining the exact moment for the web app to initiate the induction.
With further calibration the SensorBrick is a helpful tool and we hope future iGEM teams will use and improve our SensorBrick.
As said before, to automate the process of producing Virus-like particle (VLP) one needs an automation for the adding of inducer to the bioreactor. To achieve this, wee had the idea to use an easy controllable fluidic system, based on pumps and flexible tubes. This enables the flexible tubes to go directly from the pump to the bioreactor, so that the sterility of the reaction is still ensured. To have an easy, reproducible, and cheap construct we based our work on open source syringe pumps from Poseidon. We printed the three CADs that were placed at disposal from Poseidon with a Prusa MK3 printer in the open community Lab3.
For a detailed building plan see the following text!
The following text will explain how to build the syringe pump. You will need the parts, which are listed here. The first step is to fix the motor to the baseplate (no. 1). To do so, take the motor (no. 4) and fix it on the right end of the baseplate (no. 1) with the no. 10 head screws. Now that the motor is fixed take one of the M5 nuts (no. 9) and put it into the slide bar (no. 2). Screw the threaded rod (no. 8) a little through the M5 nut of the slide bar. After that, take both no. 7 and put them onto the sides of the slide bar. Now you can introduce the two no. 6 steel rods through the two holes on the right of no. 1 into the two now included bearings (no. 7) of no. 2. To fix the steel rods (no. 6) introduce the two M3 nuts (no. 12) into the two holes in no. 1, and screw the two no. 11 socket head screws into them. The rail is now stabilized. To link the stepper motor (no. 4) to the threaded rod (no. 8) take the shaft coupling (no. 5) and fix it between the motor and the threaded rod. Lastly, introduce the second no. 9 nut into no. 1 and screw the M5 knob (no. 13) as far as you need it. The syringe pump is now constructed, and you can use it for your desired application!
When thinking about upscaling the production of VLPs, one of the important parts is the purification. It is the most time-consuming segment in the manufacturing process and took us several steps before we could harvest assembled VLPs from the reaction mixture. [3] A consultation with experts resulted in the realization that cleansing of the VLPs is also a very important step towards product safety in terms of future application fields. Therefore, we investigated on how to tackle the purification in an upscaled VLP production, since a VLP based product could never be competitive if the manufacturing process is not scalable. Historically, VLPs have been harvested by ultracentrifugation on a saccharose cushion or by chromatographic processing. [4] We chose to use ultracentrifugation in combination with size-exclusion chromatography as a method for harvesting and subsequently purifying our VLPs. It is the ideal method for research and development, because it is an easy and cheap way to purify VLPs. However, due to the lack of scalability, another way of purification needs to be considered when increasing the VLP production in the future. [5]
Based on discussions with Prof. Bailer and Prof. Santi, we learned that physical separation methods like tangential flow filtration, new membrane technologies, as well as chromatography are the most promising candidates for future purification methods. [6] Separating VLPs depending on the size of the particles with filtration and membrane methods is more cautious and therefore ensures the integrity of assembled VLPs. [3] These technologies are in current development pipelines and will be able to establish a fast and time efficient downstream process for VLP-purification. [4] We are curious to try out some state-of-the-art technologies and test the compatibility with our VLPs.
The Final Setup
In Fig. 7 below you can see the final setup of our automated bioreactor system. On the left is the reactor, which is connected to an air pump. The other tubes sticking into the bioreactor direct to the syringe pumps. In the front of the picture you can see the RasberryPi with the keyboard and mouse to control it. Next to the bioreactor and connected to the RasberryPi is the OD600 sensor. The syringe pumps and the RaseberryPi are both connected to the Arduino and a power cable.
Figure 7: Setup of the bioreactor with RasberryPi, syringe Pumps, Inducer and Arduino.
After calibrating the OD600 sensor, we grew two different cultures in our bioreactor. Both witheld E. coli the dual expression vector pTeTW3con2. One was grown with induction and the other without induction, in order to control the expression of mCherry and sfGFP, as shown in the Modification-Ratio assay. Fig. 8 shows the bioreactor without having been induced, and therefore glowing green, due to the sfGFP. Fig. 9 shows a picture of the induced culture in which the red mCherry was expressed.
Figure 8: Bioreactor with sfGFP-producing E. coli containing the dual expression vector.
Figure 9: Bioreactor with mCherry-producing E. coli containing the induced dual expression vector.
References
- ↑ Yan Xiao, Hong-Ying Chen, et al., Large-scale production of foot-and-mouth disease virus (serotype Asia1) VLP vaccine in Escherichia coli and protection potency evaluation in cattle. BMC Biotechnology, 2016 [1]
- ↑ Vicente, et al., Large-scale production and purification of VLP-based vaccines. Journal of Invertebrate Pathology, 2011, p. 43-47. [2]
- ↑ Kunal Saha, et al., A simple method for obtaining highly viable virus from culture supernatant, Journal of Virological Methods, 1994 [3]
- ↑ Antonio Roldao, et al., Virus-like Paricles in Vaccine development, Expert Rev. Vaccines (9) p. 1149-1176, 2010 [4]
- ↑ R. Morenweiser, Downstream processing of viral vectors and vaccines, Gene Therapy, 2005 [5]
- ↑ David L. Grzenia, et al., Tangential flow filtration for virus purification, Journal of Membrane Science, 2008 [6]
