Introduction
Last year, we built 2 microgravity simulators and developed one space bioreactor concept. Here we will discuss one of our experimental devices in-depth. The 3D clinostat is part of our engineering group's undergraduate final year project. It was purposefully built for iGEM Concordia to run microgravity synthetic biology experiments.
Recently, we developed our experiment protocol. Our device is verifying our experiment methodologies and testing our hypothesis. We conducted a series of pilot experiments which demonstrated the effectiveness of our devices. For more information about the result of the pilot experiment, please see our genetics team page.
Figure 1: Clinostat
We learned a lot from our last year's project. We started the design phase by following this engineering flowchart:
Figure 2: Engineering Flowchart
Design
We have implemented an iterative design approach during our design phase. The design process was lengthy in order to obtain the most optimized version of the 3D clinostat. Several aspects, such as manufacturing, budget, and feasibility of the design, were our main concerns. This design validation diagram summarises our process visually.
First Design Iteration
Figure 3: First Iteration
Cons & Revisions:
- Bioreactor size & geometry could cause a large moment of inertia.
- Aluminum extrusions of the outer frame would cause poor wire management.
- Large aluminum base frames would be expensive, heavy and hard to manufacture/source.
Second Design Iteration
Figure 4: Second Iteration
Cons & Revisions:
- Large aluminum base frames would be expensive, heavy and hard to manufacture/source.
- Spherical bioreactor is not viable from a manufacturing standpoint as the two spherical halves of the bioreactor would be welded yielding a high stress concentration possibly leading to failure.
Final Design
Figure 5: Final Design
Revisions & Improvements:
- Aluminum extrusions for the base frame to support the motors & payload.
- Hollow sheet metal outer frame for easier wire management
- Rectangular shaped bioreactor made from acrylic
Build
With our final design in hand and feasibility verified, we moved on to the next stage. The finalized design components (including the electronics) are listed as:
Figure 6: Design Components
During the building phase, we have revisioned our design in 2 parts to increase device stability and reduce cost. We implemented the laser cutting process, which is an easy-to-use process as it substitutes the more conventional cutting processes due to its economic and technical benefits. We have also revisioned our frame assembly manufacturing process. This minor revision reduces manufacturing difficulties. Our outer frame assembly was welded together and assembled to the whole construct.
Figure 7: Left: Revision Drawing, Right: Outer Frame Assembly.
The total engineering cost for our 3D clinostat was 800$ CAD, excluding operational costs.
Test
In order to validate our design, we had to test all aspects extensively, as shown in the testing diagram below.
Figure 8: Testing
However, despite some minor tweaks to facilitate the assembly and quick modifications to the electronics, namely the stepper motors and drivers, to enhance the performance of the device, the 3D clinostat does achieve its function!
Figure 9: System Function Testing
Finally, we collected data from the accelerometer placed in the center of rotation of the bioreactor to evaluate the different components of acceleration in the x, y, and z directions. Here are the resulting plots after 5 minutes of rotation.
Figure 10: Acceleration Data Collected For All 3 Axes After 5mns
From the data collected from the ADXL 345 accelerometer after a sampling time of 360s, we can conclude that the rotation of the 3D Clinostat for an extended testing time of around 24h in a laboratory environment would most definitely showcase the damping or convergence of the sum of the acceleration in the x, y & z to 10^(-3) g, so as to achieve the simulation of microgravity due to its slow rotation as well as the constant dispersion of the gravity vector, as discussed given the relatively low values of acceleration in the y and z axes.
Learn
We are improving device stability after we moved the clinostat to our iGEM lab. Several issues were exposed, such as vibrations, installation error, false wiring connections, and overheating. Those issues are relatively simple and were solved by fine-tuning our device.
Figure 11: Issue With Unsupported Frame Connectors
However, the effect of mechanical vibration will be amplified on a microscopic scale which is much harder to solve.
Improve
We attached several support pieces which were 3D printed. Those pieces drastically improved our device stability.
Figure 12: Improvements
Friction is an inevitable factor when it comes to machines. Due to excess friction and vibration of the outer frame, we upgraded our stepper motor. The new motor provides more torque to sustain the unsupervised run.
Figure 13: Motor
We have also learned due to the nature of the stepping motor, vibration from the motor was conducted to our inner bioreactor box. When vibration from the stepping motor reaches the resonance frequency of the inner box, it will create an oscillating motion. Thus, we have used microstepping at 1/8 steps for the inner motor. We measured relative vibration magnitude under this setting.
Figure 14: Clinostat Inner Frame Vibrations
From our measurements, we detected several vibration regimes. Unluckily, previously determined speed introduced some mechanical vibration to the system. To reduce vibration while keep the device around optimal running speed, we determined the clinostat inner frame should run at 2.8rpm or 8.7rpm. Previously, we determined this 1.8:4 ratio is best for our device. We discussed this ratio in our modeling section. Thus, the outer frame should run at 6.2rpm or 19.3rpm respectively. The improved performance offered us some promising screening results that are mentioned in our genetics page.
Engineering Principles In The Lab
As a virtue of this being the second year of the project, we had already done a fair bit of designing in the first year of the project. This consisted of the design of the fluorescent reporter system, which is described in the Genetics page of our wiki. While we were fortunate enough that these reporters worked as expected the first time around, made evident by the results of the stress experiments, namely those of HSP30, that does not mean that we were done using the principles of engineering. In order to test our fluorescent reporters, we had to establish a protocol for our stress experiment assays. These assays went through various iterations throughout the span of roughly one month until we had an assay that worked properly and had minimal variation.
The engineering principles are as follows: design, build, test, learn, and improve. We closely followed these principles in the design of our stress experiment assay. In designing, we looked at various media to use, what concentrations of the stressors to use, how long incubation periods should be, and what instruments we should use for the measurements. When it came to building and testing, these two principles came together in the form of performing the newly designed assay in which the building would be putting the components together and testing would be running the assay itself. In learning, we looked at the results of the assay and tried to see what could have gone wrong in each iteration, and after reflection, we looked at the data we collected along with literature, to see what could be done to improve it. Finally, improvement was done by applying the changes that we thought were necessary before repeating the cycle and running the newest iteration of the assay.
In the first iteration of the stress experiments, we had incubated our strains with the fluorescent reporter added in 1.5 mL Eppendorf tubes overnight with shaking at 30oC for 24 hours. The goal of this was to see if we could measure the baseline fluorescence of the different reporters before adding any stressors. Afterward, we read the fluorescence and optical densities of the cultures, giving us a fluorescence/cell population value used to normalize against OD600 since some stressors may have affected growth differently. We had found that this method produced a high degree of variability between replicates and needed to be worked on.
After some feedback from one of our mentors, we were informed that using Eppendorf tubes limited the aeration of the cultures and that in itself could be a cause for the high degree of variation. With this information, we decided to move to our second iteration of the assay, doing the same work we had done before but instead of culturing the strains in Eppendorf tubes overnight, we cultured the strains overnight in a 96 well plate with a gas-permeable membrane. We still found a fair degree of variation so this needed to be worked on. However, we also needed to begin to test different stressors. We used this method to see if we could see any distinguishable change in genetic regulation. To do this, we added various stressors at different concentrations, namely hydrogen peroxide, ethanol, and sodium chloride. However, as was noticed with the strains without the stressors, there was a high degree of variation in our readings.
After some research into what could be causing the variation, we moved on to our third iteration of the assay. We had found that cell death from being in the stressors for too long could be a reason as to why we had nearly unusable data. To avoid this, we began to incubate our strains for only one hour as our research had suggested that that was as long as was needed to induce gene expression in the promoters used for our reporters. All other parts of the assay were kept identical as before. While we did see some clearer upregulation this way, we still saw too much variation in our samples, suggesting that there was still a major problem. After going back to an advisor for help, he suggested that we replace the YPD media being used as the media fluoresced alongside the cells, leading to interference in the data. At first, we tried using synthetic complete media, which does not fluoresce, however, we struggled to have the cells grow properly overnight in the media and needed an alternative.
In the fourth iteration of the assay, after receiving feedback from an advisor, and doing some research, we decided to replace the media with distilled water after incubation. For this, we incubated the cells in a 96 well plate with the stressors, as was done before, however, before reading the fluorescence and OD600 values, we transferred the cells incubated with the stressors to a 96 well conical plate and spun it down, removed the residual media, and replaced it with distilled water. From here, we put the cells that were resuspended in water into another 96 well plate to have their fluorescence/OD values read. While the upregulation was much clearer, there was still a considerable amount of variation between replicates, although it was much less than before.
In our fifth iteration of this assay's design, we did some more research and used our recently obtained data to determine which concentrations to use. The point of this was to be able to increase the number of replicates that we could run to see if that would decrease the overall variation by allowing us to detect outliers with a higher rejecting power. This led us to the conditions described in the "Strain Selection" part of our Genetics page on our wiki. After running the assay again, we had found that indeed the variation had decreased slightly, but not enough for us to reach any concrete conclusions about what strain to investigate further. We had investigated and altered almost all parts of this assay's design at this point: the stress conditions, the growth conditions, and the media present while measuring. The only aspects left to investigate would be the reporter design and the instrument.
In our sixth iteration of the assay, we had decided to, with the help of an advisor, to change the instrumentation being used for the measurements, as we were running out of time and would likely not be able to redesign the reporter with the time remaining. For the entirety of the assay, we had been using a spectrophotometer that can measure both fluorescence and OD600. An advisor had mentioned that maybe that was not what we should be using due to issues such as dead cells interfering with the OD600 values which would alter the resulting measurements. To avoid this, he had suggested using a flow cytometer instead. We had prepared the assay in the same way that we had done in our fifth iteration and had found much less variation in the readings. This was great as it meant that we were very close to developing the assay that we had hoped to use.
While the data obtained from our seventh iteration of the assay was definitely usable, there was still some variation that we wanted to minimize to get our assay to as close to perfect as possible. Our mentor had mentioned that since we were now using a flow cytometer, we did not have to replace the YPD media with water before running it. In theory, this would minimize potential pipetting errors from this step of the assay. To test this theory, we ran our eighth iteration of the assay. In this iteration, we ran everything the same way as we did in our seventh iteration, however, the only difference was that this time we did not replace the residual YPD media with water, we just removed the cells from the incubators and ran them through the flow cytometer. From this, we received data that had the lowest amount of variation thus far and we were able to properly use statistical tests now to see which strains had significant upregulation under various stress conditions. The results of the assays done with this iteration of the assay can be found under the "Strain Selection" portion of the Genetics page on the wiki.