Team:Washington/Design
Wetlab Design
The goal of the project is to use inducible promoters to change the color output of the violacein pathway.
To turn off VioC and VioD, we originally planned to use a one-plasmid system with inducible promoters in front of them. However, this proved to be problematic to complete, as explained below. We eventually decided to use a CRISPR/dCas9/sgRNA system. The system was designed to have CRISPR/dCas9, the violacein pathway, and sgRNA on separate plasmids. We chose to use this 3-plasmid system for the following reasons:
- The violacein plasmid is very large and difficult to assemble after being disassembled. Usually, the plasmid would be linearized to add inducible promoters before certain genes, and then recircularized. We also considered cloning each gene separately to be inserted into separate plasmids. However, this also proved difficult due to the absence of unique restriction enzyme sites flanking any of the genes in the pathway.Additionally, because each gene used the same promoter and terminator, the products of PCR amplification would form secondary structures which would be impossible to re-linearize. This means that we could not clone out the first three genes in the pathway, ABE, in one go.
- With these challenges in mind, we wanted an easy system that could be used to test other promoters. It would be difficult to linearize and re-circularize the plasmid every time we wanted to change the promoters. It is much easier to work with a small, manageable sgRNA plasmid.
- We wanted to integrate the system into the yeast genome. If we integrated everything into the genome, it would be impossible to change promoters. In our design, the violacein pathway and dCas9 gene are both integrated into the yeast genome, while the sgRNAs are on a plasmid.
When all three plasmids are transformed into yeast, the violacein pathway and CRISPR/dCas9 will be constitutively expressed on their own plasmids. In the third plasmid with the sgRNAs targeting VioC and VioD, the genes are only expressed under certain conditions based on the presence or absence of their respective inducer in the culture. Our planned inducible yeast promoters were CUP1 and ZAP1. For CUP1, as the concentration of copper in the solution increases, the presence of the copper molecule will drive the promoter forward, causing the gene attached to be transcribed. ZAP1 on the other hand works in an opposite direction. At low zinc conditions, ZAP1 promoter is active; at high concentrations, the promoter is repressed. CUP1 drives the transcription of the sgRNA for VioC while ZAP1 drives the transcription of the sgRNA for VioD. When copper sulfate solution is added to the system, the VioC enzyme will be produced. ZAP1’s function is inversely related to the level of zinc in solution, which means that as the concentration of zinc increases, VioD will be produced less.
When sgRNAs for VioC and VioD are produced, they will form a complex with CRISPR/dCas9. The complexes will bind to VioC and VioD genes and prevent transcription of these genes.
Drylab Design
Control Theory and Feedback Implementation
Control theory is a set of ideas for controlling a system based on the output from that system itself. We implemented control theory through a feedback loop which uses a sensor to measure output, compare it to an ideal value, and then change the input to get the output closer to the desired one. In our system, the input are the inducers, and the output is the color of the solution. The Chromastat adds in more of the inducer for the expression of the desired color.
Additionally, our feedback loop uses proportional control, which keeps the system stable; as the color gets closer to our desired color, the amount of inducer added decreases until the ideal value is reached. Once the output is close enough to the ideal, it will stop changing the input to the system unless there is an unexpected change in output.
Turbidostat
In order to achieve our goal of autonomously controlling the outputs of a yeast culture, we decided to use a turbidostat. We took inspiration on this from Washington’s 2012 team that made an App to control a turbidostat.
A turbidostat is a bioreactor that self-maintains a constant optical density using a laser and a photodetector. It will modulate the inflow rate of new media solution according to the outflow rate, which will outflow a bulk solution from the turbidostat. In this way, the solution can be kept in the mid-log phase. This maintains a constant population of yeast, which enables proper characterization of the population’s metabolism and maintains a uniform population.
This is important for our project, because as we change the inducer ratio present in the solution, the yeast will produce different amounts of pigment. However, we only want to know what the most recent color mix looks like, and by constantly flushing out old yeast, our color sensor readings will be a much more accurate representation of metabolite production.
Our turbidostat tank has several inputs and outputs. There is a needle going to the very bottom that feeds in filtered oxygen through an aquarium bubbler. There is a second needle, much shorter, that determines the volume in the tube. This second needle will remove bulk solution as soon as the volume reaches that level. A third needle, much shorter, will add in new media according to the current turbidity of the solution. Finally, there can be more needles added for any inducers required in the system.
The solution is mixed constantly with a spin vane. The spin vane will be spun using a magnetic spinner.
Chromastat Design
In order to build our chromastat, we followed two design objectives:
- Low-cost
- Versatile
We started with a sketch of a simple turbidostat implementation with fluid controls, based on a modified version of the turbidostat in the Klavins lab at the University of Washington. We built this simple system to test the basic components and possible points of difficulty with a full system.
We also took inspiration from Dr. Dunham in the Department of Genome Sciences, working on directed evolution in yeast with chemostats.
In our earliest design iteration, we were most worried about the possibility of using a RGB sensor to detect color changes. We used a PiCamera to test out this idea.
We realized quickly that the color sensor would be a highly sensitive piece of hardware, and that much more regulation of light and position would be required. We also found out that the current design for a syringe pump is too inaccurate, and thus the syringe pump went through numerous iterations, as described below.
For our second iteration of our device, we decided to follow the strict parts list on the turbidostat construction list for the Klavins lab, and identified parts that we could reduce costs. We also visited the Klavins lab to discuss difficulties and limitations they face with their current turbidostat.
Syringe Pump
The syringe pump is made out of 3D printed pieces of PLA plastic that are attached to a threaded rod, linear slides, and a stepper motor similar to those found on 3D printers. It is controlled by a program we created using the Raspberry Pi. It was a vital component for the Chromastat, because we needed these to input inducers into the tube. Before any long term goals were set for the syringe pump, testing needed to be done to see how well it currently performed.
From testing, we found that the pump had a constant overshoot of 0.1 mL. It was determined that the origin of the source of error was the screw we used to move the platform that depressed the syringe. We decided to switch over from a regular ¼” threaded rod to a 8mm lead screw, thereby reducing the amount of endpoint variability.
Next, we found that there were lots of issues pertaining to the linear guide rails. We remodeled the entire system in order to fit linear slide bearings on. This made the entire movement much smoother.
We decided to make the syringe pump modular, so we could introduce many different size syringes to the system. Thinner and longer syringes can be extremely accurate compared to the 3 mL syringe pumps we currently use.
Low-cost commercial syringe pumps lack programmability, accuracy, and the ability to refill. We wanted our pump to have all of these features. Programmability is achieved by using the Pi to control the pump. Because the control software is written in Java, the overall behavior of the pump can be easily customized to any situation by anyone with intermediate Java experience.
Accuracy was achieved by the above techniques of employing linear slide bearings and an 8mm lead screw to control the linear actuation of the syringe.
For automatic refilling, we employed a set of double check valves, graciously donated by Nordson Medical. These valves allows us to expel fluid when depressing the syringe and to refill the syringe by drawing it back.
