PARSE aims to address a foundational issue in molecular biology, namely the precise control of the speed of microbial growth. Our aim is to allow front-line biology researchers to optimise their workflow, and give them more flexibility and influence over the design and production within synthetic biological systems. We have extensively modelled the function of these “growth slowers” (GS) to assist aid researchers using them to more easily control and predict bacterial growth rates. To aid with this, we have also developed a modular bioreactor/turbidostat, to allow us to analyse and characterise growth under dynamic conditions. The bioreactor has been designed with input from potential real-world users - feedback from whom has allowed us to create a cheap, easy to construct, and open-source prototype - the final version of which would allow even the smallest non-commercial labs to advance their ability in controlling the growth of bacteria. Along the way, we have consulted a range of top class academics and industry workers whose advice guided and refined our project. Here is what they had to say:
A meeting with Dr. Karunakaran
In order to get the perspective of an active synthetic biology researcher, a meeting was held with Dr. Esther Karunakaran, the synthetic biology lead at the University of Sheffield to explain the project and pose questions.
Dr. Karunakaran said having a growth modulating system for cells would be “incredibly useful”. Currently she is patenting a novel bioreactor concept, where rather than having a big cylindrical vessel, the media flows around planes of stacked and very thin mosquito-coil shaped vessels. It is designed to operate as a continuous bioreactor. One of its biggest limitations, however, is the frequency with which it must be cleaned, which is estimated to be around 3 months due to cell bed formation.
Dr. Karunakaran hypothesises that with such a plasmid system, the time her bioreactor could work continuously before requiring clean-up would increase significantly. This is due to the possibility of reducing the rate of cellular growth, perhaps even with the possibility of boosting protein production at the same time.
Additionally, Dr. Karunakaran asked whether PARSE could be used in other organisms such as yeast.
It was then explained that one of the limitations of our device is that our growth modulating gene's protein must be able to interact with the organisms own metabolic machinery so it can fulfil its function. If it doesn't, the growth rate will not be manipulated. Most of our growth slowing genes are derived from E. coli bacteriophages, which (theoretically) ensures that our growth modulating mechanism will work in E. coli. This is not guaranteed in other bacteria, and unfortunately, our current growth slowing genes would not be able to work in eukaryotes due to the fundamental differences within their replication and transcription machinery that most of these proteins target. Therefore, Dr. Karunakaran advised to try and find a growth modulator that strikes at a fundamental level so that the system becomes context-independent for most (if not all) bacteria rather than only be tied to E. coli as the expression system.
As a result of this advice, a number of growth slower genes (GS) that strike the cellular replication mechanism at different levels have been chosen. The way in which the GS interact dictates what applications can be pursued by using them. For instance, gp79 could not help to slow growth while redirecting metabolic flux towards protein production since it is a transcription inhibitor, which prevents cell replication but also protein production. On the other hand gp240 is a suitable candidate for this as it is a DNA clamp inhibitor, which disrupts DNA replication but doesn't affect transcription.
Moreover, to further test for context-independence, two staphylococcus derived GSs have been cloned into E. coli, just as the rest of the E. coli phage-derived ones. It was hypothesised that the growth inhibiting mechanism of these staphylococcus genes would also work in E. coli given that they interact with structures in the DNA replication mechanism which are present in both bacteria. We tested for this via wet-lab experiments. The results can be consulted here.
One application Dr. Karunakaran devised for PARSE is in co-cultures, specifically to address an issue pertaining to co-culturing bacteria for methane production. Methanogen bacterial strains grow slowly, and this results in the wash out in a continuous system keeping them at low population levels as the other bacteria in the co-culture grow at a faster rate. Using PARSE's characterised growth slowing genes in combination with quorum communication between populations of bacteria, the methanogens could modulate the growth of the other bacteria in the culture, allowing to maintain different population ratios within a single bioreactor vessel. She was shown an initial sketch of the ideas related to the quorum sensing growth modulation system and highlighted that the signal molecule synthase could be produced by other bacteria populations, linking this to co-cultures.
Meeting with Dr. Joy Mukherjee
Dr Joy Mukherjee is a senior engineering technician at the University of Sheffield- met with primarily the hardware team to help organise a meeting between them and a representative of a company who construct bioreactors. He has worked on biofilm formation and was excited by the prospect of being able to modulate growth of bacteria. He saw the potential of the project to be used within this area of research and advised us to speak to Dr Karunakaran.
He suggested this method of bacterial growth control would be better than using engineering control methods, for instance, as temperature regulation can have an impact on enzyme activity within the cell, which in some circumstances may be vital to keeping the yield high. Furthermore, he also mentioned that such a system would allow for self-regulating cultures that would be advantageous at higher production scales where engineering control methods are less effective.
Meeting with Representatives of BPES
Matthew Wright (left) and Stuart Pope (right)
Hardware
During our discussion they explained that current methods of monitoring cell density within bioreactors include measuring viable cell density and total cells (OD). Our current approach for the OD sensor involves using an LED and a phototransistor placed directly across from each other outside the vessel to estimate the bacteria present. For this, the distance between the LED and phototransistor should be kept at a maximum of 2 - 2.5 cm to get reliable readings for the OD. Whilst this has limitations against using viable cell count, it is widely used within industry and still allows monitoring of the health of the bacterial culture through growth rate calculations highlighted in our biological modelling.
The planned fluorometer design involves using an LED and a phototransistor placed at a 90° angle from each other outside the vessel while also using exchangeable excitation and emission filters. At one end, an excitation signal would be produced, on the other, the emission interpreted.
We were informed that Aber Instruments is currently offering an OD measurement system that can take OD measurements from outside of the vessel: OpturaSpy. In order to counteract the issues with trying to get the fluorescence to measure across the entire vessel, they recommended to focus on the solution where the LED and the phototransistor are placed at a 90° angle from each other outside the vessel. This is something we shall be researching to see if we could further improve the spectral range, accuracy and compactness of such a fluorometer design.
They said the benefit of having them externally is that they're less invasive, allowing easier sterilisation, mixing etc without risking damaging probes.
One proposed use for this tool within industry is to enable GFP-tagged protein production to be visually monitored using fluorescence.
Growth Slowing
Through our discussion, they explained that current methods of slowing growth within batch fed bioreactors include altering the temperature, or starvation of nutrients or oxygen. These methods can have limitations, such as starvation reducing the growth rate and yield.
We described our system of slowing growth, potentially using quorum sensing systems or biosensors to control the expression of the GS genes. They liked the idea and suggested that if the system targets DNA replication machinery, one potential benefit of the system could be that it might reduce mutation rate in the bacterial culture. This would enable them to run the batches for longer before needing to stop and sterilise and restart the culture as it would prevent mutations that convey a greater fitness advantage from outcompeting other bacteria within the culture, causing a decrease in yield.
Another way the system could be helpful to them is in reducing clumping. They explained that within bioreactors, having a homogenous solution of culture is advantageous. Clumps can have anaerobic conditions in the centre that can reduce the yield- this is particularly problematic in bacteria with pili or other attachment mechanisms. Homogenous solutions are also important to enable successful removal of culture from the bioreactor within continuous bioreactors, as clumping can cause clogging of the outlets.
When we explained the idea of utilising the growth slowers to focus the metabolic flux of the bacteria towards protein production, they stated that this is an “area that should be explored”. It could increase productivity, and decrease the amount of energy spent producing unuseful biomass. It would also potentially allow them to size down the bioreactors they want whilst also producing the same output if they can use less cells to get the same yield due to the increased productivity of the bacteria.
Marketing
As previously mentioned, there seems to be a company creating products with a similar concept to the potential fluorometer idea. This suggests there is a demand for external monitoring systems instead of the use of probes. They suggested that if we were to market this idea, we would have to focus on our unique selling point of non-invasive fluorescence monitoring.
Meetings with Professor Dickman
To get some insight from a bioengineer, we met with Professor Mark Dickman of the University of Sheffield's department of chemical and biological engineering. He offered some insight to our biomodelling team on how to interpret the data from 96 well plate experiments, suggesting that it would be better to use a mean fluorescence/OD calculation to enable them to calculate the amount of fluorescence each cell produced.
His research area is focused on producing dsRNA as a final product, with multiple enzymes involved along the way. As a result, he didn’t think any system that targeted transcription or translation would help increase his yield. His current thoughts on how he can focus the bacteria he uses metabolic flux onto producing the desired dsRNA is using mRNA integrase enzymes. He explained to us that this digests all mRNA, causing a global transcription knockdown. It doesn’t digest any mRNA that don’t have ACA present, and this is how the protein itself is able to be produced, and through codon optimization, in theory the transcripts you want can be designed to not have any ACA sequences present.
A Meeting with Andrew Fenton
A meeting was scheduled with lecturer from the School of Biosciences, Dr Andrew Fenton. The meeting was insightful in learning more about the application of our project in the field of microbiology, with a certain focus on studying bacterial systems that have previously been tricky to understand. During the meeting, Dr Fenton also proposed the idea of using a 96 well plate to more easily determine the max fluorescence of our GFP. We then decided to incorporate this idea in our experimental design, and got some exciting results from it.
A bacterial system pointed out to us was the Toxin Antitoxin (TA) system found abundant in many bacteria. The system is made up of two components: the gene encoding the toxin protein which inhibits certain cellular processes, and its antitoxin that may be found as either a protein or RNA. The systems have been found on many plasmids, including E. Coli strain K-12 [1]. The system is often tightly regulated to maintain homeostasis in order to carry out its function. The exact role of the TA system varies from toxin to toxin. A possible application of our growth slowing system in this field is for it to be used to study the conditions in which the TA system is activated. There are also applications in studying the effect of decreasing the rate of production of an antitoxin on toxin levels.
Another application suggested to us was a potential in the field of antibiotic development, particularly when fighting the age of resistance, an issue that has been prevalent for decades. Many drugs that were once effective at fighting bacterial disease have been rendered almost useless, and new antibiotics must be developed. Our system has the potential to slow down certain molecular interactions to the point that we can more easily identify new targets that antibiotics could be used for. For example, disruption of a unique sensor histidine kinase in Streptococcus pneumoniae has been identified, and could prove to be an effective target.
Dr Fenton pointed out that genetic similarities in Shigella means that for future applications, we can test out whether our system can also be placed into its plasmid. This removes some limitations on the bacteria we can use our system in.
References
[1] Jurėnas, D., Fraikin, N., Goormaghtigh, F. et al. Biology and evolution of bacterial toxin–antitoxin systems. Nat Rev Microbiol 20, 335–350 (2022). https://doi.org/10.1038/s41579-021-00661-1