Team:Exeter/Model
Modelling
Introduction
Modelling is used to describe and explain phenomena that cannot be experienced directly, by making approximations and running simulations to predict what might occur in the real world. Models are crucial to scientific research and the communication of this research, as well as its progress. Our team utilised both biological and physical modelling to inform our research at key stages; allowing us to quantify, visualise and instruct the next steps of our project.
Hydrocyclone Model
A key part of our iterative filter design process was the inclusion of predictive modelling to assess whether key parts of our design would work for a household washing machine. Hydrocyclones are commonly used in large scale industry to separate larger particles from smaller ones within large quantities of fluid, and reduce the flow rate of liquids entering a filter system. To discover whether this would work on a smaller scale, our team physically modelled the required geometry of a hydrocyclone that would separate the microplastics from the main body of water leaving the washing machine.
From this predictive modelling we discovered that for a hydrocyclone to be able to separate the smallest microfibres from water the inner diameter would need to be 7.6mm. This value for an inner diameter makes the hydrocyclone so small that it wouldn’t be able to withstand the flow rate coming out of a washing machine. Our team speculated that this was due to the percentage weight of microplastics being miniscule, at less than 0.1% of the total weight of the fluid.
We also considered whether the upscaling of the hydrocyclone would be feasible, as the volumetric flow rates are exponentially larger within industry than in a household washing machine. However, as most of the large scale microplastic release in industry is also due to washing, the ratio of microplastics by percentage weight to the number of litres released will be the same as a washing machine. Therefore, we would encounter the same issue as in the small scale modelling.
If you are on tablet please click here to open the pdf.
As a result of this modelling process, we realised that it would not be viable to use a hydrocyclone in our filter. We therefore started to research and design different possible filter systems.
Ancestral Reconstruction
An important part of our project is engineering more stable enzymes that will last longer in our filter. In order to achieve this, we decided to make use of the method of ancestral reconstruction. This technique is used to recover ancestral traits that are useful but have been lost during the process of evolution; and relies on a large number of sequences, phylogenetic analysis, and modelling.
We started by searching current papers for phylogenetic trees of PETase that have already been built. We identified a paper that had already organised plastic-degrading enzymes into categories, and had identified a set of enzymes that were closely related to PETase1. We used BLAST software2 to search for sequences homologous to PETase and the set of enzymes previously identified in the paper; and through this process identified 243 homologous sequences. These sequences were aligned and narrowed down to 76 sequences with the help of Professor Harmer from the University of Exeter Living Systems Institute. The final alignment of the sequences was then fed into the ANCESCON software that performed the ancestral reconstruction3. The software produced a phylogenetic tree and identified 74 ancestors. In order to identify the most suitable ancestors to model and use, we used the method below to weight each one:
$$N_{w.bal} = \frac ab \times(a + b)$$
$a$ = the number of leaves in the smallest subtree immediately daughtered by the node of interest
$b$=the number of leaves in the largest subtree immediately daughtered by the node of interest.
Following this method, we identified 4 potentially suitable ancestors to further analyse, Figure 2. The YASARA software was used to model the 3D structure of each of the four ancestors. The models produced were then aligned against the structure of PETase by Professor Harmer in order to identify significant changes in the sequences. We discovered that the catalytic triad was conserved in all four ancestors, suggesting that the PET degrading activity had not been lost. Interestingly, we have also discovered that a beneficial mutations reported in past papers was already present in all four of the ancestors, namely R280A. The only significant trait lost during the reconstruction was the second disulfide bond that was present in PETase but not the ancestors. However, we reverted this by changing the two alanine residues in the ancestors with two cysteine residues. Additionally, we deleted the first five amino acids from the N-terminus that we suspected were composing the signal peptide. Once these minor adjustments had been completed, we sent the final sequences to be synthesised.
Biological Model
Manchester University iGEM Team have a reputation of excellent modelling skills. Therefore, at the UK iGEM meet up we collaborated with the Manchester team to help us model the amount of PETase or MHETase enzyme we would need to purify to achieve degradation of plastic. This was dependent on a couple of variables, including assumed enzyme activity/digestion rate, time left between washes for degradation to occur and the microplastic concentration in g/L. Other parameters used in calculation were sourced from a synthetic biology research paper entitled 'Enhanced Poly(ethylene terephthalate) Hydrolase Activity by Protein Engineering' published in December 2018.
As a result of our collaboration with the Manchester team, they gave us their initial calculations so that we could start the work in the biological wet-lab with ball-park figures in mind. Below is the spreadsheet sent to us by the Manchester Team detailing their 'back of the envelope calculations' to help us understand the factors we would need to consider whilst aiming for a final result of plastic degradation.
The Manchester Team also sent us some of the assumptions they had made whilst making these calculations:
"This models the amount of enzyme needed to digest over the course of a defined time period. This is not exclusive to PETase and could be adjusted to work with anything you have the kinetics for.The model assumes you're constantly digesting over the course of the time period and so calculates how much you need if you are going to continously digest such that everything is gone by the end of the time period. The model also assumes the microplastics are 100% PET digestable all with a constant rate. Furthermore the model assumes that the activity of the enzyme does not drop over the course of the time period."
This collaborative modelling was very valuable and informative for us at the very start of our project when we were considering the enzyme solution and how much we would need to put into our filtration system. It further informed the scale of the filter, feeding into the iterative process of our filter design. It was also useful for our team as a lot of our team members had no experience with the biological side of the project and it gave a larger sense of context to our overall aim.
