Team:UiOslo Norway/Poster

A detection system and approach for treating amoebic gill disease
Presented by Team UiOslo 2020

Alexander Refsnes-Petkovic¹, Jonas Aouay Grønbakken¹, Martin Eide Lien¹,

¹iGEM Student Team Member

Abstract

The biological and environmental concerns that the aquaculture industry faces require innovation in both detection and treatment of disease in fish. Among these is amoebic gill disease caused by Paramoeba perurans (P. perurans) affecting Atlantic salmon. P. perurans is a parasite that latches onto the gills of salmon and causes discomfort and possibly death by reducing the respirational surface area on the gills. Our project aims to create an automatic detection system. To create this system we are investigating collective behaviors and how one can use them to diagnose diseases. We also aim to use genetically modified Escherichia coli to produce salinomycin, an antiparasitic compound. The gene cluster for salinomycin production will be transferred into E. coli step by step, with the goal to provide an alternative route to production of this compound. We hope that our system will complement current diagnostic tools and treatments used in aquaculture today.
Aquaculture
Aquaculture is an essential industry in Norway and many other countries, growing rapidly worldwide, fish farming is becoming an important source of protein for the increasing human population.



References:
[1] Fish farming market outlook - 2025. https://www.alliedmarketresearch.com/fish-farming-market. Oct 2020. [Online; accessed 16. Oct 2020]
[2] Statistics Norway. Aquaculture. https://www.ssb.no/en/jord-skog-jakt-og-skeri/statistikker/skeoppdrett/aar-forelopige, Oct 2020. [Online; accessed 16. Oct. 2020].
[3] Statistics Norway External trade in goods https://www.ssb.no/en/utenriksokonomi/statistikker/muh/aar Oct 2020. [Online; accessed 16. Oct. 2020].
Loss in Aquaculture
Aquaculture is subject to 15-20% loss in terms of number of fish per year. Most of this loss is death, among the causes are parasitic diseases and their treatments.

Among the parasitic diseases are amoebic gill disease, AGD, which is our focus throughout the project. This parasite lives naturally in the sea and affects sites seasonally depending on salinity, temperature, water currents etc. Some Norwegian facilities have reported 80% mortality when AGD has been left untreated. AGD predominantly affects Atlantic Salmon, the most important fish species in Norwegian Aquaculture.


Aquaculture facilities are not required to report AGD, even then a large number of cases have been documented. Varying from year to year.


References:
[1] Noma. Velferdsindikatorer for oppdrettslaks: Hvordan vurdere og dokumentere fiskevelferd, 2018.
[2] Hjeltnes B, Bang-Jensen B, Bornø G, Haukaas A, Walde C S (Eds). The health situation in norwegian aquaculture 2018. Norwegian Veterinary institute , 8, 2019.
[3] Fish mortality and losses in production Sustainability in aquaculture. https://www.barrentswatch.no/en/havbruk/fish-mortality-and-losses-in-production, Oct 2020 [Online; accessed 16. Oct 2020]
Our approach
We have investigated an alternative to current treatment and diagnostics methods. Setting up a detection system that is capable of diagnosing a fish school as healthy or sick depending on a collective metric. We have laid plans for how to make E.coli produce the precursor to Narasin, an anti-parasitic compound.

Model
Computer modeling is an important step when designing our detection system. It allows us to evaluate and test different architectures much quicker than we can with real life experiments, one can then run experiments with the most suitable implementations later.


We have used a model of fish schooling in our project.

This model has 3 mechanisms, attraction, orientation and repulsion. It allows us to set a number of parameters, such as the radius of repulsion, reducing it might be a good approximation of less reactive fish.

References:
[1] Alethea Barbaro , Bjorn Birnir, Kirk Taylor, (2006) “Simulating The Collective Behavior of Schooling FishvWith A Discrete Stochastic Model” funded byThe National Science Foundation,The Research Fund of The University of Iceland, https://www.mrl.ucsb.edu/sites/default/files/mrl_docs/ret_attachments/research/KTaylor.pdf
Detection system

Here is what refer to as a detection system.


The features in our system is given by.


Containing relatively little information, it gives no information of school structure, direction or number of fish.
Our classifier consists of a neural net which is fed a time series of the metric given above.

Lab plans
The goal of our synthetic biology plan was to engineer E. coli to produce Salinomycin, an anti parasitic compound. This would complement our early detection system with a treatment option that is less harmful for fish in early stages of AGD. Our first step would be to clone the LuxR- family transcriptional regulator, SlnR into E. coli cells.



We would then aim to verify successful transformation via colony PCR and subsequently sequencing of transformants. To verify functional insert we would perform expression and purification of protein of interest and electrophoretic mobility shift assay.

References:
[1] Zhu, Z., Li, H., Yu, P. et al. SlnR is a positive pathway-specific regulator for salinomycin biosynthesis in Streptomyces albus . Appl Microbiol Biotechnol 101, 1547–1557 (2017). https://link.springer.com/article/10.1007/s00253-016-7918-5
[2] Liu, S., Yu, P., Yuan, P. et al. Sigma factor WhiGch positively regulates natamycin production in Streptomyces chattanoogensis L10. Appl Microbiol Biotechnol 99, 2715–2726 (2015).
Software
We have created extendable software for modeling, visualization and classification of fish behavior. Everything we have made is freely available online and we hope somebody finds it useful.
Results
With our setup we managed to achieve a reasonable accuracy, given the little amount of information required it seems reasonable that this can be done with cameras at a fish farm. In addition we found that it is easier to classify the fish when one observes the transition between behaviours, such as not schooling to schooling.
Here are three examples of correctly classified simulations, visualized with our blender script. There is no obvious sign that we can see here with our own eyes.
Sick Healthy Healthy
Future Work
There are many things to do further with our project. The model of fish behaviour can be extended significantly by adding more behaviours, one could create the environment model to evaluate different physical implementations of the detection system and extend the classification scheme.
Acknowledgements

Supervisors:

Dirk Linke, Primary supervisor and course leader
Athanasios Saragliadis supervisor. Thanks to his insight and experience, writing lab protocols without being in a lab was possible.
Kirsten B. Haraldsen supervisor. Creative and thourough support
Finn-Eirik Johansen supervisor.

Project support:

Dan Michael Heggø, Helped explaining the MediaWiki platform
Hans Jonas Fossum Moen, Expert advice that was fundamental to the direction of the modeling project
Kai Olav Ellefsen, Expert advice for evaluation of classifier and modeling
Markus Herberg Hovd, Provided a wiki template from the UiOslo_Norway 2018 team which helped us save time
Simen Kjellin, Helped us with creative ideas, filming and editing the promotion video.


Sponsors
Our involvment with iGEM was sponsored fully by the University of Oslo, where iGEM is a two semester course.