Ieva Lingytė, Eglė Vitkūnaitė, Auksė Kazlauskaitė, Austėja Sungailaitė, Edvinas Jurgelaitis, Emilija Radlinskaitė, Emilis Gaidauskas, Kamilė Liucija Vainiūtė, Paulius Sasnauskas, Liepa Šiupšinskaitė, Barbora Vasiliauskaitė, Monika Gineitytė, Denis Baronas, Povilas Šėporaitis, Rolandas Meškys, Paulius Toliušis


FlavoFlow: a comprehensive solution for exogenous fish infections detection, treatment and prevention.

Growing fish consumption rates encouraged marine culture farms to implement recirculating aquaculture systems that make intensive fish production compatible with environmental sustainability. Even if these systems reduce the use of terrestrial resources, water recirculation in such systems can cause significant losses because of bacterial or viral infections. A common pathogen of fish infections is the Flavobacterium genus bacteria, which can cause fish death in a few days after the initial infection. To detect the infection as soon as possible, we developed a rapid detection test based on helicase-dependent amplification and lateral-flow assay methods. Additionally, we created a novel treatment method which relies on a quorum sensing mechanism and exolysin protein with the aim of decreasing antibiotic consumption levels. Finally, to prevent forthcoming infections, our third goal is to provide a prevention system based on subunit vaccines encapsulated in alginate beads.


Growing fish consumption rates encouraged aquaculture farmers to implement recirculating aquaculture systems in their farms. Even if these systems reduce the use of terrestrial resources, water recirculation in such systems can cause significant losses due to bacterial and viral infections. A common pathogen of fish infections is the Flavobacterium genus bacteria, which can cause fish death in a few days after the initial infection.


To solve massive Flavobacterium problem, our team have set three goals:

  1. To develop a point-of-care detection test for F. columnare and F. psychrophilum, that is based on isothermal helicase-dependent amplification and lateral flow assay tools.
  2. Establish two different treatment strategies, which would be based on toxin-antitoxin or exolysin-endolysin systems
  3. To pack immunogenic proteins VHSV and GldJ into alginate beads




Since rapid detection test is based on nucleic acid hybridization, unique to species marker genes fragments were chosen. Chosen fragments were amplified using isothermal helicase dependent amplification method, where one primer is used in an excess concentration. Uneven primers ratio allows to obtain ssDNA molecules, which can be further used in lateral flow assay.


For point-of-care diagnostic platform development we have chosen lateral flow assay. This test consists of:

  • Sample pad – meant for HDA samples application
  • Conjugate pad – which holds functionalized gold nanoparticles and allows hybridization between those nanoparticles and ssDNA from HDA sample
  • Nitrocellulose membrane which has a test line where capture probes are immobilized via streptavidin-biotin interaction. As well as control line, where control probes are immobilized via the same bond

If the sample contains specific Flavobacterium gene fragment, then control line as well as the test line will show up. For the negative test only the control line will show up


With the aim to increase the efficiency of HDA, we have fused TteUvrD helicase and Bst polymerase I large fragment through the coiled coil amino acid structure to get a bifunctional protein called helimerase (Fig. 6).

WZB1WZA2TteUvrDBstPolStrepII Tag10xHis TagMBP TagL1L1

Figure 6. Helimerase structure. Blue color figure shows TteUvrD helicase which is fused with one part of coiled-coil structure WinZip-A2 (WZA2) through the linker L1. Yellow color indicates Bst polymerase I large fragment which is fused with one part of coiled-coil structure WinZip-B1 (WZB1) through the linker L1.



With the aim to fulfill the second FlavoFlow goal we have decided to develop two different AI-2 inducible treatment systems, based on a quorum sensing mechanism.

Exolysin-endolysin treatment system

The genetic circuit is based on bacterial interspecies communication called quorum sensing. Autoinducer - 2 is a signaling molecule of quorum sensing Type II system. After AI-2 uptake, the AI-2 inducible promoter is derepressed and the protein-coding transcription is initiated. The first protein is exolysin - Flavobacterium bacteriophage’s depolymerase that lysis pathogenic bacteria biofilm. The other one – endolysin - that targets cell membrane peptidoglycans and results in a quick cell degradation.

MazE/MazF treatment system

The second circuit used for the treatment section includes not only exolysin but mazef complex too. Same as before the protein coding genes transcription is initiated by uptaking AI-2. When MazE concentration is equal to MazF, it inhibits toxin by forming a heterocomplex. However, MazF concentration bypasses MazE concentration, MazF disrupts necessary protein synthesis and invokes cell lysis. In the end, and releases a high concentration of exolysin is released to the bacterial infection site.



However, great detection and treatment strategies only fight with Flavobacterium consequences. Due to this our third goal was to synthesize and purify immunogenic proteins which could be further used for the subunit vaccine based on alginate beads.

The protein coding sequence with 6xHis tag was inserted into an appropriate plasmid and transformed into bacterial cells. The protein was purified after induction and encapsulated into alginate by mixing the alginate with the protein and dropping the solution into a calcium chloride bath. The resulting beads were tested with digestive enzymes at different pH values, alginate lyase and pressure.




After a lot of optimization reactions symmetric and asymmetric HDA as well as PCR were performed successfully (Fig. 7). After asymmetric amplifications obtained average ssDNA concentration was 2.53-2.96 uM for PCR and 0.7 uM for HDA.

Figure 7. Left to right: A. L - gene ruler 50 bp ladder (SM0371), 1 - F. columnare symmetric PCR with F_Col and R_Col primers, 2 - F. psychrophilum symmetric PCR with F_Col and R_Col primers, 3 - E. coli symmetric PCR with F_Col and R_Col primers, 4 - F. columnare asymmetric PCR with F_Col and R_Col primers (1:15), B. L - gene ruler 50 bp ladder (SM0371), 5 - F. columnare symmetric HDA, 6 - F. psychrophilum symmetric HDA, C. 7- F. psychrophilum symmetric PCR with F_Psy and R_Psy primers , 8 - F. columnare symmetric PCR with F_Psy and R_Psy primers, 9 - E. coli symmetric PCR with F_Psy and R_Psy primers, 10 - F. psychrophilum asymmetric PCR with F_Psy and R_Psy primers (1:15), L - gene ruler 50 bp ladder (SM0371). After amplification with F_Col and R_Col primers fragment size should be 122 bp and for F_Psy, R_Psy - 104 bp.


The most specific lateral flow assay test for F. psychrophilum was based on rpoC marker gene fragment. This test was able to differentiate F. psychrophilum from F. columnare as well as from E. coli and F. piscis (Fig. 1). Meanwhile the LFA test created for F. columnare using 16S rRNA gene was not as specific as we hoped due to its inability to distinct F. columnare from E. coli (Fig. 2).


The activity of both enzymes were successfully identified during kinetic assays (Fig. 8). However, further optimization needs to be done with the aim to fuse these proteins in vitro and in vivo.

Figure 8.Time dependent analysis of helicase and BstPol activity assay. A - raw data of helicase activity assay, where fluorescence intensitivity decreases depending on helicase concentration; B - generation of dsDNA depending on BstPol amount in the assay.



The measurements showed that lsrACDBFG promoter activity can be manipulated by changing AI-2 concentration (range: 20-40μM) and the second genetic circuit with mazef complex can be used as an inducible kill-switch (Fig. 3).

After successfully producing AI-2, another aim was to test the genetic circuit J23XXX-sfGFP if it reacts to AI-2, and if so, at what concentrations the signal is increased. Before determining the most suitable promoter for treatment genetic circuit, a few promoters from the Anderson promoter collection were compared with AI-2 inducible promoters: lsrACDBFG, EP01r, EP14r. The results did not reiterate as expected. Only AI-2 inducible promoter lsrACDBFG-sfGFP signal showed a correlation with increasing AI-2 concentration. To add more, after the last experiments construct lsrACDBFG-mazF-J23117-sfGFP showed positive results: there can be seen a positive toxin mazF activity. The increase of AI-2 concentration leads to cell-lysis of E. coli (Fig. 4).



Immunogenic proteins synthesis

  • VHSV glycoprotein G coding sequence and 6xHis tag on the N terminus was cloned into pfX7 vector. Recombinant protein synthesis was tested in five different S. cerevisiae strains.
  • F. columnare gliding motility lipoprotein J (GldJ) gene was cloned into pET28a(+) vector. Synthesis of recombinant GldJ protein was tested in three different E. coli strains.

However none of these protein synthesis was successful, so further optimization needs to be done in this area.

Alginate beads testing

Nonetheless, to test whether immunogenic protein would reach the midgut of the fish undigested, physical and chemical alginate beads testing was performed. After simulating gastric conditions with trypsin in pH 4, we saw that these beads would be not affected while traveling through the fish gut (Fig. 5).

Physical alginate beads properties were tested by applying force at a rate of 10 mm/min. The stress and strain at failure values were below recommended 0.13-0.32 MPa, which is higher than the pressure in the fish digestive tract.

These chemical and physical properties proved that alginate beads could be successfully used as a tool for subunit vaccines development.


We would like to make our tests suitable for quantitative determination of analyte concentration. Thus, to solve this problem we have decided to build a strong foundation by developing a Python script that would detect strips (Fig. 9) and measure the signal intensity (Fig. 10-11) on the test line.

Figure 9. Detected strips with arbitrary labels.

Figure 10. Strips with a profile line that measures the intensity of the test line.

Figure 11. Final analysis picture with peak intensities. Blue color - intensity read from profile line. Yellow color – smoothed signal. Red triangles – detected peaks.

Such an approach would facilitate the development of quantitative LFA strip tests in the future.

This measurement procedure can be applied to many types of LFA tests and allow other further projects without any effort quantitatively measure the test line intensity.

Software & model

onFlow software

We divided our software architecture into two main parts - the back end, which is the logical component, and the front end, the graphical user interface (Fig. 12). While the software front end is only used for taking user inputs and showing the results, the back end does all the ‘heavy lifting’ - obtaining suitable strip test design and sending API requests to the KOFFI external database.

The software can be accessed through the front end, using a graphical interface with explanations and illustrations of the parameters for easy use. A more advanced user or another software can query the software through the back end's API endpoints as a HTTP POST request to

Figure 12. Diagram showing the sequence of operations in the software.


Not every LFA user is able to understand Partial Differential Equations (PDEs), not to mention correctly formulate variational problems and simulate the whole nonlinear PDEs system. Therefore, as great agreement was met with our wet lab results, we decided to fill this gap by integrating our LFA mathematical model in the software.

We chose to implement the model using the Python-friendly computational finite element method platform FEniCS. This was done because it is open-source, has a Python interface, and its back-end is implemented in C++, as a result it does not compromise speed. The model simulates a partial differential advection-convection-reaction equation system and estimates the optimal test line location (Fig. 13).

Figure 13. Simulation of the full advection-diffusion-reaction PDE system. The concentration of the A – analyte, B – the complex of analyte and detection probe, C – detection probe, D – capture probe, E – the complex of analyte, detection probe and capture probe throughout the membrane at t=180 s

Human Practices

Education & Public engagement

While planning human practices activities, our team put a great effort to include as many age groups as possible into these sub-projects, allowing people of any age to get themselves acquainted with the world of life sciences and especially – synthetic biology. We provided careful consideration into adapting the educational material to each of our target audiences, from colouring books to Augmented Reality scenes - nobody got left out. We are proud to have launched initiatives that we hope will not end with us and will continue to be a source of mutual learning and science communication.

  • The 6th SynBio sense is a mutual learning space which is based on Augmented reality technology. People in different cities can scan QR code with a mobile camera and after that, people eventually are able to view 3D models as well as to read or hear more about an exact model.
  • Educational coloring book involves kindergarten and primary school children into synthetic biology in a fun way.
  • With the purpose to help teachers spark interest in science and increase pupils’ motivation to study during a pandemic, we decided to launch an educational game called BioBlox.
  • Inviting an influencer lets us show the main lab principles for wider society, as well as to directly inspire some young people to consider a career in the life sciences field.
  • Online lessons were a great opportunity to educate thousands of students about the basics of synthetic biology, also to talk more about viruses and the importance of preventative measures in the face of pandemic.
  • Workshops helped us to link highschool students theoretical knowledge with a lot of different practical tasks.

Integrated Human Practices

The nature of our project led us to putting a lot of thought into how our team could assist the aquatic life, while simultaneously expanding our scope of impact and collaboration to the broad communities. Due to this we have decided to exclude three different integrated human practices groups: academia, companies and society.

  • Companies, such as aquaculture farm FishNet or National Food and Veterinary Risk Assessment Institute, gave a lot of practical views on detection of these bacteria, as well as they gave insights about the most common bacterial and viral pathogens in fish farms.
  • Academia helped us a lot by discussing and showing different methodologies, as well as they spent a lot of time troubleshooting failed experiments with us.
  • Society shaped our ways of thinking as well as helped us a lot in spreading science for a wider audience.



  1. Strepparava, N., Wahli, T., Segner, H. et al. Detection and quantification of Flavobacterium psychrophilum in water and fish tissue samples by quantitative real time PCR. BMC Microbiol 14, 105 (2014).
  2. Kolm, C. et al. Detection of a microbial source tracking marker by isothermal helicase-dependent amplification and a nucleic acid lateral-flow strip test. Sci Rep 9, (2019).
  3. Motré, A., Li, Y. & Kong, H. Enhancing helicase-dependent amplification by fusing the helicase with the DNA polymerase. Gene 420, 17–22 (2008).
  4. Vincent, M., Xu, Y. & Kong, H. Helicase-dependent isothermal DNA amplification. EMBO Reports 5, 795–800 (2004).
  5. Stephens, K. & Bentley, W. E. Synthetic Biology for Manipulating Quorum Sensing in Microbial Consortia. Trends in Microbiology 28, 633–643 (2020).
  6. Hauk, P. et al. Insightful directed evolution of Escherichia coli quorum sensing promoter region of the lsrACDBFG operon: a tool for synthetic biology systems and protein expression. Nucleic Acids Res gkw981 (2016) doi:10.1093/nar/gkw981.
  7. Qian, S. & Bau, H. H. A mathematical model of lateral flow bioreactions applied to sandwich assays. Analytical Biochemistry 322, 89–98 (2003).
  8. Alnæs, M. et al. The FEniCS Project Version 1.5. Archive of Numerical Software 3, (2015).
  9. Maurice, S., Nussinovitch, A., Jaffe, N., Shoseyov, O. & Gertler, A. Oral immunization of Carassius auratus with modified recombinant A-layer proteins entrapped in alginate beads. Vaccine 23, 450–459 (2004).
  10. Xu, F., Wang, P., Zhang, Y.-Z. & Chen, X.-L. Diversity of Three-Dimensional Structures and Catalytic Mechanisms of Alginate Lyases. Appl. Environ. Microbiol. 84, (2018).
  11. Caswell, R. C., Gacesa, P., Lutrell, K. E. & Weightman, A. J. Molecular cloning and heterologous expression of a Klebsiella pneumoniae gene encoding alginate lyase. Gene 75, 127–134 (1989).
  12. Egerton, S., Culloty, S., Whooley, J., Stanton, C. & Ross, R. P. The Gut Microbiota of Marine Fish. Front. Microbiol. 9, (2018).