After we discovered the need to empower farmers with the ability to swiftly detect an exact pathogen species, we set a goal to create a rapid and specific detection kit. Having these two properties in mind, we decided to base our tool on isothermal DNA amplification - HDA and lateral flow assay - LFA to visualize the results.

HDA is an easy to use method meant to amplify short DNA sequences, while LFA, based on DNA-DNA hybridization, is a fast way to identify amplified sequences specific to a single pathogen species1. We chose HDA because this amplification does not require expensive lab equipment or scientific background and due to this can be performed in rural farms.

The HDA reaction imitates a natural replication process where helicase is used to unwind DNA and allows amplification to occur at a stable temperature without the need for expensive thermocyclers. This method is also easy to design since it requires only two primers compared to other isothermal amplification methods such as LAMP2.

Our team designed this amplification in a way that the reaction is asymmetric, meaning that one primer is used in an excess concentration. This type of amplification generates ssDNA fragments which can be immediately used for our lateral flow assay tests to identify which species of Flavobacterium are present.

We performed HDA reaction according to the manufacturer's protocol. But we were not satisfied with the results since most of the time no or very little amplicon was seen in an electrophoresis gel. In order to have a good working reaction mix, we needed to optimize reaction conditions.

A fragment from F. psychrophilum genomic DNA was amplified via HDA without any modifications. However, F. columnare proved to be challenging to amplify, so we decided to try a few optimizations approaches:

• Different pairs of primers;
• Optimization of reaction temperature;
• The quantity of gDNA/primers/MgSO4/NaCl/enzyme;
• Two-step HDA reaction;
• Genomic DNA restriction;
• Additives (DMSO, PEG4000, sorbitol, dextran, ficoll, betaine, formamide, pyrophosphatase).

We performed amplification reactions using our chosen optimization methods. Based on the results seen in agarose gel, the most efficient additives were DMSO and sorbitol. The reason why these additives improved HDA assay, is their ability to minimise secondary structures of DNA which improves primer annealing3. However, some of the results did not repeat through a series of experiments, so further optimization should be performed to find the best HDA reaction conditions.

Since no optimization was efficient enough to enhance the reaction, we concluded that primers used for F. columnare gene fragment amplification were designed using wrong parameters. To combat this issue, we decided to redesign primers. Newly created primers were longer and had a greater GC% content, thus increasing their melting temperature. All parameters matched perfectly with the suggested parameters for HDA primers design. After testing those primers, we saw that the reaction efficiency was still too low. Eventually, we concluded that a more sensitive method then electrophoresis should be used to determine the concentration of amplicons such as qubit fluorometer or electrophoresis with a more sensitive dye than ethidium bromide.

After researching ways to create a point-of-use test, we decided to base our detection tool on lateral flow assay method4. The main reasons for this decision were that it requires no prior training or scientific knowledge to use and is rapid as well as robust. Thus we imagined that it would be the perfect tool for on-site detection. Keeping this in mind, our detection tool will help to monitor bacterial infections in aquaculture farms as well as serve as a foundation for creating a quantitative LFA test to be used not only in farms but in the wild environment as well.

LFA test is a universal method and with the utilization of ssDNA probes it can be designed to detect all kinds of DNA amplicons. It must not be forgotten, however, that a high level of specificity has to be achieved for the test to function properly and to get as little false-positive or false-negative results, as possible.

To assure specificity we found marker genes for F. columnare and F. psychrophilum, which were 16S ribosomal RNA genes, and created detection (BBa_K3416101, BBa_K3416107), capture (BBa_K3416102, BBa_K3416108) and control (BBa_K3416103, BBa_K3416109) probes for fragments of those marker genes5.

Our team analyzed research papers and wrote all protocols as well as designed experiments necessary to create and test a lateral flow assay detection tool. For LFA to work, gold nanoparticles conjugated to the ssDNA detection probes are crucial. To achieve this conjugation, functionalization reaction must be performed. But we ran into a problem and were unable to functionalize gold nanoparticles with ssDNA detection probes using NaCl “aging” method6. After functionalization reaction gold nanoparticles turned deep blue or colorless. These color changes identify that nanoparticles aggregated during the process.

By trying to make functionalization successful, we put a lot of efforts into improving our protocol like using a greater concentration of ssDNA detection probes. Nonetheless, it did not work, and no functionalization occurred. After these unsuccessful experiments, we talked with prof. A. Ramanavičienė and learned about different methodologies for functionalization.

One of the most promising methods, the professor suggested to us, is based on a low pH environment and is suitable for probes which have a poly-A sequence7. In pH 3 adenine becomes protonated and two poly-A sequences can make parallel duplexes8. Such structures conjugate to gold nanoparticles more efficiently. Since our team created probes with poly-A sequences we thought this method was worth trying. Besides that, we also decided to omit the sulfhydryl group activation and DTT extraction steps, because they can cause unwanted gold nanoparticles aggregation.

After testing, we saw that the functionalization reaction using a low pH method was successful because gold nanoparticles remained red and they did not aggregate during NaCl test so we carried on with our experiments. Shortly after we ran into a second problem and saw that the test created for F. psychrophilum was non-specific and became positive with F. columnare as well as E. coli DNA samples.

If the test is non-specific then it can lead to false positive results which would make our test not suitable for F. psychrophilum identification. We needed to find a solution to improve specificity. Our team read various research papers and came to a conclusion that running buffers can affect the performance of the test9. However, if different optimizations with running buffers would not improve specificity a new marker gene for F. psychrophilum identification was found - rpoC gene (JX657167)10.

Two strategies were designed. First of all, we wanted to see if buffers with SDS, SSC or formamide which are often used in lateral flow assay tests based on nucleic acid hybridization could improve specificity. Second of all, new ssDNA probes were created for the chosen rpoC gene (BBa_K3416110, BBa_K3416111, BBa_K3416112). This time ssDNA probes had a greater GC% content and were longer.

To check the first strategy we tested our lateral flow assay test strips using ultrapure water as well as phosphate buffer. Also two running buffers with different SDS and SSC concentrations and one buffer containing formamide were chosen and tested. In addition, new LFA test strips using ssDNA probes created for rpoC gene were developed and tested with DNA amplicons from different bacteria species.

Results showed that the overall performance of the test was improved by using Running buffer II which contained formamide but the specificity of the test remained unchanged, meaning that this strategy was unsuccessful.

However, after testing LFA test created using

rpoC

gene we saw that the specificity improved. The test was able to positively identify only F. psychrophilum DNA samples. Even though we had a properly working test for F. psychrophilum we ran into a third problem - the test created for F. columnare was also not specific enough.

As mentioned, the test created to identify F. columnare 16S rRNA gene was non-specific. It correctly differentiated between F. psychrophilum and F. columnare but was unable to differentiate between Flavobacterium genus and E. coli. To combat this issue we researched more and found out that cslA gene can be used for F. columnare identification as well 11. We imagine that the LFA test created using cslA gene should be of greater specificity.

We created ssDNA probes for hypothetical protein B0A56_04620 (Flavobacterium columnare, NBRC 100251 = ATCC 23463) because we found it to be homologous to cslA protein sequence. These probes were also designed to have a greater GC% and length to improve hybridization reaction (BBa_K3416113, BBa_K3416114, BBa_K3416115).

However, ssDNA probes were created only in silico. Due to the pandemic that affected the whole world our team was unable to test the specificity of the newly designed probes in the laboratory. For this reason we encourage future iGEM teams to test and if needed improve these lateral flow assay probes for F. columnare.

Since LFA requires a lot of optimization our software tool - OnFlow, based on a mathematical model, will be of immense help for future teams in deciding how to build and optimize their LFA tests. OnFlow will determine where to spray capture and control probes and the exact amount of analyte needed to obtain the most visible results on the strip. This software will help to save time and reagents.

What we learned from developing lateral flow assay tests was that a small fragment of 16S rRNA gene is not enough for accurate Flavobacterium species identification. Even though it is routinely used for phylogenetic analysis 12. Also, during the LFA development process, we gained new knowledge and skills that will be used in a further scientific career.

After we decided to base our test on nucleic acid hybridization via LFA method we started to search for a suitable isothermal ssDNA amplification method, which could be as sensitive and accurate as possible. When we came upon isothermal helicase dependent amplification (HDA) method, we started to precisely plan our experiments. However, during the planning process, our team realized that the usage of commercial HDA kits rises the price of the test.

During literature analysis we have found a significantly cheaper alternative for the HDA assay. The solution is a bifunctional protein complex called helimerase. Helimerase, according to literature, showed the ability of amplifying significantly longer fragments during HDA reaction13.

Since helimerase complex could be formed in vivo and in vitro, it became crucial for us to find the best conditions of these proteins synthesis and purification. Firstly, based on the research1, we were decided to clone TteUvrD and BstPol into different plasmids. However, after integrated meeting with dr. Giedrė Tamulaitienė, we decided to clone both genes into pETDuet-1 vector. This change of direction should help to minimise possible failure risks as much as possible.

The cloning design of the helimerase complex was based on pETDuet-1 vector, which is created for the co-expression of two target genes due to its multiple MCS.

For the development of such plasmid we performed two steps cloning. During the first cloning, in the MCS-1, WZA2-L1-TteUvrD was fused with 10xHisTag and Maltose Binding Protein (MBP) in its N terminal. During the second cloning WZB1-L1-BstPol was fused with StrepII tag. All these cloning steps were successful, however, during co-expression in E. coli BL21 (DE3) strain, only soluble TteUvrD protein was successfully synthesized.

After result analysis we concluded that the reason why there was no BstPol synthesis could be related with the differences on its copy numbers. Even if pETDuet is low copy number plasmid, 15-20 copies per cell could be too much to translate both recombinant proteins properly.

In order to tackle this issue we have decided to clone the BstPol gene into another plasmid, thus rejecting the idea to perform co-expression from one plasmid.

As synthesis of TteUvrD was successful, we have decided to use the same construct for further experiments.

However, after we concluded to change the cloning strategy of BstPol, firstly we needed to find the most suitable plasmid for co-expression. For this purpose we had two main criteria – plasmid copy number must be lower than 15-20 and it should have a compatible ori site with pETDuet. One such plasmid is pACYC, which copy number reaches only 10 copies, and its ori site is compatible with pETDuet14.

After BstPol was cloned successful into pACYC plasmid, we performed its synthesis in E. coli BL21 (DE3) strain. Regarding to TteUvrD and BstPol high-solubility, we were able to purify high amounts of both proteins, so no further optimization for protein synthesis was needed. Nonetheless, protein co-expression and fusion in vivo by using these two compatible plasmids was still unsuccessful.

Despite many attempts, further investigation on the recombinant proteins co-expression is needed. Cloning strategy could be improved by swapping TteuvrD and BstPol places in pETDuet or pACYC MCS. Also, different types of plasmids could be examined in order to find the best options for helimerase formation in vivo.