Project Achievements

Goals

We had 3 main goals:

1. We wanted to engineer Lactococcus lactis, a gut commensal bacterium of the hornet, to produce shRNA which would target an essential developmental gene in the hornets. To produce a sufficient amount of shRNA, we had to delete the RNAse gene as well as introduce a plasmid containing the shRNA production.


2. We wanted to test the efficiency of this shRNA in degrading its target, and to avoid animal testing we wanted to perform these tests in S. cerevisiae. To do that we needed to design and assemble plasmids containing the RISC complex (Argonaute and Dicer proteins) and insert them into the genome, since yeast does not naturally express this machinery. Additionally, we wanted to introduce a plasmid coding for the shRNA and a plasmid containing the target of the shRNA downstream of a GFP reporter.


3. We wanted to use well-characterised promoters, to ensure a good expression of the RISC complex. To do that, we needed to characterise a new promoter that wasn’t in the iGEM registry: pTEF1, the promoter of our DICR protein. In that mindset, we set another goal of our lab experiments and tried to characterise this promoter in yeast, with an mCherry reporter, and compared its level of expression to that of 2 other similar promoters, pCYC1, pADH1.

Additionally, we wanted to make sure that our initial project design, namely targeting larvae with a gut commensal bacteria producing RNA interference, was the most efficient solution. To do that, we designed a model that would simulate bee population dynamics confronted to predation by the Asian hornet, and Asian hornet population dynamics confronted to our solution.

Finally, we had another goal in the dry lab. We wanted to make sure that our shRNA would only be accessible to the Asian hornet, since the gene we wanted to target was very similar to other members of the Vespa genus. Our idea was to create a box which would contain a protein bait with our bacteria on it and selectively allow Asian hornets (V. Velutina) to access it. To do that we had to adapt an image recognition software (VespAI), as well as connect our software to our hardware.

Objective:

We wanted to create an E. coli strain that lacked an RNAse, since RNAse activity may interfere with shRNA production (Court et al. 2013; Paddison et al. 2002). Though the strain we wanted to use is L. lactis, we were first thinking of using E. coli as a proof of concept strain, and designed our shRNA producing plasmid with a promoter that works in both organisms.

Methodology:

Figure 2: Strategy to delete the RNAse gene in E. coli with the lambda Red recombination protocol. A kanamycin resistance cassette (KmR) with homology regions to the E. coli genome around the RNAse gene is used to delete the RNAse gene. KmR also has FRT sites to take out the kanamycin resistance out of the genome once the RNAse is knocked out.

The genomic deletion was achieved by following the lambda Red recombination protocol, as schematised in Figure 2(Datsenko and Wanner 2000). Briefly, we made an insert with homology regions to E. coli genome around the RNAse gene, and also a kanamycin resistance gene – it will be referred to as "KmR cassette". The aim was to make this cassette recombine with the RNAse gene in E. coli genome. To do so, the KanR cassette was introduced into E. coli alongside with the pkD46 plasmid, through electroporation. This plasmid codes for a recombinase, necessary to induce recombination. The recombination of the KmR cassette essentially knocked out the RNAse gene. The final step was to introduce the pCP20 plasmid encoding for a flippase that would induce recombination of the FRT sites. This last step removes the kanamycin resistance gene from the E. coli genome.

Results:

We have successfully deleted the RNAse from E. coli, as you can see in Figure 3, when we ran a PCR with primers around the RNAse region in our deletion strains the amplified fragment was around 423 bp, corresponding to the length of the scar in the genome left after the recombination of FRT sites. On the contrary, in the wild type strain, the amplified region with these primers is 956 bp long. This proves that we deleted the RNAse in E. coli.

Figure 3: PCR amplification of the region around the RNAse gene in E.coli with deleted RNAse. On the right : same primers as on the left, amplification of the region around the RNAse in wild type E.coli.

Objective:

We wanted to create a V. Velutina gut commensal L. lactis strain without an RNAse, so that we can later transform it with our shRNA-producing plasmid. We want to ensure that the shRNA that the strain will produce will not be degraded by the bacterium.

Design:

Our protocol was based on a paper by Gu et al. (Gu et al. 2013) where L. lactis gene deletion was achieved by adapting the aforementioned lambda Red recombination protocol that we used for E. coli gene deletion.

We have attempted to do this on two L. lactis strains. The first strain was a kind gift by Ghent University from Peter Vandamme's lab. That strain was extracted from an Asian hornet's gut, and would probably have a higher chance of colonising the gut. Though we would prefer the gut commensal strain, as it has never been used for synthetic biology, we also had a second, more "domesticated" strain, Lactococcus lactis subsp. cremoris MG1363 that we received through our university.

Figure 4: Steps of RNAse deletion in L. lactis with the Red recombination protocol. A kanamycin resistance cassette (KmR), flanked by homology regions matching the L. lactis genome near the RNAse gene, is used to delete the RNAse gene. The KmR cassette also includes FRT sites, allowing for the removal of the kanamycin resistance from the genome after the RNAse knockout is achieved.

Results:

The gene deletion involved the introduction of several plasmids into L. lactis, for example the integration of pkD46 plasmid is the first step of the gene deletion, as it is necessary to induce recombination between the KanR cassette and the L. lactis genome (see intermediate step in Figure 4). After consulting literature, we concluded that the most common way of transforming L. lactis is through electroporation (Papagianni, Avramidis, and Filioussis 2007; Holo and Nes 1989; Gu et al. 2013).

We have had many attempts to electro-transform L. lactis. Briefly, first we have tried to change the media in which we cultured L. lactis, both before and after electroporation. We grew both of our L. lactis strains in liquid GM17 media containing varying amounts of glycine to find the optimal concentration for growth for each strain.We also have added sucrose to both mediums, as well as MgCl₂ and CaCl₂ to the medium used to refresh the cells after the transformation.

Afterwards,we increased the amount of plasmid that we used during electroporation to 1ug. Finally, we attempted to electro-transform our Lactococcus strains with different plasmids: for instance, one had a resistance to a different antibiotic, and another one that we knew has been introduced in the Lactococcus lactis subsp. cremoris MG1363 strain before.

Despite our numerous attempts, we could not insert any plasmid into L. lactis through electroporation. This is why we have decided to try another way of introducing extragenomic DNA into our target bacterium: by conjugation.

The conjugation strategy:

Figure 5: Scheme with steps of L. lactis conjugation strategy. First the oriT is added to the plasmid that will be introduced to L. lactis through Gibson assembly. Then, that plasmid is put into E. coli, that will introduce it to L. lactis when they are co-plated. Finally, to select the l. lactis that received the plasmid, cells are plated on LB with erythromycin and with no DAP.

To do the conjugation, we assembled a plasmid by Gibson cloning that we wanted to use as a backbone for shRNA production and added oriT to it, so that it could be transferred through conjugation. oriT is a short sequence necessary for the transfer of DNA from the donor to the recipient cell during conjugation. The plasmid was then transformed into E. coli strain JKE201 (Harms et al. 2017) (Figure 5).

E. coli and l. lactis were put on the same plate together, with 5 times more L. lactis cells compared to E. coli cells. The plates contained DAP, allowing E. coli to grow and transfer the plasmid to l. lactis through conjugation.

After 3 days of co-plating, we restreaked the cells from the plate we used for conjugation (LB+DAP) on a new plate containing LB, erythromycin, but without any DAP. As E. coli JKE201 strain cannot survive without DAP, and Lactococcus is not naturally resistant to erythromycin (Figure6C ), only Lactococcus that received the plasmid was selected. Thus, the colonies (Fig 6A and 6B)we saw on LB + erythromycin had integrated the plasmid.

Figure 6: Erythromycin containing plates with plated L. lactis. Transformed L. lactis strains after 3 days of growth at 30°C: A. the hornet gut strain and B. L. lactis subsp. cremoris MG1363, both growing on erythromycin plates, confirming that they carry a plasmid conferring erythromycin resistance. C. Untransformed L. lactis strains after 4 days of growth at 30°C, with L.lactis subsp. cremoris MG1363 on the left and the hornet commensal strain on the right. No colonies are visible on either side.

To make sure that the transconjugant colonies we saw on our second plate were indeed Lactococcus we did a colony PCR with primers that bind to L. lactis genome. All colonies had bands of the expected size (around 1 kb) (Figure 7)

Figure 7:Gel of amplified fragments of L. lactis genome, with gut commensal L. lactis colonies on the left and Lactococcus lactis subsp. cremoris MG1363 on the right . The expected size is around 1 kb for both L. lactis strains.

Conclusion and outlooks:

We now have a protocol to insert plasmids into Lactococcus using conjugation and if we had more time we would proceed with our initial protocol to delete the RNAse.

Preemptive bioinformatic and research work:

For the design of the shRNA producing sequence, we identified 20 to 22 nucleotide regions within these genes that follow an AA dinucleotide sequence, as this is believed to optimise RNAi efficiency ("siRNA Design Guidelines | Technical Bulletin #506 | Thermo Fisher Scientific - CH," n.d.). These regions were then used as inverted repeats, separated by a short flexible loop, to generate our shRNAs. To choose the loop in our shRNA, we looked at a study where upon testing in human embryonic kidney 293T cells using a luciferase assay (Schopman et al. 2010), reported that shRNAs based on the mir-17 and mir-25 loops exhibited the strongest target repression, approximately 10-fold stronger than the 9-nt loop.

Objective:

We aimed to clone a plasmid that produces shRNA in order to silence specific developmental genes in the Asian hornet. While we were selecting which developmental Asian hornet genes we wanted to target, we started assembling the shRNA producing plasmids: for subsequent cloning, the backbone, promoter and terminator would be the same, and only the shRNA sequence would change. To assemble the archetype plasmid, we have used two control shRNA sequences: the first one corresponds to a DNA sequence used for CRISPR interference in the Schaerli lab (Santos-Moreno et al. 2020), while the second one corresponds to the sequence that was reported to have a high efficiency from the 2023 Estonian iGEM (“Project | Estonia-TUIT - iGEM 2023,” n.d.) pMG36E_shRNA_scaffold (Figure 8).

Figure 8: Map of our shRNA producing plasmid with the pMG36E backbone, erythromycin resistance gene for selection, the constitutive PTSIIC promoter and rrnb t1-t7te terminator.

Results:

We assembled the plasmid for both of our control shRNA sequences. However, upon sequencing, we found out that the newly assembled plasmids had some mismatches. The promoter had a dinucleotide deletion, but we speculated that it might not affect its function, as it was rather distant from the -10 and -35 regions. Both shRNA showed various mismatches. The terminator was assembled correctly (Figure 9).

Figure 9: Sequencing results of promoter-shRNA-terminator region of pMG36E backbone.

We decided to use the assembled promoter - backbone - terminator construct and replace the shRNA region, to get the correct sequence.

Since all shRNA sequences in different colonies had the exact same mismatches, we hypothesised that the shRNA inverted repeats formed a complex secondary structure during the initial PCR amplification, thereby generating incorrect sequences. To avoid possible polymerase mistakes, we ordered shRNA sequences as long single stranded oligo with their complementary sequences, and annealed them together to form a double stranded fragment. We have ordered both of our test shRNA sequences and four of our target sequences (identified in the following V. velutina genes: 2 Chitin Synthase genes, Vacuolar ATPase and VgR genes).

To assemble our new plasmids we tried in parallel two different approaches:

Figure 10: Schematic representation on the steps for the two cloning strategies to assemble our shRNA producing plasmid.

Results:

We performed ssDNA oligo annealing, PCR amplifications and Gibson Assembly, and restriction-ligation. Upon transformation of our new constructs, we got some colonies on the plates, suggesting that the cells contained either the correct plasmid, or the template, or a byproduct of the cloning. Unfortunately, we did not manage to perform colony PCR on the candidates, and we did not manage to get a plasmid concentration high enough for sequencing (neither Sanger sequencing nor whole plasmid sequencing with nanopore). These observations seemed to point towards the fact that most of the cells did not contain the plasmid.

Troubleshooting :

We realised that most of our plates and liquid cultures contained an antibiotic concentration 5-10 times lower than the one recommended in the literature. We hypothesised that, if the antibiotic selection was low upon transformation, some cells might have grown even without the plasmid.

Thereby, we did another restreak on plates with higher erythromycin concentration in order to maximise the chances to get a colony with the right plasmid. We also tested other DNA-Clean up and Miniprep kits, to rule out the hypothesis that the issues lied in the sample preparation . We managed to increase DNA concentration by 10 folds, but it remained still very low (4-8ng/ul, measured through QuBit). We tried anyway to send the plasmids for Sanger sequencing, but unfortunately they had too low quality to be analysed.

At this point, we speculated that the shRNA secondary structure might interfere with polymerase amplification or prevent the correct assembly of fragments. Hoping that such an effect might be shRNA-specific, we wanted to try with our real shRNA, as their sequence is different from the controls. We were aware that the issue might lie in the hairpin structure and not within the specific DNA sequence.

We used the Gibson assembly method and assembled four shRNAs with pMG36E backbone and then performed the transformation into E. coli.

Results:

We obtained colonies on most of the plates. We performed colony PCR, but unfortunately no colony yielded the right band size on the gel, apart from one. We purified the PCR product and sent it for sequencing, however, the sequence had many mismatches.

Outlooks:

Our hypothesis is that the complex secondary structure that the shRNA sequence forms makes it hard to amplify and to clone. We would like to continue brainstorming ideas on how to make this cloning possible, and perhaps develop the restriction-ligation strategy further, as it does not require a PCR amplification step.

Objective:

To test the efficiency of our RNA interference and to avoid testing our shRNA on live animals, we decided to test the efficiency of the target degradation in a yeast model based on the work of the 2023 Estonian iGEM team (TUIT). We aimed to create our own yeast strain that expresses the RNA interference machinery, namely Argonaute proteins (AGO) and Dicer proteins (DICR) that are part of the RISC complex. More concretely, we created plasmids containing the different parts necessary for their subsequent insertion into the yeast genome. Furthermore, to quantify the efficiency of our shRNA, the objective was to introduce the shRNA producing plasmid as well as the plasmid containing the shRNA target downstream of GFP into the yeast model.

Design:

To express AGO and DICR in our yeast model, we used constitutive promoters that would ensure a stable expression level of the RISC complex. In our design, AGO is expressed by the promoter pPGK1, which was already characterised and registered in the iGEM registry of parts. However the pTEF1 promoter of the DICR plasmid was not.

To characterise pTEF1, we used an integrative construct that expresses the mCherry reporter, similar to how we wanted to express AGO and DICR proteins in S.cerevisiae. We also included the pADH1 promoter and the pCYC1 promoter in similar configurations as the standard for strong and weak promoter in S. cerevisiae (Figure 11):

1. Plasmid pDA95 (pTEF1 promoter)
2. Plasmid pDA120 (pCYC1 promoter)
3. Plasmid pDA122 (pADH1 promoter)

We linearized and transformed these constructs into S.cerevisiae and measured their expression levels of mCherry using plate reader and flow cytometry.

Figure 11: Design of the plasmids containing distinct promoters for characterization of fluorescence levels.

To transform them in S.cerevisiae we learned a new transformation protocol for yeast, in which integrative plasmids have to be linearised before transformation. We did that with the BstBI restriction enzyme. Because these three plasmids can be integrated in the yeast genome multiple times, we first screened 12 colonies and chose the ones that showed the expected fluorescence level of one integration. To choose the right colonies, we were helped by Serge Pelet, an expert on S.cerevisiae at the University of Lausanne.

We measured the level of expression of mCherry reporter with both a plate-reader and a flow cytometer, to ensure a more reliable measurement. Plate-reader measures OD and fluorescence from a 96 well plate, by taking a measurement in each well every 10 minutes, allowing for a high throughput analysis of cell concentration and global fluorescence. Flow cytometer measures fluorescence expression by running a laser through each cell individually, allowing for a more precise quantification of fluorescence. For instance, flow cytometry allowed us to discard colonies that presented two subgroups of cells, suggesting that not all of them had the same number of plasmid integration.

Results:

After choosing colonies presenting the expected expression level of mCherry, we prepared four biological replicates for each construct and we did final measurements on plate-reader (Figure 11) and flow cytometer (Figure 12). For plate-reader measurements, we grew cells in Synthetic defined (SD) yeast liquid medium, and used SD medium that is not inoculated by any cells as a control measure of the blank. For flow cytometry, our control is an empty vector of S.cerevisiae wild-type that does not contain any fluorescent plasmid.

Figure 12: Flow cytometry results of mCherry red fluorescence measurements. The highest expression level is induced by the promoter pTEF1 (blue), as compared to pADH1 (yellow) and pCYC1 (green). The wild-type control (purple) did not contain any construct.

Figure 13: Plate-reader results of mCherry red fluorescence measurements. The highest expression level is induced by the promoter pTEF1 (blue), as compared to pADH1 (yellow) and pCYC1 (green).

Both measurements (Figure 12 and 13) suggest that pTEF1 induces a higher mCherry expression than pCYC1 and pADH1, with the control in flow cytometer measurements having the lowest fluorescence. Therefore we uploaded pTEF1 as a new basic part (BBa_K5047036) of the iGEM registry along with the results of the characterisation we performed for it.

Design:

Both AGO and DICR genes are native from S. castelii (Drinnenberg et al., RNAi in Budding Yeast, Science, 2009) and in the design of our plasmid we made sure to put them under the regulation of constitutive promoters for the insertion into the yeast genome. (Figure 14) To introduce the AGO and DICR plasmids into the genome of S. cerevisiae we first assembled two separate plasmids before combining them into one big plasmid including both parts.

Figure 14: Design of the AGO and DICR plasmids. A. The DICR assembly includes the constitutive promoter pTEF1, the DICR gene and the tSTE2 terminator. B. The AGO assembly includes the constitutive promoter pPGK1, the AGO gene and the tSIF2 promoter.

Results:

The first step in the creation of these two plasmids was the amplification through PCR of the backbone, the promoters and the terminators. Both plasmids required pN1-011 as the backbone, additionally the DICR plasmid needed pTEF1, the 2 g-blocks and tSTE2, and for the AGO plasmid we needed pPGK1, the 3 g-blocks and tSIF2.

Figure 15: A. Gel of amplified fragments pN1-011 (2603bp), pPGK1 (1060bp) and tSIF2 (552bp). B. Gel of amplified pN1-011(2603bp).

Some fragments required several rounds of amplifications with changes in certain parameters like elongation time and annealing temperature. However, most fragments were amplified and ended up being at the expected size on the gel. (Figure 15) The only fragment that never looked clean on the gel was tSIF2 (Figure 16), but we simply proceeded with the assembly of the plasmid since the lower band seemed to be the right size.

Figure 16: Gel of amplified tSIF2 (expected band at 552bp lower band, but we see higher band as well).

The DICR plasmid required only 2 rounds of Gibson assemblies, whereas the AGO plasmid required a little more troubleshooting (for details check the “engineering success” page). Ultimately we were able to create both the AGO and the DICR plasmid separately (Figure 17 and 18).

Figure 17: Sequencing results of DICR plasmid using primers PR_VV_053 and PR_VV_054.

Figure 18: Whole plasmid sequencing results of the final assembled AGO plasmid.

The next step required the amplification of both the AGO and the DICR transcriptional units (i.e. promoter, gene and terminator) from their respective plasmids, and the subsequent assembly of these fragments with the pSP599 backbone (Figure 19).

Figure 19: Design of the assembly of AGO and DICR transcriptional units into one plasmid using the backbone (pSP599) containing ampicillin resistance.

However, we were unsuccessful in the assembly of the final combined AGO and DICR plasmid, which would have resulted in a plasmid of almost 13kb in size. We suspect this final assembly was unsuccessful due to its large size.

Design:

The next steps to determine how efficiently our shRNA could degrade its target were the design of the target and shRNA plasmids. To do so we created two separate plasmids, one which produces the shRNA, and one that contains the shRNA target downstream of the gene coding for GFP. In our design, we made sure to put the target sequence between the coding sequence of the GFP and the terminator. This ensures that while testing the efficiency of our shRNA if we observe a decrease in GFP production it is purely due to RNA interference and not due to transcriptional repression of the GFP. Indeed, if the target were positioned at the start of the gene, there’s a chance the shRNA could block transcription or translation by binding to the gene or the transcript. However, by positioning it at the end, transcription and translation proceed normally, and the only way the shRNA can function is by triggering the degradation of the transcript (Figure 20).

Figure 20: Design of the shRNA and target plasmids. The shRNA plasmid includes the pGAL promoter, the shRNA and the tSIF2 terminator. The target gene plasmid included the pGAL promoter, the GFP followed by the target and the tCYC terminator.

To insert all three plasmids and validate their successful insertion and expression in yeast we made sure that each contained an auxotrophic marker. In order to test if all of these different parts were properly expressed within the yeast, we needed a triple auxotrophic yeast strain so we could grow it on minimal media and identify the triple supplemented yeast (Figure 21).

Figure 21: Schematic of the experimental design for the insertion of three plasmids into yeast.

The shRNA plasmid contains a uracil auxotrophic marker, the target plasmid contains a tryptophan auxotrophic marker, the AGO + DICR plasmid contains a leucine auxotrophic marker. Saccharomyces cerevisiae is lacking the production of all three amino acids.

We did not have enough time to finalise the assembly of the plasmid containing the target downstream of GFP and the plasmid containing the shRNA. Therefore, the final characterisation of the efficiency of our target gene degradation in yeast could also not be performed.

Future perspective:

For the insertion of AGO and DICR into the genome of S. cerevisiae we could try to insert both plasmids separately. This would however require a quadruple auxotrophic yeast strain, since the shRNA and target plasmids would still have to be inserted as well. We believe that with sufficient time, this part of the project could be carried out successfully.