Engineering Success

Introduction

To make a strain of Bacillus subtilis express fungal adhesion proteins and assimilate on the surface of baker's yeast (Saccharomyces cerevisiae), we utilized the Modular Cloning Method (MoClo). The idea was to assemble multiple plasmid parts simultaneously into a single vector by utilizing type IIS restriction enzymes (BsaI and BsmBI). These enzymes cut outside their recognition sites to generate four-base overhangs, allowing customizable vectors (Addgene: Modular Cloning Guide). For our project, a yeast MoClo toolkit found in an article by Lee et al.¹ was used for all assemblies.

The engineering principles of synthetic biology have been a guide in our process. The first step of the DBTL cycle, Design, included brainstorming sessions to narrow down our problem, identifying tools and resources, and planning. The Build phase revolved around constructing the plasmids through Modular Cloning as described. In the Test phase, we relied on methods such as gel electrophoresis, colony PCR, and restriction digestion to test if our design worked. Lastly, in the Learn phase, we checked the results from the Test phase and made adjustments. Also included in our project was the DBTL cycle of testing the organism that we engineered. We designed our final plasmids in the last DBTL cycle, then built our organism by transforming them with our plasmids. We intended to test them against each other (S. cerevisiae to B. subtilis) and in a chitin column, learning by doing statistical analysis of replicates.

Design

The two different adhesion proteins, Synthetic Peptide 1 (SP1) and Chitinase 92 (CHI92), were investigated, as well as native chitinase in the S. cerevisiae cell wall. The three final constructs are presented in the figures below. All plasmids were built in the software Benchling.

All final constructs utilized the pYTK095* backbone from the toolkit presented in Lee et al.¹ which can be seen in the figure below. The backbone contains a resistance gene, CamR, that gives the carrier of the plasmid resistance against chloramphenicol, used for selection in case of successful integration. It also contains an origin of replication as well as promoters/terminators, an ori and the sfGFP gene. The sfGFP gene, as the name suggests, produces GFP. Outside the GFP gene and its promoters sit recognition sites for BsaI cut sites. All of these parts make it a good candidate to use as a backbone for MoClo. As described before, when we perform MoClo we cut our desired gene fragment and the backbone with BsaI. This creates overhangs that then connect together, effectively swapping out the sfGFP gene with the desired gene fragment. The reason for replacing sfGFP is that we can easily check which colonies have the right insert, as the wrong ones will fluoresce because of the GFP gene remaining.

A paper by Chamas et al. explored the display of adhesion proteins on Escherichia coli Nissle for localisation of Candida albicans². This paper described two basic parts that were proven to bind to Candida. We wanted to express them in B. subtilis as it is found on the human body and might be more viable in a clinical environment. But E. coli and B. subtilis have different cell walls, which led us to seek a different expression path. During our design stage we went through several options before landing on a basic part that Kim et al. presented ³. They called the part Yuab, which produced a cell wall protein that protrudes through the cell wall. This expression pathway was native to B. subtilis. In the paper, they used a type of linker sequence together with the desired protein and Yuab, which made the protein be expressed and linked to the membrane of the bacterium. By adding our GOI to this display construct, we could express them on the B. subtilis membrane.

The transcriptional units of each plasmid were designed to produce different outcomes and will be discussed individually from this point.

CHS3

• pYTK095* backbone
• pYTK002 connector 1
• pYTK011 Promoter
• CHS3 gene
• pYTK056 terminator
• pYTK072 connector 2

For the S. cerevisiae construct, the goal was to overexpress chitin in the cell wall. This would allow for better testing and for mimicking C. albicans more accurately. For this, we chose to use CHS3, a native gene in baker's yeast responsible for chitin expression in the cell wall. The backbone consisted of different elements from Lee et al.'s paper. We constructed it this way to be able to build in E. coli and later cut the CHS3 gene out into a new backbone built for S. cerevisiae.

The main focus with the different backbones was to create recognition sites for BsaI and BsmBI, allowing us to use BsaI for the construction of plasmids in E. coli. This was done to produce more CHS3. Then we would use BsmBI to transfer it into a pre-made backbone prepared by our lab supervisor, called pLE_Ytk007. It also has its origin in Lee et al. and contains a yeast ori, which was the main reason for swapping the backbones. We also used a kanamycin resistance gene in this construction.

Build/Test

Following the Design phase, the Build phase began and took place in the wet lab. The bulk part of the build phase is the assemblies from level 1 plasmids to level 2. Naturally, tests must be run simultaneously as they are built to confirm progress, and therefore the “Test” phase will be discussed in the following section as well. All the experiential methods that were used will be presented, but the protocols can be found under the tab Experiments.
The build phase started with ordering the Sp1 and Chi92 parts. As mentioned before we started ordering the constructs in two parts. With the linker being the separation point. When the order arrived we started doing Moclo to connect the parts with our pYTK095* backbone. After every MoClo experiment we transformed the ligation mixture into E.coli. We used heat shock the majority of the time and every time we plated the cells with low and high concentration separately to ensure a high likelihood of getting colonies. Plasmid extraction was done with Thermo Fisher's plasmid miniprep kits. The testing of these plasmids varied during the project. In the first week we encountered a problem with the first test. The green white screening was hard to do as not many colonies had grown and none were green which was unusual. We put the plates in the incubator longer and more colonies grew. After that we decided to send the plasmids in for sequencing, instead of doing a test in the lab.
The result came back and after checking in Benchling we noticed that a lot was wrong. Thus sending us back to the design stage. The parts that we ordered arrived in shuttle vectors with the same resistance as our final backbone. This complicated the transformation, as some bacteria might take up the original plasmid instead of our new one and still survive the antibiotics. To try to fix the issue we went back into the lab and added more enzymes that only cut in the shuttle vector. The goal was to create a new incentive to use our correct plasmid by cutting up the shuttle vector, thus hopefully rendering it useless and unable to give resistance.
In addition we also ordered primers that extend our GOI with the cut sites, so we can do PCR and just add the resulting fragment in future MoClo. In case our first strategy did not work. We chose this approach to try to mimic real lab work where ordering new constructs that are not sponsored by iGEM, and primers are a lot cheaper than whole plasmids. To test the MoClo assembly with the extra enzymes we performed a colony PCR and ran the results on a 1% agarose gel. The result came back wrong as we had more bands than we expected and no band was the correct size. We checked Benchling again and when we added the extra enzymes to cut the shuttle vector we had accidentally added an enzyme that had a recognition site in our GOI. That recognition site had been mutated in when Twist made our order. So we decided to make three new strategies moving forward.
The first was ordering the whole construct in one piece to then cut into our backbone. The other was to inoculate and transform the vectors we had and hope for the best. The third was redoing the MoClo with the first order but remove the enzyme that had several cut sites.
We redid the MoClo several times and checked the transformed cells with colony PCR which often resulted in no bands in the gel. We initially thought that our issues had to do with the PCR reaction. Initially we used Sapphire Star polymerase and buffer, then we tried Phyre and later we used Phusion polymerase and buffer. Phyre and Phusion are more precise than Sapphire, meaning that they cause fewer mutations when replicating the DNA, however they require longer run times. The Phusion polymerase resulted in successful bands of the CHI92 plasmid but not of SP1. When we transformed these promising plasmids into E.coli again, the CHI92 plate did not grow but SP1 did. So we did a restriction digestion to further check if it was correct. Sadly it still showed too large of a band thus proving the MoClo was the problem.
This erased the option of redoing but without the extra enzymes. We then switched to our most promising plan which was to try to order the whole construct in one piece.
At the same time we started working on the S. cerevisiae construct. As we mentioned above, we wanted to over express CHS3 by placing it in an expression vector. For this we needed CHS3 with cut sites flanking the coding region. As CHS3 is a native gene in S. cerevisiae we designed primers that had recognition sites on the CHS3 gene but with overhangs that included cut sites for BsaI and BsmbI. One problem we encountered was that CHS3 has a BsaI cut site in the middle of it. We designed a mutation primer to mutate away this cut site, so it would not cause issues later on. Then we used these flaking primers and with yeast DNA as template we performed PCR to acquire our GOI. Additionally, as insurance we ordered the whole CHS3 gene without the cut site and with the flanking regions in case we encountered more issues with the PCR. In the end, after many PCR trials with different polymerases, we got a correct band on a gel and proceeded to use this result to create our E .coli construct (p001) using MoClo.
Instead of using colony PCR to check the results of the MoClo we did restriction digestion. Colony PCR can be quite tricky with large fragments which we indeed had with our CHS3 gene (~3500 bs). We did amplification PCR before mutation PCR to get more of our gene. The plan was that after mutation PCR we would digest it with Dnp1 to remove the original template DNA, check the results on a gel, then amplify with PCR, and purify the PCR results. Finally we would perform gibson assembly, and later transform into E .coli. This plasmid would then be used to create our final construct, ready to transform into yeast.
However, when we checked the results after Dnp1 digestion we noticed that the mutation primer had been ordered wrong: it missed the actual base pair mutation. We also realized that the whole CHS3 gene ordered in the beginning would arrive the next day. Therefore, we pivoted to work with our ordered CHS3 gene which we removed the cut site from. MoClo was then performed with the backbone and our ordered CHS3 gene, which worked much better. We got the correct bands with restriction digestion and the plasmid containing CHS3 could then be transformed into E.Coli using heat shock and from a green white screening the colonies looked to be successful. A colony PCR however gave vague results showing faint bands at the right site but to many bands. We sent the plasmid in for sequencing which was inconclusive, the DNA sequenced was not our gene insert but not the backbone either. Due to lack of time we decided to extract and transform it into yeast anyway since the green white screening was successful and there looked to be bands where we expected them. After transformation, S. cerevisiae grew which also indicated successful uptake of the plasmid.
Separately, when the SP1 and CHI92 orders arrived which contained the whole constructs, we changed the plan again. As time was running out we noticed that we could use the shuttle vector instead of our backbone as it already had the correct ori. By changing the resistance it became a sufficient backbone. But one problem remained, which was that the stop codons for the constructs were either lost in the manufacturing phase. Thus, we needed to order mutation primers to mutate back the stop codons. This was discovered in Benchling after double-checking the sequence of our order. What followed was many weeks of mutation PCR, including a second order of mutation primers since we ordered two of the same primers for one of the genes, instead of a forward and a reverse.
After many different settings on the thermocycler and different polymerases were tested, we succeeded by following every protocol very carefully, keeping the reaction on ice at all times, adding the polymerase last and dNTPs second to last, and preheating the PCR machine. After we checked that the correct length was found on a gel and extracted the DNA from the gel, and annealed the amplification strand. From there we moved on to attempting to transform the two constructs int B. subtilis. Several methods were tested for transforming into B. subtilis but nothing was successful. The problem seemed to be getting them competent as the growth was very slow. In the end the transformation was never successful, so we could never test it on our S. cerevisiae cells.
We decided to send our CHI92 and SP1 plasmids in for sequencing. We discovered that both the promoter (though slightly mutated), the RBS and YuaB was present in our construct which told us that it was very likely that CHI92 and SP1 were too.

Learn

Through the project a lack of effective communication and inexperience affected the wetlab work and results. Everyone had different levels of independence in the lab which we should have taken into consideration when making the schedule, pairing people who had more experience with those who felt less sure.
The project could have benefited from a more consistent approach in labeling, documenting and following protocols as small mix-ups and misinterpretations of protocols sometimes caused us to have to repeat the same protocol several times, losing valuable time.
We believe that more time would change our results drastically as it would allow us to redo the transformation of B. subtilis and the transformation of CHS3 if it turned out to be faulty. Because of the lack of time we had to assume that the CHS3 had been implemented which turned out not to matter since the B. subtilis transformation didn't work anyway.
The project planning phase in the beginning of the projects faced many issues. Our first project leader left which set us back a couple of weeks. With a more concise plan and defined roles and ambition, the project could have started earlier, which in turn may have enabled us to start the lab work earlier.
Moreover because of the lack of time we did not do the rest of the planned project. The plan was to successfully transform B. subtilis with our CHI92 and SP1 constructs, make them express the binding proteins and test them. We aimed to test B. subtilis against our transformed yeast. We would grow B. subtilis and S. cerevisiae together and then check if B. subtilis is bound to S. cerevisiae through microscope imaging. This test was to see if SP1 was better at binding as it binds to carbohydrates instead of chitin, which CHI92 binds to. We would not get data out of this test, but still confirmation that the constructs worked as intended in the real world. The other test was to use an affinity column with chitin beads. By counting the cells before adding them to the affinity column and after, we would have been able to assess how well SP1 and CHI92 bind to chitin. A negative control would also have been performed using wild type B. subtilis. The counting of the cells would have been performed using a haemocytometer. The amount of cells that would have stuck to the column over several replicates would have allowed us to test if the modified cells bound to chitin more than the wild type cells with statistical significance.

References

1. Lee ME, DeLoache WC, Cervantes B, Dueber JE. A highly characterized yeast toolkit for modular, multipart assembly. ACS Synthetic Biology. 2015;4(9):975-986. doi:10.1021/sb500366v
2. Chamas A, Aranda-Díaz A, Harousseau G, Zulauf KE, Torrealba J, Cruz C, et al. Engineering adhesion of the probiotic strain Escherichia coli Nissle to the fungal pathogen Candida albicans. mBio. 2024;15(3):e0063924. PMID: 39265099; PMCID: PMC11669158.
3. Kim D, Kim W, Kim J. New bacterial surface display system development and application based on Bacillus subtilis YuaB biofilm component as an anchoring motif. Biotechnology and Bioprocess Engineering. 2021;26(1):39-46. doi:10.1007/s12257-020-0397-7