Our project unfolds through three dynamic modules, each composed of key steps that build on one another. At its core lies the engineering cycle (Design, Build, Test, Learn) which we have embedded as a driving force behind every stage. Each iteration within every module has influenced the others, sparking innovation, refining our approach, and shaping the project into its final form. Because learning never stops, and improvement is always within reach.
Division of labour & HMO production
As a proof of concept, we began by concentrating on one HMO family, the fucosylated HMOs. To access the fucose building blocks, the first step would be to depolymerize fucoidan into fucose monomers using a surface-displayed fucanase, thereby avoiding the need to transport large polysaccharides across the cell membrane. In parallel, a second subpopulation would be engineered to express both a fucose permease and a fucosyltransferase, a combination expected to enable the first steps of HMO biosynthesis, with room for future optimization.
Iteration 1 ▶
Design
The first step of HMO production consists in converting fucoidan into fucose monomers. Therefore, we designed a genetic construction to try to express a fucanase to be displayed on the cell surface using an appropriate signal sequence. We incorporated an antibiotic resistance cassette in the construct for selection, which we flanked with loxP sites to excise it later from the strain.
Build
We ordered the DNA sequence and introduced it naturally into S. thermophilus. We selected the bacteria that had incorporated the fragment by plating them on antibiotic containing medium. We then re-isolated a clone by replating, amplified the region where the insertion had occurred by PCR and confirmed the construct by Sanger sequencing.
Test
We used the L-fucose assay kit to detect the presence of fucose. Only one test was done. To detect fucose, we used the Tecan Spark spectrophotometer.
Learn
We didn’t detect any fucose. This led us to different hypotheses; First, It is possible that our fucanase is not displayed at the surface of bacteria. This could be due to a degradation by HtrA since we used a strain in which HtrA was still present. The second hypothesis could be that our fucanase is displayed at the surface but not active due to different conditions.
The only primary result coming from these graphs is the fact that fucoidan absorb at 340nm, probably because this polymer traps ions from the extraction from Undaria pinnatifida. This absorption should not impact our usage of the kit since we do not go over the detection spectrum of our machine (abs = 2) and the information we need is the slope of the curves. For the next graph, the representation of the slope (derivative) should be preferred.
Controlled Aggregation
Controlled aggregation is expected to enhance the efficiency of HMOs production. By physically bringing the subpopulations closer together at the onset of production, we aim to enhance the diffusion of intermediates between them. However, simply forcing cells to aggregate can stress or even harm them. Therefore we wanted to develop a controllable cell–cell adhesion system, allowing us to promote or fine-tune aggregation only when and where it benefits production.
To achieve this, cells display either a nanobody or an α-rep targeting distinct epitopes of GFP, creating a “molecular glue” that promotes close cell–cell contact.
Iteration 1 ▶
Design
We began working with a strain already available in the lab, in which the three loci responsible for exopolysaccharide synthesis had been deleted. We expected that removing these surface structures would help reduce steric hindrance at the cell surface. However, we observed that this strain displayed a reduced growth rate compared to the wild type (WT). We decided to address this issue before moving forward with the strain.
Build
We performed 10 rounds of random evolution by continuously subculturing the same population for 10 days.
Test
After this adaptive evolution process, we sent the final strain for whole-genome sequencing.
Learn
The evolved strain showed improved growth, but sequencing revealed a duplication of one exopolysaccharide locus, suggesting that exopolysaccharides were still being produced. Since these structures appear to play an important role for the cells, we chose not to continue working with this strain.
Iteration 2 ▶
Design
We needed to construct two distinguishable strains, in which we wanted to split the metabolic pathway later. To do so, we wanted to express cytosolic free mScarlet and mTurquoise in distinct LMD-9 cells. We incorporated an antibiotic resistance cassette in the construct for selection, which we flanked with loxP sites to excise it later from the strain.
Build
We ordered the DNA sequences and introduced it via natural transformation into S. thermophilus. We selected the bacteria that had incorporated the fragment by plating them on antibiotic containing medium. We then re-isolated a clone by replating, amplified the region where the insertion had occurred by PCR and confirm the construction by Sanger sequencing.
Test
Visualization of both fluorescent strains under an epifluorescence microscope.
Learn
These strains are easily distinguishable under an epifluorescence microscope, we can now incorporate the nanobody and the α-rep in these fluorescent strains.
Iteration 3 ▶
Design
We designed two genetic constructions to express the nanobody and the α-rep and display them at the surface of the cell using an appropriate signal sequence. We incorporated an antibiotic resistance cassette in the construct for selection, which we flanked with loxP sites to excise it later from the strain.
We also needed purified GFP to serve as a molecular glue between the partners. Fortunately, we already had a GFP-overexpressing strain in the lab. This GFP carried a C-terminal 6His tag, which facilitated its purification via an Ni-NTA column.
Build
We ordered the DNA sequences and introduced it via natural transformation into both mScarlet and mTurquoise producing strains. We selected the bacteria that had incorporated the fragment by plating them on antibiotic containing medium. We then re-isolated a clone by replating, amplified the region where the insertion had occurred by PCR and confirmed the construction by Sanger sequencing.
In parallel, we purified the GFP on a column of nickel and measured its concentration.
Test
We performed a visual aggregation test, which consists of looking at our cells under epifluorescence microscopy after incubating our two subpopulations for 30 minutes in the presence or absence of GFP. We chose to work with 10 µM of GFP as a first trial, since we didn’t know what to expect.
Learn
There was far too much GFP, and we observed an intense GFP background. In these conditions, it is very unlikely that our cells will aggregate, since they probably bound to different GFP molecules. We cannot draw conclusions about the efficiency of our system, but we learned that we will have to use lower GFP concentrations (nM concentrations).
Iteration 4 ▶
Design
We wanted to test the functioning of our aggregation module, therefore we had to think about a suitable GFP concentration. We chose to test aggregation at 10 or 100 nM of GFP.
Build
Preparation of cocultures with or without GFP.
Test
Same visual aggregation test that in the iteration 2.
Learn
This primary data drew a first evidence of our aggregation. Statistics has been made to reinforce the microscopy images (see measurments ). Nevertheless, we lack proper controls and repetitions to be sure about our aggregation system. So, we planned to go back into microscopy, take more snaps of representative statistical analysis and same for control using mScarlet alpharep cocultured with mTurquoise alpharep (same with nano-body).
Metabolic interdependence
We needed a strategy to maintain a stable equilibrium between our bacterial subpopulations. Without some form of regulation, one strain could easily outgrow the other, jeopardizing the efficiency of the entire system.
Metabolic interdependency could help maintain this balance between subpopulations. We imagined using one S. thermophilus strain engineered to metabolize lactose into glucose and galactose, and another that can’t metabolize lactose but only galactose. So, the goal would be to engineer a Gal+ LacZ− strain. This mutual reliance would foster a stable, self-regulating consortium.
Iteration 1 ▶
Design
It was described in the literature that some strains of S. thermophilus can metabolize galactose due to spontaneous mutations. Therefore, we wanted to try to select and isolate those mutants.
Build
/
Test
We plated LMD-9 on M17gal to select spontaneous mutants.
Learn
We obtained spontaneous mutants, that we will use for this module!
