In order to show that optimizing the stoichiometry of the curli pathway can affect curli production, we attempt to demonstrate that variations in the RBS strength of csgG can impact levels of curli production. This would give a proof-of-concept foundation for the idea that curli production can be optimized by regulating expression strengths of the pathway proteins.

Generating RBS Variation

We employed a simple screening process in order to identify and select RBSs of varying expression strengths. Using the Salis Lab RBS generator, we identified a degenerate RBS sequence with a wide range of expression strengths based on whether each degenerate base crystallized to adenine, guanine, cytosine, or thymine. We sent that degenerate sequence out for synthesis and got back our library of interest.

Then, we used Gibson Assembly to position the RBSs in front of the csgG gene within the pBbB8K-CsgBACEFG plasmid used by the Joshi Lab, which was originally the consolidated curli operons with the wild-type RBSs for each protein under an arabinose promoter. The Gibson Assembly produced a library of plasmids which were identical except for the csgG modified RBS.

Detecting RBS Variation

The assembled plasmids were transformed into competent PQN4 E. coli cells and grown on plates with arabinose and Congo Red. The arabinose induced curli production while the Congo Red was there to indicate how much curli was being produced per colony.

Generally, Congo Red has been shown to bind to amyloid proteins like curli, thereby making the substance red, and can therefore act as an indicator of curli presence. However, it binds non-specifically to curli, meaning that it can also bind other extracellular components of E. coli cells such as pili. This is why we chose to transform into the PQN4 strain, which has knock-outs of most other extracellular features. Furthermore, adding just a tiny amount of Congo Red to the agar plates made them also turn red, so it was not possible to visually identify which colonies were producing curli and which weren’t.

To make such a distinction, it was necessary to take fluorescent images of the plates. After that, a MATLAB script was used to compare the average pixelation brightness between colonies, and we made a preliminary conclusion that the brightest colonies had the most curli production.

Figure 1

Figure 1: An example of a fluorescent plate image from one of our RBS variants of csgG.
The colonies have differing average brightnesses which was used to conclude relative levels of curli production.

Figure 2

Figure 2: An example of an analyzed fluorescent plate image.
The brightest colonies are circled in blue.

After plate imaging, six of the brightest colonies were selected as well as six that were not among the brightest in order to produce a range of RBS strengths. We labeled the bright colonies B1, B2, B3, B4, B5, and B6. We labeled the low brightness colonies L1, L2, L3, L4, L5, and L6. These colonies were grown up in culture and a Congo Red Pulldown Assay was performed on them.

Quantifying RBS Variation

For the Congo Red Pulldown, cells were induced overnight with arabinose so that they produced curli fibers. Then, cells were spun down and resuspended in PBS. Congo Red was added at this point and mixed so that it would bind the curli produced by the cells. Then, the cells were pelleted once more, leaving some amount of Congo Red in the supernatant. Cells which produced a lot of curli should have a clearer supernatant (i.e. with less Congo Red since most of it will have bound the curli). Cells which did not produce much curli should have a redder supernatant. The color of the supernatant was then quantified by measuring 490 nanometer absorbance and blanking with pure PBS buffer.

Figure 3

Figure 3: An example of the changing supernatant color with differing levels of curli expression.
The clearer solutions on the right represent cultures with more curli production.
The redder solutions on the left represent cultures with less curli production.
The relative redness of the solutions is quantified by measuring absorbance.

In total, the 12 RBS variants generated by our library, a positive control, and a negative control were tested in the Congo Red Pulldown. The positive control was a culture of cells transformed with the original pBbB8K-CsgBACEFG plasmid, meaning that csgG was flanked by the wild-type RBS. The negative control was a culture of untransformed PQN4 cells which should have no curli expression.
Figure 4

Figure 4: The relative curli production of each csgG RBS variant based on the reciprocal of absorbance at 490 nanometers.
The values were blanked with pure PBS buffer.
The standard deviations are included.

As demonstrated by the Figure 4, there is not a significant difference in measured curli production amongst most of the csgG RBS variants. The Congo Red Pulldown Assay does not provide too much resolution to differentiate between variants that produce slightly different amounts of curli. However, a single factor ANOVA with the B1 variant and the positive control demonstrates a significant difference in curli production between these two samples: F is greater than F critical. If only the B1, positive, and negative csgG variants are considered, there is a clear difference in curli production.

Figure 5: The relative curli production of csgG RBS variants based on the reciprocal of absorbance at 490 nanometers.
The values were blanked with pure PBS buffer.
The standard deviations are included.

Therefore, we do see a significant difference in curli production between one of the high-expressing variants and the wild-type RBS, demonstrating that modulating the RBS strength of csgG is a viable way to regulate curli expression and export.