Measurements

In science, results are very important, but those results don’t mean anything without good measurements

As explained on the other pages, we divided our project into three modules. Here, we will focus on the aggregation module and the associated measurement methods.

The first step was to purify the GFP (see protein purification protocol). After this purification, we had to quantify the concentration of the protein. To do so, we used the Bradford method. Thanks to the TECAN Infinite 200, we were able to determine the concentration of our GFP, post-purification, using albumin as a reference.

Bradford calibration curve
Figure 1: Bradford test at 595 nm. The yellow dot is corresponding to the GFP from the first batch and the red dot corresponding to the GFP from the second batch.

Both GFP were diluted 100 times for the Bradford test. The yellow dot corresponding to the GFP from the first batch represents a concentration of 10.5 mg/mL. The red dot, corresponding to the GFP from the second batch, represents a concentration of 12.5 mg/mL.

Once our GFP was purified, we were able to test whether it could induce cell aggregation by binding to the α-Rep and the nanobody. To test that, we went to the epifluorescence microscopy. Thanks to that method we were able to verify the fluorescence status of our bacteria. Once we verified the fluorescent status of our bacteria, we were able to check the aggregation thanks to GFP, mScarlet and mTurquoise.

LMD9 transformed with mTurquoise
Figure 2: LMD9 transformed with mTurquoise
LMD9 transformed with mScarlet
Figure 3: LMD9 transformed with mScarlet
Aggregation trial without GFP
Figure 4: Epifluorescence microscopy of, in red, LMD9 nano-body and, in cyan, LMD9 α-rep. This sample does not contain any GFP.
Aggregation trial with 10 mM of GFP
Figure 5: Epifluorescence microscopy of, in red, LMD9 nano-body and, in cyan, LMD9 α-rep. This sample contains 10 nM of GFP.
Aggregation trial with 100 mM of GFP
Figure 6: Epifluorescence microscopy of, in red, LMD9 nano-body and, in cyan, LMD9 α-rep. This sample contains 100 nM of GFP.

Once we managed to get fluorescent images of a co-culture of our constructions, with two different discernible fluorophores (mScarlet, represented in red and mTurquoise, represented in cyan), we wanted to have more precise data than eye-analysed images.

To be able to use these images as treatable data, we first mapped our samples in RStudio using this code:

R script — Cell detection & merging

This code, adapted to any of your samples from epifluorescence microscopy images, gave us these maps:

Aggregation trial without GFP
Figure 7: cell map of figure 1 obtained with the given code, color code kept. Coordinates in pixels.
Aggregation trial with 10 mM of GFP
Figure 8: cell map of figure 2 obtained with the given code, color code kept. Coordinates in pixels.
Aggregation trial with 100 mM of GFP
Figure 9: cell map of figure 3 obtained with the given code, color code kept. Coordinates in pixels.

Now, with mapped points associated with x and y coordinates, we can define for each cell its closest neighbor and determine their distance. We extracted this information using this second part of the code :

R script — Nearest neighbor distance

Using this to all our eleven samples, we obtained data on 12977 cells in total. By grouping our data by GFP concentration in different samples, we obtained a significant decrease in the mean distance between neighboring cells upon the increase of GFP concentration.

Mean distance between neighboring cells grouped by GFP concentration
Figure 10: Mean distance between neighboring cells from 11 samples, grouped by the GFP concentration used during co-culture.
** = p ≤ 0.01;  **** = p ≤ 0.0001;  ns = not significant between 0 and 10 nM of GFP.

These results are interesting, but they do not ensure that our assumed aggregation is heterologous as expected. To further confirm this aggregation, we plot the same dataset by the identity of the origin cell (color) and the neighbor cell. This led to this plot:

Mean distance between neighboring cells depending on the pair
Figure 11: Mean distance between neighboring cells depending on the pair. “From” (columns) indicates the color of the reference cell, and “to” the color of the neighboring cell.
n = 12 977;  ns = no significant groups.

Even if this barplot shows you equivalent means, we are quite interested in its analysis. Ideally, we would want to see significant decrease in the mean distance in heterologous couples (mScarlet to mTurquoise and mTurquoise to mScarlet) or this is not the case here. The fact is that our statistics are biased by our own chassis. Indeed, “strepto” in Streptococcus thermophilus means chain. And like a chain, S. thermophilus cells are making a small thread of 3 to 4 cells (as observed in our microscopy results), genetically the same, implying that the neighboring cell of a cell inside a chain is, except correct aggregation, the neighbor of a cell of the same color. These data are so biased towards homogenous couples (mScarlet to mScarlet and mTurquoise to mTurquoise). Taking that into account, this non-significative barpot seems more hyping for our project.

To go further, we wanted to calculate the impact of the chain effect by using proper controls like the co-culture of LMD9 mScarlet α-rep and LMD9 mTurquoise α-rep and LMD9 mScarlet nano-body and LMD9 mTurquoise nano-body. Doing this same work on these samples, we expect to see a significant increase of the distance between heterologous couples (mScarlet to mTurquoise and mTurquoise to mScarlet), corresponding to the chain effect.