Team:EPF Lausanne/Results



Characterisation of the spatiotemporal dynamics of the CpxR stress sensor

Demonstration of CpxR's dimerization & Elucidation of its dimerization orientation

Previous in vitro experiments (FRET) have shown that CpxR dimerizes. Currently little in vivo information about CpxR’s dimerization is available. To characterize our stress responsive bacteria, we had to confirm that CpxR dimerized in vivo as well as elucidate CpxR’s dimerization orientation.

We synthesized four constructs with combinations of the split IFP1.4 fragments fused to the C or N terminal of CpxR. As seen in the graph below, induction of IFP1.4 signal by 50 mM KCl was done at t=24min. We clearly see that the construct with IFP fragments on the C-terminal immediately responded to stress by emitting fluorescence. In fact we observe a 3 fold signal increase in 2 minutes. On the other hand, the three other orientations were non-responsive to KCl stress.

This 30-fold signal increase in little time from the baseline allows us to assert that only the strain synthesizing the IFP fragments on the C-terminals of CpxR responded to KCl stress. For a further analysis of this experiment check out this link!

Signal induction by various concentrations of KCl & signal shutdown by centrifugation

Having confirmed that KCl was a good inducer for our signal, we tested different concentration of KCl to modulate the signal and better characterize our biobrick. We also aimed to shut down the signal by centrifugation and medium change. The signal was measured on a plate reader 20 minutes, before the addition of KCl. After 2 hours of measurement we centrifuged the cells for ten minutes and replaced the medium with PBS to be able to get a shutdown of the signal.

We successfully showed that increasing concentrations of KCl led to stronger signals up to a saturation concentration of about 80 mM KCl. Moreover we were able to drastically shut down the signal. These results prove the reversibility of the split IFP1.4 and suggest that real-time temporal dynamics analysis is possible for our system. For a thorough analysis of this experiment check out this link!

Visualization of the the CpxR split IFP1.4 activation by KCl stress

Having shown that we were able to monitor the temporal dynamics of CpxR activation, we wanted to see if we could analyze CpxR’s spatial dynamics by microscopy. 10μl of our cells, previously stressed with 80 mM KCl were spread on a glass slide and imaged with a x100 objective and a APC (Cy5.5) filter.

We noticed various characteristics of the cells from the picture below. First, IFP signal was much more present in stressed bacteria rather than in non-stressed bacteria. Secondly, we distinguished two specific phenotypes within bacteria: elongated and normal cells. We noticed that this difference was due to CpxR overexpression as we saw this phenomenon also in non-stressed conditions.

In elongated cells, we were able to distinguish several bright bands of IFP signal that seem fairly uniformly distributed. In the normal phenotype we distinguished a single band in the centre of the bacteria. These observations led us to believe that CpxR might be involved in the division process of E. coli as it seems coherent for cells to slow down division upon stress.

After looking into the literature, similar bands were visualizable in E. coli for factors related to septum formation such as ftsZ or pbpB. Nevertheless when comparing our patterns to the ftsZ and pbpB patterns, we noticed that CpxR might be localized in opposition to these factors. Further experiments comparing the sub-localization of CpxR and ftsZ could help the scientific community better understand how E. coli monitor division under various environments. For a thorough analysis of this experiment check out this link!

results results
To the left: unstressed cells. To the right: stressed cells.

Activation of CpxR - split IFP1.4 on microfluidic chip by chamber crushing

Knowing that we were able to able to visualize CpxR-IFP activation under a microscope, we proceeded to trying to activate the pathway by mechanical stress on a microfluidic chip - the ultimate barrier to building a functional BioPad !

To induce this stress, we turned on the buttons of our SmashColi microfluidic chip at a pressure of 25 psi. We imaged chambers before stress and after stress (10 min after button activation). A drastic increase in signal was detected !

non stressed

Characterisation of the pBAD promoter and folding ability of GFP fused to CpxR

This construct aimed to evaluate the expression of our construct and the characteristics of the arabinose promoter in E. coli by fusing a superfolder GFP protein to CpxR. The sfGFP was chosen because of its higher intensity compared to GFP.

Not knowing if CpxR would react the same way if sfGFP were attached to the N or C terminus, 2 biobricks were built, one with each of the orientations: BBa_K1486002 (N terminus) and BBa_K1486005 (C terminus).

We tested both constructs CpxR-sfGFP and sfGFP-CpxR on a plate reader to characterize the pBAD promoter, measuring the sfGFP signal in function of the arabinose concentration. We also measured the signal in a microfluidic chip. You can see in the graph below that the signal is increasing as the concentration of the arabinose increases. The construct with sfGFP at the C terminal lead to a higher GFP signal.

Characterisation of the split luciferase

Split luciferase complementation assay using CheY and CheZ chemotaxis proteins

CheY and CheZ are two proteins involved in the bacterial chemotaxis pathway. It has been shown by split luciferase complementation assay that these two proteins are not interacting in presence of chemoattractant, but start to interact in absence of chemoattractant or presence of chemorepellent. Based on the work of Waldor1 Laboratory, we wanted to redo the experiment to test our own splits, with firefly (BBa_K1486055) and renilla (BBa_K1486054) luciferases.

As shown in the graphs (fig.1A and 1B), we didn't observe a high signal for our assay. However, the signal being higher than the blanks, it is an encouraging sign that the splits luciferase can be used for other experiments of this kind. A possible explanation for these results is that we didn't completely get rid of the arabinose, which acts as a chemoattractant. Moreover, CheY and CheZ being endogenously expressed in bacteria, there could be interferences with our fusion proteins and weakening of our signal. This should be tested again with CheY/CheZ knock out strains.

cheYcheZ cheYcheZ

We also could determine which of the luciferases would best suit our following experiments. As shown in fig. 2, for the same concentration of substrate, we see that firefly luciferase has a more stable and higher signal. Moreover, the difference between the background noise (negative control, non fused split luciferase) and the full luciferase is bigger for Firefly luciferase, which is also preferable.

The PBS2-HOG1 split-GFP & split Renilla Luciferase stress response

Confirmation of successful transformation via the R. Luciferase tag

Since we had never transformed S. cerevisiae, we first needed to confirm our protein tag transformations as a positive control for subsequent experiments. We cultured the two strains we got from the transformation, the HOG1-rLuc (BBa_K1486033 ) and PBS2-rLuc (BBa_K1486027), put them in a 96 well plate and tested their luminescence using a plate reader.

The test was performed using two different concentrations of the substrate, coelenterazine-h(benzyl coelenterazine): 1μM and 5μM so that we could determine the optimal concentration we would use for the next assays. The plate started to be read directly after addition of the substrate in the dark and measures were taken every 56 seconds thereafter. Three wells for each sample were measured and below are shown the graphs of their average luminescence plotted over time. We observed a signal-to-noise ratio of 4 at 1μM of substrate and 10 at 5μM for the PBS2-rLuc strain. The HOG1-rLuc strain seemed problematic and died within the plate. Unable to verify the correct transformation, we reproduced the strain but were unable to confirm its validity before the wiki deadline.

There is a clear luminescence generated by the cells, providing us with confirmation that our designs are correct and we determined which concentration of substrate to use to have a decent signal-to-background ratio.

The split-sfGFP strain stress-response

Having confirmed that our design for yeast transformations was correct via the previous experiment, we performed stress-response tests upon our PBS2-HOG1-splitGFP (BBa_K1486029) and (BBa_K1486035) strain to determine whether the pathway is reactive to stress as theorized. A co-transformation was performed using linear fragments produced in the same way as for the first experiment to obtain the strain. The N-terminal split of the superfolder GFP was tagged onto PBS2 and the C-terminal onto HOG1. The strain was then tested along with non-transformed cells in a 96 well plate with various stresses (e.g. Acetic Acid, Ethanol, Glucose...). The cells were centrifuged and resuspended in PBS before loading into the plate.

The most reactive of stresses turned out to be Acetic Acid 3.6% (shown above). Ethanol 10% was also very efficient. Other stresses tested did not seem to show conclusive results and we were unable to determine more reactants before the wiki freeze deadline. To further assess the fluorescence 10μl of our cells were spread on a glass slide and imaged with a x63 objective and a Green LP filter (Excitation BP 450-490, Dichroïc FT 510, Emission LP 515). Below is shown a comparison of before and after 3.6% Acetic Acid stress in the microscopy image merged with the fluorescence.

For a more quantitative measure, the fluorescent cells to total cells ratio was calculated and illustrated below.

From the data shown above, we can see that fluorescence increases within cells in our split sfGFP strain in response to stress while it does not in non-transformed cells. We can thus conclude that the orientation of the split sfGFP constructs added to PBS2 and HOG1 allows proper re-assembly upon interaction of the two proteins. This result shows promise for the actual application of the project as a Biopad using yeast cells. It is the first time a split gets succesfully implemented in the HOG1 pathway by direct attachment to proteins of the kinase cascade. Detection of protein–protein interactions within the pathway were previously studied using vesicle targeting.

The split-Luciferase strain stress-response

After determining that the pathway is reactive to certain stresses and that the split-GFP strain increases its fluorescence, we performed the same test in the obtained split-luciferase (BBa_K1486030) and (BBa_K1486036) strain. Unfortunately, due to the difficult nature of detecting an instantaneous luminescence signal and the unstable nature of coelenterazine-Luciferase reaction in different conditions (inhibition due to various molecules such as glucose), our attempts in detecting it on the plate reader failed. Further tests are planned, including an experiment in the microfluidic chips, to directly measure the effect of mechanical pressure, but will not be done before the wiki freeze.

Microfluidic achievements

MITOMI MITOMI modified SmashColi BioPad FilterColi CleanColi
Full chip

Unit Cell

Mold fabrication
Fabrication of the chip

To start our microfluidic experiments, we used the MITOMI chip that was designed in the laboratory of Prof. Maerkl.

Culturing E. coli with constitutive GFP on chip

We loaded E. coli, which contained constitutive GFP, in the chip. By using LabVIEW, a protocol was launched overnight to ensure the growth of the cells (the protocol can be found here).

The next morning, a scan of the chip was done to see the intensity of the GFP in the chip.

CpxR linked with GFP on the N terminal, induced by arabinose in E. coli

The experiment that was done on wetbench to show that CpxR linked with GFP was expressed with an arabinose promoter was replicated on a MITOMI chip. LB medium with arabinose was flowed in the upper half whereas LB medium without arabinose was flowed in the lower half. We scanned every hour for 5h (to know how it was done click here).

Figure 1. Scan of the microfluidic chip at t = 0min. No signal is detected

Figure 2. Scan of the microfluidic chip at t = 300min.

We analysed the scans and obtained the following results.

Figure 3.  Evolution of CpxR-GFP fluorescence over time

On chip infrared detection

As we focused part of our work on the IFP1.4, we needed confirmation of this signal detection in our microfluidic chips. Thus the aim of this experiment was to prove this fluorescence detection capability. Bacteria were loaded in the smash-coli chip. The first batch was KCl stressed and the second batch was unstressed. We then simply scanned the chip and analysed the results using ImageJ. For more details, please visit our notebook.

Figure 1. Cy5 scan of a chamber containing non-stressed CpxR-IFP bacteria.

Figure 2. Cy5 scan of a chamber containing KCl-stressed CpxR-IFP bacteria.

We analysed the scans and obtained the following results.

Figure 3.  Histogram of KCl stressed cells and non-stressed cells.

BioPad Detector: Detection of sfGFP

To get a first idea of how our BioPad was able to detect signal, we attempted to detect sfGFP signal emission by exciting our sfGFP with a 470 nm emitting LED and acquiring the images through our device. A sample of the video taken can be seen here below:

To learn more about how the detector works check out our Hardware page!