Where it started...

Upon a suggestion by Professor Maerkl, we decided to take a look at the High Osmolarity Glycerol(HOG) pathway in S. cerevisiae and see whether we could expand our project into yeast cells. Similarly to the CpxR pathway, the HOG pathway is not only responsive to hyper-osmotic stress but to also heat shock, nitrogen stress and oxidative stress.¹. Yeast cells use this MAPK (Mitogen Activated Protein Kinase) pathway to coordinate intracellular activities and to optimise survival and proliferation. It is activated mainly by two membrane proteins Sho1 and Sln1, of which Sho1's role is thought to be monitoring turgor during cell growth². Ste11, a MAPKKK, then phosphorylates Pbs2 which in turn phosphorylates the MAPK Hog1.

From this we hypothesized that by tagging downstream proteins of the pathway, Pbs2 and Hog1, with split constructs we would be able to create a touch responsive system. The two proteins' copy number is well known: 2160 and 6780 molecules/cell³ respectively for Pbs2 and Hog1.

Choosing to work with yeast

In synthetic biology, the most used organism is undoubtedly E. coli. The easy nature of its manipulation makes it ideal especially when the time for the project is limited to a few months. However, studies in bacteria are not always applicable to all cells, and an extension to a eukaryotic species adds another dimension to our project. Especially in the case of the HOG pathway, there is a very similar pair of proteins in mammalian cells, MEK5 and ERK5, meaning that the project could be extended even further⁴. Also, thinking of the safety of the 'biopad', a leak of modified yeast cells will carry less risk than bacterial cells due to the different nature of the cells and transformation methods.

Using S. cerevisiae and homologous recombination, we could also address other problems. In bacteria, inserting plasmids can disrupt the equilibrium of the pathway and activate it or deactivate it constitutively because the copy number will be changed. Also, the bacterial project consisted in attaching a split construct to a homodimerising protein there would be many unspecific interactions which would not produce the desired signal.

However, there weren't only advantages in choosing to work with yeast. First, we had no idea how to work with yeast cells and neither did our assistants; we had to do a lot of research on basic protocols. Moreover, we could not find a study which had already tested the interaction between two proteins in the pathway using split constructs so we were making a small gamble. And last but not least, working with yeasts takes a lot longer due to the slower doubling time.

Why did we choose Hog1 and Pbs2?

As mentioned above, we were unable to find a set of proteins which had already been assayed by split complementation. This meant that we would not be sure whether the split constructs would be able to interact once fused to the proteins and which orientation to use. We chose PBS2 because this is where the stress signal converges in the pathway from different membrane sensory proteins. And we chose HOG1 as the other protein because it is phosphorylated directly by PBS2 while there is a scaffold for the upstream interactions. For the orientation, there was no doubt that we had to use the C-terminal for PBS2 because the kinase domain interacting with HOG1 is on the C-terminal and the N-terminus is physically very far. We would have liked to test both orientations on HOG1 but due to limitation of time and people to invest on this part, we decided on the C-terminus. This choice was backed by the fact that kinases have their regulation tail on the N-terminus and we did not want to risk disrupting its activation.

Learning How to Handle Yeast: a Challenge

Having had several courses where yeast had been mentioned, we initially thought that we would not have too much trouble understanding how to take on the project. Little did we realise how much we were wrong. It took a whole month to fully understand what we had to do in terms of the design of the gene insert and order the plasmids we needed while we prepared everything we needed. Not knowing that the number of base pairs needed for homologous recombination varies between different species of yeast confused us, and then we were confused whether we should transform with a plasmid or linear fragments. In the end, we found plasmids specifically designed for GFP-tagging in yeast on AddGene and we consulted the professor Simanis and his assistant Andrea Krapp ([http://simanis-lab.epfl.ch/]) who helped us confirm our design.

How did we make our DNA constructs and transform our cells?

After the long discussions, the plasmids we found were the plasmids pFA6a-link-yoSuperfolderGFP-Kan (44901) and pFA6a-link-yoSuperfolderGFP-Ura (44873)⁵ which are for GFP tagging via homologous recombination in yeast. They differ by the yeast selection marker they encode: Kan or Ura. We did all the constructs linked to Pbs2 with the kan resistance gene while Hog1 related constructs were coupled to the Ura3 selection marker.

We replaced the GFP tag by using overlap PCR and created split Renilla Luciferase, split GFP and whole Renilla luciferase tags to finally produce linear fragments of the form below. The figure below illustrates how the modified genome would look like and what proteins it would express (we took the example of the GFP tagged Hog1.

It took us many PCRs to produce the 8 different linear fragments we wished to use and after a failed transformation, we hypothesized that it was due to the bad quality of the products and took our time to optimize the reactions. We preferred to do sequential transformations but, due to a lack of time, we also ended up trying the co-transformation while waiting for the verification of the first successful transformations.

Our strains and DNA constructs

We use the S.cerevisiae strain YM4271. This strain has several inactivated genes, the main ones being Ura3, Leu2, His3 and Trp1. To select transformant colonies, one must include a selection marker. For example, we incorporated the gene Ura3 in our integration vectors, to allow the transformed cells to grow on uracil lacking medium. This strain also has a mutation in the Ade2 gene, so it tends to accumulate red fluorescent metabolites if grown in medium lacking adenine sulfate

We were able to get all our transformants in time however we didn't finish the characterization of all our constructs before the wiki freeze. Tagging Pbs2 with both the yeast sfGFP and the Renilla luciferase was successful, proving that we managed to do homologous recombination in S. Cerevisiae and that our gene of interest was expressed in the cells. Moreover we were able to demonstrate that the split sfGFP we designed worked as expected : in response to an osmotic stress, when Pbs2 links and phosphorylates Hog1, the full sfGFP re-assembles and fluorescence is emitted!
It means that the BioPad can be implemented with yeast cells and fulfill the various functions we have been thinking of. The sfGFP can of course be replaced by another fluorescent protein such as the IFP.

In the table below, the eight DNA parts and the six corresponding yeast strains we aimed to obtain can be found.

	DNA sequence(s) to insert by homologous recombination	Transformation?	Characterisation?
HOG1-sfGFP		Done	Not completed
HOG1-RLuc		Done	Not completed
PBS2-sfGFP		Done	Characterised
PBS2-RLuc		Done	Characterised
Split sfGFP		Done	Characterised
Split RLuc		Done	Not completed

References

1. The HOG pathway controls osmotic regulation of transcription via the stress response element (STRE) of the S. cerevisiae CTT1 gene; Shuller and al (EMBO journal, 1994)
2. Osmotic Stress Signaling and Osmoadaptation in Yeasts; Stefan Hohmann (Microbiol Mo Biol Rev. 2002)
3. Yeast GFP library; Ghaemmaghami S, et al. (2003)
4. Scaffold proteins of MAP-kinase modules; D N Dhanasekaran et al (2007)
5. Improved Blue, Green, and Red Fluorescent Protein Tagging Vectors for S. cerevisiae. Lee et al (PLoS One. 2013 July 2)