The aim of our 2011 iGEM project is to create an E. coli kill switch using intracellular antimicrobial peptide (AMP) production (K628006).

Structure

The pBAD promoter is found in nature governing the E. coli arabinose operon, responsible for the breakdown of the sugar arabinose into D-xylulose-5-phosphate. This promoter is induced by the binding of L-arabinose to the AraC promoter region (further details on promoter structure and regulation can be found on our Modelling page). As a faster rate of protegrin-1 production would benefit our project by allowing for a wider range of potential applications, we chose to use the pBAD strong promoter (K206000), which is a mutagenized form of pBAD that induces transcription at a lower arabinose concentration and has a higher maximum expression. pBAD strong is only induced in the presence of arabinose, which is not naturally produced by the cell, making it a very stable promoter to work with and greatly reducing the chance of accidental gene activation.

The coding region of our kill switch (K628000) is responsible for synthesizing its functional aspect, the antimicrobial peptide protegrin-1.

Our ribosome binding site (RBS) is classified as a “strong” binding site (J61101), meaning that the ribosome will very readily bind to the mRNA sequence, allowing for translation to occur quickly after transcription has finished. We chose this RBS in order to help promote the rapid production of protegrin-1. Our terminator is a double terminator (B0015), standard in iGEM to decrease the chance of over-transcription by DNA polymerase.

How does it work?

Protegrin-1 is an antimicrobial peptide (AMP) first derived from porcine leukocytes. These peptides are part of the innate immune system and function by attacking the membranes and intracellular processes of invading bacteria. Protegrin-1 functions to protect the body against non-host cells by integrating itself into the phospholipid bilayer of prokaryotic bacteria, destabilizing the membrane and causing pore formation (Lam, 2006). These pores inhibit the cell’s ability to control transmembrane ion and water movement, resulting in cell death via either osmosis or cytosol loss.

The idea behind our kill switch would be to induce bacteria to produce AMPs intracellularly, and allow these peptides to integrate into the membrane. As the concentration of AMPs builds, pores will form in the membrane, inevitably leading to cell death.

Antimicrobial peptides are highly conserved within evolution and can be found throughout all classes of life. This is in part due to the fact that mutating to resist AMPs is near impossible. These proteins attack bacterial membranes, meaning that multiple chance non-detrimental mutations across multiple genes controlling phospholipid bilayer structure would be required in order to resist AMP interactions. The chances of these mutations occurring within a single replication are astronomically small. This means that the use of protegrin-1 as the functional aspect of our kill switch will never lead to the creation of an AMP-resistant strain of E. coli.

Concerns were raised about the effectiveness of intracellular AMP production. In nature, protegrin-1 attacks bacterial membranes from the outside by burrowing a hole into the outer leaf of the phospholipid bilayer. We were unsure of protegrin-1's ability to perform this action from the inside of the cell; specifically, whether or not it would interact with the inner sheet of phospholipids. Little research has been conducted on intracellular AMP production, so with no directly available answer, we decided to explore for ourselves. All the material we found provided no strong evidence that there was much structural difference between the two layers of prokaryotic membrane. Both layers have a mixture of zwitterionic and acidic phospholipids, the latter of which contain a negative charge that facilitates the electrostatic interactions between the AMP and the bilayer. Eukaryotes have almost no acidic phospholipids on their outer layer, and imbedded in between the lipids is cholesterol, which increases the stability of the phospholipids and impedes hydrophobic AMP interaction (in conjunction, these two traits provide the protection that stop AMPs from attacking host cells). So, as long as there are a sufficient concentration of acidic phospholipids to facilitate the interaction, the AMPs should be able to bind to either leaflet. We also contacted resident biomolecular scientist Dr. Peter Coote, who has authored several papers on AMPs, to provide advice on this subject. He confirmed our theory that acidic phospholipids were imperative for the interaction between bilayer and peptide to occur, and agreed that intracellular protegrin-1 production should, at least hypothetically, cause cell death.

Dr. Coote also provided us with gomesin, an AMP structurally analogous to protegrin-1, for use in testing our kill switch theory.

Experiment Planning

We devised two sets of proof-of-concept experiments: one to ensure that our kill switch was functioning, and the other to test how it would work within a drug delivery context.

As our kill switch functions by killing cells, an obstacle we came across was how exactly to measure for cell death. Our first attempt was devised with inspiration from the 2010 Berkeley iGEM team (Berkeley, 2010), where they used light absorbance as a proxy for cell death, with the idea that an intact cell will absorb more light than one that has burst due membrane disruption. An increase in optical density would correlate to cell growth, while a decrease would correlate to cell death. We also acquired gomesin, an antimicrobial peptide with an analogous structure to protegrin-1. If we could add gomesin extracellularly in one population, activate our protegrin-1 kill switch intracellularly, and show that both E. coli populations were dying in a similar, if not identical manner, we could provide that as evidence of our kill switch functionality. Weeks 6-9 were spent testing varieties of the protocol below.

pBAD Strong/Protegrin-1 Proof-of-Concept Experiment Protocol

Unfortunately, what we had not foreseen was the digestion of arabinose by the naturally occurring arabinose operon, providing energy for the cells to grow. These newly grown cells replaced the cells lost to protegrin-1 production and our absorbance kept increasing, masking the cell death we hoped to see via kill switch activation. We tried to integrate this into the experiment by measuring the rates of growth at various arabinose concentrations compared to a no-arabinose control, but we decided that the data was not sufficient to show kill switch activity. In order to reduce growth, in subsequent experiments we replaced the luria broth with sterile water and incubated the E. coli at room temperature rather than 37 degrees, but even with these changes, we were unable to find useful results.

In Week 10, with the iGEM Jamboree looming, we decided to view our E. coli under a light microscope in the hopes to watch our kill switch functioning. This inspired the idea to use fluorescent dyes to distinguish between live and dead cells, and then measure fluorescence in the same plate reader we used for measuring absorbance. We purchased the LIVE/DEAD Baclight Bacterial Viability kit, which functions via a mixture of two dyes. The first, SYTO 9, is absorbed by all cells regardless of whether they are alive or dead, and fluoresces green (at 530 nm) when excited (with 485 nm). The second, propidium iodine, is only absorbed by cells with damaged membranes. Propidium iodine competes with SYTO 9, fluorescing red (at 630 nm) when excited (with 485 nm) and reducing SYTO 9's fluorescence within that cell. Further details of how the LIVE/DEAD Baclight Bacterial Viability kit works can be found here. We integrated our characterization experiment into this experiment's protocol, which can be found below.

BBa_K628006 Characterization Experiment Protocol

We also devised a protocol to test the potential drug release of our kill switch. We hoped that we could synthesize a DNA device that would create a drug in vivo that would be released along with the cytosol upon membrane pore formation (i.e. kill switch induction). The drug would be a protein created by a synthetic gene placed on either the same plasmid as, or alongside the plasmid containing, our kill switch. This gene could be governed by either a constitutive promoter, allowing a slow, increasing concentration of the drug within the cell, or by an inducible promoter, which would only induce drug production when stimulated by an outside source. Because iGEM requires proof-of-concept experiments (and we lacked the knowledge required to synthesize and intracellular produce a real drug), we theorized the "drug" as a GFP molecule. GFP is too large (26.9 kDa) to be exported naturally from the cell, and therefore served as both a molecule of the right size (a drug that could be naturally exported would not require a kill switch to be delivered) and one that would be easily viewable and quantitatively assessed.

Not only that, but we fully intended to attach a his-tag to our final GFP molecule by adding the appropriate DNA into the biobrick sequence. This was to be completed via PCR-ing the Registry part I20260 using customized primers that contained the his-tag DNA sequence. However, this posed a problem, as the act of PCR-ing halfway into the I20260 biobrick remove the terminator in the process. We would solve this problem by performing a single-enzyme digestion of part B0015 and subsequently ligating the entire J23101 + B00032 + E0040 + his-tag sequence into it. Once transformed, this would give us a new biobrick that would serve as the subject in our experimental protocol below.

Drug Delivery Proof-of-Concept Experiment Protocol

Unfortunately, due to the constraints of the 10 week iGEM period, we were unable to proceed past the PCR stage of our drug delivery biobrick creation. With the protocol drawn up, primers in our freezer, and cells in glycerol stocks, we plan to continue with this effort before the International Jamboree in November.

Experiment Results and Characterization

Success!

Figure 1 - This graph shows fluorescence readings at 600 nm (485 nm exciting) of K628006-transformed E. coli in solutions of various arabinose concentrations.

The above graph measures the fluorescence readings of E. coli transformed with K628006 across multiple arabinose concentrations. The y-axis represents the scale of the fluorescent output of each E. coli population, while the x-axis denotes the time of each reading. The graph doesn't look very aesthetically pleasing, and in fact, may lead some to believe that our readings of fluorescence decreased over time, and subsequently, so did cell death.

However, this is due to the fact that the dye used to stain the cells reduced in fluorescing power over time, resulting in what looked like decreasing cell death. In order to show the relationship between the effects of various arabinose concentrations on fluorescence, we decided to subtract the fluorescent value of our control (E. coli transformed with K628006 in a 0 uM solution) from the rest of our values at every reading. This would remove the focus from the decreasing values, and instead place emphasis on the difference between individual arabinose concentrations. What we find is a much more interesting graph:

Figure 2 - This graph shows the fluorescence readings at 600 nm (485 exciting) of K628006-transformed E. coli in solutions of various arabinose concentrations. The value of the 0 uM control has been subtracted from the fluorescence values of each other arabinose concentration at every 5 minute reading, in order to show the scale of difference between each reading.

This graph makes the relationship between arabinose and K628006 much more clear. As we can see from the graph, lower concentrations of arabinose cause less cell death than higher ones. The presence of arabinose within the same solution as our transformed E. coli triggered the pBAD strong promoter to induce transcription of the protegrin-1 coding region. As the concentration of protegrin-1 protein rose within the cells, these antimicrobial peptides integrated into the phospholipid bilayer, eventually causing membrane destabilization and pore formation. This allowed the propodium iodine dye to enter the cells, fluorescing near the 630 nm wavelength (we were forced to measure at 600 nm due to the constraints of our microplate reader). As we can see from the graph, higher levels of arabinose induced more protegrin-1 production, resulting in higher fluorescence values.

Alongside this process, arabinose would also be broken down into ribulose-5-phosphate epimerase by the 'ara operon', naturally present within the DNA of the E. coli and the original site of the pBAD promoter. This would cause the concentration of arabinose in each well to decrease slowly over time, the rate of which may differ between various strains of E. coli. Other groups looking to repeat this experiment would need to keep this fact in mind. Additional information on this process can be found on our Modelling page.

The aim of a "characterization experiment" is to gather more data about the biobrick in question. Submitting a biobrick that has only been tested under a single condition does not reveal the full range of potential uses for that part. Not only that, but it forces other groups who wish to use that biobrick to waste valuable time further exploring a part that should have been characterized in the first place. With the number of parts being submitted increasing every year, there has been a noticeable lack of characterization in the Registry. It is important for the biobricks within the Registry to be tested under multiple conditions in order for the Registry to function as intended. We have included multiple arabinose concentrations at consecutive orders to magnitude in order to show the total cell death that each concentration will induce, as well as the rate at which that cell death occurs.

The relationship between the concentration of arabinose and the amount and rate of cell death seems linear in nature. We tested K628006 at multiple arabinose concentrations to see which would best induce activation of the pBAD strong promoter. While it might seem that the 7400 μM arabinose concentration resulted in the strongest kill switch response, the cell death of these readings is not due entirely to our kill switch. We discovered that arabinose in high concentrations would kill cells via non-kill switch related means. This was a problem that plagued the results of our original experiments, which were run with arabinose concentrations several orders of magnitude higher than they should have been. While we did not have time to identify how exactly excess arabinose caused cell death, we can speculate that there was probably an osmosis-based effect, by which the cells were placed in a hypertonic solution that interfered with various other cell regulatory functions. From the graph below, we can see that at concentrations of 7400 μM, supercompetent, non-transformed E. coli were fluorescing considerably higher levels than at 0 μM arabinose concentrations, indicating not only an increase in cell death, but that the arabinose was the cause:

Figure 3 - This graph shows the fluorescence readings at 600 nm (485 exciting) of non-transformed E. coli in solutions of 7400 μM and 0 μM arabinose.

Figure 4 - This graph shows the fluorescence readings at 600 nm (485 exciting) of non-transformed E. coli in a solution of 7400 μM arabinose concentrations. The value of the 0 μM control has been subtracted from the fluorescence value of the 7400 μM arabinose concentration at every 5 minute reading, in order to show the scale of difference between the two concentrations' cell death.

References:

Lam et al. “Mechanism of Supported Membrane Disruption by Antimicrobial Peptide Protegrin-1”. The Journal of Physical Chemistry B, Vol. 110, pg. 21282 - 21286. Published 2006. Link to paper.

Berekely. "Self-Lysis." Link to paper.

Mandard, Nicolas et al. "The Solution Structure of Gomesin, an Antimicrobial Cysteine-Rich Peptide From the Spider." European Journal of Biochemistry, Vol. 269, Issue 4, pg. 1190-1198. Published February 2002. Link to paper.