Light-Controlled KillerRed Concentration

The characterization of KillerRed showed that this protein can be used to control the number of living cells in a liquid culture using light. Indeed, we demonstrated,using (BBa_K1141001), that bacteria expressing KillerRed can grow in the dark and are killed when illuminated. In addition, we showed that the number of viable bacteria can be stabilized at different values, using different light intensity functions predicted by our mathematical model.

Our next goal was to find a way to fully automate the control of living cell density. The need to introduce IPTG into the culture was a problem that prevented our system from being fully autonomous. Automated addition of chemicals in the culture could have been a solution, but would have required using a micro pump, controlled via a computer. This approach also raised technical issues, such as the need for a reservoir containing an IPTG solution, and was consequently dropped. Using a constitutive promoter to trigger KillerRed expression inside the cells was also initially considered. However, our KillerRed characterization showed that KillerRed levels had to be high enough to enable cell killing upon illumination, but had to stay below a threshold value due to its intrinsic cytotoxicity. Thus, we decided to stick to an inducible KillerRed expression system.

Since we were already using light for cell killing, we looked for a way to kill two birds with one stone and control KillerRed expression with light as well. Indeed, in this approach, cell-machine communication could be mediated with light only and make our device much simpler, all the while its full utility. One important consideration was the wavelength at which the expression of the KillerRed gene had to be induced: indeed, to produce KillerRed without triggering its photoactivation, we had to avoid using a sensor that responded to green light. Thus we looked for a red light-inducible gene expression system. Researching the literature led to an optogenetic system that had been widely used during previous editions of the iGEM competition: the Cph8/PCB/OmpC/pompC red light-sensitive transcription system [1].

Voigt System

The Cph8/PCB/OmpC/pompC red-sensitive gene expression system was designed in the Voigt lab in 2010 (University of San Francisco, CA, USA). It is based on 2 switchable cyanobacterial phytochromes, named CcaS and Cph8. CcaS corresponds to a green light sensor and can be activated at 535 nm or deactivated at 672 nm. Cph8 corresponds to a red light sensor and can be activated at 705 nm or deactivated at 650 nm. These features allow control of the expression of two genes at different wavelengths. We figured that we could trigger KillerRed expression using the red sensor and KillerRed degradation using the green sensor, thus enabling us to fine-tune the concentration of KillerRed in E. coli.

Voigt designed and constructed three plasmids to implement his optogenetic gene expression control system:

Figure 1.
Schematic representation of the engineered two-color light induction system. [1]

The red light sensing protein Cph8, actually a Cph1-EnvZ fusion protein, is expressed from the pLTetO-1 promoter on pJT122 in its phosphorylated basal state. When illuminated, energy is transferred from Cph1 to EnvZ which allows it to phosphorylate intrinsic OmpR. OmpR then binds to and activates transcription at pompC. Because it is inactivated by red light, Cph8 can be considered a logical NOT red sensor. A genetic inverter, or logical NOT gate is used to invert the response of Cph8 to that of a red light sensor: when pompC is activated, cl is expressed, which represses pλ. pλ is linked to the red light-sensitive system’s output (here LacZ), thus creating a red light sensor by linking the output from a NOT red light sensor to that of a genetic NOT gate. The green light sensing protein CcaS is expressed in the unphosphorylated basal state. Under green light CcaS is phosphorylated, pcpcG2 is activated and the corresponding protein (here, LacZ) is expressed.

The Red Light Sensor Induces KillerRed Expression

Genetic Construct

Our idea was to substitute the initial LacZ output of the red sensor with KillerRed thus enabling a fully automated light controlled system without the need for added chemicals, like IPTG. We therefore replaced LacZ from pλ by KillerRed.

At the same time, we engineered a control and replaced LacZ also by mRFP [2], a non phototoxic red fluorescent protein. Both constructs BBa_K1141005 (KillerRed) and BBa_K1141004 (RFP) are described in our Biobrick sheets.

Using this new construct in the Voigt system, KillerRed is expressed at 650 nm when the NOT gate is disabled . We can then apply white light to induce ROS production by KillerRed and kill bacteria (Fig 2.)

Figure 2.
Schematic representation of the engineered red light-induced KillerRed expression system in E. coli. With red light (650 nm) the NOT GATE is inactived and KillerRed is expressed. Upon white light expressed KillerRed produces ROS, which damage molecules in the cell.

Characterization by Kinetics

Choice of the E. coli strain

BW25113 E. coli bacteria, a double knock-out EnvZ^-/OmpC^- mutant was chosen for characterizing the red sensor controlled KillerRed expression system. This is because endogenous EnvZ in E. coli phosphorylates the OmpR regulator and thus interferes with Cph8 which is engineered as a Cph1-EnvZ fusion protein [3]. With this system, the presence of endogenous EnvZ would then lead to continuous inhibition of KillerRed expression. Additionally, OmpC is an endogenous porin whose expression is naturally regulated via the pOmpC promoter, present in the red sensor to trigger inhibition (via the NOT gate) of the expression of the KillerRed gene.

Experimental setup

Experimental conditions

After obtaining BW25113 colonies from co-transforming plasmids pJT106b (KillerRed) or BBa_K1141005, pJT122 and pPLPCB (S), we grow cells in LB with standard antibiotic concentrations overnight. We observe that these cells grow very poorly with three different antibiotics (chloramphenicol, ampicilline and streptomycin, one for each plasmid), with lag phases exceeding 24 hours.

From the LB culture, we attempt inoculating cells in M9 minimal medium, at 1X antibiotic concentrations to follow kinetic experiments (this was selected previously as the best medium for kinetics). The cell growth in this medium was, however, extremely slow (OD610=0,7 was reached after 3 days).

A second attempt is made at growing the BW25113 transformants, this time with varying concentrations of antibiotic: a culture is made at the normal concentration (1X), one at half normal concentration (0.5X) and one with no antibiotics. Without antibiotics the cells grow quickly (OD610>2.5 in less than 24 hours). At 0.5X antibiotics, more than 36 hours are necessary to obtain OD610>2 and more than 48 hours are needed to obtain OD610>2 with cells in 1X antibiotic M9 growth medium.

To verify if our cells have kept their plasmids despite the lack of antibiotic selection, we plated them on agar plates with all three antibiotics at 1X concentration. We observed only 6 colonies after 48 hours of incubation with no prior dilution of the saturated culture. We conclude that the cells do not keep the plasmids when cultured without antibiotics. M9 medium seems to be inadapted for the culture of these cells, and we decide to attempt the kinetics experiments in LB medium.

Kinetics in LB without antibiotics

We attempted a kinetic experiment in LB without antibiotics using an initial OD610 at 0.01. The light source (P = 0.03µW/cm2) for red illumination is insured by a Wratten gelatin 26 filter (red). The maximum passing wavelength of the filter is 620 nm. Our reasoning was that faster LB growth might allow cell culture without too much plasmid loss. Like expected our cells grew quickly and reached the stationary phase after 9h. We were able to measure KillerRed fluorescence in the stationary phase (> 22 hours) of this culture indicating that some cells must have kept their plasmids even without selection pressure. However similar KillerRed fluorescence levels were observed when the culture was incubated in the dark, indicating a leaky genetic system (Fig 3.)

Figure 3.
OD610 (A) and Fluorescence (B) responses of a culture exposed to a constant red light illumination (light grey) or to darkness (dark grey).

Kinetics in LB with antibiotics

A last attempt at doing the kinetics experiment with BW25113 triple transformants was made, this time with all three antibiotics at 1X and higher starting OD610 at 0.05. Our results indicate that it takes more than 30 hours for bacteria to start growing (Fig.4A). At 30 hours we measure no KillerRed fluorescence whereas at 48 hours we have significant KillerRed fluorescence in both illuminated and non-illuminated cultures (Fig. 4B), which points again towards a leaky genetic expression system.

Figure 4.
OD610 (A) and Fluorescence (B) responses of a culture exposed to a constant red light illumination (light grey) or to darkness (dark grey).

To conclude, the BW25113 triple transformants seem to take an abnormally long time to grow in both LB and M9 media with antibiotics. If the concentration in antibiotics is lowered, growth is faster but bacteria loose their plasmids. One hypothesis is that we loose the pJT122 plasmid containing the cph8 gene. It would then be impossible to induce KillerRed production with red light. Moreover the loss of this plasmid would explain why the bacteria take so much time growing in the presence of chloramphenicol. Since it is a bacteriostatic antibiotic, it would only keep the bacteria from growing until they have found a way to work around its effects. In the stationary phase there is a significant amount of KillerRed expression, which is probably due to a leak in the genetic network. To prove this we could try to block KillerRed expression by exposing our bacteria with far red light and thus switching on the NOT gate. If, under far red light bacteria still fluoresce, then the genetic network is leaky and the observed KillerRed fluorescence is not triggered by the red sensor.

Our cultures are behind a Wratten gelatin 26 filter (red) with a maximum passing wavelength at 620 nm. Since the red sensor’s optimal stimulation wavelength is 650nm this isn’t optimal for dephosphorylation of the sensor Cph8. However it should be sufficiently below the phosphorylation wavelength of the sensor which is at 705 nm.

Green light sensor to induce KillerRed degradation

ssrA et SspB

Degradation of ssrA-tagged proteins is a central feature of protein-quality control in all bacteria [4]. Whenever E. coli ribosomes stall during translation, the tmRNA or ssrA ribosome-rescue system mediates addition of the sequence AANDENYALLAA to the C-terminus of the nascent unfinished polypeptide. This peptide sequence, called the ssrA tag, targets the modified protein for degradation, assuring protein-quality control by preventing the accumulation of aberrant, unfinished proteins. Indeed, the SspB adaptor protein enhances degradation of ssrA-tagged proteins by tethering them to the ClpXP protease (Fig. 5) [5]. ClpX binds substrate sequences known as degradation tags, unfolds the attached protein, and translocates the denatured polypeptide into ClpP for degradation. Cloned proteins with C-terminal ssrA tags are therefore rapidly degraded in the cell.

Figure 5.
SspB binds ssrA-tagged substrates and ClpXP, forming a degradation delivery complex [5].

Construct

We imagined to control KillerRed degradation by using the ssrA/SsrB system and linking it up to the green light sensing protein CcaS from the Voigt system (Fig.1). Under green light (535 nm) SspB is produced which could then rapidly decrease intracellular KillerRed concentration (Fig. 6). To do that we will replace lacZ from pcpcG2 by SspB-ssrA on pJT122. For the moment SspB-ssrA is on the plasmid pBAD described in our Biobrick sheets (BBa_K1141006). To characterize the green sensor we replaced lacZ from pcpcG2 by GFP on pJT122 described in our Biobrick sheets (BBa_K1141003).

Figure 6.
Schematic representation of the engineered green light induction system. Under green light (535 nm) CcaS is phosphorylated, and in turn phosphorylates CcaR, which then binds to and activates transcription from pcpcG2. SspB is produced and binds to ssrA-tagged KillerRed which is then readily degraded. SspB itself is also ssrA-tagged to ensure tightness of the system.

With this construction, we hope to precisely control KillerRed expression. For example, within a specific light intensity and during desired time, we can decrease the concentration of cells to a value determined by the device. This allows for a simple control of the cell population, and is due to the concentration of KillerRed determining the speed at which cells are killed.

Another important point deals with the application of green light on KillerRed within our cells. Green light was able to kill cells within our experimentation, but Y. ROULLAND, who originally sent us KillerRed, attempted to illuminate his cells with green light with little success in destroying the cells in his research. However, white light works quite effectively to kill cells and in the litterature the Voigt green sensor seems to be very sensible. That’s why we chose to use this construction for our experimental model. .

References

[1] J.J. Tabor et al., Multichromatic Control of Gene Expression in Escherichia coli, Journal of Molecular Biology, 2011.
[2] E.C. Robert et al., A monomeric red fluorescent protein, PNAS, June 2002.
[3] https://2011.igem.org/Team:TU_Munich/project/introduction
[4] C. Farrell et al., Cytoplasmic degradation of ssrA-tagged proteins, Molecular Microbiology, 2005.
[5] K. McGinness et al., Engineering Controllable Protein Degradation, Molecular Cell, June 2006.