Characterisation of the RsmA post-transcriptional regulation system components

Why should we include a translational regulation in our circuit?

The toggle we developed will switch to one or the other phenotype depending on the amount of each of the repressor proteins. When no pollutant is in the sample, the cells are receivers because the IPTG gradient on the plate induces CinR expression from the pLac promoter.

This initial setting makes the toggle harder to switch back to the sender phenotype. We therefore decided to consider and model two situations:

1. Repressors of our system can be kept to almost zero and then be rapidly induced by a translational regulation system. (RsmA-rsmY system)
2. There is no regulation of repressors and all cells transcribe TetR (MerR) before the pollutant is added to the plate. (without RsmA-rsmY system)

We modelled both the position of the switch on the plate and the visual output that result from those two situations.

Figure 1: Model of the cell switch to a stable state along a linear gradient of IPTG. The x-axis represents the position along an increasing IPTG gradient from left to right. The Y-axis indicates the final concentrations of LacI and TetR in the channels. The switch occurs at a lower concentration of IPTG when the initial amount of repressors is kept very low by a translational repression system (red curve). When the initial concentration of TetR within cells is higher, more IPTG is required for switching (blue curve). (Matlab was used for this simulation).

Figure 1 shows where the switch is expected to occur in a plate containing a given IPTG gradient and with an aTc concentration of 10−6 M. The red curve represents the position of the switch in the presence of a translational regulation system that initially keeps both repressors LacI and TetR at zero. The blue curve represents the position of the switch in the absence of a regulation system. An initial concentration of 5% of the maximum TetR (MerR) repressor concentration is considered in this example before sample addition.

We can see that the switch appears further to the left the more the initial amount of TetR in the cell is kept low. This means that we can potentially detect a lower concentration of pollutant (here aTc) if we have an efficient translation regulation system.

The visual output would also be very different depending on the initial concentration of repressors. If cells initially transcribe TetR and CinR before they switch to the other phenotype, the quorum sensing receptor will be present when they become senders. The result, as shown on figure 2, will be a plate with a visible signal wherever the cells have switched.

Figure 2: Representation of the visual ouput predicted if cells express CinR before switching to become senders and express CinI. (Matlab was used for this simulation).

These simulated predictions demonstrate that a translational control of TetR (MerR) and LacI would be useful in our device. It would allow a more sensitive and more accurate result. Wetherefore looked for a regulation mechanism that allows keeping protein translation off, until induced with a specific controllable signal. Two potential candidate systems exist:

1. The RpoS system of Escherichia coli: we extracted and cloned the leader sequence of rpoS.
2. The RsmA/rsmY system of Pseudomonas aeruginosa we cloned, and characterised several components of this system using engineered GFP reporter gene constructs.

Can rsmA be indeuced into E. coli?

The RsmA system from Pseudomonas aeruginosa has a homologue in Escherichia coli, named “CsrA”. We know that these two systems are extremely similar. Consequently we ask ourselves whether the synthesis of RsmA in E. coli interferes with its survival. Figure 3 shows growth curves of E. coli cells transformed by a plasmid containing an IPTG-inducible rsmA sequence from Pseudomonas aeruginosa and control cells carrying the same plasmid without rsmA. The superimposed curves demonstrate that the synthesis of RsmA does not interfere with the growth of E. coli.

Figure 3 Influence of RsmA production on the growth of E coli DH5α. Bacteria transformed with pVLT31-rsmA or empty pVLT31 from overnight culture were re-suspended in rich medium supplemented with tetracycline at 20mg/ml. Induction of rsmA transcription is induced by IPTG at the concentrations given in the legend. Curves are normalized to their first OD value.

Characterisation of the leader sequences

We cloned several leader sequences that contain a ribosome-binding site (RBS) in front of a reporter gene, GFP, in order to:

1. Characterise their RBS strength
2. Use them for the translational control of downstream genes by the RsmA/rsmY system.

Figure 4: :Constructs used to characterise the strength of the RBS in the leader sequences mag and fha and BBa_K256003 as a reference.

We used a FACSCalibur flow cytometer to measure the GFP fluorescence emitted by cells containing the constructs shown in Fig.4. Two negative controls were set up: a brick having the GFP reporter gene but no promoter ( BBa_E0840 ) and a cell culture containing no plasmid.

50 μl of LB containing cells were diluted into 500 μl of filtered PBS (OD600 of inoculum was 3± 0,3 for all samples) and then introduced into the FACS ten minutes after dilution.

The cytometer counts each particle that passes through the light beam. Therefore it is necessary to select an analysis window that corresponds to the size of the bacteria (see Figure5).

Figure 5: Individual flow cytometry results (at the right) for different constructs (at the left). The red number indicates an average of GFP fluorescence. The percentage in black indicates the fraction of the bacterial population within the respective analysis window M1 or M2.

The two negative controls (cells containing no plasmid (1) or plasmid without promoter (2)) show a basal fluorescence signal as expected. Cells containing the reference brick BBa_K25003 (5) show a maximum amount of GFP fluorescence with 77 % of the cell population that fluoresce more than the negative controls. The average fluorescent signal calculated for construct 5 is 1030 (vs 2,5 (neg control)). The fluorescence signal obtained with cells containing the brick with the mag leader sequence (3) does not differ very much from the negative controls (4,4 vs 2,5). Four per cent of this cell population fluoresces more than the negative control populations.

90 % of cells containing the fha leader sequence (4) present a fluorescence signal that is higher than control cells. Their average fluorescence is 98.

Figure 6: Compilation of one of the flow cytometer measurements. 1 and 2 are the negative controls, 3 : mag leader sequence, 4: fha leader sequence, 5 : reference biobrick. Number under each construct indicate mean fluorescence level of the global population. BBa_K256003 .

Figure 7 summarises one of the cytometer results for mag (in black) and fha (in red) leader sequences cloned upstream GFP reporter gene. We can see that both leader sequences mag and fha allow the translation of the GFP mRNA. They can therefore be used for further characterisation of the system. The GFP fluorescence signal is much higher using the fha leader sequence when compared to mag. Figure 7 focuses on the RBS strength of those two gene leader sequences, and compares them to the strongest RBS of the part registry: BBa_0034 .

Figure 7: Relative strength of mag (in black) and fha (in red) leader sequences compared to the strongest RBS of the registry. GFP fluorescence recordings are represented as means and their standard deviations from four independent measurements.

Effect of rsmA and rsmY on the mag and fha reporter genes

After having established the RBS strength of fha and mag leader sequences using the GFP reporter gene constructs, we quantified the translational inhibition effect of the RsmA protein and the relieve of this inhibition in presence of the rsmY RNA. In order to do these experiments, E. coli cells were co-transformed with a combination of 3 different plasmids containing:

1. A leader sequence or the reference RBS upstream of GFP (constructs 3, 4 and 5 on Fig.4 )
2. A leader sequence or RBS plus an other plasmid containing rsmA (figure 7)
3. A leader sequence or RBS plus rsmA plus rsmY (figure 7)

Figure 8: The IPTG-inducible rsmA gene is cloned in plasmid pSB3C5 and the Tet-inducible rsmY in plasmid pSB4K5

The GFP fluorescence obtained from the constructs with fha, mag and reference RBS were analysed in the absence of RsmA and rsmY (single transformants) in the presence of RsmA (double transformants) and in the presence of both RsmA and rsmY (triple transformants). Note that in our experiments rsmA and rsmY were constitutively expressed because the repressors LacI and TetR were absent from our strains.

As expected, no inhibitory effect of RsmA was observed when the reference RBS ( BBa_0034 ) was used as this leader sequence does not have a binding site for RsmA (data not shown).

In contrast, RsmA decreased the GFP fluorescence level when mag or fha leader sequences were provided upstream of GFP and this decrease could be partially relieved in the presence of rsmY for fha (Figs 9 and 10).

Effect of rsmA and rsmY on fha-GFP constructs

Figure 9: Compilation of flow cytometry measurements. The green curve shows the auto fluorescence of bacteria expressing no GFP. Red curves represent strains carrying fha-GFP alone (dashed line), fha-GFP and RsmA (continuous thin line) and fha-GFP and RsmA and rsmY (continuous bold line).

In order to compare the GFP levels, we extracted mean fluorescence values from the curves of figure 9, substracted the autofluorescence, and compared them to fha-GFP alone.

Double transformants carrying fha-GFP and an RsmA expressing plasmid (construct 4 in Fig 4 plus construct 6 in Fig 8) fluoresce about twenty times less than those with fha-GFP alone. The triple transformants expressing additionally rsmY (construct 7 in Fig 8) fluoresce at an intermediate level between that obtained for single and double transformants. Note that this intermediate fluorescence level was obtained only for 70% of this triple-transformed population, 30% showing background autofluorescence. The effect of rsmY could therefore be underestimated. The variance in fluorescence intensity in the triple transformant population does not allow us to conclude on the effective relief of translational inhibition by rsmY.

Effect of RsmA and rsmY on mag-GFP constructs

As for the fha leader sequence, we also did single, double and triple transformants combining the mag-GFP construct with the plasmids allowing expression of rsmA and rsmY, respectively. The GFP fluorescence data were recorded as previously described.

Figure 10: Compilation of flow cytometry measurements. The green curve shows the auto fluorescence of bacteria expressing no GFP. Black curves represent strains carrying mag-gfp alone (continuous thin line), mag-gfp and RsmA (dashed line) and mag-GFP and RsmA and rsmY (continuous bold line).

The FACS results obtained for the mag leader sequence analysis show that the presence of RsmA reduces the GFP fluorescence to background level (autofluorescence, green curve in Fig. 11). There is however no restoration of GFP fluorescence when providing rsmY in triple transformants.

Conclusion

The preliminary data presented above show that the RsmA system from Pseudomonas aeruginosa can be used for translational regulation in E. coli. Here we compare GFP fluorescence signals at a given time that were measured in different independent clones, carrying either one, two or three plasmids providing the different elements of the regulation system. Cell-to-cell variability has to be expected when picking different clones. Therefore, a better way to perform this analysis would be using a single strain carrying all three plasmids and successively inducing RsmA and rsmY expression. It would also be interesting to use another reporter gene such as lacZ to see weather the results show the same trend.

We have tested only two leader sequences but hundreds of others can be found in the literature. They all have a specific behaviour towards the RsmA/rsmY regulation system. Some of them are predicted to have a much higher affinity for RsmA than mag and fha. Another tethering RNA sequence, named rsmZ is also described in the literature and could be tested as an alternative to rsmY. The potential of the RsmA/rsmY translational regulation system extends thus far beyond our experiments and should be exploited in the future.

We provided four sequences (mag, fha, rsma and rsmy) to the part registry. Other teams for further characterisations and integration into any other genetic system that needs a translational regulation can now use them. The mag leader sequence can be implemented into a system in which an initial low level of protein is required, fully repressable by the expression of RsmA. The fha sequence seems to allow the translation of a higher amount of protein that can be efficiently repressed by RsmA although not completely.

Our preliminary data have to be completed to ascertain the inhibitory effect of RsmA and the relieving effect of rsmY before it can be integrated with our toggle switch. As stated above the potential of this system is far bigger than the four bricks we managed to characterize. Our work on the RsmA/rsmY system is therefore a groundbreaking start in the investigation and implementation of a versatile and tunable translational regulation system for E. coli.

Perspectives

So far only the characterisation of the leader sequences has been completed. Fha allows a better expression of GFP than mag.The next experiment will be to look at how these leader sequences react to the presence of the RsmA protein.

We did a few tests with ptet rsma and pcons-fha-gfp or pcons-mag-gfp. The first result showed an extremely low gfp signal, which suggests a repressive effect of Rsma on fha and mag. We cannot conclude yet because some controls are lacking. This however gives very promising prospects.