Results

Objective

We designed four gates, two XOR and two AND. (here)
Our objective was to characterize these gates with the recombinases as input and RFP as output, Tp901.1, Bxb1 for BBa_K1132001 and BBa_K1132002 and FimE and PhiC31 for BBa_K1132003 and BBa_K1132004. With the RFP as output, it will be easy to see the switch of the logic gates.

Conception

The gates have been inserted into pSB1C3, and the RFP gene inserted after the gate or inside it (depending of the gate type). For XOR2 and AND2, the T7 promoter was used and the polymerase T7 was also inserted in the construct, with a promoter, a rbs and a terminator.

We succeeded in assembling Bxb1 with a ribosome binding site and a promoter and FimE with a ribosome binding site, a promoter and a terminator.

The gates and the recombinases parts have been submitted, here is the detailed description of the parts.
We did not manage to finish (on time!) the cloning of PhiC31 and Tp901.1 but (thanks to Suiti’s and Bonnet’s help!!), we were able to produce the recombinases directly from the original plasmids (view other parts), and use the produced recombinases for in vitro characterization.

Results

The following table presents our plan for testing the efficiency of the recombinase-based switch and the results we should expect.

in vitro switch of the logic gates

The basic idea is to perform the switch with a cell lysis containing the recombinase and the plasmid containing the gate to be switched. After incubation, transformation of the plasmid containing the gate allows the quantification of switched versus non switched plasmids. Experiments have been performed as described in the protocol.

XOR1

In the absence of recombinase or in recombination assays with FimE, the resulting colonies are white. Therefore, the XOR1 gate does not present any transcriptional leakage and non-specific recombinases cannot switch the door. When XOR1 was put in the presence of Bxb1, the switch occurred as attested by the presence of numerous red colonies (RFP expression).

To test the time dependency of the reaction, the reaction mixture incubated was analyzed at different times, from 15 min to 2.5 hours.

From the graph above, one can see that switching occurs after 15 min. After 2 hours reaction, over 90 % of XOR1 gates were switched and the percentage of XOR1 gates switched from 2 hours to 5 hours reaction is almost constant (not shown).

In the in vitro protocol, DNAses were not eliminated. Therefore, longer times reactions also means that plasmids containing the gates can be degraded. This could explain why, when incubation lasted longer than 2 hours, the number of colonies dropped as shown in the following graph.
We also tried to switch XOR1 with the two recombinases. The XOR1 gate was tested with the Dual Controller Plasmid provided by Bonnet (KC529324) and containing Tp901.1 and Bxb1. No switch was ever obtained. This negative result might be due to low expression levels of any of the two recombinases that would lead to partial switch in in vitro conditions. Further experiments will have to be performed to understand this phenomenon.

Finally, We analyzed XOR1 switched plasmids by sequencing. The sequences correspond to the expected switches.

XOR2

In the absence of recombinase or in recombination assays with Bxb1, the resulting colonies are white. Therefore, non-specific recombinases cannot switch the XOR2 gate. When XOR2 was put in the presence of FimE, the presence of red colonies validated the switch.
However, only 5% of the plasmids were switched. Hence, the switch of XOR2 using FimE was not highly efficient. Several explanations can be made: i) low expression of FimE; ii) low activity of FimE in the tested buffer; iii) low activity of FimE on our designed gate. Improvement on FimE protocols or redesign of new sites could have been performed with our Michigan buddies but, unfortunately, time was running out...

We also performed a second switch, using the XOR2 gate that has been switched. White colonies appear as a function of time. We prepared the DNA from the XOR2 gate that has been previously switched by FimE (from a red colony). This plasmid DNA was tested in the presence of PhiC31 recombinase. The picture below demonstrates that the second switch occurred: white colonies are now present.

We also tried the switch with PhiC31 only. We used the original Siuti’s plasmid containing PhiC31, but no switch was obtained. The results might be due to low expression levels of PhiC31 leading to partial switch in in vitro conditions. A second explanation would be that the site we have designed is less efficient when the FimE recombinase has not yet performed the primary switch. More experiments are needed to verify PhiC31’s efficiency.

AND1

In the absence of recombinase, the colonies are white. As a result, AND1 gate does not have transcriptional leakage and non-specific recombinases cannot switch the door.

We tested the AND1 gate with Bxb1. After incubation with Bxb1, the colonies are still white (this is normal!), therefore visual inspection of the switch was not possible. The only possibility was to sequence independent clones (see genotype section). We analyzed five independent colonies by sequencing, all of them were switched.

AND2

In the absence of recombinase, the colonies are white. As a result, non-specific recombinases cannot switch the AND2 gate.

We tested the AND2 gate with FimE. After incubation with FimE, the colonies are still white (this is normal!), therefore visual inspection of the switch was not possible. We analyzed five independent colonies by sequencing, but none was switched. Several hypotheses can be made. First, the presence of the PhiC31 switching sequences may inhibit.

The AND2 was also tested with the Siuti’s construction parts containing PhiC31, but not switched has been obtained. As with the XOR2 gates, results might be due to low expression levels of PhiC31 that would lead to partial switch in in vitro conditions. A second explanation would be that the site we have designed is less efficient when the FimE recombinase has not yet performed the primary switch. More experiments are needed to verify PhiC31’s efficiency.

In vivo switch using cotransformation

The principle of this characterization is to cotransform the gate and the recombinase in the same strain. The gate construct (BBa_K1132032) is on chloramphenicol and the Bxb1 construct (BBa_K1132027) is on kanamycine, we used 20 ng of each plasmid to transform E. coli and let it growth on chloramphenicol and kanamycine. Some colonies were red after transformation. However, in these conditions, the genotype of the gate (presence of the switch) is hard to verify due to the presence of two plasmids in the cotransformed strain. We therefore resorted to deeper characterization with fluorescence measurements. The positive control is a plasmid expressing constitutively RFP and the negative control is the co-transformation of XOR1 with a plasmid not containing any recombinases but on the same antibiotics resistance.

On the above graph, there is a clear difference when XOR1 is cotransformed with Bxb1 contrarily to the negative control. Analysis of the fluorescence of the strain demonstrated that the level of RFP production of the negative control is probably due to light diffusion by the cells and not real fluorescence. Therefore the in vivo switch of XOR1 is confirmed. The lower fluorescence of our strains compared to the positive control could be due to the presence of a weak promoter in front of RFP and/or the longer than usual distance between the promoter and the coding sequence.

Summary
The following table summarizes our results with the different logic gates and the recombinases. Green cells are successful experiments while gray ones did not give interpretable results and the blue ones have not been done yet.

Discussion

All the constructs for this characterization have been made successfully. The characterization proves that the gates do not have transcriptional leakage and non-specific recombinases cannot switch the gates. The switch by one recombinases has been prove in vitro, by an amazing protocol, for all the gates unless AND2. It demonstrates that the gate has been well designed. The absence of result with the other recombinases has been caused by trouble on the origin construction or by none prefect condition of the in vitro protocol.
This result show that the best solution to characterize these gates would be to integrate them, indeed in vivo, the result is stochastic because of the number of plasmid copy into the cell and in vitro, it is difficult to conclude with bad result, it can be because of none well condition or because of trouble on the recombinases or on the gate. We start the construction of a strain for integration, because of lack of time; we prefer to concentrate on an in vitro characterization with a new protocol.