As IFFLs require a reporter and its inhibitor to be expressed at the same time, we needed a way to have both of these products induced at the same time. Previously, iGEM teams have found that IPTG-induced mf-Lon and ATC-induced mscarlet-I can be used together to produce a functioning IFFL circuit. This system is ideal for creating an IFFL, as using two different chemicals for induction allows for tight control of the expression parameters of both molecules. Through experimentation, this allows for the creation of a circuit with strong IFFL properties.

The first experiment we perused was ensuring that our chemically induced parts function as we expected. To do this, we transformed bacteria with a 3K3 plasmid containing the IPTG induced mf-Lon and a 1C3 plasmid containing the ATC induced mscarlet-I. We then tested these cells on the plate reader with a variety of induction conditions as per Figure 1.

While we were able to find success by doing this, However, this posed a problem, as each centrifugation step required a significant amount of time to complete. Ultimately, it was determined that having to manually centrifuge the bacteria each time we wanted to turn the IFFL on/off would be infeasible. Thus an alternate induction system would be required in order for our project to proceed. Since we were using chemical inducers, turning the circuit off posed a bit of a challenge. As we did not possess advance microfluidic technology, we would need a way to remove the chemical inducers. In order to accomplish this, we decided to centrifuge the bacteria, then wash the pellets with media. To sufficiently wash out the inducer, this step would have to repeated at least three times.

Next, we designed a separate temperature controlled system (BBa_K2680050) that expressed mf-Lon, which we hoped to combine with K860052 to form a complete IFFL. However, while cells co-transformed with both of these parts grew fine at 29C, when the IFFL was turned on by heating to 37C, no change in fluorescence values was apparent. Upon inspection of the optical density values, the reason became clear. Upon activation of the IFFL an immediate drop in cell density occurred, indicating that the constructs were toxic to the cells (Figure 3a). This was not overly surprising, as previous teams have reported that high expression of mf-Lon has a cytotoxic effect on E. coli. Given that the promoter (R0051) driving the expression of mf-Lon is phage derived and has extremely strong expression, it is reasonable to believe that an overly high expression of mf-Lon led to this result. Next, we determined that the expression of mf-Lon from our construct actually led to cell death and did not merely inhibit growth (Figure 3b).

We first attempted to get around this issue by using a lower temperature of 35C to activate the cells, but found that this still lead to cell death (data not shown). Based on these results, we reasoned that we might want to reduce the expression of mf-Lon as well as the expression of mScarlet-I-pdt, as in conjunction they appeared to strongly cytotoxic. We reasoned that one way to resolve this would be to express all transcriptional units together on a single plasmid on the low-medium copy backbone psb3K3. However, when we tested this single plasmid construct we found that while cell growth was normal, there was no appreciable gain in fluorescence upon induction (Figure 4a). As this construct was sequenced confirmed (see WM18_016 3K3 in sequencing on Data and Methods), we deemed it likely that there was simply too much mf-Lon expressed relative to mScarlet-I-pdt. This matches with previous characterization where the amount of ratio of mScarlet-I to mf-Lon needed to be far higher than what our construct would likely achieve in order to observe a pulse. We reasoned that one potential solution would be to add a strong ssrA tag (BBa_M0050 onto mf-Lon in order to increase its endogenous degradation rate, which would serve to lower its expression. However, we found that the single plasmid circuit with the ssrA performed similarly to the single plasmid circuit without the ssrA tag (Figure 4b).

After determining that these specific single plasmid IFFLs did not have appropriate parameters, we next investigated whether any single plasmid circuit made up of our existing parts was capable of generating the characteristic pulse of an appropriately tuned IFFL. Utilizing 3G Assembly to generate huge numbers of different single plasmid circuits in a single reaction, we were ultimately able to determine that no combination of our current parts would lead to a functional single plasmid IFFL circuit on psb3K3 (see the parallel 3G Assembly page for more information). While this result meant we now had to change which heat sensitive system we would use for our decoder, it ultimately saved us many weeks of experiments. Hopefully, this technique will also enable future teams to perform similar experiments to evaluate potential configurations of their circuits.

The system that we decided upon was based on tlpA₃₉ (BBa_K2680279), a version of the tlpA repressor found in Salmonella typhimurium, which was engineered by Piraner, et al to have an activation temperature of 39C [2]. We chose tlpA₃₉ as it was mechanistically similar to ts-CI, but possessed several advantages compared to ts-CI. These advantages included that our cells would not have to be grown at special temperatures during cloning, as well as that this system has two different promoters, which differ by over 200 fold, which would give us a larger range of parameters which we could use to design our IFFL circuit.

Utilizing 3G assembly, we then design a wide variety of heat inducible circuits, with both the ptlpA (BBa_K2680123), and the ~200 fold weaker ptlpAR promoters (BBa_K2680123) as well as various combinations of different mScarlet-pdts and mf-Lon with and without M0050. After testing however, we found that none of the combinations of the new parts produced the characteristic pulse of an IFFL (Figure 5).

We considered that there might be issues with one or more of the components of our constructs. So we tested some of our mScarlet-I pdt constructs and determined that in fact the degradation tags do perform appropriately in the presence of mf-Lon, with the stronger degradation tags having lower fluorescence compared to the weaker degradation tags, which were weaker than the untagged control (Figure 6a). Further, we found that the calculated degradation rates were in line with our expectations, with the tags being in order of expected strength based on Cameron and Collins [3] (Figure 6b).

We next tested whether our heat inducible mf-Lon constructs were working by combining it with the well characterized ATc inducible mScarlet construct K2333428. Using this we were able to determine that our mf-Lon constructs were indeed functional, and that there was little to no difference in degradation rate between mf-Lon (K2680060) and mf-Lon-M0050 (K2680061) (Figure 7).

Concurrently with these tests, we were also working on integrating constructs onto the genome using pOSIP[4]. We made a strains that expressed either mf-Lon or mf-Lon-M0050 under the control of the ptlpA and ptlpAR, as well as a mf-Lon under the control of the R0051 promoter. We confirmed their integration as indicated in the pOSIP protocol, and then functionally tested them with the ATc inducible mScarlet-I-pdt K2333428 (see above section). When we transformed K2333428 into the cells containing integrated mf-Lon constructs, we found that simultaneous activation of mScarlet-I-pdt (via ATc) and mf-Lon (via temperature change) generated the characteristic IFFL pulse (Figure 8). However, when we attempted to use our temperature controlled mScarlet-pdt parts in the integrated cells, no pulse was observed (Figure 9).