Mantis is a modular bacterial diagnostic device that will generate a visible fluorescent signal upon detection of antigens of infectious diseases in blood samples of infected patients. When quantifying fluorescence, a low signal-to-noise ratio hinders a proper understanding of the imaging system. Therefore, it is necessary to both increase the sensitivity of the detection and decrease the background signal to allow for a higher accuracy and precision of the quantitative fluorescence measurements [1].

While the fast Mantis reporter module (Signal Transduction and Specific Visualization) focuses on generating a signal in the shortest possible time, this project will focus on getting the most robust detection system. Due to the modular nature of Mantis, it is easy to swap out one system for the other, based on the requirements of the antigen we want to detect. For example, when highly contagious diseases are to be detected, a fast signal would be preferred. However, when the aim is not to detect diseases that are not very contagious, but pose a large health risk if untreated, the least amount of false negatives is desired. In order to achieve a robust signal generation system, a high signal-to-noise ratio is essential. Thus, a higher sensitivity of the fluorescent Green Fluorescent Protein (GFP) detection system and a lower background signal are needed.

The robust Mantis reporter module is based on a spatially separated two-component system. For that, two bacterial cell populations (BCP) will be used, each expressing only one of the two components needed to generate the output signal. Both components will be isolated inside their own BCP so the fluorescent signal is not generated. The two components will only be able to interact with each other after both BCPs lyse, which will be biologically induced after sensing the antigen (Figure 1).

Figure 1: Representation of an antigen-triggered generated output by a two-interdependent components system. In green dots an entire fluorescent molecule (GFP) is depicted.

Together with a reduction of the background signal, a higher sensitivity of the system is desired. For that, a quorum sensing mechanism will be implemented in both BCPs so the signal output produced after the detection of the antigen is propagated to all the bacteria present, no matter if they are directly detecting the antigen or not (Figure 2).

Figure 2: Representation of the antigen-triggered generated output amplified by the quorum sensing mechanism. The figure depicts the communication of a detected antigen from the directly sensing bacteria (green) to the non antigen-sensed ones (red).

In conclusion, when the antigen is detected, a lytic mechanism triggers the lysis of both BCPs. This will allow the interaction of both interdependent components leading to the generation of a fluorescent signal. Provided that both components are separated by both bacterial cell membranes, the background signal is hypothesized to be reduced as the random interaction of those two components is expected to be lower than when present within the same BCP. Therefore, by reducing the background signal which could be generated by the random interaction of both components, the signal-to-noise ratio will be increased (Figure 3).

Figure 3: Representation of the robust Mantis module.

In order to generate a fluorescent output signal, three different systems were constructed and tested. The three proposed systems are composed by two components which do not generate the output signal independently unless they interact between each other.

This system is based on the generation of two BCPs, each encoding for a complementary part of a fluorescent molecule. In this system, both BCPs will be combined in equal concentrations and a lytic process will be induced. This should lead to the fusion of both halves and hence generation of fluorescence. In order to confirm that both complementary fluorescent halves are able to fuse and lead to fluorescence, a third strain used as a positive control which was expressing both complementary halves in the same cell was also generated.

Figure 11: Assessment of a Bimolecular Fluorescence Complementation experiment triggered by sonication and performed with two BCPs. BCP1 encoding for the amino terminal half of GFP with leucine zippers (nGFP-LZ) and BCP2 encoding for the carboxi terminal half of GFP with leucine zippers (cGFP-LZ).

However, the results showed that Bimolecular Fluorescence Complementation was not able to take place in vitro. Even though the recombination of both GFP halves was taking place in the in vivo positive control, when the BCPs expressing each of the GFP halves were mixed and sonicated no fluorescence could be detected (Choosing the best reporter). The reason underlying the incapability of recombination in vitro is unknown.

Figure 12: Assessment of a Bimolecular Fluorescence Complementation experiment triggered by a biological lysis and performed with two BCPs. BCP1 encoding for the amino terminal half of GFP with leucine zippers (nGFP-LZ) and BCP2 encoding for the carboxy terminal half of GFP with leucine zippers (cGFP-LZ).

However, it was concluded that Bimolecular Fluorescence Complementation neither took place in vitro when lysis was triggered biologically. The fusion of both halves took place in vivo as it can be seen for the strain expressing both halves of the GFP, suggesting that the Bimolecular Fluorescence Complementation works, but is prevented from happening when both halves are not contained within a cell.

Because of the split GFP system not working in vivo, an alternative system which has been reported to work in vitro was tested as well [5]. Both systems are based on the generation of two BCPs, the first encoding for a dark quenched GFP molecule and the second one for a constitutively expressed TEV protease (BBa_K2387071).

Approach

FRET (Förster Resonance Energy Transfer) system

In this approach, GFP was quenched by a REACh2 dark quencher (BBa_K1319004), the GFP quenching process takes place in this system via a FRET system (Figure D). The REACh2 quenching protein (BBa_K1319002) fused to GFP reduces the observed fluorescence of GFP. This protein is a non-fluorescent but still absorbent version of EYFP [6]. GFP and REACh2 are fused with a linker (BBa_K1319016) which includes a specific TEV protease (BBa_K1319004) cleavage site. The fused protein brings GFP and REACh2 in proximity to each other which allows both GFP and REACh2 to act as donors and acceptors in a FRET (Förster Energy Transfer System) system. GFPs emission energy is absorbed by REACh2 leading to a reduction in the GFP fluorescence.

Figure D: Gene sequence depicting the GFP quenched by REACh2 construct.

Hydrophobic Quenching System

In this second approach, the GFP molecule is quenched by a hydrophobic peptide from a mutated version of the M2 fragment of influenza virus (Figure E) and a second one able to constitutively express the TEV protease. The GFP is completely quenched by the binding of the hydrophobic peptide which tetramerizes the GFP disabling the maturation of the chromophore [3].

Figure E: Gene sequence depicting the GFP quenched by M2 fragment construct.

GFP fluorescence can be recovered after the removal of the quenching peptide, leading to the generation of a robust reporter as a result of a proteolytic process. This proteolytic process is performed by the TEV protease which performs the cut of the linker fusing GFP to the hydrophobic peptide. This linker (BBa_K1319016) includes a specific TEV protease (BBa_K1319004) cleavage site.

In both system, both BCPs were co-transformed with the LysE7 lytic mechanism to allow cell lysis after the induction of L-Arabinose. Both BCPs were combined in equal concentrations and a lytic process was induced, which should allow the TEV protease to cut the linker fusing both GFP to its quenching molecule (REACh2 or the mutated M2 fragment). Once both proteins are separated, the quenching interaction is prevented resulting in an increase in GFP fluorescence. The fact that also GFP quenched by REACh2 in the absence of the TEV protease producing strains lead to fluorescence could be explained by a breakage of the linker connecting REACh2 to the GFP as a result of the lytic mechanism (Figure 13).

The induction of lysis via the addition of L-Arabinose lead to the breakage of the bacterial cell membranes (Figure 13). Fluorescence is increased in comparison to the non co-transformed strains which are not able to lyse, suggesting that the the lytic process leads to the recovery of the quenched GFP. This can indicate that both REACh2 and the mutated M2 fragment work as expected and that its quenching effect might be inhibited after the linker binding it to the GFP is cut, allowing GFP to recover its fluorescence (Figure 13).

Figure 13: Assessment of two proteolytic quenching mechanisms (GFP quenched by REACh2 or by a mutated version of M2 fragment). GFP-REACh2 stands for a strain expressing the GFP fluorophore quenched by REACH 2, GFP-M2 stands for a strain expressing the GFP fluorophore quenched by M2, TEV strain stands for a constitutively expressed TEV protease strain, TEV + GFP-REACh2 stands for a combination of TEV protease expressing strains and GFP quenched expressing strains, TEV + GFP-M2 stands for a combination of TEV protease expressing strains and GFP quenched expressing strains, GFP-REACh2 & LysE7 stands for a strain expressing the dark quenched GFP and co-transformed with the LysE7 lytic mechanism, GFP-M2 & LysE7 stands for a strain expressing the dark quenched GFP and co-transformed with the LysE7 lytic mechanism, TEV + GFP-REACh2 & LysE7 stands for a combination of TEV protease expressing strains and GFP quenched expressing strains both of them co-transformed with the LysE7 mechanism and TEV + GFP-M2 & LysE7 stands for a combination of TEV protease expressing strains and GFP quenched expressing strains both of them co-transformed with the LysE7 mechanism.

In the case of GFP quenched by REACh2, the results seen do not show an increase in fluorescence when comparing the lysed cells with the non lysed ones as was seen in the mutated M2 approach (Figure 13). The reason for this is the already high fluorescence level of GFP when quenched by REACh2. The explanation for this comes from the fact that REACh2 is not preventing the maturation of the fluorophore as M2 was doing but is absorbing the emitted fluorescence of GFP. Therefore the REACh2 quencher emits fluorescence in a close emission range as GFP as it is a version of YFP. However, if the fluorescence is measured overtime, an increase in GFP fluorescence could be seen (Figure 14).

Figure 14: Assessment overtime of the dark GFP quenched by REACh2. GFP-REACh2 stands for a strain expressing the GFP fluorophore quenched by REACH 2, TEV strain stands for a constitutively expressed TEV protease strain, GFP-REACh2 & LysE7 stands for a strain expressing the dark quenched GFP and co-transformed with the LysE7 lytic mechanism and TEV + GFP-REACh2 & LysE7 stands for a combination of TEV protease expressing strains and GFP quenched expressing strains both of them co-transformed with the LysE7 mechanism.

Four different lytic mechanisms were designed to be triggered by the antigen, although for the ease of testing in this project L-Arabinose is used to induce the signal in all constructs. In order to assess and evaluate these mechanisms, they were expressed under the inducible pBAD promoter (BBa_I0500) in two different backbones: a high copy number backbone (pSB1C3) and a low copy number backbone (pSB4K5).

Inducible versions of Colicin E7 (BBa_K2387066), T4 Holin (BBa_K2387068), T4 Endolysin (BBa_K2387069) and a double lytic mechanism (T4 Holin + T4 Endolysin) (BBa_K2387079) were tested in E. coli DH5α (Figure 15). Instead of directly quantifying cell lysis, the optical density at 600 nm was used.

Figure 15: Graphical representation of 4 lytic mechanisms layout.

From the assessment it was concluded that T4 Holin and T4 Endolysin + T4 Holin were successfully achieving bacterial cell lysis. T4 Endolysin alone was not able to trigger the lysis of bacterial cells. However, when combined with T4 Holin, the lytic effect achieved was even higher than the one achieved with T4 Holin alone (Figure 16). The observed lytic function of T4 Endolysin when acting together with T4 Holin in the double lysis gene cassette is understandable as it requires the action of T4 Holin to make the initial holes in the membrane [7]. T4 Holin starts the membrane degradation process and after holes are made in the membrane, T4 Endolysin is able to further increase bacterial membrane lysis (Figure 16).

Based on the results, it was concluded that already low inductions of L-Arabinose are able to induce a complete lysis of the bacterial cells (Figure 16). Either LysE7 system, T4 Endolysin + T4 Holin system or T4 Holin system alone could be implemented in the robust Mantis module (Figure 3). Based on the simplicity of the Inducible LysE7 system (BBa_K2387066) which only requires one gene to work compared with the double lytic casette (BBa_K2387079), this system (BBa_K2387066) was the chosen one when co-transforming strains in the proteolytic quenching experiments and the Bimolecular Fluorescence recombination.

Figure 16: Assessment of four lytic mechanisms (Colicin E7 system, T4 Endolysin, T4 Holin and T4 Endolysin + T4 Holin cassette) in two backbones (pSB1C3 and (pSB4K5)). All the mechanisms are expressed under the pBAD promoter and therefore all the mechanisms were induced at different L-Arabinose concentrations. The figure above depicts the OD after 20 hours of L-Arabinose induction.

The final mechanism will be constituted by four modules (Figure 17) expressed by two different BCPs. Both BCPs will express the four modules, being all the modules the same except module 2 which will encode for component 1 in BCP1 and for component 2 in BCP2. In the system, aiiA will be constitutively expressed under no induction preventing quorum sensing to happen. In an induced state (antigen detected or L-Arabinose added to the medium), pBAD is activated leading to the expression of TetR which inhibits the expression of pTet resulting in the downregulation of aiiA. As a result, AHL degradation is decreased and the quorum sensing mechanism becomes active. The activation of the quorum sensing mechanisms leads to both the activation of the lytic mechanism as well as the activation of a positive feedback performed by the diffusion of AHL from cell to cell. In a final state, all the cells in the medium will lysed allowing component A to B to interact and generate a fluorescent signal.

Figure 17: Graphical representation of the final system. Robust mantis is constituted by four modules ([1] Quorum sensing module, [2] Two interdependent components system module, [3] Lytic mechanism module, [4] Anti quorum sensing module) which will be expressed by the two BCPs.