To measure easily the effect of RIP on S. aureus, we designed a initial construct allowing us to simply produce RIP after IPTG induction. We thus designed a part containing a T7 promotor, ribosome binding site (RBS) upstream, and followed by a signal sequence for secretion fused to RIP sequence downstream of the coding sequence.

Eight different secretion systems have currently been identified in Escherichia coli, and more generally in Gram-negative bacteria (Ref.). Type I and type II secretion systems are the most commonly used for production of recombinant protein, and we used Type II to secrete RIP.
Type II secretion system is a two-step system: proteins are first exported to the periplasmic space, between E. coli’s two membranes, and then secreted to the medium.
We target RIP to the periplasm using the Sec-dependent secretion system (Ref.). Our peptide is fused to an amino-terminal signal sequence that gets recognized by the chaperone SecB, then addressed to SecA and then translocated accross the inner membrane through the SecYEG complex. One advantage of this system is that the signal sequence gets cleaved during translocation through SecYEG (Ref.). Several different signal sequences have been characterized (all formed of between 18 and 30 amino acids (Ref.).
Following the advice of Dr. Jean-Michel Betton, in the Structural Biology Department at Institut Pasteur, we chose to try two different signal sequences to export our peptides, which he knew about and known to be efficient: MalE and DsbA (Ref.). A secretion machinery called secreton then enables the release of proteins extracellularly. However, this system is not very well characterized yet, and it is a complex machinery composed of more than 10 proteins, so we did not plan to use it. However, since RIP is only a 7 amino acid protein (Ref.), leaky release from periplasm to the medium should be sufficient to obtain RIP in the medium.

In order not to produce RIP continuously but only in presence of Staphylococcus aureus, we engineered a sensor device in our modified E. coli capable of detecting Staphylococcus aureus and producing RIP after detection.

Our modified E. coli expresses under a constitutive Ptrc promoter the sensor device of S. aureus agr operon. It is composed of agrA and agrC. Staphylococcus communicate through a quorum sensing mechanism, which consists in producing and detecting signaling peptides called AIP (Auto-Inducing Peptide). If S. aureus approaches our system, AIP will be detected by the transmembrane protein agrC, launching the phosphorylation of agrA which then activates promotor P2 (see below). Our RIP sequence fused to a signal sequence for periplasmic export is placed after promotor P2 and will consequently be expressed only if S aureus approaches our biofilm.

We built a part that should be integrated in our final device, and permits to secrete NGF directly in the extracellular medium using E. coli type I secretion system. We used an inducible promoter T7 in order to control NGF production thanks to IPTG induction. We added an His-tag in order to purify it.

Type I secretion system transports proteins in one step across the two cellular membranes (Ref.). It is composed of an inner membrane protein HlyB, a periplasmic channel protein HlyD and an outer membrane protein TolC. As every secretion system, secretion through this one too is mediated by the specific recognition of a signal sequence, which in this case are the sixty C-terminal amino acids of alpha-haemolysin HlyA. This sequence binds with the HlyB-HlyD complex and is then translocated into the channel (Ref.).

There are several problems to address when secreting recombinant proteins through the Type I secretion system:

• First, the signal sequence HlyA is not cleaved when crossing the channel, but it needs to be cleaved to obtain a functional protein in the medium. That is why we fused the NGF sequence with this sixty-amino acid long sequence, separated by the cleavage site for Tobacco Etch Virus protease (TEV). As we co-express TEV protease, the signal sequence will be eliminated once it is out the cell, and our NGF can be active.
• Secondly, only if TolC endogenously exists within Escherichia coli. But this is not the case of the transporter complex HlyB-HlyD. We have thus co-transformed our bacteria with another plasmid pVDL 9.3, generously provided by Dr. Victor de Lorenzo, from Centro Nacional de Biotecnologia of Madrid, bearing HlyB and HlyD sequences, in order to get a chance to secrete NGF out of the cell.

Since we imagined a technology which requires to integrate engineered bacteria in humans, we need to ensure that the engineered bacteria are contained in the specific environment they are designed for. That is why we thought of using a “kill switch” that cause them to die if they are released in the environment. Our kill switch is based on temperature: it enables bacteria to survive at human body temperature (37°C) but die at lower temperatures.

The kill-switch we use is based on a toxin/antitoxin combination, CcdB/CcdA (Ref.). CcdB is a lethal toxin for E. coli and its production is placed under the regulation of a temperature-sensitive promoter (Ref.). In permissive conditions, i.e. in human body, the expression of the toxin is repressed and the antitoxin is expressed at a constitutive low level in order to counteract any leaky expression of the toxin. When the temperature goes lower, the repression is lifted and toxin expression increases. The constitutive low level of antitoxin is no longer sufficient to counter the effects of the toxin, and the bacteria die.