(a) (i) Vector-host system created in the E. coli Nissle strain by auxotrophic complementation to stabilize plasmid retention. The ‘Sense-Kill’ construct comprises of: lasR (blue), pyocin S5 (brown), E7 lysis protein (orange) and DspB (purple) (Supplementary Table 1). (ii) The chromosomal deletion of alr and dadX in E. coli Nissle was confirmed by PCR. Genes flanking the deleted region are indicated by their annotated gene names: GNBM-2454 (D-amino acid dehydrogenase small subunit), GNBM-2456 (cell volume regulation protein A), GNBM-3803 (ribosyl nicotinamide transporter, PnuC-like), GNBM-3805 (hypothetical protein). (iii) Growth of strain EcN (Nissle Δalr ΔdadX) in the presence of exogenous D-alanine (50 μg ml−1) or in the presence of complementary vector (pEaaK; Supplementary Fig. 1). The microtiter plate assay was performed in triplicate wells; the mean and s.e.m. (error bars) from three experiments are shown. (iv) Plasmid retention was tested by culturing wild-type Nissle (non-auxotrophic) and EcN (auxotrophic) cells containing plasmid pEaaK (which includes a kanamycin resistance gene) in the absence of antibiotic and in the presence or absence of exogenous D-alanine in minimal medium. Cell viability in the presence of kanamycin was used to infer plasmid retention and expression. The mean and s.e.m. (error bars) from three experiments are shown. (b) (i) Optimization of the regulatory components of the sensing system (highlighted in black in the circuit diagram) was performed. (ii) The effects of several variations in the regulatory components on GFP expression and bacterial growth were assessed. The mean and s.e.m. (error bars) from three experiments are shown. (c) GFP production per cell over time at different 3OC12-HSL inducer concentrations in Nissle and EcN cells containing the reporter construct. The optimized circuit (V5) showed a lower threshold for 3OC12-HSL detection and sustained expression in the new vector-host system.