Multilayered genetic safeguards limit growth of microorganisms to defined environments

Abstract

Genetically modified organisms (GMOs) are commonly used to produce valuable compounds in closed industrial systems. However, their emerging applications in open clinical or environmental settings require enhanced safety and security measures. Intrinsic biocontainment, the creation of bacterial hosts unable to survive in natural environments, remains a major unsolved biosafety problem. We developed a new biocontainment strategy containing overlapping 'safeguards'-engineered riboregulators that tightly control expression of essential genes, and an engineered addiction module based on nucleases that cleaves the host genome-to restrict viability of Escherichia coli cells to media containing exogenously supplied synthetic small molecules. These multilayered safeguards maintain robust growth in permissive conditions, eliminate persistence and limit escape frequencies to <1.3 × 10(-12). The staged approach to safeguard implementation revealed mechanisms of escape and enabled strategies to overcome them. Our safeguarding strategy is modular and employs conserved mechanisms that could be extended to clinically or industrially relevant organisms and undomesticated species.

Figure 1.

Design of multilayered genetic safeguards: riboregulation, engineered addiction, auxotrophy and supplemental repressors. (A) Riboregulation system: pLtetO promoter (19), repressed by TetR and induced by aTc, drives trans-activating (taRNA); pLlacO promoter, repressed by LacI and induced by IPTG, drives cis-repressed (crRNA) and essential gene. crRNA and taRNA fold through a linear loop intermediate to reveal the crRNA's RBS permitting expression (green). Supplementary TetR (purple) and LacI (green) are constitutively expressed from the genome. Carbenicillin resistance gene (bla) replaces bioAB, resulting in biotin autotrophy (blue). Constitutive EcoRI nuclease (magenta) enables inducible cell killing in the absence of EcoRI methylase (yellow), which is controlled by aTc. (B) Riboregulation system controls cell viability. Chloramphenicol (cam) resistance of strains carrying cat gene regulated in different ways, all grown without inducers. Constitutive (black, + control); pLtetO (green, non-riboregulated control); riboregulated (red); no cat (orange, - control). (C) Kinetic growth curves demonstrate dependence of engineered addiction system (strain EcEAM) on aTc.

Figure 2.

Characterization of ribo-essential and toxin safeguards. (A) Dilution series (by row) of strains carrying different riboregulated essential genes or toxins (by column) growing on permissive and non-permissive solid media. Ribo-essential strains based on ribA, nadE or glmS, respectively, require IPTG, arabinose or rhamnose plus aTc to grow. Doubling time (DT) and escape frequency (EF) are given for each strain. EAM-based safeguards depend on aTc. N/C denotes not contained, implying EF cannot be calculated. (B) Riboregulated ribA strain grown in mixed culture with ancestral strain in permissive media (+aTc, +IPTG). (C) Mixed culture experiment in non-permissive media (-aTc, -IPTG) compares fitness of ancestor and escape mutants. (D) Fitness heat map of riboregulated ribA (induced by IPTG and aTc) grown with arabinose or (E) nadE (induced by arabinose and aTc) grown with IPTG in strain EcR1ribR2nad+. Color represents maximum doubling time, as a fraction of wt.

Figure 3.

Layering of multiple safeguards reduces escape. (A) Escape frequencies (n = 3, ±SD) on solid media for strains containing one (red), two (orange), three (green) or four (blue) layers of genetic safeguards. NIH guidelines for work with engineered microorganisms advise a 10⁻⁸ escape frequency (11) (dotted horizontal line). Strains are characterized by ribo-essential gene (RE 1 or 2), presence of supplemental repressors, presence of a toxin and doubling time (DT) in minutes. Plasmid-based ribo-essential gmk safeguard (gray) does not confer containment. Square brackets denote plasmid-based constructs. Limit of detection for solid media is ∼5 × 10⁻¹⁰. Flow cytometry shows fluorescence from sodA and sulA promoters in EcR1rib+ (B) or EcTeco (C) when grown in non-permissive media (dashed) versus permissive (solid) media for 2 (purple), 6 (red) or 24 (black) h. *** denotes P ≤ 0.001, Student's one-tailed t-test with Welch's correction.

Figure 4.

Large volume and long-term challenge of multilayered safeguard strains. (A) Strains grown in permissive (P) media to ∼10⁹ cells then challenged over 96 h in 1 l of non-permissive (NP) media, which is monitored for cell density (OD) (B) and cell viability (CFU) on permissive (solid line) and non-permissive (dotted line) solid plates (C). Biological triplicate results for control strain (EcNR1, black), 2-layer strain (EcR1rib+, orange), 3-layer strain (EcR1ribR2nad+, green) and 3-layer bacteriotoxic strain (EcR1rib[Teco]+, purple) plotted. (D) Control (EcNR1, black) and 4-layer safeguard (EcR1ribR3glmEAM+, red) strains grown in permissive, then challenged in 1 l of non-permissive media. (E) Strains were monitored by OD readings (F) and by plating on permissive or non-permissive solid media over 14 days (squares, ∼10¹² cell inoculum) or 5 days (triangles, ∼10⁸ cell inoculum). No colonies were observed on NP media for the 4-layer strain.