Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Recent advances in synthetic biosafety

Recent advances in synthetic biosafety Invited Referees Synthetically engineered organisms hold promise for a broad range of medical, 1 2 environmental, and industrial applications. Organisms can potentially be designed, for example, for the inexpensive and environmentally benign version 1 synthesis of pharmaceuticals and industrial chemicals, for the cleanup of published environmental pollutants, and potentially even for biomedical applications such 31 Aug 2016 as the targeting of specific diseases or tissues. However, the use of synthetically engineered organisms comes with several reasonable safety concerns, one of which is that the organisms or their genes could escape their F1000 Faculty Reviews are commissioned intended habitats and cause environmental disruption. Here we review key from members of the prestigious F1000 recent developments in this emerging field of synthetic biocontainment and Faculty. In order to make these reviews as discuss further developments that might be necessary for the widespread use comprehensive and accessible as possible, of synthetic organisms. Specifically, we discuss the history and modern peer review takes place before publication; the development of three strategies for the containment of synthetic microbes: addiction to an exogenously supplied ligand; self-killing outside of a designated referees are listed below, but their reports are environment; and self-destroying encoded DNA circuitry outside of a not formally published. designated environment. 1 Shota Atsumi, University of California, Davis USA 2 Ichiro Hirao, Institute of Bioengineering and Nanotechnology Singapore Discuss this article Comments (0) Corresponding author: Andrew D. Ellington (ellingtonlab@gmail.com) How to cite this article: Simon AJ and Ellington AD. Recent advances in synthetic biosafety [version 1; referees: 2 approved] F1000Research 2016, 5(F1000 Faculty Rev):2118 (doi: 10.12688/f1000research.8365.1) Copyright: © 2016 Simon AJ and Ellington AD. This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Grant information: AJS is funded on NASA NAI CAN, under grant number: NNX15AF46G Competing interests: The authors declared that they had no competing interests. First published: 31 Aug 2016, 5(F1000 Faculty Rev):2118 (doi: 10.12688/f1000research.8365.1) F1000Research Page 1 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 environment, and real-life organisms addicted to natural chemi- Strategy 1: addiction cals could likely escape the same way. Wright and colleagues, for As with the lysine-deficient fictional dinosaurs in the book Jurassic example, demonstrated that ΔthyA mutants grew readily in media Park, it may be possible to employ addiction strategies to make it lacking specifically supplied thymidine but supplemented with a difficult for organisms to survive outside of their designated habi - small amount of sterilized soil, demonstrating that environmental tats. This biocontainment strategy dates back to the earliest days nutrients may complement metabolic deficiencies . Furthermore, of cloning with the development of the specialized Escherichia the engineered inability to generate a key metabolite often inhibits coli strain χ1776, which lacked functional aspartate-semialdehyde the growth of an organism even when the metabolite is heavily dehydrogenase (asd), (L-delta-1-tetrahydrodipicolinate synthetase) supplemented, rendering such engineered organisms ill-suited for (dapD), and thymidylate synthetase (thyA) genes and thus required 1,4 use in bioreactors and many other applications . Thus, the more their products, diaminopimelic acid (DAP) and thymine or thy- modern version of this strategy is to engineer organisms that depend midine, to survive . Because this strategy is so straightforward— not on a natural compound but on a synthetic one. If engineered requiring only the creation of organisms deficient in the ability to organisms require a synthetic compound for survival, they will be produce a key metabolite and this is readily achieved by targeted unable to survive outside of a laboratory or other highly special- and random mutagenesis—it has remained commonly employed ized environment in which the chemical is supplied. However, this and integrated into more sophisticated synthetic biosafety platforms 2,3 engineering feat is more difficult than simply generating organisms through recent times (Figure 1, top left). with a “broken” ability to produce a naturally occurring metabolite. Despite the straightforwardness of this simple addiction strategy, Instead, it requires engineering some sort of metabolic or functional it was not sufficient to contain Jurassic Park’s resurrected fic - connection to a synthetic chemical that was previously irrelevant to tional dinosaurs because of the ready availability of lysine in the the organism’s biology. Figure 1. A common strategy for the containment of synthetic organisms is to engineer them to require an exogenously supplied ligand. Common methods for achieving this include (clockwise from top left) knocking out a required gene and exogenously supplying the 1 9 gene product , requiring the amber-mediated incorporation of a synthetic amino acid for essential protein production , requiring the amber- 2,7 mediated incorporation of a synthetic amino acid for essential protein function , requiring a synthetic molecule as a cofactor for protein function , requiring a synthetic nucleotide for essential gene replication or translation, and requiring a synthetic molecule as a precursor for a key metabolite . Page 2 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 One strategy by which researchers have adapted organismal supplemented media but “escaped” with rates below the detectable −11 9 metabolism to require synthetic compounds is by altering the 4 × 10 frequency after 20 days of growth . This very low escape genetic code to incorporate and even require unnatural amino acids. frequency arises from the dependence of multiple proteins on these In this regard, the development of so-called orthogonal translation synthetic amino acids rather than the dependence per se of a single machinery has advanced to the point where the incorporation of protein: whereas the probability of one amber codon reverting to a −7 unnatural amino acids into proteins, at least across from amber natural codon is moderate (~10 ), the probability of getting three codons, is relatively straightforward . Although an organism can in parallel is much lower. readily be rendered dependent upon suppression of an amber codon (for example, by introducing the amber codon into an essen- A strategy to render organisms dependent on synthetic molecules tial protein), rendering a protein (and thus its organism) dependent at an even deeper level than addicting them to synthetic tRNAs on a specific synthetic amino acid is considerably more compli - or amino acids is to addict them to synthetic nucleotides. Mut- cated, as it requires the specific, introduced amino acid to be zel and colleagues demonstrated the ability to engineer organ- necessary for protein function (Figure 1, top right). isms that heavily incorporate such synthetic nucleotides into their genomes by evolving E. coli strains that can substitute the syn- Addicting a protein to a synthetic amino acid can potentially be thetic nucleotide analog 5-chlorodeoxyuridine for deoxythymidine achieved by either design or selection. On the design side, Church in their genomes . To do this, they first engineered a thymidine and colleagues engineered the essential adenylate kinase and synthase-deficient strain of E. coli containing the Lactobacillus tyrosyl-tRNA synthetase proteins to require the synthetic leichmannii nucleoside deoxyribosyltransferase gene, enabling L-4′4′-biphenylalanine amino acid in their hydrophobic cores to it to survive in low concentrations of thymine and convert exog- stably fold and thus function . Combining this synthetic amino enous 5-chlorouracil to its nucleotide analog. Next, the authors acid requirement with the classic DAP requirement employed selected for mutants capable of substituting 5-chlorodeoxyuridine in the χ1776 strain produced organisms that escape their desig- for genomic thymine by stressing the cells with media contain- nated environment (that is, grow in the absence of their exogenous ing an increasingly high proportion of 5-chlorouracil relative to –11 ligands) with frequencies of less than 6.4 × 10 . Notably, this level thymine. This produced organisms containing mutations enabling of containment is considerably tighter than the National Institutes them to substitute 5-chlorodeoxyuridine in place of 90% of −86 of Health (NIH)-suggested maximum escape frequency of 10 . On their genomic deoxythymidines and to grow equally well on the selection side, Ellington and colleagues evolved the essential 5-chlorouracil- and thymine-containing solid. Although these β-lactam antibiotic resistance protein TEM β-lactamase to require organisms were not addicted to 5-chlorodeoxyuridine per se, as they either of the synthetic amino acids 3-nitro-L-tyrosine (3nY) or retained their native ability to use thymine, this work demonstrated 3-iodo-L-tyrosine (3iY), achieving escape frequencies of as low as the ability of organisms to efficiently substitute synthetic base −117 less than 5 × 10 . They further demonstrated the generality of pairs in their genome. Conceivably, this work could be expanded this synthetic amino acid-addicted protein by demonstrating that to develop organisms with the ability not just to substitute a natural the evolved 3nY addicted protein retained its addiction in the nucleotide for a synthetic analog but to require it. diverse bacterial species E. coli, Shigella flexneri , Salmonella enterica, Yersinia ruckeri, and Acinetobacter baylyi. Although it is possible to imagine the development of an organism addicted to a single synthetic nucleotide, a more airtight strategy Orthogonal translation machinery and synthetic amino acid incor- would likely be to engineer organisms addicted not just to a single poration become especially useful in the context of organisms synthetic nucleotide but to a complementary pair (Figure 1, bottom with skewed genetic codes, such as the recoded GRO E. coli strain left). This synthetic base pair could likely be further employed to developed by Isaacs and colleagues . This strain was exhaustively encode synthetic amino acids (such as the examples described above) engineered to completely lack the amber (TAG) stop codon and required for the function of an essential protein, thus addicting the may now be readily engineered to rely on unnatural amino acids organism at multiple genetic levels. To date, multiple groups have and their cognate tRNAs that recognize ambers inserted into essen- developed such synthetic base pairs and demonstrated their robust tial genes (Figure 1, center right). In addition to their use of a GRO function and replication in vitro. For example, Hirao and colleagues strain to express rationally designed essential proteins requiring a developed another synthetic base pair, 7-(2-thienyl)-imidazo[4,5- 2 11 synthetic amino acid, as discussed above , Rovner and colleagues b]pyridine (“Ds”) and 2-nitro-4-propynylpyrrole (“Px”) , effectively 12,13 explored this strategy in a more high-throughput way by employing amplie fi d by polymerase chain reaction (PCR) amplic fi ation . multiplex automated genome engineering (MAGE) to introduce Benner and colleagues developed both the deoxyribo and ribo- amber mutations into conserved and functional aromatic residues nucleoside forms of a base-pairing set of synthetic nucleotides, in 22 essential proteins in a GRO strain containing a Methanocaldo- 6-amino-5-nitro-3-(1′-β-d-2′-deoxyribofuranosyl)-2(1H)-pyridone coccus jannaschii tRNA:aminoacyl-tRNA synthetase pair, which in (“Z”) and 2-amino-8-(1′-β-d-2′-deoxyribofuranosyl)-imidazo[1,2- turn translated the amber codons to synthetic phenylalanine-derived a]-1,3,5-triazin-4(8H)-one (“P”), that are effectively incorpo- 9 14 15 amino acids . Several strains containing a single variant protein rated into a double helix , replicated by PCR , and transcribed exhibited near-normal growth in media containing the synthetic by T7 RNA polymerase . Romesberg and colleagues developed −6 17 amino acids but escape frequencies of less than about 10 in non- a separate set of synthetic base pairs, d5SICS-dMMO2 and 18 19 supplemented media. Combining several mutations further reduced d5SICS-dNAM , that likewise are effectively amplified in PCR this escape rate; notably, a strain containing mutations in essential and T7 RNA polymerase in vitro and, in a highly notable advance, residues of the MurG, DnaA, and SerS proteins grew normally in specifically and relatively stably incorporated and maintained Page 3 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 into plasmids replicated in E. coli over multiple generations researchers have engineered synthetic organisms that actively and faithful replication in vivo in E. coli for several generations kill themselves outside of their designated environments. Nature (>15 hours of growth) . employs a similar strategy: many bacteria contain specific toxin: anti-toxin pairs that lead to selfish episome retention . If the anti-toxin activity is lost (that is, through loss of a plasmid carrying Unnatural nucleotides and amino acids are not the only possible the anti-toxin), the toxin will kill the cell. compounds to use for addicting biochemistry. Dependence on unnatural vitamins, cofactors, or other metabolites could also pre- vent growth outside of a specialized environment. In an early exam- The first synthetic, self-killing organisms employed these toxins ple of the small-molecule addiction strategy, Schultz and colleagues in simple kill switches, in which exogenously supplied small engineered a mutant interface between human growth hormone and molecules repressed toxin expression. In the absence of these its receptor containing an interfacial cavity that required binding effectors, the cell would express the toxin, killing itself. In one by the synthetic cofactor 5-chloro-2-trichloromethylimidazole for of the first such studies, Andersson and colleagues expressed the functional ligand-receptor interactions . Though not explicitly membrane-depolarizing toxin hok gene under the tryptophan- intended for synthetic biosafety applications, cells containing this repressible trp promoter in E. coli . In the absence of a high con- mutant pair exhibited a more than 1,000-fold greater response in the centration of tryptophan supplied in media, the organisms would presence of the synthetic molecule, suggesting that this approach express hok, thus killing themselves (Figure 2, top left). Later, of engineering synthetic molecule-dependent ligand–receptor pairs several authors employed LacI-based inverters to construct more could be a useful means by which to restrict their function to a modular kill switches activated by the absence or presence of a specific environment . More recently, researchers have integrated broader range of synthetic molecules. For example, Ramos and this cofactor addiction strategy with computational and selec- colleagues developed a kill switch in P. putida in which the tion methods to readily produce proteins dependent on synthetic 3-methylbenzoate-activated TOL promoter drove production of molecules for function. For example, Lopez and Anderson engi- LacI, which in turn repressed the toxin gef (Figure 2, top center). neered benzothiazole-dependent “SLiDE” mutant proteins for the An absence of 3-methylbenzoate turned off LacI expression and essential genes for phenylalanine tRNA synthetase, tyrosyl tRNA thus gef repression, resulting in gef expression-mediated cell synthetase, methionyl tRNA synthetase, DNA polymerase III, and death.Genes that sequester key metabolites, starving the cells, may adenylate kinase, which require the synthetic ligand benzothiazole be employed as an alternative to toxins. For example, Cantor and to bind as a cofactor to stabilize the hydrophobic core and thus colleagues demonstrated a switch similar to that of Andersson the folded, functional form of the protein (Figure 1, bottom and colleagues in which the killing modality is the overexpression right). Strains containing three such mutants in parallel achieved of streptavidin, which binds and sequesters the key metabolite −11 escape frequencies of less than 3 × 10 after two days in biotin, leading to cell death. Although these single-component culture. kill switches are generally robust, they are subject to failure due to point mutations inactivating the killing mechanism, which −3 −74 generally occur at frequencies of 10 to 10 . An alternative to addicting biomolecules to a synthetic chemical is to instead addict a biochemical pathway. This is achieved by modi- fying such pathways to convert an exogenously supplied synthetic As with the improvements to addiction-based synthetic biosafety compound into a required, otherwise-unavailable metabolite, thus strategies in which multiple dependencies are grouped together to addicting the organism to that molecule. For example, Quandt and ensure low reversion, recent advancements have enabled the devel- colleagues engineered E. coli to require caffeine by first knocking opment of multilayered kill switches that are both more robust and out an essential enzyme in the guanine synthesis pathway, inosine- more dependent on specialized (artificial) environments. Notably, 5′-phosphate dehydrogenase, that catalyzes the formation of the key Collins and colleagues developed several architectures for highly intermediate xanthosine-5′-phosphate from inosine-5′-phosphate . robust kill switches consisting of networks of multiple component They then introduced a refactored Pseudomonas putida alkylxan- switches that interact to reinforce the “killing” state in the absence thine degradation pathway and a Janthinobacterium marseille glu- of a strong, highly specific “don’t kill” environmental signal, pro - tathione S-transferase to demethylate caffeine into xanthine, which viding backup in case one component is mutated or otherwise the strain then converts to xanthosine-5′-phosphate and in turn non-functional . Specifically, their “DEADMAN” switch employs guanine (Figure 1, center left). Although they did not explicitly a bistable regulator with mutually reinforcing feedback loops to apply this study to synthetic biosafety, the authors demonstrated actively drive both the expression of a toxin and the degradation that cell growth was severely limited in the absence of caffeine, of an essential cell protein in the absence of a specific effector suggesting an ability to contain these organisms to a caffeine-rich (Figure 2, bottom left). Likewise, their “PASSCODE” switch environment. Although the strategy of addicting organisms to an requires the presence and absence of a specific combination of syn - exogenous, synthetic substrate via pathway refactoring has seen thetic effectors (that is, an AND/NOT gate) to repress toxin expres- little exploration as a synthetic biosafety strategy, technologies sion (Figure 2, bottom right). In the absence of these specific inputs, 25,26 for refactoring metabolic pathways are fairly well developed , these networked switches achieved escape frequencies below the −731 suggesting that this approach could prove rather straightforward. detectable limit of 10 . Although the reported escape frequen- cies do not explicitly improve upon those of previously reported 7,8,23 kill switches (that is, 30) or addiction strategies , these archi- Strategy 2: kill switches tectures presumably would show improved stability due to their In addition to engineering synthetic organisms that depend pas- bistability, resulting in quicker, more complete killing. sively on specific, synthetic environmental molecules to function, Page 4 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 Figure 2. “Kill” switches activate cell-killing proteins in the absence of specific molecular cues. Simple, early kill-switch architectures commonly used include a switch in which a synthetic molecule directly represses the expression of a toxin (top left) and a switch in which a synthetic molecule drives LacI expression, which in turn represses toxin expression (top center) . The expression of anti-toxins along with their cognate endonuclease anti-toxin has enabled the use of endonucleases in such kill switches, which destroy DNA in addition to simply 34,35 31 killing cells (top right) . Recently, Chan and colleagues developed ultra-robust kill switches . Their DEADMAN switch (bottom left) is a bistable switch that robustly activates two cell-killing modalities in the absence of a synthetic signal molecule, and their PASSCODE switch (bottom right) requires a specific combination of three synthetic molecules to block the production of a cell-killing toxin. Strategy 3: self-destroying kill-switch toxins was explored in the early days of synthetic Although the escape of live organisms represents the most obvious biosafety, their high toxicity required very low expression levels in hazard in synthetic biology, the possibility also exists that synthetic the “off” state, which is difficult with chemically driven promoters . DNA could be released from even a dead cell and make its way into Consequently, nucleases have seen most biosafety use to prevent 3,32 the environment via natural gene transfer . Thus, to make truly the dissemination of plasmids. well-contained synthetic organisms or ecologies, it is important not just to kill escaping organisms but additionally to destroy their Torres and colleagues, for example, employed nuclease-mediated DNA. toxin:anti-toxin strategies to prevent plasmids from spreading through wild populations (Figure 2, top right). Specifically, they A convenient strategy to simultaneously kill cells and destroy their co-expressed the nuclease EcoRI, encoded on a plasmid, and its genes is to employ nucleases as toxins in cellular kill switches such cognate inhibitor, EcoRI methylase, encoded genomically . as those described above. Early studies demonstrated the power Because the inhibitor could be expressed at high levels, it coun- of these nuclease-based kill switches. For example, in a 1994 teracted the potentially leaky expression of the toxin. Should the proof of concept, Ahrenholtz and colleagues demonstrated a heat- plasmid be transferred, EcoRI expression should quickly destroy responsive nuclease-driven kill switch by driving the production its circular form as well as dicing the new host chromosome of the nuc nuclease gene from Serratia marcescens with a ther- (although small linear fragments might still be transferred between moresponsive pL promoter in E. coli . Inducing by heating to cells). Later authors adapted these nuclease–inhibitor pairs to func- 42°C killed cells with an escape rate after 2.5 hours of heating of tion in circuits similar to those described in strategy 2: Gallagher, −5 2 × 10 . However, although this idea of employing nucleases as Isaacs, and colleagues developed a system which expressed EcoRI Page 5 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 constitutively and the inhibitor methylase under an aTc-responsive cataclysms, however unlikely, cannot occur. One way to study these promoter . In the absence of aTc, this single-layer switch achieved potential failure modes would be to simulate them on a microbe- −6 an escape frequency of approximately equal to 2.4 × 10 . Most containing lab-on-a-chip environment such as the microflora- recently, the “DEADMAN” and “PASSCODE” switches employed containing “gut on a chip” developed by Ingber and colleagues . EcoRI as a cell-killing toxin without coexpression of the inhibi- These model environments could be employed to measure the tor, as their bistability multiple reinforcing layers enabled the tight escape frequency of synthetic organisms or DNA into more com- control of EcoRI expression . plex physical and biological environments, such as a community of other microbes, and what the evolutionary stability of engineered Developments in genome editing, specifically by Cas9 and related organisms and communities is over time. systems, provide an alternative strategy for the direct removal and destruction of genes in response to environmental cues. For Finally, there is now the possibility that an escaped, engineered example, Caliando and Voigt describe a system termed “DNAi”, organism can be hunted down … by other engineered organisms. in which a genetically encoded, Cas9-containing circuit degrades That is, recently developed genome editing techniques have specific sections of DNA in response to a molecular effector . yielded so-called “gene drives”, genetic constructs that home Presumably, this circuit could be generalized to respond to the to and overwrite themselves at homologous loci. Church and absence of a synthetic effector, leading to destruction outside of colleagues demonstrated the ability of synthetic gene drives to a laboratory environment. The DNAi system may be employed to overwrite genes in both laboratory strain and wild-type yeast popu- target either plasmids (while not necessarily killing the cells) or the lations, specifically converting the wild-type ADE2 gene (encod- cell’s genomic DNA, including regions necessary for viability. The ing phosphoribosylaminoimidazole carboxylase) to a mutant ade2 authors achieved degradation of both plasmid and genomic target variant. They mated haploid yeast containing the ade2 gene drive −8 regions with escape frequencies of less than 10 . with wild-type (ADE2) haploid yeast of the opposite mating type and found that, while all of the resulting diploids initially inher- Future directions ited copies of both ADE2 and ade2, virtually all (>99%) of their The development of synthetic biosafety techniques that employ successive haploid progeny carried ade2 (rather than 50% carry- addiction, kill switches, and self-destroying modalities has now ing each variant as would be expected), demonstrating successful provided a framework for the development of “safe” synthetic overwriting of the wild-type ADE2 genotype. The authors further organisms and ecosystems. Recent advances in synthetic biology, demonstrated the capability of such gene drives to overwrite a particularly in the manipulation of organisms’ genomes, the devel- second, essential gene (ABD1) and to bias the inheritance of a opment of artificial biomolecules via rational and evolutionary cargo gene carried in cis with the gene overwritten by the gene design, and the construction of robust genetic switches, have in par- drive . Although this example relies on sexual mating, in an asex- ticular enabled the construction of robust safety features with meas- ual population, gene drives may likewise spread through horizon- ured escape frequencies well below those suggested by the NIH. tally transmitted genetic elements, such as broad host range vectors or phage to target specific genes or even the organisms containing However, although the construction of effective biosafety mecha- them. Citorek and colleagues, for example, developed a phage- nisms is underway, the broader question remains: how well will transmitted CRISPR/Cas9-based guided nuclease platform to they work in real-world settings? The escape of even a single, errant target and cleave specific gene sequences, either destroying the bacterium could have widespread consequences, so it is crucial that plasmid on which they were located or introducing cytotoxic these safety mechanisms be well explored. Justifiably, recent atten - genomic double-strand breaks. They demonstrated successful tion has turned from the construction of switches to understanding targeting of bacteria containing the broad-spectrum antibiotic their potential failure modes, particularly in the context of the “real” resistance gene New Delhi metallo-β-lactamase 1, reducing the world of complex physical environments and microbial communi- viability of bacteria containing it by nearly 1,000-fold . This ties outside of the laboratory. To this end, researchers have begun approach could presumably be applied to erase synthetic genes the systematic study of how the function and robustness of syn- or organisms that had escaped their designated environments into thetic biological circuits, including the kill switches discussed here, the surrounding communities. Although gene drives and phage- depend on their environmental context. In an early study, Moser, transmitted gene destruction machinery are viewed as potentially Voight, and colleagues measured the performance (via GFP output) harmful in their own right, the notion of self-propagating “code” of an AND gate and a NOR gate under different media chemistries, that can repair other “code” is prevalent in the software commu- bacterial strains, and reaction scales . They found that whereas the nity and eventually may prove tractable for biotechnology as well. NOR gate’s function output was essentially independent of these For example, a transient gene drive “sweep” through a fermentor changing conditions, the output of the AND gate varied nearly might prevent the unprogrammed escape of genetic material into 20-fold over the range of growth conditions. A similar study that non-engineered organisms. measured the dependence of escape rates of organisms contain- ing the above-described biosafety “devices” on environmental conditions would shed needed light on the environmental robust- Competing interests ness of these features. The authors declared that they had no competing interests. As effective as genetically encoded safeguards might be, there Grant information are always unexpected consequences and therefore it is important AJS is funded on NASA NAI CAN, under grant number: to ensure that “dinosaur escape”, “black swan”, or “Grey goo” NNX15AF46G. Page 6 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 References F1000 recommended 1. Curtiss R III, Inoue M, Pereira D, et al.: Construction and use of safer bacterial an expanded genetic alphabet. Nature. 2014; 509(7500): 385–88. host strains for recombinant DNA research. In Molecular Cloning of Recombinant PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | DNA. Miami Winter Symposia, Miami, FL. 1977; 13: 99–114. 22. Guo Z, Zhou D, Schultz PG: Designing small-molecule switches for protein- Publisher Full Text protein interactions. Science. 2000; 288(5473): 2042–45. PubMed Abstract Publisher Full Text 2. Mandell DJ, Lajoie MJ, Mee MT, et al.: Biocontainment of genetically | modified organisms by synthetic protein design. Nature. 2015; 518(7537): 23. Lopez G, Anderson JC: Synthetic auxotrophs with ligand-dependent 55–60. essential genes for a BL21(DE3) biosafety strain. ACS Synth Biol. 2015; 4(12): PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 1279–86. PubMed Abstract Publisher Full Text F1000 Recommendation 3. Wright O, Delmans M, Stan GB, et al.: GeneGuard: A modular plasmid | | system designed for biosafety. ACS Synth Biol. 2015; 4(3): 307–16. 24. Quandt EM, Hammerling MJ, Summers RM, et al.: Decaffeination and PubMed Abstract Publisher Full Text F1000 Recommendation | | measurement of caffeine content by addicted Escherichia coli with a refactored N-demethylation operon from Pseudomonas putida CBB5. 4. Molin S, Boe L, Jensen LB, et al.: Suicidal genetic elements and their use in ACS Synth Biol. 2013; 2(6): 301–7. biological containment of bacteria. Annu Rev Microbiol. 1993; 47: 139–66. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text 25. Medema MH, van Raaphorst R, Takano E, et al.: Computational tools for the 5. Lang K, Chin JW: Cellular incorporation of unnatural amino acids and synthetic design of biochemical pathways. Nat Rev Microbiol. 2012; 10(3): bioorthogonal labeling of proteins. Chem Rev. 2014; 114(9): 4764–4806. 191–202. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text 6. National Institutes of Health: NIH guidelines for research involving recombinant 26. Ceroni F, Carbonelli P, François JM, et al.: Editorial - synthetic biology: or synthetic nucleic acid molecules. Office of Biotechnology Activities, 78 FR. Engineering complexity and refactoring cell capabilities. Front Bioeng 2013; 66751–66. Biotechnol. 2015; 3: 120. Reference Source PubMed Abstract Publisher Full Text Free Full Text | | 7. Tack DS, Ellefson JW, Thyer R, et al.: Addicting diverse bacteria to a 27. Yamaguchi Y, Park JH, Inouye M: Toxin-antitoxin systems in bacteria and noncanonical amino acid. Nat Chem Biol. 2016; 12(3): 138–140. archaea. Annu Rev Genet. 2011; 45: 61–79. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text 8. Lajoie MJ, Rovner AJ, Goodman DB, et al.: Genomically recoded organisms 28. Molin S, Klemm P, Poulsen LK, et al.: Conditional suicide system for expand biological functions. Science. 2013; 342(6156): 357–60. containment of bacteria and plasmids. Nat Biotechnol. 1987; 5: 1315–18. PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | Publisher Full Text 9. Rovner AJ, Haimovich AD, Katz SR, et al.: Recoded organisms engineered 29. Contreras A, Molin S, Ramos JL: Conditional-suicide containment system for to depend on synthetic amino acids. Nature. 2015; 518(7537): 89–93. bacteria which mineralize aromatics. Appl Environ Microbiol. 1991; 57(5): 1504–8. PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | PubMed Abstract Free Full Text 10. Marliére P, Patrouix J, Döring V, et al.: Chemical evolution of a bacterium’s 30. Szafranski P, Mello CM, Sano T, et al.: A new approach for containment of genome. Angew Chem Int Ed Engl. 2011; 50(31): 7109–14. microorganisms: dual control of streptavidin expression by antisense RNA PubMed Abstract Publisher Full Text F1000 Recommendation and the T7 transcription system. Proc Nat Acad Sci U S A. 1997; 94(4): 1059–63. | | PubMed Abstract Publisher Full Text Free Full Text 11. Kimoto M, Kawai R, Mitsui T, et al.: An unnatural base pair system for efficient | | PCR amplification and functionalization of DNA molecules. Nucleic Acids Res. 31. Chan CT, Lee JW, Cameron DE, et al.: ‘Deadman’ and ‘Passcode’ microbial 2009; 37(2): e14. kill switches for bacterial containment. Nat Chem Biol. 2016; 12(2): 82–6. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 12. Yamashige R, Kimoto M, Takezawa Y, et al.: Highly specific unnatural base pair 32. Nielsen KM, Johnsen PJ, Bensasson D, et al.: Release and persistence of systems as a third base pair for PCR amplification. Nucleic Acids Res. 2012; extracellular DNA in the environment. Environ Biosafety Res. 2007; 6(1–2): 40(6): 2793–806. 37–53. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text 13. Okamoto I, Miyatake Y, Kimoto M, et al.: High Fidelity, Efficiency and 33. Ahrenholtz I, Lorenz MG, Wackernagel W: A conditional suicide system in Functionalization of Ds-Px Unnatural Base Pairs in PCR Amplification for a Escherichia coli based on the intracellular degradation of DNA. Appl Envion Genetic Alphabet Expansion System. ACS Synth Biol. 2016; Article ASAP. Microbiol. 1994; 60(10): 3746–51. PubMed Abstract Publisher Full Text | PubMed Abstract Free Full Text 14. Yang Z, Hutter D, Sheng P, et al.: Artificially expanded genetic information 34. Torres B, Jaenecke S, Timmis KN, et al.: A gene containment strategy based system: a new base pair with an alternative hydrogen bonding pattern. Nucleic on a restriction-modification system. Environ Microbiol. 2000; 2(5): 555–63. Acids Res. 2006; 34(21): 6095–101. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text Free Full Text | | 35. Gallagher RR, Patel JR, Interiano AL, et al.: Multilayered genetic safeguards 15. Yang Y, Chen F, Alvarado JB, et al.: Amplification, mutation, and sequencing of limit growth of microorganisms to defined environments. Nucleic Acids Res. a six-letter synthetic genetic system. J Am Chem Soc. 2011; 133(38): 15105–12. 2015; 43(3): 1945–54. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 16. Kim HJ, Leal NA, Hoshika S, et al.: Ribonucleosides for an artificially expanded 36. Caliando BJ, Voight CA: Targeted DNA degradation using a CRISPR device genetic information system. J Org Chem. 2014; 79(7): 3194–99. stably carried in the host genome. Nat Commun. 2015; 6: 6989. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 17. Leconte AM, Hwang GT, Matsuda S, et al.: Discovery, characterization, and 37. Moser F, Broers NJ, Hartmans S, et al.: Genetic circuit performance under optimization of an unnatural base pair for expansion of the genetic alphabet. conditions relevant for industrial bioreactors. ACS Synth Biol. 2012; 1(11): 555–64. J Am Chem Soc. 2008; 130(7): 2336–43. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 38. Kim HJ, Huh D, Hamilton G, et al.: Human gut-on-a-chip inhabited by microbial 18. Seo YJ, Hwang GT, Ordoukhanian P, et al.: Optimization of an unnatural base flora that experiences intestinal peristalsis-like motions and flow. Lab Chip. pair toward natural-like replication. J Am Chem Soc. 2009; 131(9): 3246–52. 2012; 12(12): 2165–74. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text 19. Malyshev DA, Seo YJ, Ordoukhanian P, et al.: PCR with an expanded genetic 39. DiCarlo JE, Chavez A, Dietz SL, et al.: Safeguarding CRISPR-Cas9 gene alphabet. J Am Chem Soc. 2009; 131(41): 14620–21. drives in yeast. Nat Biotechnol. 2015; 33(12): 1250–1255. PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 20. Seo YJ, Matsuda S, Romesberg FE: Transcription of an expanded genetic 40. Citorek RJ, Mimee M, Lu TK: Sequence-specific antimicrobials using alphabet. J Am Chem Soc. 2009; 131(14): 5046–47. efficiently delivered RNA-guided nucleases. Nat Biotechnol. 2014; 32(11): PubMed Abstract Publisher Full Text Free Full Text | | 1141–45. 21. Malyshev DA, Dhami K, Lavergne T, et al.: A semi-synthetic organism with PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | Page 7 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 Open Peer Review Current Referee Status: Editorial Note on the Review Process F1000 Faculty Reviews are commissioned from members of the prestigious F1000 Faculty and are edited as a service to readers. In order to make these reviews as comprehensive and accessible as possible, the referees provide input before publication and only the final, revised version is published. The referees who approved the final version are listed with their names and affiliations but without their reports on earlier versions (any comments will already have been addressed in the published version). The referees who approved this article are: Version 1 1 Ichiro Hirao, Synthetic Molecular Biology Team, Institute of Bioengineering and Nanotechnology, Yokohama, Kanagawa, 230-0045, Singapore Competing Interests: No competing interests were disclosed. 2 Shota Atsumi, University of California, Davis, Davis, CA, 95616, USA Competing Interests: No competing interests were disclosed. F1000Research Page 8 of 8 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png F1000Research Pubmed Central

Recent advances in synthetic biosafety

F1000Research , Volume 5 – Aug 31, 2016

Loading next page...
 
/lp/pubmed-central/recent-advances-in-synthetic-biosafety-5Ag6F2foUt

References (44)

Publisher
Pubmed Central
Copyright
Copyright: © 2016 Simon AJ and Ellington AD
ISSN
2046-1402
eISSN
2046-1402
DOI
10.12688/f1000research.8365.1
Publisher site
See Article on Publisher Site

Abstract

Invited Referees Synthetically engineered organisms hold promise for a broad range of medical, 1 2 environmental, and industrial applications. Organisms can potentially be designed, for example, for the inexpensive and environmentally benign version 1 synthesis of pharmaceuticals and industrial chemicals, for the cleanup of published environmental pollutants, and potentially even for biomedical applications such 31 Aug 2016 as the targeting of specific diseases or tissues. However, the use of synthetically engineered organisms comes with several reasonable safety concerns, one of which is that the organisms or their genes could escape their F1000 Faculty Reviews are commissioned intended habitats and cause environmental disruption. Here we review key from members of the prestigious F1000 recent developments in this emerging field of synthetic biocontainment and Faculty. In order to make these reviews as discuss further developments that might be necessary for the widespread use comprehensive and accessible as possible, of synthetic organisms. Specifically, we discuss the history and modern peer review takes place before publication; the development of three strategies for the containment of synthetic microbes: addiction to an exogenously supplied ligand; self-killing outside of a designated referees are listed below, but their reports are environment; and self-destroying encoded DNA circuitry outside of a not formally published. designated environment. 1 Shota Atsumi, University of California, Davis USA 2 Ichiro Hirao, Institute of Bioengineering and Nanotechnology Singapore Discuss this article Comments (0) Corresponding author: Andrew D. Ellington (ellingtonlab@gmail.com) How to cite this article: Simon AJ and Ellington AD. Recent advances in synthetic biosafety [version 1; referees: 2 approved] F1000Research 2016, 5(F1000 Faculty Rev):2118 (doi: 10.12688/f1000research.8365.1) Copyright: © 2016 Simon AJ and Ellington AD. This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Grant information: AJS is funded on NASA NAI CAN, under grant number: NNX15AF46G Competing interests: The authors declared that they had no competing interests. First published: 31 Aug 2016, 5(F1000 Faculty Rev):2118 (doi: 10.12688/f1000research.8365.1) F1000Research Page 1 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 environment, and real-life organisms addicted to natural chemi- Strategy 1: addiction cals could likely escape the same way. Wright and colleagues, for As with the lysine-deficient fictional dinosaurs in the book Jurassic example, demonstrated that ΔthyA mutants grew readily in media Park, it may be possible to employ addiction strategies to make it lacking specifically supplied thymidine but supplemented with a difficult for organisms to survive outside of their designated habi - small amount of sterilized soil, demonstrating that environmental tats. This biocontainment strategy dates back to the earliest days nutrients may complement metabolic deficiencies . Furthermore, of cloning with the development of the specialized Escherichia the engineered inability to generate a key metabolite often inhibits coli strain χ1776, which lacked functional aspartate-semialdehyde the growth of an organism even when the metabolite is heavily dehydrogenase (asd), (L-delta-1-tetrahydrodipicolinate synthetase) supplemented, rendering such engineered organisms ill-suited for (dapD), and thymidylate synthetase (thyA) genes and thus required 1,4 use in bioreactors and many other applications . Thus, the more their products, diaminopimelic acid (DAP) and thymine or thy- modern version of this strategy is to engineer organisms that depend midine, to survive . Because this strategy is so straightforward— not on a natural compound but on a synthetic one. If engineered requiring only the creation of organisms deficient in the ability to organisms require a synthetic compound for survival, they will be produce a key metabolite and this is readily achieved by targeted unable to survive outside of a laboratory or other highly special- and random mutagenesis—it has remained commonly employed ized environment in which the chemical is supplied. However, this and integrated into more sophisticated synthetic biosafety platforms 2,3 engineering feat is more difficult than simply generating organisms through recent times (Figure 1, top left). with a “broken” ability to produce a naturally occurring metabolite. Despite the straightforwardness of this simple addiction strategy, Instead, it requires engineering some sort of metabolic or functional it was not sufficient to contain Jurassic Park’s resurrected fic - connection to a synthetic chemical that was previously irrelevant to tional dinosaurs because of the ready availability of lysine in the the organism’s biology. Figure 1. A common strategy for the containment of synthetic organisms is to engineer them to require an exogenously supplied ligand. Common methods for achieving this include (clockwise from top left) knocking out a required gene and exogenously supplying the 1 9 gene product , requiring the amber-mediated incorporation of a synthetic amino acid for essential protein production , requiring the amber- 2,7 mediated incorporation of a synthetic amino acid for essential protein function , requiring a synthetic molecule as a cofactor for protein function , requiring a synthetic nucleotide for essential gene replication or translation, and requiring a synthetic molecule as a precursor for a key metabolite . Page 2 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 One strategy by which researchers have adapted organismal supplemented media but “escaped” with rates below the detectable −11 9 metabolism to require synthetic compounds is by altering the 4 × 10 frequency after 20 days of growth . This very low escape genetic code to incorporate and even require unnatural amino acids. frequency arises from the dependence of multiple proteins on these In this regard, the development of so-called orthogonal translation synthetic amino acids rather than the dependence per se of a single machinery has advanced to the point where the incorporation of protein: whereas the probability of one amber codon reverting to a −7 unnatural amino acids into proteins, at least across from amber natural codon is moderate (~10 ), the probability of getting three codons, is relatively straightforward . Although an organism can in parallel is much lower. readily be rendered dependent upon suppression of an amber codon (for example, by introducing the amber codon into an essen- A strategy to render organisms dependent on synthetic molecules tial protein), rendering a protein (and thus its organism) dependent at an even deeper level than addicting them to synthetic tRNAs on a specific synthetic amino acid is considerably more compli - or amino acids is to addict them to synthetic nucleotides. Mut- cated, as it requires the specific, introduced amino acid to be zel and colleagues demonstrated the ability to engineer organ- necessary for protein function (Figure 1, top right). isms that heavily incorporate such synthetic nucleotides into their genomes by evolving E. coli strains that can substitute the syn- Addicting a protein to a synthetic amino acid can potentially be thetic nucleotide analog 5-chlorodeoxyuridine for deoxythymidine achieved by either design or selection. On the design side, Church in their genomes . To do this, they first engineered a thymidine and colleagues engineered the essential adenylate kinase and synthase-deficient strain of E. coli containing the Lactobacillus tyrosyl-tRNA synthetase proteins to require the synthetic leichmannii nucleoside deoxyribosyltransferase gene, enabling L-4′4′-biphenylalanine amino acid in their hydrophobic cores to it to survive in low concentrations of thymine and convert exog- stably fold and thus function . Combining this synthetic amino enous 5-chlorouracil to its nucleotide analog. Next, the authors acid requirement with the classic DAP requirement employed selected for mutants capable of substituting 5-chlorodeoxyuridine in the χ1776 strain produced organisms that escape their desig- for genomic thymine by stressing the cells with media contain- nated environment (that is, grow in the absence of their exogenous ing an increasingly high proportion of 5-chlorouracil relative to –11 ligands) with frequencies of less than 6.4 × 10 . Notably, this level thymine. This produced organisms containing mutations enabling of containment is considerably tighter than the National Institutes them to substitute 5-chlorodeoxyuridine in place of 90% of −86 of Health (NIH)-suggested maximum escape frequency of 10 . On their genomic deoxythymidines and to grow equally well on the selection side, Ellington and colleagues evolved the essential 5-chlorouracil- and thymine-containing solid. Although these β-lactam antibiotic resistance protein TEM β-lactamase to require organisms were not addicted to 5-chlorodeoxyuridine per se, as they either of the synthetic amino acids 3-nitro-L-tyrosine (3nY) or retained their native ability to use thymine, this work demonstrated 3-iodo-L-tyrosine (3iY), achieving escape frequencies of as low as the ability of organisms to efficiently substitute synthetic base −117 less than 5 × 10 . They further demonstrated the generality of pairs in their genome. Conceivably, this work could be expanded this synthetic amino acid-addicted protein by demonstrating that to develop organisms with the ability not just to substitute a natural the evolved 3nY addicted protein retained its addiction in the nucleotide for a synthetic analog but to require it. diverse bacterial species E. coli, Shigella flexneri , Salmonella enterica, Yersinia ruckeri, and Acinetobacter baylyi. Although it is possible to imagine the development of an organism addicted to a single synthetic nucleotide, a more airtight strategy Orthogonal translation machinery and synthetic amino acid incor- would likely be to engineer organisms addicted not just to a single poration become especially useful in the context of organisms synthetic nucleotide but to a complementary pair (Figure 1, bottom with skewed genetic codes, such as the recoded GRO E. coli strain left). This synthetic base pair could likely be further employed to developed by Isaacs and colleagues . This strain was exhaustively encode synthetic amino acids (such as the examples described above) engineered to completely lack the amber (TAG) stop codon and required for the function of an essential protein, thus addicting the may now be readily engineered to rely on unnatural amino acids organism at multiple genetic levels. To date, multiple groups have and their cognate tRNAs that recognize ambers inserted into essen- developed such synthetic base pairs and demonstrated their robust tial genes (Figure 1, center right). In addition to their use of a GRO function and replication in vitro. For example, Hirao and colleagues strain to express rationally designed essential proteins requiring a developed another synthetic base pair, 7-(2-thienyl)-imidazo[4,5- 2 11 synthetic amino acid, as discussed above , Rovner and colleagues b]pyridine (“Ds”) and 2-nitro-4-propynylpyrrole (“Px”) , effectively 12,13 explored this strategy in a more high-throughput way by employing amplie fi d by polymerase chain reaction (PCR) amplic fi ation . multiplex automated genome engineering (MAGE) to introduce Benner and colleagues developed both the deoxyribo and ribo- amber mutations into conserved and functional aromatic residues nucleoside forms of a base-pairing set of synthetic nucleotides, in 22 essential proteins in a GRO strain containing a Methanocaldo- 6-amino-5-nitro-3-(1′-β-d-2′-deoxyribofuranosyl)-2(1H)-pyridone coccus jannaschii tRNA:aminoacyl-tRNA synthetase pair, which in (“Z”) and 2-amino-8-(1′-β-d-2′-deoxyribofuranosyl)-imidazo[1,2- turn translated the amber codons to synthetic phenylalanine-derived a]-1,3,5-triazin-4(8H)-one (“P”), that are effectively incorpo- 9 14 15 amino acids . Several strains containing a single variant protein rated into a double helix , replicated by PCR , and transcribed exhibited near-normal growth in media containing the synthetic by T7 RNA polymerase . Romesberg and colleagues developed −6 17 amino acids but escape frequencies of less than about 10 in non- a separate set of synthetic base pairs, d5SICS-dMMO2 and 18 19 supplemented media. Combining several mutations further reduced d5SICS-dNAM , that likewise are effectively amplified in PCR this escape rate; notably, a strain containing mutations in essential and T7 RNA polymerase in vitro and, in a highly notable advance, residues of the MurG, DnaA, and SerS proteins grew normally in specifically and relatively stably incorporated and maintained Page 3 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 into plasmids replicated in E. coli over multiple generations researchers have engineered synthetic organisms that actively and faithful replication in vivo in E. coli for several generations kill themselves outside of their designated environments. Nature (>15 hours of growth) . employs a similar strategy: many bacteria contain specific toxin: anti-toxin pairs that lead to selfish episome retention . If the anti-toxin activity is lost (that is, through loss of a plasmid carrying Unnatural nucleotides and amino acids are not the only possible the anti-toxin), the toxin will kill the cell. compounds to use for addicting biochemistry. Dependence on unnatural vitamins, cofactors, or other metabolites could also pre- vent growth outside of a specialized environment. In an early exam- The first synthetic, self-killing organisms employed these toxins ple of the small-molecule addiction strategy, Schultz and colleagues in simple kill switches, in which exogenously supplied small engineered a mutant interface between human growth hormone and molecules repressed toxin expression. In the absence of these its receptor containing an interfacial cavity that required binding effectors, the cell would express the toxin, killing itself. In one by the synthetic cofactor 5-chloro-2-trichloromethylimidazole for of the first such studies, Andersson and colleagues expressed the functional ligand-receptor interactions . Though not explicitly membrane-depolarizing toxin hok gene under the tryptophan- intended for synthetic biosafety applications, cells containing this repressible trp promoter in E. coli . In the absence of a high con- mutant pair exhibited a more than 1,000-fold greater response in the centration of tryptophan supplied in media, the organisms would presence of the synthetic molecule, suggesting that this approach express hok, thus killing themselves (Figure 2, top left). Later, of engineering synthetic molecule-dependent ligand–receptor pairs several authors employed LacI-based inverters to construct more could be a useful means by which to restrict their function to a modular kill switches activated by the absence or presence of a specific environment . More recently, researchers have integrated broader range of synthetic molecules. For example, Ramos and this cofactor addiction strategy with computational and selec- colleagues developed a kill switch in P. putida in which the tion methods to readily produce proteins dependent on synthetic 3-methylbenzoate-activated TOL promoter drove production of molecules for function. For example, Lopez and Anderson engi- LacI, which in turn repressed the toxin gef (Figure 2, top center). neered benzothiazole-dependent “SLiDE” mutant proteins for the An absence of 3-methylbenzoate turned off LacI expression and essential genes for phenylalanine tRNA synthetase, tyrosyl tRNA thus gef repression, resulting in gef expression-mediated cell synthetase, methionyl tRNA synthetase, DNA polymerase III, and death.Genes that sequester key metabolites, starving the cells, may adenylate kinase, which require the synthetic ligand benzothiazole be employed as an alternative to toxins. For example, Cantor and to bind as a cofactor to stabilize the hydrophobic core and thus colleagues demonstrated a switch similar to that of Andersson the folded, functional form of the protein (Figure 1, bottom and colleagues in which the killing modality is the overexpression right). Strains containing three such mutants in parallel achieved of streptavidin, which binds and sequesters the key metabolite −11 escape frequencies of less than 3 × 10 after two days in biotin, leading to cell death. Although these single-component culture. kill switches are generally robust, they are subject to failure due to point mutations inactivating the killing mechanism, which −3 −74 generally occur at frequencies of 10 to 10 . An alternative to addicting biomolecules to a synthetic chemical is to instead addict a biochemical pathway. This is achieved by modi- fying such pathways to convert an exogenously supplied synthetic As with the improvements to addiction-based synthetic biosafety compound into a required, otherwise-unavailable metabolite, thus strategies in which multiple dependencies are grouped together to addicting the organism to that molecule. For example, Quandt and ensure low reversion, recent advancements have enabled the devel- colleagues engineered E. coli to require caffeine by first knocking opment of multilayered kill switches that are both more robust and out an essential enzyme in the guanine synthesis pathway, inosine- more dependent on specialized (artificial) environments. Notably, 5′-phosphate dehydrogenase, that catalyzes the formation of the key Collins and colleagues developed several architectures for highly intermediate xanthosine-5′-phosphate from inosine-5′-phosphate . robust kill switches consisting of networks of multiple component They then introduced a refactored Pseudomonas putida alkylxan- switches that interact to reinforce the “killing” state in the absence thine degradation pathway and a Janthinobacterium marseille glu- of a strong, highly specific “don’t kill” environmental signal, pro - tathione S-transferase to demethylate caffeine into xanthine, which viding backup in case one component is mutated or otherwise the strain then converts to xanthosine-5′-phosphate and in turn non-functional . Specifically, their “DEADMAN” switch employs guanine (Figure 1, center left). Although they did not explicitly a bistable regulator with mutually reinforcing feedback loops to apply this study to synthetic biosafety, the authors demonstrated actively drive both the expression of a toxin and the degradation that cell growth was severely limited in the absence of caffeine, of an essential cell protein in the absence of a specific effector suggesting an ability to contain these organisms to a caffeine-rich (Figure 2, bottom left). Likewise, their “PASSCODE” switch environment. Although the strategy of addicting organisms to an requires the presence and absence of a specific combination of syn - exogenous, synthetic substrate via pathway refactoring has seen thetic effectors (that is, an AND/NOT gate) to repress toxin expres- little exploration as a synthetic biosafety strategy, technologies sion (Figure 2, bottom right). In the absence of these specific inputs, 25,26 for refactoring metabolic pathways are fairly well developed , these networked switches achieved escape frequencies below the −731 suggesting that this approach could prove rather straightforward. detectable limit of 10 . Although the reported escape frequen- cies do not explicitly improve upon those of previously reported 7,8,23 kill switches (that is, 30) or addiction strategies , these archi- Strategy 2: kill switches tectures presumably would show improved stability due to their In addition to engineering synthetic organisms that depend pas- bistability, resulting in quicker, more complete killing. sively on specific, synthetic environmental molecules to function, Page 4 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 Figure 2. “Kill” switches activate cell-killing proteins in the absence of specific molecular cues. Simple, early kill-switch architectures commonly used include a switch in which a synthetic molecule directly represses the expression of a toxin (top left) and a switch in which a synthetic molecule drives LacI expression, which in turn represses toxin expression (top center) . The expression of anti-toxins along with their cognate endonuclease anti-toxin has enabled the use of endonucleases in such kill switches, which destroy DNA in addition to simply 34,35 31 killing cells (top right) . Recently, Chan and colleagues developed ultra-robust kill switches . Their DEADMAN switch (bottom left) is a bistable switch that robustly activates two cell-killing modalities in the absence of a synthetic signal molecule, and their PASSCODE switch (bottom right) requires a specific combination of three synthetic molecules to block the production of a cell-killing toxin. Strategy 3: self-destroying kill-switch toxins was explored in the early days of synthetic Although the escape of live organisms represents the most obvious biosafety, their high toxicity required very low expression levels in hazard in synthetic biology, the possibility also exists that synthetic the “off” state, which is difficult with chemically driven promoters . DNA could be released from even a dead cell and make its way into Consequently, nucleases have seen most biosafety use to prevent 3,32 the environment via natural gene transfer . Thus, to make truly the dissemination of plasmids. well-contained synthetic organisms or ecologies, it is important not just to kill escaping organisms but additionally to destroy their Torres and colleagues, for example, employed nuclease-mediated DNA. toxin:anti-toxin strategies to prevent plasmids from spreading through wild populations (Figure 2, top right). Specifically, they A convenient strategy to simultaneously kill cells and destroy their co-expressed the nuclease EcoRI, encoded on a plasmid, and its genes is to employ nucleases as toxins in cellular kill switches such cognate inhibitor, EcoRI methylase, encoded genomically . as those described above. Early studies demonstrated the power Because the inhibitor could be expressed at high levels, it coun- of these nuclease-based kill switches. For example, in a 1994 teracted the potentially leaky expression of the toxin. Should the proof of concept, Ahrenholtz and colleagues demonstrated a heat- plasmid be transferred, EcoRI expression should quickly destroy responsive nuclease-driven kill switch by driving the production its circular form as well as dicing the new host chromosome of the nuc nuclease gene from Serratia marcescens with a ther- (although small linear fragments might still be transferred between moresponsive pL promoter in E. coli . Inducing by heating to cells). Later authors adapted these nuclease–inhibitor pairs to func- 42°C killed cells with an escape rate after 2.5 hours of heating of tion in circuits similar to those described in strategy 2: Gallagher, −5 2 × 10 . However, although this idea of employing nucleases as Isaacs, and colleagues developed a system which expressed EcoRI Page 5 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 constitutively and the inhibitor methylase under an aTc-responsive cataclysms, however unlikely, cannot occur. One way to study these promoter . In the absence of aTc, this single-layer switch achieved potential failure modes would be to simulate them on a microbe- −6 an escape frequency of approximately equal to 2.4 × 10 . Most containing lab-on-a-chip environment such as the microflora- recently, the “DEADMAN” and “PASSCODE” switches employed containing “gut on a chip” developed by Ingber and colleagues . EcoRI as a cell-killing toxin without coexpression of the inhibi- These model environments could be employed to measure the tor, as their bistability multiple reinforcing layers enabled the tight escape frequency of synthetic organisms or DNA into more com- control of EcoRI expression . plex physical and biological environments, such as a community of other microbes, and what the evolutionary stability of engineered Developments in genome editing, specifically by Cas9 and related organisms and communities is over time. systems, provide an alternative strategy for the direct removal and destruction of genes in response to environmental cues. For Finally, there is now the possibility that an escaped, engineered example, Caliando and Voigt describe a system termed “DNAi”, organism can be hunted down … by other engineered organisms. in which a genetically encoded, Cas9-containing circuit degrades That is, recently developed genome editing techniques have specific sections of DNA in response to a molecular effector . yielded so-called “gene drives”, genetic constructs that home Presumably, this circuit could be generalized to respond to the to and overwrite themselves at homologous loci. Church and absence of a synthetic effector, leading to destruction outside of colleagues demonstrated the ability of synthetic gene drives to a laboratory environment. The DNAi system may be employed to overwrite genes in both laboratory strain and wild-type yeast popu- target either plasmids (while not necessarily killing the cells) or the lations, specifically converting the wild-type ADE2 gene (encod- cell’s genomic DNA, including regions necessary for viability. The ing phosphoribosylaminoimidazole carboxylase) to a mutant ade2 authors achieved degradation of both plasmid and genomic target variant. They mated haploid yeast containing the ade2 gene drive −8 regions with escape frequencies of less than 10 . with wild-type (ADE2) haploid yeast of the opposite mating type and found that, while all of the resulting diploids initially inher- Future directions ited copies of both ADE2 and ade2, virtually all (>99%) of their The development of synthetic biosafety techniques that employ successive haploid progeny carried ade2 (rather than 50% carry- addiction, kill switches, and self-destroying modalities has now ing each variant as would be expected), demonstrating successful provided a framework for the development of “safe” synthetic overwriting of the wild-type ADE2 genotype. The authors further organisms and ecosystems. Recent advances in synthetic biology, demonstrated the capability of such gene drives to overwrite a particularly in the manipulation of organisms’ genomes, the devel- second, essential gene (ABD1) and to bias the inheritance of a opment of artificial biomolecules via rational and evolutionary cargo gene carried in cis with the gene overwritten by the gene design, and the construction of robust genetic switches, have in par- drive . Although this example relies on sexual mating, in an asex- ticular enabled the construction of robust safety features with meas- ual population, gene drives may likewise spread through horizon- ured escape frequencies well below those suggested by the NIH. tally transmitted genetic elements, such as broad host range vectors or phage to target specific genes or even the organisms containing However, although the construction of effective biosafety mecha- them. Citorek and colleagues, for example, developed a phage- nisms is underway, the broader question remains: how well will transmitted CRISPR/Cas9-based guided nuclease platform to they work in real-world settings? The escape of even a single, errant target and cleave specific gene sequences, either destroying the bacterium could have widespread consequences, so it is crucial that plasmid on which they were located or introducing cytotoxic these safety mechanisms be well explored. Justifiably, recent atten - genomic double-strand breaks. They demonstrated successful tion has turned from the construction of switches to understanding targeting of bacteria containing the broad-spectrum antibiotic their potential failure modes, particularly in the context of the “real” resistance gene New Delhi metallo-β-lactamase 1, reducing the world of complex physical environments and microbial communi- viability of bacteria containing it by nearly 1,000-fold . This ties outside of the laboratory. To this end, researchers have begun approach could presumably be applied to erase synthetic genes the systematic study of how the function and robustness of syn- or organisms that had escaped their designated environments into thetic biological circuits, including the kill switches discussed here, the surrounding communities. Although gene drives and phage- depend on their environmental context. In an early study, Moser, transmitted gene destruction machinery are viewed as potentially Voight, and colleagues measured the performance (via GFP output) harmful in their own right, the notion of self-propagating “code” of an AND gate and a NOR gate under different media chemistries, that can repair other “code” is prevalent in the software commu- bacterial strains, and reaction scales . They found that whereas the nity and eventually may prove tractable for biotechnology as well. NOR gate’s function output was essentially independent of these For example, a transient gene drive “sweep” through a fermentor changing conditions, the output of the AND gate varied nearly might prevent the unprogrammed escape of genetic material into 20-fold over the range of growth conditions. A similar study that non-engineered organisms. measured the dependence of escape rates of organisms contain- ing the above-described biosafety “devices” on environmental conditions would shed needed light on the environmental robust- Competing interests ness of these features. The authors declared that they had no competing interests. As effective as genetically encoded safeguards might be, there Grant information are always unexpected consequences and therefore it is important AJS is funded on NASA NAI CAN, under grant number: to ensure that “dinosaur escape”, “black swan”, or “Grey goo” NNX15AF46G. Page 6 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 References F1000 recommended 1. Curtiss R III, Inoue M, Pereira D, et al.: Construction and use of safer bacterial an expanded genetic alphabet. Nature. 2014; 509(7500): 385–88. host strains for recombinant DNA research. In Molecular Cloning of Recombinant PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | DNA. Miami Winter Symposia, Miami, FL. 1977; 13: 99–114. 22. Guo Z, Zhou D, Schultz PG: Designing small-molecule switches for protein- Publisher Full Text protein interactions. Science. 2000; 288(5473): 2042–45. PubMed Abstract Publisher Full Text 2. Mandell DJ, Lajoie MJ, Mee MT, et al.: Biocontainment of genetically | modified organisms by synthetic protein design. Nature. 2015; 518(7537): 23. Lopez G, Anderson JC: Synthetic auxotrophs with ligand-dependent 55–60. essential genes for a BL21(DE3) biosafety strain. ACS Synth Biol. 2015; 4(12): PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 1279–86. PubMed Abstract Publisher Full Text F1000 Recommendation 3. Wright O, Delmans M, Stan GB, et al.: GeneGuard: A modular plasmid | | system designed for biosafety. ACS Synth Biol. 2015; 4(3): 307–16. 24. Quandt EM, Hammerling MJ, Summers RM, et al.: Decaffeination and PubMed Abstract Publisher Full Text F1000 Recommendation | | measurement of caffeine content by addicted Escherichia coli with a refactored N-demethylation operon from Pseudomonas putida CBB5. 4. Molin S, Boe L, Jensen LB, et al.: Suicidal genetic elements and their use in ACS Synth Biol. 2013; 2(6): 301–7. biological containment of bacteria. Annu Rev Microbiol. 1993; 47: 139–66. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text 25. Medema MH, van Raaphorst R, Takano E, et al.: Computational tools for the 5. Lang K, Chin JW: Cellular incorporation of unnatural amino acids and synthetic design of biochemical pathways. Nat Rev Microbiol. 2012; 10(3): bioorthogonal labeling of proteins. Chem Rev. 2014; 114(9): 4764–4806. 191–202. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text 6. National Institutes of Health: NIH guidelines for research involving recombinant 26. Ceroni F, Carbonelli P, François JM, et al.: Editorial - synthetic biology: or synthetic nucleic acid molecules. Office of Biotechnology Activities, 78 FR. Engineering complexity and refactoring cell capabilities. Front Bioeng 2013; 66751–66. Biotechnol. 2015; 3: 120. Reference Source PubMed Abstract Publisher Full Text Free Full Text | | 7. Tack DS, Ellefson JW, Thyer R, et al.: Addicting diverse bacteria to a 27. Yamaguchi Y, Park JH, Inouye M: Toxin-antitoxin systems in bacteria and noncanonical amino acid. Nat Chem Biol. 2016; 12(3): 138–140. archaea. Annu Rev Genet. 2011; 45: 61–79. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text 8. Lajoie MJ, Rovner AJ, Goodman DB, et al.: Genomically recoded organisms 28. Molin S, Klemm P, Poulsen LK, et al.: Conditional suicide system for expand biological functions. Science. 2013; 342(6156): 357–60. containment of bacteria and plasmids. Nat Biotechnol. 1987; 5: 1315–18. PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | Publisher Full Text 9. Rovner AJ, Haimovich AD, Katz SR, et al.: Recoded organisms engineered 29. Contreras A, Molin S, Ramos JL: Conditional-suicide containment system for to depend on synthetic amino acids. Nature. 2015; 518(7537): 89–93. bacteria which mineralize aromatics. Appl Environ Microbiol. 1991; 57(5): 1504–8. PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | PubMed Abstract Free Full Text 10. Marliére P, Patrouix J, Döring V, et al.: Chemical evolution of a bacterium’s 30. Szafranski P, Mello CM, Sano T, et al.: A new approach for containment of genome. Angew Chem Int Ed Engl. 2011; 50(31): 7109–14. microorganisms: dual control of streptavidin expression by antisense RNA PubMed Abstract Publisher Full Text F1000 Recommendation and the T7 transcription system. Proc Nat Acad Sci U S A. 1997; 94(4): 1059–63. | | PubMed Abstract Publisher Full Text Free Full Text 11. Kimoto M, Kawai R, Mitsui T, et al.: An unnatural base pair system for efficient | | PCR amplification and functionalization of DNA molecules. Nucleic Acids Res. 31. Chan CT, Lee JW, Cameron DE, et al.: ‘Deadman’ and ‘Passcode’ microbial 2009; 37(2): e14. kill switches for bacterial containment. Nat Chem Biol. 2016; 12(2): 82–6. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 12. Yamashige R, Kimoto M, Takezawa Y, et al.: Highly specific unnatural base pair 32. Nielsen KM, Johnsen PJ, Bensasson D, et al.: Release and persistence of systems as a third base pair for PCR amplification. Nucleic Acids Res. 2012; extracellular DNA in the environment. Environ Biosafety Res. 2007; 6(1–2): 40(6): 2793–806. 37–53. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text 13. Okamoto I, Miyatake Y, Kimoto M, et al.: High Fidelity, Efficiency and 33. Ahrenholtz I, Lorenz MG, Wackernagel W: A conditional suicide system in Functionalization of Ds-Px Unnatural Base Pairs in PCR Amplification for a Escherichia coli based on the intracellular degradation of DNA. Appl Envion Genetic Alphabet Expansion System. ACS Synth Biol. 2016; Article ASAP. Microbiol. 1994; 60(10): 3746–51. PubMed Abstract Publisher Full Text | PubMed Abstract Free Full Text 14. Yang Z, Hutter D, Sheng P, et al.: Artificially expanded genetic information 34. Torres B, Jaenecke S, Timmis KN, et al.: A gene containment strategy based system: a new base pair with an alternative hydrogen bonding pattern. Nucleic on a restriction-modification system. Environ Microbiol. 2000; 2(5): 555–63. Acids Res. 2006; 34(21): 6095–101. PubMed Abstract Publisher Full Text PubMed Abstract Publisher Full Text Free Full Text | | 35. Gallagher RR, Patel JR, Interiano AL, et al.: Multilayered genetic safeguards 15. Yang Y, Chen F, Alvarado JB, et al.: Amplification, mutation, and sequencing of limit growth of microorganisms to defined environments. Nucleic Acids Res. a six-letter synthetic genetic system. J Am Chem Soc. 2011; 133(38): 15105–12. 2015; 43(3): 1945–54. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 16. Kim HJ, Leal NA, Hoshika S, et al.: Ribonucleosides for an artificially expanded 36. Caliando BJ, Voight CA: Targeted DNA degradation using a CRISPR device genetic information system. J Org Chem. 2014; 79(7): 3194–99. stably carried in the host genome. Nat Commun. 2015; 6: 6989. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 17. Leconte AM, Hwang GT, Matsuda S, et al.: Discovery, characterization, and 37. Moser F, Broers NJ, Hartmans S, et al.: Genetic circuit performance under optimization of an unnatural base pair for expansion of the genetic alphabet. conditions relevant for industrial bioreactors. ACS Synth Biol. 2012; 1(11): 555–64. J Am Chem Soc. 2008; 130(7): 2336–43. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 38. Kim HJ, Huh D, Hamilton G, et al.: Human gut-on-a-chip inhabited by microbial 18. Seo YJ, Hwang GT, Ordoukhanian P, et al.: Optimization of an unnatural base flora that experiences intestinal peristalsis-like motions and flow. Lab Chip. pair toward natural-like replication. J Am Chem Soc. 2009; 131(9): 3246–52. 2012; 12(12): 2165–74. PubMed Abstract Publisher Full Text Free Full Text | | PubMed Abstract Publisher Full Text 19. Malyshev DA, Seo YJ, Ordoukhanian P, et al.: PCR with an expanded genetic 39. DiCarlo JE, Chavez A, Dietz SL, et al.: Safeguarding CRISPR-Cas9 gene alphabet. J Am Chem Soc. 2009; 131(41): 14620–21. drives in yeast. Nat Biotechnol. 2015; 33(12): 1250–1255. PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | 20. Seo YJ, Matsuda S, Romesberg FE: Transcription of an expanded genetic 40. Citorek RJ, Mimee M, Lu TK: Sequence-specific antimicrobials using alphabet. J Am Chem Soc. 2009; 131(14): 5046–47. efficiently delivered RNA-guided nucleases. Nat Biotechnol. 2014; 32(11): PubMed Abstract Publisher Full Text Free Full Text | | 1141–45. 21. Malyshev DA, Dhami K, Lavergne T, et al.: A semi-synthetic organism with PubMed Abstract Publisher Full Text Free Full Text F1000 Recommendation | | | Page 7 of 8 F1000Research 2016, 5(F1000 Faculty Rev):2118 Last updated: 31 AUG 2016 Open Peer Review Current Referee Status: Editorial Note on the Review Process F1000 Faculty Reviews are commissioned from members of the prestigious F1000 Faculty and are edited as a service to readers. In order to make these reviews as comprehensive and accessible as possible, the referees provide input before publication and only the final, revised version is published. The referees who approved the final version are listed with their names and affiliations but without their reports on earlier versions (any comments will already have been addressed in the published version). The referees who approved this article are: Version 1 1 Ichiro Hirao, Synthetic Molecular Biology Team, Institute of Bioengineering and Nanotechnology, Yokohama, Kanagawa, 230-0045, Singapore Competing Interests: No competing interests were disclosed. 2 Shota Atsumi, University of California, Davis, Davis, CA, 95616, USA Competing Interests: No competing interests were disclosed. F1000Research Page 8 of 8

Journal

F1000ResearchPubmed Central

Published: Aug 31, 2016

There are no references for this article.