Biosafety

Biosafety is a specific set of measures which help keep genetically modified organisms from causing harm to people, other organisms or the environment. Our probiotic, AnTigone, is a living biotherapeutic which has to be administered orally and has to work inside the gut. As the product is not retained inside the body and is eventually excreted, viable cells may be released into the surrounding environment. Once released, these living cells could proliferate and transfer their genetic material to resident bacterial populations. To mitigate such dangers, we incorporated a safety design. Our target questions while building the safety design were:

  • Does it act only when and where it should?
  • Can it survive or spread outside the gut?
  • Which safety standards must be followed before the product reaches patients?

Summary

As our final product is a probiotic, consisting of engineered microorganisms, it is important that our product will not be a cause of concern, both to the environment and to the users. Since the bacterium is meant to survive, act, and eventually be cleared from a living host, we need to ensure that it behaves predictably while inside the gut and cannot cause harm when the engineered microbial strain is accidentally released to the environment. We approached this through two layers of biological control, and checked both against the regulatory framework.

These measures reduce several important risks, but AnTigone is still a proof-of-concept design rather than a clinically validated product. Further preclinical testingLaboratory and animal studies done before any trial in people. would be required before its safety could be assessed for human use.

Table 1. Overview of AnTigone's biosafety layers
LayerQuestion it answersWhat we did
In-Host ControlDoes the bacterium act only where and when it should, and how long does it stay?A hypoxia-responsiveSwitched on by low oxygen levels, as in the gut. gut gate, a testosterone-responsive feedback loop, a transient (non-colonizing)Not settling permanently in the gut: the bacteria are cleared and have to be taken again. design, and a review of how the native gut community may limit or be affected by our probiotic (resident gut communities).
Environmental BiocontainmentCan it survive or spread if it escapes the gut?AuxotrophyAuxotrophy: the bacterium cannot make a substance it needs, so it only grows where that substance is supplied., kill-switchesA built-in mechanism that kills the bacteria if they leave the intended environment. (CRISPR-based and quorum-sensing-based) and genetic firewalling (amber-suppression semantic firewallsA design where engineered genes use a recoded genetic code that only the engineered host can read, so they stop working if transferred to another organism.), plus the genetic stability of these safety circuits under selection pressureA condition that favors cells with certain traits, here mutants that lose a safety circuit because it slows their growth..
GMO RegulationsWhat must we meet before AnTigone could reach patients?EU directives (2009/41/EC, 2001/83/EC, 2001/18/EC), the EMA environmental risk assessment guideline, the Cartagena Protocol and the FDA pathway in the USA.

Beyond these two layers, we also considered risks that could affect both in-host and environmental safety, particularly horizontal gene transfer (HGT). HGT refers to the transfer of genetic material between organisms. While our engineered bacteria are in the gut, therapeutic genes could potentially spread to native gut bacteria. To reduce this risk, we considered integrating the therapeutic genes into the chromosomeInserting a gene into the bacterium's main genome means it is passed on through cell division and is harder to transfer to other bacteria. of Escherichia coli Nissle 1917 (EcN), a well-studied probiotic strain, rather than using a plasmidA small circular piece of DNA that bacteria can copy and pass to each other.. Previous studies have demonstrated successful integration at specific sites, such as flu, fim, and ecpA (Seco & Fernández, 2022). We also considered avoiding mobile genetic elementsDNA sequences that can move within a genome or between cells, such as transposons. and unnecessary antibiotic resistance markersGenes that make a bacterium survive an antibiotic, used in the lab to select cells that took up new DNA. in our construct, rather than removing EcN's naturally occurring transposonsDNA segments that can jump from one place in a genome to another., which could disrupt bacterial functions (Grozdanov et al., 2004).

For each strategy, we researched multiple possible approaches, compared them against our project's specific constraints, and selected the combination that best balances safety, feasibility, and relevance to AnTigone as a testosterone-lowering gut therapeutic.

In-Host Control

Gut Gate

Our effector should only be active when our engineered probiotic is in the gut. The distinct physicochemical signals in the gut differ from both the outside environment and between different compartments within the gut itself, and these signals can be used to build a location-based switch. Chien et al. (2022) built oxygen, pH and lactate biosensorsGenetic circuits that detect a signal, such as oxygen, and respond to it. that control essential genes (asd, glmS) in E. coli and S. typhimurium, so that growth depends on the gut-like signal, and tested them in gut-mimicking media and in mice. The three sensing options are compared in Table 2. We chose hypoxia sensing for our proof of concept.

Table 2. Sensing options compared for the gut gate
OptionHow it worksRegistry part
Hypoxia sensingA promoterThe DNA sequence in front of a gene that controls when it is switched on. (e.g. PfnrS: hypoxia-induced, FNRA bacterial protein that switches genes on or off depending on oxygen.-regulated) is strongly activated under low-oxygen conditions. The gut lumenThe hollow inside of the gut, where food and bacteria are. is anaerobicWithout oxygen. while conditions outside the gut contain more oxygen.BBa_25DMNRIB: sequence-verified match to the PfnrS promoter in our own construct
pH sensingA promoter responds to the gut's pH gradient (small intestine ~pH 6.3, colon ~pH 7).None
Lactate / bile acid sensingA promoter responds to gut-specific metabolites like lactate or bile acids, present at much higher concentration in the gut than elsewhere.None

Why Hypoxia Sensing?

Feedback-Responsive Expression

Our effector should be expressed only when testosterone levels exceed a threshold. To ensure the enzyme is produced only to lower elevated testosterone, we propose combining our gut gate with a testosterone-responsive transcription factorA protein that binds DNA to switch a gene on or off. to form an AND gateA genetic circuit that switches on only when both inputs are present., building directly on feedback from the Málaga iGEM meet-up, where judges recommended an AND gate combining our biosensor with a feedback loop.

The sensor is TeiR, a transcription factor from Comamonas testosteroni, a bacterium that uses testosterone as a carbon source (Pruneda-Paz et al., 2004). TeiR is not specific to testosterone: it also responded to eight other steroids (Göhler et al., 2008). If we implement it, TeiR would therefore have to be engineered and tested so that it responds to testosterone and not to other steroids, including non-androgens. The binding region identified in that study (around residues 189 to 209) is the natural place to start.

Figure 1 shows how the gut gate and TeiR will be engineered to work together and the Table 3 explains the steps followed.

Figure 1. Gut gate and feedback loop. Low oxygen switches on TeiR expression; testosterone-bound TeiR activates the sip48 promoter, which drives the effector enzyme that lowers testosterone and closes the loop.
Table 3. Steps of the Gut gate and Feedback loop.
StepWhat happens
1. Hypoxia promoter switches onUnder low-oxygen conditions in the gut, the hypoxia promoter (PfnrS) is strongly activated .
2. TeiR productionThe active promoter drives expression of TeiR. A teiR-knockoutA strain in which a gene has been deleted or disabled. strain lost chemotaxisMovement of a cell toward or away from a chemical. toward testosterone as well as testosterone binding and uptake, and no longer induced the steroid-degrading enzyme hsdA (Göhler et al., 2008). TeiR has been cloned and expressed in E. coli, under two different inducible promotersPromoters that are switched on by adding a specific chemical. (tac and lacZ), TeiR activated expression of its target gene, with protein levels reported at 1.42–1.80 μg/mg after IPTGA lab chemical used to switch on genes controlled by certain promoters. induction, reported as evidence that TeiR retains regulatory function outside its native organism (Chen et al., 2009).
3. TeiR binds testosterone above thresholdTeiR is a membrane-associated protein whose steroid-binding region lies around residues 189–209. Purified TeiR transfers phosphate from GTP (not ATP) to cytosolic proteins, consistent with a histidine kinaseAn enzyme that transfers phosphate groups to proteins, often as part of a signalling pathway. (Göhler et al., 2008). The testosterone concentration at which TeiR becomes active has not been measured, so the threshold in our design is a target to be tested, not a literature value.
4. sip48 promoter turns onIn our design, activated TeiR switches on the sip48 promoter, proposed as its natural target in C. testosteroni (Pruneda-Paz et al., 2004). Göhler et al., 2008 shows that TeiR drives expression of the testosterone-degrading enzyme hsdA but does not show TeiR binding DNA, so whether TeiR acts on sip48 directly has not yet been confirmed from the primary papers. This is the least well-established step of the circuit and must be checked before construction.
5. Enzyme productionsip48 drives expression of our effector enzyme, reducing elevated testosterone levels.
6. FeedbackAs testosterone levels fall, less remains available to activate TeiR, and expression is expected to decrease accordingly, closing the loop.

Our effector enzyme ci2350 is a 5β-reductase that reduced progesterone, cortisone and hydrocortisone in the study that identified it; testosterone was not assayed there (Arp et al., 2025), so its activity on testosterone must be tested directly. A related gut-bacterial reductase pathway (OsrB) was functional when expressed in anaerobically grown E. coli, and colonization by the donor bacterium lowered prednisolone 2–10-fold in mouse cecumA pouch at the start of the large intestine., colon and feces (Jacoby et al., 2025). This shows that a gut bacterium can remove a steroid in vivoIn a living organism., but it is not a sense-and-degrade circuit, and no study has built one for testosterone.

Colonization Control

Colonization control is the decision about how long our probiotic should remain in the gut after dosing. It can either be transient (not colonizing the gut, cleared between doses and requiring redosing) or engineered to be self-limiting. Residence time matters because shedding continues for as long as the bacterium stays in the gut, and a gut gate does not shorten that. We assume that lowering gut testosterone is a continuous process, so a transient design fits better: the effect is maintained by redosing and the bacterium does not have to persist.

Transient Design (relies on natural clearance)

Self-limiting (by engineering)

Our Choice: AnTigone is intended to be transient by design and the effect will be maintained by repeated dosing.

Resident Gut Communities

Resident gut communities are the native bacterial communities present in the gut. A living biotherapeutic is not alone in the gut, so we considered how it might affect this community.

The bacterium itself

Lowering Testosterone

In conclusion, we found no study that profiled the gut community in humans dosed with an engineered EcN, so we cannot say how large any effect would be. We therefore treat the effect on resident communities as an open question.

Environmental Biocontainment

Auxotrophy

Auxotrophy is a biocontainmentBuilt-in measures that stop an engineered organism from surviving or spreading outside its intended place. strategy that makes an organism unable to synthesize a substance essential for growth, so its normal growth depends on an exogenous supply of that substance. The engineered E. coli in AnTigone is designed to metabolize testosterone in the gut using enzymes encoded by AKR1D1, ci2350, and atcABC; for future therapeutic use, the bacteria must not proliferate indefinitely or persist long-term after performing their function.

Options compared

We compared six auxotrophy options before selecting this one. Each row states what the option does, where it falls short, and what the cited study showed.

OptionHow it worksAdvantagesLimitationsWhat the evidence shows
Amino-acid auxotrophyDelete a gene for synthesis of a common amino acidSimple, classic modelWeak in the gut: amino acids are available via diet, host content, or cross-feedingOne bacterium supplying a nutrient to another.Classic auxotrophic containment approach Pande et al., 2015.
ΔthyA / thymidine auxotrophyKnock out thymidylate synthaseWell-established; strong precedent in L. lactis for intestinal IL-10 deliveryThymidine may still be taken up from the environmentSteidler et al., 2003: the gut supplies thymidine, about 1 µM in porcine ileum and up to 0.075 µM in human ileal lavage. A ΔthyA B. thetaiotaomicron strain still colonized mice for at least 10 days Hayashi et al., 2024.
ΔdapA / DAP auxotrophy (chosen)Delete dapA, which encodes DHDPS, in the DAP/lysine biosynthesis pathway.Directly linked to peptidoglycan synthesis; therapeutic precedentNot absolute: mutation or HGT could restore functionIsabella et al., 2018: ΔdapA SYNB1618 was cleared from mice within 48 h and was undetectable in monkey stool within 7 days. Puurunen et al., 2021: no stool sample above the detection limit 4 days after the last dose in humans.
Δasd / upstream DAP-pathway auxotrophyKnock out asd, upstream of DAP synthesisStrong growth restriction; used in existing containment circuitsBroader metabolic impact (also affects lysine, threonine, methionine synthesis)Used in existing containment circuits Ronchel & Ramos, 2001; asd was also one of the essential genes placed under biosensor control by Chien et al., 2022.
Double / multiplexed auxotrophyCombine two independent essential dependenciesReduces escape frequencyThe fraction of cells that survive despite a containment measure.; double insuranceMore complex to build; repeated editing can make the strain more fragileChien et al., 2022: A hypoxia–lactate AND gate achieved an escapee rate of ~10⁻⁸ in vitro and was later tested for tumor targeting in Salmonella enterica serovar Typhimurium.
Synthetic auxotrophy (non-standard amino acid)Recode essential proteins/codons to require a non-natural amino acidStrongest theoretical containment; stable long-termToo complex for current project scopeDescribed as the strongest containment option, but needs extensive genome recoding Kunjapur et al., 2021.

After comparing six auxotrophy strategies, we selected ΔdapA-mediated DAP auxotrophy because of its direct link to peptidoglycan synthesis, precedent in engineered bacterial therapeutics, and relatively simple design. By disrupting DAP biosynthesis, this approach aims to allow temporary metabolic activity while restricting long-term proliferation in DAP-limited environments.

How ΔdapA auxotrophy works

The three steps below illustrate how dapA deletion disrupts DAP biosynthesis, affects cell-wall formation, and ultimately restricts bacterial growth in the absence of external DAP.

Gene deletion

Pathway blockage

Growth restriction

Stability considerations

ΔdapA creates selective pressure, so acquisition of dapA (or a functionally equivalent gene) via HGT could allow the bacteria to escape containment; cross-feeding of meso-DAP in the gut is a further, though less likely, risk. To improve stability, we are considering combining DAP auxotrophy with a second, independent essential dependency, and a CRISPR-based system (following the thyA-targeting approach described by Hayashi et al., 2024) that designed to target and disrupt exogenous dapA sequences acquired via HGT, provided they contain a matching guide RNA target sequence and a suitable PAM. Synthetic auxotrophy is not being pursued now, given the genome-editing complexity involved, but remains a candidate for a future therapeutic-grade strain. We also plan to periodically measure escape frequency and characterize any escape variants to monitor long-term stability. Given the direct link between dapA and cell-wall synthesis, its precedent in engineered bacteria, and its relatively simple construction, we selected ΔdapA-mediated DAP auxotrophy as our biocontainment strategy.

Kill-Switch

As our final product is a probiotic, it is important that it will not pose a risk to the environment or users if it is accidentally released. We looked at two families of kill-switch strategy, CRISPR-based and quorum-sensing-based, and expect combining strategies to be more effective than any single one alone.

CRISPR-based approaches

CRISPR is a gene-editing strategy naturally present in bacteria as phage defenseProtection against phages, the viruses that infect bacteria., adaptable to recognize, remove, or express genes as needed. The system has two parts: a Cas enzyme that cuts DNA, and a guide RNAThe short RNA that tells Cas9 where to cut. (crRNA + tracrRNA) that tells it where to cut, next to a short PAM motifProtospacer adjacent motif, a short DNA sequence that Cas9 needs next to its target before it will cut. near the target sequence (Lim & Kim, 2022). The resulting double-strand break is typically repaired by non-homologous end joiningA repair process that glues broken DNA back together, often introducing errors., which disrupts the target sequence and its function, with Figure 2 (Lim & Kim, 2022) displaying different CRISPR-based genome editors. We looked at four strategies for applying this to biocontainment:

  1. Environmental (condition-gated) kill switch. Rottinghaus et al., 2022 built a single-input switch responding to the chemical inducer aTc (triggering tet gene expression that kills the bacteria) and a double-input switch adding a temperature-sensing layer (PtlpA-TlpA) as an OR gateA genetic circuit that switches on when either input is present. on top. The single-input switch killed bacteria efficiently inside the gut of test mice; the double-input switch also killed bacteria that had already been excreted, extending containment beyond the gut itself.

  2. Plasmid/chromosome targeting. Caliando & Voigt (2015) designed a CRISPR array matching the bacterium's own plasmid or chromosome sequence, using Cas3 (rather than Cas9) to trigger long-range DNA breakage and bacterial death.

  3. Targeting native repeated sequences. In Pseudomonas putida KT2440 (Asin-Garcia et al., 2024), guide RNAs were designed against repeated genomic sequences (cREP, ISPpu9) using a ScCas9 variant; introducing Cas9 triggered CRISPR expression and killed the cell. A limitation is that Cas9 itself can be toxic, making controlled expression difficult.

  4. Essential-gene targeting in a eukaryotic chassisThe host organism that carries and runs the engineered genetic system.. Umashankar et al. (2026) targeted genes essential for yeast survival (ERG13, PGA3, TPI1, CDC19); disrupting even one was enough to kill the cell, unlike bacterial systems that typically need multiple genomic breaks.

    Figure 2. CRISPR-based genome editors. A: Schematics of CRISPR-Cas9 and the results of CRISPR-Cas9-mediated genome editing. B: Schematics of base editors and the results of base editing. C: Schematics of prime editors and the results of prime editing. Cas9-induced cleavage sites are indicated with red vertical dashed lines. sgRNA, single guide RNA; PAM, protospacer adjacent motif; pegRNA, prime editing guide RNA; RT, reverse transcriptase (Lim & Kim, 2022).

Our chosen approach

Quorum sensing

Quorum sensing (QS) is a bacterial communication system using signaling molecules to sense population density and regulate gene expression Sadr et al., 2025. In the LuxI/LuxR system, bacteria produce AHLAcyl-homoserine lactone, a small signalling molecule that bacteria release to sense how crowded they are., which accumulates as bacterial population density increases. Once AHL reaches a threshold concentration, it binds to LuxR and activates downstream gene expression Cai & Zhang, 2024.

Inspired by the Utrecht 2025 iGEM team, which explored using high AHL concentrations to activate MazF and inhibit bacterial growth at high population densities, we considered QS as a potential population-dependent containment strategy for AnTigone. This approach could help limit excessive accumulation of E. coli Nissle 1917 (EcN), particularly given previous findings that EcN can preferentially colonize colorectal neoplastic tissue Gurbatri et al., 2024. Alternatively, we could combine quorum sensing with our hypoxia-based gate, so that bacterial survival depends on both low oxygen availability and an appropriate population density.

Genetic Firewalling: Amber-Suppression Semantic Firewall

Figure 4 shows that amber suppression works by introducing a stop codon (UAG) into an engineered gene, which would normally cause protein production to stop early. However, in the intended host, a specially designed suppressor tRNA recognizes UAG and allows translation to continue, producing a full-length, functional protein. If the gene is transferred to another bacterium without the matching suppressor tRNA, translation stops prematurely, resulting in an incomplete protein that is unlikely to function properly. This makes the engineered gene dependent on the suppressor tRNA in the intended host (Bautista et al., 2021; Wals & Ovaa, 2014; Whitford et al., 2018).

Figure 4. Principle of Amber-Suppression Semantic Firewall (Made using BioRender)

Why we considered it

In the gut, engineered DNA could potentially transfer to other bacteria (Brito, 2021). Amber suppression would not stop that transfer, but could make the transferred gene less likely to work in a new host, without relying on a dedicated toxin, an attractive additional safety layer (Whitford et al., 2018).

Main limitations

Overall assessment

Amber suppression is a useful additional genetic safety layer, since it reduces the function of transferred genes without requiring a toxin, but it does not directly control E. coli survival, can interfere with normal UAG translation, may lose stability over long-term growth, and has not been validated as a standalone containment system in the human gut. We treat it as a supporting genetic safeguard alongside an organism-level containment system, rather than a containment system on its own.

GMO Regulations

As AnTigone is a genetically engineered live bacterial therapeutic, we considered the regulatory requirements relevant to its development and potential clinical application. These include pharmaceutical and GMO regulations covering laboratory use, human administration, environmental safety, and international transport. The table below summarizes the key regulatory frameworks and their relevance to AnTigone.

Europe

FrameworkRegulatesRelevance to our Probiotic
Directive 2009/41/EC: Contained UseUse of genetically modified microorganisms (GMMs) in labs and contained facilities: risk assessment, containment, genetic stability, waste disposalOur bacteria are developed and handled in the laboratory, requiring appropriate containment to protect human health and the environment
Directive 2001/83/EC: Human UseEU framework for human medicines: marketing authorizationOfficial permission to sell a medicine., quality, safety, efficacy, manufacturing, labeling, pharmacovigilanceMonitoring a medicine's side effects after it is on the market.Would apply if AnTigone is developed as a medicinal product for human use, before it could enter the market
Regulation (EC) No 726/2004: Centralized AuthorizationEstablishes the EMA and the EU centralized marketing-authorization systemMay apply since AnTigone is developed using genetic engineering and biotechnology
Directive 2001/18/EC: Environmental ReleaseRelease of GMOs into the environment: risk assessment, monitoringThe engineered bacteria could leave the patient via feces and enter the environment, requiring assessment of survival, spread, and gene transfer
EMA Guideline: Environmental Risk AssessmentEnvironmental risk assessment for GMO-containing medicinal productsPatient shedding could expose the environment to the engineered bacteria, requiring assessment before marketing authorization
Cartagena Protocol: International MovementSafe transfer, handling, and transport of living modified organisms across bordersWould apply if the engineered bacteria were transported internationally

USA

In the United States, AnTigone would likely follow the FDA's regulatory pathway for drugs and biological products. If classified as a biological product, an Investigational New Drug (IND) applicationThe request to start clinical trials of a new medicine. would generally be required before initiating clinical trials, followed by a Biologics License Application (BLA)The request to sell a biological product. to obtain marketing approval. As AnTigone involves genetically engineered bacteria, we would also need to assess whether an Environmental Assessment (EA)A review of a product's impact on the environment. is required or whether the product qualifies for a categorical exclusionAn exemption from a full environmental assessment for categories of actions known to have little impact. under FDA regulations.

References

  1. Arp, G., Jiang, A. K., Dufault-Thompson, K., Levy, S., Zhong, A., Wassan, J. T., Grant, M. R., Li, Y., Hall, B., & Jiang, X. (2025). Identification of gut bacteria reductases that biotransform steroid hormones. Nature Communications, 16(1). https://doi.org/10.1038/s41467-025-61425-6
  2. Asin-Garcia, E., Martin-Pascual, M., de Buck, C., Allewijn, M., Müller, A., & Martins Dos Santos, V. A. P. (2024). GenoMine: A CRISPR-Cas9-based kill switch for biocontainment of Pseudomonas putida. Frontiers in Bioengineering and Biotechnology, 12, 1426107. https://doi.org/10.3389/fbioe.2024.1426107
  3. Bautista, D. E., Carr, J. F., & Mitchell, A. M. (2021). Suppressor mutants: History and today's applications. EcoSal Plus, 9(2), Article eESP00372020. https://doi.org/10.1128/ecosalplus.ESP-0037-2020
  4. Brito, I. L. (2021). Examining horizontal gene transfer in microbial communities. Nature Reviews Microbiology, 19, 442–453. https://doi.org/10.1038/s41579-021-00534-7
  5. Cai, Y., & Zhang, X. (2024). The atypical organization of the luxI/R family genes in AHL-driven quorum-sensing circuits. Journal of Bacteriology, 206(2), e00430-23. https://doi.org/10.1128/jb.00430-23
  6. Caliando, B. J., & Voigt, C. A. (2015). Targeted DNA degradation using a CRISPR device stably carried in the host genome. Nature Communications, 6, Article 6989. https://doi.org/10.1038/ncomms7989
  7. Chen, J., Ai, Y., Zhou, Y., Pan, D., Xiong, G., Guo, Y., Pan, F., & Lin, B. (2009). Characterization and heterologous expression of testosterone-inducible regulator from Comamonas testosteroni in Escherichia coli. Chinese Journal of Agricultural Biotechnology, 6(1), 41–47.
  8. Chien, T., Harimoto, T., Kepecs, B., Gray, K., Coker, C., Hou, N., Pu, K., Azad, T., Nolasco, A., Pavlicova, M., & Danino, T. (2022). Enhancing the tropism of bacteria via genetically programmed biosensors. Nature Biomedical Engineering, 6(1), 94–104. https://doi.org/10.1038/s41551-021-00772-3
  9. Göhler, A., Xiong, G., Paulsen, S., Trentmann, G., & Maser, E. (2008). Testosterone-inducible regulator is a kinase that drives steroid sensing and metabolism in Comamonas testosteroni. Journal of Biological Chemistry, 283(25), 17380–17390. https://doi.org/10.1074/jbc.M710166200
  10. Grozdanov, L., Raasch, C., Schulze, J., Sonnenborn, U., Gottschalk, G., Hacker, J., & Dobrindt, U. (2004). Analysis of the genome structure of the nonpathogenic probiotic Escherichia coli strain Nissle 1917. Journal of Bacteriology, 186(16), 5432–5441. https://doi.org/10.1128/JB.186.16.5432-5441.2004
  11. Gurbatri, C. R., Radford, G. A., Vrbanac, L., Im, J., Thomas, E. M., Coker, C., Taylor, S. R., Jang, Y., Sivan, A., Rhee, K., Saleh, A. A., Chien, T., Zandkarimi, F., Lia, I., Lannagan, T. R. M., Wang, T., Wright, J. A., Kobayashi, H., Ng, J. Q., . . . Danino, T. (2024). Engineering tumor-colonizing E. coli Nissle 1917 for detection and treatment of colorectal neoplasia. Nature Communications, 15, Article 646. https://doi.org/10.1038/s41467-024-44776-4
  12. Hayashi, N., Lai, Y., Fuerte-Stone, J., Mimee, M., & Lu, T. K. (2024). Cas9-assisted biological containment of a genetically engineered human commensal bacterium and genetic elements. Nature Communications, 15, 2096. https://doi.org/10.1038/s41467-024-45893-w
  13. Isabella, V. M., Ha, B. N., Castillo, M. J., Lubkowicz, D. J., Rowe, S. E., Millet, Y. A., Anderson, C. L., Li, N., Fisher, A. B., West, K. A., Reeder, P. J., Momin, M. M., Bergeron, C. G., Guilmain, S. E., Miller, P. F., Kurtz, C. B., & Falb, D. (2018). Development of a synthetic live bacterial therapeutic for the human metabolic disease phenylketonuria. Nature Biotechnology, 36(9), 857–864. https://doi.org/10.1038/nbt.4222
  14. Jacoby, C., Scorza, K., Ecker, L., Nol Bernardino, P., Little, A. S., McMillin, M., Ramaswamy, R., Sundararajan, A., Sidebottom, A. M., Lin, H., Dufault-Thompson, K., Hall, B., Jiang, X., & Light, S. H. (2025). Gut bacteria metabolize natural and synthetic steroid hormones via the reductive OsrABC pathway. Cell Host & Microbe, 33(11), 1873–1885.e7. https://doi.org/10.1016/j.chom.2025.09.014
  15. Joeres-Nguyen-Xuan, T. H., Boehm, S. K., Joeres, L., Schulze, J., & Kruis, W. (2010). Survival of the probiotic Escherichia coli Nissle 1917 (EcN) in the gastrointestinal tract given in combination with oral mesalamine to healthy volunteers. Inflammatory Bowel Diseases, 16(2), 256–262. https://doi.org/10.1002/ibd.21042
  16. Kunjapur, A. M., Napolitano, M. G., Hysolli, E., Noguera, K., Appleton, E. M., Schubert, M. G., Jones, M. A., Iyer, S., Mandell, D. J., & Church, G. M. (2021). Synthetic auxotrophy remains stable after continuous evolution and in coculture with mammalian cells. Science Advances, 7(27), eabf5851. https://doi.org/10.1126/sciadv.abf5851
  17. Lim, J. M., & Kim, H. H. (2022). Basic principles and clinical applications of CRISPR-based genome editing. Yonsei Medical Journal, 63(2), 105–113. https://doi.org/10.3349/ymj.2022.63.2.105
  18. Luo, M., Chen, Y., Pan, X., Chen, H., Fan, L., & Wen, Y. (2023). E. coli Nissle 1917 ameliorates mitochondrial injury of granulosa cells in polycystic ovary syndrome through promoting gut immune factor IL-22 via gut microbiota and microbial metabolism. Frontiers in Immunology, 14, Article 1137089. https://doi.org/10.3389/fimmu.2023.1137089
  19. Moadi, L., Turjeman, S., Asulin, N., & Koren, O. (2024). The effect of testosterone on the gut microbiome in mice. Communications Biology, 7, 880. https://doi.org/10.1038/s42003-024-06470-5
  20. Pande, S., Shitut, S., Freund, L., Westermann, M., Bertels, F., Colesie, C., Bischofs, I. B., & Kost, C. (2015). Metabolic cross-feeding via intercellular nanotubes among bacteria. Nature Communications, 6, 6238. https://doi.org/10.1038/ncomms7238
  21. Praveschotinunt, P., Dorval Courchesne, N.-M., den Hartog, I., Lu, C., Kim, J. J., Nguyen, P. Q., & Joshi, N. S. (2018). Tracking of engineered bacteria in vivo using nonstandard amino acid incorporation. ACS Synthetic Biology, 7(6), 1640–1650. https://doi.org/10.1021/acssynbio.8b00135
  22. Pruneda-Paz, J. L., Linares, M., Cabrera, J. E., & Genti-Raimondi, S. (2004). TeiR, a LuxR-type transcription factor required for testosterone degradation in Comamonas testosteroni. Journal of Bacteriology, 186(5), 1430–1437. https://doi.org/10.1128/JB.186.5.1430-1437.2004
  23. Puurunen, M. K., Vockley, J., Searle, S. L., Sacharow, S. J., Phillips, J. A., III, Denney, W. S., Goodlett, B. D., Wagner, D. A., Blankstein, L., Castillo, M. J., Charbonneau, M. R., Isabella, V. M., Sethuraman, V. V., Riese, R. J., Kurtz, C. B., & Brennan, A. M. (2021). Safety and pharmacodynamics of an engineered E. coli Nissle for the treatment of phenylketonuria: A first-in-human phase 1/2a study. Nature Metabolism, 3(8), 1125–1132. https://doi.org/10.1038/s42255-021-00430-7
  24. Ronchel, M. C., & Ramos, J. L. (2001). Dual system to reinforce biological containment of recombinant bacteria designed for rhizoremediation. Applied and Environmental Microbiology, 67(6), 2649–2656. https://doi.org/10.1128/AEM.67.6.2649-2656.2001
  25. Rottinghaus, A. G., Ferreiro, A., Fishbein, S. R. S., Dantas, G., & Moon, T. S. (2022). Genetically stable CRISPR-based kill switches for engineered microbes. Nature Communications, 13, Article 672. https://doi.org/10.1038/s41467-022-28163-5
  26. Sadr, S., Zargar, B., Aucoin, M. G., & Ingalls, B. (2025). From single cells to communities: Mathematical perspectives on bacterial quorum sensing. Computational and Structural Biotechnology Journal, 27, 4340–4351. https://doi.org/10.1016/j.csbj.2025.09.030
  27. Seco, E. M., & Fernández, L. Á. (2022). Efficient markerless integration of genes in the chromosome of probiotic E. coli Nissle 1917 by bacterial conjugation. Microbial Biotechnology, 15(5), 1374–1391. https://doi.org/10.1111/1751-7915.13967
  28. Steidler, L., Neirynck, S., Huyghebaert, N., Snoeck, V., Vermeire, A., Goddeeris, B., Cox, E., Remon, J. P., & Remaut, E. (2003). Biological containment of genetically modified Lactococcus lactis for intestinal delivery of human interleukin 10. Nature Biotechnology, 21(7), 785–789. https://doi.org/10.1038/nbt840
  29. Umashankar, P., Choi, B., & Nygård, Y. (2026). Towards the development of a CRISPR-Cas9 based kill switch for Saccharomyces cerevisiae. Microbial Cell Factories, 25, Article 62. https://doi.org/10.1186/s12934-026-02959-2
  30. Wals, K., & Ovaa, H. (2014). Unnatural amino acid incorporation in E. coli: Current and future applications in the design of therapeutic proteins. Frontiers in Chemistry, 2, 15. https://doi.org/10.3389/fchem.2014.00015
  31. Wang, Q., Sun, T., Xu, J., Shen, Z., Briggs, S. P., Zhou, D., & Wang, L. (2014). Response and adaptation of Escherichia coli to suppression of the amber stop codon. ChemBioChem, 15(12), 1744–1749. https://doi.org/10.1002/cbic.201402235
  32. Whitford, C. M., Dymek, S., Kerkhoff, D., März, C., Schmidt, O., Edich, M., Droste, J., Pucker, B., Rückert, C., & Kalinowski, J. (2018). Auxotrophy to Xeno-DNA: An exploration of combinatorial mechanisms for a high-fidelity biosafety system for synthetic biology applications. Journal of Biological Engineering, 12, 13. https://doi.org/10.1186/s13036-018-0105-8