“First, do no harm.” - Hippocrates


In the pursuit of integrating our system with the genome of L. crispatus for sustained efficacy, as elaborated in the description of our project, we recognized the importance of incorporating a fail-safe mechanism, a "killswitch", to enable the controlled elimination of our engineered bacteria in case of unforeseen issues, thus ensuring user safety.


At its core, a killswitch comprises of an inducible promoter and a toxin. This can be implemented in two distinct ways:

  1. An inducible promoter that becomes active in the absence of a specific molecule, typically used for environmental safety to ensure the engineered bacterium's demise outside of the desired environment.
  2. An inducible promoter that activates in the presence of a particular molecule, commonly employed for user safety, allowing users to intentionally terminate the engineered bacterium.

Our choice was to design a killswitch of the latter type, and we opted to base it on the arabinose operon, a naturally occurring system found in various bacteria, including E. coli, B. subtilis, L. crispatus, Salmonella typhi, and Lactobacillus plantarum [1].

The arabinose operon's primary purpose is to enable bacteria to utilize L-arabinose, a pentose derived from hemicellulose degradation, as a carbon source. The operon consists of three key components:

  1. A repressor protein, induced by L-arabinose, that regulates operon expression.
  2. A transporter facilitating L-arabinose entry into the cell.
  3. Genes responsible for arabinose catabolism, including araA (L-arabinose isomerase), araB (L-ribulokinase), and araD (L-ribulose-5-phosphate 4-epimerase).
Additional note araABD

Figure 1: The metabolic pathway of araABD.

To be an effective killswitch inducer within the human body, a molecule must meet specific criteria:

Many bacteria naturally possess arabinose operons, which include arabinose-induced repressors [2].

Arabinose has not been tested for lethality in humans, but it can be inferred from studies in rats that it is safe at doses of 2g/day for individuals over 60 kg [3]. Human subjects have even safely consumed up to 15g/day (average weight 70 kg) [4]. Notably, the potential side effects of arabinose are linked to its inhibition of intestinal sucrase, leading to increased flatulence, diarrhea, and stomach discomfort [4]–[7]. However, it is also being explored as a diabetic-friendly sweetener due to its ability to lower insulin and blood sugar levels [4], [6].

Arabinose is not commonly found in its monomeric form in nature; it predominantly exists as a polysaccharide within plant cell walls, mainly as hemicellulose [8]. Isolating arabinose from the macromolecule requires extensive treatment [8].

While human arabinose metabolism has not been extensively studied, it is considered a no-calorie sweetener, with approximately 8% of consumed arabinose being excreted in urine, making it suitable for our purposes [4], [7], [8].

General Design

Our design aims to incorporate a repressor protein with a constitutive promoter and the selected toxin with an inducible promoter into the bacterial genome. Initially, we developed the killswitch for B. subtilis as a model and later adapted it for use in L. crispatus.

general killswitch design

Figure 2: The proposed general killswitch mechanism.

B. subtilis Part Design

The native system in B. subtilis comprises of:

  1. araR: Encodes the protein responsible for inhibiting the operon; it dissociates from DNA in the presence of arabinose.
  2. araE: Encodes a transporter protein that facilitates arabinose entry into the cell.
  3. araA: Encodes L-arabinose isomerase, the first enzyme required for arabinose utilization. Following the araA promoter, the rest of the operon genes follow: araABDLMNPQ-abfA (In E. coli, the system is known as pBAD due to the order araB, araA, araD) [2].


B. subtilis genetic map



Figure 3: (A) The genetic positions of the araR/araE transcriptional unit and araABDLMNPQ-abfA. The darker areas in the promoter region are the araR binding sites. (B) The transcriptional unit araABDLMNPQ-abfA.

Upon reviewing the iGEM registry, we discovered several examples of the arabinose-inducible system for E. coli, but only a single group had designed parts for this system in B. subtilis (araE promoter - BBa_K3331021, araR promoter - BBa_K3331020, araR protein - BBa_K3331022), and these parts had not been experimentally tested. After comprehensive research, we determined that these parts were unlikely to be functional, leading us to design new parts.

Comparing the sequences obtained from the iGEM registry with those available in the NCBI database (for the araR protein) and relevant literature (for the promoters), we noticed differences [2], [9], [10]. Notably, the sequences for the araR gene and its promoter in the registry were observed to be reverse complementary, likely due to the orientation of these genes on the "nonsense" strand within the B. subtilis genome [11]. Furthermore, there were differences in codon usage between the sequence found in the registry and the sequence found in NCBI.

Moreover, the araR promoter sequence in the registry appeared to lack the araR binding site, featuring only the sigma A binding sites and transcription start site. In contrast, our sequence, accessible via BBa_K4633004, includes the araR binding site. Similarly, the araE promoter in the registry also lacks araR binding sites, prompting us to improve this sequence, which is available as BBa_K4633006.

Repressor - araR Protein

In B. subtilis, the araR protein serves as the transcriptional repressor for all three parts of the arabinose operon when arabinose is absent. AraR binds to specific palindromic DNA sequences, preventing transcription [11]. However, in the presence of L-arabinose, the protein undergoes conformational changes, dissociating from the DNA and allowing transcription to proceed [11].

Table 1: araR recognition sequences

Binding SiteSequence

Sequences taken from DBTBS [12].

Constitutive Promoter - paraR

Originally, we considered using the native promoter for the araR gene as our constitutive promoter, but concerns arose regarding its potential inducibility, which could complicate our system [11]. Alternatives, such as pVeg (BBa_K143012), were considered as well, but high expression of araR may require excessive arabinose dosages to activate the killswitch. Subsequent research, elaborated on in the following section, led us to conclude that paraR, initially thought to be inducible, exhibited leakiness and thus proved suitable for our purposes.

The final paraR sequence, encompassing sigma binding sites, the araR binding site, and a ribosome binding site (RBS), was based on data from BioCyc B. subtilis 168 database, DBTBS, and related literature sources [10], [12], [13].

Inducible Promoter - paraE

Initially, we considered using the promoter for the araE gene (arabinose transporter), which we called paraE. We constructed a functional promoter containing sigma A binding sites, araR binding sites, Cre binding site (glucose inhibition) and an RBS, as found in the genome of B. subtilis [10]. During our efforts to order this sequence, we encountered an obstacle – its high A-T content, a common feature of the B. subtilis genome, which made it impossible to produce [14].

Similarly, we constructed the full promoter of the transcriptional unit araABDLMNPQ-abfA, and faced the same A-T richness issue [2].

Our goal was to use arabinose as the inducer, suitable for B. subtilis as a Gram-positive model. To achieve this, we needed to enhance the G-C content of the sequence while preserving its activity.

We explored various strategies:

This idea was eliminated due to the fact that araR achieves effective transcriptional inhibition through dimerization [11].

This idea was eliminated because it was found that the distance between the araR bonding sites was optimal [11].

This idea was eliminated for similar reasons to the previous idea [11].

Ultimately we decided to change bases in the promoter to enhance its G-C richness. For a more detailed explanation, see our part page in the registry

Toxin - MazF

Different bacteria harbor a variety of toxin-antitoxin systems, one of which is the MazF/MazE system, referred to as NdoA/NdoAI in B. subtilis [15]. MazF is characterized as an RNAse, which therefore has bacteriostatic toxin properties [16]. Notably, research has revealed that MazF from B. subtilis, when introduced into E. coli, exerts detrimental effects on the latter and is not affected by the native E. coli MazE [17]. These observations have led us to consider the potential effectiveness of this system in L. crispatus.


  1. A. Bateman et al., “UniProt: the Universal Protein Knowledgebase in 2023,” Nucleic Acids Res., vol. 51, no. D1, pp. D523–D531, Jan. 2023, doi: 10.1093/NAR/GKAC1052.
  2. I. Sá-Nogueira, T. V. Nogueira, S. Soares, and H. De Lencastre, “The Bacillus subtilis L-arabinose (ara) operon: Nucleotide sequence, genetic organization and expression,” Microbiology, vol. 143, no. 3, pp. 957–969, Mar. 1997, doi: 10.1099/00221287-143-3-957/CITE/REFWORKS.
  3. L. Hao, X. Lu, M. Sun, K. Li, L. Shen, and T. Wu, “Protective effects of L-arabinose in high-carbohydrate, high-fat diet-induced metabolic syndrome in rats,” SNF Swedish Nutr. Found., vol. 59, Dec. 2015, doi: 10.3402/FNR.V59.28886.
  4. K. Pol, K. de Graaf, M. Diepeveen-de Bruin, M. Balvers, and M. Mars, “The effect of replacing sucrose with L-arabinose in drinks and cereal foods on blood glucose and plasma insulin responses in healthy adults,” J. Funct. Foods, vol. 73, p. 104114, Oct. 2020, doi: 10.1016/J.JFF.2020.104114.
  5. K. Seri, K. Sanai, N. Matsuo, K. Kawakubo, C. Xue, and S. Inoue, “l-Arabinose selectively inhibits intestinal sucrase in an uncompetitive manner and suppresses glycemic response after sucrose ingestion in animals,” Metabolism, vol. 45, no. 11, pp. 1368–1374, Nov. 1996, doi: 10.1016/S0026-0495(96)90117-1.
  6. I. Krog-Mikkelsen, O. Hels, I. Tetens, J. J. Holst, J. R. Andersen, and K. Bukhave, “The effects of L-arabinose on intestinal sucrase activity: dose-response studies in vitro and in humans 1-3,” Am. J. Clin. Nutr., vol. 94, no. 2, p. Pages 472-478, 2011, doi: 10.3945/ajcn.111.014225.
  7. B. Tyson, C. M. Pask, N. George, and E. Simone, “Crystallization Behavior and Crystallographic Properties of dl -Arabinose and dl -Xylose Diastereomer Sugars,” Cryst. Growth Des., vol. 22, no. 2, pp. 1371–1383, Feb. 2022, doi: 10.1021/ACS.CGD.1C01329/ASSET/IMAGES/LARGE/CG1C01329_0011.JPEG.
  8. P. Mäki-Arvela, T. Salmi, B. Holmbom, S. Willför, and D. Y. Murzin, “Synthesis of sugars by hydrolysis of hemicelluloses- A review,” Chem. Rev., vol. 111, no. 9, pp. 5638–5666, Sep. 2011, doi: 10.1021/CR2000042/ASSET/IMAGES/MEDIUM/CR-2011-000042_0029.GIF.
  9. E. W. Sayers et al., “Database resources of the national center for biotechnology information,” Nucleic Acids Res., vol. 50, no. D1, pp. D20–D26, Jan. 2022, doi: 10.1093/NAR/GKAB1112.
  10. I. Sá-Nogueira and L. J. Mota, “Negative regulation of L-arabinose metabolism in Bacillus subtilis: characterization of the araR (araC) gene,” J. Bacteriol., vol. 179, no. 5, pp. 1598–1608, 1997, doi: 10.1128/JB.179.5.1598-1608.1997.
  11. L. J. Mota, L. Morais Sarmento, and I. De Sá-Nogueira, “Control of the Arabinose Regulon in Bacillus subtilis by AraR In Vivo: Crucial Roles of Operators, Cooperativity, and DNA Looping,” J. Bacteriol., vol. 183, no. 14, p. 4190, 2001, doi: 10.1128/JB.183.14.4190-4201.2001.
  12. N. Sierro, Y. Makita, M. De hoon, and K. Nakai, “DBTBS: a database of transcriptional regulation in Bacillus subtilis containing upstream intergenic conservation information,” Nucleic Acids Res., vol. 36, no. suppl_1, pp. D93–D96, Jan. 2008, doi: 10.1093/NAR/GKM910.
  13. P. D. Karp et al., “The BioCyc collection of microbial genomes and metabolic pathways,” Brief. Bioinform., vol. 20, no. 4, pp. 1085–1093, Mar. 2019, doi: 10.1093/BIB/BBX085.
  14. M. Akashi and H. Yoshikawa, “Relevance of GC content to the conservation of DNA polymerase III/mismatch repair system in Gram-positive bacteria,” Front. Microbiol., vol. 4, no. SEP, p. 59320, Sep. 2013, doi: 10.3389/FMICB.2013.00266/ABSTRACT.
  15. Y. Cui et al., “Bacterial MazF/MazE toxin-antitoxin suppresses lytic propagation of arbitrium-containing phages,” Cell Rep., 2022, doi: 10.1016/j.celrep.2022.111752.
  16. S. Brantl and P. Müller, “Toxin–Antitoxin Systems in Bacillus subtilis,” Toxins 2019, Vol. 11, Page 262, vol. 11, no. 5, p. 262, May 2019, doi: 10.3390/TOXINS11050262.
  17. J. H. Park, Y. Yamaguchi, and M. Inouye, “Bacillus subtilis MazF-bs (EndoA) is a UACAU-specific mRNA interferase,” FEBS Lett., vol. 585, no. 15, pp. 2526–2532, Aug. 2011, doi: 10.1016/J.FEBSLET.2011.07.008.