Solution and Project Design

AnTigone: engineering controlled steroid transformation

AnTigone explores whether engineered bacteria could be used to modify androgenicHaving the effects of androgens, the male sex hormones such as testosterone. steroidsHormones built on a shared four-ring skeleton, such as testosterone and progesterone. within the intestine. Building on the biological rationale presented in the Background, we propose a system that combines two functions: an enzyme capable of transforming steroid molecules and an expression system responsive to low-oxygen conditions.

Our central hypothesis is that controlled expression of a steroid 5β-modulating enzyme could support the conversion of excess androgen into products with lower or no androgenic activity and/or a reduction in potent androgen bioavailabilityHow much of a substance is free to act on the body's tissues..

Why steroid 5β-reduction?

The biological activity of a steroid depends partly on its molecular structure. Steroid 5β-reductasesEnzymes that add hydrogen to a steroid at the 5β position, changing the molecule's 3D shape and, for testosterone, greatly lowering its androgen activity. modify the double bond between carbon atoms 4 and 5The part of the steroid's first ring that the enzyme changes. Its geometry decides how the steroid fits into receptors., changing the geometry of the steroid molecule (Di Costanzo et al., 2008; Penning, 2023). A human 5β-reductases enzyme, calledEnzymes that add hydrogen to a steroid at the 5β position, changing the molecule's 3D shape and, for testosterone, greatly lowering its androgen activity. AKR1D1The human enzyme that performs 5β-reduction. We use it as our established reference. provides an established example of this chemistry and can act on testosterone and other steroid substratesThe molecule an enzyme acts on and converts..

Figure 1. Alternative outcomes of testosterone 5-reduction. Reduction of the C4–C5 double bond produces different steroid configurations depending on the stereochemistry of the reaction. Whereas 5α-reduction produces the potent androgen 5α-dihydrotestosterone, 5β-reduction is a one-way reaction that produces a bent steroid configuration associated with substantially reduced androgenic activity. AnTigone investigates the latter pathway as a potential route for modifying androgen activity. Picture made with BioRender.

Our decision to investigate 5β-reduction was informed by discussions with researchers in microbiome biology and steroid metabolism. Liisa Veerus considered direct androgen biotransformationChemical conversion of a molecule by living cells or their enzymes. an experimentally tractable starting point because it provides measurable outputs in vitroIn a test tube or culture dish, rather than in a living organism.. Trevor M. Penning supported the biochemical rationale for investigating AKR1D1-mediated testosterone conversion, while emphasizing the difficulty of achieving specificity for a single steroid. Also the reaction is one of the few that is one-way and the products generated (Veerus et al., 2026). Together, this information and these discussions strengthened our decision to begin with a defined enzymatic reaction and assess both its activity and its limitations.

Table 1. Origin of the enzyme candidates chosen for AnTigone proof-of-concept.

Elisabet Stener-Victorin further encouraged us to distinguish a useful proof-of-conceptA first test showing that an idea can work, before it is developed into something usable. enzyme from a definitive therapeutic target. We therefore approach human AKR1D1 and microbial ci2350 5β-reductases as tools for investigating the proposed mechanism, without assuming that they would be the most suitable effectorThe part of the system that carries out the intended action, here the enzyme that transforms the steroid. for a future clinical application. The full consultation summaries are available on our Human Practices page.

For AnTigone, this reaction offers a defined starting point for engineering. We can investigate enzyme production, substrate conversion and the biological activity of the resulting mixture as separate experimental questions.

Substrate selectivity: a computational rationale

A 5β-reductase acting in the gut encounters a chemically diverse steroid pool. Androgens (mainly testosterone) and progestogensA class of steroid hormones that includes progesterone. share the same core scaffoldThe shared core structure of a group of molecules., and an enzyme with broad activity could act on both. Understanding what a candidate enzyme is likely to act on and how that preference might be modified was therefore a necessary step before committing to experimental characterization.

The structural distinction between these two steroid classes lies at the C17 positionOne specific carbon atom on the steroid skeleton, where testosterone and progesterone differ.: testosterone carries a 17β-hydroxyl groupThe OH group attached at C17 in testosterone., while progesterone carries an acetyl groupThe larger two-carbon functional group attached at C17 in progesterone.. These substituents differ in size and polarity yet occupy the same region of the active siteThe pocket of an enzyme where the substrate binds and is converted.. This provides a defined structural target for engineering.

Simon Ekström and Eduardo Amo González encouraged us to approach selectivity through rational mutagenesisDesigning specific mutations based on a protein's structure, rather than mutating at random.: using the three-dimensional structure of each candidate enzyme to identify active-site residuesThe individual amino acids that make up a protein. near C17, selecting substitutions with a chemical basis for altering substrate preference, and screening variants for structural stability before assessing their predicted binding profiles.

This analysis allowed us to generate specific sequence variants with a mechanistic rationale, a structured starting point for experimental characterization rather than an exhaustive screen. The full pipeline is documented on our Model page.

A predictive model for enzyme activity

DockingA computer simulation that predicts how well a molecule fits into a protein's binding site. tells us how well a variant might bind a steroid. It does not tell us how that variant would behave in bacterial culture.

Steroid conversion in bacterial culture depends on more than binding affinityHow strongly a molecule sticks to a protein.: on how many cells are present, how much enzyme cells produce, and how the enzyme handles testosterone and progesterone when both compete for it. To connect these steps, we built a mathematical model with three linked layers: bacterial growth, enzyme expressionHow much of the enzyme the cells make from its gene. and enzyme activity. Our docking results feed into the activity layer, linking the structural analysis to the model.

The model ranks our variants by a testable readout, the time needed to consume half of each steroid, and identifies the parameters the prediction depends on most. It serves as a guide and predictive layer to support our experimental work. Its output is a prediction to be tested, not a forecast of absolute rates. The equations, parameters and results are documented on our Model page.

Connecting activity to environmental conditions

The second component of AnTigone is hypoxia-responsiveSwitched on by low oxygen levels, as in the gut. expression guided by a hypoxia-responsive promoterA stretch of DNA that switches a gene on, so that the cell starts making its protein. (Sen et al., 2025; Wichmann et al., 2023). Our intended design uses low oxygen availability as an environmental input to regulate production of the testosterone-transforming enzyme.

Figure 2. Hypoxia-responsive expression designs for the candidate steroid reductases. The schematics place ci2350–MBP and AKR1D1–SUMO fusion designs under the control of the low-oxygen-responsive PfnrS promoter. Ribosome-binding sites, 6×His affinity tags, GSG linkers and transcriptional terminators complete the illustrated expression cassettes. The purpose of these designs is to connect enzyme production to the gut low-oxygen input signal. Picture made with BioRender.

Low oxygen is only one feature of the intended intestinal environment: a response to hypoxia would demonstrate environmental regulation, but would not by itself establish exclusive expression in the gut, a relevant clinical situation or sufficient biological containmentMeasures that keep engineered bacteria from surviving or spreading outside their intended place., which is planned to be implemented in the future following further literature research and expert input.

Designing for protein production and recovery

A final product and a biochemical characterization require sufficient soluble proteinProtein that stays dissolved in the cell instead of clumping, so it can be purified and used.. As we were aware of inclusion bodies (Clumps of misfolded protein that form inside bacteria and are hard to use.Shutevska & Nestorovska, 2024) and possible problems regarding human recombinantMade in a host organism from a gene that was introduced into it. protein expression in a microbial system, we decided to incorporate and test affinityAffinity tags are short additions to a protein that let it be pulled out of a mixture and purified. and solubility tagsProtein partners fused to the target to help it stay dissolved and fold properly., documented on our Parts page. Inclusion bodies are intracellular aggregates of misfolded or partially folded proteins, which are generally inactive, requiring laborious refolding procedures to regain activity. Due to this, a significant fraction of the expressed protein may be lost during expression (Costa et al., 2014; Malik, 2016).

Figure 3. General architecture of the enzyme-expression constructs. The schematic shows a T7 promoter, ribosome-binding site, 6×His affinity tag, alternative solubility partners, a glycine–serine–glycine (GSG) linker, a candidate enzyme coding sequence and a transcriptional terminator. ci2350 and AKR1D1 were investigated using different tag configurations to support protein production and recovery. The diagram summarizes the modular design rather than indicating that every possible combination was constructed and validated; individual constructs are documented on the Parts page. Picture made with BioRender.

The construct collection includes a 6xHis affinity tagA short string of six histidines added to a protein so it can be purified. (Bornhorst & Falke, 2000) and fusionTwo proteins joined into one. designs using maltose-binding proteinA well-folding bacterial protein often fused to others to help them dissolve. (MBP) (Sachdev & Chirgwin, 2000), SUMOSmall Ubiquitin-like Modifier, a small protein fused to a target to help it fold and stay soluble. It can be cut off again afterwards. (Costa et al., 2014) and Thioredoxin AA small bacterial protein that helps others fold, used as a solubility partner. (TrxA) (Paraskevopoulou & Falcone, 2018). The chosen designs allow us to investigate whether different fusion partners improve protein availability for downstream characterization and applications. Construct sequences and architectures are documented on the Parts page; experimental iterations and informed decisions are described in Engineering.

Our Integrated Experimental Pipeline

Figure 4. Experimental workflow and development status of AnTigone. In silico design of enzymes construct and hypoxia-responsive regulation, engineering and characterization were intended to proceed alongside, with future integration into a bacterial system. Green indicates completed experimental activities, amber indicates initiated work and gray indicates planned steps. Cofactor-consumption assays provide an indirect readout, while planned LC–MS/MS and androgen-receptor reporter assays address steroid conversion and biological activity, respectively. The proposed AND gate combines hypoxia with a second input requiring selection and validation. Picture made with BioRender.

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. Bornhorst, J. A., & Falke, J. J. (2000). Purification of proteins using polyhistidine affinity tags. Methods in Enzymology, 326, 245–254. https://doi.org/10.1016/s0076-6879(00)26058-8
  3. Costa, S., Almeida, A., Castro, A., & Domingues, L. (2014). Fusion tags for protein solubility, purification and immunogenicity in Escherichia coli: The novel Fh8 system. Frontiers in Microbiology, 5, Article 63. https://doi.org/10.3389/fmicb.2014.00063
  4. Di Costanzo, L., Drury, J. E., Penning, T. M., & Christianson, D. W. (2008). Crystal structure of human liver Δ4-3-ketosteroid 5β-reductase (AKR1D1) and implications for substrate binding and catalysis. Journal of Biological Chemistry, 283(24), 16830–16839. https://doi.org/10.1074/jbc.M801778200
  5. Malik, A. (2016). Protein fusion tags for efficient expression and purification of recombinant proteins in the periplasmic space of E. coli. 3 Biotech, 6(1). https://doi.org/10.1007/s13205-016-0397-7
  6. Paraskevopoulou, V., & Falcone, F. (2018). Polyionic tags as enhancers of protein solubility in recombinant protein expression. Microorganisms, 6(2), 47. https://doi.org/10.3390/microorganisms6020047
  7. Penning, T. M. (2023). Steroid 5β-reductase (AKR1D1): Purification and characterization. Methods in Enzymology, 689, 277–301. https://doi.org/10.1016/bs.mie.2023.04.012
  8. Sachdev, D., & Chirgwin, J. M. (2000). Fusions to maltose-binding protein: Control of folding and solubility in protein purification. Methods in Enzymology, 326, 312–321. https://doi.org/10.1016/s0076-6879(00)26062-x
  9. Sen, Y., Guo, C.-H., Tong, W.-Y., Liu, X.-Y., Li, J.-C., & Kang, M. (2025). Identification and characterization of anaerobically activated promoters in escherichia coli. Journal of Biotechnology, 402, 30–38. https://doi.org/10.1016/j.jbiotec.2025.03.002
  10. Shutevska, K., & Nestorovska, A. K. (2024). Cloning, expression, and purification of recombinant AKR1D1 for therapeutic applications. Acta Pharmaceutica. https://doi.org/10.2478/acph-2025-0003
  11. Veerus, L., Subrahmanian, A., & Blaser, M. J. (2026). The testobolome in microbial testosterone metabolism and human health. npj Biofilms and Microbiomes, 12, Article 9. https://doi.org/10.1038/s41522-025-00861-0
  12. Wichmann, J., Behrendt, G., Boecker, S., & Steffen, K. (2023). Characterizing and utilizing oxygen-dependent promoters for efficient dynamic metabolic engineering. Metabolic Engineering, 77, 199–207. https://doi.org/10.1016/j.ymben.2023.04.006