A living diagnostic-therapeutic: engineered gut bacterium senses inflammation and releases a therapeutic protein on site
Synthetic biologists convert an ordinary gut-adapted probiotic — most commonly Escherichia coli Nissle 1917 or Lactococcus lactis — into a "living therapeutic" by installing a genetic circuit: a biosensor module coupled through a promoter to a therapeutic output gene. The whole circuit behaves like a biological if-then statement — if the inflammation marker is present, then produce and secrete the drug — encoded entirely in DNA and run by the cell's own transcription and translation machinery.
Engineered biosensors are not built from scratch — they are installed inside an already gut-adapted, safety-vetted host. E. coli Nissle 1917 has been used as a probiotic for over a century, tolerates the anaerobic, bile-rich, immunologically active gut environment, and transiently colonizes the intestine without permanently displacing the native microbiome.
Before any sensing circuit is added, the chassis genome is characterized and, where needed, cleaned up: cryptic prophages and mobile elements are mapped, metabolic auxotrophies can be introduced as a biocontainment safeguard, and a genomic "landing pad" (a defined attB/attP integration site) is prepared so the circuit can be inserted at a single, well-characterized locus instead of floating on an unstable plasmid.
A minimal inflammation-responsive circuit has three modular parts, each swappable independently:
• Sensor module — typically a bacterial two-component system: a membrane-bound sensor histidine kinase that physically detects an extracellular ligand, and a cognate cytoplasmic response regulator that gets phosphorylated when the ligand is bound.
• Inducible promoter — a DNA sequence recognized by the phosphorylated response regulator; in its resting state it is silent (low basal "leaky" transcription), but folds/binds regulator upon activation to recruit RNA polymerase.
• Output gene — the coding sequence for the therapeutic payload (an anti-inflammatory protein or peptide), fused to a secretion tag so the product is exported rather than trapped in the cytoplasm.
Each module is sourced from a characterized genetic parts library so its behavior — leakiness, dynamic range, activation threshold — is quantitatively known before assembly.
Because the circuit must function reliably after the bacterium has divided dozens of times inside a moving, chemically hostile gut, engineers optimize for genetic stability: low-burden expression to avoid selecting against the plasmid, toxin-antitoxin plasmid maintenance systems, or chromosomal integration for maximal stability.
Biocontainment is designed in from the start: auxotrophic dependence on a nutrient not available outside the lab or patient, kill-switch circuits triggered by loss of the target environment, and single-generation containment strategies ensure the engineered strain cannot persist or spread if excreted.
Because the sensor, promoter, and payload gene are modular, the same validated chassis and sensor can be paired with different therapeutic payloads — turning one engineering platform into a reusable toolkit for many gut conditions.
For the circuit to be useful, its sensor must respond to something that reliably rises during gut inflammation but stays low in a healthy gut. Active inflammation generates a distinctive local chemistry — reactive oxygen and nitrogen species produced by the host immune response oxidize endogenous gut sulfur compounds into byproducts such as tetrathionate, and inflamed tissue also elevates local nitrate. These byproducts diffuse into the gut lumen and are exactly the kind of small, diffusible, disease-specific molecule a bacterial two-component sensor can be tuned to detect.
Gut inflammation is not just an immunological event — it reshapes the local chemical environment. Neutrophils and other immune cells infiltrating inflamed tissue generate reactive oxygen species (ROS) and reactive nitrogen species (RNS) as part of the host defense response. These reactive molecules non-specifically oxidize abundant gut metabolites, including endogenous sulfur compounds (thiosulfate), converting them into tetrathionate — a molecule essentially absent from a healthy, non-inflamed gut lumen but reliably present at sites of active inflammation.
Because this chemistry is downstream of the host immune response itself rather than any single pathogen or cause, a tetrathionate-based (or nitrate-based) sensor responds to inflammation broadly, regardless of its underlying trigger — making it a general-purpose biomarker rather than a disease-specific one.
Detection is carried out by a bacterial two-component system, borrowed and repurposed from naturally tetrathionate-sensing organisms such as Salmonella:
• The sensor histidine kinase (e.g. TtrS) spans the inner membrane; its periplasmic domain binds tetrathionate directly. • Ligand binding triggers a conformational change that activates the kinase's cytoplasmic domain, causing it to autophosphorylate on a conserved histidine residue using ATP. • The phosphoryl group is then transferred to a matching aspartate residue on the partner response regulator (e.g. TtrR) in the cytoplasm. • Phosphorylated TtrR changes conformation, gains DNA-binding activity, and diffuses to the engineered promoter to activate transcription.
This relay converts an extracellular chemical event into an intracellular phosphorylation state within seconds to minutes — the same design principle bacteria use natively to sense osmolarity, quorum-sensing molecules, and nutrient availability.
A landmark synthetic-biology demonstration (Riglar et al., Nature Biotechnology 2017) engineered E. coli Nissle with a tetrathionate/nitrate two-component sensor wired to a stable genetic memory circuit, allowing bacteria to record — and later report on — inflammation events encountered during transit through a living mouse gut.
Sensor sensitivity is not fixed — it can be engineered. Ribosome-binding-site strength, sensor kinase copy number, and promoter architecture can all be tuned to shift the concentration at which the circuit switches from off to on, and to sharpen that switch from a gradual dial into a steep, digital-like threshold.
A lower sensitivity setting requires a higher local concentration of the inflammation signal before the circuit fires — reducing the chance of responding to minor, transient fluctuations. A higher sensitivity setting triggers activation earlier, at lower marker concentrations, favoring early intervention at the cost of a higher chance of low-level background activation. This tunability is a design trade-off engineers select deliberately for each clinical use case.
Once the response regulator is phosphorylated and bound to the engineered promoter, the bacterium's own transcription and translation machinery takes over: RNA polymerase is recruited, the therapeutic gene is transcribed into mRNA, and ribosomes translate that mRNA into the finished therapeutic protein — all using the cell's native resources, with no external dosing or manufacturing step required.
The engineered promoter is designed with operator sites recognized specifically by the phosphorylated response regulator. In the inactive state, RNA polymerase either cannot bind the promoter or binds unproductively, keeping basal expression low (a "tight" or low-leak promoter is a key design goal).
When enough phosphorylated regulator accumulates, it binds cooperatively to the operator sites, remodels the local promoter architecture, and recruits RNA polymerase holoenzyme (core enzyme plus sigma factor) to initiate transcription. Because regulator phosphorylation scales with signal concentration, and promoter activation often requires cooperative binding of multiple regulator molecules, the circuit can behave like a switch rather than a simple dial — producing a sharper on/off response than the raw chemistry alone would predict.
The gene downstream of the inducible promoter encodes the actual therapeutic molecule. Candidate payloads explored in engineered gut therapeutics include:
• Anti-inflammatory cytokines, such as interleukin-10 (IL-10), which dampens local immune activation • Protease inhibitors, such as elafin, which counteract tissue-damaging proteases elevated during flares • Trefoil factors and other mucosal-repair peptides that promote epithelial healing • Short-chain fatty acid or metabolite production pathways that support barrier function
The coding sequence is codon-optimized for the chassis organism, and is typically fused N- or C-terminally to a secretion signal sequence so the translated protein is threaded out of the cytoplasm as it is made, rather than accumulating and potentially stressing the cell.
Because the amount of phosphorylated response regulator scales with local signal-molecule concentration (within the sensor's operating range), the amount of therapeutic protein produced is not simply on-or-off — it scales, over a bounded range, with how much inflammation signal is present. A patch of mild inflammation elicits a modest secretion rate; a patch of severe, active inflammation elicits a stronger one, up to a saturating maximum set by the cell's transcriptional and translational capacity.
This graded, signal-proportional output is a core advantage of a living genetic circuit over a fixed-dose pill: the "dose" delivered locally adjusts automatically to the severity of what it encounters.
Because the engineered bacterium is physically present in the gut lumen and only activates its circuit where the inflammation signal is locally elevated, therapeutic protein is secreted directly onto the inflamed patch of tissue — typically within a few hundred micrometers — rather than being absorbed into the bloodstream and distributed to every organ in the body. This spatial precision is the central practical advantage of a living, in-situ biosensor-therapeutic over a conventional systemic biologic drug.
Conventional biologic drugs for inflammatory bowel disease — such as injected anti-TNF antibodies — must be dosed at levels high enough to reach a therapeutic concentration throughout the entire bloodstream and diffuse into inflamed tissue from outside. This exposes every organ system to the drug, driving much of the off-target toxicity, infection risk, and cost associated with systemic biologics.
An engineered bacterium sitting in direct contact with (or immediately adjacent to) the inflamed mucosa can secrete its payload straight onto the tissue that needs it. The effective local concentration at the site of disease can be therapeutic while the concentration reaching the rest of the body remains negligible — because the bacterium simply is not making the protein anywhere else.
Producing the therapeutic protein inside the cytoplasm is only half the problem; it must also be exported. Engineered strains commonly use:
• The general secretory (Sec) pathway, co-translationally threading a signal-peptide-tagged protein across the inner membrane • Type I secretion systems (e.g. the hemolysin HlyA C-terminal secretion tag), which export proteins directly from cytoplasm to the extracellular space in one step, bypassing periplasmic intermediates • Surface-display or outer-membrane vesicle release, packaging the payload into vesicles that bud off and diffuse locally
The choice of secretion route affects yield, folding fidelity, and how far the protein diffuses before being diluted or degraded — all tunable parameters in circuit design.
Because gut peristalsis and luminal flow continuously mix contents, "localized" here means concentrated near actively colonizing, actively signaling bacteria — not that a single cell treats a single fixed point. Persistent local colonization at (or near) inflamed tissue is what sustains a meaningfully local dose.
Several engineered-microbe therapeutic candidates have advanced into human clinical testing, including strains designed to locally deplete or modulate metabolites and molecules implicated in inflammatory and metabolic gut disease. Early-phase trials focus on demonstrating that an orally delivered engineered strain safely transits the gut, transiently colonizes as intended, activates its circuit in response to the correct physiological cue, and produces a measurable, geographically confined therapeutic effect — before larger efficacy trials are pursued.
Because the genetic circuit is only transcriptionally active while the inflammation-associated signal remains above the sensor's threshold, therapeutic protein production is inherently coupled to the presence of disease activity. As inflammation resolves and the local marker concentration falls, the response regulator is no longer phosphorylated, the promoter falls silent, and protein output declines on its own — producing a closed-loop, self-limiting treatment rather than a constant, unconditional drug exposure.
Two-component systems are reversible by nature: histidine kinases with phosphatase activity actively dephosphorylate their partner response regulator once the extracellular ligand is no longer bound, and phosphorylated regulators also decay over time through intrinsic and enzyme-assisted hydrolysis. As inflammation resolves and tetrathionate (or the relevant marker) production falls, less kinase remains ligand-bound, phosphorylated regulator concentration drops, and promoter occupancy — and therefore transcription of the therapeutic gene — declines proportionally.
Because mRNA and existing protein are also degraded and diluted by cell division over time, functional output tapers over a timescale of tens of minutes to a few hours after the underlying chemical trigger disappears, rather than persisting indefinitely.
Chronic inflammatory conditions such as inflammatory bowel disease are characterized by unpredictable flares punctuated by periods of remission. A fixed-dose systemic drug schedule cannot naturally track this fluctuating disease course — patients are either exposed to therapeutic levels of drug during quiet periods (unnecessary risk and cost) or under-dosed at the onset of a flare (delayed benefit).
A self-limiting, sense-and-respond genetic circuit collapses diagnosis and treatment timing into a single automatic process: therapeutic output rises only when and where disease activity is present, and recedes as it resolves — conceptually similar to how a thermostat responds to temperature rather than running on a fixed timer.
This closed-loop design also functions as a built-in safety feature: because the circuit requires continuous presence of the disease-associated signal to stay on, an engineered strain that is excreted or relocates away from an inflamed site automatically returns to a low, basal-output state.
The sense-activate-secrete-resolve cycle demonstrated here is a template, not a one-off design. The same architecture — a tunable biosensor, an inducible promoter, and a swappable therapeutic payload gene, wired through a self-terminating feedback loop — can in principle be re-targeted to other disease-associated biomarkers and other therapeutic payloads, extending the "living diagnostic-therapeutic" concept from gut inflammation to other conditions where a disease state produces a locally detectable molecular signature.