An engineered E. coli Nissle improves hyperammonemia and survival in mice and shows dose-dependent exposure in healthy humans Original paper

What was studied?

Engineered Escherichia coli Nissle 1917 for hyperammonemia was created to capture gut-derived ammonia and convert it into l-arginine, giving a live, gut-confined therapy for urea cycle disorders and hepatic encephalopathy. The research team built strain SYNB1020 on the EcN chassis, deleted the argR repressor, inserted an anaerobically driven, feedback-resistant argA, and added a thyA deletion to give biocontainment. The strain was tested in vitro for ammonia uptake under low oxygen, then in two mouse models of hyperammonemia and in nonhuman primates to confirm viability and safety. A phase 1 randomized, placebo-controlled trial in healthy adults assessed tolerability, gut exposure, clearance, and activity using 15N-ammonium tracing. The central aim was to show that a probiotic chassis can follow drug-development rules, reach steady state in stool, perform the engineered metabolic task in the lumen, and be cleared after dosing without colonizing.

Who was studied?

Preclinical work utilized ornithine transcarbamylase–deficient SPFash mice, which develop fatal hyperammonemia on a high-protein diet, and thioacetamide-injured mice that model hepatic ammonia overload. Both models mimic patients with urea cycle defects or cirrhosis who cannot detoxify portal ammonia. Mice received oral SYNB1020 at escalating doses and were monitored for blood ammonia and survival. Safety, biodistribution, and excretion were then tested in CD1 mice and cynomolgus monkeys that received up to 10¹² CFU daily for 28 days. Translation was performed in 52 healthy male and female adults in single-ascending and multiple-ascending dose cohorts, who received up to 1.5×10¹² CFU per day for 14 days. They provided serial blood and stool samples after an oral ¹⁵N-ammonium challenge to demonstrate in vivo activity and clearance.

Most important findings

SYNB1020 consumed more ammonia than parental EcN in vitro and released l-arginine in the expected stoichiometry, confirming that the engineered pathway was active. In OTC-deficient mice subjected to a protein load, oral SYNB1020 lowered circulating ammonia in a clear dose-dependent manner and improved 24-hour survival from approximately 40–55% to 100%. In contrast, heat-killed bacteria or non-arginine-producing EcN controls did not provide protection, indicating that the benefit was linked to the engineered pathway. In the liver-injury model, SYNB1020 again reduced blood ammonia levels compared with the vehicle and unmodified EcN. The strain remained within the gut, as evidenced by qPCR, which showed high fecal counts during dosing and undetectable levels in the liver, spleen, bladder, and gonads after dosing stopped. Mice and monkeys cleared the strain within 1–2 weeks.

In healthy adults, SYNB1020 was well tolerated at ≤5×10¹¹ CFU three times daily, reached a fecal steady state by day 2, and disappeared within 14 days after the last dose. Because healthy volunteers tightly regulate ammonia, venous ammonia levels did not decrease; however, 15N-ammonium was recovered as 15N-nitrate in plasma and urine in a dose-related manner, indicating that the engineered EcN was metabolically active in the human gut. These linked animal-to-human data define a microbiome signature of benefit: luminal ammonia excess, intact gut confinement, and presence of an EcN strain carrying an anaerobic arginine module and a thyA safety deletion.

Key implications

This work shows that a live EcN-based drug can be built, dosed, monitored, and cleared like a conventional medicine, while performing a non-host metabolic task that current scavengers cannot do in the lumen. For clinicians, the data suggest a future add-on for urea cycle disorders or hepatic encephalopathy where intestinal ammonia makes a significant contribution and where adherence to lactulose or rifaximin is poor. The biocontainment strategy and the absence of systemic spread address common safety objections to engineered probiotics, which is essential in cirrhotic or immunocompromised patients. The modest biomarker signal in healthy adults also signals a limit: real ammonia lowering will likely appear only in hyperammonemic patients with higher substrate loads, so disease-specific trials are needed. Microbiome databases should tag this strain by chassis (EcN), function (anaerobic ammonia-to-arginine conversion), and containment (thyA deletion) so that future engineered strains can be compared on the same metabolic and safety axes.