Improvement of a synthetic live bacterial therapeutic for phenylketonuria with biosensor-enabled enzyme engineering Original paper

What was studied?

Engineered Escherichia coli Nissle 1917 phenylketonuria therapy was developed to improve degradation of phenylalanine in the gut using a second-generation live biotherapeutic, SYNB1934, built on the earlier strain SYNB1618. The investigators reasoned that PKU patients need an oral, lumen-active, genotype-independent way to clear excess phenylalanine before it enters the bloodstream. They created a trans-cinnamate biosensor, screened a >1-million–member PAL mutant library directly in EcN, and identified an evolved PAL (mPAL) with higher whole-cell activity. They then integrated mPAL and the same auxiliary Phe-uptake and LAAD modules used in SYNB1618 into the EcN chromosome, along with auxotrophic containment. They compared activity in simulated gut conditions, in non-human primates, and against the clinical first-generation strain.

Who was studied?

Library construction and biosensor screening were conducted in the probiotic EcN, ensuring that enzyme improvements reflected the physiology of the final chassis. Functional testing was performed using in vitro gastrointestinal simulations. Nonhuman primates were then given an oral peptide plus deuterated phenylalanine load to model postprandial Phe entry, followed by a single oral dose of SYNB1618 or SYNB1934 to compare the production of the strain-specific biomarker trans-cinnamate and its hepatic product, d5-hippurate. Previous human data from the earlier strain demonstrated safety, gut confinement, and biomarker output, positioning SYNB1934 as a more potent successor for future PKU trials in patients who fail or cannot tolerate existing PAH-dependent drugs.

Most important findings

The biosensor-guided “pop’n’sort” workflow enriched PAL variants with 25–100% higher whole-cell activity than wild type, and most retained activity even after brief acid exposure, making them suitable for oral dosing. One variant carrying S92G, H133M, I167K, L432I, and V470A substitutions became the core of SYNB1934 because it improved catalytic turnover without adding phenylalanine to the cell mass. In gut simulation, lyophilized SYNB1934 generated about twice as much trans-cinnamate per 2.5×10⁹ cells as SYNB1618, confirming a genuine strain-level gain. In nonhuman primates challenged with peptide and d5-phenylalanine, SYNB1934 produced approximately twofold higher plasma exposure to TCA and d5-TCA, and more than twofold higher urinary d5-hippurate levels than SYNB1618, demonstrating superior in vivo phenylalanine capture and conversion. Because wild-type EcN does not make these metabolites, they serve as clean strain-specific readouts. The authors therefore defined a microbiome therapeutic signature for PKU: an EcN backbone, chromosomally integrated high-activity PAL, PheP-mediated substrate uptake, a LAAD backup pathway, and a dapA-based biocontainment mechanism, with biomarker recovery in both plasma and urine.

Key implications

Clinicians can read this as proof that live-engineered EcN can be iteratively improved to deliver more phenylalanine-lowering activity without changing the route or the safety concept. A more active strain should allow either lower doses, fewer capsules, or greater phenylalanine disposal in patients who stay above target despite diet or who cannot take injectable PAL. Because the system operates within the lumen, it should be effective across all PKU genotypes, including those with no residual PAH. The work also shows that future EcN therapeutics should be reported with their sensor, payload, and containment modules, since minor enzyme upgrades can translate into clear systemic biomarker gains and can be tracked for precision microbiome therapy.