A Two-Residue Swap in an Ancient Enzyme Motif Lets Biocatalysts Run on a Cofactor Cells Don't Already Use

A team led by chemical engineer Han Li at the University of California, Irvine, working with collaborators at the University of California, Davis, has redesigned a two-billion-year-old enzyme motif so that engineered biocatalysts can run on a cofactor the cell does not already use for anything else. The work, published in Nature Communications on 12 September 2026, reports that one redesigned enzyme became up to 240-fold more productive once it stopped sharing its chemical fuel with the rest of the cell's metabolism.

The fuel in question is nicotinamide adenine dinucleotide, NAD(P)H, the molecule that carries reducing power — essentially spare electrons — to a broad swath of the enzymes any cell relies on. That ubiquity is the problem for anyone trying to engineer a cell to make a single target chemical: the same pool of NAD(P)H that powers a desired reaction is also drawn on by hundreds of native reactions the engineer never asked for, capping how much of the electron supply actually reaches the product.

Researchers have spent several years trying to build biochemical pathways that use a different currency altogether — a cofactor called nicotinamide mononucleotide, or NMN(H), which is structurally similar to NAD(P)H but missing one of its two phosphate groups. If enzymes could be made to accept NMN(H) and refuse NAD(P)H, engineers could isolate a pathway's electron flow from the cell's native chemistry entirely. Han Li's group had already shown, in an earlier Nature Communications paper from 2022, that a growth-selection system built around an NMN(H)-dependent glycolytic pathway could push evolution toward enzymes with better noncanonical-cofactor activity. The new paper asks a more structural question: what does it actually take, at the level of a single conserved motif, to make an enzyme stop recognizing its natural cofactor.

A conserved motif, deliberately broken

Most NAD(P)H- and FAD-binding enzymes share a short sequence pattern called the GxGxxG motif, part of the Rossmann fold that positions the cofactor inside the enzyme's active site. The team analyzed 56 reference crystal structures — 24 NAD(H)-binding, 16 NADP(H)-binding and 16 FAD-binding enzymes — and found the motif's role in anchoring the cofactor's pyrophosphate backbone is conserved across all three groups. That conservation is exactly what let them design a general intervention rather than a one-off fix: mutating the motif's second glycine to glutamine, paired with a second, complementary substitution that forms a hydrogen bond, selectively blocks the beta-phosphate that dinucleotide cofactors carry and NMN(H) lacks, while leaving the alpha-phosphate site that both cofactors share untouched.

What changed in the two model enzymes

The team tested the design on two very different starting enzymes. Phosphite dehydrogenase, engineered into variants named NRC-01 and NRC-02, showed the clearest result: NRC-01's affinity for NAD+ dropped 4.5-fold and for NADP+ 68.5-fold relative to its parent enzyme, and the redesigned enzyme drove NMN(H)-dependent biotransformation with roughly 240-fold higher productivity than an existing engineered catalyst used for comparison. In a test reaction producing the fragrance compound levodione, crude cell lysates reached about 9.4 millimolar product at 94% conversion, with electron flow going almost exclusively through the intended NMN(H) route rather than leaking into NAD(P)H-linked side reactions.

The second enzyme, glycolytic GapA, told a less finished story. Its redesigned variant, RSQ, shifted cofactor specificity by roughly 29,000-fold away from NAD+, and its residual affinity for NAD+ and NADP+ fell so far — with Km values of 7.6 millimolar and 1.8 millimolar respectively — that both now exceed physiological concentrations, meaning the enzyme effectively can no longer use them in a cell. But the authors are explicit that this success is only half the job: in their own words, "GapA RSQ's NMN+ activity is still low," and pushing it up is, as they put it, "ongoing work in our lab." That is a preliminary result on the activity side even where the specificity switch itself is well established.

Why an isolated cofactor matters for biomanufacturing

The appeal of an orthogonal cofactor is that it removes the built-in trade-off between yield and the cell's own housekeeping demands. If a synthetic pathway's electrons flow through NMN(H) alone, the pathway's redox ratio can in principle be tuned independently of the NAD(P)/H pool the rest of the cell depends on, pushing engineered biomanufacturing routes closer to their theoretical maximum yield rather than losing product to native side reactions. Li's group has separately secured a $7.5 million award from the National Science Foundation in 2025 to build a cell-free biomanufacturing process for chiral chemicals used as pesticide precursors, one indication of the manufacturing problems the group's cofactor-engineering work is aimed at addressing.

What is still open

The paper frames the GxGxxG substitution as a general design principle rather than a finished tool: the phosphite dehydrogenase result demonstrates that an engineered enzyme can be both highly specific for NMN(H) and highly productive, but the glyceraldehyde-3-phosphate dehydrogenase result demonstrates specificity without yet demonstrating comparable activity. The authors propose that once a direct link between NMN(H)-dependent enzyme activity and cell growth is established for a given target reaction, high-throughput directed evolution — the same growth-selection strategy the group described in 2022 — could be used to close that gap. Whether the motif-swapping strategy generalizes cleanly across the wider set of NAD(P)H- and FAD-dependent enzymes it was designed around remains to be tested case by case.