Directed evolution unlocks oxygen reactivity for a nicotine-degrading flavoenzyme

Nicotine addiction makes smoking cessation difficult and is a major contributor to global mortality. Nicotine oxidoreductase (NicA2) degrades nicotine to the non-psychoactive metabolite N-methylmyosmine (NMM) and has emerged as a potential injectable therapeutic. However, wild-type NicA2 requires its native electron acceptor, the cytochrome c CycN, for efficient turnover in vivo; in its absence, such as in the bloodstream where O2 is the predominant oxidant, NicA2 turns over extremely slowly (kcat ~0.006–0.007 s−1). The research question is whether NicA2 can be evolved to efficiently use O2 as an electron acceptor, thereby increasing catalytic activity under therapeutic conditions. The study aims to overcome NicA2’s poor oxygen reactivity by directed evolution and to understand the molecular basis for altered O2 reactivity compared to related flavoenzymes that readily react with O2.

Flavin-dependent amine oxidases typically exhibit high reactivity with O2 (10^3–10^6 M−1 s−1), whereas related dehydrogenases are poorly reactive. Despite decades of study, determinants of O2 reactivity in flavoenzymes remain incompletely understood. Prior attempts to enhance NicA2’s O2-driven activity through rational design achieved only modest gains in nicotine degradation (e.g., single substitutions such as A107R yielding ~3-fold increases in kcat/Km). Structural analogs in other organisms use O2 efficiently for nicotine catabolism, and predefined or transient O2 tunnels have been implicated in governing O2 access in several flavoenzymes. This context motivates an experimental evolution approach coupled with mechanistic probing to identify features that enable O2 reactivity in NicA2.

• Genetic selection: Created a Pseudomonas putida S16 ΔcycN strain that cannot effectively use nicotine as a carbon source due to loss of NicA2’s natural electron acceptor. Bacterial growth on nicotine was thus linked to NicA2’s O2-dependent activity.
• Library construction and selection: Screened a high-diversity nicA2 mutant library (~10^6 clones; average 8.9 nucleotide mutations per gene) in ΔcycN background on nicotine minimal agar (1 g/L) with arabinose induction (0.01–0.001%). Iterative rounds of error-prone PCR (Mutazyme II), enrichment and reselection were performed to isolate variants supporting faster growth. Top variants were subcloned for purification and kinetic analysis.
• Steady-state kinetics: Determined apparent kcat and Km for nicotine under ambient O2 (~250 µM) at 22 °C in 40 mM HEPES, 100 mM NaCl, 10% glycerol, monitoring A280 and fitting to Michaelis–Menten.
• Transient kinetics: Stopped-flow spectroscopy measured oxidation of reduced flavin by O2 in the presence/absence of product (NMM) or analog (myosmine), and reduction of oxidized enzyme by nicotine under anaerobic conditions. Determined bimolecular rate constants for flavin oxidation (kO2) and rate constants for flavin reduction (kred). Assessed electron transfer to CycN.
• Structural studies: Crystallized variant v321 (apo and NMM-bound) and wild-type (NMM-bound) for X-ray crystallography; analyzed potential O2 tunnel with CAVER. Assessed protein dynamics and solvent accessibility via differential hydrogen-deuterium exchange mass spectrometry (HDX-MS).
• Conformational probes: Incorporated 4-trifluoromethyl-L-phenylalanine at Y342 and performed 1D 19F NMR to monitor conformational heterogeneity near the putative O2 tunnel; used paramagnetic relaxation enhancement with TEMPOL to gauge solvent accessibility.
• Biophysical characterization: ThermoFAD melting assay for thermostability; circular dichroism for secondary structure.
• In vivo efficacy: Generated albumin binding domain (ABD) fusions of NicA2 wild-type and v321, injected into rats exposed to chronic nicotine via osmotic minipumps. Quantified plasma nicotine by LC–MS/MS at days 1 and 5 to assess enzyme efficacy in lowering nicotine levels.

• Selection and kinetics:
  • ΔcycN-based selection yielded NicA2 variants with up to ~189-fold higher apparent kcat under ambient O2; the most active variant v417 reached kcat = 1.25 s−1 vs wild type 0.0066 s−1.
  • Despite some increases in apparent Km for nicotine, specificity constants improved up to 10-fold.
  • Frequently mutated positions: F104, A107, D130, H368, N462; beneficial single substitutions included D130S and H368R.
• Oxidase activity and O2 reactivity:
  • Variant v320: kO2 = 4,400 M−1 s−1 (NMM-bound) and 230 M−1 s−1 (ligand-free), representing a 157-fold acceleration of FADH2 oxidation in the NMM-bound state relative to wild type (which showed 120 M−1 s−1 ligand-free and 28 M−1 s−1 with NMM).
  • Myosmine (uncharged analog) yielded kO2 ≈ 2,900 M−1 s−1 for v320, indicating the positive charge of NMM is not required for enhanced O2 reactivity in variants.
  • For v320, the apparent rate constant for oxidation at ~250 µM O2 (ambient) was ~1.1 s−1, matching steady-state kcat, indicating O2 oxidation of the NMM-bound state is rate-limiting in turnover.
• Reductive half-reaction:
  • v320 showed impaired reduction by nicotine: two phases with observed rate constants 42- and 29-fold lower than wild type; the slower reductive steps can become rate-limiting as O2 oxidation is improved, explaining plateauing gains in later selection generations.
  • Electron transfer to CycN by v320 remained comparable to wild type.
• Structural and dynamical insights:
  • Mutations cluster around a putative O2-accessible tunnel near the si-side of the FAD isoalloxazine.
  • HDX-MS: v321 exhibited increased deuterium uptake in regions surrounding the tunnel in both apo and NMM-bound states, indicating greater solvent accessibility/dynamics.
  • 19F NMR at Y342 near the tunnel showed increased conformational heterogeneity in v321 vs wild type; NMM binding induced distinct chemical shift changes and narrowing consistent with a catalytically competent conformation.
  • PRE with TEMPOL indicated the tunnel region in v321 is more solvent accessible than in wild type in both apo and NMM-bound states.
  • Crystallography: NMM-bound structures of v321 and wild type are highly similar (RMSD < 0.5 Å), suggesting differences arise from conformational dynamics rather than static structure.
• In vivo efficacy:
  • In rats with chronic nicotine exposure, ABD–NicA2 v321 reduced plasma nicotine more effectively than wild type. A 0.1 mg/kg dose of v321 achieved nicotine lowering comparable to 1 mg/kg wild type (≈10-fold improvement). At 1 mg/kg, v321 reduced plasma nicotine to undetectable levels.

The study demonstrates that directed evolution can convert NicA2 from a dehydrogenase-like enzyme with poor O2 reactivity into variants that efficiently use O2, particularly when product-bound. The key mechanistic advance is the identification of mutations that increase accessibility and conformational dynamics of a putative O2 tunnel adjacent to the flavin, facilitating O2 diffusion to the active site and accelerating FADH2 oxidation to rate constants within the range of canonical flavin oxidases. Product (NMM) or analog (myosmine) binding shifts the conformational ensemble toward a catalytically competent state in variants, while wild type is inhibited by product during O2 oxidation. Although improvements in O2 oxidation can be offset by slower flavin reduction with nicotine, the evolved balance in variants like v320/v321 yields a net increase in catalytic efficiency and, crucially, improved in vivo performance. These findings provide a framework for controlling O2 reactivity in flavoenzymes via modulation of internal gas channels and conformational dynamics, with direct implications for therapeutic enzyme design.

By coupling NicA2 function to growth of a P. putida ΔcycN strain on nicotine, the authors evolved NicA2 variants with markedly enhanced O2 reactivity and up to 10-fold higher kcat/Km. Mutations near a putative O2 tunnel increase accessibility and conformational flexibility, enabling rapid oxidation of the NMM-bound enzyme by O2. Variants retain CycN reactivity and display robust biophysical properties. In vivo, an ABD-fused NicA2 v321 is approximately tenfold more efficacious than wild type at lowering plasma nicotine in rats, achieving undetectable levels at 1 mg/kg. These advances highlight a viable path to therapeutic development of nicotine-degrading enzymes. Future work should optimize the trade-off between O2 oxidation and nicotine reduction rates, further refine tunnel dynamics, assess immunogenicity and pharmacokinetics, and evaluate efficacy in addiction cessation models.

• The most O2-reactive variants often exhibited increased apparent Km for nicotine and substantially slower flavin reduction rates, which can become rate-limiting and constrain further gains.
• Structural conclusions about v320 relied on v321 due to crystallization challenges; dynamics inferred from proxies (HDX-MS, 19F NMR) rather than direct visualization of O2 pathways.
• Selection conditions (high nicotine) did not strongly select for improved kcat/Km, potentially biasing variant properties.
• Translational considerations such as immunogenicity, dosing regimens, long-term safety, and large-scale manufacturing remain to be addressed in preclinical and clinical studies.