Can engineered E. coli produce active tungsten formate dehydrogenases aerobically?

Why tungsten enzymes resist standard hosts

Tungsten-dependent formate dehydrogenases (W-FDHs) catalyze the reduction of CO2 to formate, and the paper frames them as among the most active biological catalysts for that reaction. Their active site holds tungsten inside a bis-molybdopterin guanine dinucleotide (bis-MGD) cofactor, and the enzyme also needs several iron–sulfur clusters to pass electrons to acceptors such as NAD+ or ethyl viologen.

The difficulty is that E. coli does not naturally prioritize tungsten. Its native machinery builds the molybdenum version of the cofactor, tungsten and molybdenum are chemically similar oxyanions that compete during uptake, and the standard expression strain BL21(DE3) lacks the high-affinity ModABC molybdate/tungstate transporter found in K-12 derivatives. The paper's premise is that expressing the enzyme genes alone therefore yields mostly inactive apoenzyme, and that the host's metal metabolism, not the enzyme, is the bottleneck.

Rebuilding the cofactor pathway stepwise

The authors first asked which maturation genes from M. extorquens AM1 are actually needed. Testing single and combined constructs in BL21(DE3), MeFDH1 expressed alone gave less than 1 U mg-1, and adding any one of moeA1, mobA, mobB, or fdhD alone produced no detectable activity. The combination moeA1–mobB restored about 6 U mg-1, adding mobA raised it further, and the full four-gene set (W4) reached about 27 U mg-1.

Tungsten measurements tracked the same pattern: single-gene constructs showed only basal loading, and the W4 configuration gave the highest tungsten content among the variants tested. A two-hybrid assay also showed MoeA1 interacting with both MobA and MobB, while FdhD preferentially associated with the Mob proteins. The authors interpret this as coordinated, ordered maturation rather than a simple sum of genes, and they note that even W4 did not fully match the native host, so host-level constraints remained.

Tungstate uptake sets the tungsten ceiling

Because W4 still left tungsten occupancy short of stoichiometry, the team tested three tungstate transporters. The TupBCA system from Sediminibacillus dubius NA4 gave the strongest activity increase and the highest intracellular tungsten content among those tested. Chromosomal integration of tupBCA into the W4 strain created BL1, which raised MeFDH1 activity to about 35 U mg-1, soluble yield to about 2.0 mg g-1 wet cell weight, and tungsten occupancy from roughly 25% to 50% of theoretical stoichiometry, matching native-host incorporation.

This is where the metal competition becomes visible. With 1 mM tungstate, activity was about 35 U mg-1; with 0.5 mM tungstate plus 0.5 mM molybdate it fell to about 10 U mg-1; with 1 mM molybdate alone, activity was negligible. Cellular tungsten was only slightly lower in the mixed condition (125 vs 141 µg W g-1 dry cell weight), and molybdenum stayed negligible in the purified enzyme. The authors conclude that tungstate acquisition is necessary but not sufficient, since molybdate can suppress maturation without much changing total cellular tungsten.

Iron–sulfur capacity limits yield and activity

MeFDH1 needs multiple iron–sulfur clusters in both subunits, and E. coli's ISC system is repressed by IscR. The authors deleted iscR and replaced the native isc promoter with the stronger Ptrc promoter, first on a plasmid (BL1ISC) and then chromosomally (BL2). Table 2 shows the progression in purified MeFDH1: iron rose from 1.75 mol per mol enzyme in the enzyme-only strain to 10.8 in BL2, tungsten from 0.01 to 0.92, while molybdenum stayed at 0.02. The native AM1 host gave 11.6 iron and 0.50 tungsten per enzyme.

So BL2 exceeds the native host in tungsten loading while approaching it in iron content. The paper reports that this step also improved specific activity and soluble yield, and it presents volumetric activity as a measure of whole-culture productivity. The interpretation offered is that cofactor metalation and Fe–S cluster supply are separate constraints that must be relieved together; fixing only the tungsten pathway leaves the enzyme short of the clusters it needs.

Native-level turnover but a higher apparent KM

For MeFDH1, the engineered chassis preserved maximal catalytic turnover: the paper states the BL2-produced enzyme had a comparable kcat/KM value to the AM1-derived enzyme, despite higher apparent KM values for both CO2 and the electron acceptor. That distinction matters for readers comparing preparations. Specific activity, tungsten occupancy, and kcat/KM answer different questions, and a preparation can match the native enzyme on turnover while binding substrates less tightly.

The whole-cell experiment is a separate level of evidence. Coupling MeFDH1-expressing cells with carbon monoxide dehydrogenase from Carboxydothermus hydrogenoformans, the BL2-derived system produced 44 mM formate within 4 h versus 13 mM for the AM1-derived system, roughly a 3.5-fold higher final titer and apparent rate; a no-MeFDH1 control produced 1.4 mM. The authors explicitly caution that endogenous FDHs were not removed, so this is a system-level proof of concept rather than an isolated measure of MeFDH1 activity.

One chassis matures diverse tungsten enzymes

The chassis was then tested beyond MeFDH1. CnFDW from Cupriavidus necator H16 was inactive in wild-type BL21(DE3) but reached about 38 U mg-1 and about 0.9 mol tungsten per mol enzyme in BL2, above the roughly 0.6 reported previously for the same enzyme made in a native-like host. ToFDH3 from Thermococcus onnurineus and TkFDH from Thermoanaerobacter kivui, which the paper says were previously produced in anaerobic native hosts, became functional under aerobic cultivation in BL2, at about 29 and about 9 U mg-1 respectively, with tungsten contents near 0.8 mol per mol enzyme.

LbFDH from Lutibaculum baratangense, identified through structural homology rather than prior characterization, was inactive in BL21(DE3) but showed about 8 U mg-1 and about 1.2 mol tungsten per mol enzyme in BL2; no native activity benchmark exists for it. All four accumulated as soluble protein at roughly 6–10 mg g-1 wet cell weight. Note that activity assays differ by enzyme: CnFDW and LbFDH were measured by NAD+-dependent formate oxidation, while ToFDH3 and TkFDH used ethyl viologen-dependent oxidation, so their unit values are not directly interchangeable.

What the chassis does and does not settle

The paper's own framing is that the unresolved problem is coordination, not the mere presence of a cofactor pathway: a heterologous host must combine selective tungstate acquisition, tungsten/molybdenum discrimination, bis-MGD completion, target-enzyme maturation, and Fe–S assembly. BL2 addresses these jointly, and the authors describe it as a transferable framework for aerobic W-FDH production rather than an optimized single-enzyme system. They also state that the precise biochemical order of W-bis-MGD assembly remains to be established.

Two boundaries follow directly from the results. First, the whole-cell CO-to-formate comparison is confounded by unevaluated endogenous FDH activity, which the authors flag and propose removing in future work. Second, the molybdate competition data show that performance depends on the tungsten/molybdenum balance of the medium, not on the chassis alone. Extension to other tungsten enzyme families, particularly those requiring additional maturation such as selenocysteine incorporation, is described as requiring further engineering and validation.

Where the gain actually comes from

The most decision-relevant comparison in this paper is not BL2 versus wild-type E. coli but the stepwise series within the engineered lineage. Tungsten loading in purified MeFDH1 moved from 0.01 to 0.27 with the four Wco genes, to 0.55 with the transporter, to 0.75 with plasmid-borne isc, and to 0.92 with chromosomal isc upregulation, while the native host sits at 0.50. Iron moved from 1.75 to 10.8 across the same series.

That ordering tells readers where to look first. Cofactor genes alone leave the enzyme largely tungsten-free; the transporter is what lifts occupancy toward native levels; and Fe–S reinforcement is what carries loading past the native host. The practical threshold implied by the data is the metal environment of the culture: at 1 mM tungstate the system reaches about 35 U mg-1, but halving tungstate in favor of molybdate cuts activity to about 10 U mg-1 even though cellular tungsten barely changes. For anyone adopting this chassis, tungsten supply and molybdenum exclusion belong in the experimental design alongside the strain genotype.

Sources used in this answer