Expanding the substrates for a bacterial hydrogenlyase reaction

Escherichia coli produces enzymes dedicated to hydrogen metabolism under anaerobic conditions. In particular, a formate hydrogenlyase (FHL) enzyme is responsible for the majority of hydrogen gas produced under fermentative conditions. FHL comprises a formate dehydrogenase (encoded by fdhF) linked directly to [NiFe]-hydrogenase-3 (Hyd-3), and formate is the only natural substrate known for proton reduction by this hydrogenase. In this work, the possibility of engineering an alternative electron donor for hydrogen production has been explored. Rational design and genetic engineering led to the construction of a fusion between Thermotoga maritima ferredoxin (Fd) and Hyd-3. The Fd-Hyd-3 fusion was found to evolve hydrogen when co-produced with T. maritima pyruvate :: ferredoxin oxidoreductase (PFOR), which links pyruvate oxidation to the reduction of ferredoxin. Analysis of the key organic acids produced during fermentation suggested that the PFOR/Fd-Hyd-3 fusion system successfully diverted pyruvate onto a new pathway towards hydrogen production.

In this work, pyruvate was explored as an alternative nonnatural substrate for H 2 production from E. coli Hyd-3. A rational design approach was taken to covalently attach the ferredoxin from Thermotoga maritima to Hyd-3 via the HycB subunit. T. maritima Fd and PFOR plasmids were readily available [10]. To begin, strains were constructed where the natural electron donor enzyme for FHL, FdhF, was genetically removed (Table 1) using an available DfdhF allele [11]. In vivo hydrogen production assays involved measuring the accumulation of H 2 in the headspace (10 ml) of anaerobic cultures (5 ml) in Hungate tubes containing 0.8 % (w/v) glucose. Following incubation at 37 C, H 2 was quantified using gas chromatography (Shimadzu GC-2014) with N 2 as carrier (25 ml min À1 ). The fdhF mutation resulted in a reduction in H 2 -evolution activity of 1000 times compared to the original parent strain (Fig. 1a, b). This fdhF mutant phenotype was repeated in a strain carrying a chromosomal hycE His allele ( Table 1, Fig. 1b).
FHL subunits share similarity with the respiratory NADH dehydrogenase encoded by nuoA-N [3,12,13]. A DnuoA-N allele, marked with apramycin resistance from pIJ773 [14], had no effect on the ability of E. coli FGB300 or FTD300 (Table 1) to grow under fermentative conditions or the amount of H 2 produced (Fig. 1a). Next, DfdhF and DnuoA-N alleles were combined in a single strain (MG300dZ) and the double deletion was found to reduce the residual H 2  production further still (Fig. 1b). It is therefore possible that the very low levels of residual H 2 produced in the fdhF mutants results from reversed electron transport through Hyd-2 [15].
Next, a DhycA :: 'fd-hycB allele was generated that encoded a fusion of T. maritima Fd to HycB via an HA epitope tag. Also, to upregulate expression of this fusion, the synthetic T5 promoter, lac operator and ribosome binding site from strain FZBup [11] was included to give a DhycAB :: P T5 'fd-hycB allele. Two strains, FTF2013 and FTF2015, were constructed (Table 1) and in vivo H 2 evolution activity quantified (Fig. 1c). The FTF2013 and FTF2015[pREP4] strains were transformed with pUNI-PROM (empty control vector), pUNI-Tm-POR (encoding T. maritima PFOR) or pUNI-Tm-Fd-POR (encoding T. maritima PFOR and Fd) then grown at 37 C for 24 h in anaerobic Hungate tubes containing 5 ml M9 medium supplemented with 0.8 % (w/v) glucose and 0.2 % (w/v) casamino acids. The FTF2013 strain produced H 2 at basal levels regardless of the presence of plasmids (Fig. 1c). This basal level was mirrored in the FTF2015[pREP4]/pUNI-PROM strain (Fig. 1c). However, when the PFOR plasmid was introduced into FTF2015 [pREP4] hydrogen, evolution increased to >40 nmol H 2 OD À1 ml À1 (Fig. 1c). Moreover, the vector encoding both PFOR and extra Fd induced H 2 production to a maximal level of >60 nmol H 2 OD À1 ml À1 in the presence of IPTG (Fig. 1c).
The levels of the most common organic acids produced during mixed-acid fermentation were investigated for strains producing active Fd-Hyd-3/PFOR (Fig. 2). Strains were grown for 24 h in 16 ml LB medium supplemented with 0.8 % (w/v) glucose. Culture supernatants were then passed  through a 0.2 µm filter and analysed with an Aminex HPX-87H organic-acid column at 55 C and 0.5 ml min À1 . Organic acids were detected by A 210 nm and compared to standard curves. Representative HPLC traces are shown in Fig. S1 (available in the online Supplementary Material). The starting concentration of glucose added to the rich medium was 44 mM D-glucose and, under the growth conditions chosen, the MC4100 FHL-positive strain produced 1.5 mM OD 600 À1 of formate (Fig. 2a) compared with 30.6 mM OD 600 À1 for the FTF2013/pUNI-PROM strain (inactive for FHL). Importantly, when the PFOR, Fd and Fd-Hyd-3 system is produced at its maximum level, the extracellular formate was observed to drop back to 5.7 mM OD 600

À1
, which is indicative of pyruvate being directed away from the endogenous pyruvate formatelyase (PFL) enzyme to the fusion protein.
Extracellular lactate levels were found to be high in FTF2013 (Fig. 2b). This may mean that the higher formate levels (Fig. 2a) are inhibiting PFL leading to an accumulation of pyruvate and thus extra substrate for lactate dehydrogenase. Indeed, pyruvate can be detected in the growth medium (Fig. 2c) and its level does follow that of formate and lactate in the FTF2013 mutant strains (Fig. 2c). Although pyruvate would not normally be located outside the cell, E. coli is known to possess a pyruvate exporter to balance metabolite levels [16], and so any extracellular pyruvate levels may also correlate somewhat with that inside the cell cytoplasm. Importantly, in all cases, when the PFOR, Fd and Fd-Hyd-3 system is maximally produced (FTF2015/pUNI-Tm-Fd-POR + IPTG), the balance of pyruvate/lactate/formate returns to the low levels seen in the FHL-positive strain (Fig. 2c).
Normally, acetate levels are linked to that of acetyl CoA via phosphate acetyltransferase and acetate kinase. The observed increase in extracellular acetate (Fig. 2d) may mean a concomitant increase in cytoplasmic acetyl CoA, which is normally competed for by the AdhE-dependent ethanol production pathway. This is feasible and could be a consequence of the increased activity of the NADH-dependent lactate dehydrogenase already noted for these strains, which would reduce the requirement for AdhE to recycle NAD + and allow acetyl CoA to be used for ATP and acetate production instead.
Together, these data demonstrate the successful repurposing of E. coli Hyd-3 to accept electrons from a new substrate: reduced ferredoxin linked to pyruvate :: ferredoxin oxidoreductase. Examples of native [NiFe]-hydrogenase :: ferredoxin interactions are not common; however, Synechocystis sp. PCC 6803 does contain such a system [17]. Physical tethering of a ferredoxin to an [FeFe]-hydrogenase, as opposed to a [NiFe]-hydrogenase, from Chlamydomonas reinhardtii showed that photosystem I could be coupled directly to H 2 production [18,19]. Similarly, photosynthesis-linked ferredoxins have been fused to cytochromes P450 [20]. Functional fusion of a ferredoxin to [NiFe]hydrogenases has resulted in some activity in vitro [21]; however, the 'Fd-Hyd-3 enzyme described in these experiments is one new example of a functional fusion that is active in the living cell.