Constructing and testing the thermodynamic limits of synthetic NAD(P)H:H2 pathways

Abstract

NAD(P)H:H(2) pathways are theoretically predicted to reach equilibrium at very low partial headspace H(2) pressure. An evaluation of the directionality of such near-equilibrium pathways in vivo, using a defined experimental system, is therefore important in order to determine its potential for application. Many anaerobic microorganisms have evolved NAD(P)H:H(2) pathways; however, they are either not genetically tractable, and/or contain multiple H(2) synthesis/consumption pathways linked with other more thermodynamically favourable substrates, such as pyruvate. We therefore constructed a synthetic ferredoxin-dependent NAD(P)H:H(2) pathway model system in Escherichia coli BL21(DE3) and experimentally evaluated the thermodynamic limitations of nucleotide pyridine-dependent H(2) synthesis under closed batch conditions. NADPH-dependent H(2) accumulation was observed with a maximum partial H(2) pressure equivalent to a biochemically effective intracellular NADPH/NADP(+) ratio of 13:1. The molar yield of the NADPH:H(2) pathway was restricted by thermodynamic limitations as it was strongly dependent on the headspace:liquid ratio of the culture vessels. When the substrate specificity was extended to NADH, only the reverse pathway directionality, H(2) consumption, was observed above a partial H(2) pressure of 40 Pa. Substitution of NADH with NADPH or other intermediates, as the main electron acceptor/donor of glucose catabolism and precursor of H(2), is more likely to be applicable for H(2) production.

Figure 1

Graphic illustration of anaerobic central carbon metabolism of importance for fermentative H₂ production in E. coli BL21(DE3). Pathways introduced in the present study are surrounded by a dotted rectangle. The plasmids employed to generate the pathways in the transformed host are indicated below. G6P, glucose‐6‐phosphate; GAP, glyceraldehyde‐3‐phosphate; PYR, pyruvate; PPP, pentose phosphate pathway; PntAB, membrane‐bound NADPH:NADH transhydrogenase encoded by pntAB; SthA, soluble NADPH:NADH transhydrogenase encoded by sthA; NFOR, NAD(P)H:ferredoxin oxidoreductase; HydA, ferredoxin‐dependent FeFe hydrogenase.

Figure 2

A. The BsNFOR‐ (grey squares) and CtNFOR‐ (unfilled diamonds) dependent H₂ accumulation. The Δ partial H₂ pressure value is calculated by subtracting the partial H₂ pressure value of BL21(DE3)ΔydbK pCDOPFEGA pCpFd pET‐Duet from partial H₂ pressure values of cultures of BL21(DE3)ΔydbK pCDOPFEGA pCpFd pCtNFOR or BL21(DE3)ΔydbK pCDOPFEGA pCpFd pBsNFOR respectively.
B. Same as (A), except the actual partial H₂ pressure values for all three strains are shown [BL21(DE3)ΔydbK pCDOPFEGA pCpFd pET‐Duet, black circles; BL21(DE3)ΔydbK pCDOPFEGA pCpFd pCtNFOR, empty diamonds; BL21(DE3)ΔydbK pCDOPFEGA pCpFd pBsNFOR, grey squares].
C. Growth as determined by optical density (600 nm absorbance), for the same cultures as in (B).
D. Residual glucose (mM) in the media, for the same cultures as in (B).
Error bars indicate standard deviation (n = 3). All experiments were in vessels with a headspace : liquid ratio of 0.60.

Figure 3

A. The theoretical equilibrium point for a NADPH:H₂ (dashed line) and NADH:H₂ (solid line) pathway estimated using Eqn A2 in Appendix S2 assuming that the NAD(P)H/NAD(P)⁺ ratio (cofactor ratio) is fixed. The equilibrium points calculated using cofactor ratios reported for E. coli cultured under anaerobic conditions (Alexeeva et al., 2003; Brumaghim et al., 2003) are highlighted with arrows.
B. In vitro reconstitution of NADPH‐dependent H₂ synthesis using affinity‐purified CtNFOR and HydA in closed N₂‐sparged serum vessels. Standard reaction (filled diamonds) starts with 2.5 mM NADPH and uses serum vessels with a headspace of 5.5 ml. Variations: filled circles, 1.25 mM NADPH and 5.5 ml headspace; unfilled triangles, 2.5 mM NADPH and 12 ml headspace; unfilled squares, same as standard reaction except for addition of 4.5 U ml⁻¹ G6PDH and 0.5 mM G6P. H₂ accumulation was also observed when NADPH was exchanged with NADH (standard reaction conditions: 42.3 ± 18.3 Pa H₂ at the 15 h sampling point). Error bars indicate standard deviation (n = 3).

Figure 4

A. The effect of headspace volume (filled diamonds, headspace : liquid ratio 0.6:1; grey squares, headspace : liquid ratio 5:1; open triangles, headspace : liquid ratio 31:1) on BsNFOR‐dependent molar H₂ production. Molar differences in H₂ production were calculated by subtracting the average partial H₂ pressure value at each time point for cultures of strains harbouring pBsNFOR with average partial H₂ pressure values obtained from cultures of strains harbouring pET‐Duet (i.e. no recombinant NFOR). All BL21(DE3)ΔydbK strains also harboured pCDOPFEGA and pCpFd. The values are averages of differences between replicate culture pairs (n = 3). Error bars display standard deviation of differences between replicate culture pairs.
B. Consumption of H₂ by cultures of BL21(DE3)ΔydbK harbouring pCDOPFEGAFdx and pBsNFOR (grey squares), pCtNFOR (unfilled triangles) or pET‐Duet (filled circles). The headspace of independent replicate cultures was sparged with N₂ 24 h after induction, followed by addition of three different levels of H₂ (0, ∼125 Pa, ∼300 Pa). The actual partial H₂ pressure of each individual culture measured directly following H₂ addition can be seen at t = 24, and the changes in partial H₂ pressure was thereafter monitored for an additional 24 h.
C. The relationship between the BsNFOR‐dependent molar yield after 28 h of cultivation and the headspace to liquid ratio of the cultures shown in (A).
The values are averages of replicate cultures (n = 3).