Redox mediators modify end product distribution in biomass fermentations by mixed ruminal microbes in vitro

One of the greatest needs in developing sustainable alternative energy systems is an economical means of producing energy-dense, infrastructure-compatible liquid fuels (Granda et al. 2007). Mixed cultures of microorganisms can degrade biomass to mixtures of volatile fatty acids (the carboxylate platform; Agler et al. 2011) that can then be converted by further chemistry to useful bioproducts, including liquid fuels (Holtzapple and Granda 2009; Lange et al. 2010; Levy et al. 1981). The carboxylate platform has been highlighted for its ability to be operated non-aseptically, generate high yields, and utilize a wide range of feedstocks, owing to the metabolic diversity of the undefined microbial community (Agler et al. 2011). Within the carboxylate platform, efforts have been made to improve its economics by increasing the value of products while still working within reasonable operating costs and run-times (Agler et al. 2012).

Ruminal fermentation is the means by which ruminant animals convert plant biomass to volatile fatty acids (VFA) that serve as energy source for the host animal. When conducted outside the animal (i.e., in bioreactors), the ruminal fermentation can be considered as a type of consolidated bioprocessing system that has the potential to improve upon the existing carboxylate platform for fuels and chemical production. In a manner similar to other undefined mixed cultures, such as sewage sludge or aquatic sediments, the mixed ruminal bacteria are capable of digesting a wide range of substrates (Weimer 2011). Of particular note for the ruminal bacteria is their capability to produce large amounts of short-chain volatile fatty acids (VFA) from cellulosic substrates in run times as short as 1–3 days. Likewise, the accumulation of methane in these systems is considerably lower than in most anaerobic digestion processes due to a lack of aceticlastic methanogens and proton-reducing acetogens, which require more time for growth than the short 1–3 day ruminal retention times allow, limiting their effect (Weimer et al. 2009). The ability of ruminal bacteria to produce substantial VFA yields has been well-documented and efforts have been made to elongate these short-chain VFAs to medium-chain VFAs, which are more energy-dense and more easily extractable (Singhania et al. 2013). The propensity of the ruminal bacteria towards VFA generation at considerable yield in such short run times deems it worthy of continued study with respect to chain elongation, and what low-cost methods can be used to manipulate the ruminal end-products to more valuable alternatives without altering the initial microbial composition, i.e. without addition of other bacteria. End product manipulation may also be beneficial within the rumen itself, as part of strategy for decreasing methane emissions and retaining feed energy in VFA in vivo, as this is a recognized as a fundamental goal of economically and environmentally sustainable animal agriculture (Hristov et al. 2013).

One important means of contributing to these goals is to better characterize the inherent manipulability of the ruminal fermentation under in vitro (“extraruminal”) conditions. The use of redox mediators has the potential to improve our understanding of how readily fermentation end product ratios may be altered in this system. This study examines the ability of several redox mediators—neutral red, methyl viologen, safranin O, and tannic acid—to decrease methanogenesis and shift ruminal fermentation end-products in vitro in short (72 h) run-times, as a means of demonstrating the inherent manipulability of the ruminal fermentation system.

Fermentations of switchgrass by the mixed ruminal inoculum yielded varying amounts of methane and C₂–C₆ VFA. Product concentrations were notably altered by the addition of all three redox dyes MV, NR, or safranin O (Table 1). MV addition resulted in slightly increased concentrations of caproate (C₆) and a ninefold increase in the sum of the molar amounts of C₄ and C₅ branched-chain VFA (isobutyrate, isovalerate and 2-methylbutyrate), but decreased concentrations of methane and C₂–C₄ VFA (acetate, propionate, and butyrate), and also decreased total carbon in VFA. Safranin O largely inhibited total VFA carbon and numerically decreased caproate and valerate to near their detection limits. Butyrate production did increase, however, which led to an increase in average carbon chain length to 2.61, compared to 2.42 in the mediator-free control. NR inhibited methane production (by 56%) and significantly increased valerate, caproate, and C₄–C₅ branched chain VFA, and numerically increased total VFA carbon production. Comparison of the data on a per substrate added basis revealed that only NR numerically increased total end product C yield while enabling a shift in fermentation to yield longer-chain products (Fig. 1).

Product	Net umol producta				SEDb	Model P > F
	Methyl viologen (0.5 mM)	Neutral red (1.0 mM)	Safranin O (1.0 mM)	Control (no dye)
Methane	122.1 A	62.1 B	38.5 B	139.2 A	9.9	<0.0001
Acetate	269.8 A	182.0 B	142.9 B	305.6 A	13.7	<0.0001
Propionate	60.4 C	119.7 A	48.0 C	91.0 B	5.0	<0.0001
Butyrate	24.0 B	40.2 AB	48.8 A	38.8 AB	5.9	0.018
Isobutyrate	9.7 A	4.3 B	0.8 C	1.8 C	0.7	<0.0001
Valerate	6.8 B	35.9 A	1.5 C	5.3 BC	1.4	<0.0001
Isoval + 2 MBc	19.1 A	14.4 B	0.1 C	1.4 C	1.0	<0.0001
Caproate	6.7 A	8.1 A	0.3 B	2.1 B	1.6	0.003
Total VFA	399.1 A	404.1 A	242.3 B	441.5 A	15.4	<0.0001
Total VFA-C	1025.6 B	1201.3 A	629.6 C	1,092.0 AB	40.5	<0.0001
ACLd	2.58 B	2.97 A	2.61 B	2.42 C	0.04	<0.0001
Enthalpy of combustion, kJ	0.602 B	0.667 A	0.343 C	0.633 AB	0.009	<0.0001

Vials contained 100 mg switchgrass and 3.0 mL of diluted mixed ruminal inoculum in a total liquid volume of 10 mL.

^aValues are least-square means from triplicate cultures after 72 h incubation, after correction for inoculated blank vials lacking switchgrass. Values within row having different letters (A, B, C) differ (P < 0.05).

^bPooled standard error of the difference.

^cIsovalerate plus 2-methylbutyrate (chromatographically co-eluting isomers).

^dAverage chain length of VFA.

Owing to the significant improvement in C₅ and C₆ VFA production resulting from NR addition, the effect of NR concentration on the fermentation was examined (Fig. 2). Addition of NR in increasing concentrations up to 0.6 mM resulted in linear increases in propionate (P < 0.001), butyrate (P = 0.099), valerate (P = 0.002), caproate (P < 0.001) and average VFA chain length (ACL; P < 0.0001). Caproate concentrations increased linearly (P < 0.001) from an undetectable concentration (< 0.05 mM) in the absence of NR to 2.2 mM at the highest NR concentration, and valerate levels increased by over fourfold to 8.8 mM. These increases were accompanied by linear decreases (P < 0.0001) in both methane and acetate. At 0.6 mM NR, the highest concentration tested, methane production decreased by 52% relative to the mediator-free control. Total VFA carbon did not significantly differ among the range of NR concentrations tested (range 434–459 mM, SED = 24.4 mM, P = 0.59).

In the above fermentations inhibition of methanogenesis by NR was accompanied by accumulation of modest amounts of both H₂ and formate. Fermentations containing either 0 or 0.008 mM resazurin with NR (Table 2) averaged 34.9 μmol H atoms in measured fermentation products (mg switchgrass digested)⁻¹, numerically less than the 40.5 μmol H atoms (mg switchgrass digested)⁻¹ in parallel fermentations lacking NR, but this difference was not significant. However, total H atoms in measured fermentation products (mg switchgrass digested)⁻¹ tended (P = 0.089) to be lower in the presence versus absence of NR without resazurin, and was nearly so (P = 0.115) with 0.008 mM resazurin, suggesting that there may be additional fermentation products not accounted for in the NR-supplemented cultures. As expected, all cultures had similar enthalpies of combustion on a per mg switchgrass digested basis.

Shifting metabolic products of plant matter degradation to decrease methanogenesis and increase VFA is desirable within the rumen itself (as it should increase energetic efficiency and decrease greenhouse gas emissions) and in extraruminal bioreactors (as it should foster chain elongation to higher-energy, medium chain-length carboxylates). The inherent manipulability of the ruminal fermentation has largely been explored by altering substrate (i.e., diet composition) or by adding specific metabolic inhibitors (e.g., of methanogenesis), but the limits to which the fermentation can be altered, and the mechanisms underlying these alterations, have not been firmly established (Ungerfeld 2015). In this study we used artificial electron acceptors (redox mediators) to examine fermentation end product distribution on a single plant biomass substrate (ground switchgrass not subjected to chemical pretreatment).

The capability of redox dyes to shift anaerobic fermentation towards formation of longer end-products was first examined in non-ruminal systems by Hongo (1958). By diverting reducing equivalents that are normally released as H₂ towards formation of NAD(P)H, redox dyes can inhibit methanogenesis by decreasing the supply of H₂, and provide NAD(P)H for use as reducing power to synthesize longer chain molecules. Whether a redox mediator may divert electron flow depends on its capability to compete with a natural electron carrier in vivo. MV has a similar redox potential to ferredoxin and thus may replace ferredoxin in hydrogenase-catalyzed reactions (Peguin et al. 1994). That MV may replace ferredoxin allows it to act as a substitute for the direct reduction of NAD⁺. Accumulation of NADH inhibits further NAD⁺ reduction by either MV or ferredoxin (Rao and Mutharasan 1987). This is the mechanism for production of longer chain products with redox dyes: because a second electron carrier is present, the natural electron carrier will accumulate unless the reaction is coupled to the further synthesis of longer chain molecules that decreases the amount of NADH present in the cell.

An increased availability of NADH leads to increased production of VFAs, such as valerate and caproate. This is accomplished by reverse β-oxidation, which sequentially adds acetyl (C₂) units to form butyrate from two molecules of acetate, valerate from acetate and propionate, and caproate from acetate and butyrate (Agler et al. 2012). This is most clearly demonstrated by a sevenfold increase in valerate upon the addition of 1 mM NR (Table 1). Interestingly, the levels of propionate were higher in the NR cultures than in the control, while acetate levels decreased by almost half. Not only did NR increase the concentration of C₅–C₆ VFAs, but also C₃–C₄ VFAs (propionate and butyrate) as well. So while the average chain length of end-products increased from 2.42 to 2.97, decreases in end-products only occurred with acetate and methane, the least desirable products in terms of energy content and commercial value.

The levels of total carbon in VFA resulting from MV or NR additions were nearly identical. Of interest is that MV did a significantly poorer job of inhibiting methanogenesis and facilitating chain elongation of acetate. Although Bauchop (1967) demonstrated that the related dye benzyl viologen decreased methanogenesis by mixed ruminal microbes in vitro, the effects on the amounts and distribution of other end products were not reported. The capability of MV to serve as an electron carrier for hydrogenase and increase the availability of NADH for VFA chain elongation was observed by Peguin et al. (1994) who showed MV to increase butyrate concentrations to 0.65 mol (mol glucose)⁻¹ in pure cultures of the non-ruminal solventogenic bacterium, Clostridium acetobutylicum. In our experiments, the amount of butyrate was numerically lowered by MV, but the caproate concentration rose to over three times that of the control. This confirms not only that MV is capable of generating butyrate at the expense of shorter products, but also that a second chain elongation to caproate took place despite the short run times, although to relatively modest concentrations.

Of further interest is that NR shifted the fermentation to longer-chain end-products in a concentration-dependent manner. NR was also unaffected by the addition of resazurin as a redox indicator for reducing conditions of the ruminal media. Within the range of concentrations tested, linear decreases in acetate and methane along with increases in C₃–C₆ VFA occurred upon additions of greater amounts of NR. This demonstrates that there may be no specific threshold for NR activity. At what NR concentration the linear functions of methane inhibition and VFA chain extension dissipate may depend on the toxicity of the end-products.

At low concentrations, TA had no effect on fermentation product profiles, probably due to its relatively positive redox potential. At the highest concentration tested (1.0 g L⁻¹), TA displayed a slight inhibition of switchgrass degradation and a corresponding decrease in methane and propionate formation, and in the ACL of VFA products. In addition, it decreased yield of C₄–C₅ branched-chain VFA, which are considered to be produced exclusively by fermentation of branched-chain amino acids (Russell 2002). It is likely that the TA-mediated inhibition of branched-chain VFA production reflects the known ability of tannins to bind proteins and protect them from ruminal degradation (Min et al. 2003).

In this report, we show that certain redox mediators may be an effective means of extending carbon chain length, and we highlight NR as the most effective of the dyes examined. NR is capable of increasing the average carbon chain length of end-products from 2.42 to 2.97 and inhibiting methanogenesis by over half, all while including no additionally added substrates/reagents and keeping fermentation run times at 72 h. The fact that shifts in end product distribution vary dramatically among different redox mediators that differ only slightly in their reduction potential reinforces bioenergetics studies that the ruminal fermentation is primarily under thermodynamic control (Ungerfeld and Kohn 2006). The manipulability of the ruminal system to form higher value fermentation end-products in vitro is clear, and it is a system worthy of further study to identify other, more practical interventions for shifting end product distribution, for potential use on an industrial scale.