Soluble sugar substrates can enter C. saccharolyticus cells either as mono-, di- or oligo-saccharides via several ABC-transporters, which facilitate substrate transport across the membrane at the expense of ATP. In addition, C. saccharolyticus contains one fructose specific phosphotransferase system (PTS). During PTS-mediated transport the substrate is both transported and phosphorylated at the expense of phosphoenolpyruvate (PEP). The substrate specificity of the 24 sugar ABC transport systems, identified in C. saccharolyticus, has been assigned based on bioinformatic analysis and functional genomics [53]. Most of the identified ABC transporters have a broad predicted substrate specificity and for some transporters the annotated substrate specificity was confirmed by transcriptional data obtained from cells grown on different mono-saccharides [53]. Some substrates can be transported by multiple transporter systems (Figure 1).

The growth of C. saccharolyticus on sugar mixtures revealed the co-utilization of hexoses and pentoses, without any signs of carbon catabolite repression (CCR) [33,53]. The absence of CCR is in principle a very advantageous characteristic of C. saccharolyticus, as it enables the simultaneous fermentation of hexoses and pentoses [33]. Substrate co-utilization has also been confirmed for biomass derived hydrolysates [54,55]. Despite the absence of CCR, a somewhat higher preference for the pentose sugars (xylose and arabinose) with respect to the hexose sugars (glucose, mannose and galactose) was demonstrated, but the highest preference was observed for the PTS-transported hexose fructose [53].

Once inside the cell, the sugar substrates are converted into a glycolytic intermediate. NMR analysis of the fermentation end-products of C. saccharolyticus grown on ¹³C-labeled glucose showed that the Embden-Meyerhof pathway is the main route for glycolysis [56]. All genes encoding components of the Embden-Meyerhof and non-oxidative pentose phosphate pathway have been identified in the genome. There is no evidence for the presence of the oxidative branch of the pentose phosphate pathway or the Entner-Doudoroff pathway [33] (Figure 1).

Each sugar substrate, with the exception of rhamnose is completely catabolized to glyceraldehyde 3-phosphate (GAP) (rhamnose catabolism will be discussed separately in more detail below). The subsequent conversion of GAP to pyruvate, via the C-3 part of glycolysis, results in the formation of the reduced electron carrier NADH. Pyruvate can be further oxidized to acetyl-CoA by pyruvate: ferredoxin oxidoreductase (POR), which is coupled to the generation of reduced ferredoxin (Fd_red). Finally, acetyl-CoA can be converted to the fermentation end-product acetate (Figure 1).

Both types of reduced electron carriers (NADH and Fd_red) can be used by hydrogenases for proton reduction, thus forming H₂. The genome of C. saccharolyticus contains a gene cluster coding for an NADH-dependent cytosolic hetero-tetrameric Fe-only hydrogenase (hyd) and a cluster encoding a membrane bound multimeric [NiFe] hydrogenase (ech), which presumably couples the oxidation of Fd_red to H₂ production [33]. Under optimal cultivation conditions, when all reductants are used for H₂ formation, the complete oxidation of glucose yields 4 mol of H₂ per mol of glucose consumed (Equation 1).

However, suboptimal growth conditions lead to a mixed acid fermentation with ethanol (Equation 2) and lactate (Equation 3) as end products in addition to acetate and H₂. Lactate is produced from pyruvate, using NADH as electron donor and catalyzed by lactate dehydrogenase (LDH). The corresponding ldh gene could be easily identified in the genome [33,57]. However, the identity of the enzymes and genes involved in ethanol formation, are less clear. Two alcohol dehydrogenase (ADH) genes have been identified in the genome which, based on transcriptional data, can both be involved in ethanol formation from acetaldehyde. The way acetaldehyde is produced, is however, not known. Acetaldehyde can be produced from pyruvate by a pyruvate decarboxylase, as described for yeast, or from acetyl-CoA by an acetaldehyde dehydrogenase, as is commonly seen in fermentative bacteria. In several thermophilic ethanol-producing bacteria, acetaldehyde is produced by a bifunctional acetaldehyde/ethanol dehydrogenase [58,59]. However, no candidate gene could be identified for any of these alternatives [33,60]. A third option might be that acetaldehyde is formed from pyruvate in a CoA-dependent side reaction of the pyruvate:ferredoxin oxidoreductase as described for Pyrococcus furiosus [61]. There is no experimental evidence for such a side reaction in C. saccharolyticus, but the absence of a dedicated enzymatic acetaldehyde-forming step might explain the low ethanologenic capacity of C. saccharolyticus. The difference in the observed lower ethanol to acetate ratio for C. saccharolyticus with respect to Clostridium thermocellum [16] can be explained by the fact that the C. saccharolyticus genome does not have a similar pathway for ethanol formation as identified in C. thermocellum [58]. Similar to other high yield ethanol-producing thermophiles, like Thermoanaerobacter ethanolicus [62] or Thermoanaerobacterium saccharolyticum [63], Clostridium thermocellum has a bifunctional acetaldehyde-CoA/alcohol dehydrogenase, which catalyzes ethanol formation from acetyl-coA [58,59]. Such a pathway is absent in low level ethanol-producing thermophiles like C. saccharolyticus [33], Thermoanaerobacter tengcongensis [64] and Pyrococcus furiosus [65]. Although C. saccharolyticus is able to produce some ethanol, the flux through the ethanol-forming pathway is apparently limited resulting in the lower ethanol to acetate ratio.

NADPH is assumed to be the preferred substrate for the ethanol-forming ADH reaction [33,57]. A potential source of NAPDH could be the isocitrate dehydrogenase in the oxidative branch of the incomplete TCA cycle (Figure 1). Alternatively, NADPH can be produced from NADH, but so far no candidate genes coding for such transhydrogenase have been identified in C. saccharolyticus [33,60]. With respect to H₂ production, it is important to realize that only the production of acetate is coupled to H₂ formation since no net reducing power remains for H₂ formation when ethanol or lactate is produced.

Compared to glucose, fermentative growth on the deoxy sugar rhamnose is associated with a different carbon and electron metabolism. The proposed pathway for rhamnose degradation (Figure 1) implies that during rhamnose catabolism half of the generated reduced electron-carriers is used for the reduction of lactaldehyde to 1,2-propanediol, while the other half can be recycled through H₂ formation [33]. Indeed, fermentation of rhamnose by C. saccharolyticus results in the production of 1,2-propanediol, acetate, H₂ and CO₂ in a 1:1:1:1 ratio ([66], Figure 2a). This ratio suggests that all NADH is used for 1,2-propanediol formation and that all Fd_red is used for H₂ formation. However, when C. saccharolyticus is grown on rhamnose under a headspace of carbon monoxide (CO), which is an established competitive inhibitor of both NiFe- [67] and Fe-only [68] hydrogenases, H₂ evolution is significantly inhibited ([66], Figure 2b). The CO cultivation condition does not affect 1,2-propanediol formation, but less H₂ is produced and the remaining reduced electron-carriers are now recycled by the formation of lactate and ethanol, consequently leading to a decrease in the acetate level. These findings suggest that, based on the substrate specificity of the lactate dehydrogenase (NADH, [57]) and the ethanol dehydrogenase(s) (NADPH, [57]), electron exchange between the Fd_red and NAD(P)⁺ is required. However, no genes have been identified in C. saccharolyticus coding for an enzyme capable of catalyzing such a reaction [33,60]. Transcript levels of the genes from the gene cluster containing both the L-rhamnose isomerase and L-rhamnulose-1-phosphate aldolase are highly upregulated during growth on rhamnose (Figure 3) [33]. Although the function of the other genes in the cluster with respect to rhamnose catabolism remains unclear, the absence of such gene cluster and the rhamnose kinase gene from the genome of other Caldicellulosiruptor species, strongly correlates with their inability of rhamnose degradation (Table 1). This difference in rhamnose degrading capability between Caldicellulosiruptor species reflects the open nature of the Caldicellulosiruptor pan genome [29].

From a bioenergetics point of view, glucose fermentation is optimal when acetate is the only end product, because an additional ATP is generated during the final acetate-forming step (Figure 1). When it is assumed that the ABC-transporter mediated substrate uptake requires 1 ATP, overall ATP yields become 1.5 ATP per acetate versus only 0.5 ATP per lactate or ethanol. However, ATP yields might even be higher when the involvement of pyrophosphate (PP_i) as an energy carrier is considered.

PP_i is a by-product of biosynthesis reactions like DNA and RNA synthesis or is generated when the amino acids are coupled to their tRNAs during protein synthesis [69]. Since these reactions are close to equilibrium accumulation of PP_i is believed to have an inhibitory effect on growth, and only the effective removal of PP_i drives these biosynthetic reactions forward [70]. When PP_i is hydrolysed to P_iby a cytosolic inorganic pyrophosphatase (PPase), the free energy just dissipates as heat. C. saccharolyticus does not contain a cytosolic PPase but possesses a membrane bound H⁺-translocating PPase [71], which allows the free energy released upon PP_i hydrolysis to be preserved as a proton motive force. The high-energy phosphate bond of PP_i can also be used for the phosphorylation of fructose 6-phosphate, catalyzed by a PP_i-dependent phosphofructokinase (Figure 1).

Furthermore, PP_i is consumed during the catabolic conversion of phosphoenolpyruvate to pyruvate, catalyzed by pyruvate phosphate dikinase (PPDK). Such a catabolic role for PPDK was proposed based on the increase in transcript level of the ppdk gene under increased glycolytic fluxes [33,71]. Altogether, the use of PP_i as an energy donor could be a way for the organism to deal with the relative low ATP yields which are usually associated with fermentation [72].

H₂ has been reported as a growth inhibitor for C. saccharolyticus [16,73] and a critical dissolved H₂ concentration, which leads to the complete inhibition of growth, of 2.2 mmol/L has been determined for controlled batch cultivations [74]. In addition, an elevated P_H2 has been shown to cause a switch in the fermentation profile, leading to increased formation of ethanol and lactate, in both controlled batch and chemostat cultivations [12,60,74,75]. Willquist et al. reported that in controlled batch fermentations the initiation of lactate formation coincided with an increment in both the internal NADH/NAD⁺ ratio and the P_H2 of the system [57,75].

An increase of the overall carbon flux through glycolysis results in a higher NADH production rate. To maintain a constant NADH/NAD⁺ ratio a subsequently higher H₂ formation rate is required. However, when the H₂ formation rate (volumetric H₂ production rate) exceeds the H₂ liquid to gas mass transfer rate, H₂ will accumulate in the liquid phase. In some cases this can even lead to super saturation of the liquid, meaning that the dissolved H₂ concentration exceeds the maximal theoretical H₂ solubility [74,76]. Such high dissolved H₂ levels inhibit H₂ formation and presumably cause an increase in the NADH/NAD⁺ ratio. An increased NADH/NAD⁺ ratio has been shown to have an inhibitory effect on GAPDH activity thus limiting the glycolytic flux [75] and consequently leads to a decrease in substrate consumption and growth rate. A switch to lactate formation alleviates the inhibitory effect of an increased NADH/NAD⁺ ratio by causing a decrease of the NADH/NAD⁺ ratio through the reduction of pyruvate.

A clear switch to lactate formation caused by an increase in glycolytic flux can be observed for C. saccharolyticus grown in a chemostat under high P_H2 ([77], Figure 4a). When the glucose load is increased from 20 mM to 40 mM, end-product formation is switched from mainly acetate to mainly lactate, respectively. After the switch to 40 mM glucose some adaptation time is required before a new steady state is achieved. During this adaptation period substantially higher amounts of glucose were detected in the culture effluent, probably indicating an inhibition of the glycolytic flux at the level of GAPDH. When a new steady state was achieved, residual glucose was only slightly higher with respect to the 20 mM condition. As a consequence of the increase in glycolytic flux, the lactate concentration dramatically increased while the acetate concentration hardly changed, indicating that during both substrate loads a similar volumetric H₂ productivity was maintained. These data indicate that under these conditions the organism is not capable of dealing with the increased glycolytic flux by increasing its H₂ productivity, but requires a switch to lactate formation. Although oxaloacetate formation was discernible under a 20 mM glucose load, significantly higher and therefore quantifiable levels of oxaloacetate were detected under the 40 mM glucose load condition. This observation together with the increased flux towards lactate suggests that in the newly achieved steady state, a bottleneck exists at the level of pyruvate.

Ethanol formation can also serve as reductant sink in C. saccharolyticus. For some chemostat cultivation conditions a decrease in H₂ yield is only associated with ethanol formation and not with lactate formation [56,60,75]. These conditions concern low substrate loads, and the observed increase in ethanol formation was triggered by an increase of the dilution rate [56,75] or a change in mode of gas flushing [60], both potentially generating a moderate increase in the H₂ level of the system.

Hydrogen, ethanol and lactate formation are the main routes for reductant disposal in C. saccharolyticus. The gene expression of the hydrogenases and both alcohol dehydrogenases involved in ethanol formation are proposed to be under the control of the NADH/NAD⁺ sensitive transcriptional regulator Rex [60] (Figure 5). Cultivation of C. saccharolyticus under high P_H2 conditions was shown to lead to an upregulation of both the hydrogenases and alcohol dehydrogenases [60]. In addition, the ldh transcript level is also upregulated under high P_H2 conditions, but in silico analysis of the ldh promoter region did not reveal a likely Rex operator binding sequence [60]. Thus, the exact regulatory mechanism triggering ldh transcription remains therefore unclear. Nonetheless, lactate dehydrogenase activity has been shown to be regulated at the enzyme level, with fructose 1,6-bisphosphate and ATP acting as allosteric activators and both NAD⁺ and PP_i as competitive inhibitors [57]. Under high P_H2 cultivation conditions, enzyme activity assays revealed increased LDH activity with respect to low P_H2 conditions [57,75]. A hampered glycolytic flux at the level of GAPDH, potentially triggered by NADH build-up, might lead to the accumulation of fructose 1,6-bisphosphate. In turn, the accumulated level of fructose 1,6-bisphosphate could stimulate lactate formation (Figure 5), explaining the switch to lactate under high P_H2 at the enzyme level.

Interestingly, LDH enzyme activity can also be measured under non-lactate producing conditions, which suggest that additional factors control lactate formation [57,75]. Low levels of lactate formation can be observed at the end of growth during the transition to stationary phase for cultures grown in controlled batch systems [55,57,75,78,79]. This initiation of the lactate formation coincided with a relative increase in ATP levels and a relative decrease in PP_i levels [57,71]. These observed changes in ATP and PP_i levels could release the inhibitory effect of PP_i and stimulate LDH activity [57] (Figure 5). For this latter mode of LDH activity regulation, the lactate formation is proposed to be coupled to the energy metabolism. Accordingly, lactate formation is prevented during exponential growth, which is associated with a high anabolic activity and high PP_i levels. Whereas, during the transition to stationary phase, which is associated with low anabolic activity and relative low PP_i levels, lactate formation inhibition is alleviated [12,57].

Studies on biomass hydrolysates and mono-saccharide mixtures, mimicking the biomass hydrolysates, showed that, while at low substrate loads fermentation performances were comparable, at higher substrate loads H₂ yields were higher on the mixed mono-saccharides compared to the biomass hydrolysates. The difference in yields was caused by a shift to lactate formation during the growth on high substrate load hydrolysates. Interestingly, for the higher substrate concentrations, the total sugar consumption was higher during growth on hydrolysates compared to growth on sugar mixtures [55,91]. For C. saccharolyticus, grown on carrot pulp hydrolysate (20 g/L), a lower cumulative H₂ production was found, compared to growth on a glucose/fructose mixture (20 g/L), while a relatively higher maximal H₂ productivity was observed for the hydrolysate compared to the sugar mixture. During the growth on carrot pulp hydrolysates the relative higher H₂ productivity preceded the switch to lactate formation [55]. These results demonstrate the relation between high H₂ productivity, lactate formation and an overall low H₂ yield. However, these described phenomena were not observed for fermentations on Miscanthus hydrolysates. For each tested substrate load (10, 14 and 28 g/L) fermentation performance during growth on the Miscanthus hydrolysate was similar to the glucose/xylose sugar mix and only moderate levels of lactate were formed even at high substrate loads [54]. The differences in H₂ production characteristics between the discussed hydrolysates might be related to the difference in sugar composition of the hydrolysates or the differences in pretreatment applied prior to hydrolysis.

In general, biomass derived hydrolysates might contain substances which negatively affect growth or fermentation performance. For example, the dilute-acid pretreatment of lignocellulosic biomass releases undesirable inhibiting compounds like 5-hydroxymethylfurural (HMF), furfural, phenolic compounds and acetate [89] and a growth inhibition of 50% was reported for C. saccharolyticus in the presence of 1–2 g/L HMF or furfural [54].

For chemostat cultivation on glucose (4.4 and 5 g/L) it was shown that a dilution rate (D) exceeding 0.1 h⁻¹ gave rise to an incomplete substrate conversion but also to a lower H₂ yield. The concomitant decrease in both biomass level and H₂ yield caused the volumetric H₂ productivity (mmol/L*h) to level off at higher dilution rates [56,75]. The observed lower H₂ yield reflects the shift in end product formation, from mainly acetate at a low dilution rate of 0.05 h⁻¹ to a mix of acetate and ethanol at a D = 0.15 h⁻¹ [75]. Interestingly, no lactate was produced during these chemostat cultivations [56,75]. The observed incomplete substrate conversion indicated that another factor was limiting under those conditions. Indeed, increasing the yeast extract to glucose ratio, from 0.25 to 1 g/g, resulted in the almost complete consumption of glucose and also a doubling of cell density (D = 0.3 h⁻¹), which led to an increase in volumetric H₂ productivity (20 mmol/L*h). The H₂ yield was, however, not affected [56]. This finding indicated that these observed changes in H₂ yields were a function of the growth rate and did not depend on the substrate conversion efficiency.

Incomplete substrate conversions can also be observed in controlled batch fermentations with high initial substrate levels. Growth on 10 g/L of both glucose and fructose resulted in complete substrate consumption, while higher initial substrate levels of glucose (20 and 31 g/L) and fructose (20 g/L) resulted in incomplete conversions [55,91]. Similar observations were done for different sugar mixtures [54,55,94]. These incomplete conversions were attributed to the inhibitory effect of accumulating organic acids like acetate or lactate. Inhibition experiments showed that an acetate concentration of 200 mM and higher prevented acid production by C. saccharolyticus grown on glucose (10 g/L) in batch [91], which is in line with the earlier findings of van Niel et al., who observed critical sodium acetate and potassium acetate concentrations of 192 mM and 206 mM, respectively, for batch growth on sucrose [73]. However, similar inhibitory effects were observed for NaCl and KCl, with critical concentration of 216 mM and 250 mM respectively [73], suggesting that increased osmolarity is the cause of inhibition and not the end products per se. This relative low osmotolerance in comparison to marine organisms, like Thermotoga neapolitana [91], probably reflects the terrestrial origin of C. saccharolyticus. Ljunggren et al. designed a kinetic model for the growth of C. saccharolyticus incorporating the inhibitory effect of a high osmolarity and determined a critical osmolarity (no growth) in the range of 270 to 290 mM. They also showed that osmolarity is of minor influence on fermentations with low initial glucose levels [74]. This low tolerance to osmotic pressure also prevents the application of CO₂ as a cheaper and more convenient stripping gas than N₂. The use of CO₂ as stripping gas during C. saccharolyticus cultivations negatively affects growth rate and hydrogen productivity. CO₂ sparging led to a higher dissolved CO₂ concentration, which required addition of extra base to maintain a constant pH, overall leading to an increase in osmotic pressure [79].

Controlled batch cultivations were used to investigate the influence of NH₄⁺ on the performance of C. saccharolyticus grown on molasses. These experiments revealed that the omission of NH₄⁺ gave rise to a higher H₂ yield and maximal H₂ productivity [90]. Although C. saccharolyticus is able to grow on a medium without NH₄⁺, containing only YE as a nitrogen source, it becomes very sensitive to changes in P_H2 ([77], Figure 4b). Chemostat cultivations showed complete glucose (20.7 mM) consumption and high acetate yields (1.87 ± 0.02 mol/mol) under low P_H2. However, when the cultivation condition changed to a high P_H2 a new steady state could be achieved, but substrate consumption was incomplete (55%) and acetate yields decreased (1.68 ± 0.01 mol/mol).

Omission of yeast extract (YE) during growth of C. saccharolyticus on molasses did not affect the H₂ yield but led to a lower volumetric H₂ productivity [90]. Similar observations were made for controlled batch fermentations on glucose, where the absence of YE did not affect the H₂ and biomass yields [100]. Contrary to the molasses study the volumetric productivity was not affected [100] but this might be due to the lower substrate load used. C. saccharolyticus is able to grow on a defined minimal medium with additional vitamins, but without additional amino acids [100]. Growth in the absence of YE helps to reduce the production costs. However, increased biomass levels and especially growth on high substrate loads might augment medium requirements. So fine-tuning of the medium composition with respect to the specific substrate and substrate load is required.

C. saccharolyticus has many properties that make it an excellent candidate for biohydrogen production via dark fermentation. However, for biohydrogen production to become economically feasible major improvements should be made with respect to the H₂ productivity [104,105,106]. Productivity is maximized by improving the substrate consumption rate but also the H₂ yield.

C. saccharolyticus is able to produce H₂ close to the theoretical maximum of 4 H₂ per hexose. In this respect, it is important to realize that the theoretical yield refers to the pure catabolic component of glucose conversion. The glucose used for anabolism should not be incorporated. Generally, this distinction is not considered in literature, and reported experimental data therefore reveal H₂ yields lower than 4. For example, a yield of 3.5 H₂ per consumed glucose has been reported by de Vrije et al. (chemostat cultivation with a 23.0 mM glucose load and a dilution rate of 0.1 h⁻¹) [56]. According to their data 16% of the consumed glucose is used only for biomass formation indicating that only ~19.4 mM glucose was available for ATP generation. Given the reported H₂ production this results in a theoretical conversion efficiency of 4.15 mol H₂ per mol glucose, which approximates the theoretical maximum of 4 H₂ per hexose as indicated by Thauer et al. [83]. These high yields are only achievable when the organism ferments the substrate solely to acetate. However, non-ideal growth conditions lead to a mixed fermentation profile (acetate, lactate and ethanol), and a consequently lower H₂ yield. From the organism’s perspective switching to lactate or ethanol is profitable since it allows the organism to continue to grow under elevated P_H2 conditions, albeit with a lower growth rate because ATP yields are lower under lactate and ethanol forming conditions. Obviously, the lower H₂ yield under a mixed end-product fermentation, is not desirable from a biotechnological point of view. To maximize the H₂ yield and productivity, the dissolved H₂ concentration should be kept as low as possible, which requires an optimization of the liquid to mass transfer rate [74,76], which is mainly a matter of reactor design. In continuous stirred-tank reactor systems low dissolved H₂ concentrations could be achieved by increasing the sparging rate [107,108] or the stirring speed [109,110]. In addition, reduction of internal reactor pressure [111,112,113] and enforced bubble formation [113,114] could potentially lead to a lower dissolved H₂ concentration. Addition of zeolite particles, which enhances bubble formation, allowed a reduction of the N₂ stripping rate from 5 L/(h*L) to 1 L/(h*L), without affecting the H₂ productivity and H₂ yield of C. saccharolyticus. N₂ stripping could even be completely omitted when an internal reactor pressure of 0.3 bar was used, which sustained a similar fermentation performance compared to cultivation at atmospheric pressure (1 bar) using a N₂ gas stripping rate of 5 L/(h*L) [113].

Most research on C. saccharolyticus has been performed in serum bottles and suspended continuous stirred-tank reactor systems, where only relatively low cell biomass levels can be achieved. Higher cell biomass levels would cause an increase in substrate consumption rates leading to an increase in H₂ productivity when H₂ yields are maintained. To realize higher biomass levels a deeper insight into growth limiting medium compounds should be acquired. Additionally, other reactor types, like a trickle bed reactor or a fluidized bed system might allow cell biomass accumulation. C. saccharolyticus could be cultivated in a non-axenic 400 L trickle bed reactor [102] out-competing other organisms, with H₂ yields around 2.8 mol H₂/mol hexose and a productivity of 22 mmol H₂/L*h. These results showed that C. saccharolyticus can be used in a large scale non-sterile industrial setting.

Combining different organisms in a co-cultivation setup allows the exploitation of the hydrolytic capability of each individual species and could enhance the overall range of useable substrates. Batch co-cultivations of C. saccharolyticus with either C. owensensis, C. kristjanssonii or an enriched compost microflora were performed on a glucose-xylose mixture [115]. The co-cultivation with the enriched compost microflora resulted in a fast, simultaneous consumption of both glucose and xylose with a relatively high specific hydrogen production rate, but with a lower H₂ yield [115]. A stable co-culture consisting of two closely related Caldicellulosiruptor species, C. saccharolyticus and C. kristjanssonii, could be established in a continuous cultivation system [116]. These findings demonstrate the possibility to create co-cultures for H₂ formation and reveal an apparent synergistic effect of the strains, which lead to improved fermentation performances.

Overall H₂ yields from biomass derived substrates can be increased when the dark fermentation is coupled to a second stage like electrohydrogenesis or photofermentation [117,118]. The former system uses a microbial electrolysis cell (MEC), in which electricity is used to convert acetate or other organic acids to hydrogen. In the latter case the main end product of the dark fermentation, acetate, is further converted by an anaerobic non-sulfur purple photosynthetic bacterium, forming a maximum of 4 mol H₂ per mol acetate, giving an overall H₂ yield of 12 mol H₂ per mol glucose. The effluent of C. saccharolyticus has been successfully used as a feed for photofermentative growth and H₂ production [90,119,120]. Alternatively, dark fermentation end-products H₂ and acetate could serve as substrates for hydrogenothropic methanogens in a biogas generating system. The addition of C. saccharolyticus to natural biogas-producing consortia led to an improvement of biogas production and a stable co-cultivation could be maintained for several months [98].

The first steps in the development of a genetic system for Caldicellulosiruptor species have been made. Chung et al. have shown that methylation with an endogenous unique α-class N4-Cytosine methyltransferase is required for transformation of DNA isolated from E. coli into Caldicellulosiruptor bescii [121]. Furthermore, an uracil auxotrophic C. bescii mutant strain was generated by a spontaneous deletion in the pyrBCF locus [121]. This nutritional deficiency was exploited as a selection marker of C. bescii transformants [121]. A similar strategy might be applied to develop a genetic system for the other members of this genus.

To improve the H₂ producing capabilities of C. saccharolyticus or other Caldicellulosiruptor species metabolic engineering strategies could focus on improving the H₂ yields. Additionally it could be aimed at altering intrinsic properties of Caldicellulosiruptor species limiting H₂ productivity like the enhancement of their H₂ tolerance or osmotolerance. For example, for industrial applications increased substrate concentrations are favored since it reduces the fresh water demand, thus reducing the overall costs and the environmental impact of the process. However, for C. saccharolyticus the maximum substrate load is limited by its sensitivity to osmotic stress [12,73,74].

With respect to the mixed acid fermentation, both lactate and ethanol formations lead to a lowered H₂ yield. Because ethanol formation is not the major reductant sink under redox stress and in general is only produced at low levels, knocking out the alcohol dehydrogenase responsible for ethanol formation will probably not significantly alter the fermentation performance of C. saccharolyticus. However, lactate formation can be seen as the main mechanism to alleviate redox stress. Targeting the lactate formation pathway for complete knockout will probably make C. saccharolyticus less resilient to fluctuations in dissolved H₂ concentration and is inadvisable.

Alternatively, one could alter glycolysis in such a way that substrate conversion is less energy efficient. So to generate the same amount of ATP, essential for biosynthesis and maintenance, a higher glycolytic flux to acetate is required. This would result in a higher H₂ yield because the glycolytic flux to acetate is increased with respect to the carbon flux to biomass. A less energy efficient glycolysis can be achieved by eliminating some ATP generating steps from the central metabolic pathway. For example, exchanging the NADH-dependent GAPDH in C. saccharolyticus with the ferredoxin-dependent GAPOR, would decrease the overall ATP yield of glycolysis. In addition, since the H₂ formation from the generated Fd_red is energetically more favourable than H₂ formation from NADH, the organism would become less sensitive to increased H₂ levels [9,12].

H₂ yields on rhamnose and fucose could be increased if the carbon flux from the intermediate lactaldehyde is redirected via methylglyoxal to pyruvate, by the insertion of a lactaldehyde dehydrogenase and a methylglyoxal dehydrogenase. When rhamnose is completely oxidized to acetate, via this pathway, the H₂ yield increases from 1 to 5 H₂ per rhamnose. With respect to alternative substrates, glycerol could serve as a good substrate for H₂ production because of the relative high reduced state of its carbon atoms. A maximum H₂ yield of 3 mol/mol glycerol is achieved if glycerol is completely oxidized to acetate. So far, growth or co-consumption on/of glycerol has, however, not been observed for C. saccharolyticus [122].