High yield, final concentration and productivity of the desired product are the major objectives for optimizing microorganisms used in an industrial scale bioprocess. Beyond that, it has to be ensured that the microorganism can still cope with process constraints, which might expose the microorganism to severe stress. Generally speaking, finding an acceptable trade-off between these opposing requirements is a major challenge for successful strain engineering. In particular modifications of the central carbon metabolism are inherently coupled to energy and redox issues , which might cause severe side effects on the cell’s robustness towards environmental stress. One prominent example for such a challenge is the reduction of glycerol formation in Saccharomyces cerevisiae in order to increase the ethanol yield. Glycerol is one of the main by-products in ethanol fermentation and may account for up to 5% of the substrate carbon . Therefore, the abolishment or at least a substantial reduction may lead to a significant increase in ethanol yield. This issue has been on the scope for a long time and has been addressed by both, process optimisations  and genetic engineering [1, 4]. Despite of their success in reducing glycerol formation, both approaches often resulted in severe side effects on growth and performance. These studies also substantiated the importance of glycerol as a major player in the cell metabolism [5, 6], as a central element of the cell redox balance , as essential precursor for phospholipids and triacylglycerolipids , and lastly as an essential constituent of the cell stress resistance system .
Concerning the role of glycerol in the redox balance, glycerol is involved in the transfer of the reducing power from the cytosol to the mitochondria in aerobic condition, but more importantly it is mainly used as a sink for electrons under anaerobic conditions . Indeed the coupling of glycolysis and ethanol production presents a null oxydo-reductive balance , however the synthesis of organic acids as well as some anabolic reactions produce an excess of NADH [10, 11]. In anaerobiosis, the glycolytic intermediate dihydroxyacetone phosphate (DHAP) is reduced to glycerol-3-phosphate (G3P) at the expense of one NADH  and subsequently, G3P is dephosphorylated into glycerol as the final metabolite [5, 12].
Furthermore, glycerol is also known for its broad implication in stress resistance, particularly in osmotic stress. Glycerol is the main compatible solute accumulated in S. cerevisiae. Intracellular accumulation of glycerol contributes to maintain turgor pressure and prevents the loss of water under hyperosmotic conditions. Intracellular glycerol concentrations are regulated by the High Osmolarity Glycerol (HOG) MAP kinase pathway , which enhances glycerol formation under hyperosmotic stress, and by the plasma membrane channel Fps1, which regulates the efflux rate of glycerol during a hypo-osmotic shock . Apart from osmotic stress, a potential role of glycerol in resistance to a wide range of stress types such as temperature, thawing, oxidative stress as well as stress by high ethanol concentration has been suggested in literature . These broad implications in central cellular functions make it difficult to engineer mutant strains, showing not only the desired reduction of glycerol but also stress robustness.
Approaches aiming at redirecting the main glycerol pathway, mostly targeted the genes encoding for the enzymes directly involved in glycerol formation, namely the glycerol-3-phosphate dehydrogenase (GPDH) and the glycerol-3-phosphatase (GPP). Both enzymes exist in two iso-forms encoded by their corresponding isogenes, which show highly similar sequences. However the physiological role within the cell differs quite considerably among the isoforms [15, 16]. The GPDH isoform, Gpd1 is involved in the response to osmotic stress  and its activity increases in condition of hyperosmotic stress. Strains with deleted GPD1 are osmo-sensitive . Gpd2 is involved in the response to anaerobiosis; strains with deleted GPD2 showed an altered growth under anaerobic conditions and its activity was found increased in absence of oxygen . Mutants being deleted in either one or both of the two isogenes, GPD1 and GPD2, were constructed in different backgrounds [4, 12, 15, 18–20]. Under anaerobic conditions, the gpd1Δ and gpd2Δ mutant showed an increase in ethanol yield of 2.8% and 4.7% respectively, while the gpd1Δ gpd2Δ mutant strain was not able to grow. Under aerobic conditions, gpd1Δ and gpd2Δ increased their ethanol yield by 2.2% and 3.3%, respectively. The gpd1Δ gpd2Δ mutant increased the ethanol yield by 12.7%, however not only due to abolished glycerol formation but also due to reduction in the biomass yield by 28.8% . Moreover, the assessment of the ethanol production capacity of the gpd1Δ gpd2Δ mutant in an aerobic high ethanol production showed that its tolerance to ethanol was reduced . Gpp1 and Gpp2 are involved in osmotic stress. However, strains with a deletion of GPP1 are not able to grow under anaerobiosis. A recent study targeted both GPP iso-enzymes, Gpp1 and Gpp2. This study showed that, in aerobic conditions, deletion of one gene did not affect growth or glycerol production, while deletion of both genes only lead to a 50% decrease in the glycerol formation, suggesting for unspecific glycerol dephosphorylation, or activation and reversion of the catabolic glycerol pathway [21, 22].
Alternative strategies to reduce glycerol investigated an altered cofactor use to decrease the need for NADH re-oxidation in the cell, by engineering the redox metabolism. This was either done by i) decreasing the NADH produced or by ii) introducing new reaction consuming NADH. In the first case, an attempt to modify the redox metabolism was made by by-passing the NAD+-dependent glycolytic conversion of glyceraldehyde to glycerate through the heterologous expression of a NADP+-dependent glyceraldehyde-3-phosphate dehydrogenase. This strategy replaced a NADH producing reaction by a NADPH producing reaction and resulted in a reduction in glycerol yield of 40% and an increase in the ethanol yield by 3%. The biomass yield was not constant throughout the tested strains . One example for new reactions, which consumed NADH and replaced glycerol as redox sink, was carried out by Nissen et al.. In this study, the ammonium assimilation was modified by deleting the gene GDH1 encoding the NADP+-dependent glutamate dehydrogenase and over-expressing the genes for the NAD+-dependent ammonium assimilation pathway GLN1/GLT1. This allowed decreasing the need for NADH re-oxidation via glycerol formation and resulted in a reduction in glycerol yield by 38% and an increase of ethanol yield by 10% . In a recent approach, a new pathway for NADH reoxidation was introduced by overexpression of the Escherichia coli gene mhpF, encoding the acetylating NAD-dependent acetaldehyde dehydrogenase, in a gpd1Δ gpd2Δ mutant. The reduction of acetate to acetaldehyde in S. cerevisiae consumed one NADH instead of the usual NADPH. This reaction provides an alternative redox sink to reoxidize excess NADH. Therefore, it was possible to partly restore growth of the gpd1Δ gpd2Δ mutant under anaerobic conditions. In this mutant, NADH was completely re-oxidized by the reduction of acetic acid to ethanol via the new NADH-dependent reaction. The co-fermentation of acetic acid together with glucose represents an interesting strain property in ethanol production from lignocellulosic hydrolysates, which contains a significant concentration of acetic acid .
Other recent studies combined the modification in the glycerol synthesis pathway, redox metabolism engineering, the modification of yeast glycerol transport systems and the overexpression of trehalose synthesis genes [25–30]. The best results were obtained by deleting the GPD1gene, over-expressing the trehalose synthesis genes TPS1 and TPS2 and expressing the Bacillus cereus glyceraldehyde-3-phosphate dehydrogenase GAPN, the strain showing a 75% reduction of glycerol yield concomitant to a 8% ethanol yield increase . Though, those studies were obtained on rich medium (YPD) or without complete product analysis (CO2 for example), which did not allow a close monitoring of carbon fate during the fermentation leaving gaps in the understanding of the metabolism in those strains.
Advances in yeast promoters engineering have recently allowed to finely grade gene expression allowing to circumvent a complete gene deletion, which might cause severe side effects . In order to study S. cerevisiae strains which have a glycerol formation capacity ranging between that of the gpd2Δ single mutant (100%) and the gpd1Δ gpd2Δ double mutant (0%), we recently replaced the native GPD1 promoter in a gpd2Δ background by two well-characterized TEF1 promoter mutant versions [31, 32]. The genetic modifications were accompanied by 61% and 88% reduction in glycerol yield on glucose and by 20 and 30% reduction in maximal aerobic growth rate compared to the wild type. Interestingly, the engineered (“intermediate”) strains referred to as TEFmut2 and TEFmut7 showed a 2 and 5% increase in ethanol yield and could well cope with process stress, which is in remarkable contrast to a gpd1Δ gpd2Δ mutant. These results were obtained in a Very High Ethanol Performance (VHEP) fed-batch process with aeration . Flux calculation based on a metabolic model [3, 32, 33] showed that the carbon flux through the glycerol pathway was sufficient to provide enough G3P as biomass precursor and to sustain the maximal growth yield, observed in the wild type strain. Under fully aerated conditions, we did not observe a negative impact of low glycerol production upon the industrial relevant traits of the production strain. Results showed that, in such conditions, it was possible to widely decrease the glycerol yield, increase the ethanol yield and limit the negative impact of the deletion in regards to biomass, viability and tolerance to ethanol . Recently, we constructed a collection of different strains with different combinations of residual GPD1 and GPD2 expression levels controlled by the TEFmut2 and TEFmut7 engineered promoters . Among our engineered strains we identified four strains showing improved ethanol yields compared to the wild type. In contrast to the gpd1Δ gpd2Δ mutant, these strains were able to completely ferment the sugars under quasi-anaerobic conditions in both minimal medium and during Simultaneous Saccharification and Fermentation (SSF) of liquefied wheat mash (wheat liquefact) . In the current study, the two strains, TEFmut7 and TEFmut2, were grown in a VHEP fed-batch process under high productivity anaerobic ethanol fermentation. The quantitative kinetic analysis was applied to evaluate the impact of reduced glycerol formation on the overall yeast metabolism and the cell viability.