Ioannis Papapetridis

Mankind’s energy requirements, which are currently mainly covered by the combustion of fossil fuels, have been steadily increasing in the past half century. While fossil fuels have a high energy content, their use results in significant emissions of greenhouse gases (mainly CO2, methane and nitrous oxide). As the industrialization of developing nations continues, the requirement for a paradigm shift is becoming increasingly evident. Microbial fermentation can provide an alternative by enabling the sustainable production of transport fuels that combine a lower carbon footprint with compatibility with current internal combustion engine technology. Bioethanol is, by volume, the biofuel with the highest annual production (ca. 100 billion liters in 2016). Current ‘first generation’ industrial bioethanol production processes are mainly based on fermentation of hydrolysed corn starch or sugar-cane sucrose by the budding yeast Saccharomyces cerevisiae and capitalize on the naturally high sugar-uptake rates and ethanol yield of this microorganism. The first full-scale ‘second generation’ ethanol production plants that are now coming on line use lignocellulosic hydrolysates, derived from agricultural ‘’waste’’ such as corn stover or wheat straw, as feedstocks. Second-generation bioethanol production can have a smaller carbon footprint than first-generation processes. Moreover, it uses feedstocks that are not a part of the human food chain. However, yeast-based second-generation bioethanol production poses multiple challenges for scientists. Lignocellulosic hydrolysates contain significant amounts of pentose sugars (mainly D-xylose and L-arabinose) which are not naturally fermentable by S. cerevisiae. Further, during biomass pretreatment, inhibitors of yeast performance (phenolics, aldehydes and organic acids) are released into the hydrolysates. To mitigate the negative effects of these inhibitors, yeast strains used in second-generation bioethanol production processes need to maintain high rates of sugar fermentation, both for hexoses and for pentoses. In both first- and second-generation bioethanol production, the price of the hydrolysed feedstock represents the single largest factor in production costs. Therefore, in an industry that generally operates at low profit margins, maximization of the ethanol yield on fermentable sugars is of paramount importance. Chapter 1 of this thesis discusses past research and recent advances in strain engineering for improved ethanol production in both first- and second-generation processes.

During industrial bioethanol production, carbon losses occur due to the formation of biomass, CO2 and fermentation by-products, with glycerol formation accounting for up to 4% of consumed sugars. Glycerol plays multiple roles in yeast metabolism. It forms the backbone of glycerolipids, is a stress protectant (mainly against osmotic stress) and, in anaerobic yeast cultures, its production plays a key role in redox metabolism. Formation of glycerol from dihydroxyacetone-phosphate in the reactions catalysed by glycerol-3-phosphate dehydrogenase and glycerol-3-phosphate phosphatase requires input of NADH. The coenzyme pairs NAD(P)+/NAD(P)H play a vital role in mediating >200 cellular redox reactions in S. cerevisiae. NAD(P)+ and NAD(P)H are conserved moieties; when reduction of NAD(P)+ in oxidation reactions is not matched by oxidation of NAD(P)H in reductive reactions, growth rapidly ceases. Anaerobic cultures of S. cerevisiae require glycerol formation to reoxidize excess NADH formed in biosynthetic reactions, as glycolysis and alcoholic fermentation already form a redox-neutral pathway. Elimination of glycerol formation by deletion of GPDN and GPD2, the two genes encoding glycerol-3-phosphate dehydrogenase in S. cerevisiae, results in abolishment of anaerobic growth unless an external electron acceptor, such as acetoin that can be reduced to 2,3-butanediol, is provided. However, in industrial processes, such additions would increase operational costs. Recently, functional expression of ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO, Thiobacillus denitrificans CbbM) and phosphoribulokinase (Spinacea oleracea PRK) in S. cerevisiae was shown to enable the use of CO2 as an alternative electron acceptor. As CO2 production is stoichiometrically linked to alcoholic fermentation, its external supply to growing cultures is not required. The modified strain displayed a 90% decrease in glycerol yield in anaerobic glucose/galactose-grown chemostat cultures and a 60% decrease in glycerol yield in anaerobic galactose-grown batch cultures, with concomitant increases in the ethanol yield. The use of galactose as an inducer of PRK expression in the proof-of-principle strains, along with the slow anaerobic growth rate and fermentation kinetics remained points of optimization before industrial implementation could be considered. In Chapter 2 a metabolic engineering strategy for optimization of the RuBisCO/PRK pathway in S. cerevisiae is presented and experimentally tested. CRISPR-Cas9 genome-editing technology, with which complicated genetic modifications can be introduced in one or a few steps, was used to modify carbon and redox metabolism in a RuBisCO/PRK-expressing strain. The resulting strain grew at wild-type rate in anaerobic batch cultures on glucose, while displaying a ca. 90% lower glycerol yield and a 15% higher ethanol yield than a non-engineered S. cerevisiae strain. As strains engineered in this way do not require specific media compositions or process modifications, this concept may be implemented in industrial yeast strains and used to increase the ethanol yield in bioethanol production processes.

Acetic acid can also be used as an alternative electron acceptor for reoxidizing NADH in anaerobic yeast cultures. Lignocellulosic hydrolysates invariably contain acetic acid, which is released during deconstruction of the hemicellulose fraction during biomass pre-treatment. Lower concentrations of acetic acid have been reported in first-generation feedstocks. Acetic acid is a potent inhibitor of yeast fermentation, as it causes weak-organic acid uncoupling and abolishment of growth at higher concentrations. Expression of a heterologous acetylating-acetaldehyde dehydrogenase (Escherichia coli MhpF) in S. cerevisiae was previously shown to enable use of acetic acid as an external electron acceptor, by completing an acetate-to-ethanol reduction pathway in combination with the native yeast acetyl-CoA synthetases and alcohol dehydrogenases. This approach not only resulted in higher ethanol yield by replacing glycerol formation with acetate reduction, but also enabled partial in situ detoxification of the medium by the engineered strain. As the reduction of acetic acid requires cytosolic NADH, the amount of additional ethanol that can be produced and the extent of the medium detoxification by the engineered strains is limited by the amount of NADH formed in yeast anabolism.

Chapter 3 explores a redox engineering strategy for increasing cytosolic NADH generation in S. cerevisiae. Replacement, in an acetate-reducing strain, of the native, NADP+-dependent, 6-phosphogluconate dehydrogenases by a heterologous NAD+-dependent enzyme (Methylobacillus flagellatus GndA), in combination with deletion of Ald6, resulted in 44 and 3% increases in acetate consumption and ethanol yield, respectively. Replacement of MhpF by the alternative acetylating acetaldehyde dehydrogenase EutE (Escherichia coli) significantly improved the specific growth rates of the engineered strains.

The acetate-reducing S. cerevisiae strains discussed in Chapter 3 harboured deletions in GPDN and GPD2, resulting in the absence of a functional glycerol formation pathway. The inability to produce glycerol could decrease the stress tolerance of engineered strains in industrial media, especially against high osmotic pressure. Metabolic engineering strategies to enable acetate reduction in strains still capable of glycerol formation are discussed in Chapter 4. Only expressing EutE was found to be insufficient to enable optimal acetate reduction in the presence of a fully functional glycerol formation pathway. Deletion, in an EutE-expressing strain, of GPD2, which is upregulated under anaerobic conditions, resulted in a fourfold lower glycerol production and concomitant increases in acetate consumption and ethanol yield in low-osmolarity media. In high-osmolarity media (1 mol L-1 glucose), acetate reduction and anaerobic growth was enabled by replacement of GPDN and GPD2 by the archaeal, NADP+-dependent, glycerol-3-phosphate dehydrogenase GpsA (Archaeoglobus fulgidus). Expression, in an EutE-expressing strain, of GpsA, in combination with deletion of Ald6, enabled immediate growth under high-osmolarity conditions. Moreover, the GpsA-expressing strain exhibited equivalent acetate reduction to a Gpd- strain without the associated osmosensitivity, and a 13% higher ethanol yield than observed in a non-acetate reducing S. cerevisiae. The metabolic engineering strategies discussed in Chapters 3 and 4 should facilitate the transfer of the acetate-reducing pathway to industrial strains and media for testing in process conditions and lignocellulosic hydrolysates.

In addition to the economic importance of increased ethanol yields, second-generation bioethanol production can benefit from decreased fermentation times that can increase overall productivity and process robustness against contaminations. S. cerevisiae strains that express functional pentose utilization pathways preferentially consume glucose when media also contain pentoses, resulting in sequential utilization of sugar mixtures and increased overall fermentation times in second-generation bioethanol production processes. Chapter 5 discusses a combinatorial metabolic and laboratory evolution strategy that was designed and successfully applied towards the identification of genetic mutations that facilitate increased xylose utilization by S. cerevisiae in the presence of glucose. Deletion of PGIN and RPEN, encoding phosphoglucose isomerase and ribulose-5-phosphate epimerase respectively, in xylose-consuming S. cerevisiae with a modified pentose-phosphate pathway, forced co-consumption of glucose and xylose in batch cultures. Laboratory evolution in media with increasing glucose concentrations, followed by whole-genome sequencing, identified mutations in HXK2, RSPS and GAL8P in evolved strains. Combined introduction of the HXK2 and GAL8P mutations in a non-evolved xylose-consuming strain resulted in a 2.5-fold higher xylose consumption rate in the presence of glucose in anaerobic batch cultures on glucose-xylose mixtures. This led to a shorter xylose consumption phase and an overall reduction of the length of anaerobic fermentation experiments of over 24 h. The combinatorial metabolic and evolutionary engineering strategy developed in Chapter 5 should be applicable to similarly identify relevant beneficial mutations in different yeast strain backgrounds and/or process conditions.