Engineering Enzymes for Cofactor Recycling and Carbon Fixation

Enzymes can catalyze reactions with high selectivity under mild conditions, and are therefore especially suited for the upgrading of C1 feedstocks into value-added products. Linear carbon ligation routes are of particular interest due to their simplicity and potential for high carbon efficiencies. A linear carbon fixation pathway can be constructed using a combination of NADH-dependent oxidoreductase enzymes and a formaldehyde carboligation enzyme, through which CO₂ is upgraded into C₂ and C₃ products. The stoichiometric NADH requirement imposed by the oxidoreductase enzymes and the poor performance of two core pathway enzymes (formaldehyde dehydrogenase and formolase) are the main obstacles to the efficient application of this linear carbon fixation pathway. In this dissertation, a host of fundamental enzyme engineering and characterization techniques are applied to study and address the thermodynamic, transport, and kinetic challenges arising from the use of enzyme cascades for multistep catalysis. In Chapter 2, a modular approach for the design of cofactor-independent transhydrogenases was explored and developed to enable catalysis and cofactor recycling in a single protein. Individual, unmodified active sites were modularly assembled and their activity catalytically coupled using biomimetic PEG-NAD(H) swing arms. Protein engineering and molecular design were used to increase the swing arm content and increase the activity of the transhydrogenases without detriment to their selectivity, circumventing the typical tradeoffs associated with modifications to the active site. The modularity of this approach was illustrated through the creation and characterization of four novel transhydrogenase enzymes with behavior that was predictable from that of the parent enzyme active sites. In Chapter 3, the kinetic behavior of the formolase (FLS) enzyme was comprehensively characterized to facilitate its use in carbon fixation cascades. A mechanistic rate equation and theory-based figures of merit were derived from first principles and used to capture and rank the full catalytic performance of 8 FLS variants under different conditions. The transition state specificity constant derived in this chapter was used to quantify product preference. In Chapter 4, the limiting performance of the formaldehyde dehydrogenase enzyme was explored within the context of a NADH-dependent pathway for the reduction of CO₂ to methanol. Protein engineering experiments targeting the elongated cofactor binding loop of the enzyme were investigated as a means of enhancing NADH-dependent formate reductase activity, but all mutations in the loop region dramatically reduced protein expression levels. Pathway flux was increased through the substitution of the formaldehyde dehydrogenase with a better performing homolog. In Chapter 5, a model carboligation pathway for the conversion of formaldehyde to glycerol and ethylene glycol was built using the FLS, glycerol dehydrogenase, and phosphite dehydrogenase enzymes. The impact of the low activity and substrate affinity of the FLS enzyme on pathway carbon and energy efficiencies was examined in purified proteins and crude lysates. High energy efficiencies, as quantified through the efficiency of NADH utilization, were achieved only with purified proteins. Cofactor regeneration was also shown to lower the cofactor requirement from stoichiometric to catalytic concentrations. In this work, we utilized a range of characterization techniques to study the limitations and challenges involved in the use of both NADH-dependent oxidoreductases and designed enzymes for multistep catalysis, and used protein engineering to address them. The insights gained from this work will facilitate the efficient use of these enzymes for linear carbon fixation as well as other biocatalysis applications.