With the expanding global population and the associated growing demand for food and energy, researchers have been looking into ways to improve plants productivity, and more particularly the way they convert carbon dioxide to biomass during photosynthesis. Targeting carbon capture efficiency seems a good opportunity to raise crop productivity both for food and biomass-based biofuels. Most approaches have focused on optimizing the CBB cycle and more particularly the key carboxylating enzyme, Rubisco, because of its rather low catalytic rate and poor substrate specificity. Here we propose an alternative approach by engineering an orthogonal artificial carbon fixation pathway into photosynthetic organism. We choose cyanobacteria, a model system for phototrophic eukaryotes, while also implementing the new pathway into the model plant organism, Arabidopsis thaliana. The new CO2 fixation cycle would not rely on Rubisco and would fix carbon with only 30% of the energy requirement compared to the native CBB cycle. Over the course of this study we encountered a number of logistical problems that thwarted the implementation of the orthogonal carbon fixation cycle. First, the Rubisco mutant strain that could have served as a platform for efficient screening of our pathway was too sick to be used as such. Secondly, our pathway relied on a pyruvate ferredoxin:oxidoreductase for one of its carboxylation step, which is known to be oxygen sensitive. Efforts to evolve an oxygen tolerant mutant of this enzyme were unsuccessful. Lastly, a possible kinetic trap that could prevent the cycle from running was identified. The accumulation of setbacks forced us to modify our initial pathway cycle design. The final pathway, termed rGS2, does not use pyruvate ferredoxin:oxidoreductase for carbon fixation and avoids the kinetic trap. It does lose any energy efficiency gain over the CBB cycle but could nevertheless still increase carbon fixation efficiency as it does not solely depend on Rubisco. Implementation of the rGS2 in cyanobacteria increased the intracellular acetyl-CoA pool and led to secretion of α- ketoisocaproate. The engineered strain was genetically unstable, which was probably the result of toxic intermediates being built up from the rGS2 pathway. In the future, we hope to optimize the rGS2 pathway by testing new homologues enzymes or alternative steps to improve the rGS2 related toxicity.

In plants, the original synthetic carbon fixation pathway, the rGS cycle, was engineered in both wild type Arabidopsis and in two mutant strains defective in their endogenous CBB cycle. In the three different backgrounds, mutants with enhance biomass and/or height were identified, suggesting that the rGS pathway may be functional and could be working as an optional pathway for CO2 fixation. But like in cyanobacteria, genetic instability was observed, and genotyping of the T3 mutant plants could not confirm the presence of all heterologous rGS genes, possibly the results of DNA recombination between homologous regions on the T-DNA causing some of the fragment to loop out.

Numerous obstacles were encountered throughout the course of this work, but the results and lessons learned could later be applied to build more robust orthogonal carbon fixation cycles, and improve production of downstream pathways.

Simulation of metabolic systems with kinetic models requires a large number of parameter values, which are either difficult or impossible to obtain experimentally. Network information, such as stoichiometry, reversibility and steady state flux, can be used to determine mechanistically realistic rate laws, and these can be used to constrain the parameter space to only those values which satisfy the constraints. Until now, stability has been overlooked when considering kinetic metabolic models. However, dynamical stability and robustness to perturbation are important qualities for living organisms, since they may encounter changing environments or stochastic variation across time or within populations. Considering stability can both provide constraints on the parameter space and be used to interpret the response of the model to queries about the performance of the metabolic system under various perturbations. I have used stability analysis to predict the performance of many metabolic systems, with an emphasis on providing guidance for experimental efforts and uncovering biological significance.

The uses of stability analysis have encompassed several projects. Optimization of a novel methanol condensation cycle (MCC) was accomplished by tuning the amount of an irreversible phosphoketolase enzyme to a local productivity and stability maximum, as predicted by stability analysis and confirmed by in vitro experimentation. Several other in vitro enzymatic were subjected to stability analysis, and predictions matched previously published experimental results.

Stability analysis was also applied to several microbial systems to maximize production of a desired compound: n-butanol in Escherichia coli, isobutanol in Clostridium thermocellum and lipids in Yarrowia lipolytica. In these systems, production simulations matched the observations and predictions for further production improvements were made.

Stability analysis was also applied to gain biological understanding of the significance of structural features of the Calvin-Bassham-Benson (CBB) pathway in plants. The phosphate/glyceraldehyde-3-phosphate translocator was identified as more important for stability than a proposed glucose-6-phosphate shunt. Further, productivity was increased after overexpression of sedoheptulose-1,7-biosphosphatase, but not RuBiSCO, in agreement with previous experimental reports.

The importance of stability in analysis of metabolic systems is affirmed by this work, and the techniques demonstrated here pave the way for even further explorations.

Global climate change, caused by the emission of greenhouse gas (GHG) to the atmosphere, is one of the key challenges facing humanity in the coming decades. Carbon dioxide (CO2) is one of the leading sources of overall GHG emission, and it is released to the environment in large quantities by the burning of petroleum or other fossil fuels as liquid fuel. Sustainable alternatives to fossil fuels can be produced from biomass using microbial whole-cell catalysis. Since biomass fixes atmospheric CO2, burning these biomass-derived fuels as energy sources reduces the net emission of carbon to the environment. However, current biofuel processes often struggle to compete economically with fossil fuels. A key contributing factor to this problem is the inefficient utilization of feedstock carbon in biomass. Since microbes use conserved metabolic pathways to catabolize carbohydrate feedstocks, inherent limitations in these pathways often cannot be avoided. Moreover, these pathways evolved in nature to prioritize growth rather than production considerations. A notable example of this is the carbon loss when forming two-carbon (C2) metabolites, such as acetyl-CoA. C2 metabolites are precursors to many industrial products, including important biofuels such as ethanol and butanol. Essentially all organisms degrade sugar to pyruvate, a three-carbon (C3) metabolite using the conserved glycolytic pathways. To produce C2, pyruvate is decarboxylated to produce one C2 equivalent and one CO2, thus wasting a third of input carbon and effectively limiting the maximum carbon yield of C2-derived products to 67%. This carbon loss imposes a significant economic constraint on biorefining. While no natural pathway can bypass pyruvate formation, a synthetic non-oxidative glycolysis (NOG) was recently developed that directly degrades sugar phosphates into stoichiometric amounts of C2 equivalents. Here, we first constructed a strain of E. coli, a model microbial organism, which uses NOG for sugar catabolism rather than glycolysis. This was achieved by a combination of rational design and evolution. By coupling the pathway to growth, we were able to use evolution to develop a robust pathway and fine tune the relative activity of pathway enzymes. The resulting strain grew at a comparable rate to wild-type E. coli and could produce acetate, a C2-derived product, at yield exceeding the theoretical maximum using glycolysis. Since it uses NOG for sugar catabolism, this strain has fundamentally rewired metabolism, as it generates C2 metabolites before C3 metabolites. C3 metabolites are generated from C2 using carbon scavenging pathways. Thus, we believe this strain has considerable potential as a host for the production of C2-derived products. Following evolution of the NOG strain, the chromosome was sequenced to elucidate mutations acquired through evolution. The effect of several of these mutations was also investigated here. We found that a transposon insertion inactivating PtsG, a component of the glucose-specific phosphotransferase system, was essential to the NOG strain’s ability to grow in minimal media. When ptsG is deleted, the catabolic activator protein (CAP) is able to upregulate many genes, which is likely necessary for the NOG-dependent growth phenotype. Another transposon insertion in alternative sigma factor rpoS contributed to growth in the NOG strain, likely through the upregulation of enzymes that benefit NOG growth, such as the TCA cycle. We also evaluated mutations where the cell had fine-tuned the activity of pathway enzymes. A mutation knocking down activity in key NOG enzyme phosphoketolase (F/Xpk) was found to be beneficial for growth by potentially alleviating a kinetic trap in the NOG cycle. However, fixing a promotor truncation that had knocked down activity of Pck, a key gluconeogenic enzyme, actually improved growth rate in glucose media by about 15%. Since the mutation occurred before NOG growth was developed, it was reasoned that changing intracellular conditions after the adaptation to NOG-based growth caused an increase in Pck activity to be beneficial. We also explored using the NOG strain for the production of ethanol, a C2-derived liquid fuel. Since ethanol is more reduced than acetate, it cannot be generated without the supply of external reducing power using NOG. Using formate as the electron donor, we were able to improve ethanol yield from 0.26 to 0.82 through the addition of formate in anaerobic production. Further improvements in ethanol yield were limited by the cells ability to produce ATP to maintain sugar phosphorylation. Since the cell is dependent on acetate production to make ATP anaerobically, ethanol production will need to be integrated with respiration, which can convert excess reducing power into ATP using oxygen as an electron acceptor. Finally, we offer strategies for establishing robust, high yield ethanol production using the NOG strain in microaerobic conditions.

The increasing atmospheric carbon dioxide (CO2) level is calling for more efficient CO2 fixation systems to re-balance the carbon cycle. At the same time, atmospheric CO2 can in principle be a cheap and abundant carbon source for synthesizing various organic compounds. Converting CO2 into useful chemicals can be an ideal solution for a sustainable future. However, due to its high thermodynamic stability and kinetic inertness, the chemical transformation of CO2 into organic compounds usually requires quite harsh conditions. In contrast, nature has evolved several CO2 fixation pathways for synthesizing longer carbon chain compounds from CO2 under ambient conditions. To date, six CO2 fixation pathways have been reported. However, they all contain key carboxylases with intrinsic properties such as structural complexity and oxygen sensitivity that hinder their further engineering and applications. To circumvent these problems, we designed a synthetic CO2 fixation cycle composed of all efficient, simple and oxygen tolerant enzymes and demonstrated its function in vitro.

In this work, we utilized the two most active and oxygen-insensitive carboxylases, crotonyl-CoA carboxylase/reductase (Ccr) and phosphoenolpyruvate carboxylase (Ppc) to fix CO2 and bicarbonate, respectively. Our designed cycle is composed of the reductive glyoxylate synthesis (rGS) pathway and the reductive pyruvate synthesis (rPS) pathway. The rGS pathway converts one pyruvate to one glyoxylate and one acetyl-CoA fixing one bicarbonate, while the rPS pathway converts one acetyl-CoA to pyruvate with one CO2 fixed. Together, we obtained a CO2 fixation cycle termed the reductive glyoxylate/pyruvate synthesis (rGPS) cycle.

We first validated the feasibility of rGPS with its stepwise reconstitution in vitro. Using a two-pot immobilized enzyme system, we were able to run the rGPS cycle semi-continuously. To further improve its efficiency, we developed electrochemical and enzymatic FAD regeneration systems to run the rGPS cycle continuously in one pot. Using the enzymatic FAD regeneration method increased the CO2 fixation rate of rGPS to 66.9 nmol min-1 mg-1 of core cycle proteins.

We further explored the construction of rGPS in vivo. The rGS part has been demonstrated in vivo. The rPS part has not worked in E. coli for growth rescue, although all ten enzymes of rPS were confirmed to be active in crude extract assays. Finally, strategies for constructing a fully functional rGPS in vivo have been proposed.

Man-made photovoltaic device is relatively efficient in converting sunlight to electricity, but the electrical energy generated is difficult to store. Current methods, such as chemical batteries, hydraulic pumping, and water splitting, suffer from low energy density or incompatibility with current transportation infrastructure. The biological systems, on the other hand, can store sunlight in high energy-density carbon-carbon bonds. However, the photosynthesis has low solar energy harvesting efficiency, for which no near-term improvements are in sight. One way to solve both problems is to combine man-made solar cells to biological Carbon dioxide (CO2) fixation and fuel production. Therefore, a microbial process to produce biofuels from CO2 and electricity is needed. In this work, we first used metabolic engineering methods to achieve the production of isobutanol, n-butanol, and 3-methyl-1-butanol (3MB) in a lithoautotrophic bacterium Ralstonia eutropha H16, which can utilize formate as the sole carbon and energy source. We then demonstrated electrochemical production of formate from CO2 and finally integrated the electrochemical reactions with the microbial fuel production. Besides achieving the goals in engineering, this work explored important design principles of metabolic pathway and developed new tools in synthetic biology.

To produce higher alcoholos, two different carbon chain building routes were constructed. First, Coenzyme A (CoA)-dependent pathway was used to synthesize isobutanol and n-butanol in R. eutropha H16. We demonstrated for the first time the production of isobutanol, a branched-chain alcohol, using the CoA-dependent pathway in recombinant R. eutropha H16. The isobutanol production pathway deviates from the CoA-dependent n-butanol production pathway in that it contains an extra carbon chain rearrangement step catalyzed by the isobutyryl-CoA mutase in R. eutropha, which has not been characterized previously. Metabolic engineering methods such as heterologous gene expression, codon optimization, and promoter strength altering were applied to first achieve the production of ~200mg/L n-butanol from fructose or ~30mg/L from formate by engineered R. eutropha. The isobutyryl-CoA mutase was then added to the pathway to achieve the production of ~30mg/L isobutanol from fructose. The carbon skeleton rearrangement chemistry explored here might be used to expand the repertoire of the chemicals accessible with the CoA-dependent pathway.

Next, keto acid-dependent pathway was used to synthesize isobutanol and 3MB in R. eutropha H16. We integrated the set of genes for isobutanol and 3MB production into R. eutropha H16 genome, namely alsS from Bacillus subtilis, and ilvC and ilvD from Escherichia coli. The genes kivd from Lactococcus lactis and yqhD from E.coli were then introduced using a plasmid. R. entropha uses poly[R-(-)-3-hydroxybutyrate] (PHB) as a storage compound and as the metabolic sink for carbon and reducing equivalents. We disrupted the PHB synthesis and used the synthetic isobutanol and 3MB production pathway as the new metabolic sink. In a pH-coupled formic acid feeding fermentor, the engineered strain LH74D produced fuels with the final titer of over 1.4 g/l (~846mg/l isobutanol and ~570mg/l 3MB) and peak productivity of 25 mg/l/h.

Most natural metabolic pathways are regulated on transcriptional level or on protein level by allosteric effectors. In this work, non-native regulatory mechanisms were introduced to the synthetic pathways. In particular, on transcriptional level, a synthetic anhydrotetracycline (aTc)-controllable gene expression system in R. eutropha H16 was developed and applied in regulating the biofuel production gene. The system is composed of a controllable promoter containing the operator tetO, the repressor tetR, and the inducer aTc. The active hybrids between the tetO operators and the native PrrsC were first identified and shown to be repressable by tetR. Next, two mutants of the native PphaC1 promoter were obtained from a high-throughput screening of 300 candidates to tune the tetR expression. The optimized system, which contains the PrrsC-O1-O1 hybrid promoter and the PphaC1-G3::tetR cassette, has decreased leaky expression level and can be regulated gradually by different aTc concentration with a ~11 fold dynamic range. The system was used to alleviate cellular toxicity caused by AlsS overexpression, which impeded our metabolic engineering work on isobutanol production in R. eutropha H16. The system reported in this study may be a useful tool for future research and engineering work in this organism.

Finally, we combined the electrochemical formate production with the microbial fuel synthesis by the engineered host in an integrated process. The challenge was that the growth of the microbial host was inhibited by the electrochemical reaction in a transient manner, suggesting that unstable compounds such as reactive oxygen or nitrogen species might be responsible for the growth inhibition. Reporter assays in R. eutropha H16 showed that sodC and norA promoters, which drives the genes for O2− and NO defense pathway, respectively, were induced when electricity was on. To circumvent this toxicity problem, a porous ceramic cup was used to shield the anode. This inexpensive shield provides a tortuous diffusion path for chemicals. Therefore, the reactive compounds produced by the anode may be quenched before reaching the cells growing outside the cup. Using this approach, healthy growth of Ralstonia strains and production of over 140mg/l biofuels were achieved with the electricity and CO2 as the sole source of energy and carbon, respectively.

Synthetic Biology and Metabolic engineering research aim to elucidate the underlying principle of biological systems and enable a predicable cellular behavior through artificial design. In this work, the discipline of synthetic biology was applied to photosynthetic organism to improve carbon intake efficiency.

After survey and analysis of existing carbon fixation metabolic pathways, the design of a synthetic Reverse Glyoxylate Shunt (rGS) core module was invented. This core module could potentially pair with multiple metabolic routes to generate various synthetic carbon incorporation/fixation pathways. The design of rGS-Glycerate pathway, which could theoretically double the Acetyl-CoA yields from C3 metabolites, was reduced to practice at first due to its technical feasibility simulated through computation. The performance of rGS-Glycerate pathway reached its theoretical value in vitro. The pathway was further proved to be functional in E.coli through growth auxotroph test. Finally, rGS-Glycerate pathway was implemented into a model photosynthetic microorganism, Synechococcus elongatus PCC7942, after iterative “Design-Build-Test” engineering cycles. The higher Acetyl-CoA yield was characterized in terms of higher production of ketoisocaproate (KIC) that required Acetyl-CoA as its precursor in cyanobacteria. Further metabolomics analysis found out that one of the most abundant fatty acids, Oleic acids, reached to 2.5 folds of its content in WT. Another rGS core module contained pathway, rGS-Serine pathway, was also constructed in Synechococcus elongatus PCC7942. Similar level of KIC production was detected in those rGS-Serine strains as well.

In sum, this work is the initial demonstration that the rGS core module contained pathways were compatible with photosynthetic machinery and could potentially lead to enhanced carbon incorporation in photosynthetic organisms.

Harnessing the adaptive nature of cell metabolism presents an opportunity to understand the function of biological systems, how they adapt, and how they may respond when challenged. It also stands as a tool that can aid in restoring impaired metabolic function caused by engineering microbial production. Many studies have demonstrated the ability of the cell to overcome metabolite auxotrophies and have elucidated underlying mechanisms. However, these studies have primarily focused on mechanisms that directly replace mutant function. In this work, we first aim to expand this view by evolving and elucidating more complex adaptive mechanisms. As an example, we used a ΔpanD strain of E. coli, a β-alanine auxotroph, to demonstrate that entire metabolic pathways can evolve to repair auxotrophy. Using directed strain evolution, we showed that E. coli successively evolved three distinct metabolic pathways to synthesize β-alanine. The first involved significant rewiring and repurposing of the uracil synthesis and degradation pathways. The second relied on a gain-of-function mutation in ornithine decarboxylase (SpeC) which altered substrate and reaction specificity. The third pathway emerged that relies on synthesis of polyamines.

This work also serves as a demonstration of how metabolism can be evolved to overcome impaired metabolic function that may be incurred through engineering microbes for production. As an example, we focused on a modified strain of E. coli that is capable of producing high titers of butanol in rich media using an anaerobic, growth-coupled, modified Clostridial CoA-dependent pathway. For unknown reasons, strain modifications impaired metabolic function. Using directed strain evolution, a strain was acquired that has improved growth, titers and butanol yields. We further identified several mutations that adapted energy and carbon

vi

metabolism and optimized expression of pathway enzymes. These works collectively demonstrate the elucidation of adaptive mechanisms of cell metabolism and further, they demonstrate applications in strain engineering.

Direct conversion of CO2 into chemical production through cyanobacteria is a desirable solution to utilizing the excess amount of greenhouse gases released by burning fossil fuel. Several important chemical products including iso-butanol, n-butanol and 3-hydroxypropionate have been demonstrated in our and other laboratories. However, several issues remain unsolved before practical deployment of such technologies. These problems require both scientific and engineering investigations. In this thesis, we start by reporting an effort on improving n-butanol production from Synechococcus elongatus PCC7942 by determining the bottleneck in a production strain. Then, we focus on determining the physiological role of phosphoketolase (Xpk), which is an enzyme that can be used in the synthetic pathway, non-oxidative glycolysis. Although this enzyme is prevalent in cyanobacteria, its physiological role or biochemical properties are still unknown. Therefore, the second part of my thesis describes the characterization of the biochemical and physiological roles of Xpk in cyanobacteria.To improve n-butanol production, pathway optimization for heterologous expression is one of the important strategies to enhance the carbon flux toward the targeted metabolite. Here, butyryl-CoA is identified as the bottlenecks in the original n-butanol producer through heterologous expression of Clostridial CoA dependent butanol production pathway in Synechococcus elongatus PCC7942. The combined strategy of enhancing the alcohol dehydrogenase (PduP) activity and overexpression of the single unit acetyl-CoA carboxylase from Yarrowia lipolytica, increases n-butanol production from 285mg/L in the original producer to 355mg/L in the strain DC11 after induction 10 days. To further improve metabolite production derived from acetyl-CoA, Xpk expression has been suggested to form the non-oxidative glycolysis pathway that conserves carbon. Xpk is an enzyme that links several primary metabolic pathways in cyanobacteria, including Calvin-Benson-Bassham (CBB) pathway, oxidative pentose phosphate (OPP) pathway, Embden–Meyerhof–Parnas (EMP) pathway and the tricarboxylic acid (TCA) cycle. Putative sequences annotated as Xpk are widespread in most well-studied cyanobacteria strains. However, little is known about its physiological functions. Here, we show that Xpk from S. elongatus PCC7942 is inhibited by ATP. Further surveying Xpk’s from different organisms indicates that ATP inhibition is conserved among cyanobacteria strains, suggesting its potential roles under low ATP conditions. Moreover, Xpk in S. elongatus PCC7942 and Cyanothece ATCC51142 show that they are dusk-peaked genes in day-night (12hr/12hr) cycle grown cultures, suggesting their roles in dark conditions. Finally, we find that knocking out xpk in S. elongatus PCC7942 abolished acetate production under dark and anaerobic condition. Interestingly, under this condition, Xpk pathway is particularly important in response to osmotic shock such as NaHCO3, KHCO3 and NaCl and its deletion causes viability decrease and glucose secretion as part of the incomplete glycogen degradation.