Biological carbon fixation is a vital process in the global carbon cycle, annually assimilating over 380 gigatons of CO₂. This natural process, undertaken by plants, algae, and other autotrophs through various CO₂-fixation pathways, is crucial for mitigating climate change. However, nature has explored a limited solution space for CO₂ fixation, suggesting that many theoretically possible pathways remain undiscovered. Synthetic biology provides the tools to design and construct these new-to-nature pathways by combining enzymes and reactions from diverse sources. While some oxygen-insensitive synthetic CO₂-fixation pathways like CETCH and rGPS-MCG have been successfully demonstrated in vitro, their in vivo implementation remains a challenge. This challenge increases with the complexity of the pathway, as the potential for unwanted interactions with the host cell's metabolic network grows. While some naturally occurring CO₂-fixation pathways have been successfully transferred to *E. coli*, the implementation of large, orthogonal synthetic pathways in vivo is still largely unproven. This study aims to address this gap by designing, optimizing, and implementing a novel synthetic CO₂-fixation pathway, the THETA cycle, in vivo. The THETA cycle, which converts CO₂ directly into acetyl-CoA, a central metabolic building block, has the potential to serve as a versatile platform for synthesizing various value-added compounds. This work showcases the design and construction of the THETA cycle, its in vitro optimization using rational design and machine learning, and its modular in vivo implementation in *E. coli*, demonstrating a significant step towards fully integrating this complex pathway into a living cell.

Seven naturally occurring CO₂ fixation pathways are known, each with unique physiological properties and environmental adaptations. The limited diversity of these pathways suggests significant unexplored potential for improved CO₂ fixation. Theoretical studies have proposed over 30 new-to-nature CO₂ fixation pathways. Previous work has successfully demonstrated two oxygen-insensitive designs, the CETCH and rGPS-MCG cycles, in vitro. These in vitro systems, while valuable, highlight the challenge of transferring complex, orthogonal metabolic pathways into living cells. Existing work has shown the successful transfer of naturally occurring CO₂ fixation pathways into model organisms like *E. coli*, but this success hasn't been replicated for complex, synthetic pathways. The challenge lies in the potential for interactions between the synthetic pathway and the host cell's existing metabolic network. This study builds upon these advancements by focusing on a novel synthetic pathway and demonstrating its implementation in vivo.

The THETA cycle was designed based on two highly efficient CO₂-fixing enzymes: phosphoenolpyruvate carboxylase (Ppc) and crotonyl-CoA carboxylase/reductase (Ccr). These enzymes were selected for their kinetic, thermodynamic, and mechanistic properties. The THETA cycle was conceptually divided into three modules: Module 1 (pyruvate to succinate), Module 2 (succinate to crotonyl-CoA), and Module 3 (crotonyl-CoA to acetyl-CoA and pyruvate). Each module was constructed in vitro using purified enzymes, and their activity was assessed through liquid chromatography-mass spectrometry (LC-MS). The in vitro construction of the complete THETA cycle (THETA 1.0) demonstrated initial functionality, but also revealed the accumulation of an unwanted byproduct, malyl-CoA. This issue was addressed through rational design (THETA 2.0), which included the addition of malyl-CoA thioesterase. Further optimization involved implementing a fumarate reductase bypass (THETA 3.0) using dihydroorotate dehydrogenase (DHOD1a/b), addressing a bottleneck identified by testing the cycle with different starting intermediates. Subsequently, machine learning-guided optimization (METIS workflow) was employed to refine the reaction conditions and enzyme concentrations, leading to a significant increase in acetyl-CoA production (THETA 3.9.9). The optimized THETA cycle was expanded to produce other valuable molecules, malonyl-CoA and glycolate, by integrating additional enzymes. For in vivo implementation, three *E. coli* strains were engineered to test the function of each module individually. Module 1 was implemented in a succinyl-CoA auxotrophic strain. Module 2 was integrated into an acetyl-CoA auxotrophic strain by creating a pathway to convert crotonyl-CoA to acetyl-CoA, and short-term evolution was utilized to select for strains that could grow on succinate. Module 3 was tested in two different acetyl-CoA auxotrophic strains. In these in vivo experiments, the activity of each module was confirmed through growth rescue experiments and ¹³C-labeling experiments. ¹³C-labeling experiments, using [U-¹³C]glucose and non-labeled acetate/CO₂, were used to validate the flux through the introduced modules. Whole-genome sequencing was used to analyze the genetic changes in evolved strains. Detailed methods are described in the supplementary information.

This research successfully designed, constructed, and optimized a novel synthetic CO₂-fixation pathway, the THETA cycle. The in vitro implementation of the THETA cycle demonstrated the production of acetyl-CoA from CO₂, initially at low yields, and subsequently improved significantly through rational design and machine learning optimization. Rational design improvements, including implementing a fumarate reductase bypass and adding malyl-CoA thioesterase, increased acetyl-CoA production by 20-fold. Machine learning-guided optimization further increased production by a factor of five. The optimized THETA cycle (THETA 3.9.9) produced approximately 890 µM acetyl-CoA in vitro, a 135-fold improvement over the initial version (THETA 1.0). The versatility of the THETA cycle was demonstrated by expanding its output to include malonyl-CoA and glycolate. Each module of the THETA cycle was successfully implemented in *E. coli* in vivo. The in vivo functionality was demonstrated using growth rescue experiments and ¹³C-isotope labeling. Growth rescue experiments confirmed that each module could support the growth of specific auxotrophic *E. coli* strains. ¹³C-isotope labeling analysis showed the expected labeling patterns in various metabolites, confirming the activity of the introduced modules in vivo. Unexpected findings included the discovery of native *E. coli* enzymes, YciA and TesB, capable of hydrolyzing methylsuccinyl-CoA and the unexpected activity of succinate dehydrogenase (Sdh) in oxidizing methylsuccinate. These unexpected findings were incorporated into the in vivo implementation of Module 3. Overall, this study demonstrates a successful example of constructing a complex synthetic metabolic pathway and implementing it in vivo.

This study successfully addressed the challenge of implementing a complex, synthetic CO₂-fixation pathway in vivo. The significant improvement in THETA cycle yield, achieved through a combination of rational design and machine learning, demonstrates the power of these approaches for optimizing complex in vitro systems. The modular design of the THETA cycle facilitated its in vivo implementation, allowing the investigation of individual modules in the context of the host cell's metabolism. The successful in vivo implementation of each module, validated by both growth experiments and isotopic labeling, validates the feasibility of this approach. The unexpected findings highlight the complexities of transferring synthetic pathways into living cells, emphasizing the need for iterative design and optimization strategies that account for interactions with existing cellular components. The in vivo experiments revealed the importance of accounting for the activity of native cellular components, which can both hinder and enhance the activity of the introduced pathway. This work represents a significant advance in the field of synthetic biology and metabolic engineering, offering a blueprint for creating more complex and efficient synthetic CO₂ fixation systems for applications in sustainable bioproduction. Future studies should focus on the complete integration of the THETA cycle in vivo and its further optimization for autotrophic growth.

This study demonstrates the successful design, in vitro optimization, and modular in vivo implementation of the THETA cycle, a novel synthetic CO₂-fixation pathway. The significant increase in acetyl-CoA production through rational design and machine learning optimization, along with the successful implementation of individual modules in *E. coli*, represent a major step forward in synthetic biology. Future research could focus on integrating the full THETA cycle into a single *E. coli* strain, optimizing for autotrophic growth and exploring the potential for producing a wide range of valuable compounds directly from CO₂.

The study primarily focused on the modular implementation of the THETA cycle in *E. coli*. While this demonstrated the in vivo functionality of the individual modules, future work is needed to fully integrate the entire cycle in vivo. The in vivo experiments require further optimization to improve the yield and efficiency of CO₂ fixation. The study utilized specific *E. coli* strains and genetic modifications. The generalizability of these results to other organisms requires further investigation. The current in vitro CO₂ fixation rate is comparable to other synthetic systems but still lower than naturally occurring pathways. The study did not address the long-term stability of the THETA cycle in vivo.