Integrating manual and automated metabolic engineering methods

metabolic modeling;metabolic reconstruction;manual curation;media database;MediaDB;methanococcus maripaludis;model morphing;integrated methods;anaerobic methane oxidation (AOM)

Richards, Matthew Aaron

关键词: metabolic modeling;    metabolic reconstruction;    manual curation;    media database;    MediaDB;    methanococcus maripaludis;    model morphing;    integrated methods;    anaerobic methane oxidation (AOM);   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/90748/RICHARDS-DISSERTATION-2016.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Metabolic engineering—the process of altering an organism's metabolism to achieve a desired goal—presents an alternative to established chemical processes. By tapping into the enormous range of metabolic transformations carried out by microbes, we can create microbial cell factories that improve upon chemical processes efficiencies while eliminating potential waste products. One of the greatest obstacles to realizing this potential is our incomplete picture of the possibilities of metabolism, a challenge we address by developing automated tools and databases to guide our inquiries. Though these high throughput studies have greatly accelerated the timeline to go from sequencing a genome to piecing together overall metabolic functions, these accelerated results generally come at the price of higher error rates. In my thesis work, I investigated ways to mitigate this loss of precision by more deeply integrating manually curated information into automated approaches. In the first part of my thesis, I focused on defined microbial growth media, the essential substances that comprise the raw materials of biochemical processes. The vast majority of microbes cannot currently be cultured in a laboratory, a formidable obstacle to characterizing these organisms and their metabolisms. Methods that predict new defined media could expedite culturing experiments; however, such efforts require a repository of known defined media that collects successful growth conditions. To address this need, I created MediaDB, an open access database of chemically defined microbial media from published biochemical literature. MediaDB enables studies across different media that can reveal emergent trends in known media formulations across organisms. By examining media in the database, I found that they often contain similar trace mineral and vitamin solutions, suggesting a measure of uniformity in the way that biologists have traditionally created growth media. Clustering organisms based on their media compounds, I found no connection between media similarity and organism phylogeny, though several cases demonstrated a link connecting media to specific metabolic functions.For the second part of my thesis, I built a genome scale metabolic reconstruction for Methanococcus maripaludis, an archaeon that produces methane from CO2 and H2. M. maripaludis provided an excellent engineering target, both for modifying forward methanogenesis as well as for working to oxidize methane to methanol, a first step towards building a pathway to liquid fuel that is of interest to the Department of Energy. I reconstructed my metabolic network model by relying chiefly on manual, resulting in the first network to correctly depict hydrogenotrophic methanogenesis. My reconstruction demonstrates the importance of electron bifurcation in central metabolism, providing both a window into hydrogenotrophic methanogenesis and platform to generate metabolic engineering hypotheses. I validated my model on growth yield and gene knockout data, showing its strong ability to reproduce experimentally measured results. Using the completed network, I predicted the previously unknown gene for glycine biosynthesis, a hypothesis I am now verifying with auxotrophic growth experiments. Moreover, I generated strain designs to achieve energetically feasible conversion of methane to methanol and in doing so, further demonstrated the vital role of manual curation for these predicted engineering strategies. For the final piece of my thesis, I explored how to leverage manual curation to improve automated metabolic reconstruction. To this end, I created a method that “morphs” a manually curated metabolic model to a draft model of a closely related organism. My method combines genes from the original manually curated model with genes from an annotation database to create a final structure that contains gene-associated reactions from both sources. I used this method to create morphed models of three methanogens from iMR540 and showed that phylogenetic similarity between the source and target organisms correlated with the similarity of their models. I also found that gene annotations from iMR540 showed very low intersection with those from the annotation database, demonstrating the volume of information added by my manual curation. The morphing method could provide a viable alternative to other automated reconstruction methods for organisms that are dissimilar from those that form the foundation of annotation databases.Together, my work exemplifies the advantages conferred by integrating manual methods with automated tools. My studies demonstrate the importance of maximizing the information we glean from manually curated data and blending that data with automated tools that accelerate large scale studies of metabolism. Such approaches mitigate the pitfalls associated with relying solely on automated methods, ensuring the high quality and depth of data as we work to characterize the space of microbial metabolism.