2014 ME-X Conference Program Book Poster Abstracts

1 Poster Abstracts

Session Chairs: Irina Borodina, Novo Nordisk Biosustainability Center; Michael Jewett, Northwestern University; Julius Lucks, Cornell University; Caroline Peres, DuPont

1. The Genetic and Metabolic Regulation of 2. Isotopically Nonstationary 13C Flux Analysis of Rhamnolipids Biosynthesis in Pseudomonas Isobutyraldehyde Production in Synechococcus Aeruginosa Reveals New Engineering Strategies Elongatus for Heterologous Expression Adeola Adebiyi Ahmad M. Abdel-Mawgoud*, François Lepine Chemical and Biomolecular Engineering, Vanderbilt and Eric Déziel University, Nashville, TN INRS-Institut Armand-Frappier, Laval, QC, Canada Recent studies have demonstrated the feasibility of convert- Rhamnolipids (RLs) are bacterially-produced surfactants ing energy from sunlight and carbon from CO2 directly into of growing biotechnological importance. The first-known biofuels using photosynthetic microorganisms. Despite the producer is the opportunistic pathogen Pseudomonas advances made in cyanobacterial biofuels production, the aeruginosa. Although much is known about RLs biosyn- productivities achieved by cyanobacterial fermentations are thesis, the exact metabolic and genetic regulation of RLs currently too low for industrial feasibility and few tools are biosynthesis in this prototypic producer is still insufficient for available that specifically address the challenges of redirect- the development of a successful RLs metabolic engineering ing and enhancing metabolic flux in photosynthetic microbes. strategy in non-pathogenic host. In this study, we characterize the metabolic and genetic changes naturally taking Our group is developing novel approaches that use isotope place in P. aeruginosa PA14 in mineral salts medium (MSM) tracers and metabolic flux analysis (MFA) to quantitatively as- in which RLs is 100 times more produced compared to a sess in vivo metabolic phenotypes of photoautotrophic hosts. rich, undefined medium, tryptic soy broth (TSB). In MSM, Although 13C is the preferred isotope tracer for mapping production of the lipid precursor of RLs 3-hydroxydecano- central carbon metabolism in heterotrophic hosts, photo- ate is at least 30 times higher and exclusively in the R- chiral autotrophs assimilate carbon solely from CO2 and therefore form, compared to growth in TSB. We hypothesize this is produce a uniform steady-state 13C-labeling pattern that is attributed to an increase (2-3 times) in R-specific enoyl-CoA insensitive to fluxes. However, transient measurements of iso- hydratase (ECH) activity. Interestingly, the expression of tope incorporation following a step change from unlabeled to rhlYZ, encoding the R-ECH proposed to be implicated in labeled CO2 can be used to map photoautotrophic fluxes by R-3-hydroxydecanoate precursor supply, is under control of applying newly developed techniques of isotopically nonsta- the quorum sensing system together with the other precur- tionary MFA (INST-MFA). We have recently developed a novel sor supplying operon, rmlBDAC, coding for rhamnose, and software package called INCA to facilitate model generation with the genes coding for precursor linking enzymes, rhlAB. and computational solution of INST-MFA models, which is Furthermore, comparative expression experiments using now publicly available to the scientific community. We have qRT-PCR in MSM compared to TSB demonstrated that 13 also established experimental protocols for performing CO2 RhlAB is the key element in RLs overproduction in MSM as labeling experiments and mass isotopomer analysis that are it is 85 times more expressed than in TSB. This transcrip- required for INST-MFA of autotrophic hosts. tional data, together with the results of homologous expression of the two operons coding for RLs precursors, rhlYZ To establish proof-of-concept, we first applied13 C I and rmlBDAC, further corroborate that the RhlAB-mediated NST-MFA to map fluxes in the model cyanobacterium step is rate-limiting in the biosynthesis of RLs rather than Synechocystis sp. PCC 6803 growing under photoautotro- the precursor supply step. Although low RLs yields were phic conditions. Comparison of the INST-MFA flux map to obtained, heterologous expression of rhlAB was successful theoretical values predicted by a linear programming model in hosts furnishing the two RLs precursors, namely P. putida revealed inefficiencies in photosynthesis due to oxidative and P. fluorescens. Yet, E. coli DH5α host required rmlBDAC pentose phosphate pathway and malic enzyme activity. co-expression. This study enhances the understanding of Our ongoing work involves extending the 13C INST-MFA regulation of RLs biosynthesis and proposes a diagnostic approach to examine engineered strains of Synechococ- tool that helps in the selection of the appropriate non-patho- cus elongatus PCC 7942, with the goal of identifying novel genic hosts for heterologous RLs production. genetic targets that control production of isobutyraldehyde 2 Poster Abstracts

(IBA, a direct precursor of isobutanol). Quantification of pho- microorganism for the production of lactic acid is one that tosynthetic carbon fluxes in IBA-producing cyanobacteria is produces pure enantiomeric forms, with high yield, and low expected to pinpoint pathway bottlenecks that can be sub- cost substrate. sequently removed in further rounds of metabolic engineering, thus leading to maximal productivity by redirecting flux The development of an efficient process for the industrial pro- into biofuel-producing pathways. duction of poly (lactic acid) directly depends on the economically viable production of lactic acid. Currently, the production 3. Characterization of LDH Genes for L-Lactic Acid of lactic acid is mainly made from starch6. However with the Production in Pichia Pastoris increasing production of biodiesel there is an overproduction Pollyne B. A. Almeida* of glycerol. This is the major waste produced during the con- Molecular Biology, Universidade de Brasília, Brasília, Brazil version of vegetable oil into biodiesel. Furthermore, the use Recently, the demand for plastic material is growing substan- of glycerol is considered advantageous to reduce the cost of tially. Since most of it nowadays is not recyclable, the amount biodiesel production in the biorefinery7. of plastic deposited in the environment reaches 140 million tons annually1. Those are estimated to degrade in about a Different microorganisms have been genetically modified to thousand years when derived from petrochemical sources. produce lactic acid, but in most cases sugar is used as sub- Therefore due to environmental and economical issues, strate. Pichia pastoris, a metilotrophic yeast, can reach high efforts turned into the development of green alternatives to cell density using crude glycerol as carbon source, survive in petrochemical derived plastics. low pH levels, although it is not able to produce lactic acid. Genetic engineering can make possible this production, from Biomaterials are natural products that are synthesised and overexpression of the gene that encode an enzyme able to catabolised by different organisms and that have found broad catalyze the conversion of pyruvate to lactate. The enzyme biotechnological applications. Bioplastics, also known as lactate dehydrogenase (LDH) is encoded by specific genes green-plastic, are a special type of biomaterial made from from different species. It determines yield and purity of pro- renewable biomass feedstock such as corn and sugar-cane2. duced lactic acid. They are polyesters, produced by a range of microbes, cultured under different nutrient and environmental condi- In this study, different strains are constructed to enable high tions3. During its synthesis less carbon is emitted and lower yield and productivity of lactic acid from crude glycerol. At energy is required. Furthermore bioplastics are biodegradable least 4 different genes encoding for LDH have been tested which facilitates recycling thus reducing the amount of plastic for l-lactic acid production in P.pastoris. In parallel genetic deposited in the environment. engineering strategies for increased pyruvate inside de the cell and secreted lactic acid have also been tested. After The development of a more sustainable society is expected strain development, fermentations are carried out to deter- to gradually replace petroleum derived plastics by bioplas- mine which recombinant strains results in higher yield and tics. Thus, the demand for this type of material is estimated productivity of L-lactic acid. to grow at 20-30% by 2016. L-Lactic acid (2- hidroxy References propionic acid) is widely used in food, chemical, cosmetics and pharmaceutical industries. Also, it is the monomer of 1. Nampoothiri, K. M., et al., 2010. An overview of the recent developments n polylactide (PLA) research. Bioresource Technology. polylactic acid (PLA), which is a bioplastic. PLA products 101, 8493-8501. can be used in a wide variety of applications ranging from 2. Iles, A., Martin, A.N. 2013. Expanding bioplastics production: packaging to fibers and foams4. sustainable business innovation in the chemical industry. Journal of Cleaner Production. 45, 38-49. The purity of the lactic acid is important as it influences 3. Luengo, J.M., et al., 2003. Bioplastics from microorganisms. the physico-chemical properties of the poly (lactic acid), Current Opinion in Microbiology. 6, 251–260 for example, the polymer L-lactic acid is the desired form 4. Abdel-Rahman, M. A., et al., 2013. Recent advances in lactic 5 for the production of packages . Consequently the ideal acid production by microbial fermentation processes. Biotechnol-

3 Poster Abstracts

ogy Advances. Article in press. levels. We approximated this space with well-determined 5. Jamshidian, M., et al., 2010. Poly-Latic Acid: Production, sub-spaces, generated by classification rules on the me- Applications, Nanocomposites and Release Studies. Comprehen- tabolite concentrations values. Through this approximation, sive Reviews in Food Science and Food Safety. 9, 552-571. we achieve two important objectives: (i) the computation of 6. Adsul, M. G., et al., 2011. Development of biocatalysts for the stable physiological volume, as the sum of the volumes production of commodity chemicals from lignocellulosic biomass. of the sub-spaces; (ii) the characterization of all the topologi- Bioresource Technology. 102, 4304-4312. cal properties of the space, with the most important being 7. Fernando, S., et al., 2006. Biorefineries: Current satatus, the contiguity of the space. The latter helps to discriminate challenges, and future direction. Energy & Fuels. 20, 1727-1737. the quantities that determine the attainment of selected 8. Okano, K., et al., 2010. Biotechnological production of production profiles. We applied the novel approach against enantiomeric pure lactic acid from renewable resources: recent reported experimental studies of glucose-fed E. coli for the achievements, perspectives, and limits. Applied Microbiology and production of 1,4-butanediol. We have found the metabolite Biotechnology. 85, 413-423 concentrations and enzyme saturations, and the relations between these conditions, which are the most decisive 4. Identification of Key Metabolite Concentrations and Enzyme Saturations Determining the in realizing a specific growth and production phase. The Physiological States of Glucose-Fed E. coli for the proposed approach offers a new way to identify in advance Production of 1,4-Butanediol which quantities should be measured to better characterize Stefano Andreozzi*1,2, Ljubisa Miskovic1,2 and a particular physiological state, and it can guide experiments Vassily Hatzimanikatis1,2 for the improvement of biocatalyst performance. 1Swiss Institute of Bioinformatics, Lausanne, Switzerland 2Laboratory of Computational Systems Biotechnology, 5. New Advances in 13C Metabolic Flux Analysis: EPFL, Lausanne, Switzerland Complete-MFA, Co-Culture MFA and Dynamic MFA

Metabolic Engineering is fostered by an increasing effort Maciek R. Antoniewicz* Department of Chemical and Biomolecular Engineering, and progress in the collection of large and accurate ‘omics’ University of Delaware, Newark, DE datasets. The abundance of available measurements and

advances in measurement techniques only alleviate to a cer- Measuring fluxes by13 C metabolic flux analysis 13( C-MFA) tain extent the uncertainty around the physiological states of has become a key activity in metabolic engineering, biotech- an organism. In fact, due to intrinsic complexity of metabolic nology and medicine. In this talk, I will present three major

networks, it is not possible to determine their exact intracel- new advances in the field of13 C-MFA that are extending the lular states such as metabolic flux distributions by integrat- range of biological systems that can be analyzed with this ing only the currently available fluxomics and metabolomics technique and the types of biological questions that can data into a model. For these reasons, it is desirable to have be addressed. The three major advances that I will discuss a method that further reduces the uncertainty in depicting are: 1) COMPLETE-MFA (or complementary parallel labeling the actual intracellular states. To tackle this problem, we experiments technique for metabolic flux analysis), which propose a novel computational approach, based on Monte improves the precision and accuracy of flux estimates by Carlo sampling and machine learning classification. We about one-order-of-magnitude; 2) co-culture MFA, which start from the integration of known physiological informa- allows metabolic fluxes to be measured in multi-cellular tion, in the form of cultivation data, thermodynamics, kinetic systems; and 3) dynamic MFA, which extends flux analysis rate laws of the constituent reactions and available kinetic to dynamically changing biological systems. parameters. This allows the formulation of kinetic models that describe the possible behaviors, realized by the inte- 1. The COMPLETE-MFA methodology that we have devel- grated information. We exhaustively sampled the space of oped is based on combined analysis of multiple isotopic metabolite concentration levels and enzyme saturations. labeling experiments, where the synergy of using comple- We performed a consistency check on the models of the mentary tracers greatly improves the precision of estimated analyzed physiology to unravel conditions on these quanti- fluxes. Here, I will demonstrate the COMPLETEMFA

ties that reflect particular physiological states, reducing approach using all singly labeled glucose tracers, [1-13C], [2-

even more the space of realizable metabolite concentration 13C], [3-13C], [4-13C], [5-13C], and [6-13C]glucose to determine 4 Poster Abstracts

precise metabolic fluxes for wild-type E. coli, and using all integrating timeseries of metabolite measurements and non- singly labeled xylose tracers, [1-13C], [2-13C], [3-13C], [4-13C], stationary 13C-labeling data to quantify flux changes in time. and [5-13C]xylose to determine precise metabolic fluxes for Thus, this allows us to measure for the first time dynamically Thermus thermophilus. In these studies, cells were grown changing metabolic fluxes, for example, in fed batch fermen- in multiple parallel cultures on defined medium with glucose tations. Three key advantages of our 13CDMFA method are: and/or xylose as the only carbon source. Mass isotopomers 1) time-series of metabolite concentration and labeling data of biomass amino acids were measured by gas chromatog- can be applied directly for estimating dynamic fluxes, mak- raphy-mass spectrometry (GC-MS) and the data from all ing data smoothing unnecessary; 2) characteristic metabolic experiments were then fitted simultaneously to a single flux phases during a culture are identified automatically by the model to determine accurate intracellular fluxes. In all cases, algorithm; 3) 13C-labeling data provides insights into tran- we obtained a statistically acceptable fit with more than 300 sients in fluxes of parallel and cyclic pathways that cannot redundant measurements. As I will demonstrate, the flux be observed without labeling. maps that we have determined here are the most precise flux results obtained thus far (by about order-of-magnitude) 6. Parallel Labeling Experiments: A Novel Approach for for any biological system. Validating Metabolic Network Models Jennifer Au* and Maciek R. Antoniewicz 2. Microbial communities play an important role in biofuel Department of Chemical and Biomolecular Engineering, production, biomedical research, food production, and University of Delaware, Newark, DE waste water treatment. Co-culture systems particularly have Metabolic network models play a central role in metabolic unique advantages in optimizing product yield as a result of engineering applications. However, these models are often synergistic interactions. To gain insight into these systems not rigorously validated, but rather based on automated we have developed the first methodology for measuring annotations with limited experimental supporting data. metabolic fluxes in multicellular systems. Here, I will demon- Here, we present a novel comprehensive experimental and strate our novel co-culture 13C-MFA framework that does not computational framework for validating metabolic network require any physical separation of cells or proteins. Specifi- models using parallel labeling experiments, in combination cally, we have developed a new computational approach with 13C metabolic flux analysis 13( C-MFA) and statistical for modeling isotopic labeling in biological systems that analysis. We have applied the methodology to elucidate core allows fluxes in individual populations to be computationally metabolism of Clostridium acetobutylicum. After multiple deconvoluted from the overall co-culture 13Clabeling data. rounds of model testing and updating we present a new We show that the overall 13C-labeling data has abundant invalidated metabolic network model for C. acetobutylicum formation not only to estimate the fluxes in multiple popula- that can be used for unbiased metabolic flux measurements tions, but also to determine the fraction of each cell using 13C-MFA. The model was used to study the metabolic population in the co-culture (e.g. to visualize co-culture stress response of C. acetobutylicum under butanol and dynamics). I will demonstrate the coculture flux analysis butyrate stresses. methodology using a co-culture system of two E. coli knockout strains, Δzwf (knockout of the first step in the pentose Although the biochemistry of C. acetobutylicum has been phosphate pathway) and Δpgi (knockout of the first step in extensively reviewed, the central metabolic pathways have glycolysis pathway), using a yeast/E. coli co-culture, and us- remained only partially resolved. Two recent reconstructions ing a thermophilic co-culture system. The new flux analysis of genome-scale models have proposed different mecha- methodology that we have developed for analyzing co-cul- nisms for the biosynthesis of α-ketoglutarate, the precursor ture systems adds a new dimension to the field of for glutamate, glutamine and proline. Initial stable-isotope labeling experiments and qualitative 13C-isotopomer analysis 13C-MFA and provides an enormous resource to the metabolic engineering and biotechnology communities. have supported the idea of an incomplete TCA cycle and suggested a Re-stereospecificity for the citrate synthase 3. Finally, we have developed new methods for dynamic reaction. However, quantitative analysis of metabolic fluxes had not been performed. In this work, we have rigorously 13C-MFA, for biological systems that are not at metabolic validated, for the first time, the metabolic model of steady state. The 13C-DMFA methodology is based on

5 Poster Abstracts

C. acetobutylicum. Using the novel parallel labeling experi- production of DHB have been developed that employ the ments approach we quantitatively elucidated core metabo- previously unreported enzymatic activities malate kinase, lism (including amino acid metabolism) of C. acetobutylicum. malate semialdehyde dehydrogenase and DHB dehydroge- Contrary to previously proposed hypotheses, we found that nase1 whereas the second pathway employs malate synthe- while the TCA cycle runs in the oxidative direction, there is tase or malyl-CoA lyase, previously unreported malyl-CoA2 no notable flux betweenα -ketoglutarate and succinyl-CoA or reductase, and malate semialdehyde reductase activities2. succinate and fumarate, and that the conversion of succinyl- Most of these enzymes were obtained either by rational CoA to succinate proceeds independently. Using multiple design based on structural and mechanistic knowledge 13C-labeled amino acid tracers, we additionally showed that of candidate enzymes acting on sterically similar cognate there is no flux between malate and oxaloacetate, and that substrates, or they were identified by screening of natural there exists a previously unknown metabolic cycle where enzymes and further improved by rational design. As the carbon flows from aspartate to threonine, serine, pyruvate, two pathways depart from malate or glyoxylate, metabolic oxaloacetate and back to aspartate. Finally, we identified engineering of E. coli for optimizing these DHB precursors a putative citramalate synthase gene that is in fact the first was carried out that differed from previous reported works3. step in isoleucine biosynthesis in C. acetobutylicum. Finally, the synthetic pathways were expressed in the genetically optimized E. coli strains, and significant in vivoproduc- The validated metabolic network model was used to mea- tion of DHB was obtained. sure metabolic fluxes in C. acetobutylicum under butanol and butyrate stress. The metabolic flux distributions result- [1] Walther, T., Cordiez, H., Topham Ch., Andre I, Remaud-Simeon, ing from this analysis provide the basis for ongoing work on M., Huet, R. & François, J. (2012) A novel method of production of improving the tolerance of C. acetobutylicum to these fer- 2,4-dihydroxybutyric. Patent WO2012/056318A1 mentation products. Results from this ongoing effort will be (2]Walther, T., Cordier, H., Dressaire, C. François, J. (2013). A presented. We note that the systematic approach employed production of 2,4-dihydroxybutyric by malyl-CoA pathway . Patent WO2013/160762A1 in this study can be easily applied to elucidate metabolism of [3] Zhang X, Wang X, Shanmugam KT, Ingram LO (2011) L-malate other poorly characterized organisms that are of importance production by metabolically engineered Escherichia coli. Appl to metabolic engineering. Environ Microbiol 77:427-434.

7. Synthetic Microbial Metabolism Refactoring for 8. Implementation of a Disassociated Fatty Acid the Production of a Chemical Synthon, Synthase System (FAS type II) in Saccharomyces 2,4-Dihydroxybutyric Acid cerevisiae for Fatty Acid and Wax Ester Production Clément Auriol1-4; Audrey Baylac1-4, Romain Irague1-4, Flávio Azevedo* and Björn Johansson Christopher Topham1-4, Clémentine Dressaire1-4, Jean Marie CBMA - Center of Molecular and Environmental Biology/ François1-4 and Thomas Walther1-4 Department of Biology, University of Minho, Braga, Portugal 1Université de Toulouse; INSA, UPS, INP; LISBP, 135 Avenue de Rangueil, 31077 Toulouse, France 2NRA, UMR792 Ingénierie des Systèmes Biologiques et des Microbial biosynthesis of fats and oils from renewable Procédés, Toulouse, France carbon sources has attracted significant attention in re- 3CNRS, UMR5504, Toulouse, France cent years for the potential production of biofuel and other 4TWB, 3 rue des Satellites, Canal Biotech Building 2, 31400 commodities. Almost all organisms synthesize de-novo Toulouse, France fatty acids via a well conserved cyclic series of four reac-

tions involving the condensation, reduction, dehydration and 2,4-Dihydroxybutyric acid (DHB) is a molecule with consid- reduction of carbon-carbon bonds. In nature there are two erable potential as a versatile chemical synthon. Notably, main types of fatty acid synthase systems (FAS), type I and it may serve as a precursor for chemical synthesis of the type II. FAS I systems utilize a single large, multifunctional methionine analogue 2-hydroxy-4-(methylthio)butyrate, thus, polypeptide and are common to both mammals and fungi, targeting a considerable market in animal nutrition. However, although with some structural differences. On the other petrochemical synthesis of DHB is not economically viable, hand, Plants and Bacteria’s (as well in mitochondria and and no natural metabolic pathways exist for the biochemical chloroplasts) utilize the disassociated FAS type II system. production of DHB. Synthetic metabolic pathways for the The reactions of the type II FAS, unlike FAS I, are catalyzed

6 Poster Abstracts

by discrete and monofunctional enzymes, the proteins are ogy Development Program to Solve Climate Changes on all expressed as individual polypeptides from separated Systems Metabolic Engineering for Biorefineries from the genes. The organization of FAS II enables the synthesis of Ministry of Science, ICT and Future Planning (MSIP) through several fatty acid products and is more amenable to modifi- the National Research Foundation (NRF) of Korea (NRF- cation of chain length. The separation of functions in single 2012-C1AAA001-2012M1A2A2026556).] polypeptides also facilitates the metabolic optimization of 10. Challenges in Reverse Engineering of Industrial each reaction step along the sequence, which is impossible Fermentation Strains with FAS I, since the subunits are stoichiometrically fixed. In Kirsten R. Benjamin*, Annie Tsong, Chris Reeves, Benjamin this work we implemented a FAS II system consisting of 12 Kaufmann, Amoolya Singh, Jeff Ubersax, Michele Fleck, individually expressed genes in a S. cerevisiae strain carry- Lauren Pickens, Maxime Durot and Darren Platt ing a conditional allele of the fatty acid synthase genes of the 1Amyris Inc., 5885 Hollis Street, Ste. 100, Emeryville, endogenous FAS system. The physiological consequences CA94608, United States of America of this expression will be discussed. At Amyris, we optimize S. cerevisiae strains for industrial This work was funded by the FCT project MycoFat PTDC/ fermentations to manufacture biofuels and chemicals. Apply- AAC-AMB/120940/2010. F.A. was supported by an FCT ing a combination of random mutagenesis, high-throughput fellowship SFRH/BD/80934/2011. This work was supported Automated Strain Engineering (ASE), enzyme engineering, by FEDER through POFC – COMPETE and by Portu- high-throughput phenotyping, and a tiered screening ap- guese funds from FCT through the project PEst-OE/BIA/ proach, we identify strains with improved yield, productiv- UI4050/2014 ity, and product recovery. Strain “winners” serve as parents for further rounds of strain improvement, and are dissected 9. Production of 4-Hydroxybutyric Acid By Metabolically using ‘omics technologies, whole genome sequencing, and Engineered Mannheimia Succiniciproducens and Its identification of causal mutant alleles. We will discuss the Conversion to Gamma-Butyrolactone By Acid Treatment merits, challenges and lessons learned from three different Junho Bang*, Sol Choi, Hyun Uk Kim, Won Jun Kim and approaches to strain improvement. First, we will discuss the Sang Yup Lee relative contributions of mutagenic vs. rational genetic engi- Chemical & Biomolecular Engineering, Korea Advanced neering approaches and the benefits of combining the two. Institute of Science and Technology (KAIST), Daejeon, Second, we will illustrate the use of whole genome sequenc- South Korea ing and causal mutation identification to accelerate strain γ-Butyrolactone (GBL) is an important four carbon (C4) improvement. Third, we will describe attempts to reverse chemical, which has a wide range of industrial applications. engineer top strains by building “clean” strains containing GBL can be produced by acid treatment of 4-hydroxybutyric beneficial causal alleles and lacking “collateral damage” of acid (4-HB), which is a derivative of succinic acid. Heterolo- mutagenesis. gous metabolic pathways were designed and established in succinic acid overproducing M. succiniciproducens 11. Metabolic Engineering for Ricinoleic Acid Production LPK7 by the introduction of heterologous genes that en- in Y. Lipolytica code succinyl-CoA synthetase, CoA-dependent succinate Athanasios Beopoulos*1, Jonathan Verbeke1, Florence 2 2 2 2 semialdehyde dehydrogenase and either 4-hydroxybutyrate Bordes , Marie Guicherd , Melusine Bressy , Alain Marty and Jean-Marc Nicaud1 dehydrogenase in LPK7 (p3S4CD) or succinate semialde- 1Micalis, INRA, France hyde reductase in LPK7 (p3SYCD). Fed-batch cultures of 2LISBP, INSA, Toulouse, France LPK7 (p3S4CD) and LPK7 (p3SYCD) resulted in the production of 6.37 and 6.34 g/L of 4-HB, respectively. Finally, GBL Biofuels and oleochemicals produced by microorganisms was produced by acid treatment of the 4-HB obtained from have similar chemical structure and properties to petroleum- the fermentation broth. This study demonstrates that 4-HB, based products. However, producing oleochemicals in yields and potentially other four carbon platform chemicals, can permitting their economic exploitation requires the engineer- be produced by the engineered rumen bacterium M. suc- ing of the microorganism’s metabolism. Such engineering ciniciproducens. [“This work was supported by the Technol- cannot be based on just one specific feedstock or host

7 Poster Abstracts

organism. Synthetic-biology approaches should be used to the heterologous host necessitated the fine-tuning of the Cp- optimize both the host and pathways to maximize specific FAH12 hydroxylase expression to the native Y. lipolytica PDAT oleochemical and/or fuel production. Even though there acyltransferase (Lro1p) expression level to maximize meta- are still challenges so as microbial biodiesel production can bolic flux. The strain co-expressing 3 copies of the CpFAH12 compete with conventional fuels, the production of an added with 2 copies of YlLRO1, always under the control of the TEF value-oleochemical could be efficient enough to attain com- constitutive promoter, not only restored the lipid accumulation mercialization standards. capacity of the host strain, but also drove RA accumulation to over 50% of total lipid content. This preliminary work leads to Here, we present the attempts on the metabolic engineering the production of 12 g/L of RA in fermentor. of the fatty acid pathway of the oleaginous yeast Yarrowia lipolytica for the production of the economically important Furthermore, using functional-genomics approaches we ricinoleic acid (RA). Whereas numerous oleochemical ap- identified the rate-limiting steps in RA production in an at- plications for RA and its derivatives exist, their production is tempt to balance the flux between the oleic acid biosynthetic limited and subject to various safety legislations, mainly due pathway and the oleic to ricinoleic hydroxylation pathway to the extremely poisonous protein ricin, found in abundance that is part of the Lands phospholipid cycle. Overexpression in castor seeds. Y. lipolytica’s innate de novo lipogenesis of the native diacylglycerol:cholineophosphotransferases may be modest (around 40% of cell dry weight in wild type (Ept1p, Cpt1p) accelerated the oleic acid flux from diacylg- strain) but nonetheless the rewiring of its lipid metabolism lycerols (DAG) towards the phosphatidyl-choline (PC) where conducted by several groups, including ours, enhanced lipid hydroxylation takes place. Similar results were obtained by accumulation up to 90% of the engineered strains’ cell dry overexpression of the native acyl-CoA: lysophosphatidyl- weight. Adding this to its genetic tractability and the avail- choline acyltransferase (LPCAT) that facilitated the transport ability of genetic tools, make Y. lipolytica a good candidate of fatty acids between lysophosphatidylcholine (LPC) and as a platform organism for the production of oleochemicals. PC. Overexpression of the native A2 phospholipase permit- ted the discharge of free RA from PC and resulted in the Although Y. lipolytica is unable to naturally produce RA, secretion of RA coupled with increased RA titres. Each of the main fatty acids it synthesizes are oleic acid (the direct these modifications increased ricinoleic production of the precursor of RA) and linoleic acid, the latter being the Δ12 engineered strain by at least 20%. The combination of all desaturation product of the former. The study of RA incor- the aforementioned modifications should provide a robust poration in wild-type, triacylglycerol acyltransferase de- ricinoleic producing cell factory with possible industrial ap- leted (Dga1p, Dga2p and Lro1p) and fatty acid degradation plications. To our knowledge, this work reports the most invalidated (pox1-6Δ) Y. lipolytica strains led us to understand efficient production of RA in an organism other than R. com- the enzymes specificities and the mechanisms governing munis described to date. In order to achieve these rates of the metabolic fate of RA. We therefore proceeded to the production, the lipid metabolism of Y. lipolytica necessitated heterologous expression of the codon optimized Ricinus substantial rerouting. The limiting steps of RA synthesis and communis Δ12 hydroxylase (RcFAH12), under the control of accumulation are identified and will be discussed in an effort the TEF constitutive promoter, after deletion of the sole na- to provide fundamental understanding on the regulation and tive Δ12 desaturase (YlFAD2), in a strain combining all of the the interconnection of the lipogenic metabolic processes. above modifications. However, RA constituted only 7% of the total lipids produced by this engineered strain. By contrast, 12. Fatty Acid Overproduction in a Genetically expression of the codon optimized Claviceps purpurea (Cp- Engineered Puryvate Decarboxylase-Negative Strain FAH12) in the same genetic background resulted in a strain of Saccharomyces cerevisiae able to accumulate RA to 29% of total lipids, and expression Alexandra Bergman*, Verena Siewers and Jens Nielsen of an additional copy of CpFAH12 drove RA accumulation Department of Chemical and Biological Engineering, up to 35% of total lipids. The co-expression of the C. pur- Chalmers University of Technology, Gothenburg, Sweden purea or R. communis type II diacylglycerol acyltransferase The energy supply of modern society is majorly based on (RcDGAT2 or CpDGAT2) had negative effects on RA accumu- fossil fuels. Yet petroleum is a declining resource and easily lation in this yeast, with RA levels dropping to below 14% of accessible reserves are assumed to be depleted in a fore- total lipids. The construction of the RA synthetic pathway in 8 Poster Abstracts

seeable future, which is why current petroleum based fuels to today’s mesophilic white biotechnology work horses are will have to switch into ones established from renewable the thermophilic nature, the anaerobic fermentation capaci- resources. One alternative is to develop efficient cell factories, ties and the possibility to use simultaneous saccharification which can directly convert plant-based material into fuels, and fermentation (SSF) of lignocellulosic substrates at the for example alkanes or fatty acid ethyl esters. To obtain such optimum temperature of commercial hydrolytic enzymes. high-energy fuel compounds, the production host needs to The thermophilic nature will lower production costs via more synthesize high amounts of fatty acids. During fermentation efficient SSF, less contamination risk and less cooling costs. processes, however, the formation of unwanted byproducts will reduce the yield of the sought fuel product. We have isolated a collection of thermophilic Bacilli from environmental samples using several different enrichment The overall objective of this project is to obtain a fatty acid techniques. An initial selection of strains was made based on overproducing strain of Saccharomyces cerevisiae, which product formation and sugar utilization, after which the se- through several chromosomal gene deletions will have lost lected isolates were tested for genetic accessibility. Genetic function to produce ethanol, storage lipids as well as its ca- accessibility was found to be highly strain-specific and only pability of catabolizing fatty acids. Such a yeast strain could a very limited number of strains was reproducibly transform- be valuable as a platform strain for the production of energy able. These strains were subsequently tested in lab-scale dense biofuels. Heterologous pathways will be installed fermentations, resulting in one strain that performed best to redirect carbon flux towards cytosolic acetyl-CoA, the and was selected for further studies. For this strain, the precursor for synthesis of fatty acids. Furthermore, enzymes transformation protocol was optimized and its genome was involved in producing fatty acids from acetyl-CoA will be sequenced. We have developed an advanced genetic tool- overexpressed in order to create a “pull”, which poten- box for this strain which allowed us to make multiple clean tially can increase flux through the heterologous pathways. gene deletions using a counter-selection system, as well as Integrative analysis of multi-omics data will be performed overexpress genes from a plasmid. As the wild type strain to characterize the established strains, and the obtained re- produces mainly lactate, the first target was to remove the sults will in combination with predictive computer modeling ldh-gene, which resulted in a viable mutant. Results of the act as tools to identify important metabolic flux controlling metabolic engineering of multiple knockouts and overexpres- factors. Such insights could potentially be used for opti- sion of genes in central metabolism, as well as biochemical mization of the created system or implementation of other analyses of related enzyme activities will be discussed. biosynthetic pathways.

13. Isolation, Characterization and Metabolic 14. Controlled Protein Degradation for Development of Engineering of a Thermophilic Bacillus for Green Metabolite Valves Chemical Production Irene Brockman* and Kristala Prather Elleke F. Bosma*1, Antonius H.P. van de Weijer1, John van Chemical Engineering, Massachusetts Institute of der Oost1, Willem M. de Vos1 and Richard van Kranenburg1,2 Technology, Cambridge, MA 1Laboratory of Microbiology, Wageningen University, Wageningen, Netherlands The expression of heterologous pathways in microbial hosts 2 Corbion, Gorinchem, Netherlands offers interesting opportunities for production of both com-

modity chemicals and specialty compounds. Gene knockouts Biomass is an attractive renewable resource as alternative are often employed to redirect flux into these pathways, but to fossil-fuel-based chemicals for the production of green in cases where the competing enzymes are part of central chemicals such as organic acids that can be processed into metabolism of the host, this may result in poor cell health and for example bio-plastics. Major challenges for the transition slow growth on the desired substrate. We would like to de- to a bio-based economy are to develop cost-effective pro- velop strategies for dynamically modulating the abundance of duction processes and to avoid the use of food substrates. native enzymes within the host cell, allowing for switching be- Therefore, we are developing a new generation of microbial tween growth and production modes. One system that could platform organisms based on moderately thermophilic, fac- potentially benefit from implementation of such a “metabolite ultatively anaerobic Bacilli. Industrial advantages compared

9 Poster Abstracts

valve” strategy is the pathway for production of glucaric acid the network graph. The developed tools exploit OptFlux’s in E. coli previously developed in our lab. The initial substrate capabilities in terms of model interaction, simulation meth- for the glucaric acid pathway, glucose-6-phosphate, is utilized ods and visualization features. by the cell in both glycolysis and the pentose phosphate pathway, and static knockout strategies involving these The first proposed software, named MetabolIc NEtwork central metabolic pathways can be detrimental to growth and Ratio AnaLysis (MiNeRAl) aims at analyzing labeling experi- recombinant protein expression. ments to infer flux constraints that for stoichiometric models.

From a set of measurements of a 13C-labelling experiment, To develop the metabolite valve concept for this system, we mass isotopomer distribution vectors (MDV) are calculated. If have implemented a strategy based on controlled degrada- aminoacids are measured, the measured fragments, coupled tion of a key glycolytic enzyme, phosphofructokinase. Control with a carbon transition map provided by the user, are used of enzyme abundance at the post-translational level through to determine their precursors, and the corresponding MDVs degradation allows for rapid response time and can help over- are calculated. Based on the set of MDVs, the software come growth-mediated buffering effects seen with transcrip- uses the carbon transitions to determine the flux ratios that tional control. Combining this with tuning of gene expression produce a given metabolite through the different pathways. levels, we have been able to develop an E. coli strain with These ratios are probabilistic equations that translate how the

a “growth mode” very close to wild type and a “production 13C-labeling pattern is distributed throughout the metabolic mode” with decreased glycolytic flux. Ongoing work focuses network2. Since the calculation of the flux ratios is indepen- on induction of the system in response to culture conditions to dent of the flux distribution, this software can be used inde- allow for autonomous switching. pendently of other flux calculation processes, and the ratios can be further exploited to reduce the degrees of freedom 15. Flexible and User Friendly Tools for the Incorporation of systems obtained in other MFA approaches 3, 4. The main of Fluxomics Data into Metabolic Models differentiating characteristics of this tool are, besides being Rafael Carreira*1, Marcellinus Pont2, Jean-Francois Tomb2, usr-friendly, the fact that it is generic for any type of metabo- 3 1 1 Silas G. Villas-Bôas , Miguel Rocha and Isabel Rocha lite fragmentation originating from GC-MS techniques and 1Centre of Biological Engineering, Department of Biological metabolic network topology. Furthermore, the software is Engineering, University of Minho, Braga, Portugal 2E.I. du Pont de Nemours and Company, DE also able to investigate what flux ratio constraints are pos- 3Centre for Microbial Innovation, School of Biological sible to be inferred for a certain experiment beforehand. Sciences, University of Auckland, Auckland, New Zealand On the other hand, the second software application here The measurement of fluxes and the understanding of their described, jMFA, is focused on using different types of control are at the core of Metabolic Engineering (ME). In experimental flux data to constrain metabolic models and this context, this work presents two integrated opensource improve their predictions with a variety of tools. It allows software tools that allow to perform tasks of metabolic flux users to define constraints associated with measured fluxes analysis (MFA). Both are platform independent, written in and/ or flux ratios, together with environmental conditions Java, and interact with the OptFlux framework [1], which also (e.g. media) and reaction/ gene knockouts. The applica- facilitates their communication. tion identifies the set of applicable methods based on the constraints defined from user inputs, allowing to select the OptFlux is a modular open-source software that incorporates desired approach, encompassing algebraic and constraint- tools for strain optimization, i.e., the identification of ME based simulation methods (such as Flux Balance Analysis targets. It also provides tools to use stoichiometric meta- and its variants). Anytime a set of constraints is selected, the bolic models for phenotype simulation of both wild-type and software calculates the degrees of freedom of the configured mutant organisms, using methods such as the well known system, and updates the admissible methods depending Flux Balance Analysis (FBA). Graphical user interfaces are on whether the system is underdetermined, determined or made available for every operation and to check the results overdetermined. A method to perform robustness analysis is that are obtained. Moreover, a network visualization system also implemented. The integration of jMFA within the OptFlux is offered, where simulation results can be added to overlap framework allows the use of different model formats and

10 Poster Abstracts

the integration with complementary methods for phenotype tic acid. For this objective, first we developed a consistently simulation and visualization of the results. Moreover, the flux reduced core model (279 metabolites and 382 reactions) for S. ratio constraints can be obtained from previous calculations cerevisiae, based on the recent genome scale model. Sub- in MiNeRAl, or manually defined by the user. The first option sequently, we integrated thermodynamic and experimentally provides a straightforward way to integrate both applications measured information about the metabolites concentrations in a ME workflow. and reaction fluxes, to identify thermodynamically feasible References operational configurations of the network under different experimental conditions using the novel Flux Directionality Profile [1] Rocha, Isabel, et al. “OptFlux: an open-source software plat- Analysis (FDPA) technique. We then used the ORACLE frame- form for in silico metabolic engineering.” BMC systems biology 4.1 (2010): 45. work, to develop sets of log-linear kinetics models, which are stoichiometrically, thermodynamically, kinetically and physi- [2] Sauer, U. W. E., et al. “Metabolic flux ratio analysis of genetic and environmental modulations of Escherichia coli central carbon ologically consistent, to explore the flexibility and robustness metabolism.”Journal of bacteriology 181.21 (1999): 6679-6688. of the operational states. Analysis of the models allowed us to [3] Zamboni, Nicola, Eliane Fischer, and Uwe Sauer. “FiatFlux–a identify: 1) the differences of the flux profiles between differ- software for metabolic flux analysis from 13C-glucose experi- ent doses of acetate during ethanol production in the CEN. ments.” BMC bioinformatics6.1 (2005): 209. PK strain; and 2) the optimal strategies to improve ethanol [4] McAnulty, Michael J., et al. “Genome-scale modeling using flux production under these different conditions. ratio constraints to enable metabolic engineering of clostridial metabolism in silico.”BMC systems biology 6.1 (2012): 42. 17. Understanding and Optimizing Free Fatty Acid Production in Synechocystis Sp. PCC 6803 16. Unraveling the Inhibitory Effects of Acetate on Yi Ern Cheah* and Christie A.M. Peebles Ethanol Production in Cen.PK Chemical and Biological Engineering, Colorado State Anirikh Chakrabarti*1,2, Alexandros Kiparissides1,2, Keng University, Fort Collins, CO Cher Soh1,2, Bart R. B. H. van Rijsewijk3, Jason W. Hickman3, Tarek S. Najdi3, Dan Halim3, Vasiliy A. Portnoy3, Matthew A. Cyanobacteria are attractive systems for metabolic engineer- 3 3 3 Lorenz , Rogelio Oseguera , Adam M. Burja , Ljubisa Mis- ing due to the relative ease of genetic manipulation and their kovic1,2 and Vassily Hatzimanikatis1,2,4 potential of producing a variety of valuable molecules directly 1Laboratory of Computational Systems Biotechnology, EPFL, Lausanne, Switzerland from carbon dioxide. So far, engineering pathways in cyano- 2Swiss Institute of Bioinformatics, Lausanne, Switzerland bacteria has been a challenge due to our limited understanding 3BP Biofuels, Global Technology Centre (GTC), of the function and regulation of their genetic components. San Diego, CA On-going research in our lab focusses on the development and 4 Institute of Chemical Sciences and Engineering, Ecole characterization of molecular biology tools primarily for use in Polytechnique Fédérale de Lausanne (EPFL), Lausanne, the cyanobacterium Synechocystis sp. PCC 6803. Switzerland

Saccharomyces cerevisiae, besides being a model eukaryotic As a proof of concept, we applied our knowledge to manipu- organism, is one of the most commonly used yeast species late the fatty acid synthesis pathway of Synechocystis sp. for production of bioethanol and other biofuels. Even though PCC 6803. Based on what we know about fatty acid synthe- S. cerevisiae is a natively proficient ethanologenic yeast sis in E.coli, we constructed a synthetic metabolic sink by strain, the presence of weak acid inhibitors, such as acetic expressing genes leading to free fatty acid (FFA) production acid, found in cellulosic feedstocks for 2nd generation bio- in Synechocystis sp. PCC 6803. We then use this strain as ethanol production has detrimental effects on its productivity. a platform to demonstrate the use of our metabolic tools to study and optimize for FFA production. This presentation will In this study, we used the ORACLE (Optimization and Risk focus on our research efforts to date. Analysis of Complex Living Entities) framework to unravel the 18. Novel Acetyl-CoA Transfer Route in Saccharomyces metabolic impact of extracellular acetic acid on S. cerevisiae cerevisiae metabolism in order to generate metabolic engineering targets Yun Chen*, Yiming Zhang and Jens Nielsen for improving ethanol production rates in the presence of ace- Chemical and Biological Engineering, Chalmers University of

11 Poster Abstracts

Technology, Göteborg, Sweden fatty acyl-CoA synthetase, Clostridium acetobutylicum fatty acyl-CoA reductase and Arabidopsis thaliana fatty aldehyde Acetyl-coenzyme A (acetyl-CoA) is not only an essential in- decarbonylase. The final engineered strain produced up to termediate in central metabolism, but also an important pre- 580.8 mg l-1 of SCAs consisting of nonane (327.8 mg l-1), cursor for a range of commercially interesting chemicals that dodecane (136.5 mg l-1), tridecane (64.8 mg l-1), 2-methyl- can be used as pharmaceuticals, chemical building blocks, dodecane (42.8 mg l-1) and tetradecane (8.9 mg l-1) togeth- or biofuels. In Baker’s yeast Saccharomyces cerevisiae, er with small amounts of other hydrocarbons. [This work acetyl-CoA is compartmentalized and not direct interchange- was supported by the Advanced Biomass Research and able between the different compartments. Using either the Development Center of Korea (ABC-2010-0029799) through acetyl-carnitine or glyoxylate shuttle, acetyl-CoA produced in the Global Frontier Research Program of the Ministry of Sci- peroxisomes or the cytoplasm can, however, be transported ence, ICT and Future Planning (MSIP) through the National into the cytoplasm or the mitochondria, respectively. How- Research Foundation (NRF). Systems metabolic engineer- ever, whether acetyl-CoA generated in the mitochondria can ing work was supported by the Technology Development be exported to the cytoplasm is still unclear. In the current Program to Solve Climate Changes on Systems Metabolic study we found a mechanism for the transfer of acetyl-CoA Engineering for Biorefineries (NRF-2012-C1AAA001- from mitochondria to the cytoplasm. We further character- 2012M1A2A2026556) by MSIP through NRF]. ized this novel metabolic route using genetic and biochemical approaches. The results will increase our fundamental 20. High-Throughput Screening System and Its understanding of intracellular transport of acetyl units, and Biotechnological Applications hereby enable development of better microbial cell factories Jong Hyun Choi*, Kyong-Cheol Ko, YunJon Han, Dae-Eun that can be used for production of many kinds of acetyl-CoA Cheong, Jin Sun Kim and Jae Jun Song derived products. Korea Research Institute of Bioscience and Biotechnology (KRIBB), South Korea 19. Microbial Production of Short-Chain Alkanes Novel enzyme and/or metabolic pathway screening tech- So Young Choi*, Yong Jun Choi and Sang Yup Lee niques from various environmental resources and its biotech- Chemical & Biomolecular Engineering, Korea Advanced nological applications have widely expanded in the past few Institute of Science and Technology (KAIST), Daejeon, years. However, it is difficult to screen new enzymes due to South Korea limited assay protocol, expensive substrate, library, low sensitivity, low activity of target enzymes in host strain, and time- Our increasing concerns on limited fossil fuels and global consuming and laborious process. Therefore it was difficult environmental problems are urging us to develop sustain- for mass screening of target enzymes from various resources. able biofuels from renewable resources. Although micro- We had been developed the HTS system for enzyme screen- bial production of diesel has been reported, production of ing from various environmental resources including metage- another much demanded transport fuel, gasoline, has not nome. This HTS system composed with Robotic Integration yet been demonstrated. Here we report the development of System, JANUS Liquid Handling System, VICTOR3 Multilabel platform Escherichia coli strains that are capable of produc- Microplate Reader, K3 Colony Picking System, Automated ing short chain alkanes (gasoline), free fatty acids (FFAs), Incubator, Carousel, Lidding/Delidding System, Bar Code fatty esters, and fatty alcohols through the fatty acyl-[acyl Reader, FACSAria High-speed Flow Cytometer. Finally opti- carrier protein (ACP)] to fatty acid to fatty acyl-CoA pathway. mized screening methods for high throughput system led to The β-oxidation pathway was blocked by deleting the fadE over 10 clone active verification per day were available. gene to prevent the degradation of fatty acyl-CoAs gener- 5 ated in vivo and the activity of 3-oxoacyl-ACP synthase (FabH), which is inhibited by unsaturated fatty acyl-ACPs, In this study, we developed mass screening methods for vari- was enhanced to promote the initiation of fatty acid biosyn- ous enzymes from metagenomic libraries using HTS system thesis by deleting the fadR gene. A modified thioesterase based on robot. We could new enzymes such as cellobiohy- was employed to convert short chain fatty acyl-ACPs to the drolases, glycosyltransferases, BVMO, cold-adapted ester- corresponding FFAs, which were consequently converted ase and deoxyribose 5-phosphate aldolase(DERA) based to short chain alkanes by the sequential reactions of E. coli on fluorescence intensity. These new strategies combined

12 Poster Abstracts

with HTS system could screen various new enzyems/meta- Seoul, South Korea bolic pathways more fast, sensitive and easy than previously reported screening methods. This approach would be applied Phenol is an industrially versatile commodity chemical and, for other useful enzyme and metabolic pathway screening its annual production exceeded 8 million tons worldwide in from metagenomic resources. 2008. Currently, phenol is produced via the chemical oxidation of cumene derived from benzene. Due to our increasing 21. Overcoming Inefficient Cellobiose Fermentation By concerns on global warming and depletion of fossil resources, Cellobiose Phosphorylase in the Presence of Xylose biological production of phenol has been attracting much Kulika Chomvong*1, Vesna Kordic2, Xin Li3, Stefan Bauer4, attention. However, phenol’s biological production from Abigail E. Gillespie3, Suk-Jin Ha5, Eun Joong Oh5, Jonathan renewable resources has been limited due to its toxicity to 3 5 3 M. Galazka , Yong-Su Jin and Jamie H.D. Cate microorganisms and complex biosynthetic network of aro- 1Plant and Microbial Biology, University of California at matic compounds. To address these issues, we simultane- Berkeley 2Chemistry, University of California at Berkeley ously engineered 18 Escherichia coli strains for the production 3Molecular and Cell Biology, University of California at of phenol by taking advantage of synthetic regulatory sRNA Berkeley technology. sRNA-based knock-down of the two regulators 4Energy Biosciences Institute and overexpression of the genes involved in the tyrosine bio- 5 Food Science and Human Nutrition, University of Illinois synthetic pathway together with tyrosine phenol-lyase in E. coli at Urbana-Champaign, Urbana, IL strains resulted in the production of phenol from glucose. The

18 engineered E. coli strains showed significant differences in Cellobiose and xylose co-fermentation holds promise for the production of tyrosine (the immediate precursor for phenol), efficiently producing biofuels from plant biomass. Cellobi- tyrosine phenol-lyase activity, and tolerance to phenol. These ose phosphorylase (CBP), an intracellular enzyme generally led to much variation in their phenol-producing capabilities. found in anaerobic bacteria, cleaves cellobiose to glucose Among the engineered E. coli strains, the BL21(DE3) strain and glucose-1-phosphate, providing energetic advantages produced phenol most efficiently: 419 mg/L by flask culture under the anaerobic conditions required for large-scale and 1.69 g/L by fed-batch culture. The final titer and productiv- biofuel production. However, its efficiency to ferment cel- ity were further improved through biphasic fed-batch fermenta- lobiose in the presence of xylose is unknown. This study tion using glycerol tributyrate as an extractant of phenol. The investigated the effect of xylose on anaerobic CBP-mediat- concentration of phenol in the glycerol tributyrate phase and ed cellobiose fermentation by Saccharomyces cerevisiae. fermentation broth reached 9.84 and 0.3 g/L, respectively, in Yeast capable of fermenting cellobiose by the CBP pathway 21 h, which translates into the final phenol titer and productivity consumed cellobiose and produced ethanol at rates 59% of 3.79 g/L and 0.18 g/L/h, respectively. This is the highest titer and 47% slower, respectively, in the presence of xylose than achieved by microbial fermentation. Although further engineer- in its absence. The system generated significant amounts ing is required to be competitive with the current petro-based of the byproduct 4-O-β-d-glucopyranosyl-d-xylose (GX), process, the strategies used for the development of the engi- produced by CBP from glucose-1-phosphate and xylose. neered strain and fermentation process will provide a valuable In vitro competition assays identified xylose as a mixed- framework for the microbial production of toxic chemicals. inhibitor for cellobiose phosphorylase activity. The negative [This work was supported by the Intelligent Synthetic Biology effects of xylose were effectively relieved by efficient cellobi- Center (2011-0031963) through the Global Frontier Research ose and xylose co-utilization. Alternatively, GX cleavage can Program of MEST.] be carried out by an intracellular β-glucosidase.

22. Production of Phenol from Glucose in Escherichia 23. Towards Synthetic Phototrophy: Engineering Proton- coli through Metabolic Engineering Approach Pumping Rhodopsins into E. coli Hannah Chung*1, Byoungjin Kim1, Hyegwon Park1, DoKyun Nico J. Claassens*1, Michael Volpers2,3, Vitor A. P. Martins Na2 and Sang Yup Lee1 Dos Santos2, John van der Oost1 and Willem M. Vos1,4 1Chemical & Biomolecular Engineering, Korea Advanced 1Laboratory of Microbiology, Wageningen University, Institute of Science and Technology (KAIST), Daejeon, Wageningen, Netherlands South Korea 2Laboratory of Systems and Synthetic Biology, Wageningen 2School of Integrative Engineering, Chung Ang University, University, Wageningen, Netherlands

13 Poster Abstracts

3Center for Biological Systems Analysis, University of 2Industrial Microbiology, Dept. of Biotechnology, TU Delft, Freiburg, Freiburg, Germany Julianalaan 67, 2628BC Delft, The Netherlands 4Department of Bacteriology and Immunology, Helsinki University, Helsinki, Finland In an industrial setting, anaerobic conditions are especially attractive as costs related to energy input for mixing and oxygen Light-driven, microbial production of biochemicals and transfer are drastically reduced. In addition, highest theoretical biofuels is an interesting alternative to fossil resource-based product yields can be achieved. Synthetic biology nowadays production. However, non-photosynthetic microorganisms enables the introduction of extensive changes in the metabolic are often better characterized and more suitable for metabol- capabilities and regulation of microorganisms. ic engineering than natural photosynthetic microorganisms. Hence, we aim to develop synthetic phototrophic microor- The budding yeast Saccharomyces cerevisiaeis well known for ganisms by introducing photosystems into non-photosyn- its high fermentative capacity and robustness in large-scale thetic, microbial hosts. fermentation processes. However, it has not yet been used for industrial amino acid production. The aim of our project is to The photosystem we focus on is the family of proton-pumping reprogram yeast metabolism to achieve efficient amino acid rhodopsins (PPRs) [1]. A wide range of these PPRs have synthesis under anaerobic conditions. recently been discovered in nature. All these PPRs are trans- membrane proteins harbouring a retinal pigment. PPRs are Current processes rely on aerobic cultivations, using bacteria simple photosystems that can easily be introduced in non-pho- mainly C. glutamicum and E. coli. To reach anaerobic produc- tosynthetic hosts. PPRs generally use light energy to generate tion of amino acids with sufficient yield, synthetic biology is an outward proton flux, and the resulting proton motive force applied to reprogram yeast central carbon metabolism, espe- can subsequently generate ATP. This light-generated ATP can cially to obtain: be used to boost ATP-consuming production pathways or CO 2 1) Redox-balanced production pathways. fixation cycles. Recently, the introduction of PPRs in microbial 2) Biosynthetic pathways including the export have to produce production hosts, such as E. coli, has successfully led to some free energy in the form of ATP. light-driven biotechnological conversions. 3) The export mechanism has to be energy-efficient and thermodynamically favorable. In our research, the model bacterium E. coli is the host of 4) Other fermentative pathways have to be eliminated. interest for establishing PPR photosystems and light-driven 5) Intracellular accumulation in compartments has to be metabolic production pathways. We focus on two main avoided for an efficient transport to the extracellular space. aspects of PPR-based phototrophy in E. coli: (i) optimizing expression of different PPR photosystems including pig- To evaluate the success of genetic modifications, extra- and ment biosynthesis, and (ii) application of the PPR-generated intracellular metabolite measurements are performed. Analyti- proton gradients to drive production pathways or CO fixation 2 cal protocols for the quantification of pathway intermediates cycles. We use an integrated experimental and genome-scale have been developed (GC- and LC-MS). Thermodynamic modelling approach to investigate the potential of PPRs analysis is performed to identify limiting reaction steps. to boost metabolic pathways in E. coli. Development of a synthetic phototrophic E. coli is a promising development Homoalanine fermentation is a straightforward pathway that towards a useful photosynthetic, microbial production host. requires the expression of one gene: Alanine dehydroge- 1Claassens, N.J. et al. (2013) Potential of proton-pumping rhodopsins: engineering photosystems into microorganisms. Trends nase NADH-dependent. The developed strain expresses the Biotechnol. 31, 633–42 enzyme Alanine Dehydrogenase (ADH) of Bacillus subtilis. However, no difference in amino acid production could be 24. Engineering Anaerobic Amino Acid Production in observed between mutant and wild-type strain. No accumu- Saccharomyces cerevisiae: Alanine As Case of Study lation in the intra- and extracellular space was observed in Hugo Federico Cueto-Rojas1, Anisha Goel1, Nick Milne2, glucose-limited anaerobic chemostats. Two hypotheses were 2 1 1 Jean Marc Daran , J.J. Heijnen , Aljoscha Wahl formulated to explain these results: 1) alanine excretion is the 1Cell Systems Engineering, Dept. of Biotechnology, TU Delft, limiting step of the production process and 2) as ammonium Julianalaan 67, 2628BC Delft, The Netherlands 14 Poster Abstracts

is used as N-source no ATP is produced in the homoalanine symport mechanism has to be engineered to achieve higher fermentation pathway, due to the fact that 1 mole ammonium extracellular alanine concentrations. requires 1 mole ATP for the excretion of the charge imported with the ammonium ion and only 1 ATP is produced per mole Furthermore, bioprocess optimization requires a thorough of alanine (zero in total). characterization of transport processes of both product and substrate. Characterization of the ammonium transporters Hess et al. (2006) observed excretion of amino acids when requires accurate data of the intra- and extracellular space. yeast cells are cultured at high ammonium and low potas- An enzymatic reaction kit (Roche) was modified a) to be used sium concentrations. Potassium (K+) and ammonium (NH4+) in 96-well format for high throughput measurements and b) have a similar ionic radius and the authors suggested that this to increase its sensitivity for low ammonium concentrations leads to unspecific import of NH4+ by K+ channels (especially (between 5 µmol/l to 300 µmol/l). With the developed assay, at very low K+concentration). it was possible to measure intra- and extracellular concentrations of ammonium in S. cerevisiae CENPK 113-7D under Accumulation of NH4+ in the intracellular space is toxic aerobic nitrogen-limited conditions. The intracellular concen- and cells have to export ammonia in order to maintain the tration found was 1.36 +/- 0.07 mmol/lIntracellular at an extracel- intracellular homeostasis. Several mechanisms are possible: lular residual concentration of 61.0 ± 8.1 µmol/l. The measured urea production, overproduction of amino acids or excretion ratio (22.95) is within the expected range (between 4.9 and of ammonium using active transport. To this date neither 67.1) according to thermodynamic calculations assuming + urea synthesis nor NH4 active transporters have been a pmf between -150 mV and -190mV, and intracellular pHi reported for S. cerevisiae (Hess et al., 2006); which supports between 6.5 and 6.8. To our knowledge, this is the first time the hypothesis of overproduction of amino acids as defence that intracellular concentrations of ammonium are measured mechanism against high concentrations of ammonium in the directly in vivo. intracellular space. References: Alberty RA (2003). Thermodynamics of biochemical reactions. Using an anaerobic glucose-limited chemostat with 500 Hoboken, NJ: Wiley Interscience. mmol/l ammonium and 2.5 mmol/l potassium (as suggested Duarte N, et al. (2004). Genome Res. 14: pp. 1298-1309 by the work of Hess et al.) the alanine excretion was in- Hädicke O, Klamt S (2010) J Biotechnol., 147 pp. 88-101 creased from 8.2 mmol/Cmol X/h to 178.9 mmol/Cmol X/h in the WT strain and from 14 mmol/Cmol X/h to 274 mmol/ Hess DC, et al. (2006). PLoS Biology 4:11 pp. 2012-2023 Cmol X/h in the recombinant strain. Other differences were Ikeda M, (2003) Adv. Biochem. Eng. Biotechnol. 79: pp.1-35 observed in biomass and glycerol production. Ishimoto, M., et al. (2012). Biosci. Biotechnol. Biochem., 76:9 pp.1802-1804 Other experiments using high-ammonium low potassium Marini et al. (1997) Mol. Cell. Biol., 17 pp.4282-4293 conditions showed intracellular alanine concentrations as high Nookaew I., et al.(2008). BMC Systems Biology 2: 7 as 100 mmol/lintracellular. These results suggest that the limiting Rocha M., et al. (2008). BMC Bioinformatics 9:499 process in L-alanine production seems to be the transport from the cytosol to the extracellular space. Sekito T., et al. (2008). IUBMB Life 60:8 pp. 519-525

It is assumed that the measured extracellular alanine is ex- 25. Cytosolic Acetyl-CoA Platform in Yeast for ported via the reversed activity of the alanine importer proteins Biochemicals Production Zongjie Dai*, Yiming Zhang, Anastasia Krivoruchko, Verena (symporters of one molecule of alanine and one proton). Using Siewers and Jens Nielsen this mechanism, the intracellular concentration is expected Department of Chemical and Biological Engineering, Chalm- to be 1440 times higher than the extracellular concentration ers University of Technology, Gothenburg, Sweden at equilibrium conditions. Experimental observations suggest that L-ala concentration in the intracellular space is around Acetyl-CoA is not only a key precursor for cell metabolism, 20 mmol/lintracellular; being at equilibrium with an extracellular but also is the important intermediate for production of bio- concentration close to 14 µmol/lextracellular. The unfavourable fuels and biochemicals. Saccharomyces cerevisiae is a very

15 Poster Abstracts

important cell factory and has been widely used for many cost-efficient route to sclareol and other diterpene analogues. biomolecules production, especially for bioethanol production. However, the synthesis of cytosolic acetyl-CoA is not 27. Advanced Production of Faee in a S. cerevisiae efficient enough for industrial application due to the organelle Cell-Factory compartmentalization and hallmarks of yeast metabolism. Bouke Wim de Jong*1, Shuobo Shi2, Verena Siewers1 and 1 In yeast, part of pyruvate was converted to acetyl-CoA in Jens Nielsen * 1Chemical and Biological Engineering, Chalmers University mitochondria by pyruvate dehydrogenase and pyruvate of Technology, Gothenburg, Sweden decarboxylase catalyzed major pyruvate to acetaldehyde for 2Metabolic Engineering Research Laboratory, Institute of ethanol and acetyl-CoA production in cytoplasm. In order Chemical Engineering and Sciences, Singapore, Singapore to eliminate ethanol accumulation, endogenous pyruvate decarboxylases (Pdc1p, Pdc5p, Pdc6p) were firstly deleted Due to an increasing demand of transportation fuels and a creating mutant strain S. cerevisiase E1. However, strain lower availability of crude oils, a gradual shift from oil based E1 could not grow in glucose medium due to no cytosolic fuels towards alternative and renewable fuel resources will be acetyl-CoA produced for cell metabolism. Base on this required in the near future. In addition, currently existing re- platform, a heterogenous combination module was secondly newable alternatives, bioethanol and conventional biodiesel, introduced into this strain. In this module, pyruvate oxidase cannot cover the increasing demand for biofuels and there and phosphate acetyltranferase were coexpressed in cyto- is therefore a need for new biofuels, like advanced biodiesel, plasm to convert pyruvate to acetyl-CoA. Growth of E1 strain with superior fuel properties. Novel yeast-cell factories will in glucose medium was rescued by the modifications, which provide production platforms for advanced fuels. Deep cel- indicate that this pathway could convert pyruvate to cyto- lular understanding and the use of an integrated systems ap- solic acetyl-CoA for cell growth. In addition, there was no proach are driving the development of several new process- ethanol be detected in culture medium. With this cytosolic es. Here, metabolic pathways leading towards the metabolite acetyl-CoA producing platform, a new non-ethanol produc- acetyl-CoA are crucial pathways for the production of many ing yeast cell factory could be constructed to produce many promising renewable biofuels and chemicals1. industrially relevant chemicals. In the first part of this project, Saccharomyces cerevisiae is

metabolically engineered to increase the internal carbon flow 26. Toward a Biosynthetic Route to Sclareol and towards acetyl-CoA and the final biodiesel product, fatty Amber Odorants acid ethyl esters (FAEEs). Different metabolic pathways were Laurent Daviet* compared: In the first pathway endogenous alcohol and Biotechnology R&D, Firmenich SA, Geneva, Switzerland aldehyde dehydrogenase and a heterologous acetyl-CoA Ambergris, a waxy substance excreted by the intestinal synthetase were over-expressed together with a wax ester tract of the sperm whale, has been a highly prized fragrance synthase (WS) from Marinobacter hydrocarbonoclasticus, re- ingredient for millenia. Because of supply shortage and sponsible for FAEE synthesis. A second alternative pathway price inflation, a number of ambergris substitutes have been uses the heterologous phosphoketolase and acetate kinase, developed by the fragrance industry. One of the key olfactory gained from Aspergillus nidulans and several steps from the components and most appreciated substitutes of amber- pentose phosphate pathway (PPP) to reach the intermediate gris, Ambrox is produced industrially by semisynthesis from metabolite acetyl-CoA. Several strains were constructed and sclareol, a diterpene-diol isolated from Clary sage. In the resulted in a 3 fold improvement for the ethanol degradation present work, we report the cloning and functional charac- pathway (first strategy) and a 1.7 fold improvement for the terization of two diterpene synthases that act sequentially in PHK pathway (second strategy) times improvement of FAEE sclareol biosynthesis from the universal geranyl geranyl pyro- production in yeast2. phosphate (GGPP) precursor. Furthermore, we reconstructed A second part of the project involves the combination of the sclareol biosynthetic pathway in bacterial platform strains the up-regulated genes from the central carbon metabolism engineered to overproduce GGPP. Peak titers of 400 mg/L (described above) and up-regulation of acetyl-CoA car- and 1.5 g/L were obtained by cultivation in shake flasks and boxylase, acyl-CoA-binding protein and NADP-dependent bench-scale bioreactors, respectively. Our work provides a glyceraldehyde-3-phosphate dehydrogenase towards the basis for the development of an alternative, sustainable, and

16 Poster Abstracts

formation of free fatty acids and eventually FAEEs. Additional and taking advantage of the genome-scale science, i.e. ap- strategies for stable and reliable chromosomal integration of plying omics technologies for in depth physiological strain the pathways are presented. characterization and metabolic modeling for data analysis Reference: and strain design. 1. Bouke de Jong, Verena Siewers and Jens Nielsen; Systems biology of yeast: enabling technology for development of cell factories High overproduction of the triterpenoid will reduce down- for production of advanced biofuels, Curr Opin Biotechnol (2011), stream processing efforts and will allow the economic and doi:10.1016/j.copbio.2011.11.021 sustainable production of this promising compound. 2. Bouke Wim de Jong, Shuobo Shi, Verena Siewers and Jens Nielsen; Improved production of fatty acid ethyl esters in The project TRITERP is carried out in collaboration with the Saccharomyces cerevisiae through up-regulation of the ethanol Novo Nordisk Foundation Center for Biosustainability (CFB) degradation pathway and expression of the heterologous at the Technical University of Denmark and financed by the phosphoketolase pathway, Molecular Cell Factories, submitted. DBU – Deutsche Bundesstiftung Umwelt (Germany).

28. Metabolic Engineering of Cyclic Triterpenoid References Production in Saccharomyces cerevisiae Muffler K, Leipold D, Scheller MC, Haas C, Steingroewer J, Bley T, Birgitta E. Ebert*1, Kerstin Walter1, Christine Lang2 and Lars Neuhaus HE, Mirata MA, Schrader J, Ulber R. 2011. Biotransforma- M. Blank1* tion of triterpenes. Process Biochemistry 46(1):1-15. 1 iAMB - Institute of Applied Microbiology, RWTH Aachen Mullauer FB, Kessler JH, Medema JP. 2010. Betulinic acid, a natural University, Aachen, Germany compound with potent anticancer effects. Anti-Cancer Drugs 2 Organobalance GmbH, Germany 21(3):215-227. Triterpenoids are terpenoids derived from squalene and consist of six isoprene units (C30). These compounds can 29. Development & Exploitation of Gene Tools for Metabolic Engineering in Saccharolytic Clostridia be isolated from many different plant sources. They occur in countless variations and can be subclassified into several Muhammad Ehsaan*, Klaus Winzer and Nigel Minton, School of Life Sciences, The University of Nottingham, groups including squalenes, lanostanes, dammaranes, lu- Nottingham, United Kingdom panes, oleananes, ursanes, hopanes, cycloartanes, friedel- anes, cucurbitacins, and miscellaneous compounds (Mul- C. acetobutylicum ATCC 824 is a well characterized micro- lauer et al. 2010). Many of them or their synthetic derivatives organism known for its ability to produce solvents using the are currently being investigated as medicinal products for Acetone-Butanol-Ethanol (ABE) fermentation process. It can various diseases, including cancer. Despite their obvious utilize a variety of C5 and C6 sugars, but cannot directly interest for the industry, their wide applications are often access the complex lignocellulose plant cell wall material hindered by the presence of these compounds in only which is the most abundant source of carbon in nature. minute amounts in natural sources. This poses challenges in Sophisticated genetic tools are required to enhance the a biosustainable production of such compounds since per substrate utilisation ability of the organism by incorporat- gram active ingredient produced a high volume of solvent is ing synthetic operons using a synthetic biology approach. needed in the purification process. Efficient tools were developed for making precise alterations to the C. acetobutylicum genome using either heterologous Within the project TRITERP we establish a biotechnological pyrE or codA genes as counterselection markers. In the process for the production of betulinic acid, a cyclic triterpe- case of the former, the utility of the method was also demon- noid, using tailored Saccharomyces cerevisiae strains. The strated in Clostridium difficile. The robustness and reliability plant metabolite betulinic acid has antiretroviral, antimalarial, of the methods were demonstrated through the creation of and anti-inflammatory properties and has potential as an in-frame deletions in four genes (Cac2071 (spo0A), amylase, anticancer agent and is of high interest for the pharmaceuti- granulose glgA and Riboflavin operon) using pyrE and also cal and nutritional industry (Muffler et al. 2011). two genes (Cac1502 and Cac2071 (spo0A) using codA. The pyrE system is reliant on the initial creation of a pyrE dele- The strain re-engineering is accomplished using advanced tion mutant using Allele Coupled Exchange (ACE) that is and modern molecular biology and synthetic biology tools 17 Poster Abstracts

auxotrophic for uracil and resistant to fluoroorotic acid (FOA). 31. Targeted Proteomics Enabled Metabolic Engineering This enables the subsequent modification of target genes of Clostridium Cellulolyticum for n-Butanol Production by allelic exchange using a heterologous pyrE allele from C. Stefan M. Gaida*, Andrea Liedtke, Andreas H.W. Jentges sporogenes as a counter-/negative-selection marker in the and Stefan Jennewein presence of FOA. Following modification of the target gene, Industrial Biotechnology, Fraunhofer IME, Aachen, Germany the strain created is rapidly returned to uracil prototrophy The sustainable production of biofuels and chemicals require using ACE, allowing mutant phenotypes to be characterised the use of environmentally friendly carbon neutral feed- in a pyrE proficient background. Crucially, wild-type copies stock’s, which most importantly do not compete with food of the inactivated gene may be introduced into the genome production. An abundant and widely available source of using ACE concomitant with correction of the pyrE allele. such feedstock’s is lignocellulosic material, like agricultural This allows complementation studies to be undertaken at an residues, woody material like energy crops or municipal appropriate gene dosage, as opposed to the use of multi- solid waste (MSW). Unfortunately, these materials are highly copy autonomous plasmids. The rapidity of the ‘correction’ resistant towards microbial digestion as they are composed method (5–7 days) makes pyrE strains attractive hosts for of complex polymers of various compositions, e.g. cellulose mutagenesis studies. and hemicellulose. Common pretreatment processes to 30. Novel Methods to Investigate Solvent Toxicity in break down these polymers require harsh conditions leading Bacteria to the formation of inhibitory substances like HMF as well as Eugene Fletcher*1 and Chris French2 the addition of costly degrading enzymes. Clostridium cel- 1Systems and Synthetic Biology, Chalmers University of lulolyticum naturally excretes a cellulosome, a multi subunit Technology, Göteborg, Sweden enzyme complex able to degrade lignocellulosic compo- 2Synthetic and Systems Biology Centre (SynthSys), nents. Thus the cost and condition of the pretreatment University of Edinburgh, Edinburgh, United Kingdom process could be tremendously reduced by using C. cellulolyticum as a production organism for various compounds. Toxicity of organic solvents to microbial hosts is a major consideration in the industrial production of biofuels such As a proof of concept, we report on metabolic engineering as ethanol and especially butanol, with low product con- C. cellulolyticum to produce n-butanol, an advanced biofuel centrations leading to high recovery costs. Understanding and industrial important bulk chemical directly from crystal- the mechanisms involved in solvent tolerance is crucial for line cellulose. Also, great progress has been made towards rational engineering of robust microbes. We have developed developing better and novel genetic tools for Clostridia, a a bioluminescence assay to determine the effects of different still troublesome bottleneck is the functional expression genes on survival in four model inhibitors-ethanol, n-butanol, and regulation of heterologous genes. Moreover, fine tuning acetone and furfural. Adopting a synthetic biology approach, of engineered pathways is difficult, because little is known a library of potential solvent tolerance BioBricks (genes) was about genetic regulatory elements especially on the post- generated and tested as proof-of-concept. Using this meth- transciptional level. Only a few promoters are currently used od, we have generated a set of tolerance modules suited for in engineering heterologous pathways, which are mostly these inhibitory compounds, which can be combined with constitutive. Here, we report on developing a method involv- genetic modules encoding substrate breakdown and product ing targeted proteomics combined with metabolite profiling formation pathways. Ultimately, we hope to generate im- to facilitate metabolic engineering of heterologous pathways proved biofuel-producing systems which can generate higher and fine tuning thereof. Combining the information of the product concentrations, greatly improving process econom- protein level (targeted proteomics) and the flux through the ics. Further tests carried out using enzyme and fluorescence- pathway (fluxomics), changes to regulatory elements in the based assays to characterise the effect of n-butanol on the engineered pathway towards transcription, translation or cell envelope showed that the alcohol released lipopolyss- post translation can easily be accessed with our newly de- sacharides from the outer membrane of E. coli and resulted in veloped method and used to fine tune the entire biosynthetic ‘leakiness’ in both outer and inner membranes. pathway. The information about the regulation of expression can furthermore be used in the metabolic engineering of ad-

18 Poster Abstracts

ditional heterologous pathways leading to the production of Reference: various desirable chemicals. Brat, D. et al., 2012. Cytosolic re-localization and optimization of valine synthesis and catabolism enables inseased isobutanol pro- 32. Isobutanol Production By an Industrial duction with the yeast Saccharomyces cerevisiae. Biotechnology Saccharomyces cerevisiae Strain for Biofuels, 5(1), p.65. Wesley Cardoso Generoso*, Heiko Dietz, Mislav Oreb and Eckhard Boles 33. Metabolic Engineering of Photorespiratory Bypass Institute of Molecular Biosciences, Goethe University Pathways to Enhance Novel Biofuel Production in Frankfurt am Main, Frankfurt am Main, Germany Transgenic Plants Sheba Goklany*1, Yong Kyoung Kim2, Hong Ma2, Yi Cheng 2 3 3 2 Among the next-generation biofuels, isobutanol has attract- Liu , Eiji Takahashi , Don Ort , Joshua Yuan and Joe Chappell1 ed research’s attention due to a superior combustion energy, 1Pharmaceutical Sciences, University of Kentucky, Lexing- a weaker corrosive action and a reduced aqueous miscibility. ton, KY Moreover, it can be separated from fermentation media by 2Plant Pathology and Microbiology, Texas A&M University, less intensive energy processes. Naturally, Saccharomyces TX cerevisiae is able to produce isobutanol. Basically, isobutanol 3Plant Biology, University of Illinois at Urbana-Champaign, IL can be formed from pyruvate via the L-valine biosynthesis (Ilv2, Ilv5, Ilv3) and degradation (Aro10, Adh2) pathways. Industrialization and the global population growth rate have Nevertheless, due to a low yield in the natural occurring placed a tremendous burden on food, energy, and other isobutanol production, this pathway has been focus of opti- natural resources, including land and water. The current mizations. In our previous results, the first three enzymes of world population of ~7.1 billion is expected to increase to 9.6 valine biosynthesis were overexpressed as cytosolic iso- billion by 2050, while the global energy demand is anticipat- forms, in order to optimize the pathway flow from pyruvate. ed to increase more than 3-fold over the same time period. The cytosolic isobutanol pathway could rise the isobutanol In contrast, global oil production will peak before 2030, production to about 650 mg/L (Brat et al. 2012). This same leading to diminishing supply and increasing cost. Hence, strategy was incorporated into an industrial S. cerevisiae there is a clear and urgent need to harness the full potential strain (Ethanol Red), resulting in an isobutanol production of of sustainable energy sources and convert these efficiently to just 200 mg/L in 24 hours cultivation. Then, one allele of the resources required for human and industrial applications. PDC1 gene was replaced by a bacterial transhydrogenase (UdhA), for enhancing of the NADPH supply for Ilv5, as well The current project is addressing this worldwide challenge as Aro10 and Adh2, all under strong promoters. This strain by focusing on the development of plant systems to capture was able to produce 300 mg/L. Interestingly, high concen- sunlight energy into more direct biofuels rather than sugars trations of acetoin (200 mg/L), 2,3-butandiol (1.0 g/L) and or fatty acids requiring extensive downstream processing. dihydroxy-isovalerate (2.5 g/L) were found in the medium, Our work is also focused on remediating a longstanding inef- but no keto-isovalerate or isobutanal. Drawing on these ficiency in photosynthesis, the process known as photores- results, endorsed by a still high production of ethanol and piration. About 20% of the time, photorespiration rather than glycerol, Ilv3 could be assigned as one of the bottlenecks photosynthesis occurs wherein instead of CO2 being fixed of the isobutanol pathway. The dihydroxy-acid dehydratase and converted to precursors for carbohydrate biosynthesis,

(Ilv3) is a member of the ILVD_EDD protein family, which is O2 is condensed with RuBP to yield glycolate. The glyco- known to comprise iron-sulfur cluster enzymes. However, late is subsequently decarboxylated in an energy intensive further overexpression of either a cytosolic or a mitochondri- process resulting in the net loss of one CO2 returned to the al isoenzyme did not increase isobutanol production. These atmosphere. results show us that the primordial bottleneck surround the low activity of Ilv3 can be probably an inefficient build of the While attempts to engineer the oxygenation reaction out of iron-sulfur cluster. the RuBP Carboxylase/Oxygenase have been unsuccessful, attempts to recycle carbon in the photorespiratory glycolate Sponsored by CNPq, Brazil. have shown some promise. We too have pursued the latter strategy with the intent of recycling the carbon into the 19 Poster Abstracts

production of high-value, triterpene biofuels. Our focus on specific genes. For example, only in 2010, the genes coding production of linear, branched-chain hydrocarbon triterpenes for the central pyruvate transporter proteins were identified. is driven by the ease of their catalytic cracking into all class of fuels: gasoline, disease, and jet fuel. Chinese Hamster Ovary (CHO) cells are the major workhorse in biopharmaceutical production. They are also an important To evaluate various photorespiratory constructs, we first mammalian cell model system. To study metabolic activi- generated transgenic lines engineered for novel triterpene ties at the cytosol-mitochondria interface, we used a novel production in the chloroplast (G1 line). These transgenic systems-oriented approach combining different methods lines were then re-engineered with 3 different constructs and to elucidate rate limiting steps and robustness of metabo- the regenerated lines evaluated for triterpene accumulation lism. These methods were mainly dynamic metabolic flux by standard GC-MS analyses and for the operation of the analysis and the measurement of compartment-specific putative bypass pathways by measuring the incorporation of enzyme activities. Metabolic fluxes represent in vivo rates or radioactivity from glycolate into triterpenes. fluxes that can be related to maximum possible in vitro rates. Comparing these two values provides valuable information pT1 lines containing a partial bypass pathway showed a about metabolic control in a metabolic network. We ap- slight increase (33%) in the glycolate incorporation into trit- plied dynamic metabolic flux analysis using measurements erpenes as well as triterpene accumulation (44%) compared of biomass formation, extracellular carbohydrates, organic to the G1 control line. These levels were further enhanced acids and amino acids. Batch cultivation of CHO resulted in the pT3 lines incorporating a complete bypass pathway in different distinct growth phases with significantly differ- where higher levels of glycolate incorporation into triterpenes ent metabolic flux distributions. The first phase (phase I) (500%) correlated with higher triterpene accumulation (97%) was characterized by high glycolytic activity with secretion compared to the G1 controls. The highest triterpene lev- of lactate and consumption of pyruvate, glutamine, aspara- els (~2500 µg/g FW) were obtained in the pT5 lines, where gine, aspartate, serine and secretion of alanine and glycine. glycolate incorporation and triterpene accumulation were The ratio of lactate production to glucose consumption was enhanced by 1200% and 220%, respectively, compared to about 1.5 mol/mol and the ratio of TCA cycle flux to glycolyt- G1 controls. ic flux varied around a value of 1 mol/mol. After consumption of glutamine, metabolism switched to consumption of lactate Our data provides evidence for engineering known and novel and glutamate. The ratio of lactate consumption to glucose photorespiratory bypass pathways in plants for efficient consumption was now around a value of 1 mol/mol while the biofuel production. These metabolic engineering strategies TCA/glycolysis ratio stabilized at a value of about 3 mol/mol. demonstrate the use of sustainable energy to meet the grow- Hexokinase activities were low in phase I, only slightly higher ing need for fuels and renewable resources. than in vivo glycolytic fluxes indicating their rate controlling function. All other glycolytic fluxes were much smaller than 34. Metabolic Activities and Their Control at the corresponding in vitro activities. All glycolytic ratios increased Mitochondria-Cytosol Interface in CHO Cells significantly in phase II. TCA cycle was clearly limited by Elmar Heinzle*, Judith Wahrheit and Averina Nicolae mitochondrial isocitrate dehydrogenase with an in vitro/in Biochemical Engineering, Saarland University, Saarbrucken, vivo activity ratio of close to one. Interestingly, the fraction Germany of in vitro phosphofructokinase (PFK), that was measured after digitonin treatment leading to selective permeabiliza- Mitochondrial activity and its regulation are of major im- tion of the cell membrane but leaving mitochondrial mem- portance for eukaryotic cell factories as well as in many brane intact, increased with increasing cultivation time. This health-related aspects. Mitochondria play a central role in shows that a large fraction of PFK was initially quite strongly energy generation that is crucial for the energy-intensive attached to mitochondrial proteins thus blocking its active biosynthesis of proteins. The respiratory chain and oxidative center. Later, an increasing fraction was directly accessible af- phosphorylation require TCA cycle activity supplying NADH, ter digitonin treatment. This behavior is consistent with earlier succinate and ADP. Various transporters are involved in the indications of glycolytic substrate channeling. Flux analysis exchange of metabolites between cytosol and mitochondria. and compartment specific in vitro enzyme activities showed Interestingly, not all transport activities are clearly assigned to 20 Poster Abstracts

that in phase I mitochondrial pyruvate is primarily derived via a wide range of harsh chemicals. Last, but not least, many malate and alanine that are transported into the mitochondria LAB can us alternative carbon source of no use as food for and there converted to pyruvate. Glycolytic pyruvate primarily humans or as feed for domestic animals, opening for the reacts to lactate. This is supported but channeling of glyco- possibility of producing 2nd generation biofuels. lytic intermediates directly to lactate. Anaplerotic supply of C4 units is then accomplished by glutamine feeding into the A major problem facing the biotechnological exploitation of alpha-ketoglutarate pool via glutamate dehydrogenase. In LAB for the production of high value products such as higher phase II, lactate was taken up and converted to pyruvate that alcohols is related to the toxicity of these compounds, and was now directly transported to the mitochondrial matrix. The the lack of knowledge about their effects. Studies in LAB major anaplerotic reaction was now by transport of aspartate stress responses have until now focused on oxidative, os- into mitochondria and conversion to oxaloacetate. motic, nutritional, temperature, and pH stresses (Tsakalidou & Papadimitriou, 2011). Our findings show that there is a subtle control of glycolytic reactions and there connection to mitochondrial activities In this study, we have characterized the stress response of by channeling. Such findings were possible by a combined Lactococcus lactis subspecies cremoris strain MG1363 to- systems-oriented approach combining dynamic metabolic wards the exogenous short even-chain alcohols; ethanol, n- flux analysis and compartment specific enzyme activities. butanol and 1-hexanol. A detailed physiological, metabolic, We think that simultaneous reporter gene studies will help to and comparative transcriptional analysis, provided valuable further refine mechanisms of such type of control and later to insights into the stress response of LAB towards these alco- derive methods for intervention to improve energy metabo- hols and gave clues to potential targets for improvement of lism in cells producing recombinant proteins but also to cure L. lactis alcohol tolerance. We found that the stress response diseases related to such processes. towards the three alcohols showed both common response patterns as well as patterns that separated ethanol stress 35. General and Specific Stress Responses Towards from n-butanol/1-hexanol stress. Short Even-Chain Alcohols in Lactic Acid Bacteria Pro- vide Clues for Improving Second Generation Biorefineries Guillot, A., Gitton, C., Anglade, P., & Mistou, M.-Y. (2003). Proteomic Anne-Mette Hviid*1, Peter R. Jensen2 and Mogens Kilstrup1 analysis of Lactococcus lactis, a lactic acid bacterium. Proteomics, 1Institute for Systems Biology, Technical University of 3(3), 337–354. doi:10.1002/pmic.200390047 Denmark Tsakalidou, E., & Papadimitriou, K. (2011). Stress Response of 2National Food Institute, Technical University of Denmark Lactis Acid Bacteria.

Lactic acid bacteria (LAB) have the potential to become 36. Isotopically Nonstationary 13C Metabolic Flux efficient production organisms for large-scale bio-refinery Analysis of Arabidopsis thaliana Rosettes at Altered processes. LAB are currently used extensively by the dairy Light Conditions industry for fermentation of milk; thus their industrial impor- Lara J. Jazmin*1, Fangfang Ma2, Douglas K. Allen2,3 and 1 tance is indisputable. The biotechnological exploitation of Jamey D. Young 1Chemical and Biomolecular Engineering, Vanderbilt Univer- LAB has multiple benefits. Firstly, LAB possess high glycolyt- sity, Nashville, TN ic capacities, with up to 20% of the total proteome consist- 2Donald Danforth Plant Science Center, St. Louis, MO, ing of glycolytic enzymes, turning glycolysis into a metabolic 3USDA-ARS, St. Louis, MO highway (Guillot et al., 2003). Secondly, many LAB use a homofermentative metabolism, converting more than 90% Photoautotrophic metabolism represents the primary source of of their pyruvate into a single product, lactate. Re-routing of all food on earth as well as raw materials for bio-based produc- this major flux into alternative fermentation products, such as tion of fuels and chemicals. In particular, the ability to quantify small-chain alcohols opens for the opportunity of producing the flow and fate of carbon in biochemical networks using13 C such potential bio-alcohols at high yields. Since LAB primar- flux analysis is important for engineering desired metabolic ily produce lactate, they have evolved a tolerance towards phenotypes. Therefore, efforts to improve photosynthetic ef- acidic environments, through a multi-stress resistance ficiency would be aided by more quantitative descriptions of mechanism, resulting in slightly elevated resistance towards primary metabolism in plants. There are few comprehensive

21 Poster Abstracts

methods that quantitatively describe leaf metabolism, though of Technology, Gothenburg, Sweden such information would be valuable for increasing photosyn- 2LISBP - INSA de Toulouse, Toulouse, France thetic capacity, enhancing biomass production, and rerouting carbon flux toward desirable end products. In the last decades the environmental, production and social sustainability of the oil based industry has become increas- Isotopically nonstationary 13C metabolic flux analysis (INST- ingly questioned. Major efforts are hence being put into place MFA) has been previously applied to map carbon fluxes in to find alternative production methods for traditional petro- photoautotrophic bacteria, which involves model-based leum derived products. For the fuel sector several bio-alter- regression of transient 13C-labeling patterns of intracellular natives are being developed, however approximately 15 % of metabolites. We have recently conducted in vivo isotopic all oil is used to produce petrochemicals other than fuels and 13 bio-production methods of these are rare. One of the major labeling of Arabidopsis thaliana rosettes with CO2 at two differing light intensities of 200 (normal light) and 500 (high petrochemicals is ethylene (ethene), a hydrocarbon which light) µmol/m2/s, with and without prior acclimation, in order due to its chemical structure makes it susceptible to a wide to quantify fluxes through photosynthetic leaf metabolism. range of chemical conversions. In nature ethylene is most Using LC-MS/MS and GC-MS profiling techniques on isoto- known as a plant hormone. However, there are also microor- pically labeled intracellular metabolites, along with measure- ganisms which produce ethylene, foremost plant pathogens and soil bacteria. There are hence biological pathways which ments of net photosynthetic CO2 uptake and starch production rate, we have created comprehensive flux maps of the can be utilized for the production of ethylene. Through meta- central carbon metabolism in Arabidopsis rosettes under bolic modelling it was concluded that the bacterial pathway normal and high light conditions. employing the ethylene forming enzyme (EFE) is the most promising when S. cerevisiae is employed as the host organ- The resulting flux maps provide a subcellular compartmen- ism. The overall aim of this project was to elucidate the func- talized description of the carbon fixation pathways, which is tionality of this cell factory. In our previous study we showed the first application of INST-MFA to a terrestrial plant system that a good or increased provision of the EFE substrates in planta. The results suggest that subcellular compartmen- oxygen and 2-oxoglutarate is positive for ethylene produc- tation was necessary to include products from glucose- tivity. However, increased provision of the EFE-substrate 6-phosphate (i.e. starch and sucrose) that are spatially arginine contrary to our expectations reduced ethylene for- resolved in the chloroplast and cytosol. Photorespiration flux mation. To study the effect of arginine on the system further, we employ two metabolic strategies to reduce intracellular increased from 15% to 34% of net CO2 assimilation with increasing light, despite concomitant increases in carboxyl- arginine availability; overexpression of CAR1 and deletion of ation flux that led to more sucrose production. Interestingly, ARG4. Through metabolic network reconstruction of the EFE we observed an inverse relationship between intermediate reaction together with the S. cerevisiae metabolism and sub- pool sizes and Calvin cycle fluxes as light intensity increased. sequent flux balance analysis of the system, we have further Additionally, we identified enhanced hexose exchange be- identified several other targets predicted to increase ethylene tween the chloroplast and cytosol as a potential short-term formation. Three candidates from the list – ALT1, CIT1 and adaptation to high light that was suppressed by acclima- IDP2 – were overexpressed and their effect on ethylene pro- tion. This study shows that INST-MFA is a feasible approach ductivity in vivo evaluated. Our current research is aimed at to quantifying photosynthetic metabolism in Arabidopsis developing a deeper understanding of the enzyme functional- rosettes and gives the framework for comparison in future ity. Structure determination of the EFE has proven difficult, studies involving perturbed environmental conditions, as well however through sequence alignment of three EFE, of which as transgenics with improved photosynthetic efficiency. one had lost the ethylene formation capability, 10 amino acid residues emerged which are seemingly important for the 37. Biotechnical Production of Ethylene in S. cerevisiae capability of the enzyme to form ethylene. Taken together this — Insights from Metabolic Modeling, Cultivation Studies work elucidates the possibilities and challenges of turning S. and Enzyme Engineering cerevisiae into a cell factory for ethylene production. This re- Nina Johansson*1, Karl Persson1, Paul Quehl1, Gilles Vieira2, search is funded by the EU 7th Framework Programme under Stéphanie Heux2, Joakim Norbeck1 and Christer Larsson1 1Chemical and Biological Engineering, Chalmers University the grant agreement n°FP7-241566 - BIOCORE.

22 Poster Abstracts

38. Microbial Production of Cis,Cis-Muconic Acid By theoretical maximum. However, the growth rate of the pflB Klebsiella Pneumoniae mutant was slightly reduced compared to its parental strain. 1 2 1 Hwi-Min Jung* , June-Hyung Kim and Min-Kyu Oh Intracellular acetyl-CoA and redox balance was measured in 1Deparment of Chemical and Biological Engineering, Korea pflB deficient K. pneumoniae to find the reason of reduced University, Seoul, South Korea 2Department of Chemical Engineering, Dong-A University, cell mass production. Transcriptomic analysis with RNAseq Busan, South Korea was also conducted and the pflB mutation in K. pneumoniae significantly repressed the expression of genes involved in The microbial production of bio-based chemical becomes an the formate hydrogen lyase (FHL) system. Our results indi- important approach in the chemical industry. Here we report cate that the growth defect of pflB deficient K. pneumoniae the microbial production of cis,cis-muconic acid(CCMA) is primarily due to the reduction in the size of the acetyl-CoA that is the precursor of adipic acid and terephthalic acid. pool that redistributes the carbon flux to diol production. The CCMA was produced through the aromatic amino acid synthesis route in Klebsiella pneumoniae. This microorgan- 40. Bio-Hydrogen Production By Continuous Culture of ism has the genes related to the production of CCMA from Hyperthermophilic Archaeon from Carbon Monoxide 1 1 2 glucose as a carbon source, which are not normally activat- Tae Wan Kim* , Seung Seob Bae , Jeong Geol Na , Hyun Sook Lee1, Jung-Hyun Lee1 and Sung Gyun Kang1 ed in wild type condition. When we blocked the synthesis of 1Marine Biotechnology Research Division, Korea Institute of aromatic amino acid, Klebsiella pnuemoniae could produce Ocean Science and Technology, Ansan, South Korea not only CCMA but catechol that is the last intermediate of 2Biomass and Waste Energy Laboratory, Korea Institute of CCMA production pathway. After endogeneous catechol Energy Research, Daejeon, South Korea 1,2-dioxygenase was overexpressed, the accumulated catechol is converted to CCMA. We further improved the The hyperthermophilic archaeon, Thermococcus onnurineus pathway by inactivating several unnecessary genes and NA1 was known for a good bio-hydrogen producer through developed the strain to produce 2g/L of CCMA in flask the water-gas shift (WGS) reaction (CO + H2O → CO2 + H2). cultivation. Furthermore, we optimized aeration and pH for Recently, we reported that H2 productivity from carbon fed-batch fermentation. monoxide (CO) of the metabolically engineered T. onnurin- eusNA1, MC01, was enhanced significantly (at least 2-fold 39. Effects of Pyruvate Formate Lyase Inactivtation in increase), compared to that of wild-type strain. Klebsiella Pneumoniae and Its Application to Diol Production In this study, we assessed kinetic parameters for cell growth Moo-Young Jung* and Min-Kyu Oh and bio-H production from CO by conducting the continu- Department of Chemical and Biological Engineering, Korea 2 university, Seoul, South Korea ous cultivation of MC01 with varying dilution rate, CO supply rate, or agitation speed. Practically, the CSTR fermentation The fermentative production of 2,3-butanediol and 1,3-proof MC01 using CO as a sole carbon could be performed panediol has been intensively studied because of the over a month without any technical problem, implicating that potential of these products as platform chemicals, such as the process is feasible to produce hydrogen continuously polymers, fuels, and plastics. Among several natural diol for a long-term basis. Based on the results, both cell growth producers, Klebsiella pneumoniae can effectively synthe- and H2 productivity were linearly correlated with dilution rate size both 2,3-butanediol and 1,3-propanediol from a wide unless CO was not limited. At a fixed dilution rate of 0.3h-1 , range of carbon sources. In this study, three genes, includ- H2 production rate was linearly increased depending on CO ing wabG, ldhA, and pflB, were removed in K. pneumoniae supply rate up to 240 ml/min, which is corresponded to 0.12 genome to reduce both its pathogenicity and the production vvm. However, at CO rate of 300 ml/min, both cell growth of byproducts. In flask cultivation with minimal medium, the and H2 production rate considerably decreased, indicating yield of 2,3-butanediol and 1,3-propanediol from rationally CO inhibition toward cell growth and H2 production. Ad- engineered K. pneumoniae (¢wabG ¢ldhA ¢pflB) increased ditionally, the increase in agitation speed was also effective significantly. In particular, the inactivation of pflB signifi- in enhancing H2 production. It is considered that enhanc- cantly decreased the synthesis of byproducts and improved ing mass transfer by increasing CO supply or high agitation the yield of 2,3-butanediol from glucose to 92.2% of the speed seemed likely a critical determinant for cell growth

23 Poster Abstracts

and H2 production rate from CO. experimental quantification of intracellular metabolic fluxes is challenging not only because of the extensive instrumen- 41. Prediction and Design of Novel Metabolic Pathways tation required for these methods but also because of the for the Production of Desired Chemicals limited number of fluxes that can be measured. Dong In Kim*, Ayoun Cho, Hongseok Yun, Jin Hwan Park, Sang Yup Lee and Sunwon Park Chemical & Biomolecular Engineering, Korea Advanced One of the alternative methods that is widely used for sys- Institute of Science and Technology (KAIST), Daejeon, South tem-level analysis of metabolic networks is a computational Korea modeling approach called flux balance analysis (FBA), which predicts metabolic flux distributions at steady state by mak- Many systematic platforms have been developed in the ing use of in silico genome-scale metabolic models. Since past decades to predict the metabolic pathways for efficient these models are underdetermined in general, context-spe- production of desired chemicals. Here, we developed the cific and physiologically meaningful flux solutions need to be framework to generate pathways and suggests enzyme narrowed down from the space of all possible distributions. candidates through prioritization process to evaluate feasibility of novel pathways. The systematic framework consists Transcriptomic data can be used to define flux bounds, of two parts, route generation and prioritization process. objective functions, or both, to enhance the predictive power Route generation process generates pathways based on of in silico genome-scale metabolic models. Transcriptomic the reaction rule sets which were constructed based on data has significant advantages for this purpose over other the logics acquired from analysis of reaction mechanism ‘omics’ platforms such as proteomics, metabolomics, and of existing biochemical reactions. After route generation, fluxomics because the layer of RNA transcripts is the only five prioritization factors, binding site covalence, chemical layer where a complete quantitative snapshot of all genome- similarity, thermodynamic favorability, pathway distance and wide molecular species is currently possible. In addition, organism specificity are evaluated to estimate the feasibil- RNA amount changes can be precisely measured in a highly ity of generated reaction pathways. As a validation, novel automated process at a low cost compared to the amount of synthetic pathways for the three chemicals, isobutanol, data gathered. 3-hydroxypropionate (3HP) and butyryl-CoA, were predicted using our system to evaluate the capacity and reliability of Because of these advantages of RNA transcript profiling, the framework. This systematic framework should be an there have been previous studies to integrate transcriptomic additional resource useful for the practice of synthetic biol- data with in silico genome-scale metabolic models. However, ogy and metabolic engineering.[This work was supported previous methods have limitations such as the requirement by the Technology Development Program to Solve Climate of multiple input datatypes for analysis, discretization or Changes on Systems Metabolic Engineering for Biorefineries binarization of gene expression measurement data accord- from the Ministry of Science, ICT and Future Planning (MSIP) ing to the user-defined thresholds, or requirement for a priori through the National Research Foundation (NRF) of Korea knowledge of biomass production rate or carbon source. (NRF-2012-C1AAA001-2012M1A2A2026556).] Moreover, the accuracy of the intracellular fluxes predicted by these methods have not been validated by measured 42. Integration of Transcriptomic Data in Genome-Scale intracellular fluxes. Metabolic Models Predicts in vitro Intracellular Central Carbon Metabolic Fluxes with High Correlation in Escherichia coli and Saccharomyces cerevisiae In this study, we developed a computational tool for pre- Min Kyung Kim*1 and Desmond Lun2 dicting intracellular metabolic flux distribution of E. coli and 1Rutgers University S. cerevisiae by integrating transcriptomic data with their 2Computer Science, Rutgers University in silico genomic-scale metabolic models.We suggest two different template models to be integrated with gene ex- Determination of systemic changes in intracellular metabolic pression data, which are Yes Carbon source (YC) and No fluxes of microorganisms is important to understand funda- Carbon source (NC) models, and two kinds of optimization mental mechanisms of their metabolic responses as well as strategies, Yes Biomass (YB) and No Biomass (NB) strate- to identify molecular targets for metabolic engineering. The gies, which can be chosen and combined depending on the

24 Poster Abstracts

availability of knowledge on the carbon source or biomass flux solution; the method is easy to implement. Our method flux. The YC model allows only the known carbon source would make it possible to identify fundamental mechanisms to be taken up by the cell, while the NC model allows all of metabolic responses and to find reasonable molecular tar- carbon sources in the model to be taken up by the cell. The gets for metabolic engineering in an efficient way. The pres- YB strategy tries to find a metabolic flux distribution that ent study may be extended to test for metabolism of other allows the cell to achieve maximum growth rate in an energy organisms, especially multi-cellular eukaryotic organisms. efficient way by solving a two-step of optimization problem: minimization of the Euclidean norm after maximizing bio- 43. Model-Driven Metabolic Engineering of Escherichia mass. The NB strategy can be applied when biomass flux is coli for Improving Conversion of Lignocellulose-Derived Sugars to Ethanol not a suitable objective function. It maximizes the Pearson Joonhoon Kim*1,2, Mary Tremaine1, Robert Landick1,3,4, correlation between metabolic flux and transcriptomic data. Patricia Kiley1,5 and Jennifer L. Reed1,2 Transcriptomic data are used to constrain fluxes in the model 1DOE Great Lakes Bioenergy Research Center, University of for the YB strategy, while for the NB strategy, they are used Wisconsin-Madison, Madison, WI to define the objective function. 2Department of Chemical and Biological Engineering, University of Wisconsin-Madison, Madison, WI 3Department of Biochemistry, University of Wisconsin – The computational method and its validation process can be Madison, Madison, WI classified into 5 steps, which are 1) obtaining transcriptomic 4Department of Bacteriology, University of Wisconsin – data measured under conditions in which we are interested, Madison, Madison, WI 2) obtaining in vitro fluxes data measured under exactly the 5Department of Biomolecular Chemistry, University of same conditions as that of the transcriptomic data (for the Wisconsin-Madison, Madison, WI purpose of comparing them with the predicted fluxes later), 3) integrating the transcriptomic data into the genome-scale Computational models of genome-scale metabolic networks metabolic model according to the model’s gene-protein-re- have been successfully used to study and engineer microbial action (GPR) association relationship, 4) solving FBA opti- metabolism for production of valuable chemicals. Constraint- mization problems with one of the two objective functions based approaches such as flux balance analysis can predict depending on the availability of information on biomass flux metabolic flux distributions in genetically perturbed strains, or carbon source, and 5) calculating correlation between the and they have been used to identify novel metabolic engi- predicted fluxes and the measured fluxes. neering strategies.

On average, we observed good correlation between the We used genome-scale metabolic models of Escherichia coli predicted fluxes and the measured fluxes. The average un- to identify gene knockout strategies to improve co-utilization of centered Pearson correlation between predicted and mea- glucose and xylose, which are major sugars in lignocellulosic sured fluxes was in the range of 0.7 to 0.9 in most cases, hydrolysate. We constructed the gene knockout mutants and significantly outperforming existing methods. Our method inserted Zymomonas mobilis pyruvate decarboxylase and was able to predict intracellular fluxes with an average un- alcohol dehydrogenase genes to increase ethanol production. centered Pearson correlation of around 0.6 even in the case where carbon source and objective function are unknown. The constructed E. coli mutants co-utilized glucose and It took less than 5 seconds to calculate the flux distribution xylose anaerobically in minimal media, but the growth and using our algorithm in all cases. glucose uptake rates were much slower than wildtype E. coli strain. The mutants were adaptively evolved in minimal me- Aside from this predictive accuracy over competing meth- dia containing glucose, xylose, or both. The evolved mutants ods, our method improves other limitations of the previous were able to simultaneously convert glucose and xylose into approaches described above in that it needs only one set of ethanol when grown in synthetic hydrolysate. gene expression data as an input; it utilizes absolute gene expression data without arbitrary discretization of it; and it can be used when the objective function or the carbon 44. Yeast Cell Factories for Production of Biobutanol Anastasia Krivoruchko*, Verena Siewers and Jens Nielsen source of an organism is not known; it produces a unique Department of Chemical and Biological Engineering, 25 Poster Abstracts

Chalmers University of Technology, Gothenburg, Sweden limitations of single species. The fermentative bacterium

ferments the carbohydrates to H2 and fermentation products Increasing oil demands and diminishing resources, as well while the photoheterotroph oxidizes fermentation products as concerns over global warming, have stimulated interest to H2. These studies have confirmed that coculturing can in- in the biosynthesis of fuels from renewable sources. Biobu- crease the H2 yield. Unfortunately, the coculturing techniques tanols (1-butanol, 2- butanol and iso-butanol) have gener- used did not promote culture stability, as conditions al- ated attention as alternative gasoline additives. Butanols lowed for unrestrained growth of the fermentative bacterium. are superior to ethanol as a fuel source due to their higher Overall, these cocultures have given highly variable trends energy content and lower volatility, hygroscopicity and cor- and have been inadequate as an experimental system. We rosiveness. In addition, they have sufficiently similar charac- hypothesized that engineering metabolic interdependency teristics to gasoline to be compatible with existing engines between microbes would resolve these issues by stabilizing and infrastructure. species population dynamics.

While a natural pathway for 1-butanol production exists in Using defined genetic mutations and environmental con- the Clostridia species of bacteria, difficulties associated with ditions, we developed a stable coculture between E. coli cultivation and genetic manipulation of these bacteria have and R. palustris that functions via obligate, bi-directional called for reconstruction of this pathway in more commonly exchange of nutrients. R. palustris acquires carbon from used industrial microorganisms. The yeast Saccharomyces fermentation products excreted by E. coli while it simulta- cerevisiae makes for an attractive host for 1-butanol produc- neously fixes 2N gas and provides essential nitrogen to E. tion due to its robustness under industrial conditions, avail- coli; growth of each species is dependent on the metabolic ability and abundance of information regarding its physiol- activity of the other. Under these conditions, the coculture ogy, and a broad selection of tools available for its genetic achieves higher H2 yields than E. coli monocultures while ac- manipulation. The goal of this project is therefore to use vari- cessing electron sources otherwise unavailable to R. palus- ous synthetic biology and metabolic engineering tools to cre- tris. Importantly, unlike traditional cocultures, this coculture ate efficient yeast cell factories for the production of 1-buta- is stable and gives reproducible results over multiple serial nol. This includes introduction of a heterologous pathway for transfers, essentially serving as a coculture that can be 1-butanol production into S. cerevisiae in combination with treated like a monoculture. We have found that we can alter various modifications to acetyl-CoA metabolism to improve environmental conditions to limit N2 availability to a subpopu- flux towards 1-butanol. We demonstrate that such manipula- lation of bacteria in the coculture, resulting in even higher tions significantly increase 1-butanol production in yeast. H2 yields while still maintaining stability over serial transfers. We have also determined that the R. palustris Calvin cycle 45. Engineering of a Stable, Syntrophic Microbial competes for electrons against nitrogenase under coculture Coculture for Enhanced H2 Production conditions. Eliminating Calvin cycle activity results in higher Breah LaSarre*, Alexandra L. McCully, Amanda L. Posto and H yields and improved coculture growth rates, likely through James B. McKinlay 2 a redirection of electrons to nitrogenase. Department of Biology, Indiana University, Bloomington, IN

The reproducible metabolic traits of this coculture make it Hydrogen gas is an important commodity chemical and a an attractive platform to adapt 13C-labeling approaches to promising fuel that can be produced biologically from inex- determine metabolic fluxes in single species within a simple pensive renewable resources. Anaerobic fermentative bacte- defined community. Furthermore, the modularity of this ap- ria, such as Escherichia coli, produce H from carbohydrates 2 proach will allow for the incorporation of other microbes to but at a low yield due to obligate co-production of organic make use of alternative feedstocks and to produce other acids and alcohols. In contrast, anoxygenic photoheterotro- chemicals of value to society. phic bacteria, such as Rhodopseudomonas palustris, cannot consume carbohydrates but use electrons in fermentation 46. Redirecting Photosynthetic Reducing Power into products to produce H via nitrogenase. Several studies have 2 Light-Driven Biosynthesis of Bioactive Natural combined the complementary metabolic traits of fermen- Compounds tative and photoheterotrophic bacteria to circumvent the Lærke Marie Münter Lassen*, Agnieszka Zygadlo Nielsen, 26 Poster Abstracts

Carl Erik Olsen, Birger Lindberg Møller and Poul Erik Jensen 1Department of Environmental Engineering and Energy, Copenhagen Plant Science Centre, Department of Plant and Myongji University, Republic of Korea Environmental Sciences, University of Copenhagen, DK-1871 2Korea Research Institute of Chemical Technology, Republic Frederiksberg C, Copenhagen, Denmark of Korea 3Metabolic and Biomolecular Engineering National Research Plants can synthesize around 200,000 different specialized Laboratory, Department of Chemical and Biomolecular Engineering (BK21 Program), KAIST, Republic of Korea metabolites, many of which are of commercial interest as they can be utilized as pharmaceuticals. Often these com- Polyhydroxyalkanoates (PHAs) are microbially derived bio- pounds are produced at low levels in the plants, and due to based polyesters synthesized by many microorganisms. their complex structures, chemical synthesis is not feasible. Since material properties of PHAs are highly dependent The cytochrome P450s (P450s) are among the enzymes that upon their types and compositions of monomers, researches participate in the biosynthesis of these compounds. have been focused on the development of microbial system to design polyesters containing novel monomers, which We are aiming at coupling photosynthetic electron transport might confer superior material properties. directly to a metabolon of P450s to develop a system in which the electrons derived from water can be used directly We have metabolically engineered E. coli strains for the pro- in the reactions of the P450s. As a model system we are duction of 2-hydroxyacids containing PHAs such as lactate- working with the enzymes producing the cyanogenic glu- containing PHAs, polylactic acid (PLA), and 2-hydroxybutyr- coside dhurrin in the endoplasmic reticulum membranes ate (2HB)-containing PHAs from structurally unrelated carbon of Sorghum bicolor. Three enzymes are involved in dhurrin sources by one-step fermentative process. PHAs consisting biosynthesis, two P450s (CYP79A1 and CYP71E1) and a of 2-hydroxyacids as monomer units have attracted much glycosyl transferase, UGT85B1. Recently, we have found attention, but their production has not been efficient. that these enzymes can be expressed in the chloroplasts of Nicotiana benthamiana, where they show light-dependent Since Ralstonia eutropha is the most efficient host strain for dhurrin production with the P450s driven by electrons deliv- the production of PHAs from renewable resources, here, we ered from photosystem I (PSI) by ferredoxin1. report the metabolic engineering strategies for the development of recombinant Ralstonia eutropha strains to synthesize In this project we have genetically fused the catalytic domain PHAs containing 2-hydroxyacids as monomers. This could be of CYP79A1 to a PSI subunit in Synechococcus sp. PCC achieved by the construction of base R. eutropha strains that 7002, thereby making it an integral part of PSI to facilitate express engineered PHA synthase able to use 2-hydroxyacyl- efficient electron transfer. Preliminary results indicate that the CoAs (2HA-CoAs) as substrates and engineered propionyl- PSI-P450 fusion is active both in isolated thylakoids and in CoA transferase to synthesize 2HA-CoAs. Detailed metabolic living cyanobacteria. It is envisioned that the PSI-P450 sys- engineering strategies for the construction of versatile recom- tem can be developed as a light-driven and environmentally binant R. eutropha strains to produce PHAs containing various friendly production system of bioactive compounds relevant 2-hydroxyacid monomers will be presented. to the pharmaceutical or chemical industries.

References: [This work was supported by the Technology Development [1] Nielsen, A. Z. et al. Redirecting Photosynthetic Reducing Power Program to Solve Climate Changes (Systems Metabolic toward Bioactive Natural Product Synthesis. ACS Synth. Biol., Engineering for Biorefineries) from the Ministry of Science, 2013, 2 (6), pp 308–315, DOI: 10.1021/sb300128r ICT and Future Planning (MSIP) through the National Re- Funding from UNIK Center for Synthetic Biology, Interdisciplinary search Foundation (NRF) of Korea (NRF-2012-C1AAA001- Research Center “bioSYNergy” and the VILLUM Research Center 2012M1A2A2026556)]. “Plant Plasticity” is gratefully acknowledged. 48. Biosynthesis of Polyhydroxyalkanoates in 47. Biosynthesis of 2-Hydroxyacid Containing Recombinant Ralstonia Eutropha Engineered to Utilize Polyhydroxyalkanoates in Metabolically Engineered Sucrose As a Carbon Source Ralstonia Eutropha Si Jae Park*1, Young Hoon Oh*2, Jung Eun Yang3, So Young Si Jae Park*1, Seung Hwan Lee2, Young Hoon Oh2, Jung Eun Choi3, Seung Hwan Lee*2 and Sang Yup Lee*3 Yang3, So Young Choi3, Sang Yup Lee3 27 Poster Abstracts

1Department of Environmental Engineering and Energy, nent stresses are accumulation of acetate under sub-optimal Myongji University, Yongin, South Korea glucose feeding regimes, and high osmolarity resulting from 2 Korea Research Institute of Chemical Technology, Daejeon, pH neutralization and high substrate, product, and byprod- South Korea uct concentrations. Different laboratory strains of E. coli 3Chemical & Biomolecular Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South exhibit remarkable variation in their growth phenotypes under Korea these applied stresses, however virtually nothing is known to explain the source of this variation and to enable rational Polyhydroxyalkanoates (PHAs) are potential alternatives of engineering to impart stress tolerance. We have constructed petroleum-based polymers due to their material properties diverse transposon insertion libraries (>50000 mutants) in 4 which are much similar to those of chemical plastics. Since strains of E. coli: K-12 MG1655, BL21(DE3), W, and Crooks. the cost effective production of PHAs is one of the most Short-term growth selections were performed in media important factors for the commercialization of PHAs, devel- supplemented with high acetate concentrations, high NaCl opment of recombinant microorganisms able to efficiently concentrations, and both high acetate and high NaCl concen- utilize cheap carbon sources have been examined for the trations. Insertions frequencies in each gene were determined production of PHAs. Sucrose is one of the most abundant by massively parallel sequencing of transposon-chromosome and least expensive carbon sources extracted from sugar- junctions, allowing an analysis of both conditionally essential cane and sugarbeet. Therefore, microorganisms capable and conditionally detrimental genes. Many differences and of utilizing sucrose as carbon source might support cost- similarities in resistance mechanisms at the genetic level competitiveness of fermentation-driven products. Here, could be revealed across strains, allowing correlations to be we report recombinant Ralstonia eutropha strains able to made with growth phenotypes. Cross-strain comparisons produce PHAs from sucrose as a carbon source. Sucrose of conditionally essential genes and their relative essential- utilization pathway was constructed in Ralstonia eutropha ity also suggest a large degree of variation in metabolic flux NCIMB11599 and R. eutropha 437-540 by introducing distributions and regulation of gene expression between the Mannheimia succiniciproducens MBEL55E sacC gene strains. A number of direct targets for metabolic engineering encoding β-fructofuranosidase and was examined for the of stress resistance via loss-of-function mutations were also production of poly(3-hydroxybutyrate) [P(3HB)] and poly(3- discovered, and we show that deletion of a selection of these hydroxybutyrate-co-lactate) [P(3HB-co-LA)], respectively. genes results in increased growth rates and/or decreased lag Detailed results will be presented in this presentation. times under the original selection condition.

[This research was supported by the Basic Science Re- 50. Construction of a Efficient Xylose Metabolic search Program through the National Research Foundation Pathway in Saccharomyces cerevisiae for Ethanol Production of Korea (NRF) funded by the Ministry of Education (NRF- Yunjie Li*1 and Tianwei Tan2 2013R1A1A2058379) and by the Technology Development 1Beijing Key Lab of Bioprocess, Beijing University of Program to Solve Climate Changes (Systems Metabolic Chemical Technology, Beijing, China Engineering for Biorefineries) from the Ministry of Science, 2Beijing University of Chemical Technology, Beijing, China ICT and Future Planning through the NRF (NRF-2012- C1AAA001-2012M1A2A2026556)]. Nowadays the low ethanol yield and productivity of xylose from plant biomass by Saccharomyces cerevisiae is a big 49. Comparative Cross-Strain Analysis of Stress problem, so that it seriously hinders the development of Resistance Mechanisms Revealed By Transposon industrial cellulose bioethanol. To overcome this bottle- Insertion Sequencing neck, we introduced two key genes of Calvin-Cycle to Rebecca M. Lennen* and Markus J. Herrgård improve the non-oxidative phase of the pentose phosphate The Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Hørsholm, Denmark pathways(PPP), which is the necessary steps of xylose metabolism. In the engineered S.cerevisiae, we expressed Escherichia coli is exposed to a number of stress conditions codon-optimized phosphoribulokinase (PRK) , and ribu- during high-cell density fed batch fermentation due to non- lose-1,5-biphosphate carboxylase/oxygenase (Rubisco) from ideal mixing and pH maintenance. Two of the most promi- Ralstonia eutropha H16, and codon-optimized GroEL/GroES

28 Poster Abstracts

from Escherichia coli, coupled with the overexpression of A bacterial population which is exposed to stressful industrial mutant XYL1(R276H) and XYL2 from Scheffersomyces stipi- production environments may convert into subpopulations tis and XKS from S.cerevisiae. We observed improved etha- with reduced or stopped product formation properties. Con- nol yield (0.45 g/g) and productivity (0.79 g/L/h) from cofer- sequently, overall process performance deteriorates, be- mentation experiments performed with glucose (80g/L )and cause tools to prevent these developments are still unknown. xylose (50g/L), suggesting this is a promising glucose and Using P. putida KT2440 as a model system, we quantified xylose metabolic strategy for cellulose biofuel production. population heterogeneities under standard conditions and The successful integration shortened PPP in S.cerevisiae is a in stress response to oxygen limitation and solvent stress in critical step towards enabling economic biofuel production. controlled continuous cultivations.

51. Development of a Yeast Cell Factory for Resveratrol Stressed cells showed an early wash-out in comparison Production to non-stressed cells. Single cell analysis revealed a split- Mingji Li*1, Kanchana R. Kildegaard1, Edith A. R. Prado1, up of the clonal population into two subpopulations. The Christian B. Jendresen1, Steen G. Stahlhut1, Irina Borodina1 and Jens Nielsen1,2 ratio of subpopulations with single DNA and double DNA 1Novo Nordisk Foundation Center for Biosustainability, content were investigated at steady-state in stressed and Technical University of Denmark, Hørsholm, Denmark non-stressed set-ups. Stressed populations showed a higher 2Department of Chemical and Biological Engineering, pronounced continuous shift of this ratio towards cells con- Chalmers University of Technology, Göteborg, Sweden taining a double DNA content.

Resveratrol is a natural polyphenol with antioxidative, cardio- DNA distribution reflects information of growth rate on protective, and antitumor activities. In the present study, we subpopulation level and mirrors changes in environmental turn to produce resveratrol directly from glucose in Saccha- surroundings. The observed shifts in subpopulation ratios romyces cerevisiae. Phenylalanine/tyrosine ammonia lyase indicate influences on the level of cell cycling. Steady state activity was combined with expression of p-coumarate-CoA analysis of DNA contents is integrated into a mechanistic ligase from Arabidopsis thaliana and of resveratrol synthase model of cell cycling. Cell cycle parameters were estimated from Vitis vinifera. The resulting strain produced 39.70±1.08 by comparison of simulated to experimentally derived DNA mg L-1 resveratrol in mineral medium. In order to increase content histograms. The comparison of the relevant param- the supply of malonyl-CoA precursor, we have introduced eters in stressed versus non stressed growth conditions acetyl-CoA synthetase from Salmonella enterica and over- gives valuable insights into quantitative understanding of expressed native aldehyde dehydrogenase and inactivation- subpopulation split-up and highlights the influence of typical resistant version of acetyl-CoA carboxylase. Furthermore, production stress conditions on population dynamics. to produce resveratrol directly from glucose, tyrosine supply was improved by overexpression of feedback-resistant This project is funded within the ERA-IB / ERA-NET Scheme versions of chorismate mutase and 3-deoxy-D-arabino- of the 6th EU Framework Programme (0315932B) heptulosonate-7-phosphate synthase. Finally, we optimized the expression of genes involved in the resveratrol pathway 53. Carbon Flux-Associated Redox Rebalancing By by using multiple integrations in the genome at TY elements. Static and Dynamic Control The results show a potential of industrial resveratrol produc- Jae Hyung Lim*1 and Gyoo Yeol Jung1,2 tion from glucose by engineered S. cerevisiae. 1School of Interdisciplinary Bioscience and Bioengineering (I-Bio), POSTECH, Pohang, South Korea 52. Modelling Population Dynamics of Pseudomonas 2Department of Chemical Engineering, POSTECH, Pohang, Putida KT2440 Under Various Growth Conditions South Korea Sarah Lieder*1, Michael Jahn2 and Ralf Takors3 1Institute for Biochemical Engineering, University of The intracellular redox state plays an important role for the Stuttgart, Stuttgart, Germany efficient production of chemicals and biofuels by microbial cell 2 Environmental Microbiology, UfZ Leipzig, Leipzig, Germany factories. However, it is difficult to achieve optimal redox rebal- 3University of Stuttgart, Institute of Biochemical Engineering, ancing of synthetic pathways due to the sensitive responses of Stuttgart, Germany cellular physiology with regard to the intracellular redox state

29 Poster Abstracts

changes as well as overall carbon fluxes. Here, we propose enhanced utilization of both galactose and alginate and optimal rebalancing of the intracellular redox state enables production of biochemicals. Finally, we could confirm the maximizing cellular performance and demonstrated using possibility of marine biomass as carbon source to produce n-butanol producing E. coli as a model system. Redox rebal- many valuable bioproducts instead of fossil fuel. ancing was achieved by tuning the expression level of NAD+- dependent formate dehydrogenase (fdh1 from Yeast) through 55. Comprehensive Study of Metabolic Flux Rewiring in rational UTR engineering. In the static control of redox state, E. coli Knockout Strains Christopher P. Long* and Maciek R. Antoniewicz flux of n-butanol was enhanced up to 35.4% in glucose-sup- Department of Chemical and Biomolecular Engineering, plemented medium, and 44.6% in case of galactose-supple- University of Delaware, Newark, DE mented. Interestingly, efficient production of n-butanol required different amounts of reducing equivalents depending on the Cellular metabolic and regulatory systems are of fundamen- substrate. This intriguing finding was expanded to construct tal interest to systems biology and metabolic engineering, genetic circuits for redox rebalancing in a dynamic fashion that but incomplete understanding of their complex properties re- respond to metabolic fluxes, enabling optimize cellular perfor- mains an obstacle to progress in those fields. An established mance even under various environmental conditions by the method for obtaining information on network structure, automatic redox rebalancing. Collectively, this work suggests regulation, and dynamics is to study the system following a that redox rebalancing depending on cellular demands is a key perturbation such as a genetic knockout. The Keio collection determinant for strain improvement and optimal production of all viable Escherichia coli single-gene knockouts is now fa- of chemicals and fuels can be easily achieved by the redox cilitating a systematic investigation of the regulation and me- rebalancing along with the carbon fluxes. tabolism of E. coli. Of all omics measurements available to observe the perturbed phenotype, the metabolic flux profile 54. Enhanced Utilization of Non-Favored Sugars from (the fluxome) provides the most direct representation and the Marine Biomass By Re-Designed Escherichia coli most relevant to metabolic engineering. Recent advances in Hyun Gyu Lim*1, Jae Hyung Lim2 and Gyoo Yeol Jung1,2 13 13 1Department of Chemical Engineering, POSTECH, Pohang, C-metabolic flux analysis ( C-MFA) are permitting highly South Korea precise and accurate flux measurements for investigating 2School of Interdisciplinary Bioscience and Bioengineering cellular systems and guiding metabolic engineering efforts. (I-Bio), POSTECH, Pohang, South Korea In this project, the core metabolism of E. coli is studied by Marine biomass, known as the third generation of biomass measuring metabolic fluxes of all single gene knockout mu- for replacing fossil fuels, has many advantages rather than tants associated with central carbon metabolism (glycolysis, conventional biomass due to its high biomass productiv- pentose phosphate pathway, TCA cycle, and anaplerotic ity, carbon fixation rate and easy pretreatment process, etc. reactions). First, a parallel labeling experimental design was Macroalgae, which is main part of marine biomass, contains optimized to improve (by about 10-fold) flux estimation preci- various kinds of sugars such as glucose, galactose, and sion in the E. coli network model. Our optimized experiment alginate. However, the challenging problem for using mac- design is based on performing parallel experiments with [1,2- roalgae as a substrate is that these sugars are not usually 13C]glucose, [1,6-13C]glucose and [4,5,6-13C]glucose tracers, favored and utilized by many microorganisms such as Esch- coupled with mass spectrometry measurement of isotopic erichia coli. Slow utilization rate of non-favored sugar limits labeling and 13C-MFA. In addition to these isotopic labeling reduced carbon flux and results in low productivity and titer measurements, we also measure changes in biomass com- and production using non-utilizable sugar is virtually impos- position and cellular yields for each Keio strain. These mea- sible. In this research, we tried to develop E.coli to utilize surements constitute a thorough assessment of the metabol- galactose and alginate from marine biomass and to produce ic states of each knockout. This information is then used to some commodity chemicals. For utilization of these sugars, learn about possible novel or unannotated reactions in central rationally designed pathway with synthetic promoter, 5’UTR, carbon metabolism. The metabolic fluxes are also compared terminator for maximum catalytic activity were introduced. to predictions made by commonly used constraint-based And downstream pathway for production were fine-tuned at models including FBA, MOMA, and ROOM. a key node. With this approach, the engineered strain shows

30 Poster Abstracts

In this presentation, we will present our first rigorous assess- limiting reactions9,10. The availability of PEP and erythrose- ment of wild-type E. coli and two knockout strains (pgi-KO 4-phosphate (precursors for DAHP synthesis) was enhanced and zwf-KO). These two enzymes catalyze the two sides of through changes in the central glucose catabolism. the branch point between glycolysis and the pentose phosphate pathway: pgi (phosphoglucose isomerase) and zwf In shake flask experiments, up to 2 mM of chorismate (from (glucose-6-phosphate dehydrogenase). We will also present 50 mM glucose) were measured in the supernatant of the ongoing work on the knockout strains of the pentose phos- constructed strains. Depletion of accumulated chorismate phate pathway and glycolysis. When completed, this project was observed during the course of cultivation, indicating its will provide the first comprehensive assessment of the regula- subsequent consumption. Based on the chorismate pro- tion of E. coli central carbon metabolism; it will allow detailed ducing strains, further genes were added on newly created analysis of the accuracy of constraint-based models such expression plasmids to obtain chorismate-derived products. as FBA, MOMA, and ROOM to predict metabolic fluxes in Examples will be given. perturbed systems, which has been difficult to date because 1. K.E. Nelson et al (2000) Environ. Microbiol. 4(12), 799-808. of limitations in existing knockout flux data; and finally, it will 2. N. Graf, J. Altenbuchner (2011) Appl. Environ. Microbiol. 77(15), provide valuable information for new model development, for 5549-5552. example, in training and fitting parameters in mechanistic (i.e. 3. F. Gibson, J. Pittard (1968) Bacteriol. Rev. 32(4), 465-492. kinetic) metabolic models and regulatory models. 4. F. Gibson (1964) Biochem. J. 90, 256-261.

56. Metabolic Engineering of Pseudomonas Putida 5. J. Bongaerts et al. (2011) Angew. Chem. 123, 7927-7932. KT2440 for the Production of Compounds Derived from 6. M.I. Gibson et al. (1962) Nature 195, 1173-1175. the Shikimic Acid Pathway 7. M.A. Molina-Henares et al. (2009) Microbial. Biotechnol. 2(1), Silvia Lorenz*1, Natalie Trachtmann1, Nadja Graf2, Josef 91-100. Altenbuchner2 and Georg A. Sprenger1 1Institute of Microbiology, University of Stuttgart, Stuttgart, 8. R. Schoner, K.M. Herrmann (1976) J. Biol. Chem. 251, 5440-5447. Germany 9. G.A. Sprenger (2007) Appl. Microbiol. Biotechnol. 75, 739-749. 2Institute of Industrial Genetics, University of Stuttgart, Stutt- 10. K.A. Dell, J.W. Frost (1993) J. Am. Chem. Soc. 115, 11581- gart, Germany 11589.

We show an engineered biosynthetic pathway for aromatic 57. Metabolic Engineering of Yeast for Commercial substances based on the simple sugar glucose in Pseu- Production of Succinic Acid domonas putida KT2440. P. putida KT2440 is the best Alrik Los*, Rene Verwaal, Ben den Dulk, Mickel Jansen, characterized strain of the saprophytic Pseudomonads and Zheng Zhao, Theo Geurts and Roel Bovenberg has been certified as a biosafety host for cloning of foreign DSM Biotechnology Center, Delft, Netherlands genes1. Chromosomal changes were implemented using a DSM and ROQUETTE have started a joint venture by the method for markerless gene insertion and deletion, based name of “Reverdia” (www.reverdia.com) for the fermenta- on homologous recombination and counterselection with tive production and commercialization of succinic acid the antimetabolite 5-fluorouracil in an upp mutant strain2. from renewable resources, to be marketed under the name Making use of the inherent shikimate pathway1,3 we aimed BiosucciniumTM. Succinic acid has been identified as a to accumulate the intermediate chorismate as building block potential key building block for deriving both commodity and for further biosynthesis. Chorismate is the branching point in specialty chemicals from biomass. While existing markets the biosynthesis of a diverse set of aromatic compounds4-6. for chemically produced succinic acid include pharmaceuti- The chorismate-utilizing reactions for the aromatic amino acid cals, food, coatings and pigments, bio-based succinic acid biosynthesis were blocked by constructing pheA and trpE is envisioned to drive the emergence of new applications deletion mutants7. To channel carbon flow into the shikimate such as polyester polyols for polyurethanes, polybutylene pathway, the gene aroF for the (tyrosine-sensitive) DAHP succinate (PBS), plasticizers, 1,4-butanediol and resins. The synthase (AroF) of E. coli8 was chromosomally integrated fermentative production of succinic acid at low pH results in a into the P. putida KT2440 genome. Additionally, genes aroB substantially lower environmental footprint compared to both and aroL, for the DHQ synthase (AroB) and shikimate kinase the current petrochemical process, the near-neutral pH used (AroL) were chromosomally integrated to increase known rate 31 Poster Abstracts

in bacterial fermentation processes, as well as the process for min analogs appears to be economical since their precursor petrochemical adipic acid, which is the conventional chemical molecules, vitamins, are already available in producer cells and used for production of for example polyester polyols. merely modifying enzymes are needed to convert the vitamin into an antibiotic. We present the metabolic engineering strategy of the yeast 2. Many microorganisms (target organisms) have efficient vita- used in this process. Heterologous genes, optimized for ex- min transporters. These transporters also catalyze the uptake pression in the host, were introduced for achieving high level of vitamin analogs and ensure the delivery of the antibiotic to production of succinic acid from sugar. Expression of these the target molecules in a cell. genes was verified at the protein level with LC-MS. Systems- 3. Since most vitamins are active at more than one site, vi- level analysis (transcripts, proteins and metabolic fluxes) pro- tamin analogs in principle have multiple cellular targets and vided insight into the physiology and metabolism of the strain the frequency of appearance of resistant strains is expected during fermentation, and generated various new leads for to be lower. strain and process optimization. The strain was additionally 4. Many vitamins are synthesized on an industrial scale using mi- improved by classical means with mutagenesis and crossing. croorganisms. Once the biosynthetic pathway for a certain vita- Re-sequencing of a selection of strains provided additional min analog is known, an already established vitamin bioprocess information and targets for further strain improvement. can be used as a starting point for a vitamin analog bioprocess. 5. Natural antimetabolites are thought to in general have a Initial shake flask protocols for strain testing were successfully lower toxicological potential since these compounds have down-scaled to micro titer plates, and the process for produc- coevolved in close contact to cellular structures. tion of succinic acid has been scaled up in our demonstration Three examples of naturally occurring vitamin analogs with an- plant. In this process the strain produces succinic acid at a tibiotic/toxin function have been featured in publications: Baci- high titer at a low pH. Furthermore, Reverdia is the first in the methrin, ginkgotoxin and roseoflavin. Bacimethrin is a thiamine world to have a large scale facility for the commercial produc- (vitamin B1) analog produced by Bacillus megaterium and tion of bio-based succinic acid. The new facility, based in Cas- Streptomyces albus. Ginkgotoxin is a neurotoxin occurring in sano Italy, commenced operation in December 2012. Ginkgo biloba and is structurally related to vitamin B6. Roseo- flavin is a riboflavin (vitamin B2) analog produced by the Gram- 58. Vitamin Analogs As Antiinfectives: Occurrence, positive bacteria Streptomyces davawensis and Streptomyces Mode of Action, Metabolism and Production 1 Danielle Biscaro Pedrolli, Frank Jankowitsch, Simone Langer, cinnabarinus . Roseoflavin is studied in our laboratory as a Julia Schwarz and Matthias Mack* model compound. We investigate the biosynthesis of roseofla- Institute for Technical Microbiology, Mannheim University of vin, its possible large-scale production, its metabolization, its Applied Sciences, Paul-Wittsack-Str. 10, 68163 Mannheim, mode of action and the resistance mechanism of the producer Germany. organisms in order to pave the way for the structured analysis of other vitamin analogs to be discovered. It was postulated Antimetabolites are compounds, which are structurally similar that roseoflavin is synthesized from GTP and ribulose-5-phos- to molecules needed to carry out primary metabolic reac- phate through riboflavin, 8-amino-riboflavin and 8-methyl- tions. The inhibitory activity of an antimetabolite depends on amino-riboflavin2. Recently, the first enzyme of the roseoflavin its successful competition with the natural substrate, ligand, biosynthetic pathway has been identified in S. davawensis, an modulator or cofactor of a given biomolecule. Antimetabolites (S)-adenosyl methionine-dependent N,N-8-amino-riboflavin are indispensable as molecular tools in order to understand dimethyltransferase (EC 2.1.1.93), which converts 8-amino- biological processes. Beyond that, antimetabolites are used as riboflavin in two steps to roseoflavin3. The corresponding gene anti-infectives and anti-cancer drugs and also are employed as rosA was found to be located in a cluster comprising a total preserving agents, pesticides or insecticides. of 10 genes. The remaining genes of this cluster were shown to not be involved in roseoflavin biosynthesis. Employing Natural vitamin analogs with antibiotic function can be consid- cluster deletion and heterologous expression experiments a ered as natural antimetabolites and aroused our interest for the second gene cluster was identified which directs the synthesis following reasons: of 8-amino-riboflavin. The genes/enzymes of this cluster are 1. From an evolutionary standpoint the “development” of vita- currently analyzed with regard to their function, which eventu-

32 Poster Abstracts

ally should lead to the elucidation of the complete roseoflavin other mammal. However, since human flavokinase and FAD biosynthetic pathway. About eight thousand tons of riboflavin synthetase convert roseoflavin to RoFMN and RoFAD it is pos- are produced annually and cost-effective biotechnological ribo- sible that flavin analogs (and/or degradation products of flavin flavin production processes were developed using e.g. Bacillus analogs) negatively interfere with human metabolism12. subtilis and Ashbya gossypii. We anticipate that an economic bio-production process can also be established for roseoflavin, S. davawensis is naturally resistant to its own antibiotic and once its biosynthetic pathway is fully understood. exemplarily was studied with regard to roseoflavin resistance, an issue, which is crucial for developing flavin analogs as The mechanism of action of roseoflavin was studied in con- anti-infectives. S. davawensis contains an FMN riboswitch siderable detail in our laboratory1. For a variety of microorgan- which is not affected by RoFMN and which was found to be isms it was shown that roseoflavin is an excellent substrate for responsible for roseoflavin resistance9. Moreover, a roseofla- riboflavin transporters4. Moreover, it was found that roseoflavin vin exporter appears to be present which confers roseoflavin quickly and almost quantitatively is converted to the flavin resistance as well. cofactor analogs roseoflavin mononucleotide (RoFMN) and roseoflavin adenine dinucleotide (RoFAD) by flavokinases (EC Although, until now, only very few vitamin analogs with antibi- 2.7.1.26) and FAD synthetases (EC 2.7.7.2)5,6. These enzymes otic function have been identified, we expect that a multitude are present in all organisms. Thus, RoFMN and RoFAD (rather of yet unknown vitamin analogs exists. These compounds than roseoflavin) are the effector molecules in target cells. could help to replenish the arsenal of antimicrobials urgently RoFMN was reported to reduce expression of genes involved needed to fight multiresistant bacterial pathogens. in riboflavin biosynthesis and/or transport in B. subtilis, Strep- References tomyces coelicolor and the human pathogen Listeria monocy- 1. Pedrolli, D. B., Jankowitsch, F., Schwarz, J., Langer, S., Nakanishi, togenes7,8. These genes are all controlled by FMN riboswitch- S., Frei, E., and Mack, M. (2013) Curr Pharm Des 19, 2552-2560 es, regulatory genetic elements, which are negatively affected 2. Otani, S., Matsui, K., and Kasai, S. (1997) Osaka City Med J 43, 107-137 by RoFMN. This reduction of gene expression explains at least 3. Jankowitsch, F., Kuhm, C., Kellner, R., Kalinowski, J., Pelzer, S., in part why roseoflavin acts as an antibiotic. For example, Macheroux, P., and Mack, M. (2011) J Biol Chem 286, 38275-38285 reduced expression of the FMN riboswitch controlled ribofla- 4. Hemberger, S., Pedrolli, D. B., Stolz, J., Vogl, C., Lehmann, M., and vin biosynthetic genes ribEMAH in S. coelicolor (caused by Mack, M. (2011) BMC Biotechnol 11, 119-129 RoFMN) led to a significantly decreased level of riboflavin syn- 5. Grill, S., Busenbender, S., Pfeiffer, M., Kohler, U., and Mack, M. thase (RibE, EC 2.5.1.9) activity and consequently to reduced (2008) J Bacteriol 190, 1546-1553 supply of riboflavin9. The addition of roseoflavin to riboflavin 6. Langer, S., Nakanishi, S., Mathes, T., Knaus, T., Binter, A., Mach- auxotrophic L. monocytogenes resulted in reduced expres- eroux, P., Mase, T., Miyakawa, T., Tanokura, M., and Mack, M. (2013) sion of the FMN riboswitch controlled riboflavin transporter Biochemistry 52, 4288-4295 gene lmo1945 and to reduced supply with riboflavin as well 7. Mansjo, M., and Johansson, J. (2011) RNA Biol 8, 674-680 7. However, RoFMN and RoFAD were found to be also active at other sites in target cells. Approximately 1-3% of all bacte- 8.Ott, E., Stolz, J., Lehmann, M., and Mack, M. (2009) RNA Biol 6, 276-280 rial proteins depend on the riboflavin derived cofactors FMN or FAD10 and thus naturally constitute additional targets for 9. Pedrolli, D. B., Matern, A., Wang, J., Ester, M., Siedler, K., Breaker, R., and Mack, M. (2012) Nucleic Acids Res 40, 8662-8673 RoFMN and RoFAD. Indeed, some of these FMN- or FAD-dependent proteins (“flavoproteins”) were found to be less active 10. Macheroux, P., Kappes, B., and Ealick, S. E. (2011) Febs J 278, 2625-2634 or completely inactive in combination with RoFMN or RoFAD (6). Moreover, 37 out of 38 Escherichia coli flavoproteins were 11. Langer, S., Hashimoto, M., Hobl, B., Mathes, T., and Mack, M. (2013) J Bacteriol 195, 4037-4045 shown to contain either RoFMN or RoFAD when cells were treated with roseoflavin11 indicating that roseoflavin indeed has 12. Pedrolli, D. B., Nakanishi, S., Barile, M., Mansurova, M., Carmona, multiple targets and has the potential to exert a broad negative E. C., Lux, A., Gartner, W., and Mack, M. (2011) Biochem Pharmacol 82, 1853-1859 effect on cellular physiology. No FMN riboswitches or similar flavin-binding control elements were found in humans or any 13. Serganov, A., Huang, L., and Patel, D. J. (2009) Nature 458, 233- 237

33 Poster Abstracts

59. Genome-Scale Strain Designs Based on Regulatory simultaneous modifications due to the explosion in the com- Minimal Cut Sets binatorial space. The model-based strain designs have been 1, 2 Radhakrishnan Mahadevan* Axel von Kamp and Steffen experimentally validated for lactate, butanediol, malonyl-CoA Klamt3 and fatty acid production in E. coli, and vanillin production in S. 1Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, ON, Canada cerevisae highlighting the value of these in silico strain designs 2Max Planck Institute for Dynamics of Complex Technical for metabolic engineering. Systems, Magdeburg, Germany 3Analysis and Redesign of Biological Networks, Max Planck An alternative approach which does not use a cellular objec- Institute for Dynamics of Complex Technical Systems, tive function is the use of minimal cut sets in which all flux Magdeburg, Germany distributions with a yield lower than the desired value are eliminated by the identification of minimal combinations of Recent advances in genome-wide characterization of cellular reaction or gene deletions. Previously, such minimal cut sets systems have allowed the opportunity to catalog a significant were identified after the calculation of elementary modes for fraction of the metabolic reactions in the cell. These advances a given metabolic network3. However, a recent study showed coupled with the development of metabolic modeling methods the equivalence between the elementary modes of the dual have enabled the construction of detailed models for many and the minimal cut sets of the primal problem4. This advance industrially relevant microbial hosts including Escherichia coli, enabled the direct identification of these minimal cut sets using and Saccharomyces cerevisiae. In parallel, the increasing only the network stoichiometry and constraints. Recently, this price and volatility associated with petroleum based feed- method, coupled with the use of binary indicator variables that stocks stimulated the use of biological processes for renew- represented whether a continuous variable was non-zero, was able chemical synthesis. Together, these two factors further used to identify thousands of minimal cut sets for intervention motivated the development of computational algorithms that problems in genome-scale metabolic networks5. Once the facilitate the engineering of metabolism for enabling renewable minimal cut sets were identified, additional constraints that chemicals synthesis. allow, for example, biomass synthesis with a minimum yield can be used to filter the set and identify constrained minimal In the past, several algorithms have been developed for cut sets (cMCS) fulfilling this desired behaviour. The findings computational strain designs including the series of bilevel in [5] highlighted the ability of this cMCS approach to identify optimization algorithms such as OptKnock, OptStrain, Opt- novel deletion strategies that are not typically found by the Gene, OptReg, CosMos and OptORF1. In all of these methods, growth-coupled bilevel optimization problems. However, one of the inner optimization problem involves the formulation of the current restrictions of the cMCS approach is that these are genome-scale metabolic model with growth rate maximiza- limited to identifying only deletion modification. tion as the objective, whereas the outer optimization involves the identification of the specific gene deletion/addition (represented as a binary variable) that lead to the maximization of the flux to the target chemical. Typically, the duality theory is used to convert the bilevel optimization problem into a single level mixed integer linear optimization problem with both the primal and dual version of growth maximization linear programs. However, the solution of the bilevel optimization problem for genome-scale models for cases with more than 6 or so modifications can be computationally prohibitive, even more if one wants to enumerate alternate intervention strategies. Finally, we have also previously developed an integer-free optimization approach based on successive linear programming that was shown to be highly efficient (EMILiO2). However, most of these algorithms use growth rate maximization as the cellular objective and use integer variables, which limit the number of

34 Poster Abstracts

60. A Bayesian Design of Experiments for Ensemble using the parameter ensemble as a priori information. Modelling of Metabolic Networks Erica Manesso* and Rudiyanto Gunawan The design of experiments is based on approximate Bayesian Institute for Chemical and Bioengineering, ETH Zurich, Zurich, computation design (ABCD) [3]. Briefly, the procedure consists Switzerland in four steps: (1) select a experimental design d, (2) draw a sample p from the parameter ensemble (prior distribution) P(p), Model-based discovery in biology is an iterative process that (3) for every parameter sample, simulate the model and gener- integrates wet-lab experiments, in silicoanalysis, and opti- ate a sample of noisy data y* (assuming the data noise statistic misation. Most modelling studies in the literature involve the is known), thus producing a sample from the joint distribution creation of an accurate (high fidelity) model of the system, de- P(y,p|d), (4) for every y in the marginal distribution P(y|d), gather fined by a set of equations and parameter values that describe p for which y* is within a neighbourhood of y, (5) compute important or interesting behaviour of the system. There are design criterion based on the neighbourhood p, and (6) repeat many challenges in performing the iterative model-based dis- these steps while optimizing the design criterion. We evaluated covery. The bottlenecking step is typically encountered during several design criteria including maximization of the mode of the estimation of unknown kinetic parameters from experimen- posterior distribution and Fisher information. We demonstrated tal data. This has led to the development of a large number of the utility of the design on a few case studies of metabolic parameter estimation techniques [1]. networks: a generic branched pathway [4] and the trehalose pathway in Saccharomyces cerevisiae [5]. The estimation of kinetic parameters by fitting model simulations to biological data is usually ill posed. There often does not References exist a single (best-fit) solution to the data fitting problem, but [1] I. C. Chou and E. O. Voit. Recent developments in parameter rather one can find many parameter combinations, i.e. an en- estimation and structure identification of biochemical and genomic semble of parameters, that can fit the data statistically equally systems. Math Biosci. 219(2):57-83, 2009. [2] G. Jia, G. Stephanopoulos, and R. Gunawan. Ensemble kinetic well. Here, the parameter ensemble represents the uncertainty modeling of metabolic networks from dynamic metabolic profiles. of the model parameters. We have recently introduced an algo- Metabolites 2(4):891-912, 2012. rithm for constructing such an ensemble from a given dataset [3] M. Hainy, W. Müller, and H. Wynn. Approximate Bayesian of time-series metabolite concentrations for kinetic models of Computation Design (ABCD), an Introduction. In: D. Ucinski and A. metabolic networks [2]. The ensemble corresponds to the pa- C. Atkinson; Patan, Maciej (Eds.): mODa 10 – Advances in Model- rameter subspace defined by the contour of the likelihood ratio Oriented Design and Analysis, Proceedings of the 10th Interna- tional Workshop in Model-Oriented Design and Analysis Held in for a specified statistical significance level. In practical applica- Łagów Lubuski, Poland, June 10–14, 2013, Series Contributions tions, it is often desired and necessary to reduce the size of the to Statistics, Springer International Publishing, Page(s) 135-143, parameter ensemble by performing additional experiments and Springer, 2013. gathering new data. [4] E.O. Voit and J. Almeida. Decoupling dynamical systems for pathway identification from metabolic profiles. Bioinformatics The goal of the present work is to design the experiments that 20(11):1670-1681, 2004. [5] I. C. Chou and E. O. Voit. Estimation of dynamic fluxprofiles would lead to a significant reduction in the ensemble size. The from metabolic time series data. BMC Syst. Biol. 6:84-106, modern technique of model-based experimental design aims 2012. at obtaining the most informative data from an experiment in order to validate the predictions of a model (model outputs). 61. Developing an Integrated Systems and Synthetic For this purpose, the experimental conditions, including for Biology Platform for Gas Fermenting Acetogens 1 2 2 example the sampling times and the time-varying controls or Esteban Marcellin* , Michael Koepke , Wayne Mitchel , Simon Segovia2 and Lars K, Nielsen1 inputs of the system, are optimized to obtain the maximum 1Australian Institute for Bioengineering and Nanotechnology, information from the data. Many designs of experiments have The University of Queensland, St Lucia, Australia previously been developed based on the Fisher information 2LanzaTech matrix optimality criterion without considering the uncertainty associated to the model parameters. Here, we have used a Gas fermentation offers numerous advantages for the produc- Bayesian approach to optimize the experimental conditions tion of sustainable fuels and chemicals without compromising

35 Poster Abstracts

food security or being dependent on the availability of arable quantities of ATP (metabolomics data). This high efficiency of land. Advantages over the use of conventional (sugar, oil or the electron chain transfer, coupled to the recently discovered algae) feedstocks also include minimal water and nutrient electron bifunctional hydorgenases is potentially what makes demand and the capability to capture greenhouse gases that C. autoethanogenum so unique and a great industrial platform would otherwise be emitted into the atmosphere. Practically for the conversion of syngas into valuable fuels and chemicals. all carbon-containing feedstocks, including industrial off-gases and gasified agricultural and municipal waste, can be readily 62. Rational Genome Engineering with Genetically utilised or converted to syngas for subsequent transformation Encoded Biosensors at Single-Cell Scale Jan Marienhagen*, Stephan Binder, Georg Schendzielorz, into fuels and chemicals. Lothar Eggeling and Michael Bott Institute of Bio- and Geosciences, IBG-1: Biotechnology, Acetogens such as Clostridium Ljundahlii or Clostridium auto- Forschungszentrum Juelich, Juelich, Germany ethanogenum use the reductive acetyl-CoA (Wood-Ljungdahl) pathway as a terminal electron-accepting, energy-conserving, The engineering of microbial strains for the production of

CO2-fixing process. This pathway is speculated to be the small molecules is a time-consuming, laborious and expen- first biochemical pathway in existence on Earth and contin- sive process. This can be mostly attributed to the fact that ues to play a key role in the global acetate cycle with annual good producers cannot be readily obtained by high-through- acetogenesis in sediments and termite hindguts estimated to put (HT) screening approaches since increased product for- amount to several trillion kg of acetate. Energy metabolism mation usually does not confer a clear phenotype to produc- in acetogens is complex and many aspects are only partly ing strain variants. understood. Recently, advances were made in the design and construc- While all acetogens use the Wood-Ljungdahl pathway to fix tion of genetically encoded biosensors for detecting small

CO2, they vary in terms of redox coupling (cytochromes, sodi- molecules at the single-cell level [1]. In combination with um translocating Rnf or proton translocating Rnf) and co-factor fluorescent-activated cell sorting (FACS) we already demon- use in the bifurcating hydrogenase (NADPH or NADH). Under- strated the value and potential of these new tools for microbial standing the complex energy metabolism is critical, since ATP strain development by screening large libraries of chemically availability is a fundamental limitation in engineering acetogen mutagenized Corynebacterium glutamicum cells for L-lysine metabolism. C. autoethanogenum offers a robust engineering producers with the L-lysine sensor pSenLys [2]. Whole genome system and a flexible platform for syngas fermentation. Fer- sequencing of selected clones identified a single mutation mentation of C. autoethanogenum has high product selectiv- leading to an amino acid substitution from glycine to glutamate ity, tolerates a broad range of gas compositions and accepts at position 81 in the UDP-N-acetylmuramyl-tripeptide synthe- contaminants well, making it the preferred microorganism for tase (MurE). This enzyme consumes the L-lysine precursor D, industrial gas fermentation. L-diaminopimelate and is involved cell wall synthesis, rendering MurE essential for C. glutamicum. Using C. autoethanogenum as the system for this study, we used a multi-omics approach in order to gain a measure of en- Motivated by the assumption that other amino acid substi- zyme capacity (transcriptomics and proteomics) and thermo- tutions in murE might lead to even higher Llysine titers, we dynamic driving force (metabolomics). By comparing gene ex- developed RecFACS for the site-directed saturation muta- pression and transcription when cells are grown autotrophically genesis of microbial genomes. RecFACS combines targeted

and heterotrophically (CO CO2 and H2 vs fructose fermentation) genome mutagenesis via recombineering with biosensor- we found that, compared to C. ljungdahlii, in C. autoethano- guided HT screening for improved producers (Fig. 1) [3]. genum the RNF complex is extremely efficient and is able to We successfully used RecFACS to generate and screen a maintain high ATPase activity at the transcriptional (RNA-seq) site-saturation library of murE of C. glutamicum via FACS and and translational (iTRAQ) levels even when cells are fermented identified 12 different amino acid substitutions causing differ- exclusively using gas. Furthermore, our RNA-sequencing ent L-lysine production titers. data show that under autotrophic conditions, genes in the RNF complex are highly transcribed, resulting in equi-molar RecFACS is highly suitable to generate targeted genetic

36 Poster Abstracts

diversity in microbial genomes and screen for phenotypes in structures that do not occur in nature. To diversify the types of a HT-format in a single step, thus establishing a new general chemicals that can be made by biological synthesis, we must concept in metabolic engineering at genome-scale. rely on enzyme promiscuity to catalyze reactions on non-native substrates for which limited data currently exists. Aldehydes 63. An Integrated Multi-Omics and Computational are commonly used in polymerization chemistry as well as Characterization of Seven Unique Escherichia coli many flavoring and perfume additives; however, the diversity Production Strains Commonly-Used in Industrial Biotechnology of aldehydes is low in native metabolomes, particularly com- Jonathan M. Monk* pared to related structures like alcohols and carboxylic acids. University of California, San Diego, La Jolla, CA We use a dual computational and experimental approach to assess enzyme reactivity for a range of enzymes that reduce A major challenge in industrial biotechnology is the selec- carboxylic acids to aldehydes. Using two of our own com- tion of an appropriate platform organism to be engineered in putational tools, the biochemical network simulator BNICE order to maximize the production of the desired chemical in as and the enzyme prediction tool SimZyme, we were able to little time as possible. Often times, researchers resort to their identify a group of enzymes that convert carboxylic acids to in-house favorite strain due to a lack of a standardized and aldehydes, mostly through a thermodynamically favorable, functional comparison across strains within a species. In this conserved phosphopantetheine-mediated reduction. We then study, we set out to clarify some of the strain-specific differ- expanded the scope to homologues of the original enzymes. ences in the most common laboratory organism and produc- We are now able to give an expanded view of several carbox- tion chassis, Escherichia coli. Specifically, we performed an ylic acid precursors that can be converted to aldehydes, as integrated functional characterization of seven commonly well as which enzymes have preferences for different car- used industrialEscherichia coli strains (BL21 (DE3), C, Crooks, boxylic-acid-containing substrates. By exploring the catalytic DH5α, MG1655, W and W3110) through phenomics, transcrip- limits of these enzymes we are able to provide novel insight tomics, and genome-scale modelling under both aerobic and for the production of diverse bio-aldehydes. anaerobic conditions. Genome-scale metabolic models were generated for each of the strains and integrated with the phe- 65. K-Optforce: Strain Design Using Kinetic Information nomic data to determine reactions that carry high flux in each Anupam Chowdhury1, Ali Khodayari2, Costas D. Maranas3 and 3 of the strains. High-flux reactions and their encoding genes Thomas J. Mueller* 1Chemical Engineering, Pennsylvania State University, Univer- were compared to differentially expressed genes in each of the sity Park, PA strains, as well as across the strains, to characterize shared 2Chemical Engineering, The Pennsylvania State University, and strain-specific behaviour. Furthermore, the wild-type phe- University Park, PA notypes of each of the strains were compared to production 3Department of Chemical Engineering, The Pennsylvania State phenotypes for a variety of industrial compounds to make a University, University Park, PA prediction on what strains are inherently better suited to pro- Existing computational strain-design approaches relying solely duce a given compound or class of compounds. The result of on stoichiometry and on-off regulation ignore the effects of the study is a classification that can be used to guide selection metabolite concentrations, enzyme activity and substrate-level of a given strain for the biosynthesis of a desired product. enzyme regulation while identifying metabolic interventions. In this work, we implemented the recently developed k-OptForce 64. Computationally Guided Characterization of procedure, which integrates the available kinetic descriptions Carboxylic Acid Reductases for Expanding Aldehyde of metabolic reactions with stoichiometric models, to sharpen Bioproduction the prediction of intervention strategies for improving the bio- Matthew Moura*, Stephen Lenzini, Dante Pertusi, Linda J. production of a chemical of interest. This procedure seam- Broadbelt and Keith E.J. Tyo Department of Chemical and Biological Engineering, North- lessly integrates the mechanistic detail afforded by kinetic western University, Evanston, IL models within a constraint-based optimization framework tractable for genome-scale models. Instead of relying on While industrial biosynthesis has primarily focused on overpro- surrogate fitness functions such as biomass maximization duction of naturally-occuring metabolites, there is an urgent or worst-case simulation for predicting flux re-directions, k- need to expand our biochemical capabilities toward chemical OptForce uses kinetic rate expressions to (re)apportion fluxes

37 Poster Abstracts

in the metabolic network. The interventions suggested by k- framework paves the way for an integrated analysis of kinetic OptForce are comprised of both direct enzymatic parameter and stoichiometric models and enables elucidating system- changes (for reactions with available kinetics) and indirect wide metabolic interventions while capturing regulatory and reaction flux changes (for reactions with only stoichiometric kinetic effects. information). In some cases, additional modifications are needed to overcome the substrate-level regulations imposed 66. Rational Design & Optimization of a Synthetic by the representative kinetic model. The mechanism of action Entner-Doudoroff Pathway for Improved & Controllable NADPH Regeneration of these modifications is often subtle by alleviating substrate Chiam-Yu Ng*, Iman Farasat, Costas D. Maranas and Howard inhibition or draining away cofactors from competing path- Salis ways. In other cases, kinetic expressions shape flux distribu- Department of Chemical Engineering, The Pennsylvania State tions so as to favor the overproduction of the desired product University, University Park, PA requiring fewer direct interventions. NADPH is an essential cofactor for the biosynthesis of several k-OptForce requires as input kinetic expressions that accurate- high-value chemicals, including isoprenoids, fatty acid-based ly capture the substrate-level regulation of metabolic fluxes. fuels, and biopolymers. Tunable control over all potentially To this end, we constructed a kinetic model of E. coli core rate-limiting steps, including the NADPH regeneration rate, will metabolism that satisfies the fluxomic data for wild-type and be crucial to maximizing production titers. We have engineered seven mutant strains by making use of the recently introduced a synthetic Entner-Doudoroff (ED) pathway that increases the Ensemble Modeling (EM) concepts. This model consists of NADPH regeneration rate by 10-fold, compared to a wild-type 138 reactions, 93 metabolites and 60 substrate-level regulatory E. coli MG1655 strain, exceeding alternative approaches. To interactions and accounts for glycolysis/gluconeogenesis, pen- do this, we employed the newly developed Operon Calculator tose phosphate pathway, TCA cycle, major pyruvate metabo- to design the 5-enzyme pathway, removing all cryptic genetic lism, anaplerotic reactions and a number of reactions in other elements and optimizing sequences for maximum expression parts of the metabolism. Parameterization of the model was control. We integrated the synthetic bacterial operons into performed using a formal optimization algorithm that minimizes the E. coli MG1655 EcNR2 genome, and employed MAGE uncertainty-scaled discrepancies between model predictions genome mutagenesis to introduce ribosome binding site muta- and flux measurements. The predicted fluxes by the model tions. To direct genome mutagenesis, we employed the RBS are within the uncertainty range of experimental flux data for Library Calculator to identify the smallest combinatorial muta- 78% of the reactions (with measured fluxes) for both the refer- tion library that maximized coverage of the 5-dimensional ex- ence (wild-type) and seven mutant strains. The remaining flux pression level space. We quantitatively measured the NADPH predictions fall within three standard deviations of measured regenerate rate of genome variants using a NADPH-dependent values. The predicted metabolite concentrations are also within fluorescent protein, a reporter assay using a NADPH-sensitive uncertainty ranges of metabolomic data for 68% of the me- promoter, and a furfural reductase absorption assay. Com- tabolites. Converting the EM-based parameters into a Mi- bining sequences and measurements, we are developing a chaelis-Menten equivalent formalism revealed that 80% of Km sequence-expression-activity model to forward-design new and kcat parameters are within one order of magnitude of the ED pathways with proportional control over a wide range of literature available values. A leave-one-out cross-validation test NADPH regeneration rates. The ability to rationally control the performed to evaluate the predictive capability of the model NADPH regeneration rate will substantially improve the biore- also showed good agreement with test experimental data. newable production of diverse chemical products.

Application of k-OptForce for overproduction of bio-chemicals 67. Strategies for Improving Renewable Phenol Biosynthesis in Engineered Escherichia coli recapitulated existing intervention strategies, while identify- Brian Thompson and David R. Nielsen* ing additional and alternate manipulations for improving the Chemical Engineering, Arizona State University, Tempe, AZ yield of target chemical. k-OptForce identified key regulatory bottlenecks preventing the redirection of flux towards the tar- Phenol is a non-renewable petrochemical building block used get chemical, and suggested manipulations, often non-intuitive widely in the production of fine and commodity chemicals (e.g., and distal to the point of regulation to overcome them. This bisphenol A, caprolactam, and salicylic acid), as well as for a

38 Poster Abstracts

number of important plastics (e.g., phenolic resins). Whereas suggested building blocks, which can be prepared biologi- a renewable route to phenol from glucose-derived tyrosine cally, for synthesis of bio-polyamides. 5AVA is a metabolite of via tyrosine phenol lyase has previously been demonstrated, the so-called aminovalerate pathway of L-lysine catabolism in the pathway and overall strategy suffers from several inherent Pseudomonas putida. In the aminovalerate pathway, L-lysine limitations. Most notable among these is that the key path- is converted to 5AVA via 5-aminovaleramide by lysine 2-mo- way enzyme, namely tyrosine phenol lyase (tpl), suffers from nooxygenase and delta-aminovaleramidase encoded by the both equilibrium and feedback inhibition limitations that limit davB and davA genes, respectively. metabolite flux and achievable product titers. With the goal of improving renewable phenol bioproduction, we present here In this study, we have developed a hybrid process composed our most recent progress related to understanding and ad- of biological production of 5AVA and chemical synthesis of dressing these and other critical factors. To this end, we have nylon 6,5. A recombinant Escherichia coli strain expressing focused on the engineering of feedback-resistant tpl mutants P. putida davAB genes was employed in large-scale produc- with the aid of a novel, high throughput screening assay that tion of 5AVA by fed-batch fermentation. A cyclization reaction couples apparent enzyme activity with growth rate in minimal was conducted to convert the biologically prepared 5AVA into media. Isolated mutants are also being reversed engineered of δ-valerolactam. Finally, nylon 6,5 could be synthesized by to provide an improved understanding of the evolved phe- bulk polymerization of δ-valerolactam and ε-caprolactam. notypes. Furthermore, through complementary efforts have also engineered two alternative and non-natural pathways [This work was supported by the Technology Development linking phenol biosynthesis with native function of the shikimic Program to Solve Climate Changes (Systems Metabolic acid biosynthesis pathway. Unlike the tpl-derived pathway, Engineering for Biorefineries) from the Ministry of Science, ICT both novel pathways terminate with irreversible, phenylacrylic and Industrial Strategic Technology Development Program acid decarboxylase reactions so as to drive flux towards the (10047910, Production of biobased cadaverine and polymer- final product. Both routes were systematically constructed in ization of Bio-polyamide 510) funded By the Ministry of Trade, Escherichia coli following screening studies to identify the most industry & Energy(MI, Korea).] robust pathway enzymes resulting in initial shake flask titers of 50 mg/L. Although all three routes possess identical theoreti- 69. Metabolic Flux Analysis of Isopropyl Alcohol- cal yields, comparison of the relative pathway performance Producing Escherichia coli Nobuyuki Okahashi*1, Katsunori Yoshikawa1, Yoshiko Matsumoto2, provides interesting insights into pathway design strategies for Tomokazu Shirai3, Fumio Matsuda1, Takashi Hirasawa1,4, Chikara non-natural chemical production. Current strategies to increase Furusawa1,5, Mitsufumi Wada2 and Hiroshi Shimizu1 final product titers through all three routes include pathway 1Information Science and Technology, Osaka University, Osaka, Japan de-bottlenecking by development of optimal expression strate- 2Synthetic Chemicals Laboratory, Mitsui Chemicals Inc. gies and re-engineering of the host metabolism to maximize 3Biomass Engineering Program, Riken, Kanagawa, Japan 4 production and availability of respective precursors. Bioscience and Biotechnology, Tokyo institute of Technology 5Quantitative Biology Center, Riken 68. Synthesis of Nylon 6,5 from Biologically Prepared AVA By Metabolically Engineered Escherichia coli The replacement of petrochemicals with biofuels produced Young Hoon Oh*1, Si Jae Park2, Bong Keun Song1, Jonggeon from renewable resources has attracted growing attention. The Jegal1, Seung Hwan Lee1 and Sang Yup Lee3 dehydration of isopropyl alcohol (IPA), a secondary alcohol, 1Korea Research Institute of Chemical Technology, Daejeon, produces a useful material, propylene. Its polymerized form, South Korea polypropylene, is widely used to manufacture commodity 2Department of Environmental Engineering and Energy, Myongji plastics because of its superior characteristics such as high University, Yongin, South Korea 3Chemical & Biomolecular Engineering, Korea Advanced Insti- mechanical strength and heat and chemical resistance. To tute of Science and Technology (KAIST), Daejeon, South Korea accomplish the bioproduction of IPA, we developed an IPA- producing strain of Escherichia coli by expressing heterologous Production of bio-based polymers has been attracting atten- acetoacetate decalboxylase and IPA dehydrogenase. To fur- tions as fossil fuel depletion and environmental pollution are ther increase IPA production, we constructed an Entner-Dou- accelerated. 5-aminovalerate (5AVA) is one of the recently doroff (ED) pathway-dependent IPA-producing strain (hereafter

39 Poster Abstracts

termed “ED strain”) by enhancing the NADPH supply required 71. Identifying Bottlenecks in Engineering Efficient for IPA production. In the ED strain, pgi and gnd were deleted Cellobiose Metabolism (Evidence for putative promoters to catabolize glucose via the NADPH-generating oxidative within operon and TCA cycle imbalance) Vinuselvi Parisutham* and Sung Kuk Lee pentose phosphate (PP) pathway. An understanding of the School of life sciences, Ulsan National Institute of Science metabolic state of the cell provides information that is valuable and Technology, Ulsan, South Korea in developing metabolic engineering strategies. 13C metabolic flux analysis (MFA) is a powerful tool to quantify the intracellular Despite the availability of sophisticated genome editing tools, carbon flow. In this study, we performed MFA in the ED strain it is a seemingly difficult task to optimize the metabolic path- to elucidate its metabolic state. way and obtain mutants with desired functions. Often the failure to conquer the complexity of biological systems is because The ED strain was aerobically cultured at 30°C in M9 minimal of the gaps in our understanding of the regulatory mechanisms medium containing 13C-labeled glucose, which is composed governing the system under investigation. For example, even of [1-13C] and [U-13C] glucose in the ratio of 1:1. For MFA, with a well-studied organism (i.e. Escherichia coli) about 50% cells were harvested at the mid-exponential phase and 13C of the regulatory mechanisms remain unknown. Indeed, it enrichments of proteinogenic amino acids were measured would take another decade to reduce the percentage of un- by gas chromatograph mass spectrometry. Analysis of 13C knowns from 50% to 45%. enrichment of alanine indicated that glucose was catabolized through the ED pathway. Based on the MFA results, the most While the main purpose of such complex regulatory mecha- of the glucose was estimated to flow into the ED pathway via nisms is to choose the best carbon source that demands a the PP pathway. Moreover, fluxes of the reactions correspond- minimal cellular resource for a maximal growth, metabolic ing to the deleted genes pgi and gnd were very small. This engineers strive to engineer cells with deregulated metabo- result indicated that the metabolic flux distribution in the upper lism for cellulose, a complex carbon source that requires the part functioned as expected. In the TCA cycle, low flux through investment of a vast amount of cellular resources for cellulase the glyoxylate shunt and high flux through the TCA cycle were production. The increase in demand for the production of estimated, in agreement with previous MFA results for wild value-added biochemicals from lignocelluloses urges to an- type E. coli grown under aerobic culture conditions. nihilate the regulatory networks (rather than to decipher them) to efficiently express several cellulases. Understanding the 70. The D494G Point Mutation in the Bifunctional Alco- molecular mechanism of cellulose utilization in native cellu- hol and Aldehyde Dehydrogenase (adhE) of Clostridium lolytic organisms helped envision the fact that celluloses are Thermocellum Leads to Improved Ethanol Production Daniel Olson* metabolized in the form of cellodextrins and cellobioses (rather Thayer School of Engineering, Dartmouth College, Hanover, than as glucoses), in order to save the resources spent on NH secreting one of the enzymes, β-glucosidase. Recently, E. coli Previously we have engineered Clostridium thermocellum is engineered for internal cellobiose metabolism using cryptic for increased ethanol production by eliminating lactate and genes of E. coli. acetate production. The resulting strain did not grow well and carbon flux was diverted to amino acid production In native E. coli, cellobiose metabolism is pertained to two instead of ethanol. The strain was evolved by serial transfer cryptic operons: the chb operon and the asc operon. Since and ethanol production increased. To understand the reason promoters of these two operons are relatively insensitive to cel- for this change, the strain was resequenced and compared lobiose, wild-type E. coli is unable to grow on cellobiose as a to the unevolved strain. A SNP in the bifunctional alcohol/ sole carbon source. We replaced the native promoters of these aldehyde dehydrogenase adhE was found that creates a two operons with synthetic constitutive promoters and the D494G substitution in the amino acid sequence. The muta- resulting strain (SVC01) acquired the ability to grow on cellobi- tion appears to alter the cofactor specificity for the alcohol ose. But the growth rate of SVC01 was very low (μ= 0.21/hr). dehydrogenase reaction and may explain the reason for the Adaptive evolution on cellobiose for 30 days helped in enhanc- additional ethanol production. ing the growth rate of SVC01 (μ= 0.40/hr).

40 Poster Abstracts

Even when E. coli is engineered to metabolize cellobiose of cells grown on these carbon sources were compared. In- efficiently, their cellobiose metabolic rate was lower than the terestingly, the expression of glyoxylate pathway enzyme was glucose metabolic rate. In addition, these strains suffered very high whereas the expression of ED pathway enzyme was from repression of cellobiose metabolism in the presence of reduced in cellobiose grown cells compared to cells growing glucose thus failing to mimic the native cellulolytic organisms. on glucose minimal medium. The expression level of glyoxylate Here, we identified two major bottlenecks in efficient cellobiose pathway protein, AceB was followed by western blotting. As metabolism: 1) regulation of the internal putative promoters of expected, the glyoxylate pathway in the adapted strains was asc operon and 2) difference in the pathway for the metabo- constitutively expressed at a higher level in cells growing on lism of glucose and cellobiose. We show that it is possible to cellobiose-minimal medium when compared to cells grown on rewire the regulations of the cryptic genes of E. coli in order to glucose-minimal medium. increase the rate of cellobiose metabolism. This result indicates an imbalanced distribution of flux around Uncharacterized putative promoters within asc operon: the glyoxylate pathway and the ED pathway in cellobiose when Use of synthetic constitutive promoters or modularized as- compared to glucose metabolism. Rerouting the flux of cel- sembly of genes as an operon became a potent alterna- lobiose metabolism to mimic the glucose metabolic pathway tive to circumvent the host-dependent regulations such might be one way to optimize the metabolic pathway. Hence, as stationary phase effect, sigma-factor dependence or the flux distributions between these two pathways were altered carbon catabolite repression of the desired pathway. Here, through the modulation of the promoter strength of the genes we show with cellobiose metabolism as a proof-of-concept encoding these two pathways. Thus, the putative promoter that despite the use of synthetic promoters or operons, the in the asc operon and imbalanced flux distribution between presence of putative, uncharacterized promoters within an ED pathway and the glyoxylate pathway are two major bottle- operon is another bottleneck in engineering the metabolic necks controlling efficient cellobiose metabolism in E. coli. In pathways. The mutations responsible for the enhanced this study, we highlight that the putative promoters within an growth rate of SVC01 were biased around the RBS of ascB operon could be an important factor to be considered in gene of the asc operon. Optimization of the RBS of all genes metabolic engineering practices. of the two operons and their promoters through single stranded oligo-mediated recombineering also enhanced the 72. Building Metabolic Engineering Tools to Better Un- growth rate (μ= 0.32/hr) of SVC01. The asc operon encodes derstanding Product Production from Microbial Sources: Using the Cyanobacterium Synechocystis Sp. PCC 6803 two genes where ascF, the first gene of the operon, encodes for Astaxanthin Production the incomplete cellobiose transporter and the second gene, Stevan Albers1 and Christie A.M. Peebles*2 ascB encodes the enzyme β-phosphoglucosidase. Analysis 1Department of Cell and Molecular Biology, Colorado State of the mRNA level between SVC01 and the adapted strain University, Fort Collins, CO indicates no change in the expression of the first gene, ascF 2Chemical and Biological Engineering, Colorodo State but a 5-fold increase in the expression of the second gene, University, Fort Collins, CO ascB. The differential expression of the first and the second gene intrigues the possibility of an internal promoter in this Product production in microorganisms has great potential to operon. A promoter region was identified between -70bp and fulfill many industries requirements for biologically derived -570bp from the start codon of ascB gene. The regulations molecules. Because of the simplicity of bacteria, cyanobacteria of this putative promoter were being deciphered in order to make good candidates as a platform for product production. enhance the growth rate of SVC01 on cellobiose. For photosynthetic organisms to fulfill this potential, gene expression will need to be manipulated, controlled, and ultimately Difference in the routes for metabolism of glucose and managed within the cell. Currently, metabolic engineers utilize cellobiose molecular tools that are poorly understood and therefore con- We hypothesized that the cellobiose metabolic rate could tribute to inconsistent results and product titers. The objective be enhanced, if cellobiose is metabolized through the same of my research is to better understand how commonly utilized pathway as that of glucose. In order to identify the pathway promoter constructs operate during typical environmental difference between cellobiose and glucose, the total proteome conditions within cyanobacterial cells.

41 Poster Abstracts

Using the photosynthetic cyanobacterium Synechocystis sp. the expression platform. This unblocks the RBS which allows PCC 6803, we have been able to successfully quantify gene translation of the gene located downstream. expression from several suites of promoter constructs. We have developed and tested variants based on the commonly In this work, several types of riboswitches are designed and used E. coli tac suite of promoters, which includes an inducible evaluated using a fluorescent protein as reporter gene. Since construct capable of inducing gene expression in Synechocys- gene regulation by riboswitches requires conditional base- tis. Additionally, we have quantified expression of promoters pairing, RNA secondary structure prediction algorithms are commonly believed to control gene expression during circa- used during the design phase. As such, computational tools dian rhythms. As a proof of concept, we have begun to use are used to determine suitable RNA sequences with the de- these promoter constructs to engineer the carotenoid biosyn- sired properties. After designing different riboswitch configu- thesis pathway found in Synechocystis. This poster will detail rations the various designs were tested in vivo to determine the specifics related to our research. their performance.

73. Steering Prokaryotic Gene Expression Using From the performance of the different designs systemic rules Engineered Riboswitches are derived which can lead to a generic guideline for the Gert Peters*1, Jo Maertens1, Joeri Beauprez1, Jeroen Lammertyn2 and Marjan De Mey1 design of riboswitches in general. Ultimately, this approach 1Inbio.be, Department of Biochemical and Microbial can be used for the design of tailor made riboswitches with Technology, Ghent University, Ghent, Belgium excellent performance. 2MeBioS, Biosystems Department, University of Leuven, Leuven, Belgium The authors would like to acknowledge Inbiose for their contributions. Several techniques originating from synthetic biology have a big impact on the field of metabolic engineering. These 74. Sensor-Selector Strategy for Directed Evolution of methods are successfully being used to reprogram cell Biosynthetic Pathways behaviour for example by constructing genetic circuits or Srivatsan Raman*1 and George Church2 introducing heterologous biosynthetic pathways. In this 1Synthetic Biology, Wyss Institute, Harvard University, context, specific and controlled gene expression based on Boston, MA 2 the presence of a certain molecule is of great value. Such Department of Genetics, Wyss Institute, Harvard University, Boston, MA regulatory elements are abundantly present in nature.

Metabolites biosynthetically produced in nature hold enor- Most of the engineered devices use the post-transcriptional mous potential for the sustainable production of chemicals, level (RNA) to exert control over gene expression because materials and therapeutics. This potential remains largely of the versatility and designability of RNA. Despite these unfulfilled because technological limitations severely restrict advantages, the engineering process of these RNA devices the types of metabolites that can be readily detected and remains very challenging and relies mostly on high-through- the number of biosynthetic pathway prototypes that can be put screening. The current lack of in-depth knowledge limits evaluated. Thus, despite advances in genetic diversification the widespread application of these promising devices. methods, phenotype evaluation remains the major rate- limiting step. Here we present a broadly extensible synthetic Here, we propose a framework for the design and construc- selection system, known as sensor-selector, that simulta- tion of riboswitches. These RNA devices exhibit translational neously expands our phenotype evaluation capabilities in control by conditionally blocking the ribosome binding site two ways: 1) a wide range of metabolites are detectable by (RBS) when the specific molecule is absent and releasing drawing on the diversity of natural protein and RNA sensors this blockage when the same molecule becomes available. and 2) directed evolution is harnessed to select the highest- For this purpose, riboswitches are composed of two struc- producing biosynthetic pathways from libraries orders of turally linked domains: an aptamer and a gene expression magnitude larger than current screening methods permit. platform. Upon the ligand binding the aptamer part structur- To overcome the problem of unproductive escape mu- ally rearranges which leads to RNA conformational changes in tants that has previously plagued selection-based pathway

42 Poster Abstracts

optimizations, we employ a toggled selection scheme with 76. Production of Aromatic Compounds in E. coli Strains dual selective markers to robustly eliminate cells capable of Lacking Interconversion of PEP and Pyr When Glucose circumventing selection. We characterize sensor-selectors and Acetate Are Coutilized Andrea Sabido*1, Juan Carlos Sigala*2, Georgina for 15 metabolites belonging to diverse classes – polyketide Hernández-Chávez1, Noemí Flores1, Guillermo Gosset1 and antibiotics, flavonoids, polymer-precursor diacids, vita- Francisco Bolívar1 mins, long-chain alkanes and sugars. We have developed a 1Ingeniería Celular y Biocatálisis, Instituto de Biotecnología- general suite of genetic interventions capable of expanding UNAM, Cuernavaca, Morelos, Mexico library sizes by four orders of magnitude and programming 2Procesos y Tecnología, Universidad Autónoma Metropolita- sensor-selectors to function across user-specified metabolite na-Cuajimalpa, D. F., Mexico concentration ranges. Evaluating more than a billion cells per Phosphoenolpyruvate (PEP) is a precursor involved in the day, we deploy sensor-selector to optimize biosynthesis of biosynthesis of aromatics and other valuable compounds in the industrially useful metabolite, naringenin, yielding nearly E. coli. However, the PEP:carbohydrate phosphotransferase 50 fold improvement, through four iterations of directed evo- system (PTS), is the largest PEP consumer. In this regard, our lution. Sensor-selectors allow multiplexed phenotype evalu- group has generated E. coli JM101 mutants devoid of PTS ation, obviating the need to characterize designs individually, by deleting the ptsHIcrr operon (PB11 and PB12 strains), a thereby facilitating rapid design-build-test cycles in biological strategy that, in theory, may double PEP availability. In these engineering. We envision further application of sensor-selec- ptsHIcrr- strains, the glycolytic and gluconeogenic pathways tors to metagenomic pathway discovery, enzyme engineer- function simultaneously, allowing the coutilization of second- ing, and the evolution of robust genetic circuits. ary carbon sources in the presence of glucose due to the Glc 75. Rapid Evaluation of Itaconic Acid Production absence of the EIIA component. Taking into account this Strategies in Saccharomyces cerevisiae capacity, the physiological and transcriptional response of Zheng Zhao*, Ben Meijrink, Bianca Gielesen, Burhan Ozalp, blocking carbon skeletons interchange between PEP and Rob van der Hoeven, Liang Wu, Roel Bovenberg and Hans pyruvate (PYR) in these ptsHIcr- strains was investigated by Roubos deleting the pykA, pykF and ppsA genes, during simultane- DSM Biotechnology Center, Delft, Netherlands ous utilization of glucose and acetate. It was shown that under this condition, in the PB11 pykAF- ppsA- strain glycoly- The recent explosion of strain optimization algorithms based sis and the TCA cycle appear to coexist independently. The on genome-scale metabolic models (GSMM) allows for expression profile of this derivative showed that all metabolic rapid generation of a large number of metabolic engineering central pathways are downregulated in the mixture. Appar- strategies for maximizing metabolite production. However, ently, the increase in PEP availability could inhibit some gly- implementation and validation of these strategies remains colytic genes. In contrast, a partial separation of glycolysis a bottleneck in the design-build-test-learn cycle. Here we and the TCA cycle was achieved in the PB12 pykAF- ppsA- demonstrate a method for targeted high-throughput strain strain, which upregulates the aceBAK operon and the sfcA transformation using exchangeable building blocks. Using gene in order to reroute the local flux towards the synthesis this method and a high-throughput analysis platform, we of PYR. In order to determine the effects of the modifications evaluated multiple itaconic acid producing strategies simul- at the PEP-PYR node on PEP availability, ptsHIcrr- pykAF- taneously in Saccharomyces cerevisiae wild type strains. The ppsA- engineered derivatives were generated and tested for initial design included differences in compartmental strategy, total aromatic compounds (TAC) production. The engineered a set of gene variants, differences in promoter strength and PB12 pykAF- ppsA- tyrR- pheAev2+ pJLB aroGfbr tktA de- two strain backgrounds. The results revealed further poten- rivative achieved a 4-fold higher TAC yield on glucose and tial for improving the productivity of itaconic acid. Overall, we acetate (YTAC/Glc+Ace) compared with its control strain, demonstrate that rapid prototyping by targeted engineering representing 65% of the theoretical maximum. In contrast, has become a valuable tool to quickly explore metabolic en- in the PB11 pykAF- ppsA- tyrR- pheAev2+ /pJLB aroGfbr tktA gineering scenario’s including host strains for the production derivative there was no benefit on aromatics production since of a heterologous product. this strain reduced its qGlc by 47%, and it could cause lower intracellular PEP concentrations. However, when we overex-

43 Poster Abstracts

pressed the glk and galP genes in the engineered PB11 de- can be finely tuned for optimal or desired level performance, rivative in order to increase glucose consumption, the PB11 in the manner of a transistor where a small current change is pykAF- ppsA- tyrR- pheAev2+/pJLB aroGfbr tktA/pv5GalP5Glk used to control a big current flow. strain increased 3-fold its YTAC/Glc+Ace compared with its control strain, representing 48% of the theoretical maximum. 78. Ubiquinone Accumulation Improves Osmotic-Stress Furthermore, the higher q of the former allowed an increase Tolerance in Escherichia coli Glc Daniel C. Sevin* and Uwe Sauer of 6-fold in the YTAC/ +Ace with respect to the PB11 pykAF- Glc Institute of Molecular Systems Biology, ETH Zurich, Zurich, - - ev2+ fbr ppsA tyrR pheA /pJLB aroG tktA, the same derivative Switzerland without the glk and galP genes. Microbial production organisms in bioprocesses are often 77. Metabolic Transistor Strategy for Controlling challenged by medium conditions that unfavorably change Electron Transfer Chain in Escherichia coli over time. High product concentrations achieved by mod- Hui Wu1,3, Leepika Tuli1, George N. Bennett2, Ka-Yiu San1,2* 1Department of Bioengineering, Rice University, Houston, ern production strains – often a prerequisite for economic Texas feasibility – may hamper growth and productivity of the 2Department of Chemical and Biomolecular Engineering, biocatalyst, thus reducing overall process efficiency. Besides Rice University, Houston, Texas product-specific inhibitory effects, general issues caused by 3 State Key Laboratory of Bioreactor Engineering, East China product accumulation are solvent stress frequently encoun- University of Science and Technology, Shanghai 200237, tered in biofuel production and osmotic stress imposed by China up to molar concentrations of produced bulk chemicals. A detailed understanding of the microbial response mecha- A novel metabolic transistor was built to finely control a large nisms to such general stresses can guide metabolic en- metabolic flux by a small change in the level or availability of gineering strategies by identifying cellular targets, thereby a key participant in the large flux. The fine-tuning of the key leading to more effective production organisms. participant in the large flux can be accomplished by adding a competitive reaction of a precursor or an intermediate A principal mechanism microbes use to cope with osmotic in the biosynthetic pathway of the key participant. For use stress is to adjust their intracellular osmolality by accumu- of oxygen there is the electron transport chain (ETC), also lating compatible solutes. To gain detailed insights into this called the electron transport system (ETS), in E. coli and process, we analyzed the global metabolic response of the many industrially important organisms. This ETC contains bacterium Escherichia coli to salt-induced hyperosmotic various cytochromes and electron carriers, such as qui- stress using nontargeted high-resolution mass spectrom- nones- ubiquinones. Our recent strategy, which can control etry (ref. 1), and unexpectedly found the respiratory electron the level of oxidation via the ETC, is through regulating the carrier, antioxidant and isoprenoid lipid ubiquinone-8 (Q8) production of the quinone by competing for an intermediate as the by far most accumulating metabolite. Using various within its biosynthetic pathway. One approach is using the in vivo and in vitro experiments, we demonstrated that Q8 is lycopene synthesis pathway to drain the isopentenyl diphos- required for acute and sustained osmotic-stress tolerance phate (IPP) pool; Another strategy is fine-tuning the reaction and that the main mechanism of Q8-mediated osmoprotec- catalyzed by the geranyl diphosphate:4-hydroxybenzoate tion is the stabilization of the cytoplasmic membrane. Thus, geranyltransferase from Lithospermum erythrorhizon (leP- we find that besides regulating intracellular osmolality, E. GT-1) to drain both the isopentenyl diphosphate (IPP) and coli enhances the stability of its cytoplasmic membrane to 4-hydroxybenzoic acid (4-HB) pools. The achievement of a withstand long-term osmotic stress (ref. 2). theoretical yield on lactate production under aerobic conditions via this metabolic control strategy upon manipulation of To investigate how widespread osmotic-stress induced quinone synthesis pathway in E. coli strain, thus provides an ubiquinone accumulation is, we are currently analyzing the in vivo means to genetically control the activity of the elec- metabolic responses of diverse prokaryotic and eukaryotic tron transfer chain and manipulate the production of reduced microbes, including several established production hosts. products. The advantage of this metabolic control approach Furthermore, we are extending our research to a range of is that it uses very little cell energy, protein quantity, and additional stresses including temperature and solvent stress carbon flux to control the major metabolic flux of the cell and 44 Poster Abstracts

to understand whether ubiquinone accumulation contributes acid was observed, indicating towards the presence of pas- to general stress tolerance. At the conference, we will report sive or facilitated diffusion of the undissociated species of our findings and provide a perspective on how they might fumaric acid. It was found that the permeability coefficient contribute to future metabolic engineering approaches. of the cells for fumaric acid was 10 times higher than for References succinic acid. From flux analysis using stoichiometric model, we found that no additional ATP dissipation occurred at high Ref. 1: Fuhrer T., Heer D., Begemann B. & Zamboni N. High- throughput, accurate mass metabolome profiling of cellular extracts extracellular concentrations of undissociated fumaric acid by flow injection-time-of-flight mass spectrometry. Analytical in the wild type strain. This indicates the absence of futile Chemistry 83, 7074-7080 (2011). cycling and thus the absence of an active export mechanism Ref. 2: Sévin D. C. & Sauer U. Ubiquinone accumulation improves of fumaric acid in wild strain. Implications of our experimen- osmotic-stress tolerance in Escherichia coli. Nature Chemical Biol- tal results for the industrial scale production of fumaric acid ogy, published online ahead of print (2014). will be discussed.

79. Transport and Metabolism of Fumaric Acid in 80. Implementing the Formose Pathway for Conversion

Saccharomyces cerevisiae of Electricity and CO2 to Biofuel Precursors Via Formate Mihir Shah*, J.J. Heijnen and Walter Gulik in Escherichia coli Department of Biotechnology, Delft University of Technology, Amanda Lee Smith*1, Justin Siegel2,3, Adam Wargacki2, Delft, The Netherlands Jacob Bale4, David Baker2,3 and Mary E. Lidstrom1,5 1Chemical Engineering, University of Washington, Seattle, Currently, research is focussed on the production of fumaric WA 2 acid and other relevant organic acids via fermentation using Biochemistry, University of Washington, Seattle, WA 3Biomolecular Structure and Design Program, University of Saccharomyces cerevisiae as preferred organism because Washington, Seattle, WA of its robustness at low pH and the availability of genetic 4Molecular and Cellular Biology, University of Washington, tools for metabolic engineering. Industrial fermentation of Seattle, WA fumaric acid is advantageous to be carried out at low pH 5Microbiology, University of Washington, Seattle, WA for cost effective downstream processing and reduction of waste production. However, at low cultivation pH most of Engineering fuel-producing microorganisms that utilize the fumaric acid is present in undissociated form, which can renewable energy in the form of electricity could reduce diffuse through the plasma membrane into the cytoplasm. dependence on water and land resources. Here we pres- In the presence of an efficient export mechanism for fumaric ent a metabolic pathway module in Escherichia coli with the acid this will lead to futile cycling resulting in ATP dissipation, potential for reducing the life cycle greenhouse gas emis- and will significantly reduce the product yield. Therefore it is sions of microbial biofuels. In contrast to other naturally relevant to study the importance of futile cycling in the pres- occurring carbon dioxide fixation pathways, this proposed ence of high concentration of undissociated fumaric acid chemotrophic system, the Formose Pathway, is the shortest and to get insight into the transport mechanism and metabo- and one of the most efficient pathways ever realized. The lism of fumaric acid in S. cerevisiae. Formose Pathway consumes formate, which can be generated at high yield from carbon dioxide, the most abundant In this work we studied the uptake mechanism and me- greenhouse gas, and neutral water through electrochemical tabolism of fumaric acid at a pH of 3.0 in a glucose limited means. The formate is partitioned to provide NADH in addi- aerobic chemostat at a dilution rate of 0.10 h-1 . Our study tion to carbon flux, which flows through two steps catalyzed was done using the wild type strain of S. cerevisiae CEN. by overexpressed enzymes, acylating acetaldehyde dehydro- PK 113-7D. It was observed that fumaric acid was efficiently genase and acetyl-CoA synthetase, making formaldehyde. used as a carbon and energy source in the presence of For the carboligation of formaldehyde, we employ a novel glucose, there was significant increase in the steady state enzyme, formolase, which catalyzes the conversion of three biomass concentration and respiratory quotient with increase one-carbon molecules (formaldehyde) into one three-carbon in biomass specific uptake rate of fumaric acid. A linear molecule (dihydroxyacetone), a reaction not yet observed relationship between the steady state extracellular fumaric in nature. Dihydroxyacetone can be phosphorylated to acid concentration and the specific uptake rate of fumaric dihydroxyacetone phosphate, a central metabolic interme-

45 Poster Abstracts

diate, allowing this pathway to plug into many preexisting the glyoxylate shunt. In addition, the three known fumarase biofuel production pathways. This pathway was previously genes (fumA, fumB and fumC) were also deleted to enhance established using purified proteins, which revealed pathway fumaric acid formation. The resulting strain was able to pro- bottlenecks. These have been addressed, and the modified duce 1.45 g/L of fumaric acid from 15 g/L of glucose in flask version has been introduced as a complete pathway into E. culture. This base strain was further engineered by plasmid- coli. Clarified lysates of this strain have been demonstrated to based overexpression of the native ppc gene, encoding convert formate into central metabolic intermediates, show- phosphoenolpyruvate carboxylase (PPC), based on in-silico ing that the Formose pathway is functional in a single strain. aided prediction strategy, which resulted in the production of 4.09 g/L of fumaric acid. And then, the arcA and ptsG genes 81. Engineering the Valine Assimilation Pathway to were sequentially deleted to reinforce the oxidative TCA cy- Produce Biochemicals and Fuels in S. cerevisiae cle flux, and the aspA gene was deleted to block the conver- Kevin Solomon*, Elisa Ovadia and Michelle A. O’Malley sion of fumaric acid into L-aspartic acid. Since it is desirable Department of Chemical Engineering, University of California, Santa Barbara, Santa Barbara, CA to avoid the use of inducer, the lacI gene was also deleted. The native promoter of the galP gene was replaced with Growing energy demand and sustainability concerns the strong trc promoter to increase glucose uptake rate and have stimulated a renewed interest in the engineering of fumaric acid productivity. Fed-batch culture of the final strain branched-chain amino acid assimilation pathways. The Eh- CWF812 allowed production of 28.2 g/L fumaric acid in 63 rlich pathway, in particular, has been the subject of recent ef- h with the overall yield and productivity of 0.389 g fumaric forts for its ability to produce branched-chain alcohols such acid/g glucose and 0.448 g/L/h. This study demonstrates as isobutanol. In this study, we explore alternative valine the possibility for the efficient production of fumaric acid by assimilation routes as a production platform for renewable metabolically engineered E. coli. (Development of systems biochemicals and fuels in the yeast Saccharomyces cerevi- metabolic engineering platform technologies for biorefiner- siae. We were able to successfully reconstitute a heterolo- ies; NRF-2012-C1AAA001-2012M1A2A2026556) funded by gous valine assimilation pathway in yeast and target expres- the Ministry of Education, Science and Technology) sion to the cytoplasm or mitochondria through the use of an N-terminal tag from a mitochondrial ATPase (Su9) found in 83. Engineering the Glycolytic Pathway of E. coli K12 Mutants By Gene Deletions and Introduction of a N. crassa. While pathway in vitro activities were comparable Fructose 6-Phosphate Aldolase after 48h, product titers were observed to be much larger Marcel Wolfer, Katrin Gottlieb, Nathalia Trachtmann and when expression was targeted to the mitochondria as com- Georg A. Sprenger* pared to the cytoplasm. In this poster, I further discuss the Institute of Microbiology, University of Stuttgart, Stuttgart, potential and performance of this platform and examine the Germany effects of pathway localization. Escherichia coli K-12 is a model organism in microbiology 82. Systems Metabolic Engineering of Escherichia coli and genetics. Its genes and enzymes which govern the cen- for the Production of Fumaric Acid tral metabolic Embden-Meyerhof-Parnas (glycolysis) route Chanwoo Song*, Sol Choi, Dong In Kim, Jae won Jang and and pentose phosphate pathway (PPW) are well understood Sang Yup Lee 1,2. The systematic knockout of single genes has furthermore Chemical & Biomolecular Engineering, Korea Advanced Insti- allowed to assess essential reactions in the E. coli central tute of Science and Technology (KAIST), Daejeon, 3,4 South Korea metabolism . Combination of gene knock-outs for enzymes of the upper glycolytic pathway is presented. Triple-negative Fumaric acid is a naturally occurring organic acid which is mutant strains of E. coli K12 W3110 which lack both genes an key intermediate of the tricarboxylic acid cycle. It can be (pfkA, pfkB) for phosphofructokinase (PFK) and the gene pgi used for diverse purposes such as food additives, resins, for phosphoglucoisomerase were created. The triple mutant plasticizers, and precursor for diverse chemicals. In this strain was analyzed (in comparison with its parent strains) for study, Escherichia coli was metabolically engineered for its growth behaviour in LB, various mineral salt media (MM) the production of fumaric acid under aerobic condition. To with both, PTS- and non-PTS substrate, and on MacConkey design optimal production pathway for fumaric acid, firstly agar plates. Whereas growth on LB and MM with fructose the iclR gene was deleted to redirect the carbon flux through as C-source was nearly unaffected in the pfkA pfkB pgi triple 46 Poster Abstracts

mutant, growth on MM with C-sources which are catabo- 84. The Potential of Lactic Acid Bacteria As Microbial lised via the intermediate, fructose 6-P, was either abolished Factory for Pentanol Isomer Production 1 1 2 or severely affected. Growth on glucose was still possible, Karen I. Starlit* , Brian Koebmann , Jan Martinussen , Steen T. Joergensen1 and Peter R. Jensen3 presumably via the shunt to the PPW. The triple mutant was, 1Bacterial Strain Technology, Novozymes A/S, 2880 however, severely inhibited for its growth on MacConkey Bagsværd, Denmark agar plates containing bile acids. Efficiency of plating (eop) 2Dept. of Systems Biology, Technical University of Denmark, on MacConkey agar dropped for at least three orders of 2800 Lyngby, Denmark magnitude compared to the wild type strain. This reflect that 3National Food Institute, Technical University of Denmark the lack of PGI activity leads to an altered composition of the lipopolysaccharide (LPS) envelope of E. coli which protects Due to modern society´s reliance on petroleum-derived fuels these gram-negative bacteria from the effects of bile acids5. and a limited supply of this non-renewable resource, recent The most severe effect could be seen on MacConkey agar approaches have aimed at constructing microbial hosts for plates with 0.5% of mannitol added. The eop was reduced sustainable production of alcohol by engineering a number for more than 4 orders of magnitude in comparison to the of organisms such as Clostridium acetobutylicum, Escherich- wild type, presumably due to the combined effects of cholate ia coli and Saccharomyces cerevisiae. However, lactic acid sensitivity and the accumulation of fructose 6-P due to the bacteria (LAB) may be more suitable as host organisms since block in PFK and PGI activities. they potentially tolerate organic acids, alcohol and low pH better. With a metabolic engineering-, screening–, and adap- We transformed the triple-negative mutant strain with a plas- tive evolution approach this project proposes to utilize C5 and mid-borne copy of the gene (fsaA) which encodes a fructose C6 rich waste streams as substrate for microbial conversion to 6-phosphate aldolase enzyme (FSA)6 variant with improved alcohol. The research plan for elucidating the potential of alco- affinity for fructose 6-phosphate, FSA A129S7. Then the cells hol production in LAB include screening of tolerance towards were subjected to an adaptive evolution experiment in shake alcohol and lignocellulosic inhibitors, adaptive evolution of flasks with minerals salts medium with fructose (0.02%) and LAB for improved alcohol tolerance and metabolic engineering mannitol (0.5%) as second C source. By serial transfers we of both a LAB model strain as well as an industrial strain. By could evolve strains which regained the ability to grow on exploiting the native pathways for amino acids, several higher mannitol as sole carbon source with a generation time of alcohols may be produced upon insertion and overexpression about 4 hours. Data on the in-depth analysis of this strain of an alcohol decarboxylase and a ketoacid decarboxylase. and its further improvement and alterations of the glycolytic A model LAB strain is currently under construction. Two LAB bypass pathway will be presented and discussed. strains were selected for adaptive evolution of which one showed improved tolerance towards a pentanol isomer after References approximately 200 generations of growth in the presence of [1] D.G. Fraenkel, Glycolysis in (Neidhardt FC. et al. eds.) E.coli and increasing selective pressure. Salmonella, ASM Press 1996. [2] G.A. Sprenger (1995) Arch. Microbiol. 164, 324-330 85. Engineering Saccharomyces cerevisiae for the Production of Hexadecanol and Octadecanol [3] T. Baba et al. (2006) Mol. Syst. Biol. 2006.0008 David Stuart*, Xiao Dong Liu and Isabella Wong [4] K.A. Datsenko, B.L. Wanner (2000) Proc.Natl.Acad.Sci. USA 97, Biochemistry, University of Alberta, Edmonton, AB, Canada 6640-45. [5] L. Eidels, M.J. Osborn (1971) Proc.Natl.Acad.Sci. USA 68, The fatty alcohols hexadecanol and octadecanol are widely 1673-1677. used compounds in the cosmetic industry and find extensive [6] M. Schürmann, G.A. Sprenger (2001) J. Biol. Chem. 276:11055- application as platform chemicals. These fatty alcohols oc- 11061. cur in their free form rarely in nature and are produced in very [7] Sprenger, G.A., Schürmann, Me., et al. (2007) C-C-bonding low abundance by a few plants and algae where they are microbial enzymes: thiamine diphosphate-dependent enzymes incorporated into waxes. The primary source of hexadecanol and class I aldolases. pp. 312-326. In Asymmetric Synthesis with and octadecanol is palm and coconut oil. The use of these Chemical and Biological Methods (D Enders, K-E Jaeger, Hg.) food oils for industrial chemical and cosmetic production has Wiley-VCH, Weinheim. lead to extensive deforestation in Asia and competition with

47 Poster Abstracts

the food industry for supply. and not as the pure L- enantiomer. Obviously, this is the consequence of the demanding precursor, energy and reduction We have engineered yeast as a microbial cell factory to equivalent needs for the biosynthesis of L-methionine using produce hexadecanol and octadecanol from sugar, starch, e.g. glucose as carbon source. or cellulosic hydrolysates. Production of free fatty alcohols was achieved by over expressing key enzymes in the fatty This contribution considers metabolic engineering studies acid biosynthetic pathway. This was coupled with deletion for L-methionine formation by glucose-based fermentation. of the POX1 gene encoding fatty acyl-CoA oxidase, the Analyzing network topologies, first, the coupling of cellular first step in degradation of fatty acyl-CoA molecules. These energy management and metabolic performance for L-me- modifications increased fatty acid accumulation in the cells. thionine biosynthesis is outlined considering recombinant C. Conversion of fatty acids and fatty acyl-CoA to fatty alcohol glutamicum and E. coli strains. Constraints for metabolic and was achieved by expression of a thioesterase from E. coli and biochemical engineering are derived. a fatty acyl reductase from mouse. Surprisingly, we found that production could be further improved by over expres- Promising L-methionine producers were investigated by so- sion of FAA3 a fatty acyl-CoA synthetase. These manipula- phisticated stimulus response experiments triggering prod- tions yielded strains capable of producing up to 194 mg/L uct synthesis in vivo. Rapid sampling technologies with fast hexadecanol and octadecanol in rich medium at a ratio of metabolism inactivation were applied to monitor metabolic 47.8% hexadecanol to 52.2% octadecanol. This yield was dynamics as a function of external stimuli. As stimulators dif- further increased by up to 91% in nitrogen poor fermenta- ferent compounds were tested that fulfilled the criteria of (i) tion medium. In screening for strains with greater yields of fast cellular uptake and (ii) significantly strong perturbation of hexadecanol we identified mutants in the SNF2 gene. Snf2 is the complex L-methionine biosynthesis. Based on newly depart of a chromatin-remodeling complex that influences the veloped analytical protocols, LC—QQQ-MS studies enabled expression of many genes so the mechanism by which snf2 the detailed monitoring of intracellular metabolism dynam- mutation increases fatty alcohol synthesis is not clear. Inter- ics providing the basis for the model-based identification of estingly in this strain background the unsaturated C18-1 and metabolic engineering targets. In essence, dynamics of all in- C16-1 are the most abundant fatty acids yet the major fatty termediates of L-methionine biosynthesis were documented alcohols produced are C18-OH and C16-OH, likely reflecting and studied. Tools of metabolic control analysis were applied specificity of the fatty acyl reductase expressed in the cells. to unravel details of metabolic regulation. Thereof strategies We identified mutants that could shift the balance of alcohol for further strain improvement are derived. produced to an increase in octadecanol over hexadecanol. In addition to producing fatty alcohols from refined sugar we L-methionine producers were tested in lab-scale fermenta- show that the strains are capable of using cellulosic hydroly- tions applying e.g. different process scenarios and using sates as a feedstock. various sulphur sources. Process phases were analyzed by metabolic flux analysis. Fermentation performance data 86. L-Methionine Production with Recombinant E. coli revealing e.g molar product/substrate yields >20% will be Ralf Takors*1, Maria Rahnert1, Attila Teleki1, Horst Priefert2 presented that underline the feasibility and attractiveness of 3 and Brigitte Bathe glucose-based L-methionine production. 1University of Stuttgart, Institute of Biochemical Engineering, Stuttgart, Germany 87. Engineering of Artificial Enzyme Complexes 2Evonik Industries AG, HN-BP-RD, Halle, Germany Mediated By Heterospecific Coiled-Coil Zippers or 3Health & Nutrition, R&D Bioproducts, Halle, Germany Synthetic Protein-Protein Interaction Domains in Saccharomyces cerevisiae The sulphur containing amino acid L-methionine (C H1 NO S) belongs to the small group of amino acids that 5 1 2 Cora Mignat*, Thomas Thomik*, Eckhard Boles and are not (yet) produced by fermentation. While current market Mislav Oreb volumes of ~ 800.000 tons per year are highly attractive, the Institute of Molecular Biosciences, Goethe University feed additive is still sold as chemically synthesized D/L-me- Frankfurt am Main, Frankfurt am Main, Germany

thionine or as the methionine hydroxy analog (C5H10O3S) -

48 Poster Abstracts

In metabolic engineering approaches, yield and synthesis for example the carbon substrate degradation pathways rate of a desired product are often compromised by diver- beta-ketoadipate and Entner-Doudoroff are transcription- sion of precursor compounds by competing pathways. One ally regulated [1]. From this finding we deduce ametabolic possibility to circumvent this problem is the so called sub- engineering strategy that relies on metabolic demand: strate channeling, whereby the metabolite is directly passed between the enzymes. An additional benefit of this strategy “High transcriptional activity in a peripheral pathway that is results from increasing the local concentration of the me- translated into high enzymatic activity (high metabolic de- tabolite, which leads to improved substrate-saturation of the mand) allows significant rerouting of metabolic resources to enzymes and, consequently, faster reaction kinetics. the product of choice”.

In order to develop a toolbox for substrate channeling in Indeed, we have evidence from extreme growth conditions engineered S. cerevisiae, we fused several enzymes in- (e.g., growth in the presence of a second phase of octanol) volved in the central carbon metabolism with protein-protein that P. putida can match an increased metabolic demand interaction modules. For establishing a direct interaction by tripling the glucose uptake rate without producing any between enzymes, we used heterospecific coiled-coil zip- side products, thus only biomass and CO2 [2]. We tested this pers. As an alternative approach, modular scaffold proteins, metabolic engineering strategy using rhamnolipid produc- comprising serially arranged protein-protein interaction tion as an example. Rhamnolipid synthesis relies on two domains, can be used to mediate complex formation. This pathways: Fatty acid de novo synthesis and the rhamnose strategy has been successfully employed in E. coli with the pathway, providing the required precursors hydroxyalkanoy- production of mevalonate (Dueber et al. 2009). The effect of loxy-alkanoic acid (HAA) and activated (dTDP-) rhamnose, fusing protein-protein interaction modules to the enzymes respectively. Hence, opposing to single pathway molecules, on their activity and productivity of selected pathways is rhamnolipid synthesis causes demand of two central carbon discussed. metabolism intermediates; glucose-6-phosphate is required for rhamnose synthesis, while the lipid moiety is derived from Financial support by the German Federal Ministry of Food, lipid de novo synthesis, thus consuming acetyl-CoA. Agriculture and Consumer Protection following a decision of the German Bundestag is gratefully acknowledged (project We previously presented a P. putida KT2440 able to pro- ECO-FERM, FKZ 22031811). duce up to 0.2 g/L rhamnolipids by recombinant expression of the two responsible genes (rhlAB) from Pseudomonas 88. Driven By Demand Metabolic Engineering — aeruginosa [3]. Here we show that we could raise rham- Recombinant Rhamnolipid Synthesis in Pseudomonas nolipid concentration to greater than 3 g/L. Following the Putida As an Example above sketched strategy of driven by demand, a synthetic Till Tiso*1, Andreas Wittgens2, Frank Rosenau2 and promoter library was developed using an approach reported Lars M. Blank1 1iAMB - Institute of Applied Microbiology, RWTH Aachen in literature based on degenerated primers [4]. This tech- University, Aachen, Germany nique yields an array of transcriptional activity, potentially 2Institute of Pharmaceutical Biotechnology, Ulm University, allowing the identification of optimal enzyme activity for high Ulm, Germany flux towards the rhamnolipid synthesis. The best producing strain was able to reach a titer of 3 g/L with a yield of 40% Metabolic engineering of secondary metabolite producers [CmolRL/CmolGlc], which is around 60% of the theoretical implicitly relies on high flux through central carbon metabo- yield. While native producer P. aeruginosa synthesizes 40 g/L lism. This high flux caused by the demand for carbon and [5], the carbon yield is significantly lower. With around 10% energy for the synthesis of the molecule of interest, however [CmolRL/CmolGlc] it ranges below 10% of the theoretical yield. is rarely matched, requiring substantial improvements of Notably, also the specific rate of rhamnolipid production is central carbon metabolism operation. significantly higher using this novel recombinant P. putida (43

mgRL/(gCDW h) opposing to 27 mgRL/(gCDW h)). It was recently shown that central carbon metabolism of Pseudomonas putida is metabolically regulated, while This high rhamnolipid synthesis rate is possible, because the

49 Poster Abstracts

activated rhamnose pathway triples its flux, while the flux chemical overproduction. The SMET method combines both through de novo fatty acid synthesis increases by at least elementary mode analysis and ensemble metabolic model- 40%. We here show that P. putida´s central carbon metabo- ing to derive SMET metrics including c-values and l-values lism is capable of meeting metabolic demand generated by that can identify rate-limiting reaction steps and suggest engineering transcription in peripheral pathways, thereby which enzymes and how much these enzymes to manipu- enabling significant rerouting of carbon fluxes towards the late to enhance product yields, titers, and productivities. target compound, here, industrially interesting rhamnolip- Here, we will present the application of the SMET method ids. Hence, the engineering strategy of driven by demand for analyzing Escherichia coli and Clostridium Thermocel- is highly applicable to P. putida, arguing for this intriguing lum metabolic networks for aromatic acid and lignocellulosic organism as a host in industrial biotechnology. ethanol overproduction. References 1Flowers, D., R. A. Thompson, D. Birdwell, T. Wang, and C. T. Trinh. 1. Koebmann BJ, Westerhoff HV, Snoep JL, Nilsson D, Jensen 2013. SMET: systematic multiple enzyme targeting–a method to rationally design optimal strains for target chemical overproduction. PR: The glycolytic flux in Escherichia coli is controlled by the Biotechnology Journal 8:605-618 demand for ATP. J Bacteriol 2002, 184:3909-3916. .

2. Blank LM, Ionidis G, Ebert BE, Bühler B, Schmid A: Metabolic 90. Mathematical Modelling of Apoptosis for GS-NS0 response of Pseudomonas putida during redox biocataly- Cell Culture Secreting Monoclonal Antibody: Linking sis in the presence of a second octanol phase. FEBS J 2008, Gene to Growth, Metabolism and Metabolic Stress 275:5173-5190. Chonlatep Usaku*1, Efstratios N. Pistikopoulos2 and 3. Wittgens A, Tiso T, Arndt TT, Wenk P, Hemmerich J, Muller Athanasios Mantalaris1 C, Wichmann R, Kupper B, Zwick M, Wilhelm S, et al: Growth 1Biological Systems Engineering Laboratory, Department of independent rhamnolipid production from glucose using the Chemical Engineering, Imperial College London, London, non-pathogenic Pseudomonas putida KT2440. Microb Cell Fact United Kingdom 2011, 10. 2Dept. of Chemical Engineering, Centre for Process Sys- 4. Jensen PR, Hammer K: The sequence of spacers between the tems Engineering, Imperial College London, London, United consensus sequences modulates the strength of prokaryotic Kingdom promoters. Appl Environ Microbiol 1998, 61:82-87. 5. Müller MM, Hörmann B, Syldatk C, Hausmann R: Pseudomo- Mammalian cell culture systems are widely utilised for bio- nas aeruginosa PAO1 as a model for rhamnolipid production in manufacturing, including the production monoclonal antibod- bioreactor systems. Appl Microbiol Biotechnol 2010, 87:167-174. ies (mAbs). Recently, mAbs became ones of the pharmaceuticals of great importance due to their capabilities to efficiently 1 89. SMET: Systematic Multiple Enzyme Targeting for treat chronic diseases, such as cancers . Up to 30 mAbs are Rational Design of Optimal Strains successfully commercialised with the sale value of 18.5 billion Cong T. Trinh*1 and Adam Thompson2 dollars in 2010, which is also expected to be continuously 1 Department of Chemical and Biomolecular Engineering, increasing within the next years1,2. This fact reflects a steadily University of Tennessee, Knoxville, Knoxville, TN constant rise in global demand for mAbs over the last decade 2Energy Science and Engineering, University of Tennessee, and attracts attention in bioprocess intensification for the pro- Knoxville, TN duction of mAbs by mammalian cell culture systems. Several Since the inception of metabolic engineering more than two approaches have been introduced to large scale mammalian decades ago, metabolic engineering has played a signifi- cell culture systems to meet the demand for mAbs: genetic cant role in optimizing microbial biocatalysts and has had a engineering, medium optimisation and manipulation of extra- significant impact on biotechnological applications related to cellular conditions. Nevertheless, mAb production is subject health, food, energy, and environment. One of the key ques- to metabolic stress, and in particular nutrient deficiency and tions that metabolic engineers face is to identify which genes toxic metabolite accumulation, which normally occur as a should be targeted to develop a robust and efficient strain result of cellular metabolism. The stresses have detrimental to achieve desirable phenotypes, e.g., production of a target impact on cell proliferation and favour cell death, thus reduc- compound at high yields, titers, and productivities. We have ing mAb titre. Therefore, intensifying mAb secretion is still a developed a Systematic Multiple Enzyme Targeting method, challenging mission since there is much room for relieving called SMET1, to rationally design optimal strains for target metabolic stress and ultimately increasing mAb final titre. 50 Poster Abstracts

Attempts have been made in order to enable cells to toler- ple, the production of alanine during the exponential phase ate metabolic stress, resulting in a longer culture viability and and the production of ammonium could point out a meta- ultimately a higher mAb production. Previously, apoptosis, bolic shift caused by glutamate and aspartate deprivation12. or programmed cell death, was found to be a main cause of cell death in the culture [3], and it is believed to be triggered In order to take into account a complete picture of the by metabolic stresses, especially nutrient deprivation4,5. Cell observed experimental data, we propose here a systematic lines with transfection of anti-apoptotic gene(s), such as bcl- framework to build a mathematical model toward model- 2 and bcl-xL exhibit resistance to apoptosis, consequently based optimisation. This model was experimentally validated leading to a prolonged culture viability6,7. However, the based on our previous data. The model aims to capture extension of culture viability lifespan does not always result apoptosis at a genetic level and its link to cell growth, cell in an increase in mAb secretion7,8. Cell cycle arrest at G1/ metabolism and extracellular nutrient and metabolite profiles, G0 phase was also found to be where mAb productivity was and aims to incorporate observed features from the ex- enhanced and has an intricate relation to apoptosis. Trans- perimental data. Our mathematical model was constructed fection of p21 inducing cell proliferation arrest at G1/G0 tran- based on the batch with fair balance between biological sition phases of the cell cycle shows promise for enhancing information and model complexity. It can be divided into mAb secretion; several fold increase in mAb production was three main parts; cell growth, cell metabolism and apopto- obtained by the culture of NS0 cell line with p21 over-ex- sis, which is a combination of unstructured and structured pression9. Though cell cycle arrest at G1/G0 phase was also sub-models. The sub-model for cell growth and metabolism found to coexist with bcl-2 and bcl-xL over-expression10, no parts was adopted from previously introduced adapted significant rise in mAb production is often acquired by the Monod kinetics, an unstructured mathematical model linking culture of cell lines with bcl-2 or bcl-xL transfection. These cell growth to substrate utilisation, including glucose and contradictory data together with the high complexity of cellu- key amino acids, and the production of toxic metabolites: lar systems emphasises the need for a systematic approach ammonium and lactate13-15. The apoptosis sub-model was in order not to focus only on a specific point of view, but to inspired by Koutinas, M. et al. (2011) and aims to account for rather take into account a complete picture of the problem/ transcriptional and translational regulation of apoptosis and system. Therefore, a complete map of interactions between/ links to growth and metabolism kinetics as a death factor among cell proliferation, cell metabolism and apoptosis (cell induced by metabolic stresses16. Parameter estimation was death) is needed in order to design an efficient strategy for performed to gain parameter values as regards the batch achieving a higher mAb production. data. Global sensitivity analysis (GSA) was employed to the model in order to identify sensitive parameters to model Previously, we conducted a batch culture study of GS-NS0 outputs and reduce unnecessary model complexity. Our producing cB72.3 mAb to reveal the culture profiles from model shows a good fit to the batch experimental data. Our the genetic level to extracellular metabolic conditions. More preliminary GSA also confirms state variables, such as viable specifically, an interplay map between cell cycle and apop- cells and apoptotic cells, is dependent on their parameters tosis at a transcriptional level, and their links to extracellular relating to these state variables, such as Monod constants metabolites, including glucose, glutamate, ammonium and for glucose, glutamate and caspase3. The model is validated lactate was introduced11. The map was enhanced with ex- with an independent experiment of fed-batch culture which is tracellular amino acid profiles, consequently revealing amino designed based on the batch experimental data. The model acid deprivation effects on cell proliferation inhibition and refinement can be then considered at this stage. In the future, the induction of apoptosis. These data show strong connec- Model-based optimisation will be applied to the finalised tions between amino acid deprivation, especially glutamate model for suggesting feeding strategies. An experimental vali- and aspartate, to the induction of apoptosis. The sudden dation with the suggested feeding schemes will be performed increase in atf5, casp8 and casp3 expression suggests that and might guide the direction in the final model modification cells initially die via the extrinsic pathway of apoptosis as a and significant parameter re-estimation. The advantages of response to the amino acid deprivation. The observed extra- this approach over others are the less expensive cost from cellular amino acid profiles also suggest metabolic pathway experimental trial and possible error. This approach can be adaptation as a result of amino acid deprivation. For exam- a basis for improvement of bio-manufacturing applications,

51 Poster Abstracts

especially mammalian cell culture systems producing recom- Mammalian Cells-Toward the Development of a Model Library. binant proteins and antibodies. Biotechnology Progress, 2007. 23(6): p. 1261-1269. References 16. Koutinas, M., et al., Linking genes to microbial growth kinetics—An integrated biochemical systems engineering approach. 1. Buss, N.A.P.S., et al., Monoclonal antibody therapeutics: Metabolic Engineering, 2011. 13(4): p. 401-413. history and future. Current Opinion in Pharmacology, 2012. 12(5): p. 615-622. 91. Strong Reduction of Acetate Overflow in Escherichia 2. Elvin, J.G., R.G. Couston, and C.F. van der Walle, Therapeutic an- coli By Systems Metabolic Engineering tibodies: Market considerations, disease targets and bioprocessing. Kaspar Valgepea*1,2, Ranno Nahku1,2, Karl Peebo1,2, Petri- International Journal of Pharmaceutics, 2013. 440(1): p. 83-98. Jaan Lahtvee1,2, Liisa Arike1,2 Gethe Riis1,2, Mikk Oun1,2, 1,2 1,2 3. Goswami, J., et al., Apoptosis in batch cultures of Chinese Ham- Kaarel Adamberg and Raivo Vilu 1 ster Ovary cells. Biotechnology and Bioengineering, 1999. 62(6): p. Tallinn University of Technology 2 632-640. Competence Centre of Food and Fermentation Technologies 4. Simpson, N.H., et al., In hybridoma cultures, deprivation of any single amino acid leads to apoptotic death, which is suppressed by The biotechnology industry has extensively exploited Esch- the expression of the bcl-2 gene. Biotechnology and Bioengineer- erichia coli for producing recombinant proteins, low molecu- ing, 1998. 59(1): p. 90-98. lar weight compounds etc. However, high growth/glucose 5. Mercille, S. and B. Massie, Induction of apoptosis in nutrient- uptake rate aerobic E. coli cultivations are accompanied by deprived cultures of hybridoma and myeloma cells. Biotechnology carbon wasting as acetate i.e. acetate overflow. Elimination and Bioengineering, 1994. 44(9): p. 1140-1154. of acetate overflow would improve many bioprocesses as 6. Figueroa Jr, B., et al., A comparison of the properties of a acetate accumulation in the growth environment leads to nu- Bcl-xL variant to the wild-type anti-apoptosis inhibitor in mamma- merous negative effects as it inhibits growth, diverts valuable lian cell cultures. Metabolic Engineering, 2003. 5(4): p. 230-245. carbon from biomass/product formation and is detrimental 7. Tey, B.T., et al., Bcl-2 mediated suppression of apoptosis in for target product synthesis. Despite decades of studies, myeloma NS0 cultures. Journal of Biotechnology, 2000. 79(2): p. mechanism and regulation of acetate overflow are still not 147-159. completely understood. Hence, the aim of this work was 8. Tey, B.T. and M. Al-Rubeai, Suppression of apoptosis in perfusion to elucidate the mechanism and regulation behind acetate culture of Myeloma NS0 cells enhances cell growth but reduces overflow using a comprehensive systems biology approach antibody productivity. Apoptosis, 2004. 9(6): p. 843-852. leading to the design of reduced overflow strains. 9. Watanabe, S., J. Shuttleworth, and M. Al-Rubeai, Regulation of cell cycle and productivity in NS0 cells by the over-expression of In this work, we continuously monitored specific growth rate p21CIP1. Biotechnology and Bioengineering, 2002. 77(1): p. 1-7. (μ)-dependent acetate overflow dynamics of wild-typeE. 10. Janumyan, Y.M., et al., Bcl-xL/Bcl-2 coordinately regulates coli using advanced continuous cultivation methods (A-stat apoptosis, cell cycle arrest and cell cycle entry. EMBO J, 2003. 22: p. 5459-5470. and D-stat). Absolute quantitative exo-metabolome, transcriptome and proteome analyses coupled to metabolic 11. Usaku, C., et al., Cell cycle and apoptosis: a map for the GS-NS0 cell line at the genetic level and the link to environmental flux analysis revealed that acetate overflow (at μ=0.27±0.02 stress. BMC Proceedings, 2013. 7(Suppl 6): p. P54. h-1) in E. coli is triggered by carbon catabolite repression- 12. Nelson, D.L. and M.M. Cox, Lehninger principles of biochemis- mediated down-regulation of acetyl-CoA syntethase (Acs) try 4th edition ed2004: W. H. Freman. resulting in decreased assimilation of acetate produced by phosphotransacetylase (Pta) and disruption of the PTA-ACS 13. Jang, J.D. and J.P. Barford, An unstructured kinetic model of macromolecular metabolism in batch and fed-batch cultures of node. This was confirmed by two-substrate A-stat and D-stat hybridoma cells producing monoclonal antibody. Biochemical Engi- experiments which showed that acetate consumption capa- neering Journal, 2000. 4(2): p. 153-168. bility of E. coli decreased drastically, just as Acs expression, 14. Kiparissides, A., E.N. Pistikopoulos, and A. Mantalaris, To- before the start of acetate overflow. These results suggested wards energy-based dynamic optimization of monoclonal antibody insufficient Acs activity for completely consuming the acetate producing GS-NS0 Cultures, in Computer Aided Chemical Engi- produced by Pta, leading to disruption of acetate recycling neering, S. Pierucci and G.B. Ferraris, Editors. 2010, Elsevier. p. in the PTA-ACS node where constant acetyl-phosphate or 589-594. acetate regeneration is essential for E. coli chemotaxis, prote- 15. Kontoravdi, C., et al., Modeling Amino Acid Metabolism in 52 Poster Abstracts olysis, pathogenesis etc. regulation. The latter indicated that To the best of our knowledge, this is the first successful ap- E. coli actually synthesizes acetate constantly at all μ under plication of modification of protein acetylation for metabolic aerobic conditions and no acetate overflow occurs at low μ engineering in E. coli. Furthermore, we see this work being since acetate is fully recycled in the PTA-ACS node. This in- a good example for proving the value of systems biology deed turned out to be the case as both E. coli mutants (Δacs study of metabolism for successful metabolic engineering of and ΔcobB) unable to consume acetate in the PTA-ACS strains with potential interest for industrial use. node excreted acetate at all μ studied, further highlighting the relevance of Acs in acetate overflow regulation. 92. Cytosolic Acetyl-CoA Synthesis By Pyruvate- Formate Lyase in Yeast Harmen M. van Rossum*1, Barbara U. Kozak1, Kirsten R. The following obvious effort to diminish acetate overflow Benjamin2, Liang Wu3, Jean-Marc G. Daran1, Jack T. Pronk1 based on the latter results was to increase the expression and Antonius J.A. van Maris1 of Acs. However, a strain overexpressing acs did not lead 1Department of Biotechnology, Delft University of Technology, to reduced acetate overflow which could have arisen from a Delft, The Netherlands substantial part of the Acs protein pool being inactive as Acs 2Amyris Inc., Emeryville, CA, United States of America 3 activity in vivo is known to be actively repressed by post- DSM Biotechnology Center, Delft, The Netherlands translational acetylation. Thus we next analyzed growth of E. Many industrially relevant products, like farnesene, arte- coli lacking protein lysine acetyltransferase (Pka), known to misinin and 1-butanol have acetyl-CoA as a precursor. The inactivate Acs, which resulted in postponed start of acetate yeast Saccharomyces cerevisiae is a widely used metabolic overflow. We surmised that acetate overflow could be further engineering platform because of its robustness and genetic reduced by slightly increased levels of active Acs in Δpka and accessibility. This makes acetyl-CoA produced in the cytosol increased throughput of downstream pathways (e.g. TCA of yeast an important building block for the formation of cycle) to recycle more acetate into acetyl-CoA and process many (heterologous) products. Therefore fast and efficient the latter downstream to divert carbon away from acetate. production of cytosolic acetyl-CoA in brewer’s yeast is of Exactly this was achieved by deleting the TCA cycle regulator industrial interest. arcA in Δpka background leading to further postponed onset and strongly reduced acetate overflow by the coordinated In this yeast, cytosolic acetyl-CoA synthesis and growth activation of PTA-ACS and TCA cycles. The double deletion strictly depend on expression of either the Acs1 or Acs2 Δpka ΔarcA strain showed 4-fold reduced acetate overflow (2 isoenzyme of acetyl-CoA synthetase (ACS). Since these vs 8% from total carbon) at fastest growth compared to wild- enzymes consume the equivalent of 2 ATP, it constrains type, did not accumulate any other detrimental by-product maximum yields of acetyl-CoA-derived products. Therefore, besides acetate, showed identical μmax and only ~5% lower the replacement of ACS by the ATP-independent pathway biomass yield compared to wild-type. Moreover, this strain of pyruvate-formate lyase (PFL) for cytosolic acetyl-CoA would enable production of target compounds in the absence synthesis was explored. of acetate at considerably higher growth/glucose uptake rates most probably resulting in higher volumetric productivi- After evaluating expression of five different PFL genes, ties (~22% higher gDCW L-1 h-1 compared to wild-type). We acs1Δ acs2Δ S. cerevisiae strains were constructed in which conclude that a fine-tuned coordination between increasing PFL successfully replaced ACS. In PFL-dependent strains, the recycling capabilities of acetate in the PTA-ACS node anaerobic growth rates were stoichiometrically coupled to through higher pool of active acetate scavenging Acs protein formate production. The biomass yield on glucose was lower and downstream metabolism throughput in the TCA cycle are than those of the reference strain. Transcriptome analysis necessary for diminishing acetate overflow in E. coli. suggested that this reduced biomass yield was caused by formate. Transcript profiles also indicated that a previously This work demonstrates that a simple genetic overexpres- proposed role of Acs2 in histone acetylation is probably sion does not work in all cases for achieving the desired linked to cytosolic acetyl-CoA levels rather than to direct effects but the expression level of the relevant fraction of the involvement of Acs2 in histone acetylation. While demon- protein pool (active Acs in this case) has to be fine-tuned strating that yeast ACS can be fully replaced, further modi- together with downstream throughput (TCA).

53 Poster Abstracts

fications are needed to achieve optimal in vivo performance Seeking alternative energy sources has gained much popu- of the PFL pathway for supply of cytosolic acetyl-CoA as a larity due to the shortage of fuel and increasing global product precursor. environmental concerns. During the past decades,ethanol See also Kozak, B. U., Van Rossum, H.M. et al. Metabolic has be seen the major renewable substitute of fuel. How- engineering 21C, 46–59 (2013). ever, ethanol is not the ideal one, because of its lower energy density and hygroscopicity-resulting storage and 93. Itaconic Acid Production in Escherichia coli By distribution problems. Biobutanol, produced by using vari- Overexpression of Citrate Synthase, Aconitase, and ous renewable resources, is considered one kind of better Cis -Aconitate Decarboxylase alternative fuel. It is able to be synthesized by the engi- Kiira Vuoristo*1, Astrid Mars2, Jan Springer2, Gerrit Eggink2, Johan Sanders2 and Ruud Weusthuis1 neered Klebsiella pneumoniae using crude glycerol from 1Bioprocess Engineering, Wageningen University, the production of biodiesel as the sole source. Comparing Wageningen, Netherlands the relative pathways published on KEGG between Clous- 2Biobased Products, Wageningen University and Research tridium acetobutylicum and K. pneumonia, we found that Centre, Wageningen, Netherlands in K. pneumonia there exist some enzymes which have similar catalytical roles with some essential genes needed Interest in sustainable development has led to efforts to re- in butanol synthesis in C. acetobutylicum. Only additional place petrochemical-based monomers with biomass-based bcd/etfAB, bdhB and bdhA, absent in K. pneumonia, are ones. Itaconic acid (IA) is a C5-dicarboxylic organic acid, to be introduced. When constructing the engineered strain, which can be polymerized at a high conversion rate, and as the gene ter, instead of the bcd/etfAB, was introduced with such, a building block for the chemical industry with many the other two genes. The engineered K. pneumonia has the potential applications. To make production of these biobased ability of yielding 18 mg/L butanol using glycerol as the only building blocks competitive over their petrochemical coun- carbon source by flask fermentation while the native strain terparts, microbial processes have to reach high yields (g/g) produces no butanol. Moreover, the antisense RNA strategy and productivities (g/l/h). was adopted to reduce the level of relative genes’ mRNA to interfere the synthesis of 1, 3-propanediol and 2, 3-butane- IA has traditionally been produced by Aspergillus terreus, but diol that were native products during the fermentation. The as the process has not reached cost-efficiency in large scale, resultant yield of 1, 3-propanediol was reduced by 90%, also recombinant production hosts such as Escherichia coli 2, 3-butanediol was by 10%, compared to those of the K. have been investigated. In the work of Li et al. (2011), a key pneumonia control strain. gene for itaconate biosynthesis, cis-aconitate decarboxylase (cadA) from A. terreus was successfully expressed in E. coli, 95. Single-Cell Bioreactors Boost Bioprocess but product titre remained low. Development: New Insights into Cellular Metabolism Wolfgang Wiechert*, Dietrich Kohlheyer, Alexander Grün- The purpose of our work was to study the IA production berger, Christopher Probst, Stephan Helfrich, Julia Frunzke, Lothar Eggeling, Katharina Nöh and Stephan Noack mechanism and thus increase the efficiency of the pathway IBG-1: Biotechnology, Forschungszentrum Jülich, Jülich, in E. coli. We started by optimizing cadA expression condi- Germany tions and also improved the availability of the precursors citrate and cis-aconitate by overexpression of heterologous Cell-to-cell heterogeneity caused by biological (intrinsic) and citrate synthase and aconitase. As a result, IA production in environmental (extrinsic) fluctuations can have severe impact E. coli was significantly improved, up to 750 mg/L, which is on the productivity of biotechnological processes. Neverthe- among the highest titres reported for this organism. less, till now the complex interplay between environmental reactor dynamics, genetic noise and cellular activity is hardly 94. Metabolic Engineering of Klebsiella Pneumoniae for understood. Specially designed microfluidic systems of- 1-Butanol Production By Using Crude Glycerol 1 Miaomiao Wang*1 and Tianwei Tan2 fer unique possibilities to tackle these problems . Aiming 1Beijing key lab of bioprocess, Beijing University of Chemical at metabolic engineering and bioprocess development, we Technology, Beijing, China developed an innovative microfluidic platform for microbial 2Beijing University of Chemical Technology, Beijing, China single-cell investigations2. Different reactor types from femto-

54 Poster Abstracts

to pico-litre scale facilitate spatio temporal analysis of single Interestingly, under certain complex environmental cultivation microbial cells under well-defined, externally controlled and conditions, isogenic micro colonies split up into to producing and even dynamic environmental conditions. For the first time non-producing sub populations. Other microfluidic investigation aims at spontaneous prophage induction in C. glutamicum strains several central questions regarding growth, productivity, and its influence to bioprocess performance12. substrate uptake as well as population heterogeneity and in- • Process characterization: C. glutamicum strains exhibiting 1.5 fluence of bioreactor inhomogeneity can be addressed within fold improved growth rates were analyzed during microfluidic cul- well-defined single-cell cultivation experiments. tivations13. The bioreactor environment could be simulated on chip by applying bioreactor supernatant from distinct cultivation time The Jülich single-cell bioreactor technology platform devel- points. In combination with extensive substrate screening experiments, the constant availability of protocatechuate (PCA), typically oped in the recent years exhibits several specific advantages added as iron chelator, was identified to be responsible for the for applications in metabolic engineering and bioprocess elevated growth rates14. This reveals that microbial diets in industrial development: bioreactors can strongly differ from the wanted substrate usage. • sub-micrometer structures needed for bacterial investigations Thus, using microfluidic chips single-cell performance can be al- can be fabricated in a PDMS molding process3 located to specific process conditions. • many single-use chips can be replicated at low cost from • Cell manipulation for strain screening: Till now most single-cell one master mould3 microbial cultivation systems suffer from limited control during cell • new designs can be quickly realized and, thus, tailor-made inoculation and sampling methods during cultivation. To enable indi- chips can be adapted to the experimental questions and vidual cell selection and manipulations we combined laser tweezers not vice versa4 with microfluidic cell cultivation environments specifically tailored • environmental conditions (temperature, pH, medium for micrometre sized bacteria. For the first time a filamentous E. coli composition) can be tightly controlled and also WT (MG1655) was safely relocated from its growing microcolony by dynamically changed on a sub second time scale 2, 5, 6 laser manipulations. The cell was transferred to an empty cultivation 15 • hundreds of single-cell bioreactors can be operated site allowing single cell growth and morphology investigations . simultaneously in one chip, thus generating the required Summarizing, the advent of tailor-made single-cell bioreac- statistical data for meaningful results tors which can precisely simulate bioprocess relevant condi- • microfluidic chips are fixed on an automated x-y stage to tions and allow the exact monitoring of growth, morphology enable high throughput microscopy including fluorescence imaging2, 5, 6, 7 and biosensor signals related to metabolism and gene ex- • hundreds of interlaced time-lapse videos can be automatically pression will most certainly become an indispensable tool for evaluated by a specially developed image analysis pipeline strain characterization and screening, metabolic engineering The following applications from our institute demonstrate and bioprocess development. that microfluidics has become mature as a highly valuable References tool to understand hitherto non understood metabolic [1] Grünberger, A,: Wiechert, W.; Kohlheyer, D. phenomena: Single-Cell Microfluidics: Opportunity for Bioprocess Development. Current Opinion Biotechnology. (2014) In Press. • Strain characterization: Growth, one of the most important performance indicators in biotechnological production processes, [2] Grünberger, A.; Paczia, N.; Probst, C.; Schendzielorz, G. ; is still one of the “most underexploited assays of cellular heteroge- Eggeling, L. ; Noack, S.; Wiechert, W.; Kohlheyer, D. neity studies to date”8. The link between metabolism and growth A disposable picolitre bioreactor for cultivation and investigation of or morphology is currently intensely investigated8. Microfluidics in industrially relevant bacteria on the single cell level. combination with image analysis produces precise lineage-trees Lab on a Chip 12, 2060 - 2068 (2012) and single-cell growth curves revealing population inhomogeneity [3] Grünberger, A.; Probst, C.; Heyer, A.; Wiechert, W.; Frunzke, J.; 10 and cell aging effects . Strong differences in the growth behavior of Kohlheyer, D. Microfluidic Picoliter Bioreactor for Microbial Single- 10 E. coli and C. glutamicum production strains could be revealed [6]. cell Analysis: Fabrication, System Setup, and Operation. • Population heterogeneity: Genetically encoded fluorescence bio- Journal of Visualized Experiments 82, 50560 (2013) sensors and reporter systems are ideal tools for time-lapse imaging during microfluidic single-cell analysis. In particular transcriptional [4] Probst, C.; Grünberger, A.; Wiechert, W.; Kohlheyer, D. regulator based metabolite reporters proved to be a versatile tool in Polydimethylsiloxane (PDMS) Sub-Micron Traps for Single-Cell monitoring intracellular metabolites5, 6, 7, 11. These reporter systems Analysis of Bacteria. transform intracellular metabolite concentrations into a detect- Micromachines 4(4), 357 - 369 (2013) able fluorescence readout and were used to investigate single-cell [5] Mustafi, N.; Grünberger, A. ; Kohlheyer, D. ; Bott, M.; Frunzke, J. production of amino acids at various environmental conditions. The development and application of a single-cell biosensor for the

55 Poster Abstracts

detection of L-methionine and branched-chain amino acids. 2Institute of Biochemical Engineering, University of Stuttgart, Metabolic Engineering 14, 449 - 457 (2012) Germany 3 [6] Mustafi, N.. ; Grünberger, A. ; Mahr, R.; Helfrich, S.; Nöh, K.; Institute of Microbiology and Biotechnology, University of Blombach, B. ; Kohlheyer, D.; Frunzke, J. Application of a Geneti- Ulm, Germany cally Encoded Biosensor for Live Cell Imaging of L-Valine Produc- tion in Pyruvate Dehydrogenase Complex-Deficient Corynebacte- Corynebacterium glutamicum is a non-pathogenic, Gram- rium glutamicum Strains. PLoS One 9(1), e85731 (2014) positive organism that grows on a variety of substrates and [7] Schallmey, M.; Frunzke, J.; Eggeling, L.; Marienhagen, J.; is used for the production of amino acids (e.g. L glutamate, Looking for the pick of the bunch: high-throughput screening of L lysine and L valine). The aim of the present work was to producing microorganisms with biosensors. engineer C. glutamicum to produce pyruvate aerobically and Current Opinion in Biotechnology 26, 148-154 (2014). exploitation of the resulting strain as platform for anaerobic [8] Lecault,V.; White, A.K.; Singhal, A.; Hansen, C.L.; succinate production. Microfluidic single cell analysis: from promise to practice. Current Opinion in Chemical Biology 16, 381-390 (2012). In our study we modified C. glutamicum for the production of [9] Klumpp, S.; Hwa, T.; Bacterial growth: global effects on gene e pyruvate and decreased formation of byproducts. By step- xpression, growth feedback and proteome partition. Current wise inactivation of the pyruvate dehydrogenase complex, Opinion in Biotechnology 28, 96–102 (2014). the pyruvate:quinone oxidoreductase, the L lactate dehydro- [10] Wang, P.; Robert, L.; Pelletier, J.; Dang, W.L.; Taddei, F.; Wright, genase and attenuation of the acetohydroxyacid synthase A.; Jun, S. Robust growth of Escherichia coli. Current Biology 20, [AHAS], efficient pyruvate production was achieved. The 1099-1103 (2010). deletion of the genes encoding alanine aminotransferase [11] Schendzielorz, G.; Dippong, M.; Grünberger, A.; Kohlheyer, D.; and pyruvate:valine aminotransferase led to a strong re- Yoshida, A. ; Binder, S.; Nishiyama, C. ; Nishiyama, M. ; Bott, M.; Eggeling, L. Taking Control over Control: Use of Product Sensing in duction of the side product L alanine and together with the Single Cells to Remove Flux Control at Key Enzymes in Biosynthe- attenuation of the AHAS to decreased L valine formation. sis Pathways. ACS Synthetic Biology. 130715122931005 - (2013) In fed-batch fermentations with adjusted oxygen supply [12] Nanda, A.; Heyer, A.; Kramer, C.; Grünberger, A.; Kohlheyer, D.; during growth and production (0 5% DO), the engineered Frunzke, J. strain produced more than 500 mM pyruvate with a yield Analysis of SOS-Induced Spontaneous Prophage Induction in of 0.97 mol per mole of glucose and a productivity of 0.92 Corynebacterium glutamicum at the Single-Cell Level. Journal of mmol∙g(CDW)−1∙h−1 (i.e., 0.08 g∙g(CDW)−1∙h−1) in the pro- Bacteriology 196(1), 180 - 188 (2014) duction phase1. [13] Grünberger, A.; van Ooyen, J.; Paczia, N.; Rohe, P.; Schen- dzielorz, G.; Eggeling, L.; Wiechert, W.; Kohlheyer, D.; Noack, S. For production of succinate with this mutant, a tri-phasic Beyond growth rate 0.6: Corynebacterium glutamicum cultivated fed-batch fermentation process in a single bioreactor was in highly diluted environments. Biotechnology & Bioengineering 110(1), 220 - 228 (2013) established: Aerobic growth on acetate was followed by a self-induced microaerobic phase at the end of growth by [14] Unthan, S.; Grünberger, A.; van Ooyen, J.; Gätgens, J.; Hein- rich, J.; Paczia, N.; Wiechert, W.; Kohlheyer, D.; Noack, S. minimal aeration. Subsequently, an anaerobic production

Beyond growth rate 0.6: What drives Corynebacterium glutamicum phase was realized by CO2 gassing. Under these conditions to higher growth rates in defined medium? a final succinate concentrations above 330 mM from 325 Biotechnology & Bioengineering 111(2), 359–371 (2014) mM glucose was obtained, with a YP/S of 1.02 mole succi- [15] Probst, C.; Grünberger, A.; Wiechert, W.; Kohlheyer, D. nate per mole of glucose2. Microfluidic growth chambers with optical tweezers for full spatial References: single-cell control and analysis of evolving microbes. [1] Wieschalka S, Blombach B, Eikmanns BJ (2012) Engineering Journal of microbiological methods 95(3), 470 - 476 (2013) Corynebacterium glutamicum for the production of pyruvate. Appl Microbiol Biotechnol. doi: 10.1007/s00253-011-3843-9 96. Corynebacterium Glutamicum Engineered As a De- signer Bug for the Production of Pyruvate and Succinate [2] Wieschalka S, Blombach B, Bott M, Eikmanns BJ (2013) Bio- Stefan Wieschalka*1, Bastian Blombach2 and Bernhard J. based production of organic acids with Corynebacterium glutami- Eikmanns3 cum. Microb Biotechnol 6:87–102. doi: 10.1111/1751-7915.12013 1Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Denmark

56 Poster Abstracts

97. Systems Metabolic Engineering of Corynebacterium These substrates are renewable, produced via biomass Glutamicum to Overcome the Cellular Toxicity Derived gasification or natural gas reformation, thus the biologi- from Cellulosic Hydrolysate cal gas-to-liquids (GTL) process represents a sustainable Hong-Sil Park1,2, Youngsoon Um1, Sang Jun Sim2,3 and Han platform that will be important for reducing dependence on Min Woo*1,2 1Clean Energy Research Center, Korea Institute of Science fossil fuels. However, the range of products available from and Technology, Seoul, South Korea syngas fermentation, as exemplified by the commercial 2Green School, Korea University, Seoul, South Korea processes of LanzaTech, Coskata and INEOS, is currently 3Department of Chemical and Biological Engineering, Korea limited by the absence of robust genetic engineering tools University, Seoul, South Korea for the organisms capable of metabolizing the gases in question. We present here our progress in developing these Pretreatment of cellulosic biomasses by acid hydrolysis tools to enable the metabolic engineering of the acetogen generates cellular inhibitors such as Furfural, HMF, and Moorella thermoacetica. This thermophilic, gram-positive acetic acid etc. Therefore, it is necessary to overcome the anaerobe naturally produces acetate from H /CO or CO via toxicity in the hydrolysate during fermentation. Our goal is to 2 2 the canonical Wood-Ljungdahl pathway in order to provide develop the strain capable of being tolerant to the inhibitors. the energy required for cell growth. As a homoacetogen, We investigated the effect of the inhibitors to Corynebacte- producing only acetate as its metabolic byproduct, Moorella rium glutamicum, which is widely known as an amino acid is an ideal candidate for a production platform, but stable producer and potential microbial cell factory. Cellular growth transformation of this organism has only recently been re- rates of C. glutamicum were decreased with various cellular ported, and more development is required to establish a set inhibitors, including furfural that is a derivative of furan in the of tools for convenient, tunable heterologous expression. To hydrolysate. C. glutamicum grown with furfural showed over that end, we describe the development of a Flow Cytometry 2-folds of reactive oxygen species stresses. In addition, the assay to optimize DNA delivery conditions, as well as our lower conversion rate of furfural and glucose uptake rate progress improving the robustness of the transformation were measured. We have engineered C. glutamicum by over- protocol and expressing heterologous genes. expressing the target genes identified through both bioinformatics search and DNA microarray. When we cultivated the 99. Construction of Fast Xylose-Fermenting Yeast Based engineered strains with furfural, the growths were recovered on Industrial Ethanol-Producing Diploid Saccharomyces as about 70% percent of growth rate of wild-type grown cerevisiae By Rational Design and Adaptive Evolution without furfural. Several metabolic engineering approaches Liuyang Diao, Yingmiao Liu, Fenghui Qian, Junjie Yang*, could further improve the tolerance in C. glutamicum. This Yu Jiang and Sheng Yang CAS Key Laboratory of Synthetic Biology, Institute of Plant work was supported by the National Research Foundation Physiology and Ecology, Shanghai Institutes for Biological of Korea Grant funded by the Korean Government (MEST) Sciences, Chinese Academy of Sciences, Shanghai, China (2014, University-Institute cooperation program) and ‘Cre- ative Allied Program (CAP)’ through the Korea Research It remains a challenge for recombinant S. cerevisiae to Council of Fundamental Science and Technology (KRCF) convert xylose in lignocellulosic biomass hydrolysates to and Korea Institute of Science and Technology (KIST). ethanol. Although industrial diploid strains are more robust compared to laboratory haploid strains, however, indus- 98. Development of Genetic Tools for the Metabolic trial diploid S. cerevisiae strains have been less pursued in Engineering of the Thermophilic Acetogen Moorella previous studies. This work aims to construct fast xylose-fer- Thermoacetica Benjamin Woolston*, Devin Currie, Hamid Rismani-Yazdi and menting yeast using an industrial ethanol-producing diploid Greg Stephanopoulos S. cerevisiae strain as a host. Chemical Engineering, MIT, Cambridge, MA Fast xylose-fermenting yeast was constructed by genome Synthesis gas (syngas) fermentation by anaerobic bacte- integration of xylose-utilizing genes and adaptive evolution, ria is a novel method for the biological production of fuels including 1) Piromyces XYLA was introduced to enable the and chemicals from gaseous substrates such as carbon host strain to convert xylose to xylulose; 2) endogenous monoxide (CO), carbon dioxide (CO2), and hydrogen (H2). genes (XKS1, RKI1, RPE1, TKL1, and TAL1) were overex-

57 Poster Abstracts

pressed to accelerate conversion of xylulose to ethanol; 3) deleted, and in addition the two key enzymes responsible for Candida intermedia GXF1, which encodes a xylose trans- the conversion, namely IolG and IolW, were overproduced porter, was introduced at the GRE3 locus to improve xylose under the control of one of the strongest promoters. All of uptake; 4) aerobic evolution in rich xylose media was carried 1% (w/v) MI contained in the medium was converted into SI out to increase growth and xylose consumption rates. The at the rate of 10 g/ L/ 48h at least. The efficient conversion best evolved strain CIBTS0735 consumed 80 g/l glucose was achieved only in the presence of enriched nutrition in the and 40 g/l xylose in rich media within 24 hours at an initial form of 2% (w/v) Bacto soytone, which may be due to the OD600 of 1.0 (0.63 g DCW/l) and produced 53 g/l ethanol. increasing demand for regeneration of cofactors including NADPH. The results from our transcriptomic and fluxomic Based on the above fermentation performance, we conclude analyses suggested that the regeneration of NADPH might that CIBTS0735 shows great potential for ethanol production be enabled by generic alternation in central carbon me- from lignocellulosic biomass tabolism including enhancement in hexose monophosphate shunt and gluconeogenesis. 100. Bacterial Cell Factory for Production of Scyllo- Inositol, a Potential Therapeutic Agent for Alzheimer’s 101. Construction of a Hybrid Pathway for Selectively Disease Removing Nitrogen Atom from Carbazole Kosei Tanaka1, Yoshihiro Toya2, Shinji Takenaka3, Hiroshi Jianfeng Hou and Bo Yu* Shimizu4 and Ken-ichi Yoshida*3 Institute of Microbiology, Chinese Academy of Sciences 1Organization of Advanced Science and Technology, Kobe University, Kobe, Japan Acid precipitation resulting from oil combustion has driven 2Graduate School of Information Science and Technology, enormous efforts to remove sulfur and nitrogen contami- Osaka University, Suita, Japan 3Department of Agrobioscience, Kobe University, Kobe, nants from fossil fuels, including biological treatment as Japan alternatives. Unlike biodesulfurization where sulfur is selec- 4Department of Bioinformatic Engineering, Osaka University, tively removed from substrates leaving the hydrocarbon por- Suita, Japan tion of the molecule intact, the pathway of carbazole (typical nitrogen compound in oils) degradation only liberates nitro- Inositol (1,2,3,4,5,6-cyclohexanehexol) has nine possible gen in the course of completely degrading the substrates, stereoisomers. One of the stereoisomers, myo-inositol (MI), is leading a loss of the oil combustion values. While there have most abundant in nature and supplied cheap from rice bran. been no publications concerning pathway engineering to On the other hand, another stereoisomer, scyllo-inositol (SI), investigate the selective removal of nitrogen atom from car- is rare in nature, and precious because of being expected bazole. In the native pathway, carbazole was firstly metabo- as a disease-modifying therapeutic agent for Alzheimer’s lized into 2’-aminobiphenyl-2,3-diol catalyzed by carbazole disease, which is one of the most common and problematic 1,9a-dioxygenase. A novel deaminase from chloronitroben- forms of dementia. It is known that aggregation of amyloid- zene degradation is expected to be evolved, which could beta in the brain is one of the key pathological features of selectively remove amino-group from 2’-aminobiphenyl-2,3- Alzheimer’s disease. SI directly interacts with amyloid-beta diol. Combination of the carbazole 1,9a-dioxygenase and the and blocks the development of its fibrous aggregation. In evolved deaminase, a ‘hybrid’ pathway for carbazole me- fact, oral administration of SI to a mouse model of Alzheim- tabolism is proposed, in which carbazole should be selec- er’s disease attenuated amyloid-beta-induced impairments tively transformed into hydroxylated biphenyl. The synthetic of spatial memory, reduced cerebral amyloid-beta pathology, pathway will be capable of simultaneously and selectively and decreased the rate of mortality. And thus SI has received removing nitrogen atom to retain the carbon skeleton of the a fast-track designation from the US Food and Drug Adminis- fuels and therefore keep the combustion values. tration for treatment of mild to moderate Alzheimer’s disease. 102. Protein Design for a De Novo Synthetic Pathway of We demonstrated a cell factory, which enables bio-conver- Microbial Production of 1,3-Propanediol from Sugar sion from MI to SI, made of Bacillus subtilis with the modi- Feng Geng, Zhen Chen and An-ping Zeng* fied inositol metabolic pathway. In the B. subtilis cell factory, Institute of Bioprocess and Biosystems Engineering, Hamburg University of Technology, Hamburg, Germany all “useless” genes involved in MI and SI metabolism were

58 Poster Abstracts

Low energy-cost and environmentally friend bioproduction Since no naturally efficient enzymes are known for the first processes are a major goal of sustainable bio-based econo- reaction starting from homoserine, the engineering of a suit- my. The limitation of natural biosynthetic pathways hints the able enzyme for an altered substrate specificity is the most development of efficient bioprocesses for the production of important step towards the new PDO biosynthesis route. We certain desired products. Resent years, protein engineering chose glutamate dehydrogenase (GDH) as a candidate. The and synthetic biology opens doors to efficient or completely wildtype GDH had a relatively low activity towards homoser- new bioprocesses by De novo design of non-natural path- ine. Several mutants were designed by a structure-based ap- ways. This is demonstrated in this work for the realization proach which showed significantly increased activities. For a of a non-natural pathway and reengineered enzyme for the proof of concept we used an in vitro enzyme cascade con- production of 1,3-propanediol (PDO) from sugar. sisting of mutated GDH, pyruvate decarboxylase (PDC) from Zymomonas mobilis DSM 3580 and alcohol dehydrogenase PDO is an important chemical monomer that can be utilized (yqhD) from E. coli MG1655 to demonstrate the production for the synthesis of high value polymers and coating materi- of PDO rom homoresine. Initial test resulted in the production als. However, no natural microorganisms have been found of 35 mg/L PDO from homoserine. which can directly use sugars to produce PDO. The Dupont Bio-PDO process was considered as a major achievement Subsequently, an E. coli strain was constructed to test the of metabolic engineering and industrial biotechnology in the new pathway in vivo. Genes for the three key enzymes (gdhA last years. In the Dupont-PDO a synthetic pathway was suc- or its mutant M1, kdcA, yqhD) were integrated into plas- cessfully developed to produce PDO from glucose, in which mids and overexpressed in E. coli MG1655 ∆ThrB. E. coli the glycerol synthesis pathway from Saccharomyces cere- MG1655∆ThrB (pZA-gdhA-M1-kdcA-yqhD) with the mutated visiae and the metabolic pathway of converting glycerol to GDH produced 51.5±4.9 mg/L PDO which was 110% higher PDO from Klebsiella pneumonia were integrated into E. coli. than the strain MG1655∆ThrB (pZA gdhA-kdcA-yqhD) with However, several major burdens have to be overcome in de- the widetype GDH whereas the blank control MG1655∆ThrB signing such a synthetic metabolic pathway such as (1) the did not produce any PDO, demonstrating the possibility of need of coenzyme B12 by one of the key enzymes glycerol producing PDO from sugar with the new pathway. dehydratase (GDHt) and (2) substrate-suicide of GDHt which could limit the productivity. Furthermore, E. coli can normally 103. Systematic Characterization of Protein–Protein only use C6 sugar. Interface for the Development of Artificial Biomachinery for Metabolic Engineering Cheng-Wei Ma and An-Ping Zeng* In our work, protein engineering is used to design a new and Institute of Bioprocess and Biosystems Engineering, de novo route for the biosynthesis of PDO from sugars. It Hamburg University of Technology, Hamburg, Germany is based on pathways of amino acid biosynthesis and uses homoserine as a key intermediate. Homoserine is converted Protein–protein interactions (PPIs) play crucial roles in many into PDO in an unnatural pathway involving three enzymatic fundamental biological processes such as signal transduc- steps: (1) the deamination of homoserine to 4-hydroxy-2-ke- tion, transport, cellular motion, and most regulatory mecha- tobutyrate; (2) the decarboxylation of 4-hydroxy-2-ketobu- nisms. Owing to such overwhelming biological significance, tyrate to 3-hydroxypropionaldehyde; and (3) the reduction of elucidating the mechanisms of the interactions presents a 3-hydroxypropionaldehyde to PDO. The theoretical maxi- challenge in molecular biology. Since PPIs are locally medi- mum yield (1.5 mol PDO/ glucose) of the new PDO pathway ated through protein–protein interfaces, systematic charac- is the same as that of the Dupont route. Since homoserine terization of the physico-chemical properties of interfaces synthesis is a common pathway in most of the bacteria, the through which PPIs take place will provide valuable informa- proposed route can be engineered into selected hosts with tion for the rational design of novel PPIs. In the meanwhile, the more favorable ability to utilize different and cheap sug- the huge number of biomolecular structures stored in the ars. Moreover, the proposed pathway does not utilize GDHt Protein Data Bank (PDB) makes it possible to discover fine and thus can avoid the serious problems associated with details about the interface that enables the interaction be- vitamin B12 and substrate suicide. This non-natural pathway tween proteins by systematic investigation of those molecu- is thus very appealing for PDO production. lar structures. In this study, characteristics of protein–protein

59 Poster Abstracts

interfaces are analyzed based on the non-redundant data- solvated free energy under different allosteric states. For the base in order to obtain statistical meanings. The non-redun- ligand binding process, the state parameter is calculated dant database is derived from the PDB by clustering chains from molecular dynamics simulation. Contributions of each into groups according to their amino acid sequence similari- residue can be illustrated through energy decomposition. ties and selecting a representative from each of those groups. The allosteric process is simulated by steered molecular When characterizing PPI interfaces, the stereo complemen- dynamics followed by dynamic correlation analysis and fluc- tarity evaluated includes interface size measured in solvent– tuation analysis. Three key enzymes in the biosynthesis of accessible surface area (SAS) and the volume formed among amino acids, E. coli aspartokinase III (AK III), E. coli 3-Deoxy- the interface. Interface residues could also be found accord- d-arabino-heptulosonate-7-phosphate synthase (DAHPS) ing to the SAS change of each surface residue during the and C. glutamicum homoserine dehydrogenase (HSDH) are formation of the protein complex. The energetic complemen- used as model systems to demonstrate the novel strategy tarity takes into account both the contributions from Van der with totally 14 metabolites as non-natural ligands. This strat- Waals interactions and that from electrostatic interactions. egy makes it possible to rational design of allosteric regula- Moreover, contributions of each interface residue are also ex- tions useful in the dynamic control of metabolism. amined to find hot spots. Considering the significance of PPIs in the construction of biomachinery such as metabolic chan- 105. Development of the First Scalable Rubbery Polyester neling and biofunctional compartmentalization, it is expected Kechun Zhang* and Mingyong Xiong that these results could be helpful for the artificial design of Chemical Engineering and Materials Science, University of novel biomachinery for metabolic engineering. Minnesota, Minneapolis, MN

104. Dynamic Control of Metabolism through While holding great promise as sustainable and biodegrad- Engineering Ligand-Induced Allosteric Regulation Based able, current biobased polyesters encounter significant difficul- on a New Concept of Thermodynamic Cycle of Protein ties in commercialization because of their brittleness and rigid- Dynamics Cheng-Wei Ma, Feng Geng, Sugima Rappert and An-Ping ity. This characteristic precludes their use in products required Zeng* to be soft, elastic, or flexible. Recently we have developed Institute of Bioprocess and Biosystems Engineering, a biobased approach to manufacturing a branched lactone, Hamburg University of Technology, Hamburg, Germany β-methyl-δ-valerolactone (MVL), and demonstrated its potential as the first scalable rubbery polyester. We designed an artifi- Allosteric regulation is the regulation of an enzyme or other cial metabolic pathway to MVL by extending the mevalonate protein by binding an effector molecule at the protein’s al- pathway: introduction of fungal siderophore proteins to convert losteric site, a site other than the protein’s active site. It is mevalonate into anhydromevalonolactone; and reduction of a natural example of control loops, such as feedback from anhydromevalonolactone to MVL by enoate reductases. The downstream products or feed-forward from upstream sub- total biosynthetic pathway produced 300 mg/L of MVL directly strates. Control of allosteric regulation is required in order to from glucose in shake flask experiments. While directed evolu- drive metabolic flux toward desired levels. Although static tion is ongoing to optimize the MVL pathway, as an alternative structures are known for many proteins, the functions of pro- solution we developed a semisynthetic approach for the im- teins are governed ultimately by their dynamical character- mediate commercial-scale production of MVL. In this approach istics. Here, a novel strategy to engineer the ligand-induced we introduced the archaeal mevalonate pathway into produc- allosteric regulation is proposed based on a new concept tion host E. coli and engineered the pathway to produce ~80 of thermodynamic cycle of protein dynamics. In this con- g/L mevalonate in a bioreactor. We then performed two simple cept, allosteric process is considered as a thermodynamic chemical reactions, dehydration & hydrogenation, to produce process which is the energetic development of a thermody- MVL from mevalonate. The overall purification and reaction ef- namic system proceeding from an initial state to a final state. ficiency reached 90% . With the biobased MVL, we generated The thermodynamic cycle consists of a linked sequence of an amorphous rubbery polymer with a Tg at -51oC, which can thermodynamic processes: conformational change, ligand be used to not only prepare property-enhancing additives for binding and the allosteric process. A serial of computational polyester packaging, but also offers unprecedented opportuni- approaches are employed to obtain the state parameter of ties for developing thermoplastic elastomers, agricultural mulch

60 Poster Abstracts

films, chew gums and pressure sensitive adhesives with a Eukaryotic metabolism consists of a complex network of combined market size of more than $50 billion per year. enzymatic reactions and transport processes which are distributed over different subcellular compartments. Cur- 106. Co-Culture Based Modular Engineering for Aro- rently, available measurement techniques allow to measure matic and Aromatic-Derived Compounds Production in whole cell amounts which in several cases is not sufficient E. coli to describe the in-vivo kinetics in different compartments, Haoran Zhang*, Brian Pereira and Gregory N. Stephanopoulos which are driven by compartment specific concentrations. Chemical Engineering, Massachusetts Institute of Phosphate (Pi) is considered an essential component for the Technology, Cambridge, MA metabolic behaviour of glycolysis. Specially, the cytosolic Pi level is important for the balance between the upper and Biosynthesis of aromatic and aromatic-derived compounds lower glycolysis. One prominent example supporting the role in engineered microbe provides a robust means to produce of Pi is the disregulation of glycolysis in the tps1 knockout value-added molecules from inexpensive and renewable strain of Saccharomyces cerevisiae. TPS1 is expected to carbon sources. Here, we present a novel co-culture strategy act as cytosolic phosphate recovery mechanism, building a to biosynthesize aromatic and aromatic-derived compounds safe-guard to glycolysis. Understanding the concentration using engineered E. coli. As a proof of concept, we show of Pi in different compartments under dynamic conditions is that production of industrially important compound muconic critical to study the Pi metabolism, function and regulation. acid can be achieved in E. coli at high titer and yield. Spe- cifically, the shikimate pathway of E. coli was first engi- Fractionation, permeabilization, in vivo NMR, and metabolic neered to produce the key intermediate, dehydroshikimate sensor reactions have been applied to monitor the ratios of (DHS), as a starting molecule for aromatics biosynthesis. conserved moieties in subcellular compartments. Here we The downstream heterologous biosynthetic pathway and a developed a method that enables to monitor the dynamic DHS transporter were then engineered in E. coli to produce cytosolic Pi in S. cerevisiae by an equilibrium sensor reac- muconic acid at a titer of 500 mg/L in test tube. Next, the tion: maltose + Pi <=> D-glucose +D-glucose 1-phosphate. whole pathway was split into two modules, each of which The required enzyme, maltose phosphorylase from L. san- was carried by an individually engineered E. coli cell. The franciscensis was overexpressed in S. cerevisiae. Additional- upstream cell was responsible for DHS production, whereas ly, β-phosphoglucomutase from L. lactis was overexpressed the downstream cell was dedicated to importing DHS to to convert G1P to G6P which will have a higher concentra- produce muconic acid. Such a modular engineering strategy tion and is therefore measured more accurately. Furthermore, reduced the metabolic stress on each cell and resulted in a maltose hydrolase was knocked out to prevent a cycle. 40% production improvement, representing 12% of theoreti- The cytosolic free Pi concentration can then be calculated cal maximum yield. Furthermore, the two-cell system was from the measured intracellular glucose, G6P and optimized so that each cell grew on different sugars, which maltose concentrations. enabled the co-culture to simultaneously consume glucose and xylose, the major two components of lignocellulose, The sensor was applied to study Pi level in S. cerevisiae in to biosynthesize muconic acid without the glucose repres- several different conditions. The cytosolic free Pi concen- sion effect. The proposed strategy is readily applicable for tration in batch condition was 13.7mM in S. cerevisiae. We microbial production of a variety of valuable aromatics or also studied the Pi under different Pi supply conditions: Pi- aromatic-derived compounds from inexpensive and renew- excess (glucose-limited) chemostat and Pi-limited chemo- able carbon sources. stat at D=0.1h-1 and under dynamic pulse conditions e.g. from glucose-limited to glucose excess and from Pi-limited 107. A Fast Metabolic Sensor for in vivo Cytosolic to Pi-excess condition. Phosphate Concentration in Saccharomyces cerevisiae. Jinrui Zhang*1,2, Joseph. J. Heijnen1,2 and S. Aljoscha Wahl1,2 1Department of Biotechnology, Delft University of Technology, There was a significant difference between the cytosolic Delft, Netherlands and whole cell concentration. The steady-state cytosolic 2Kluyver Centre for Genomics of Industrial Fermentation, free Pi concentration in glucose-limited condition was 14.7 Delft, Netherlands mM, which corresponds to 31.4% of the whole cell free Pi

61 Poster Abstracts

concentration (46.8 mM). While in Pi-limited condition, the Fatty-Acid-Based Biofuel Production steady-state cytosolic free Pi concentration was 2.6 mM, Yongjin Zhou*, Buijs Klaas, Verena Siewers and Jens Nielsen which was about 15.7% of the whole cell Pi concentration Department of Chemical and Biological Engineering, Chalmers University of Technology, Göteborg, Sweden (16.6 mM). Additionally, dynamic pulse experiments from glucose-limited to glucose excess and from Pi-limited to Pi- Volatile energy costs and pressure to conserve fossil fuel excess condition revealed fast dynamics of Pi. The cytosolic resources have ignited efforts to produce biofuels and re- free Pi concentration increased 2.5 fold within 10 seconds and newable commodity chemicals via microbial fermentation of then decreased, which was unexpected as especially sugar- biomass. The fatty acid-based biofuels (FABB, biodiesel, al- phosphate are rapidly formed in the first seconds. The Pi kanes, alcohol, etc.) are considered as ideal alternative fuels balance was constructed from cytosolic Pi, Pi in nucleotides, for high-power machines as well as materials for biochemical Pi in glycolytic intermediates, Pi in other measured intracel- synthesis due to their high energy density and environmental lular metabolites, Pi in the vacuole which was calculated by friendliness. Though several FABB molecules have been mi- using whole cell Pi minus cytosolic Pi, and polyPi. The results crobialy synthesized,the titers remains to be largely improved showed that the immediately increase of cytosolic Pi level for industrial process. The natural oleaginous microorgan- was not caused by polyphosphate hydrolysis but due to the isms can accumulate cellular lipid to more than 60% (w/w) transport of Pi from the vacuole, as well as Pi uptake from the of dry cell weight, however, the scarcity of genetic tools and extracellular space. From these compartment specific Pi data metabolic intractability prevent their engineering for lipid hypothesis about the metabolic regulations can be derived derived FABB production. As Saccharomyces cerevisiae is a and integrated into thermodynamic and kinetic models. well-studied industrial model microorganism and the overall metabolism in S. cerevisiae, including the lipid metabolism 108. Cooperative Co-Culture of Escherichia coli and Saccharomyces cerevisiae for Overproduction of Pacli- is well studied, it is feasible to engineer an intracellular lipid taxel Precursors biosynthesis pathway for FABB overproduction into S. cere- Kang Zhou*, Kangjian Qiao, Steven Edgar and Gregory visiae. Therefore, nature inspired reconstruction of the lipid Stephanopoulos metabolism regulation of oleaginous microorganism in S. Department of Chemical Engineering, Massachusetts Insti- cerevisiae should provide a novel platform for FABB produc- tute of Technology, Cambridge, MA tion. According to the tans-omics study in the candidate’s former lab, we partition the overall reconstructed FABB In nature, there are many examples of microbial consortia biosynthesis into four modules: module 1, upstream/precur- that can efficiently accomplish chemically difficult processes sor; module 2, relative pathways for lipid anabolism and through the division of labor among different species. To catabolism; module 3, downstream modification module for date, such strategy has not been fully explored in engineer- FABB synthesis from the lipid intermediates; module 4, regu- ing microbes for production of valuable metabolites. In this lation module. We will increase and balance metabolic flux study, we demonstrated a synthetic cooperative microbial to FABB biosynthesis by the multivariate engineering of the system to produce precursors of anti-cancer drug paclitaxel modules. The multidisciplinary and inter-sectorial approach by using two model laboratory microbes, E. coli and S. cere- of this project will create conditions for high yield conversion visiae. The two species with distinct advantages cooperated of biomass into target biofuels, inhibition of by-product for- synergistically in two levels, cell growth and taxane produc- mation (e.g. ethanol) and improvement of FABB productivity tion. This design could also make the engineering of paclitaxel biosynthetic pathway to be more modular, because 110. Engineering a Balanced Mevalonate Pathway each species carrying part of the pathway can be optimized in E.coli individually and then simply mixed together to complete the Jorge Alonso-Gutierrez* whole pathway. Because of these advantages, the concept Fuel Synthesis, Joint Bioenergy Institute (JBEI), Emeryville, CA demonstrated in this project, cooperation of multiple cells with their own specialties, should be generally applicable to In the present study we showed two different approaches for many other similar processes. balancing heterologous MEV gene expression in E.coli and improve production from 1% glucose: i) Engineer a dynamic 109. Systematic Engineering of Lipid Metabolism for regulation of the pathway using promoters responding to an

62 Poster Abstracts

accumulation of the toxic intermediate FPP, which improved Advanced Water Management Centre (AWMC), The University bisabolene titer to 875 mg/L ii) Explore expression of the of Queensland, St Lucia, Australia 2 MEV pathway enzymes using targeted proteomics and its Australian Institute for Bioengineering and Nanotechnology (AIBN) associated limonene production to define engineering strategies; a principal component analysis (PCA) of the proteomics Adipic acid is a main precursor for the production of nylon-6,6 and the production at various conditions suggests that an and polyurethanes. To date it is still derived from the precursor over-expression of the bottom portion of the pathway and a benzene in a process of oxidation of cyclohexane to cyclo- moderate expression of the top portion are keys to improve hexanol/cyclohexanone. While these precursors are all petro- the production. Indeed, extra copies of the terpene synthase chemistry based and thus non-sustainable, the further oxida- and a modified top portion led to 665 mg/L and 1150 mg/L tion with nitric acid has an additional environmental impact as of limonene and bisabolene productions, respectively. This the greenhouse gas N O is a large by-product in this process. represents a 40% increase in production titer over previous, 2 unbalanced systems using the same genes. The bio-production of adipic acid offers a promising alternative to chemical synthesis. There is currently a strong interest in the 111. Yeast Mitochondrial Engineering: Targeting the Powerhouse of the Cell for Advanced Biofuel Production metabolic engineering community (both commercial and sci- Jose L. Avalos*1,2, Gerald Fink2 and Greg Stephanopoulos3 entific) to develop a biological replacement process. There are 1Department of Chemical Engineering, Massachusetts Insti- mainly three types of approaches that can be distinguished: (i) tute of Technology, Cambridge, MA Direct production of adipic acid, (ii) production of the unsaturat- 2 Whitehead Institute, Cambridge, MA, ed adipic acid precursor cis,cis-muconic acid and (iii) produc- 3Chemical Engineering, MIT, Cambridge, MA tion of D-glucaric acid. Muconic and glucaric acid are subsequently reduced to adipic acid by chemical hydrogenation. Yeast metabolic engineering has mostly focused on the construction of metabolic pathways in the cell cytoplasm. In the first part of this study, all possible sub-routes of the However, there is huge potential in harnessing the diversity plenty of available pathways to adipic acid were identified. of environments, metabolites and enzymes that exist in the In total 16 different production routes based on dextrose or different organelles of eukaryotic cells, to enhance engineered fatty acids as carbon-feed exist. These routes were compared metabolic pathways. We developed new molecular tools to under aerobic and anaerobic conditions in E. coli and S. cere- expedite the simultaneous construction of multiple metabolic visiae, using a new approach combining elementary flux mode pathways targeted to either the cytoplasm or the mitochondria analysis with network embedded thermodynamic analysis, for of yeast. With these tools, we show that targeting biosynthetic feasibility and maximum yield. pathways to the mitochondria of the yeast Saccharomyces cerevisiae improved the production of advanced biofuels by as One approach that is currently of particular attraction is to much as 500%, compared to identical pathways targeted to derive the precursor muconic acid from shikimate pathway. the cytoplasm. The mechanisms of this enhancement include After early works by the Frost laboratory in the 1990’s there has (1) the elimination of metabolic bottlenecks, (2) increased been a renaissance in interest in this topic as can be seen from availability of intermediates to the engineered pathways, and recently published work by three independent laboratories (3) increased local concentrations of enzymes due to their con- (Boles, Alper and Yan). We noticed that all three approaches finement inside the smaller volume of mitochondria. Our study suffer from very low product titres (in the mg / L range) and opens multiple new avenues to target metabolic pathways to decided to complement our study with an in silico comparison yeast mitochondria and affect mitochondrial physiology for the of the strain construction strategies. benefit of engineered pathways.

112. Combining Elementary Mode Analysis with a Net- Among the 16 different routes the theoretical mass yields work Embedded Thermodynamic Approach for Analysis ranged from 0.07 to 1.16 g g-1 on glucose and/or palmitate, of Microbial Adipic Acid Production highlighting the importance of pathway choice. Infeasibility Nils J. H. Averesch*1, Verónica S. Martínez2 and Jens O. patterns were determined for each pathways flux distribution, 1 Krömer revealing that all but one pathway contained reactions that 1Centre for Microbial Electrosynthesis (CEMES) within the limit product formation through a thermodynamic equilibrium 63 Poster Abstracts

lying on the substrate side. Almost half of these appeared to engineered by knocking-out the adhE1 encoding aldehyde/ have an equilibrium that deemed the whole pathway infeasi- alcohol dehydrogenase to prevent solvent-production. The ble. This problem also affected recently patented routes. The simultaneous deletion of the pta-ctfB-adhE1 in C. acetobu- pathway feasible over a large range of concentrations and pH tylicum resulted in metabolic switch from biphasic to acido- values was the production of muconic acid via the shikimate genic fermentation, which enhanced butyric acid-production. pathway derived tryptophan precursor anthranilic acid (496 [This work was supported by the Technology Development mgmuconic gglucose-1). In the second part of this study we Program to Solve Climate Changes on Systems Metabolic used in depth elementary mode analysis of shikimate path- Engineering for Biorefineries from the Ministry of Science, way based routes to determine new knock-out strategies to ICT and Future Planning (MSIP) through the National enhance muconic acid production. Research Foundation (NRF) of Korea (NRF-2012-C1AAA001- 2012M1A2A2026556).] We found that the availability of sensible knock out targets depends not only on the pathway but also on the organism, 114. Development of Next Generation Yeast Strains for allowing significantly different maximum yields. In addition it Ethanol Production from Lignocellulosic Feedstocks Paul de Waal1, Wouter Hartman1, Hans de Bruijn1, Aloys was possible to create a scenario where product formation is Teunissen1, Rene de Jong1, Paul Klaassen11, Wouter directly coupled to central metabolism, allowing the introduc- Wisselink2, Jack T. Pronk2, Ton van Maris2 and Viktor Boer*1 tion of a minimum yield constrain of 45%. This is particularly 1DSM Biotechnology Center, Delft, The Netherlands applicable in S. cerevisiae. While this is an in silico analysis, 2Department of Biotechnology, Delft University of Technology, we believe it adds an important angle to the current efforts Delft, The Netherlands in creating bio-based muconic acid production and will help researchers to prioritize their research efforts. Lignocellulosic feedstocks are considered to be of great economic and environmental significance for sustainable pro- The outcomes may also be applicable to other target com- duction of valuable fermentation products. For cost-effective pounds in different pathways where similar bioconversion and efficient industrial processes, complete and fast fermen- steps are involved. tation of all biomass sugars is a prerequisite. In this respect, the main challenge emerging from the use of lignocellulosic 113. Metabolic Reconstruction of Clostridium feedstocks for ethanol production by Saccharomyces cerevi- Acetobutylicum for Enhanced Production of Butyric Acid siae is the efficient fermentation of the pentose sugars xylose Junho Bang*, Yu-Sin Jang and Sang Yup Lee and arabinose, as these sugars cannot be utilized by natural Chemical & Biomolecular Engineering, Korea Advanced S. cerevisiae strains. Another significant challenge is the Institute of Science and Technology (KAIST), Daejeon, South Korea inevitable presence of inhibitors, such as furfural and acetic acid, which are formed during pretreatment and hydrolysis of Butyric acid production by fermentation of renewable re- the feedstocks and severely inhibit yeast growth and product sources has been receiving much attention as an alternative formation. to petroleum-derived butyric acid because of the consumers’ preference towards natural ingredients for food, pharmaceu- DSM has developed advanced yeast strains by introducing tical, animal feed supplement, and perfume. Clostridium ace- heterologous pathways into robust S. cerevisiae hosts which tobutylicum has been considered an attractive platform-host resulted in the ability to ferment xylose and arabinose in for biorefinery due to its metabolic diversity. Considering its lignocellulosic hydrolysates. Subsequently, by the application capability to overproduce butanol through butyrate, it was of evolutionary engineering the total time required to ferment thought that butyric acid can also be efficiently produced by hexoses and pentoses in lignocellulosic sugar mixtures was this bacterium through metabolic engineering. The pta-ctfB- significantly reduced. deficientC. acetobutylicum CEKW, in which genes encoding phosphotransacetylase and CoA-transferase were knocked- In order to take these developments to the next level, a out, was assessed for its potential as a butyric acid-producer toolbox for the new generation of advanced yeast strains in fermentations with four controlled pH-values at 5.0, 5.5, is in development, in which the main challenges are further 6.0, and 6.4. Furthermore, the CEKW strain was further addressed by accelerating pentose fermentation through

64 Poster Abstracts

expression of engineered pentose transporters, as well as enzyme, 1-deoxy-D-xylulose-5-phosphate synthase (dxs), converting lignocellulosic inhibitors and biorefinery waste releases the primary main bottleneck in isoprenoid biosyn- streams into additional ethanol. A first glance at these new thesis6, 7; however, we have observed significant differences developments will be presented. in the “engineerability” of even genetically closely related strains through dxs overexpression. In particular, an isolate of 115. Identifying the Source of Strain-to-Strain Variabil- the widely used laboratory strain E. coli MG1655 produced ity in Isoprenoid Production Capacity of E. coli Using a 5-fold less lycopene than its parental wild-type strain WG1, Systems Biology Approach suggesting a loss of isoprenoid production capacity dur- Mareike Bongers*, James B.Y.H. Behrendorff, Claudia E. Vickers and Lars K. Nielsen ing the genetic modifications made to obtain this laboratory Australian Institute for Bioengineering and Nanotechnology, strain. In order to understand these phenotypic differences, The University of Queensland, St Lucia, Australia we are currently conducting systems biology analyses using genomics and proteomics. Full genome sequences were ob- Isoprenoids are a large class of natural compounds with tained for several strains of interest for which genomes were diverse industrial applications ranging from pharmaceuticals not publicly available. Comparative genomics using these to biofuels. Extraction of isoprenoids from their native hosts sequences was identified mutations in candidate genes that is often costly and inefficient since many high-value iso- might be related to the observed phenotypic differences. In prenoids are produced in low quantities. Therefore heter- the WG1 family (including K-12 strains), no single or simple ologous production in microorganisms presents a desirable set of gene mutations can explain the phenotypic differences alternative. In spite of their remarkable chemical diversity, all and proteomics across the lycopene-production period was isoprenoids are derived from the same five-carbon precur- used to identify the effect of more subtle differences. We are sors: isopentenyl diphosphate (IPP) and dimethylallyl diphos- currently evaluating target genes through complementation phate (DMAPP). This enables metabolic engineers to access and expect this to deepen our understanding of MEP path- a myriad of valuable compounds through overproduction of way flux regulation and suggest new targets for isoprenoid these basic building blocks. Of the two different metabolic engineering in E. coli. routes leading to isoprenoid precursors, the methylerythritol References phosphate (MEP) pathway used by E. coli and many other bacteria has a higher theoretical maximum yield than the 1. Rude, M.A. & Schirmer, A. New microbial fuels: a biotech perspective. Current opinion in microbiology 12, 274-281 (2009). mevalonate (MVA) pathway1. However, despite intensive metabolic engineering efforts, isoprenoid product yields via 2. Ajikumar, P.K. et al. Isoprenoid pathway optimization for Taxol precursor overproduction in Escherichia coli. Science 330, 70-74 the E. coli MEP pathway remain far below the theoretical (2010). maximum. The factors influencing isoprenoid biosynthesis in 3. Zou, R., Zhou, K., Stephanopoulos, G. & Too, H.P. Combina- E. coli are currently poorly understood. Unpredictable results torial engineering of 1-deoxy-D-xylulose 5-phosphate pathway from MEP pathway engineering – including inexplicable using cross-lapping in vitro assembly (CLIVA) method. PloS one 8, production thresholds and reduced product formation when e79557 (2013). all pathway enzymes are highly over-expressed – suggest 4. Boghigian, B.A., Salas, D., Ajikumar, P.K., Stephanopoulos, G. 2, 3 complex regulatory mechanisms . While it is known that & Pfeifer, B.A. Analysis of heterologous taxadiene production in K- different E. coli strains vary in their ability to produce isopren- and B-derived Escherichia coli. Applied microbiology and biotech- oids 4, 5, most previous studies have focussed on E. coli nology 93, 1651-1661 (2012). strains from the K-12 lineage. We have systematically tested 5. Chae, H.S., Kim, K.H., Kim, S.C. & Lee, P.C. Strain-dependent E. coli laboratory strains from different phylogenetic groups carotenoid productions in metabolically engineered Escherichia coli. for their ability to produce lycopene, a model compound Applied biochemistry and biotechnology 162, 2333-2344 (2010). for evaluating isoprenoid production in microorganisms. We 6. Harker, M. & Bramley, P.M. Expression of prokaryotic 1-deoxy-d- observed differences in lycopene production of up to an or- xylulose-5-phosphatases in Escherichia coli increases carotenoid der of magnitude between strains. Of particular significance and ubiquinone biosynthesis. FEBS Letters 448, 115-119 (1999). for metabolic engineering applications, we also show that 7. Wang, H.H. et al. Programming cells by multiplex genome engi- different wild-type E. coli strains respond differently to MEP neering and accelerated evolution. Nature 460, 894-898 (2009). pathway engineering. Overexpression of the rate-limiting

65 Poster Abstracts

116. Rational Metabolic Engineering of Baker’s Yeast for University of Queensland, Brisbane, Australia Production of 3-Hydroxypropionic Acid

Irina Borodina* Monoterpenes are ten-carbon (C10) liquid hydrocarbons The Novo Nordisk Foundation Center for Biosustainability, that can serve as immediate precursors for replacement Technical University of Denmark, Hørsholm, Denmark jet fuels. In a demonstration flight in June 2012, Amyris Inc. showed that their biojet fuel, AMJ-700, met engine Metabolic engineering distinguishes itself from the classi- performance requirements and required no changes to the cal engineering disciplines, such as electrical or chemical aircraft. Monoterpenes make up 60% of this jet fuel mix- engineering, by lower predictability of design outcome ture and are essential for meeting the strict physiochemical and long turnover times of the build and testphases. This specifications. Fermentative production of monoterpenes situation has however been rapidly improving over the past in industrially relevant microbial hosts, such as Saccharo- two decades.* Advanced systems biology and genome- myces cerevisiae, is severely hindered by product toxic- scale modeling approaches allow more precise simula- ity. The first part of this work describes the physiological tions of the complex biological systems, thus improving impact of limonene on the yeast cell envelope. Membrane predictions of the metabolic engineering targets. The novel fluidity, composition and structural integrity were found synthetic biology tools accelerated genome editing and undisturbed during limonene stress. Surprisingly, limonene enabled high-throughput screening using synthetic signal- caused damage to the cell wall lattice structure. The data ing circuits, i.e., biosensors. is the first report demonstrating that cell wall, rather than the plasma membrane, deterioration is the source of toxic- We engineered baker’s yeast for production of 3-hydroxy- ity1. Secondly, we describe two strategies to overcome propionic acid (3HP). 3HP can be chemically dehydrated limonene toxicity: (1) extractive fermentation employing into acrylic acid and thus serve as a biosustainable build- farnesene as a co-product and extractant2 and (2) adaptive ing block for acrylate-based products (diapers, acrylic laboratory evolution (ALE). For ALE, strains were evolved paints, acrylic polymers, etc.) We considered several bio- through serial passage achieving a 5-fold improvement synthetic pathways leading to 3HP, screened for efficient in solvent tolerance. Strains collected across the evolu- enzyme variants and optimized gene expression using tion process were sequenced and mutations identified. novel synthetic biology tools that we developed for stable The original strain was engineered to carry one or more single and multi-copy integration of genes into the yeast of these mutations, and it was found that a single muta- genome. Engineering of precursor and co-factor supply by tion could confer the tolerant phenotype. We are currently using both rational and model-guided approaches and op- characterizing this mutation using a full complement of timizing fermentation parameters helped to further improve systems biology tools. 3HP titer, production rate and yield. Adaptive laboratory 1. Brennan, T.C.R., Krömer, J.O. & Nielsen, L.K. Physiological evolution followed by genome re-sequencing, transcrip- and transcriptional responses of Saccharomyces cerevisiae to tome analysis and reverse engineering allowed to decipher d-limonene show changes to the cell wall but not to the plasma 3HP tolerance mechanism. membrane. Appl Environ Microbiol 79, 3590-3600 (2013). 2. Brennan, T.C.R., Turner, C.D., Krömer, J.O. & Nielsen, L.K. In summary, we show how modern synthetic biology and Alleviating monoterpene toxicity using a two-phase extractive metabolic engineering tools enable rapid strain optimiza- fermentation for the bioproduction of jet fuel mixtures in Saccha- tion in iterative design-build-test cycles. romyces cerevisiae. Biotechnol Bioeng 109, 2513-2522 (2012). *Borodina I, Nielsen J (2014). “Advances in Metabolic Engineer- ing of Yeast for Production of Chemicals”. Biotechnol J. DOI: 118. Molecular Approaches to Improve 1-Butanol 10.1002/biot.201300445 Tolerance and Production in Escherichia coli Le Minh Bui*, Ziaur Rahman, Almando Geraldi, Kyo Hun 117. Understanding and Overcoming Monoterpene Tox- Kang, Jun Hyoung Lee and Sun Chang Kim icity in Yeast for the Production of Renewable Jet Fuels. Biological Sciences, Korea Advanced Institute of Science Timothy Brennan*1, Jens Kroemer2 and Lars K. Nielsen1 and Technology, Daejeon, South Korea 1Australian Institute for Bioengineering and Nanotechnol- ogy (AIBN), University of Queensland, Brisbane, Australia 1-butanol toxicity is a major limiting factor for improving 2Centre for Microbial Electrosynthesis (CEMES), The 1-butanol production in either a native 1-butanol pro- 66 Poster Abstracts

ducer (Clostridium) or non-native 1-butanol producing 20 g/L of glucose and 10 g/L of l-lysine, 3.6 g/L of 5AVA was microbial hosts, especially Escherichia coli. Because of a produced by converting 7 g/L of l-lysine. When the davAB low 1-butanol tolerance of 1 – 2% [v/v], the capability of genes were introduced into recombinant E. coli strain XQ56 these strains to produce 1-butanol with titers higher than allowing enhanced l-lysine synthesis, 0.27 and 0.5 g/L of the tolerance limits is severely reduced. While the 1-buta- 5AVA were produced directly from glucose by batch and nol tolerance of Clostridium strains has been intensively fed-batch cultures, respectively. Further conversion of 5AVA studied and improved via different approaches, concerns into glutarate could be demonstrated by expression of the on enhancing the tolerance of E. coli just has been raised P. putida gabTD genes encoding 5AVA aminotransferase recently and efforts so far did not result in any significant and glutarate semialdehyde dehydrogenase. When recom- improvement. These researches, moreover, did not show binant E. coli WL3110 strain expressing the davAB and any linkage between the improved 1-butanol tolerance and gabTD genes was cultured in a medium containing 20 g/L production, since the host, E. coli, is incapable of producing glucose, 10 g/L l-lysine and 10 g/L α-ketoglutarate, 1.7 g/L 1-butanol naturally. Previously, we screened and identified of glutarate was produced. [This work was supported by the a 1-butanol tolerant Artificial Transcription Factor (ATF) that Technology Development Program to Solve Climate Chang- increases the tolerance of E. coli in 1.5% [v/v] 1-butanol es on Systems Metabolic Engineering for Biorefineries from (published). On the other hand, we also selected rational the Ministry of Education, Science, and Technology (MEST) approaches which were successfully applied to other hosts through the National Research Foundation of Korea(NRF- and evaluated their effects on E. coli tolerance. In this work, 2012-C1AAA001-2012M1A2A2026556). Further support by therefore, we study the impact of different groups of genes, the World Class University program (R32-2008-000-10142- including ATF, molecular chaperones, solvent efflux pump, 0) of the MEST are appreciated.] etc. on both improving 1-butanol tolerance and production of E. coli by co-expressing them with the clostridial 120. Synthetic Regulatory Small RNAs for Genome-Wide 1-butanol pathway. The result of this work would show a Metabolic Engineering Tong Un Chae*1, DoKyun Na2, Seung Min Yoo1, Hannah more practical evaluation on the significance of suggested Chung1, Hyegwon Park1 and Sang Yup Lee1 approaches for improving 1-butanol tolerance in E. coli as a 1Chemical & Biomolecular Engineering, Korea Advanced tool to increase the production capacity. Institute of Science and Technology (KAIST), Daejeon, South Korea 119. Metabolic Engineering for Production of 5- 2School of Integrative Engineering, Chung Ang University, Aminovalerate and Glutarate Using Escherichia coli Seoul, South Korea Tong Un Chae*1, Si Jae Park2, Eun Young Kim3, Won Noh3, Hye Min Park1, Young Hoon Oh3, Seung Hwan Lee3, Bong Optimized modulation of metabolic fluxes through the 3 3 1 Keun Song , Jonggeon Jegal and Sang Yup Lee control of gene expression is one of the key challenges in 1Chemical & Biomolecular Engineering, Korea Advanced metabolic engineering. Here, we developed rational de- Institute of Science and Technology (KAIST), Daejeon, South Korea sign principles for synthetic regulatory small RNAs (sRNAs) 2Department of Environmental Engineering and Energy, for adjustable expression control. We then expanded our Myongji University, Yongin, South Korea method to create a system utilizing synthetic sRNAs as a 3Korea Research Institute of Chemical Technology, Daejeon, portable and conditional chromosomal gene controller, and South Korea engineered Escherichia coli to produce tyrosine and cadaverine as a model. An engineered E. coli strain (tyrR csrA 5-Aminovalerate (5AVA) is the precursor of valerolactam repressed S17-1 strain) capable of producing 21.9 g/L of which is a potential building block for nylon 5, and is a tyrosine was developed by combinatorial knockdown ex- C5 platform chemical for synthesizing 5-hydroxyvalerate, periments on various candidate genes in 14 different strains glutarate, and 1,5-pentanediol. Escherichia coli was meta- using respective synthetic sRNAs. As another example, this bolically engineered for the production of 5-aminovalerate strategy was applied to an already metabolically engineered (5AVA) and glutarate. When the recombinant E. coli WL3110 strain producing cadaverine by applying a library of 130 strain expressing the Pseudomonas putida davAB genes synthetic sRNAs. Repression of murE allowed 55% increase encoding delta-aminovaleramidase and lysine 2-monooxy- in cadaverine production. The feasibility of using synthetic genase, respectively, were cultured in a medium containing

67 Poster Abstracts

sRNAs to modulate gene expression holds great promise in physiological conditions and it can be used for the design of next-generation metabolic engineering and synthetic biol- any metabolic network of interest. ogy applications. [This work was supported by the Technol- ogy Development Program to Solve Climate Changes on 122. Direct Fermentation for Isobutene, Butadiene and Systems Metabolic Engineering for Biorefineries from the Propylene Production : A Highway to Renewable Plastics, Synthetic Rubber and Fuels. Ministry of Science, ICT and Future Planning (MSIP) through Romain Chayot* the National Research Foundation (NRF) of Korea (NRF- Strain Construction, Global Bioenergies, EVRY, France 2012-C1AAA001-2012M1A2A2026556).] As of today, most industrial bioproduction processes are 121. Analysis of Aerobic-to-Anaerobic and Anaerobic- based on naturally existing metabolic pathways, limiting to-Aerobic Switches in E. coli Using Large-Scale Dy- the scope of industrial biology, and preventing the access namic Metabolic Models Anirikh Chakrabarti*1,2, Geogios Fengos1,2,, Meric Ataman1,2, to many of the chemistry’s largest markets. The purpose Keng Cher Soh1,2,, Ljubisa Miskovic1,2, and Vassily Hatzi- of Global Bioenergies, is to develop innovative metabolic manikatis1,2,3 pathways for the production of light olefins from renewable 1Laboratory of Computational Systems Biotechnology, resources, by direct fermentation. EPFL, Lausanne, Switzerland 2Swiss Institute of Bioinformatics, Lausanne, Switzerland Light olefins (ethylene, propylene, linear butylene, isobu- 3Institute of Chemical Sciences and Engineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, tylene and butadiene) are the core of the petrochemical Switzerland industry, and each one represent a multi-billion dollars market. These markets have been shaped by the historical Dynamic nonlinear models of metabolism offer significant co-production of light olefins from fossil oil in steam crack- advantages as compared to traditional steady-state coun- ers. The recent use of large quantities of shale gas in steam terparts for the analysis of metabolic networks and identifi- crackers, resulting mostly in the production of ethylene, has cation of metabolic engineering strategies. Uncertainties in changed the market landscape, emphasizing the need for the metabolic network structure, kinetic rate laws, and their alternative production routes for the other olefins. However, corresponding parameters are limiting the development of light olefins are not naturally produced by microorganisms systematic methodologies for construction of large-scale, and no bioprocess to convert renewable resources to these dynamic metabolic models. In this study, we employed molecules has been industrialized so far. a novel methodology to construct dynamic, large-scale, nonlinear models of E. coli metabolism. The parameters of Global Bioenergies has developed an artificial metabolic these models were computed in two steps: (i) the ORACLE pathway including all the necessary enzymatic reactions framework is used to integrate thermodynamics, avail- from feedstock to isobutene. In contrast with most for- able omics, and kinetic data and construct a population mer approaches, the metabolic route leading to isobutene of log-linear kinetic models; (ii) the log-linear models are includes non-naturally occurring reactions as key steps, for used to compute kinetic parameters of nonlinear enzymatic example the decarboxylation of hydroxyisovaleric acid into mechanisms in the network and we developed a population isobutene. Since all these reactions are enzymatic, isobu- of nonlinear kinetic models. We used a metabolic network tene can be obtained by direct fermentation, e.g. a process composed of 140 metabolites and 225 reactions, and we wherein all the chemical transformations are carried on by analyzed the dynamic responses of E. coli upon switch the host microorganism. from aerobic respiration to anaerobic, and vice versa. This analysis allowed us to characterize the time constants of the The scale-up of this process is in progress. Importantly, different functional parts of E. coli metabolism such as gly- production of a volatile compound such as isobutene (and colysis, TCA, and oxidative phosphorylation. Furthermore, other light olefins) by direct fermentation presents two major we analyzed the network responses upon various perturba- advantages: first, in contrast with what would happen with tions of the network components to explore the robustness ethanol or isobutanol for example, the product is spontane- and sensitivity of the operational states. The methodology ously removed from the culture broth, which alleviates the presented here is independent of the organism and the limitations linked with titer issues. Second, the purification

68 Poster Abstracts

process is considerably easier and cheaper since no energy systems-level metabolic engineering and enzyme engineer- consuming methods such as distillation or phase separation ing allowed one-step production of PLA and its copolymers are necessary to purify the end product. in E. coli. [“This work was supported by the Technology Development Program to Solve Climate Changes from Scale-up will be carried out in two industrial pilots. A first pi- National Research Foundation of Korea (Development of lot installed in Pomacle-Bazancourt in France with an annual systems metabolic engineering platform technologies for capacity of 10 tons of oxidation-grade isobutene will target biorefineries; NRF-2012-C1AAA001-2012M1A2A2026556) applications such as methacrylic acid (a paint additive) and and Intelligent Synthetic Biology Center (2011-0031963) of organic glass. The second pilot, installed on the refinery Korea through the Global Frontier Research Program of the site of Leuna, Germany, will have an annual capacity of 100 Ministry of Education, Science and Technology (MEST).”] tons of polymer-grade isobutene compatible with the rest of isobutene’s wide product tree including fuels and rubber 124. Production of Native-Sized Spider Dragline Silk applications. Protein through Metabolic Engineering Approach in Escherichia coli Hannah Chung*1, Xiao-Xia Xia2 and Sang Yup Lee1 Finally, while the isobutene process is progressing towards 1Chemical & Biomolecular Engineering, Korea Advanced industrial scale, Global Bioenergies is also developing new Institute of Science and Technology (KAIST), Daejeon, artificial metabolic pathways enabling direct bio-production South Korea of Butadiene and Propylene. 2School of Life Sciences and Biotechnology, Jiao Tong University, Shanghai, China 123. Biosynthesis of Lactate-Containing Polymers in Metabolically Engineered Escherichia coli Naturally found spider silk and elastin protein attract at- So Young Choi*1, Yu Kyung Jung1, Si Jae Park2 and Sang tention due to its outstanding physical properties coming Yup Lee1 from the highly repeated structure and size. However, the 1 Chemical & Biomolecular Engineering, Korea Advanced exceptional structure and size limits expression in heter- Institute of Science and Technology (KAIST), Daejeon, ologous hosts, where the repetitive sequences in mRNA South Korea 2Department of Environmental Engineering and Energy, create extensive secondary structures. And these structures Myongji University, Yongin, South Korea decrease ribosome processivity and assist mRNA degradation. Using the naturally found protein, spider dragline silk Polylactic acid (PLA) has been considered as a good alter- protein, we present techniques to solve biological problems native to petroleum-based plastic as it possesses several that occurred: using metabolic engineering and increasing desirable properties such as biocompatibility, biodegrad- the cellular resource, more specifically, particular amino acid ability, and compostability. However, current industrial PLA tRNA pool. Newly synthesized native-size spider dragline production depends on the two-step process: fermentative silk protein produced increased titer than those reported production of lactic acid (LA) followed by chemical polym- previously, therefore proving that the strategies used were erization with several catalysts. In this study, we were able efficient. The results provide insight into approaches to to produce PLA and PLA-containing biopolymers using control expression of recombinant proteins containing high metabolically engineered Escherichia coli. Introduction of molecular weight and highly repetitive sequence. [This work the heterologous metabolic pathways involving engineered was supported by the Intelligent Synthetic Biology Center of propionate CoA-transferase and polyhydroxyalkanoate Global Frontier Project funded by the Ministry of Education, synthase resulted in synthesis of PLA and P(3-hydroxy- Science and Technology. Also this work was supported by butyrate-co-LA) [P(3HB-co-LA)] in E. coli. Metabolic path- the Converging Research Center Program (2009-0082332) ways of the E. coli strain were further engineered based on of the Ministry of Education, Science, and Technology in silico genome-scale metabolic flux analysis. Using the (MEST) through the National Research Foundation (NRF)] metabolically engineered strain, PLA homopolymer and P(3HB-co-LA) copolymers containing up to 70 mol% lactate 125. Splitting the E. coli Metabolism for the Production of Fructose-6-P Derived Chemicals could be produced up to 11 wt% and 46 wt%, respectively, Pieter Coussement*, Jo Maertens, Joeri Beauprez, Mathias from glucose. Thus, in this study, the strategy of combined Duhamel, Wouter Van Bellegem, Dries Van Herpe, Marjan

69 Poster Abstracts

De Mey and Wim Soetaert The authors like to acknowledge Inbiose. Inbio.be, Department of Biochemical and Microbial Technology, Ghent University, Ghent, Belgium 126. Engineering S. cerevisiae Metabolism for Efficient Production of Acetyl-CoA Derived Products In the past, coupling product formation with growth has prov- Charles Denby* en to be a highly successful metabolic engineering strategy Fuels synthesis, Joint Bioenergy Institute (JBEI), for the biotechnological production of numerous molecules, Emeryville, CA e.g., succinate (Hong, 2002). The guaranteed (high) product An ideal microbial platform for efficient biosynthesis of valu- yield, the fast recuperation of cofactors and abundance of able chemicals would rapidly convert sugars from biomass intermediates are some of the main advantages. directly into a desired chemical compound. The yeast S. cerevisiae has many attractive features as a biosynthesis Such a strategy was developed for the production of fruc- platform, however, its inability to convert glucose directly tose-6-phosphate derived molecules in E. coli, as many high to acetyl-CoA-derived compounds at high glucose con- added value molecules, such as for example chito-oligomers, centrations limits its productivity. My research focuses on are derived from fructose-6-P. replacing the endogenous cytoplasmic acetyl-CoA production pathway with an exogenous pathway suited for rapid However, compared to previous examples of this strategy, conversion of glucose to acetyl-CoA-derived compounds. fructose-6-phosphate is a key metabolite in central metabo- Acetyl-CoA serves as a molecular building block for many lism, that intervenes in the glycolysis, the pentose phosphate important and valuable compounds, and this research will shunt, etc. which renders such a strategy intrinsically more establish a platform with higher productivity for biosynthesis challenging. of acetyl-CoA-derived compounds.

The coupling between fructose-6-phosphate production 127. Engineering Yeast to Produce Fatty Acid-Derived and growth was achieved by using the heteromeric sub- Fuels and Chemicals. strate sucrose, which is cleaved by a heterologously ex- Leopold d’Espaux*, Weerawat Runghphan and Jay pressed sucrose phosphorylase into glucose-1-phosphate D. Keasling and fructose. A knock-out strategy was developed using Lawrence Berkeley National Lab, CA metabolic models and C13 labeled flux analysis to preserve the fructose moiety for the production of fructose-6-phos- A rising world population and rapid economic development phate derived molecules, while growth is fueled by metabo- continue to increase energy demands. Currently, most of lizing glucose-1-phosphate, hence, resulting in guaranteed our energy needs are met by fossil fuel technologies, which product yields and a reasonable productivity. are unsustainable, ecologically risky, and often associated with volatile pricing and challenging political factors. Micro- To achieve this split metabolism, since fructose-6-P is a bial-based fuels production technology offers a greener, key intermediate in the central metabolism, the fluxes of the sustainable alternative. Here we engineer baker’s yeast to central carbon metabolism have to be redirected severely, produce fatty acid-derived fuels and chemicals. We opti- e.g., the glycolysis has to be split by knocking out the genes mize the endogenous fatty acid pathway naturally used for pgi and pfkAB, which prevents the flow from fructose- energy storage and membrane composition, and redirect 6-phosphate through the glycolysis to the Krebs cycle. it using terminal enzymes to produce free fatty acids, fatty Different additional knockouts were evaluated in order to acid ethyl esters, and fatty alcohols. achieve a complete split metabolism. 129. Engineering a Functional Deoxyxylulose Phosphate (DXP) Pathway in Saccharomyces cerevisiae. C13 labeled flux analysis, using OpenFlux, was performed to Kevin Dietzel*, Eugene Antipov, Gale Wichmann, Nathan evaluate the fluxome of the various knock-out mutants, which Moss, Peter Jackson, Sara Gaucher, Shayin Gottlieb, Jack ultimately showed that the fructose moiety of sucrose can D. Newman and Lishan Zhao be completely preserved for the formation of fructose-6-P Amyris Inc, Emeryville, CA derived products, rendering such a strain the ideal base strain for the development of a wide variety of production hosts. Isoprenoids and their chemical derivatives are used in many 70 Poster Abstracts

commercial and industrial products. Producing these com- 131. The Importance of the Lipid Biosynthetic Pathway pounds via the engineering of microbial systems is an attrac- for Glycolipids Production in Engineered E. coli Cells. tive alternative to extraction from their native source because Neus Mora-Buyé, Magda Faijes* and Antoni Planas Institut Químic de Sarrià - Universitat Ramon Llull it can provide a more sustainable source, a more stable supply, and/or lower production costs. All isoprenoids are formed The production of glycolipids as biosurfactants is attract- from the two activated hydrocarbon monomers isopentenyl ing much interest due to their potential advantages over diphosphate (IPP) and dimethylallyl diphosphate (DMAPP). their synthetic counterparts but large-scale application Depending on the organism, these monomers are either pro- is limited by the high production cost and low yields1. Of duced from acetyl-CoA by using the mevalonate pathway or special interest for industrial use are microbial glycolipids from pyruvate and glyceraldehyde-3-phosphate by using the and among them, we focus on glycoglycerolipids (GGL) for deoxyxylulose phosphate (DXP) pathway. The baker’s yeast their drug delivery applications. Our aim is to turn E. coli into Saccharomyces cerevisiae exclusively utilizes the mevalonate a platform host for GGL industrial production. E. coli does pathway. In this work, we describe the use of synthetic biol- not produce GGL but when the glycolipid synthase MG517 ogy, proteomics, in vitro biochemistry and metabolomics to from Mycoplasma genitalium is expressed in E. coli, this engineer a functional heterologous DXP pathway in Saccha- enzyme is functional2,3. It is able to produce glycoglycero- romyces cerevisiae and demonstrate the biosynthesis of IPP lipids from UDPGlc and diacylglycerol, precursors which and DMAPP through the DXP pathway in yeast. are E. coli common metabolites. The metabolic engineering design was based on the introduction of the glycolipid 130. Expression Optimization of Multi-Enzyme Pathways for Xylose Utilization and Chemical Production. synthase MG517 and the overproduction of the sugar John E. Dueber*1, Michael Lee1 and Luke Latimer2 nucleotide and the phosphatidic acid, which is precursor 1Bioengineering, U.C. Berkeley, Berkeley, CA of diacylglycerol4. For the modulation of the glycolytic flux, 2Chemical Biology, U.C. Berkeley the overexpression of the GalU enzyme that forms UDP-Glc from glucose-1-phosphate was chosen. For diacylglycerol Engineered metabolic pathways often suffer from flux overproduction, the biosynthetic pathway that converts imbalances that can overburden the cell and accumulate glycerol-3-phosphate into phosphatidic acid was evaluated. intermediate metabolites, resulting in reduced flux and/or Among the different enzymes involved, the PlsC acyltrans- accumulation of side products. One method for alleviating ferase that incorporates the second acyl chain to form these imbalances is to adjust the expression levels of each phosphatidic acid was selected to avoid an increase of the pathway enzyme in a combinatorial manner. We have devel- anionic phospholipid biosynthesis and guarantee the avail- oped a method for rapidly assembling combinatorial expres- ability of phosphatidic acid for DAG formation. sion libraries and applied this towards the optimization of both xylose utilization as well as production of compounds Seven engineered strains were obtained combining mg517, in the highly branched violacein biosynthetic pathway in S. galU and plsC genes. While the UDPGlc pool was enough cerevisiae. Xylose comprises approximately 30% of the total and not limiting, the increased flux to phosphatidic acid was biomass and accordingly has attracted a great deal of at- the keypoint to enhance DAG availability and GGL synthe- tention for optimization. We explore the benefits of expres- sis. The strain overexpressing the mg517 and plsC genes sion co-optimization of the entire pathway under different produced the highest GGL yield, with a 39% increase of relevant conditions using growth as a selection to enrich for GGL compared to the strain only expressing MG5174. Sur- optimal genotypes. The branched violacein pathway was prisingly, the MG517 glycosyltransferase activity is signifi- used as model for pathways where a selection is not avail- cantly decreased compared to other strains and the strain able. We tested the efficacy of employing a linear regression coexpressing the three genes is not the highest producer. modeling approach on a small training set of samples to The presence of GGL contributes to membrane proper- predict what expression combinations in the overall library ties and perturbs phospholipid biosynthesis decreasing the is optimal such that a manageable number of strains can phosphatidylethanolamine amounts and giving new mem- be synthesized and tested. This approach promises to be brane properties to these E. coli cells. Therefore this engi- valuable for biofuel biosynthetic pathways that lack a high- neered strain has proven to be effective on GGL production throughput screen or selection. and a suitable microorganism platform to generate a variety 71 Poster Abstracts

of glycolipids with different regio- and stereoselectivity using References their corresponding glycolipid synthases. Stable inhibition of mmu-miR-466h-5p improves apoptosis resistance and protein production in CHO cells. Druz A, Son YJ, References Betenbaugh M, Shiloach J.Metab Eng. 2013 Mar;16:87-94 1 Banat, I.M., Franzetti, A., Gandolfi, I., Bestetti, G., Martinotti, Large-scale screening identifies a novel microRNA, miR-15a-3p, M.G., Frachhia, L., Smyth, T.J., Marchant, R., Appl. Microbiol. which induces apoptosis in human cancer cell lines. Druz A, Chen Biotechnol. 2010, 87, 427-444. YC, Guha R, Betenbaugh M, Martin SE, Shiloach J. 2 Andrés, E., Martínez, N., Planas, A., J. Biol. Chem. 2011, 13, 286 RNA Biol. 2013 Feb;10(2):287-300. (41), 35367-79. 3 Andrés, E., Biarnés, X., Faijes, M., Planas, A., Biocat. Biotrans- 133. Implications of the Assumptions on Intracellu- form. 2011, 30, 274-287. lar Metabolic Operational States in Metabolic Control 4 Mora-Buyé, N., Faijes, M., Planas, A., Metab. Eng. 2012, 14, Analysis 551-559. Geogios Fengos*1,2, Anirikh Chakrabarti1,2, Keng Cher Soh1,2, Ljubisa Miskovic1,2 and Vassily Hatzimanikatis1,2,3 132. Micrornas and Apoptosis in Cell Culture - Applica- 1Laboratory of Computational Systems Biotechnology, tion for Enhanced Biological Production and Cancer EPFL, Lausanne, Switzerland Treatment 2Swiss Institute of Bioinformatics, Lausanne, Switzerland Joseph Shiloach*, Aliaksandr Druz, Yu-Chi Chen Michael 3Institute of Chemical Sciences and Engineering, Ecole Betenbaugh, Rajarshi Guha, and Scott Martin Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Biotechnology Lab, NIDDK /NIH, Bethesda, MD Switzerland

Joseph Shiloach MicroRNAs are global regulators of gene Modeling and analysis of metabolic networks is inherently expression that are involved in multiple cellular processes biased by assumptions regarding the intracellular metabolite such as apoptosis, cell development, differentiation, metab- and flux states. Despite the advanced metabolomics and olism, and proliferation. Simultaneous alteration of multiple fluxomics methods at our disposal, the sheer complexity pathways to modify cells performance may be possible by of metabolic networks prevents us from determining their changing the miRNA expression profiles. Two examples will exact operational state. To address this shortcoming, we be given; the first will describe the enhancement of recom- have developed a novel framework called Flux Directionality binant protein production from CHO cells by the alteration Profile Analysis (FDPA) for characterizing systematically all of the expression profile of microRNAs involved in apoptosis possible intracellular flux and thermodynamic states us- regulation. In this case the inhibition of miR-466h expression ing available experimental information, such as metabolite caused a delay in apoptosis initiation, increased cell viabil- concentration measurements and flux split ratios. In the ity decreased Caspase-3/7 levels and contribute to higher current study, we demonstrated the utility of this frame- recombinant protein expression. In the second example work in elucidating the different physiologically relevant, yet large-scale screening of the complete microRNA mimics operationally different, functioning states of wild-type E. coli library demonstrated that the human microRNA, hsa-miR- under optimal aerobic growth condition. Using the available 15a-3p, has a pro-apoptotic role in several human cancer information, we explored the set of possible states to iden- cells. This novel member of the pro-apoptotic miRNA tify the most important and consistent internal operational cluster, miR-15a/16, was found to cause viability loss in B/ states for interpreting the observed physiology. To that end, CMBA cells during preliminary screening. Microarrays and we used the ORACLE framework to study the implications bioinformatics analysis showed that hsa-miR-15a-3p acti- of the choice of different operational paradigms by analyz- vates Caspase-3/7 and reduce cell viability by inhibiting the ing their effects on control coefficients. Further analysis of expression of bcl2l1 (bcl-xL) in difference cell lines. It may the control coefficients allows us to classify the enzymes in therefore be considered for apoptosis modulating therapies terms of their impact across all, or only specific operational in cancers associated with high Bcl-xL expression (cervical, configurations. Using this approach, we can knowledgeably pancreatic, breast, lung, and colorectal carcinomas). identify the right choice of enzyme targets, which shall help us in meeting metabolic engineering specifications.

72 Poster Abstracts

134. Novel Biosensors for Optimizing Yeast Cell Microbial communities play an important role in biofuel Factories production, biomedical research, food production, and Florian David*, Verena Siewers and Jens Nielsen waste water treatment. The capabilities of multi-microor- Department of Chemical and Biological Engineering, ganism systems are often enhanced by synergistic inter- Chalmers University of Technology, Gothenburg, Sweden actions at different levels. Co-culture systems particularly The development of efficient cell factories is fundamen- have unique advantages over mono-culture systems in tal for establishing a biosustainable economy. Inverse optimizing product yield as a result of synergistic interac- metabolic engineering is used to uncover new targets by tion, communication, division of labor, and separate and high-throughput screening of versatile genetic libraries. It diverse metabolic pathways of its components. Among relies on the selection of a high-producing phenotype from the many omics tools used to gain insight into the physiol- a diverse, previously generated, population of cells. While ogy of microbial systems, fluxomics is the most direct and techniques for generating these diverse cell populations relevant method to study the actual in vivo metabolic state. are well established, inverse metabolic engineering suf- 13 13 fers from the lack of sufficiently sensitive, selective high- C-Metabolic flux analysis ( C-MFA) is the most widely throughput measurement technologies to screen efficiently used model-based experimental approach to study flux 13 for high producers. In order to generate a multi-dimension- distributions within metabolic pathways. In the past, C- al sensor system, we are engineering new transcription MFA for co-culture systems has required physical separa- factor based sensors for key metabolites and combine tion of proteins and/or cells to resolve individual popula- them with fluorescent protein reporters. This novel technol- tions in a co-culture. In this work, we demonstrate a novel 13 ogy is allowing a direct dynamic feedback of the produc- co-culture C-MFA framework that does not require any tion of the target compound. Both upstream and down- physical separation of cells or proteins. Rather, fluxes for stream optimizations of the particular metabolic pathways individual populations are computationally deconvoluted 13 become possible. from the overall co-culture C-labeling data. We show that the overall 13C-labeling data has abundant information not We are aiming at optimizing the production of fatty acid only to estimate the fluxes in the two populations, but also derived compounds in Saccharomyces cerevisiae. Intra- to determine the fraction of each cell population in the 13 cellular biosensors sensing limiting key intermediates like co-culture. We also demonstrate how optimal C-tracers Malonyl-CoA and Acyl-CoA are used for multi-dimensional should be selected to maximize the resolution of fluxes in high producer screening. The used transcription factor co-culture systems. Significantly, we show that commonly 13 13 based sensors originate from bacterial systems. They were used tracers such as [1- C]glucose and [1- C]glucose/ 13 characterized in terms of sensitivity and quantitative re- [U- C]glucose are poor tracer choices for co-culture sys- 13 sponses to extracellular and intracellular fatty acids. Yeast tems. Instead, the less commonly used [1,2- C]glucose cells equipped with the sensor system were transformed tracer provides optimal resolution of fluxes in co-cultures. with a cDNA library thereby creating a phenotypic diversity We demonstrate our methodology experimentally us- of cells. After sorting and enriching the cells by Fluorescent ing a co-culture system of two E. coli knockout strains, Activated Cell Sorting (FACS) cell populations were exten- ∆zwf (knockout of the first step in the pentose phosphate sively characterized by PCR and followed up high through- pathway) and ∆pgi (knockout of the first step in glycolysis put sequencing. The information gained will be applied to pathway). We also present results from currently on-going redesign naïve strains for verification of target genes and work on yeast/E. coli co-culture and thermophilic co-cul- further knowledge based rational strain improvement. ture systems.

135. 13C Metabolic Flux Analysis of Co-Culture Systems: The new flux analysis methodology that we have devel- A Novel Approach oped for analyzing co-culture systems adds a new dimen- Nikodimos A. Gebreselassie* and Maciek R. Antoniewicz sion to the field of13 C-MFA and provides an enormous Department of Chemical and Biomolecular Engineering, resource to the metabolic engineering and biotechnology University of Delaware, Newark, DE communities.

73 Poster Abstracts

136. Targeted Omics Informed Engineering to Improve by a low exposure of molecular chaperones to the newly C5 Alcohol Production in E. coli synthesized recombinant proteins due to the crowded and Kevin W. George* viscous environment of E. coli cytoplasm. To further improve Fuels Synthesis, Joint Bioenergy Institute (JBEI), the chaperone-mediated solubilization of recombinant pro- Emeryville, CA teins in E. coli, we designed a novel RNA scaffold system C5 alcohols are attractive targets for microbial production in which the 3’-UTR of the recombinant protein-encoding due to their importance as “platform” chemicals and poten- mRNA serves as a scaffold for localizing the molecular tial use as advanced biofuels. Here we leverage targeted chaperone in proximity with the newly synthesized proteins. omics data and standardized parts to construct a modified In our system, a molecular chaperone DnaJ was fused with mevalonate pathway in E. coli for the improved produc- an RNA binding domain that specifically binds a unique tion of three C5 alcohols: 3-methyl-3-butenol, 3-methyl- RNA sequence in an engineered RNA hairpin loop structure 2-butenol, and 3-methyl-1-butanol. By correlating pathway on the 3’-UTR of the recombinant protein-encoding mRNA. proteomics with diagnostic metabolites, we first assessed This design can prevent the formation of inclusion bodies by pathway behavior and revealed the primary determinants promoting the rapid interaction between molecular chap- of efficient precursor production and carbon utilization. erones and newly synthesized recombinant proteins. As Informed by this methodology, we increased 3-methyl- expected, our RNA scaffold system successfully increased 3-butenol titer to 1.5 g/L, identified toxic metabolites, and the solubility of selected aggregation-prone proteins over- prioritized future engineering targets. Further improvement produced in E. coli (UDP-6-glucose-dehydrogenase, anti of 3-methyl-3-butenol titer to ~2 g/L (70% theoretical) was p21-Ras ScFv, and anti p21-Ras ScFv fused with a cell achieved through selective RBS engineering of NudB—a penetrating peptide), which was better than the solubil- promiscuous phosphatase—and targeted gene replace- ity increment obtained by their simple co-expression with ments. Expression of a fusion protein and reductase result- DnaJ molecular chaperone. Our RNA scaffold system would ed in the production of all three C5 alcohols including over provide a valuable tool for the production of recombinant 250 mg/L of 3-methyl-1-butanol, a >20-fold improvement proteins in soluble and active forms in E. coli with minimal over a previous engineering effort. This targeted approach to post-production process. engineering should prove useful for the analysis and optimi- 138. 2-Butanol and Butanone Production in Saccharo- zation of increasingly complex metabolic pathways. myces cerevisiae through the B12 Dependent Dehydra- tase Pathway Using a Tev-Based Expression System 137. A Novel Design of a Translation Coupling-RNA Payam Ghiaci*1, Joakim Norbeck2 and Christer Larsson2 Scaffold System to Improve the Efficiency of Molecular 1Chemical and Biological Engineering, Systems and Chaperone on Recombinant Proteins Solubilization Synthetic Biology, Chalmers University of Technology, Almando Geraldi*, Le Minh Bui, Ziaur Rahman, Kyo Hun Gothenburg, Sweden Kang, Jun Hyoung Lee and Sun Chang Kim 2Chemical and Biological Engineering, Chalmers University Biological Sciences, Korea Advanced Institute of Science of Technology, Gothenburg, Sweden and Technology, Daejeon, South Korea Butanone and 2-butanol are chemicals with rather decent Inclusion bodies formation upon the overexpression of potential to be used as either biofuels or biochemicals. In recombinant proteins in Escherichia coli is among the major this study, an application of a TEV-protease based expres- obstacles in therapeutic proteins production and metabolic sion system is delineated with ultimate goal of attaining engineering. Numerous approaches have been attempted equimolar expression of the individual subunits of a B12- for minimizing the formation of misfolded and aggregated dependent dehydratase used for constructing a 2-butanol recombinant proteins, such as the use of solubility-enhanc- pathway in Saccharomyces cerevisiae. The dehydratase ing tags, the modification of physicochemical conditions, is taken from Lactobacillus reuterii and it is supposed to and the overexpression of molecular chaperones. Among convert meso-2,3-butanediol to butanone. Subsequently, an those approaches, the overexpression of molecular chap- NADH dependent secondary alcohol dehydrogenase, from erones is considered as the most economical and time-effi- Gordonia sp, was introduced to catalyze the conversion of cient approach for producing soluble recombinant proteins butanone to 2-butanol. Exploiting a Δgpd1,2 strain, we could in E. coli. However, the efficacy of this strategy is still limited

74 Poster Abstracts

provide the needed NADH for the alcohol dehydrogenase tomized algorithm for quickly generating unique molecular and therefore enforce the flux towards 2-butanol (by culti- graphs and detecting symmetries for all metabolites in the vating the cells under anaerobic conditions). A final concen- database. This is used to create atom transition information tration of 2 mg/l of butanone and 4±0.2 mg/l 2-butanol was between reactants and products for all reactions contained found. in MetRxn. This information is leveraged for the construction of genome-scale size mapping models to support metabolic 139. Using Metrxn for Flux Elucidation and Model flux elucidation using C13 labeled substrates through meta- Reconstruction bolic flux analysis (MFA). Akhil Kumar*1, Saratram Gopalakrishnan*2 and Costas D. Maranas2 1The Huck Institutes of the Life Sciences, Penn State, State Models used for MFA typically include only central metabo- college, PA lism and simplified reactions for amino acids synthesis. We 2Department of Chemical Engineering, The Pennsylvania decided to explore how including additional reactions may State University, University Park, PA affect the flux inference process using MFA. To this end we first appended onto a popular core metabolic model amino MetRxn (http://www.metrxn.che.psu.edu/) is a searchable acid synthesis and degradation reactions from the iAF1260 electronic resource for published metabolic models and model and we de-lumped a handful of central metabo- databases from a wide variety of organisms. The MetRxn lism reactions. This modest modification led to significant project aims to organize and disseminate standardized changes in the confidence level for glycolysis and TCA metabolite and reaction information to improve metabolic cycle reactions and non-zero flux through arginine degrada- modeling by accurately describing reaction stoichiometry, tion. Computational challenges and results obtained upon directionality, atom mapping from reactants to products, expansion to a core genome-scale model for E. coli will be and gene to protein to reaction relations. Standardiza- described. tion algorithms automatically curate information to remove Supported by funding from the U.S. Department of Energy incompatibilities in content representation, fix stoichiometric to Dr. Costas D. Maranas grant DE-FG02-05ER25684. errors such as elemental or charge imbalances and resolve incomplete atomistic details. The standardization proce- 140. Use of Transporter Plug-Ins for Enhanced Produc- dure follows a workflow that starts by matching metabolite tivity and Reduced Byproduct Formation of Bioalkanes entries using lexicographic and phonetic techniques, and and Related Compounds structure comparison using atomistic details. The reac- Chris Grant*, Phattaraporn Morris and Frank Baganz tions are first charge and mass balanced and subsequently Advanced Centre for Biochemical Engineering, University atom/bond mapping resolution algorithms are applied. For College London, London, United Kingdom each reaction, metabolite stoichiometry, atom transition and Synthetic biology hopes to predictably design from scratch metabolite compartment information is stored. The reac- or reprogram cells into cellular factories for the production of tion and metabolite information is downloadable in SBML important chemicals, pharmaceuticals and fuels from cheap 3.0 and in a tabular format. The current MetRxn update renewable feedstocks. A critical, yet understudied, element includes recently published metabolic data for a total of 112 of this is the control of compound transport across the cell metabolic models and 8 metabolic databases. The number membrane both into and out of the cell as a means of avoid- of distinct reactions that have been mapped is greater than ing issues such as substrate access limitations, substrate 20,000 and MetRxn contains tools that allow users to down- and product inhibition and aiding product recovery1. load atom mapping data for each reaction.

We report here a successful strategy for the contextual As part of our ongoing effort we have enhanced the MetRxn evaluation of this topic using a library of transport modify- knowledgebase with additional information such as reac- ing plug-ins combined with multifactorial characterisation. tion atom transition information and reaction standard free Auxiliary plasmids designed for analogue expression of energies. In accordance with our data integration goal, we membrane proteins (pUMP) were used to express a library have integrated the ncbi taxonomy database, uniprot gene of transport proteins as plug-ins alongside two biosynthesis id’s and ncbi gene id’s within MetRxn. We developed a cus- plasmids pGEC41, which oxidises hydrocarbons into prima-

75 Poster Abstracts

ry alkanols and pADAR7942 which synthesizes bioalkanes is hydrolysed by endoxylanases resulting in the formation

from metabolic free fatty acid precursors. We demonstrate of XOS, primarily xylobiose (X2) but also xylotriose (X3) and

here benefits of the transporter plug-in approach for: xylotetraose (X4), of which small amounts enter the cell and (i) Facilitated delivery of bulky and hydrophobic substrates are then cleaved by beta-xylosidases to release monomeric to improve whole-cell biocatalysis rates by up to 70 fold for xylose. Compared with xylose, butanol-producing bacteria hydrocarbon substrates growing on xylan are energy limited with slow growth rates resulting in low fermentation productivity. Engineering such (ii) Industrially relevant product yields of over 40g/Lorganic microbes to use this material more efficiently would increase phase (8g/Ltotal) achieved using this strategy their growth rates as oligo-xylans are the primary carbon (iii) Reducing byproduct formation in whole-cell bioconver- source available when growing on hemicellulose and hence sion of alkanes to alkanols these reactions have high flux. Central to our design there- (iv) Improving bioalkane synthesis yields from glycerol by >5 fore are xylooligomer transporters. By using transporters for X -X , which take up one of these molecules for one unit of fold 2 4 energy, bacteria are much more efficient than those which (v) Reducing alkanol and aldehyde intermediate formation in need 2-4 units of energy for uptake of equivalent amounts biosynthesis of bioalkanes by >10 fold of sugar present in the monomeric form. Once inside the (vi) The integration with in-situ product removal strategies to cell, there are no energetic costs for the breakdown of X2-X4 improve bioalkane yields by 10 fold compared to the start- to X1. At the bioprocess level, the use of XOS would allow ing process. for milder pretreatment procedures in contrast to chemical This library and plug-in approach is of broad appeal for methods typically required for the complete hydrolysis of xy- biological production of hydrophobic compounds and could lose which can release inhibitors and retard fermentation. be a key enabling technology for biological routes for pro- 142. Using Protein Scaffolds to Redirect Photosynthetic ducing a wider range of hydrophobic compounds such as Reducing Power for Biosynthesis of Natural Products biofuels, fine and specialty chemicals and pharmaceutical Maria Henriques de Jesus*, Agnieszka Zygadlo Nielsen, intermediates. Birger Lindberg Møller and Poul Erik Jensen Copenhagen Plant Science Centre, Department of Plant and 1. Chen, R. R. Permeability issues in whole-cell bioprocesses and Environmental Sciences, University of Copenhagen, DK- cellular membrane engineering. Appl. Microbiol. Biotechnol. 74, 1871 Frederiksberg C, Copenhagen, Denmark 730–738 (2007). Photosynthesis in plants provides ATP and NADPH as well as 141. Engineering Efficient Xylose Metabolism Using Synthetic Biology carbon sources for primary metabolism. Cytochrome P450 Rosanna Hennessy*1, Henrique I. Neves1, Preben Krabben2, monooxygenases (P450s) in the plant endoplasmic reticulum Elizabeth Jenkinson2 and Gavin H. Thomas1 (ER) are essential in the synthesis of many bioactive natural 1Dept. of Biology, University of York, York, United Kingdom products, powered by single electron transfers from NADPH. 2 Green Biologics Ltd, Abingdon, United Kingdom We have recently demonstrated that it is possible to break the evolutionary compartmentalization of energy generation Volatile crude oil prices coupled with global warming con- and P450-catalysed biosynthesis, by relocating an entire cerns emphasise the urgent need for alternative technolo- P450 dependent pathway to the chloroplast and driving the gies that can replace fossil fuels and reduce environmental pathway by direct use of the reducing power generated by impact. Plant biomass is the most abundant renewable photosystem I in a light-dependent manner [1]. This demon- feedstock in the world with a low production cost, mak- strates the potential of transferring pathways for structurally ing it ideal for the biological production of liquid fuels (e.g. complex high-value natural products and directly tapping into biobutanol) and bulk chemicals. However, recalcitrance the reducing power generated by photosynthesis to drive the of lignocellulosic feedstocks to degradation is a press- P450s using water as the primary electron donor. ing challenge. We are using a synthetic biology approach to engineer industrially important bacterial strains for the Current work is directed towards exploring different strate- efficient breakdown and uptake of hemicellulose-derived gies to optimize channeling of product formation. One ap- xylooligomers (XOS). During enzymatic degradation, xylan 76 Poster Abstracts

proach to ensure co-localization of the enzymes in the thyla- It is has been shown that volatile fatty acid (VFA) production koid membrane is the fusion of the enzymes to components increases during sudden increases in the organic loading of the Twin-arginine translocation pathway – TatB and TatC. rate in anaerobic reactors6,7. In particular, mixed cultures These are membrane anchored and have inherent self-orga- under these conditions are known to produce elevated nizing properties which will allow us to recruit the enzymes levels of lactate and propionate, each valuable chemicals into close proximity and thereby reduce metabolic load. for bioplastics production and in the food industry. These An alternative approach is scaffolding which aims at more “shock loads” have previously been maintained for short efficient channeling of the substrate and intermediates. For periods of time, but have not yet been sustained for produc- this we utilize the protein-protein binding properties of PDZ tion at steady state. It is notable, however, that lactate and domains for building a modular synthetic scaffold. These propionate production remained elevated throughout the are around 80-90 amino acids and bind to the C-terminus duration of the shock loads. Cells use production of lactate of their target proteins (~10 aa) with different specificity. By and propionate to sink excess electrons and maintain a making fusion proteins of the required enzymes with the metabolic redox balance. By contrast, pathways for acetate ligand peptides, we can spatially recruit them in a desirable and butyrate production are used for ATP generation to manner. We expect that both strategies should reduce for- support cell growth and maintenance. Extended elevated mation of intermediaries, limit cross-talk between signaling production of lactate and propionate is therefore counterin- pathways, improve substrate channeling and consequently tuitive, especially under conditions where more substrate is increase product yields. available for biomass generation. The ultimate goal of this References experiment is to determine the product formation effects of long-term elevated organic loading rates, and whether [1] Zygadlo-Nielsen, A. et al. Redirecting Photosynthetic Reducing Power toward Bioactive Natural Product Synthesis. ACS Synth. lactate and propionate production remain elevated beyond Biol., 2013, 2 (6), pp 308–315, DOI: 10.1021/sb300128r a particular loading rate threshold. Product formation trends Funding from UNIK Center for Synthetic Biology, Interdisciplinary will then be correlated to enzyme expression to assess Research Center “bioSYNergy”, and the VILLUM Research Center metabolic function and signalling dynamics. “Plant Plasticity”, is gratefully acknowledged. A 1.3L chemostat, inoculated with biomass from an anaero- 143. Assessing Metabolic Response to Increased Substrate Loading Rate in Mixed-Culture Fermentation bic digester in Brisbane, Australia, has been operated at pH of Waste Water 5.50, 30.0°C, and a dilution rate of 1.0 d-1 for over 90 days. Robert D Hoelzle*1, Bernardino Virdis1,2 and Damien J. The substrate is glucose in basal anaerobic (BA) media8, 1 Batstone acting as a synthetic waste water feed. Initially, the glucose 1 Advanced Water Management Centre, University of concentration in the feed was 5g/L, and has been stepped Queensland, Brisbane, Australia up to 10g/L and then 20g/L, each after a period of steady 2Centre for Microbial Electrosynthesis, University of Queensland, Brisbane, Australia state was established. The glucose feed concentration will also be increased to 50g/L and then 80g/L. These corre- Mixed-culture fermentation (MCF) is a key central process to spond to loading rates of 5, 10, 20, 50, and 80g/L*d. enable next generation biofuels and biocommodity production due to economic and process advantages over appli- Temperature, pH, and gas flow production rate were record- cation of pure cultures. However, a crucial limitation to the ed continuously using LabVIEW. Liquid samples were taken application of MCF is predicting culture product response regularly for chemical, microbiome, and metatranscriptome to environmental stimuli. Certain effects, such as product analysis. VFA concentration was measured via HPLC. All formation response to pH1,2 and microbiome response to cell samples were stored in RNAlater® solution (Life Tech- substrate type3, have been well characterized. Other stimuli, nologies, NY, USA) at -20˚C prior to analysis. The microbi- such as substrate loading rate, have not been characterized ome was analysed via fluorescent in situ hybridisation (FISH) despite some intriguing reports in the literature. Additionally, supplemented with 16s rRNA pyrotag sequencing. DNA mixed culture metabolic response is not well understood was extracted for sequencing with the PowerSoil® DNA under any of these conditions, limiting the accuracy of MCF Isolation Kit (MO BIO, CA, USA). RNA was extracted with models which currently assume static metabolic arrays4,5. the RiboPure™ RNA Purification Kit, and then mRNA was

77 Poster Abstracts

purified from total RNA for metatranscriptome sequencing of Klebsiella. Species of Clostridia have also been shown with the MICROBExpress™ Bacterial mRNA Enrichment Kit to be dominant in similar operating conditions2, however (both by Life Technologies, NY, USA). 0.0 0.5 1.0 1.5 2.0 0 2 our probes for Clostridia have so far been inconclusive. 4 6 8 10 12 14 16 Concentration (g/L) Time (days) Butyrate Propionicimonas and Psuedomonas were also observed Acetate Lactate Propionate a) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 18 with FISH analysis, each making up 1% or less of the total 24 30 36 42 48 54 Concentration (g/L) Time (days) Butyrate microbiome. Klebsiella and Psuedomonas are both known Acetate Lactate Propionate b) 0.0 4.0 8.0 12.0 16.0 57 61 65 for 2,3-butanediol production following exponential growth 69 73 77 81 85 89 Concentration (g/L) Time (days) Butyrate phase9. Clostridia is generally associated with butyrate Acetate Lactate Propionate c) The initial phase of operation, production, but certain species are also associated with at a loading rate of 5g/L*d, butyrate (averaging 0.19g/L), butanol and acetone production10. Further chemical analy- acetate (0.33g/L), lactate (0.66g/L), and propionate (0.21g/L) sis will reveal whether these products are present, and 16s are all primary products. Butyrate and acetate production rRNA pyrotag results will guide further FISH analysis. These are relatively stable, whereas propionate and lactate pro- results will then be used to construct a metagenome for duction are more variable. metatranscriptome analysis. It is anticipated that fermentation enzyme expression levels will be associated with During the second phase, with a loading rate of 10g/L*d, activity of redox repressor enzymes11. This set of enzymes is butyrate (0.61g/L) and acetate (0.63g/L) become the primary known to regulate transcription of enzymes associated with products after reaching steady state, whereas lactate (0.16g/L) electron sinking in response to changes in NAD redox state. and propionate (0.16g/L) become secondary products. Inter- estingly, lactate and propionate each spike initially in response While further experimentation is needed to fully character- to the increased loading rate before giving way to butyrate and ize the effect of substrate loading rate on the MCF product acetate production. This is consistent with the literature reports spectrum, the results to this point are consistent with continu- of lactate and propionate spiking during shock loads6. It is not ous MCF systems run at similar conditions. Additionally, the clear at this point why lactate and propionate production again spikes in lactate and propionate production in the transition increase toward the end of this phase. stages between steady states are consistent with known shock load effects. These results, along with shock load stud- During the third phase, with a loading rate of 20g/L*d, ies in the literature, suggest the possibility of elevated and lactate and propionate again spike in response to the load- sustained lactate and propionate production at high loading ing rate increase, though the lactate response is notably rates. It is anticipated that metatranscriptomic analysis will more pronounced than in the second phase. Following this reveal correlation of product spectrum to activity of redox spike, the culture takes considerably longer to reach steady repressor enzymes. Full characterization of the effects of sub- state production in the third phase compared to the second strate loading rate will add valuable knowledge to the under- phase (roughly 10 days compared to 5 days). However, standing of mixed-culture fermentation dynamics.

upon reaching steady state butyrate (6.03g/L) becomes the References primary product, acetate (1.14g/L) the secondary product, 1. Temudo, M. F., Kleerebezem, R. & van Loosdrecht, M. Influence and lactate (0.13g/L) and propionate (0.37g/L) the tertiary of the pH on (open) mixed culture fermentation of glucose: a che- products. It is anticipated that the lactate and propionate mostat study. Biotechnol Bioeng 98, 69–79 (2007). spikes will become more pronounced and longer lasting 2. Lu, Y., Slater, F. R., Mohd-Zaki, Z., Pratt, S. & Batstone, D. J. Im- with continuing research into substrate loading rates of 50 pact of operating history on mixed culture fermentation microbial and 80g/L*d. ecology and product mixture. Water Sci. Technol. 64, 760 (2011). 3. Zhang, W., Werner, J. J., Agler, M. T. & Angenent, L. T. Substrate FISH has so far revealed preliminary information about the type drives variation in reactor microbiomes of anaerobic digest- microbiome. Roughly 25% of the bacteria present are of ers. Bioresour. Technol. 151, 397–401 (2014). the class Gammaproteobacteria, and of these greater than 4. Rodriguez, J., Kleerebezem, R., Lema, J. M. & van Loosdrecht, 90% are of the family Enterobacteriaceae. Bacteria of this M. C. Modeling product formation in anaerobic mixed culture fer- family have previously been shown to be a dominant spe- mentations. Biotechnol Bioeng 93, 592–606 (2006). cies under these operating conditions2, particularly species 5. Kleerebezem, R., Rodriguez, J., Temudo, M. F. & van Loos-

78 Poster Abstracts

drecht, M. C. M. Modeling mixed culture fermentations; the role of to isobutanol, 2,3-butanediol is also a promising alcohol as different electron carriers. Water Sci Technol 57, 493–497 (2008). a commodity chemical due to its extensive industrial appli- 6. Eng, S. C., Fernandes, X. A. & Paskins, A. R. Biochemical ef- cations. These metabolic pathways for producing isobutanol fects of administering shock loads of sucrose to a laboratory-scale and 2,3-butanediol are crucially discriminable in terms of anaerobic (UASB) effluent treatment plant. Wat Res 20, 789–794 cofactor balance such as NADH and NADPH. In this study, (1986). we constructed the genetically engineered yeast S. cere- 7. Voolapalli, R. K. & Stuckey, D. C. Hydrogen production in an- visiae to produce these higher alcohols, and evaluated the aerobic reactors during shock loads-influence of formate produc- difference in the alcohol production levels of the constructed tion and H2 kinetics. Water Res. 35, 1831–41 (2001). strains. This work was supported by Special Coordination 8. Angelidaki, I. & Sanders, W. Assessment of the anaerobic biode- Funds for Promoting Science and Technology, Creation of gradability of macropollutants. Re/Views Environ. Sci. Bio/Technol- ogy 3, 117–129 (2004). Innovation Centers for Advanced Interdisciplinary Research Areas (Innovative Bioproduction Kobe; iBioK), MEXT, Japan, 9. Garg, S. K. & Jain, A. Fermentative production of 2,3-butanediol: and in part by Industrial Technology Research Grant Pro- A review. Bioresour. Technol. 51, 103–109 (1995). gram in 2011 from New Energy and Industrial Technology 10. Jones, D. T. & Woods, D. R. Acetone-Butanol Fermentation Development Organization (NEDO) of Japan. Revisited. Microbiol. Rev. 50, 484–524 (1986).

11. Ravcheev, D.A., Li, X.Q., Latif, H., Zengler, K., Leyn, S.A., 145. Directed Evolution of Terpene Synthases Using Korostelev, Y.D., Kazakov, A.E., Novichkov, P.S., Osterman, A.L., High-Throughput Colorimetric Screening Based on & Rodionov, D. Transcriptional Regulation of Central Carbon and Substrate Consumption Energy Metabolism in Bacteria by Redox-Responsive Repressor Miki Iwasaki*1, Akira Fujii1, Maiko Furubayashi1, Shifei Wang1 Rex. J Bacteriol. 194,1145–1157 (2012). and Daisuke Umeno1,2 1Department of Applied Chemistry and Biotechnology, Chiba 144. Genetic Engineering to Produce Higher Alcohols in University, Chiba, Japan Yeast Saccharomyces cerevisiae 2Precursory Research for Embryonic Science and Technolo- Jun Ishii*1, Fumio Matsuda2, Akihiko Kondo3 and Kengo Ida4 gy (PRESTO), Japan Science and Technology Agency (JST), 1Organization of Advanced Science and Technology, Kobe Kawaguchi, Japan University, Kobe, Japan 2 Information Science and Technology, Osaka University, Terpenoids are one of the largest groups of natural prod- Osaka, Japan ucts with huge structural diversity and molecular functions 3Graduate School of Engineering, Department of Chemical Science and Engineering, Kobe University, Kobe, Japan including pharmaceuticals, fragrance, synthetic rubber and 4Department of Chemical Science and Engineering, Gradu- diesel/jet fuels. Integrated efforts by synthetic biologists ate School of Engineering, Kobe University, Kobe, Japan have dramatically improved the production levels of valuable terpenoids in heterogonous hosts such as microbes The production of higher alcohols including C3-C5 alco- or plants. This improvement has primarily been achieved hols in engineered bacteria has received significant atten- by metabolic engineering efforts, reactor designs, and host tion, because they can be used as fuels, fuel additives, or breeding, but not by engineering the catalytic capacity of commodity chemicals. The budding yeast, Saccharomyces terpene synthases (TPSs). This is despite that TPSs are gen- cerevisiae, has considerable potential as a producer of erally slow enzymes (kcat values typically range from 1 min-1 higher alcohols because of its capacity to naturally fabricate to 1 sec-1). Given the maturation of metabolic engineering fusel alcohols, in addition to its robustness and tolerance of isoprenoid pathways, the major limiting factor in the total to low pH. However, because its natural productivity is not production levels of terpene compounds has been shifting significant, we considered a strategy of genetic engineering to the TPSs. to increase production of higher alcohols. For example, the branched-chain higher alcohol isobutanol, which is involved TPSs catalyze complex and multi-step carbon-carbon in valine biosynthesis, offers higher octane values than their bond formation reactions including carbocation formation, straight-chain counterparts with equivalent carbon numbers rearrangements, and quenching of reactive intermediates and can be used with current infrastructure in addition to into various terpene structures. The complexity of these possessing almost the same capabilities (energy density, processes renders rational engineering of terpene syn- hygroscopicity and octane number) as gasoline. In addition thases nearly impossible. In addition, forward engineering of 79 Poster Abstracts

TPSs have been severely limited due to the lack of reliable The anticipation for substituting conventional fossil fuels screening/selection systems for TPSs: both substrates and with cellulosic biofuels is growing in the face of increasing products of TPSs are colorless, diverse in structure, and demand for energy and rising concerns of greenhouse gas volatile. A systematic and high-throughput screening of TPS emissions. However, commercial production of cellulosic activities should greatly accelerate the advancement of the biofuel has been hampered by inefficient fermentation of forward engineering of TPS. xylose and the toxicity of acetic acid, which constitute substantial portions of cellulosic biomass. Here we use a redox In this study, we developed a high-throughput colorimetric balancing strategy to enable efficient xylose fermentation assay for TPS based on substrate consumption. Here, the and simultaneous in situ detoxification of cellulosic feed- expression of active TPSs results in decreased building stocks. By combining a nicotinamide adenine dinucleotide blocks for carotenoid biosynthesis, thereby reducing the (NADH)-consuming acetate consumption pathway and an pigmentation of the host cells. We show that this strategy NADH-producing xylose utilization pathway, engineered enables selection for various types of TPSs and/or isoprenyl yeast converts cellulosic sugars and toxic levels of acetate transferases, irrespective of their product types. together into ethanol under anaerobic conditions. The results demonstrate a breakthrough in making efficient use Using this method, we attempted to the forward engineering of carbon compounds in cellulosic biomass and present an of TPSs. Various TPSs for monoterpnene, sesquiterpene, and innovative strategy for metabolic engineering whereby an diterpene, were subjected to error-prone PCR to generate undesirable redox state can be exploited to drive desirable large pool of TPS variants. The resultant TPS libraries were metabolic reactions, even improving productivity and yield. then transformed into E. coli cells harboring either of C30- or C40-carotenoid pathway enzymes. In order to obtain TPS 147. Lysate of Engineered Escherichia coli Supports mutants with improved cellular activity, the colonies with Conversion of Glucose to 2,3-Butanediol with Near-Theoretical Yields and Ultrahigh Productivity least pigmentation were selected. The selected mutants thus Jennifer E. Kay* and Michael C. Jewett obtained were then re-screened to eliminate false positives Chemical and Biological Engineering, Northwestern from them, and were analyzed for their performance in the University, Evanston, IL cell harboring another plasmids for elevating the precursor supply. The production level of terpene was higher for TPS Traditionally, metabolic engineering has been applied variants when compared with the cells expressing the wild for biochemical production in one of two ways; cell-free type, indicating our screening method indeed isolate the TPS systems of purified enzymes and in vivo systems using variants with elevated cellular activity. engineered organisms. In vitro, the high cost of purified enzymes and cofactors has limited industrial applications, In this talk, we present a series of screening constructs particularly when considering enzyme ensembles. In vivo, specially made for isolating active variants of monoterpene, over-production of naturally high-flux nodes or byproducts sesquiterpene, and diterpene synthases. Also presented are in a host’s metabolism is often successful and has resulted our current efforts to make specific/selective screen for the in many industrial processes. Conversely, generating high size-variant TPSs. flux through a novel pathway can be extremely difficult and development for many desirable products remains costly 146. Enhanced Biofuel Production through Coupled and slow. A key limitation is the conflict between cellular Consumption of Acetic Acid and Cellulosic Sugars By growth and adaptation objectives and engineering goals. Engineered Yeast In this work, we begin to explore a hybrid model in which Na Wei1, Guochang Zhang2, Joshua C. Quarterman2, Soo Rin Kim2, Jamie H.D. Cate3 and Yong-Su Jin*2 metabolic engineering techniques are used to tailor a 1Institute for Genomic Biology; Food Science and Human crude cell lysate for high rates, yields, and titers. Nutrition Deptartment, University of Illinois at Urbana Champaign, Urbana, IL As a model pathway, we selected conversion of glucose to 2 Food Science and Human Nutrition, University of Illinois at 2,3-butanediol (23BDO). 23BDO is a medium level com- Urbana-Champaign, Urbana, IL modity chemical with many industrial applications. First, 3Molecular and Cell Biology, University of California at Berkeley a three-enzyme pathway for production of 23BDO from

80 Poster Abstracts

pyruvate was expressed in BL21(DE3) cells. Second, lysates SFA1 gene encoding S-(hydroxymethyl)glutathione de- were prepared by high-pressure homogenization and clari- hydrogenase. Introduction of the mutated SFA1 alleles or fication of these cells. Third, lysates were combined with overexpression of any of the SFA1 alleles in a sfa1Δ strain glutamate salts, catalytic (1 mM) cofactors NAD and ATP, enabled growth in the presence of above 40 g/L 3HP. and glucose to produce 23BDO. Importantly, most soluble We further found that aldehyde dehydrogenase (ALD6), native enzymes are present and active in the lysate, allow- S-formylglutathione hydrolase (YJL068C) and glutathione ing the endogenous glycolytic enzymes convert glucose to play a role in 3HP detoxicification. Addition of glutathione pyruvate, the starting intermediate for 23BDO synthesis. We relieved growth inhibition by 3HP for several yeast species observed a maximal synthesis rate of meso-2,3-butanediol and for E. coli; but glutathione could not enable growth of of 4.2 g/L/h with a theoretical yield of 80% (0.4 g m23BDO a S. cerevisiae sfa1Δ strain. Based on our findings we pro- / g glucose). Titers reached 66 g/L m23BDO in a 30 hour pose a 3-hydroxypropionic aldehyde-mediated mechanism batch reaction. Simply by removing the cellular structure underlying 3HP toxicity as well as a glutathione-dependent and avoiding genomic regulation, productivity almost as route for detoxification of 3-hydroxypropionic aldehyde high as any native producer and titers several fold above (reuterin). The identified molecular response to 3HP and any previous studies in E. coli in vivo have been achieved. reuterin may well be a general mechanism for handling Our results highlight the ability for co-factor regeneration in resistance to organic acids and aldehydes by living cells. cell-free lysates. Further, they suggest exciting opportunities for use of lysate-based systems to (i) rapidly prototype 149. Optimality of Microbial Metabolism with metabolic pathways and (ii) carry out molecular transforma- Biosynthetic Heterologous Reactions Dong In Kim*, Hyun Uk Kim and Sang Yup Lee tions when bioconversion yields (g product/L), productivities Chemical & Biomolecular Engineering, Korea Advanced (g product/L/h), or cellular toxicity limit commercial feasibility Institute of Science and Technology (KAIST), Daejeon, of whole-cell fermentation. South Korea

148. Evolution Reveals a Glutathione-Dependent As more of microorganisms are metabolically engineered Mechanism of 3-Hydroxypropionic Acid Detoxification to produce chemicals beyond their native metabolism, it Kanchana R. Kildegaard*1, Björn M. Hallström2, Thomas H. has become important to systematically study the effects Blicher3, Nikolaus Sonnenschein2, Niels B. Jensen2, Svet- lana Sherstyk2, Scott J. Harrison2, Jerome Maury2, Markus of introducing heterologous reactions on microbial me- J. Herrgård2, Agnieszka S. Juncker2, Jochen Förster2, Jens tabolism. However, in contrast to gene knockout mutants Nielsen2,4 and Irina Borodina2 and microorganisms under environmental perturbations, 1Novo Nordisk Foundation Center for Biosustainability, Tech- microbial metabolism with heterologous reactions has not nical University of Denmark, Hørsholm, Denmark been studied sufficiently. To this end, we use genome- 2Science for Life Laboratory, KTH Royal Institution of Tech- scale metabolic models representing Escherichia coli nology, Solna, Sweden 3The Novo Nordisk Foundation Center for Protein Research, strains engineered to produce 1,4-butanediol, 1,3-pro- University of Copenhagen, Copenhagen, Denmark panediol, and amorphadiene with previously reported 4Department of Chemical and Biological Engineering, 13C-based flux data in order to investigate effects of foreign Chalmers University of Technology, Gothenburg, Sweden reactions on microbial metabolism. Overall, the simulated microbial metabolism with heterologous pathways shows Biologically produced 3-hydroxypropionic acid (3HP) is a either very optimal or non-optimal statuses with respect to potential source for sustainable acrylates and can also find the theoretically optimal metabolic space in a strain-spe- direct use as monomer in the production of biodegrad- cific manner. This observation contrasts with wild-type and able polymers. For industrial-scale production, high titer, single gene knockout mutants whose metabolism tends to rate and yield are essential; thus there is a need for robust be slightly sub-optimal, but rather stable. Variables caus- cell factories tolerant to high concentration of 3HP, pref- ing more complex effects of introducing heterologous erably at low pH. Through adaptive laboratory evolution reactions than single gene knockouts or environmental we selected S. cerevisiae strains with improved tolerance perturbations will be discussed. [This work was supported to 3HP at pH 3.5. Genome sequencing of three indepen- by the Technology Development Program to Solve Climate dent clones identified single-nucleotide changes in the Changes on Systems Metabolic Engineering for Biorefiner-

81 Poster Abstracts

ies from the Ministry of Science, ICT and Future Planning single reaction, a metabolic pathway or an entire metabolic (MSIP) through the National Research Foundation(NRF)of network. This is, to our knowledge, the first attempt to use Korea (NRF-2012-C1AAA001-2012M1A2A2026556).] metabolic modeling in order to provide a significance rank- ing of metabolites to guide experimental measurements. 150. Deciphering Thermodynamics in Metabolic Net- works: A Priority List of Candidates for Metabolomics 151. Malic Acid Production By Aspergillus Oryzae Alexandros Kiparissides*1,2 and Vassily Hatzimanikatis1,2 Christoph Knuf*1, Intawat Nookaew1 and Jens Nielsen2 1Laboratory of Computational Systems Biotechnology, 1Department of Chemical and Biological Engineering, EPFL, Lausanne, Switzerland Chalmers University of Technology, Göteborg, Sweden 2Swiss Institute of Bioinformatics, Lausanne, Switzerland 2The Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Hørsholm, Denmark Almost every process or action that occurs within a cell involves a metabolic reaction. Knowledge of the metabolic Aspergilli are widely used as cell factories for the produc- state of a cell and how it responds to various stimuli and tion of food ingredients, enzymes and antibiotics. The extracellular conditions can offer significant insight in the advancement of methods for performing directed genetic regulatory functions and how to manipulate them. The modifications has enabled the use of metabolic engineering increasing availability of large metabolomics datasets en- in order to improve the organisms performance. With the hances the need for computational methodologies that can recent availability of genome sequences for several industri- organize the data in a way that can lead to the inference ally relevant Aspergilli it has become possible to implement of meaningful relationships. Metabolic models comprising systems biology tools for advancing metabolic engineering. the entirety of reactions known within a pathway and/or

cell provide an increasingly popular and effective platform Malic acid belongs to the group of C4 dicarboxylic acids, to study the internal states of a cell. Constraint based ap- which are structurally similar to maleic acid and maleic proaches are commonly employed to define a, usually near anhydride, important building blocks of the petrochemical

infinite, set of equally optimal feasible internal states the industry. The C4 dicarboxylic acids are therefore expected to cell can operate in. Integration of experimental measure- replace petro chemically derived compounds in the future, ments of intracellular metabolite concentrations in genome when increased oil and gas prices favor the production of scale models restricts the thermodynamically feasible flux chemicals from renewable feedstocks. In previous stud- space and reduces uncertainty regarding the net outcome ies A. flavus wildtype strains have shown great malic acid

of by—directional reactions. However a systematic meth- secretion capacities and are among the best C4 dicarboxylic odology to prioritize and incorporate metabolomics data acid producers. The cultivation conditions for high malic within constraint based metabolic models in order to refine acid production were optimized and nitrogen limitation was the large solution space is currently lacking. identified as main factor for reaching high malic acid yields. Nevertheless, A. flavus is known to produce the mycotoxin By combining Thermodynamics—based Flux Balance aflatoxin and is consequently disqualified for large scale Analysis (TFBA), Marcov Chain sampling, Experimental industrial production. Lately it has been shown that the wild- Design and Global Sensitivity Analysis we present an type A. oryzae strain NRRL3488 exhibits similar production efficient algorithm to quantify the effect of intracellular capabilities and was therefore chosen for futher optimization metabolites on the thermodynamic flexibility of cellular of its metabolism towards malic acid production. metabolism. Metabolites are ranked based on their ability to constrain the range of possible solutions to a limited, In this study, we present the physiological characterization thermodynamically consistent set of internal states. The of an engineered A. oryzae strain in which the reductive proposed methodology effectively defines the amount of cytosolic TCA branch from pyruvate via oxaloacetate to experimental information required to reduce uncertainty in malate and a malate transporter have been overexpressed. defining the state of cellular metabolism (i.e the flux distri- The use of glucose and xylose as carbon sources in this bution and displacement from thermodynamic equilibrium) study bridge over to the use of lignocellulosic biomass in by providing a ranked list of targets for metabolomics. The future applications of this organism in a truly sustainable proposed approach is modular and can be applied to a biorefinery.

82 Poster Abstracts

152. Feasibility Studies of New Strategy for Ultra-High- made this microorganism a popular metabolic engineering Throughput Screening (uHTS) of Novel Enzyme By in platform for the production of biotechnologically relevant vitro Compartmentalization (IVC) Using Microbeads compounds1. In this yeast, cytosolic acetyl-CoA synthe- from Metagenomic Resources sis and growth strictly depend on expression of either the Kyong-Cheol Ko*, Binna Lee, Dae-Eun Cheong, Jae Jun Song and Jong Hyun Choi Acs1 or Acs2 isoenzyme of acetyl-CoA synthetase (ACS). Korea Research Institute of Bioscience and Biotechnology Since hydrolysis of ATP to AMP and pyrophosphate in the (KRIBB), South Korea ACS reaction constrains maximum yields of acetyl-CoA- derived products, this study explores replacement of ACS In general, it is very difficult to screen for novel enzymes us- by ATP-independent pathway for acetyl-CoA synthesis2. The ing traditional methods owing to limitations in resources and functional expression of different bacterial genes encoding screening methods. This study is the first to demonstrate acetylating acetaldehyde dehydrogenase (A-ALD) was stud- direct screening for novel enzymes activity using an ultra- ied in strains, in which native route of cytosolic acetyl-CoA HTS (uHTS) system by in vitro compartmentalization (IVC) formation was disrupted by deletion of all five known genes using microbeads from metagenomic resources. Recently, coding for acetaldehyde dehydrogenases (ALD). In the next a screening protocol for robotic HTS system have been en- step, Acs- Ald- S. cerevisiae strain was constructed in which abled us to perform activity verifications on more than 104 A-ALD from Escherichia coli - eutE successfully replaced clones per day and identify four distinct clones with enzyme the two step reaction of cytosolic acetyl-CoA synthesis. In activity. However, we demonstrate to process screening this strain, aerobic growth rate of 0.27 h-1 was observed, on more than 107 clones per hour by single cell level. In which was equal to 79% of the Ald+ Acs+ reference strain. the present study, we can generate very easily microbeads In glucose-limited chemostat cultures biomass yield on glu- without microfluidic system and show a rapid, simple, and cose of A-ALD-dependent strain was lower than that of the efficient method for screening enzyme including hydrolase reference strain. Subsequently, a systems biology approach using an uHTS system by using fluorescence-activated cell was used to investigate the physiological impact of these sorting (FACS) analysis. This IVC method using agarose- interventions in cytosolic acetyl-CoA metabolism. Tran- based microbeads can detect and screen various enzyme/ scriptome analysis suggested that reduced biomass yield pathway in the single cell level without diffusion of fluores- was caused by acetaldehyde. This hypothesis was further cent substrate in IVC using water-oil-water (w/o/w) emulsion supported by the increased levels of acetaldehyde in A- method. This new uHTS system is a model system for ultra- ALD-dependent strain. Transcript profiles also indicated that speedy, sensitive, multiplex screening of enzyme/pathway a previously proposed role of Acs2 in histone acetylation is from various genetic libraries and holds promise for provid- probably linked to cytosolic acetyl-CoA levels rather than to ing new enzymes/pathways for bioindustrial applications. direct involvement of Acs2 in histone acetylation [3]. Despite these interventions in acetyl-CoA metabolism, the intracel- 153. Replacement of the Saccharomyces cerevisiae lular acetyl-CoA concentration did not change significantly Acetyl-CoA Synthetases By Acetylating Acetaldehyde between strains, which suggest a strong homeostatic regu- Dehydrogenase for Cytosolic Acetyl-CoA Synthesis 4 Barbara U. Kozak*1, Harmen M. van Rossum1, Kirsten R. lation of its concentration . Nonetheless, the levels of some Benjamin2, Liang Wu3, Jean-Marc G. Daran1, Jack T. Pronk1 compounds that have cytosolic acetyl-CoA as a precursor and Antonius J.A. van Maris1 (such as lysine) were elevated, suggesting an increased 1Department of Biotechnology, Delft University of Technol- availability of this molecule. Although further modifications ogy, Delft, The Netherlands are needed to achieve optimal in vivo performance of the 2Amyris Inc., Emeryville, CA alternative reaction for supply of cytosolic acetyl-CoA as a 3DSM Biotechnology Center, Alexander Fleminglaan 1, 2613 AX Delft, The Netherlands product precursor, this study demonstrates for the first time that yeast ACS can be fully replaced.

Cytosolic acetyl-coenzyme A is a precursor for many com- References pounds whose production from sugars is already imple- 1. Hong, K. K., Nielsen, J., 2012. Metabolic engineering of Saccha- mented or under investigation by industry. The robustness romyces cerevisiae: a key cell factory platform for future biorefiner- of Saccharomyces cerevisiae combined with fast develop- ies. Cell Mol Life Sci. 69, 2671-2690. ments in yeast synthetic biology and systems biology have 2. Kozak, B. U., van Rossum, H. M., Benjamin, K. R., Wu, L., Daran, 83 Poster Abstracts

J. M., Pronk, J. T., van Maris, A. J., 2014. Replacement of the Sac- presence of a functional PG wall in this bacterium (common- charomyces cerevisiae acetyl-CoA synthetases by alternative path- ly referred to as the “Chlamydial anomaly”)[2]. More recently, ways for cytosolic acetyl-CoA synthesis. Metab Eng. 21, 46-59. we have improved upon the current probes and introduced 3. Takahashi, H., McCaffery, J. M., Irizarry, R. A., Boeke, J. D., 2006. a third class of labeling method, opening the way to three Nucleocytosolic acetyl-coenzyme a synthetase is required for his- distinct fluorogenic PG labeling strategies. tone acetylation and global transcription. Mol Cell. 23, 207-217.

4. Cai, L., Sutter, B. M., Li, B., Tu, B. P., 2011. Acetyl-CoA induces In the near future, we believe that these novel approaches cell growth and proliferation by promoting the acetylation of his- for PG labeling will find a common use as a diagnostic tool tones at growth genes. Mol Cell. 42, 426-437. for infections, will expand the knowledge base required for 154. Decorating Bacterial Surfaces By Designer new antibiotic discovery, and will represent the core de- Molecules Advances the Fundamental Knowledge signs for new antibacterials themselves. More generally, we about Bacterial Growth believe that our success in hijacking PG biosynthesis with a 1 2 3 Erkin Kuru* , Edward Hall , George Liechti , Srinivas library of distinct molecules suggests that PG pathway may Tekkam2, Anthony Maurelli3, Yves Brun4 and Michael be an inherently promiscuous scaffold for metabolic engi- VanNieuwenhze2 1Molecular and Cellular Biochemistry, Indiana University, neering of various polypeptide compounds. Furthermore, Bloomington, IN our work on fluorogenic D-amino acids, i.e. molecules that 2Chemistry, Indiana University ‘turn on’ once incorporated into the PG, may eventually 3Uniformed Services University of the Health Sciences lay the groundwork for the design of genetically encodable 4 Biology, Indiana University fluorogenic L-amino acids as the preferred protein tags as opposed to much larger and fainter fluorescent proteins. In addition to the usual protein synthesis and unlike other domains of life, bacteria can and must make a second type References of proteinaceous structure, namely their peptidoglycan (PG) 1. Kuru, E., et al., In Situ probing of newly synthesized peptidogly- cell walls. PG significantly differs from natural proteins by can in live bacteria with fluorescent D-amino acids. Angewandte the presence of D-centered amino acids, it defines bacterial Chemie, 2012. 51(50): p. 12519-23. shape, growth and division and it is therefore an essential 2. Liechti, G.W., et al., A new metabolic cell-wall labelling method surface element. These properties in turn make PG also reveals peptidoglycan in Chlamydia trachomatis. Nature, 2013. the best target for antibiotic development. Due to the lack of tools for spatiotemporal tracking of PG biosynthesis, 155. Metabolic Modulation in Response to Chemical- Induced Signal Transduction in Chlamydomonas we have developed a set of simple, yet versatile, ways to Reinhardtii tag PG growth of virtually all bacterial species. Exploiting Jung-eun Lee*1, Do Yup Lee1, Jeong-Jin Park2 and the surprisingly promiscuous nature of the PG metabolic Oliver Fiehn3 enzymes and following a modular approach, we have 1Kookmin University designed and synthesized unnatural substrate analogs, 2Washing State University 3 such as so-called fluorescent D-amino acids (FDAAs)[1], Genome Center, UC Davis that can hijack PG biosynthesis at different and mostly es- Chlamydomonas reinhardtii is a eukaryote algae model sential points of the pathway. Because they are non-toxic system for photosynthetic organisms. Chlamydomonas is D-amino acid derivatives, these molecules specifically and suitable for studying complicate signaling pathways that are efficiently label bacterial growth, opening the way to specifi- evolutionary or clinically important (e.g. target of rapamy- cally engineer surfaces of live bacteria with any functionality cin, jasmonate, and salicylic acid signaling pathways). In conceivable even in complex environments. These tools current report, we elicitated signal transduction pathways have already had a profound impact on the field of bacterial of Chlamydomonas by various chemicals, and monitored cell biology and helped us to address crucial questions in physiology-relevant metabolic modulation using metabolo- the field. Case in point, we recently settled a 50 years long mic approaches. Comprehensive quantitative metabolomic controversy and revealed presence of PG in sexually trans- tools have been used for reconstructing the dynamic net- mitted pathogen Chlamydia, which are sensitive to anti-PG work in living cells in time-dependent and dose-responsive drugs despite the failure of all previous attempts to show the metabolic changes by applying analytical chemistry and

84 Poster Abstracts

informatics tools in order to provide a multi-dimensional a genetically-encoded biosensor based on the transcriptional gateway to cellular responses that is otherwise inaccessible regulator Lrp (leucine-responsive protein) of Corynebacterium to genomics. glutamicum1. The sensor allows the intracellular detection of methionine and branched-chain amino acids at the single cell 156. Metabolically Engineered Escherichia coli for level by converting this information into a fluorescent signal. Isoprene Biosynthesis The suitability of the biosensor to mirror different intracellular Chun-Li Liu*1 and Tianwei Tan2 concentrations of effector amino acids was applied to study 1College of Life Science and Technology, Beijing University of Chemical Technology, Beijing, China population dynamics of the valine producer strain C. glutami- 2Beijing University of Chemical Technology, Beijing, China cum ΔaceE lacking the pyruvate dehydrogenase complex.

As an important feedstock in petrochemistry, isoprene is Flow cytometry based monitoring of biosensor cells during used in a wide range of industrial applications. It is pro- lab-scale fed-batch cultivation revealed the appearance of duced almost entirely from petrochemical source, however subpopulations varying in productivity and viability. In addi- depleting progressively. A reliable biological process for iso- tion, live cell imaging studies using microfluidic chip devices prene production utilizing renewable feedstocks will be an displayed different types of non-producing cells within industry-redefining development. There are two biosynthetic isogenic microcolonies of C. glutamicum ΔaceE2. The ap- pathways producing isoprene: the mevalonate (MVA) path- pearance of non-productive subpopulations might strongly way and the methyl erythritol 1-phosphate (MEP) pathway. impact the performance and stability of the production In this study, the MEP pathway was modified in Escherichia process; therefore, genetically-encoded biosensors have the coli BL21 (DE3) to produce isoprene. The gene isoprene great potential to reveal bottlenecks for improving fermenta- synthase (IspS) chemically synthesized after codon optimization processes. tion from Populus alba was heterologously expressed. The endogenous genes of 1-deoxy-D-xylulose-5-phosphate In further studies, the Lrp-biosensor was applied to improve synthase (DXS) and 1-deoxy-D-xylulose-5-phosphate reduc- growth and productivity of C. glutamicum ΔaceE by in vivo toisomerase (DXR) were over-expressed. The gene isopen- evolution. Using fluorescent-activated cell sorting (FACS), sta- tenyl pyrophosphate isomerase (Idi) from streptococcus tionary phase cells with the highest fluorescent output were pneumoniae was exogenously over-expressed, and farnesyl iteratively isolated and (re-)cultivated. Isolated strains revealed diphosphate synthase (ispA) was weakened to enhance the a significantly increased growth rate, shortened lag-phase yield in E. coli BL21 (DE3). It was shown that the control strain and an increased final optical density. The L-valine production harboring empty plasmids did not emit any isoprene at all. of some strains was increased up to 100% compared to the The overexpression of single DXR gene had little impact on parental strain while the formation of by-products (L-alanine) the yield of isoprene. Idi from streptococcus pneumoniae was reduced. These results emphasize biosensor-based played a significant role in the improvement of isoprene strain evolution as a straightforward approach to improve production. The highest yield was achieved by ispA weak- growth and productivity of microbial production strains. ened DXS-IDI-IspS recombinant with 19.9 mg /L, i.e., 33-fold References enhancement of isoprene yield of the IspS recombinant. (1) Mustafi et al., 2012 Met Eng 14 (4), 449-457. (2) Mustafi et al., 2014 PLoS One 9 (1): e85731. 157. Application of a Genetically-Encoded Metabolite Sensor for Single Cell Analysis and Development of Production Strains 158. On the Effects of Phenotype Prediction Methods Regina Mahr*, Alexander Gruenberger, Nurije Mustafi, over Strain Design Algorithms. a Multi-Objective Dietrich Kohlheyer and Julia Frunzke Approach Institute of Bio- and Geosciences, IBG-1: Biotechnology, Paulo Maia1,2, Miguel Rocha1 and Isabel Rocha1 Forschungszentrum Juelich, Juelich, Germany 1Centre of Biological Engineering, Department of Biological Engineering, University of Minho, Braga, Portugal. The analysis of microbial metabolite production is typically 2SilicoLife — Computational Biology Solutions for the Life performed using bulk techniques, which obscure informa- Sciences, Rua do Canastreiro, 15, 4715-387 Braga, tion with respect to single cell behavior. We have developed Portugal.

85 Poster Abstracts

The past two decades have witnessed great advances in use of enzyme-associated flux5) or LMOMA (a linear imple- the computational modeling and systems biology fields. mentation of MOMA6) (box colors) are simulated with the Soon after the first models of metabolism were developed, other (x-axis). Besides the great variation in fitness for the several methods for the prediction of phenotypes were also different phenotype simulation methods, we veri_ed that in put forward. With the ever-growing information provided by some cases less than 10% of the solutions generated by such methods, new questions arose. Metabolic Engineering pFBA are valid in LMOMA (BPCY _ 0:0001). in particular posed some interesting questions. Recently, Schuetz and co-workers proposed that the metabolism of Assumptions regarding the cellular objectives of an organ- bacteria operates close to the Pareto-optimal surface of a ism when subjected to distinct conditions (environmental, three-dimensional space definedned by competing genetic, etc.) are still the object of active discussion. This objectives and demonstrated the validity of their claims for fact motivated us to develop a method capable of sug- various environmental perturbations1. gesting designs compliant with more than one phenotype prediction method. Solutions generated by our method are However, phenotype prediction methods have all been simulated using pFBA and LMOMA and plotted by BPCY developed to operate based on the assumption of a given for both phenotype simulation methods. The ad-hoc clus- single-objective, as an example Flux Balance Analysis ters reveal a group of interesting solutions (cluster 2). An (FBA) often assumes that the organisms are evolutionarily analysis on the flux distribution of the solutions presented in optimized towards optimal growth. On the other hand, Mini- these clusters is also provided and a rational for robust mization of Metabolic Adjustment (MOMA) proposes that solution design is derived.

after a perturbation, the goal of the organisms shifts from References optimal growth to the minimization of the global metabolic [1] R. Schuetz, N. Zamboni, M. Zampieri, M. Heinemann, and U. adjustment relative to the wild-type. Albeit multi-objective Sauer, \Multi-dimensional optimality of microbial metabolism,” Sci- approaches focused on the bio-engineering objectives ence, vol. 336, no. 6081, pp. 601{604, 2012. have been proposed2, 3, none tackles the multi-objective [2] Y.-G. Oh, D.-Y. Lee, S. Y. Lee, and S. Park, \Multiobjective ux nature of the cellular objectives. balancing using the nise method for metabolic network analysis,” Biotechnology progress, vol. 25, no. 4, pp. 999{1008, 2009. In this work we analyze the inuence of several phenotype [3] J. Costanza, G. Carapezza, C. Angione, P. Li_o, and G. Nicosia, prediction methods on the strainsdesigned by metaheuristic \Robust design of microbial strains,” Bioinformatics, vol. 28, no. algorithms and suggest a multi-objective approach capable 23, pp. 3097{3104, 2012. of finding designs compliant with the cellular objectives as- [4] A. M. Feist, C. S. Henry, J. L. Reed, M. Krummenacker, A. R. sumed by the various phenotype prediction methods. Joyce, P. D. Karp, L. J. Broadbelt, V. Hatzimanikatis, and B. _. Palsson, \A genome-scale metabolic reconstruction for Escherichia Using a recent model of Escherichia coli K124, we observed coli k-12 mg1655 that accounts for 1260 orfs and thermodynamic information,” Molecular systems biology, vol. 3, no. 1, 2007. the effect of different phenotype prediction methods in the convergence of metaheuristic algorithms performing strain [5] N. E. Lewis, K. K. Hixson, T. M. Conrad, J. A. Lerman, P. Charu- optimization, evolving growth-coupled production mutants santi, A. D. Polpitiya, J. N. Adkins, G. Schramm, S. O. Purvine, D. Lopez-Ferrer, et al.,\Omic data from evolved E. coli are consistent in aerobic and anaerobic conditions. A critical analysis of with computed optimal growth from genome-scale models,” Mo- the different mutant ux distributions was performed, and lecular systems biology, vol. 6, no. 1, 2010. we concluded that, for a selected phenotype prediction [6] S. A. Becker, A. M. Feist, M. L. Mo, G. Hannum, B. _. Palsson, method, the strain designs proposed by the optimization and M. J. Herrgard, \Quantitative prediction of cellular metabolism algorithms were generally not robust when another method with constraint-based models: the cobra toolbox,” Nature proto- was used to predict their phenotypes. cols, vol. 2, no. 3, pp. 727{738, 2007.

There is variation in the Biomass-product coupled yield (BPCY) of aerobically succinate producing mutants with glucose as carbon source, when solutions generated with either pFBA (a variation of FBA that minimizes the overall

86 Poster Abstracts

159. Exploring Bacterial Microcompartments to knowledge of reaction rate equations and their associated eEstablish Orthogonal Metabolism parameters. The model is generally written as: Joerg Mampel* dX/dt = Sv(x,p) (1) BRAIN AG, Zwingenberg, Germany

Where X is the vector of species concentrations, S is the Post genomic system analyses of apparently simple micro- stoichiometric matrix, v(X,p) is the vector of rate equations, bial life reveal an ever increasing complexity that must be and p is the vector of kinetic parameters. Here, the estimation managed in order to design reliable and robust cell factories. of unknown kinetic parameters from time-series concentration Managing this complexity using systems biology approaches data frequently becomes the bottlenecking step, motivat- is an attractive option but reduction of complexity (ROC) is the ing the development of a large number of methods1. Exist- more promising solution. ROC could be achieved by a true ing methods can generally be divided into two approaches: de novo design of microbial cells built only from the essential integral and differential approach, based on whether or not components, or by sequential genome reduction of naturally the ODE model is integrated during the estimation. In the evolved platform organisms. integral approach, the parameter estimation is formulated as a constrained optimization to minimize model prediction error. Here, we introduce another ROC-concept, i.e. orthogonalisa- Meanwhile, the differential approach involves a pre-processing tion of cellular metabolism1. Recent developments in meta- step, in which the time-series data are smoothen and dif- bolic engineering and synthetic biology offer novel design ferentiated to give estimates of dX/dt and Eq1 is then used principles towards the introduction of spatial orthogonality in to compute dynamic reaction flux estimates v. Afterwards, living cells. Towards this goal, scaffolding of enzyme cascades the parameters are estimated by minimizing flux prediction is especially appealing due to the ease and flexibility of this errors. The main advantage of the differential over the integral design principle. Bacterial microcompartments (MCPs) are approach is computational speed, as the ODEs need not be subcellular proteinogenic organelles found in a broad variety of integrated and the parameter estimation can be done one flux bacteria 2. MCPs enclose proteins and enzymes of biosynthet- at a time. However, the differential approach is known to give ic or biodegradative pathways and serve to improve multistep biased parameter estimates, whose the parameter accuracy is biosynthesis or to insulate (toxic) reaction cascades within sensitive to the data-smoothing step. cells. We demonstrate successful rerprogramming of bacterial microcompartments (MCPs) to enclose a b-galactosidase, In metabolic networks, the number of metabolites often ex- an esterase and an oxidoreductase with broad substrate ceeds that of reactions, and thus the matrix S does not have specificity. Demonstration of in vitro substrate conversion a full column rank. Consequently, the differential approach catalyzed by reprogrammed and purified MCPs challenge the cannot be directly applied to metabolic networks, as the dy- view of MCPs as substrate-specific, tight metabolic insulators. namic flux estimation becomes underdetermined. Despite this Therefore, spatial orthogonality by scaffolding of enzymes into restriction, it may be possible to estimate the dynamic fluxes microreactors appears to be the natural function of MCPs. when the data are sufficiently dense or by taking minimum Above all, our findings open the door to a broad range of in norm solution2, 3. In a new class of methods, we have used the vitro and in vivo MCP-applications in (nano)biotechnology. relationship between dX/dt and v in Eq.1 to reduce the dimen- References sionality of the parameter estimation4. Specifically, we first (1) Mampel, J et al., 2013. Trends Biotechnology (31), 52-60 divided the fluxes into independent and dependent subsets, (2) Shively, JM et. al., 2009. Encyclopedia of Microbiology, 404-424 vi and vd, such that the dependent fluxes could be uniquely determined from the independent fluxes according to Eq. (1). 160. Redemption: Reduced Dimensional Ensemble Given the values of the (independent) parameters of vi(X,p) Modeling and Parameter Estimation and data of X, the flux vi could be calculated according to the Yang Liu, Erica Manesso* and Rudiyanto Gunawan rate equations vi(X,p). Subsequently, the flux vd were obtained Institute for Chemical and Bioengineering, ETH Zurich, by solving Eq.1 using the estimates of dX/dt from data and the Zurich, Switzerland vi values above. The (dependent) parameters of vd(X,p) could The creation of kinetic ordinary differential equation (ODE) then be estimated by least square optimization of the differ- models of biological networks is often hampered by imprecise ences between the flux vd and the flux function vd(X,p), per-

87 Poster Abstracts

formed one flux at a time. Therefore, the parameter estimation method above. Here, instead of finding the global minimum was done only over the independent parameters of vi(X,p). solution of the error function, we used a combination of We have shown using case studies that such a strategy could an out-of-equilibrium Adaptive Monte Carlo method and a provide more than two orders of magnitude improvement in multiple ellipsoids-based sampling method9 to identify a set the computational speed over the traditional parameter esti- of parameters whose error functions were within a speci- mation using the same error function4. fied threshold. The threshold error value could be set using statistical score (e.g. F-distribution) or estimated by boot- However, the strategy above still depends on time-series data strapping the original data using Monte Carlo approach. preprocessing and thus suffers from the same issues affect- Our ensemble method has the advantage that the ensemble ing any differential estimation method. Bhatt and coworkers5-7 could be compactly defined over only the space of the inde- recently proposed differential estimation methods that avoided pendent parameters. time-series data smoothing and differentiation, by estimating the extents of reactions, not the reaction fluxes. Nevertheless, The methods mentioned above are available as a MATLAB the estimation of reaction extents requires the S matrix to have based toolbox, called REDEMPTION (REduced Dimension- a full column rank. In this work, we have developed a new al Ensemble Modeling and Parameter estimaTION). Here, reduced dimensional parameter estimation that can generate we have used eSS (Enhanced Scatter Search) algorithm unbiased parameter estimates. Motivated by the extents of re- from SSmGO (Scatter Search for Matlab Global Optimiza- actions, the new method involved the calculation of integrated tion) Toolbox for the global optimization10, 11, and CVODE fluxes V, based on rewriting Eq.1 in the integral form: from SUNDIALS (SUite of Nonlinear and DIfferential/ALge- X(t) – X(0) = SV(X,p)2 braic equation Solvers) for solving ODEs12. For the explora- tion of parameters for ensemble modeling, we have used The method was implemented following the procedure the HYPERSPACE toolbox9. The beta version of the toolbox described above. In other words, we divided the integrated is available upon request.

fluxes V into independent and dependent components, and References performed the parameter search only for those that ap- 1. Chou, I.C. and E.O. Voit, Recent developments in parameter peared in the independent subset. We have demonstrated estimation and structure identification of biochemical and genomic and compared the performance of the integrated flux method systems. Mathematical Biosciences, 2009. 219(2): p. 57-83. using two case studies: an ODE model of branched metabolic 2. Chou, I.C. and E.O. Voit, Estimation of dynamic flux profiles from network and a lin-log model of L. lactis metabolic pathways. metabolic time series data. BMC Systems Biology, 2012. 6. The case studies showed that the integrated flux strategy 3. Voit, E.O., Characterizability of metabolic pathway systems from could provide unbiased parameter estimates and better ac- time series data. Mathematical Biosciences, 2013. curacy than our previous method, at the cost of slightly higher 4. Jia, G.J., G. Stephanopoulos, and R. Gunawan, Incremental computational speed. For lin-log models, the integrated flux parameter estimation of kinetic metabolic network models. BMC however performed faster than the previous method, by taking Systems Biology, 2012. 6. advantage of the structure of the model equations where the 5. Amrhein, M., et al., Extents of Reaction and Flow for Homo- integrations of log(X) were only computed once. geneous Reaction Systems with Inlet and Outlet Streams. Aiche Journal, 2010. 56(11): p. 2873-2886. The difficulty in the parameter estimation above has been 6. Bhatt, N., M. Amrhein, and D. Bonvin, Incremental Identification attributed to the parameter identifiability, or the lack thereof. of Reaction and Mass-Transfer Kinetics Using the Concept of Ex- In other words, there can exist multiple parameter combina- tents. Industrial & Engineering Chemistry Research, 2011. 50(23): tions that are indistinguishable according to the data. This p. 12960-12974. issue has motivated the application of ensemble modeling 7. Bhatt, N., et al., Incremental identification of reaction systems-A through the creation of a family of models. Within the umbrella comparison between rate-based and extent-based approaches. of ensemble modeling, we have created a method for the Chemical Engineering Science, 2012. 83: p. 24-38. generation of an ensemble of parameters that can provide sta- 8. Jia, G., G. Stephanopoulos, and R. Gunawan, Ensemble Kinetic tistically equivalent fit to the experimental data8. The method Modeling of Metabolic Networks from Dynamic Metabolic Profiles. Metabolites, 2012. 2(4): p. 891-912. was based on the reduced dimensional parameter estimation

88 Poster Abstracts

9. Zamora-Sillero, E., et al., Efficient characterization of high- malate dehydrogenase, and malic enzyme were constructed dimensional parameter spaces for systems biology. BMC Systems in BY4741lpd1Δ strain. In a batch fermentation test from 100 Biology, 2011. 5. g/L glucose using the engineered strains, the isobutanol titer 10. Egea, J.A., et al., Scatter search for chemical and bio-process reached 1.62 ± 0.11 g/L at 24 h after the start of fermentation, optimization. Journal of Global Optimization, 2007. 37(3): p. 481-503. which corresponds to the yield at 0.016 ± 0.001 g/g glucose 11. Rodriguez-Fernandez, M., J.A. Egea, and J.R. Banga, Novel consumed. This work was supported by industrial technology metaheuristic for parameter estimation in nonlinear dynamic bio- research grant program in 2011 from NEDO of Japan. logical systems. BMC Bioinformatics, 2006. 7. 12. Hindmarsh, A.C., et al., SUNDIALS: Suite of nonlinear and 162. The Impact of Orthogonal Gene Expression on Het- differential/algebraic equation solvers. Acm Transactions on erologous Pathway Productivity. Mathematical Software, 2005. 31(3): p. 363-396. George H. McArthur IV* and Stephen S. Fong Department of Chemical and Life Science Engineering, 161. Metabolic Engineering of Yeast Central Metabolism Virginia Commonwealth University, Richmond, VA for Higher Alcohol Production 1,2 3 1 Fumio Matsuda* , Jun Ishii , Syunsuke Nishino , Keisuke The goals of metabolic engineering are often at odds with the Morita1, Akihiko Kondo2,4 and Hiroshi Shimizu1 host’s own cellular objectives. For example, the heterologous 1Graduate School of Information Science and Technology, Osaka University, Suita, Japan overproduction of a secondary metabolite such as lycopene in 2RIKEN CSRS, Yokohama, Japan Escherichia coli directly competes with the growth objective of 3Organization of Advanced Science and Technology, Kobe E. coli by diverting carbon flux away from the native metabolic University, Kobe, Japan network (i.e., biomass formation) and toward lycopene ac- 4 Graduate School of Engineering, Department of Chemical cumulation through the DXP pathway. A significant amount of Science and Engineering, Kobe University, Kobe, Japan work in the metabolic engineering community has been carried out to address this problem largely through investigating Isobutanol is a target for biorefinery research as a next-gen- carbon and redox balancing between heterologous pathways eration biofuel and a building block for commodity chemical and the host’s metabolism. While this is essential for production. Metabolically engineered Saccharomyces cere- successful pathway engineering, very little attention has been visiae strains to produce isobutanol have been developed by given to balancing the energy and material flux through the introducing the Ehrlich pathway with respect to its advanta- process of heterologous gene expression, which is geous characteristics for cost-effective production. In this typically shared with the host. That is, heterologous expres- study, the central metabolism in S. cerevisiae was engineered sion traditionally relies on native RNA polymerases and ribo- to improve isobutanol production based on understanding somes. If the energetic cost or burden of expression is S. cerevisiae metabolism by flux balance analysis (FBA) and at least partially due to the process of gene expression itself, absolute quantification of intermediate. and not solely the product (e.g., pathway enzymes), then expressing a heterologous pathway using orthogonal FBA using S. cerevisiae metabolic model pointed out that a cellular machinery (i.e., RNAPs and ribosomes that act only on redox imbalance caused by an activation of isobutanol bio- the heterologous pathway genes) should be less costly to the synthesis could be relieved by introducing the Enter-Doudoroff host than expressing the same pathway enzymes via native (ED) pathway or pyridine nucleotide transhydrogenase. Meta- transcription and translation. In particular, orthogonal gene bolically engineered strains expressing Escherichia coli eda expression systems that provide a stable, dedicated supply and edd genes in BY4742gnd1Δ showed a faster growth rate of cellular machinery should alleviate burden and increase indicating a function of ED pathway in S. cerevisiae. An ab- growth rate (and therefore productivity) if the concentrations solute quantification of the metabolite levels in BY4742pfk1Δ of available native RNA polymerases and ribosomes in the cell strain showed that a tight control of glucose-6-phoshate are rate limiting. Building from recent work in synthetic biology dehydrogenase reaction hampers a redirection of glycolytic that has demonstrated functional orthogonal transcription— flux into ED pathway, since levels of 6-phosphogluconate in translation processes, our work experimentally investigates BY4742pfk1Δ (0.08 mmol/g fw) was essentially identical with the impact of orthogonally expressing three heterologous wild type strain. In order to implement a transhydrogenase enzymes for lycopene biosynthesis in E. coli on fermentation function, metabolic shunts including pyruvate carboxylase, productivity, yield and titer.

89 Poster Abstracts

163. Tools to Resolve Compartmentalized Metabolism in way were expressed in several heterologous organisms such Mammalian Cells as E. coli B or K12 or Saccharomyces cerevisiae to convert Christian M. Metallo* acetylCoA to butanol. Reported results showed that the seven Department of Bioengineering, University of California San C. acetobutylicum known genes of the pathway are function- Diego, La Jolla, CA ally expressed whatever the heterologous organism is, but Eukaryotic cells compartmentalize biochemical processes the final butanol titer and yield are rather low. Neither species in different organelles, often relying on metabolic cycles to could produce more than 1g/l of butanol. A hypothesis could shuttle reducing equivalents across intracellular membranes. be that some key enzymes involved in butyryl-CoA synthesis NADPH serves as the electron carrier for the maintenance of in C. acetobutylicum, unknown up to now, were missing in the redox homeostasis and reductive biosynthesis, with separate heterologous host. cytosolic and mitochondrial pools providing reducing power in each respective subcellular location. This cellular organization To solve this problem and further improve final titer and yield is critical for numerous functions but complicates analysis of of heterologous n-butanol production, our objectives were to metabolic pathways using available methods. To address this i) identify all the enzymes involved in butyryl-CoA synthesis need we have developed an approach to resolve NADP(H)-de- in C. acetobutylicum ii) demonstrate the neediness of these pendent pathways in subcellular compartments within intact enzymes for in vivo butanol synthesis in C. acetobutylicum mammalian cells. By tracing hydrogen in compartmentalized through knockout mutants construction and characteriza- reactions that use NADPH as a cofactor, including the production iii) express the gene(s) encoding the novel enzyme (s) tion of 2-hydroxyglutarate by mutant isocitrate dehydroge- with genes encoding the complete already known metabolic nase enzymes, we can observe metabolic pathway activity butanol pathway from the C. acetobutylicum in a E. coli in distinct cellular locations. Using this system we determine MG1655 strain. Results showed that the novel recombinant E. the direction of several metabolic cycles that are channelled coli strain was able to produce n-butanol from glucose at high between the mitochondria and cytosol, highlighting the ability yields never obtained before with any microorganisms (natu- of this approach to resolve compartment-specific redox reac- ral or non-natural n-butanol producers). Moreover, when the tions in intact cells. strain was grown in continuous culture, a stable production (for more than four months) of butanol with a low amount of 164. Identification of the Missing Enzymes for the other by products was obtained. Heterologous Production of n-Butanol at High Yield Isabelle Meynial-Salles*, Antoine Riviere, Liang Tian, Céline 166. Engineering Improved Productivity of 1,4- Foulquier and Philippe Soucaille Butanediol in E. coli – a Kinetic Modeling Approach LISBP, Toulouse University, Toulouse, France Stefano Andreozzi1,2, Anirikh Chakrabarti1,2, Keng Cher Soh1,2, Anthony Burgard3, Stephen Van Dien3, Ljubisa N-butanol, an important industrial intermediate chemical Miskovic*1,2 and Vassily Hatzimanikatis1,2,4 1 and a potential biofuel, can be produced in slow-growing Laboratory of Computational Systems Biotechnology, EPFL, Lausanne, Switzerland native hosts that belong to genus Clostridium, as a mixture 2Swiss Institute of Bioinformatics, Lausanne, Switzerland with acetone and ethanol in biphasic fermentation. Even if 3Genomatica, Inc., San Diego, CA the acetone-butanol-ethanol (ABE) fermentation was one the 4Institute of Chemical Sciences and Engineering, Ecole first large scale industrial fermentation and despite the long Polytechnique Fédérale de Lausanne (EPFL), Lausanne, research history on the fermentation process, obstacles still Switzerland remain today. Butanol titer, yield, and productivity are still low mainly due to the co-production of acetone, ethanol and Microbial cell factories are becoming a norm for commer- acetic and butyric acids. Advances in genetic manipulation cially viable production of chemicals for pharmaceutical, of clostridia have led to large increases in yield and produc- biotechnology, food and beverage industries. However, tivity. However, the regulation of solvent production is highly engineering of microbial cell factories requires a simultane- complex and cannot yet be achieved in a single-stage fermen- ous optimization of several criteria such as productivity, tation at high yields. To overcome some of the problems as- yield, titer, stress tolerance, all the while retaining the ef- sociated with clostridial n-butanol production processes, the ficient, cost-effective and robust process. One of the most genes encoding the clostridial n-butanol fermentation path- prominent examples where a rational metabolic engineering 90 Poster Abstracts strategy played a key role is in the production of 1,4-bu- sume intracellular oxygen, followed by nitrogen fixation, dur- tanediol (BDO) in E. coli. In this study, we used the ORACLE ing the night. While Cyanothece 51142 already up-regulates (Optimization and Risk Analysis of Complex Living Entities) genes associated with nitrogen fixation, iron uptake, and framework to analyze possible enhancements of the E. coli iron-sulfur cluster biogenesis during the dark period, it is strain engineered for improved production of BDO. ORACLE difficult to genetically manipulate. Synechocystis PCC 6803 framework allowed us to integrate thermodynamics, avail- is the most thoroughly studied cyanobacterium and a close able omics and kinetic data into a population of large-scale relative of Cyanothece 51142. A detailed study found evi- kinetic models. Analysis of the engineered E. coli strain led dence for a nitrogen-fixing ancestor of Synechocystis 6803 to the identification of three critical modules within the met- and the subsequent loss of nitrogen-fixing ability. This close abolic network which contained the enzymes that primarily phylogenetic relationship and ability to perform targeted control the fluxes leading to BDO production. The enzymes genetic modifications makes Synechocystis 6803 an ideal in these modules are focused around: a) central glycolysis, target for the incorporation of nitrogen-fixing capabilities. b) the lower branch of tricarboxyclic acid cycle, and c) novel BDO production route. However, the manipulation of the The nif gene cluster in Cyanothece 51142 is the largest intact enzymes in the identified modules - while possibly leading contiguous cluster of nitrogen fixation related genes com- to the increased BDO production - had complex effects on pared to other nitrogen-fixing cyanobacteria. The transfer other intracellular states like redox state, energy charge, of this approximately 28 kb cluster into Synechocystis 6803 cofactro levels, cellular growth and byproduct formation. We will introduce the genes necessary for nitrogen fixation. The used the large-scale kinetic models generated by ORACLE incorporation of nitrogen fixation into Synechocystis also to postulate successfully metabolic engineering alternatives requires the development of other genetic elements, such for optimal performance with reduced byproduct secretion as promoters that enable Synechocystis gene expression and fine-tuned redox balance, energy charge and cofactor in a similar diurnal pattern as in Cyanothece 51142. Several levels. While in the current study, the aim was to improve promoters that exhibited diurnal rhythms have been char- BDO production, the methodology presented here can read- acterized, and their output strengths have been fine tuned ily be applied to other products and organisms of interest. for use in nitrogen fixation. To facilitate the characterization and engineering of nitrogen fixation in Synechocystis, several 167. Designing a Nitrogen Fixation Circuit in an ammonia biosensors have also been constructed using the Oxygenic Photosynthetic Organism glnRA operon from Lactococcus lactis. These biosensors Andrea Balassy1, Allison G. Hoynes-O’Connor1, Cheryl M. were shown to lower targeted gene transcription given the Immethun1, Deng Liu2, Thomas J. Mueller*3, Rajib Saha3, Yi Xiao1, Tae Seok Moon1, Fuzhong Zhang1, Costas D. presence of ammonia in E. coli cells. Given the high sensitivi- Maranas3 and Himadri B Pakrasi1,2 ty of the nitrogenase enzyme to oxygen, the use of an oxygen 1Department of Energy, Environmental, and Chemical Engi- sensor to control the transcription of the nif cluster would neering, Washington University in St. Louis, St. Louis, MO mimic the regulation found within Cyanothece. 2Department of Biology, Washington University in St. Louis, St. Louis, MO The in silico modeling of metabolism offers a number of 3Department of Chemical Engineering, The Pennsylvania State University, University Park, PA insights into the metabolism of an organism and genetic modifications that would optimally perturb the organism to Cyanobacteria are oxygenic photosynthetic prokaryotes of achieve desired results. Genome-scale metabolic models considerable interest given their ability to use sunlight to have been created for both Cyanothece 51142 and Synecho- cystis 6803, and the Synechocystis model has been updated convert CO2 into a number of useful chemicals including fuels. Certain cyanobacteria are also capable of fixing nitrogen to incorporate recent refinements in the modeling of the through the use of either specialized cells called heterocysts organism. These advances are aimed towards implementing or by temporally separating photosynthesis and nitrogen and optimizing nitrogen fixation in Synechocystis 6803. fixation. Cyanothece ATCC 51142 is one such strain that is capable of fixing nitrogen by performing photosynthesis Supported by funding from the NITROGEN program of the and storing carbon in glycogen granules during the day and National Science Foundation using the stored carbon for high rates of respiration to con-

91 168. Effect of CO Transfer on Hydrogen Production thought. Whereas enterobacteria constitute a paradigmatic By Carboxydotrophic Archeon Thermococcus case of EMP-operating microorganisms, other bacteria Onnurineus NA1 favor the use of alternative glycolytic pathways. Modern Jeong Geol Na*1, Hee Seok Jeong1, Tae Wan Kim2, Sang metabolic engineering endeavors call for adequate microbial Goo Jeon1, Gwang Ho Kim1, Soo Hyun Chung1 and Sung Gyun Kang2 chassis to plug-in and -out genetic circuits and new-to- 1Biomass and Waste Energy Laboratory, Korea Institute of nature functionalities. However, bacteria that are the easi- Energy Research, Daejeon, South Korea est to manipulate are seldom the most suitable or the most 2Marine Biotechnology Research Division, Korea Institute of appropriate for specific large-scale endeavors or industrial Ocean Science and Technology, Ansan, South Korea applications. Among them, environmental Pseudomonas strains, such as P. putida, are interesting microbial platforms In recent years biological conversion of syngas or steel mill since they are pre-endowed with metabolic, physiological, gas containing carbon monoxide (CO) has been receiv- and stress-endurance traits optimal for biotechnological ing much attention since it offers an exciting opportunity needs. P. putida KT2440 operates an entirely ED-based gly- to produce fuels and chemicals from non-food and non- colysis, and the EMP pathway is non-functional due to the fossil resources. However, there are some obstacles for the absence of a 6-phosphofructokinase activity. Interestingly, commercialization of this process. One major bottleneck EMP outcompetes ED in ATP yield from glucose, making is regarding the CO transfer since it has the poor aqueous the former preferable over the later from an industrial point solubility and it is important to design a strategy to enhance of view. Activating an EMP glycolysis in P. putida is therefore gas transfer efficiency. In this study, the effect of CO transfer of paramount interest. Building on the tenets of synthetic on the performance of syngas fermentation was investi- biology, a streamlined glycolytic pathway was designed by gated. Hydrogen production by Thermococcus onnurineus taking advantage of the functional glycolytic elements of NA1, a carboxydotrophic hyperthermophilic archeon using a Escherichia coli. By following a systematic, bottom-up ap- model gas containing 60% of CO was carried out at 80 oC in proach, the relevant genes were cured from commonly used a stirred tank reactor with Rushton turbine impeller. Reactor restriction enzymes, and all non-coding DNA sequences. performance with various agitation speeds and CO flow rate The genes were then organized in two independent operons was evaluated in terms of cell growth and hydrogen produc- and placed under the transcriptional control of well charac- tivity. After short lag period depending on agitation speed terized synthetic parts. As a proof-on-concept, the system and CO flow rate, T. onnurineus NA1 actively produced was assayed in E. coli and P. putida mutant strains as a first hydrogen with growth–associated kinetics, governed by step towards the activation of a flawless linear glycolytic Luedeking-Piret model. Severe mass transfer was occurred pathway in virtually any bacterium. within several hours and hydrogen productivity remained constant. From mass transfer coefficient of CO, determined 170. Towards a Chassis Organism for Synthetic Biology using the hydrogen productivity in the mass transfer limiting Stephan Noack*1, Simon Unthan1, Meike Baumgart1, stage, we derived a correlation equation between CO trans- Marius Herbst2, Gerd Seibold3, Christian Rückert4, Volker fer and reaction conditions. Wendisch2 and Wolfgang Wiechert1 1IBG-1: Biotechnology, Forschungszentrum Jülich, Jülich, 169. Streamlining Central Catabolism for à La Carte Germany 2 Activation of Glycolysis Faculty of Biology & CeBiTec, Bielefeld University, Biele- Pablo I. Nikel* and Víctor de Lorenzo feld, Germany 3 Systems and Synthetic Biology Program, Centro Nacional Institute of Biochemistry, University of Cologne, Cologne, de Biotecnología (CNB-CSIC), Madrid, Spain Germany 4Center for Biotechnology, Bielefeld University, Bielefeld, Germany The process known as glycolysis encompasses several biochemical sequences, including the Embden-Meyerhof- Synthetic biology is expected to introduce engineering Parnas (EMP) and the Entner-Doudoroff (ED) pathways. principles to the field of life sciences. A prerequisite to en- Not all the microorganisms capable of growing on glucose able the rational assembly of biological devices is not only use the canonical EMP glycolytic pathway. In fact, glucose a library of well-characterized genetic fragments (BioBricks) metabolism in bacteria is more diverse than previously but also a robust structural basis (chassis organism). Such

92 a chassis should display minimal biological complexity to Improved L-lysine production with Corynebacterium glutamicum ensure a predictable behavior and can either be constructed and systemic insight into citrate synthase flux and activity. Biotech- in a bottom-up or top-down strategy. The latter is followed in nol Bioeng 109(8):2070-81 our joined project in which we construct a chassis organism [2] Voges R, Noack S (2012) Quantification of proteome dynamics 15 by step-wise reduction of the natural genome of Corynebac- in Corynebacterium glutamicum by N-labeling and selected reaction monitoring. J Proteomics 75(9):2660-9 terium glutamicum. This soil bacterium is a promising target organisms for a genome-reduction project, since deep in- [3] Unthan S., Grünberger A., van Ooyen J., Gätgens J., Heinrich J, Paczia N, Wiechert W., Kohlheyer D., Noack S. (2014) Beyond sights into its physiology have already been gained in various growth rate 0.6: What drives Corynebacterium glutamicum to 1, 2, 3 systems biology approaches in the last decade . higher growth rates in defined medium? Biotechnology & Bioengi- neering 111(2):359-371 As a starting point we deleted three prophages (CGP1-3) [4] Baumgart M., Unthan S., Rückert C., Sivalingam J., Grünberger which make up almost 6 % of the wild type genome of C. A., Kalinowski J., Bott M., Noack S., Frunzke J. (2013) Construc- glutamicum. The prophage-cured strain (ΔCGP123) showed tion of a Prophage-Free Variant of Corynebacterium glutamicum no impaired phenotype with respect to growth rates or ATCC 13032 for Use as a Platform Strain for Basic Research and biomass yields during cultivation under various conditions. Industrial Biotechnology, Applied and environmental microbiology. 79(19):6006-6015 Interestingly, the loss of the prophages and their restriction- modification system resulted in a significantly increased [5] Rohe P., Venkanna D., Kleine B., Freudl R., Oldiges M. (2012) plasmid stability and transformation efficiency4. An automated workflow for enhancing microbial bioprocess optimization on a novel microbioreactor platform. Microbial cell factories 11(1):144 Based on ΔCGP123 with improved genetic stability, we evaluated further genome-wide targets for subsequent dele- 171. Computing Proteome Abundance and Activity tion steps. In cooperation with project partners, we classi- with a Genome-Scale Model of Metabolism and Gene fied more than 50 % of the wild type genes as non-essential Expression and identified several gene clusters for simultaneous dele- Edward J. OBrien*, Ali Ebrahim, Jose Utrilla Carreri, Joshua A. Lerman and Bernhard O. Palsson tion. These clusters were deleted one at a time in CGP123 Δ Bioengineering, University of California, San Diego to establish a strain library for subsequent phenotypic characterization and evaluation of each genome reduction step. We have constructed a genome-scale model for Escherichia coli that seamlessly integrates metabolic and gene product To meet the demand of high-throughput cultivations we expression pathways, termed an ME-Model. Metabolism conducted the growth experiments in a novel Mini-Pilot-Plant and gene expression are interdependent processes that (MPP) by embedding a BioLector-system in a robotic environ- affect and constrain each other. We formalize these con- ment to automate complete workflows for upstream develop- straints and apply the principle of growth maximization to 5 ment . To gain a deeper process understanding, cultivation show that the ME-model accurately predicts biomass com- samples are harvested and centrifuged automatically to position, carbon source uptake rates, by-product secretion provide supernatants for subsequent quantitative analysis rates, and central carbon metabolic fluxes. Importantly, the with various fully automated assays in MTP scale. ME-model rectifies a number of failure modes of metabolic models by accounting for enzyme synthesis costs. As a result, the majority of gene cluster deletions did not impair the phenotype with respect to growth rate or biomass We have furthermore developed a method that utilizes the yield. These findings strongly support our prior classification ME-Model to integrate gene expression and physiologi- of non-essential genes and thus prove the high potential to cal data to predict condition-specific enzyme activities. significantly reduce the wild type genome of C. glutamicum. We first utilize proteomic data from E. coli growth in batch In ongoing work we are combining multiple promising dele- culture with different carbon sources to derive a consistent tions steps to finalize our construction of a less complex and genome-wide parameter set of effective enzyme activities. robust chassis organism for synthetic biology. We show that the parameter set agrees with low-throughput References kinetic assays and allows for prediction of absolute protein [1] van Ooyen J, Noack S, Bott M, Reth A, Eggeling L (2012) expression levels. We then utilize gene expression data from 93 Poster Abstracts

chemostat culture at various dilution rates to characterize that catalytic components of fungal cellulosomes are highly the changes in enzyme activity under nutrient limitation. regulated in response to simple sugars, which is supported by recently obtained proteomic data. We will further dis- The uses of the ME-Model described here will have impor- cuss the transcriptional regulation patterns observed for tant implications for metabolic engineering. We envision other important enzyme families under catabolic regulatory the ME-Model will aid in guiding genome reduction strate- conditions, and connect these regulation patterns to protein gies, quantifying proteome burden, utilizing omics data to secretion and lignocellulose degradation. iteratively improve strains, and understanding the effects of culture conditions on organismal physiology. 173. Aspergillus Terreus Isolated from the Brazillian Diversity: Genomic Variability Associated to Lovastatin 172. Deciphering Dynamic Regulation Patterns of Production Cellulose-Degrading Enzymes in Anaerobic Fungi Nádia Parachin, N.S.*1, Flávia Mullinari1, Kelly Assis1, Beatriz Kevin Solomon1, John K. Henske1, Charles Haitjema1, Diego Magalhães2, Gabriel Fernandes2 and Maria Sueli Felipe1 Borges-Rivera2, Dawn A. Thompson2, Aviv Regev2 and Mi- 1Biologia Celular, Universidade de Brasília, Brasília, Brazil chelle A. O’Malley*1 2Pós-Graduação em Ciências Genômicas e Biotecnologia, 1Chemical Engineering, University of California, Santa Universidade Católica de Brasília, Universidade Católica de Barbara, Santa Barbara, CA Brasília, Brasília, Brazil 2Broad Institute of MIT and Harvard, Cambridge, MA The World Health Organization (WHO) has determined that Anaerobic gut fungi are attractive microbes to adapt for cardiovascular diseases are the leading cause of deaths bio-based chemical production from lignocellulose. These worldwide. In 2008, about 17.3 million people died from car- fibrolytic, invasive microbes secrete an array of cellulases diovascular disease accounting for about 30% of total world and cellulolytic complexes (fungal cellulosomes) for syner- deaths. This number is expected to increase 34% by 2030 gistic hydrolysis of plant biomass. Though fungal hydrolytic (OMS 2008). According to the Brazilian Society of Cardiol- activity has been shown to be substrate dependent, the ogy, in Brazil, it has been estimated that 300-400mil people underlying regulation mechanisms that coordinate the ac- die of cardiovascular disease being the leading cause of tion of cellulases and cellulosomes from gut fungi remain mortality in the country (SBC 2012). unknown. To address this issue, we have combined next- generation sequencing and proteomic approaches to exam- One of the causes that may lead to cardiovascular disease ine cellulose-degrading enzyme production under several is hypercholesterolemia which represents high blood cho- fungal growth conditions. A new species of gut fungus from lesterol levels (> 200mg/dL). In U.S. one in every six Ameri- the Piromyces genus was isolated from the digestive tract of cans has high cholesterol level in the blood (CDC 2012) and a horse, and its proliferation was monitored via fermentation in Brazil a study including Brazillians from different cities gas production. Fungi exhibited high enzymatic reactivity had showed that about 40% of this population had choles- against a range of cellulosic and lignocellulosic substrates terol above normal levels (Martinez, Santos et al. 2003). (filter paper, Avicel, reed canary grass), which was repressed in the presence of simple sugars. Through strand-specific Statins are the most used drugs for hypercholesterolemia RNAseq and use of the TRINITY assembly platform, we treatment. These are inhibitors of HMG-CoA reductase, the were able to assemble hundreds of novel cellulase genes de first enzyme in the cholesterol biosynthesis pathway, which novo from >27,000 transcripts without the need for genomic catalyzes the reduction of HMG-CoA to mevalonate. Statin sequence information. The fungal transcriptome is particu- treatment reduces cholesterol synthesis preventing the larly rich in GH6 and GH43 enzymes, and we find that 27 of buildup of plaque inside the arteries (Barrios-Gonzalez and 54 diverse glycosyl hydrolase families are transcriptionally Miranda 2010). Among naturally occuring statins, Lovastatin repressed during growth on glucose relative to reed canary is the active principle of drugs such as Mevacor, Altocor and grass (lignocellulose). Within the majority of these tran- Altoprev. Also, is the precursor of Simvastatin, the second scripts, dockerin-tagged elements of fungal cellulosomes best-selling statin world being the active drug like Zocor and are abundant, and 15% of dockerin-containing transcripts Lipex (Barrios-Gonzalez and Miranda 2010). Simvastatin is are repressed in the presence of glucose. This suggests produced chemically from the direct alkylation of lovastatin.

94 Poster Abstracts

However, the chemical conditions are not environmentally 1School of life sciences, Ulsan National Institute of Science friendly and the final product is difficult to be purified (Barri- and Technology, Ulsan, South Korea 2 os-Gonzalez and Miranda 2010).The fermentation process Ulsan National Institute of Science and Technology, Ulsan, South Korea, 3Korea University, South Korea was then developed and patented where the simvastatin is produced by fermentation of genetically modified strains of The central carbon metabolism of Eshechrichia coli includes Aspergillus terreus (Van Den Berg, Hans et al. 2009). Thus the glycolysis, tricarboxylic acid cycle (TCA) and pentose strains that hyper produce lovastatin allow the production phosphate pathway (PPP) providing 12 essential precursor not only of this statin but its derivatives. metabolites and cellular energy (ATP, NADH and NADPH) needed for building the entire biomass of the cell. In addition Up to now a single strain of the fungus Aspergillus terreus to the central carbon metabolism, there are several bypass (ATCC20542) is used for industrial lovastatin production. and ancillary pathways such as Entner-Doudroff pathway Additionally strains producing different amounts of lovastatin (ED), gluconeogenesis, methylglyoxal shunt and the glyox- allow a comparative study in order to improve knowledge ylate shunt that help in maintaining cellular homeostasis. on the biosynthetic pathway of this compound. Therefore Robustness in the central metabolic pathway is controlled the present study aimed at isolation of A. terreus strains that by several transcription factors: crp, fnr, arcA, rpiB, cra, superproduce lovastatin. Screening results enabled iden- iclR, and pdhR etc. However, very less is known about the tification of strains with similar production as control strain transcriptional regulation of the ancillary reactions. Even ATCC20542 (~300mg/L) and isolates that produced traces when all necessary precursors are derived from the central amount of lovastatin (~6mg/L). Those strains were further carbon metabolic pathway, ancillary reactions are necessary identified as A. terreus using both molecular and protein for the sustainability of the cellular process. For example, finger printing profile. Genomic sequences from lovastatin glyoxylate shunt is indispensable for growth on acetate as cluster were obtained from strains producing different lov- it helps reduce the futile cycling of carbon through the TCA astatin levels and are correlated to statin production. Up to cycle. On the other hand, deregulated activation of ancillary date this is the first time that A. terreus strains are compared reactions will lead to energy and metabolite level imbalance. in genomic level and correlated to lovastatin production. Here, we show that such ancillary reactions are also es- References sential to shorten the lag phase for growth and to combat Barrios-Gonzalez, J. and R. U. Miranda (2010). “Biotechnological the accumulation of metabolic intermediates which other- production and applications of statins.” Applied Microbiology and wise would disturb cellular homeostasis and growth. We Biotechnology 85(4): 869-883. elucidate the transcriptional regulation of ancillary reactions CDC, C. f. D. C. a. P. (2012). “(http://www.cdc.gov/cholesterol/ through a novel transcription factor, YebK. facts.htm) “. Martinez, T. L., R. D. Santos, D. Armaganijan, K. P. Torres, A. Deletion of the transcription factor, yebK, affects the lag Loures-Vale, M. E. phase up on shifting cells from rich medium to minimal Magalhaes, J. C. Lima, E. Moriguchi, C. Amodeo and J. Ortiz medium with different carbon sources like acetate, lactate (2003). “National alert campaign about increased cholesterol: and propionate. Lactate is first converted to pyruvate and determination of cholesterol levels in 81,262 Brazilians.” Arq Bras metabolized through gluconeogenesis whereas acetate is Cardiol 80(6): 635-638, 631-634. metabolized via the TCA cycle, and glyoxylate pathway. The OMS. (2008). “http://www.who.int/mediacentre/factsheets/fs317/ strains deleted for the transcription factor yebK exhibited en/.” from http://www.who.int/mediacentre/factsheets/fs317/en/. antagonistic effect for growth on acetate and lactate. For SBC, S. B. d. C. (2012). “(http://socios.cardiol.br/noticias/coles- growth on lactate minimal medium, wild type strains had a terol.asp.” significantly long lag phase (~15 hours) whereas yebK de- Van Den Berg, M., M. Hans and H. Streekstra (2009). METHOD leted strains did not exhibit any lag phase. A long lag phase FOR THE PRODUCTION OF SIMVASTATIN. US 2009/019731. was evident in acetate minimal medium with yebK deleted 174. Regulation of Ancillary Reactions Around the strains when L-glutamine was used as a nitrogen source Central Carbon Metabolism of E. coli whereas wild-type strains did not have any lag phase. Vinuselvi Parisutham*1, Sangwoo Kim1, Young-Kyo Seo2, In-Geol Choi3 and Sung Kuk Lee1 Lag phase is a distinct growth phase that helps bacteria to 95 Poster Abstracts

prepare for the new environment. Lag phase is the poorly 175. Examining the Complex Transcriptional Response characterized phase in bacterial growth regime mainly be- of Perturbing Anthranilate Synthase in the Terpenoid cause of low number of cells in that phase. The length of the Indole Alkaloid Pathway in Catharanthus Roseus lag phase determines the efficiency to activate all the meta- Jiayi Sun and Christie A.M. Peebles* bolic enzymes needed for growth on a particular medium. Chemical and Biological Engineering, Colorado State Longer lag phase is usually indicative of failure to activate University, Fort Collins, CO key metabolic genes. However, it could also be possible that the failure to rapidly turn off pathways that would lead Catharanthus roseus (Madagascar periwinkle) produces two to flux imbalance in the new medium could lead to longer clinically important chemotherapy drugs, vinblastine and lag phase. Since there is no evidence to support the notion vincristine. These drugs are produced through the terpenoid that YebK regulates the metabolic operons for propionate, indole alkaloid (TIA) pathway. While researchers have been acetate or lactate directly we hypothesized that YebK may studying the TIA pathway genes and regulators for the past be regulating such futile ancillary metabolic reactions. 40 years, there are still many aspects of the pathway that Consistent with that the long lag phase observed in wild are unknown. Previous study on this system has shown type strains growing on lactate could be rescued with the that over expression of a feedback insensitive anthranilate deletion of α-ketoglutarate/proton symporter protein, KgtP. synthase (AS) under the control of an inducible promoter in- ChIP-seq identified the presence of YebK binding site in the crease the levels of some direct metabolites, while decrease middle of the coding region of otherwise constitutive gene, the levels of some downstream metabolites. kgtP. Thus, we hypothesize that YebK might be acting as a feed forward loop to activate the ancillary reaction catalyzed Although genetic modification is specifically targeted to by KgtP. To test if YebK could control ancillary reactions glob- a particular gene AS, we hypothesized that it may cause ally, conditional mutants were made that could grow only with unexpected transcriptional responses of other TIA pathway the activation of ancillary reactions. For example, deletion genes and regulators. In this research, RT-qPCR was ap- of the gene pfkA encoding the dedicated step of EMP path- plied to analyze the transcriptional response of all known way would demand the activation of ED or PPP pathway for TIA genes and regulators after 12, 24, and 48 hours induc- efficient growth. Deletion of yebK in strains deleted for pfkA tion of AS expression. The results show that overexpression exhibited a faster growth on glucose than strains with only of AS caused an increase in known terpenoid pathway gene pfkA deleted indicating the possibility of regulating the ED transcripts, indole pathway gene transcripts, and some TIA pathway. Similarly, deletion of sucA and sucB genes of TCA pathway gene transcripts. Notably two TIA gene transcripts cycle blocks the conversion of α-ketoglutarate to succinyl- were significantly down-regulated under the same condi- coA. Triple knockout strain (ΔsucA/ΔsucB/ΔyebK) experi- tions. Similarly transcripts of TIA regulators were differen- enced a longer lag phase in glucose minimal medium than tially expressed when AS was overexpressed. These results the double knockout strain (ΔsucA/ΔsucB). Despite the longer explain why this AS engineered root line did not result in lag phase the triple knockout strains experienced a higher a higher level TIA accumulation and further indicate the specific growth rate indicating a delayed activation of ancillary complex regulation of TIA pathway after perturbing AS gene reactions consuming alpha ketoglutarate. These results clearly expression. Therefore, further elucidation of the TIA path- indicate the importance of regulation of the ancillary reactions way and regulation is required to facilitate the future rational depending on the accumulation of the metabolic intermediates. engineering of this plant for increased TIA production.

176. Discovery of Gene Overexpression Targets for In this study, we elucidate the transcriptional regulation of Biofuel Product Tolerance in Yeast the ancillary reactions of E. coli by a novel transcription Pedro V. Peña*1 and Friedrich Srienc2 factor, YebK. YebK acts as a feed forward loop to activate 1BioTechnology Institute, University of Minnesota, or repress several ancillary reactions around TCA cycle Minneapolis/St. Paul, MN and ED pathways. Knowledge of such regulations is highly 2Department of Chemical Engineering and Materials important while re-designing the metabolic pathways as Science, and BioTechnology Insititute, University of they would serve to reduce the lag phase and increase the Minnesota, Minneapolis/St.Paul, MN specific growth rate on the desired carbon source.

96 Poster Abstracts

Production of biofuels such as ethanol induces a potent we validate these pathways and demonstrate the micro- stress on the microorganisms that produce them. Thus, bial production of MEG from various sugars, including one engineering product tolerance is essential to improve yields engineered strain that produces 40 g/L MEG from xylose at and process efficiency. We demonstrate the discovery of a yield of 0.35 g-MEG/g-xylose. gene overexpression targets to significantly improve ethanol tolerance in Saccharomyces cerevisiae. Using the Cytostat 178. Improving the Internal Flux Distributions from continuous culture technique we screened for a sustainable Genome Scale Metabolic Models of S. cerevisiae 1 2 3 ethanol tolerance mechanism in a pool of S. cerevisiae cells Rui Pereira* , Jens Nielsen and Isabel Rocha 1IBB-Institute for Biotechnology and Bioengineering/Cen- transformed with a genome-wide overexpression library. We tre of Biological Engineering, University of Minho, Braga, discovered that an overlapping sequence containing genes Portugal HAP4 and SLD2 represented 90% of selected clones from 2Department of Chemical and Biological Engineering, a starting pool of 1,588, and that co-overexpressing these Chalmers University of Technology, Gothenburg, Sweden genes is sufficient to confer an increase of nearly 50% in 3Centre of Biological Engineering, Department of Biological both the growth rate and ethanol productivity under ethanol Engineering, University of Minho, Braga, Portugal stress. In addition, overexpression of these genes prevents the drastic effects of ethanol stress on cell size and the Genome Scale Metabolic Models can be used to simulate cell cycle. Co-overexpression of HAP4 and SLD2 leads to the intracellular flux distributions under steady state condi- significantly improved kinetics of glucose consumption and tions using different simulation tools such as Flux Balance 1 ethanol productivity during fed-batch growth under ethanol Analysis (FBA) . Lately, these models proved to be useful stress. As demonstrated for the case of ethanol production for predicting gene knock-outs that optimize the production 2–4 the approach presents a general strategy for the improve- of important industrial targets . To obtain a good correla- ment of a wide range of biotechnology processes. tion between simulations and in vivo results, it is important to validate beforehand how well the model allows to predict 177. Microbial Production of Renewable the metabolic fluxes for the wild-type organism. Monoethylene Glycol Brian Pereira*1, Marjan De Mey1,2, Chin Giaw Lim1, Haoran In this work, the accuracy of the simulated intracellular flux 1 1 Zhang and Gregory N. Stephanopoulos distributions in Saccharomyces cerevisiae was evaluated. 1Chemical Engineering, Massachusetts Institute of The results revealed that steady-state simulations per- Technology, Cambridge, MA 2Inbio.be, Department of Biochemical and Microbial formed with FBA and the available genome-scale models Technology, Ghent University, Ghent, Belgium [5–8] under fully aerobic conditions contained relevant mismatches in important areas of central metabolism, when Monoethylene glycol (MEG) is an important commodity compared with in vivo data9, 10 and physiological knowl- chemical with such applications as antifreeze and as a raw edge, namely the absence of flux in the Pentose Phosphate material for poly(ethylene terephthalate) which is utilized for Pathway. Since many of these mismatches are associated plastic packaging and polyester fabric. Currently, MEG is with reactions involving the cofactors NADP+/NADPH and produced in large volumes (approximately 19 million metric NAD+/NADH, all the enzymatic reactions that included these tons in 2010), primarily from fossil fuels. As a sustainable cofactors were manually curated. alternative, we propose a single-step bioprocess in which plant-derived carbohydrates are converted by engineered Because under fully aerobic conditions the ratios of NADPH/ microorganisms into renewable MEG. Toward this goal, we NADP+ and NAD+/NADH are high, it was assumed that the have engineered novel metabolic pathways for the biologi- concentration of this cofactors would drive reactions near cal production of MEG into strains of E. coli. The general equilibrium in one direction. Therefore, if a reaction was found metabolic engineering strategy for the conversion of sug- to be near equilibrium, its reversibility was constrained in the ars into MEG is that pentoses are split into 2-carbon and direction of NADPH consumption or NADH production. 3-carbon compounds, hexoses are split into two 3-carbon compounds, and the respective 2-carbon and 3-carbon To verify if the modifications applied had any effect on the intermediates are independently converted into MEG. Here, predicted fluxes for the central carbon metabolism, the

97 Poster Abstracts

models5–8 were used for FBA simulations and the results age and improves model performance. Database (Oxford). 2013, were compared with experimental fluxes. The simulations 2013:bat059. performed with the curated models revealed several 9. Jouhten P, Rintala E, Huuskonen A, Tamminen A, Toivari M, improvements in the Pentose Phosphate Pathway and Wiebe M, Ruohonen L, Penttilä M, Maaheimo H: Oxygen depen- other parts of NADPH metabolism, resulting in a flux dence of metabolic fluxes and energy generation of Saccharomy- ces cerevisiae CEN.PK113-1A. BMC Syst. Biol. 2008, 2:60. distribution much closer to experimental values9, 10. The new flux distribution was then used as a reference for the 10. Gombert AK, Moreira dos Santos M, Christensen B, Nielsen J: Network identification and flux quantification in the central MOMA11 methodology in an in silico optimization of the metabolism of Saccharomyces cerevisiae under different condi- production of two organic acids to evaluate its impact on tions of glucose repression. J. Bacteriol. 2001, 183:1441–51. 11. quality the results. It was observed that the knock-out mu- Segrè D, Vitkup D, Church GM: Analysis of optimality in natural tants obtained were consistent with experimental evidences and perturbed metabolic networks. Proc. Natl. Acad. Sci. U. S. A. in the literature and were only valid when the curated model 2002, 99:15112–7. was used. 179. The Yeast Pathway Kit: A Method for Rational or In sum, it was shown that a careful curation of the wild-type Combinatorial Metabolic Pathways Design in Saccharomyces cerevisiae network can improve the simulation accuracy, resulting in a Filipa Pereira*1, Nadia S. Parachin2, Barbel Hähn-Hagerdal3, better correlation with experimental data. Since in vivo strain Marie-Francoise Gorwa-Grauslund3 and Björn Johansson4 design is very time consuming, these results can prove impor- 1Research Centre of Molecular and Environmental Biology tant to boost the reliability of in silico optimizations. However, (CBMA), Department of Biology, University of Minho, Braga, since the proposed changes are only valid under specific Portugal 2 conditions (full aerobiosis) it can also be concluded that the Departamento de Biologia Celular, Instituto de Ciências Biológicas, Universidade de Brasília, Brasília, Brazil accuracy of flux distribution prediction in large-scale models 3Department of Applied Microbiology, Lund University, Lund, might be dependent on condition-specific modifications. Sweden References: 4CBMA - Center of Molecular and Environmental Biology / 1. Edwards JS, Palsson BO: Metabolic flux balance analysis and Department of Biology, University of Minho, Braga, Portugal the in silico analysis of Escherichia coli K-12 gene deletions. BMC Bioinformatics 2000, 1:1. Metabolic engineering often requires multiple genetic modi- 2. Brochado AR, Matos C, Møller BL, Hansen J, Mortensen UH, fications in order to alter the properties of a target organism, Patil KR: Improved vanillin production in baker’s yeast through in especially if an entirely new metabolic pathway has to be silico design. Microb. Cell Fact. 2010, 9:84. established. Pathway elements can be rationally or randomly 3. Fowler ZL, Gikandi WW, Koffas MAG: Increased malonyl co- assembled, the latter facilitates selection of the best perform- enzyme A biosynthesis by tuning the Escherichia coli metabolic ing pathway from a set of randomly generated pathways if a network and its application to flavanone production. Appl. Environ. suitable screening or selection procedure is available. Meth- Microbiol. 2009, 75:5831–9. ods described so far allow either rational or random assembly 4. Otero JM, Cimini D, Patil KR, Poulsen SG, Olsson L, Nielsen J: of metabolic pathways in in vivo or in vitro, but not both and Industrial systems biology of Saccharomyces cerevisiae enables usually providing few reusable genetic elements such as novel succinic acid cell factory. PLoS One 2013, 8:e54144. promoters and genes. We present here the Yeast Pathway 5. Förster J, Famili I, Fu P, Palsson BØ, Nielsen J: Genome-scale Kit (YPK) that aids rational or random metabolic pathway as- reconstruction of the Saccharomyces cerevisiae metabolic net- sembly using the same genetic parts. The system is based on work. Genome Res. 2003, 13:244–53. efficient and rapid cloning using positive selection vector in 6. Mo ML, Palsson BO, Herrgård MJ: Connecting extracellular me- combination with hierarchical in-vivo gap repair. As a proof of tabolomic measurements to intracellular flux states in yeast. BMC principle, we assembled xylose metabolic pathways with up Syst. Biol. 2009, 3:37. to eight genes producing a recombinant S. cerevisiae strain 7. Österlund T, Nookaew I, Bordel S, Nielsen J: Mapping condition- able to grow on xylose with a specific growth rate of 0.181 dependent regulation of metabolism in yeast through genome- scale modeling. BMC Syst. Biol. 2013, 7:36. h-1. YPK relies on PCR reactions with short primers resulting in a relatively low cost of construction compared to other 8. Heavner BD, Smallbone K, Price ND, Walker LP: Version 6 of the consensus yeast metabolic network refines biochemical cover- protocols such as Gibson assembly or DNA assembler. 98 Poster Abstracts

180. Efficient Searching and Annotation of Metabolic facilitate more efficient translational work in the laboratory. Networks Using Chemical Similarity Dante Pertusi*, Andrew Stine, Linda J. Broadbelt and 181. Development of a Highly Efficient Gene Delivery Keith E.J. Tyo System for Syngas Fermenting Clostridia Department of Chemical and Biological Engineering, Gabriele Philipps*, Sebastian de Vries, Christian Janke, Northwestern University, Evanston, IL Nicole Schnaß and Stefan Jennewein Department of Industrial Biotechnology, Fraunhofer Institute The ability to quickly search large biochemical networks to for Molecular Biology and Applied Ecology, Aachen, discover or design novel metabolic pathways is urgently Germany needed as metabolic engineers both become more aggres- Several microorganisms belonging to the genus Clostridia sive with biosynthetic goals and have access to exponen- can convert synthesis gas (syngas), a mixture of carbon tially increasing amounts of biochemical data. In particular, monoxide, carbon dioxide and hydrogen, into more com- novel metabolic routes have the potential to offer higher plex organic molecules, including biomass, acetate and yields and better economy for pharmaceuticals and fine ethanol. The conversion of syngas into ethanol is particularly chemicals. Existing network search and generation tools are relevant in the context of biofuel production, and several powerful, but are also limited by the combinatorial explosion start-up companies are already pursuing this technology. of potential compounds and reactions in pathways that can The commercial development of syngas-based ethanol originate from a single compound node in the biochemical fermentation processes is challenging because additional network. Moreover, existing algorithms cannot efficiently an- energy is needed to purify the resulting alcohols by distilla- notate reactions with non-native enzyme substrates, despite tion. Based on the known metabolic capabilities of different this being a core activity for selecting enzymes for engineer- Clostridium strains, we can already envision the production ing. To facilitate both speed and novel pathway prediction, of several other chemicals and biofuels, most representing two algorithms, SimIndex and SimZyme, have been writ- high-value products that cannot be synthesized via estab- ten to efficiently search novel biochemical networks and to lished chemical routes such as the Fischer-Tropsch process. predict particular enzymes that—via enzyme promiscuity— can catalyze reactions required in de novo pathways that However, for the establishment of these biosynthetic path- proceed through novel intermediates. ways (beyond syngas based ethanol production) metabolic engineering approaches involving complex pathways A marked improvement in network search performance is become necessary. Up to today the metabolic engineering achieved with SimIndex when it is coupled with the predic- of Costridia in general and of syngas fermenting Clostridia tive power of the biochemical network generation software strains in particular proved rather challenging. BNICE. For test pathways with three and four compounds, SimIndex offers order-of-magnitude reductions in the number We developed a highly efficient gene delivery system ca- of nodes searched while still reaching the desired target mol- pable for introducing complex (large) gene clusters encod- ecule. The reduction of both nodes and edges (compounds ing entire biosynthetic pathways into syngas fermenting and reactions, respectively) searched enables faster pathway Clostidia such as C. ljungdahlii and C. autoethanogenum. identification within the putative network. Our method of This development will not only deliver syngas-based fer- assigning specific enzymes to putative reactions, SimZyme, mentation systems for chemicals and biofuels but will also ranks enzymes likely to perform a desired chemical transfor- foster the development of further Clostridia fermentation mation based on information in the BRENDA Database. We processes using other feed stocks such as cellulose (e.g. by validated SimZyme using a leave-one-out approach to deter- Clostridia cellulolyticum). mine how the algorithm would predict the enzymatic reaction we removed. Across four enzyme classes, the correct en- 182. Retooling Glycolysis in Saccharomyces cerevisiae zyme was ranked in the top three results in 80%-95% of the for More Efficient Isoprenoid Production 100 trial runs in each class. Our preliminary results indicate Lauren Pickens*, Kristy Hawkins, Yoseph Tsegaye, Adam that our approaches to network reduction and edge annota- Meadows, Eugene Antipov, Lan Xu, Madhukar Dasika, Anna tion in biochemical networks are useful strategies for efficient Tai, Tina Mahatdejkul-Meadows, Savita Ganesan, Jefferson assembly of novel metabolic pathways in silico, which in turn Lai, Lily Chao, Patrick Westfall, Youngnyun Kim, Timothy

99 Poster Abstracts

Gardner and Annie Tsong observed physiology or in developing metabolic engineer- Amyris ing strategies. Towards this end, we developed a framework called Flux Directionality Profile Analysis (FDPA) that enables Microbially derived complex anabolic compounds such as both the complete enumeration and the characterization isoprenoids have previously been confined to applications of all possible intracellular flux states. The proposed meth- in the pharmaceutical or specialty chemical industries, in odology allows for the seamless incorporation of available part because the energetic costs of biosynthesis translate experimental data, when available, such as intracellular to high production costs. Here, we design and construct an reaction fluxes, uptake rates, and intra—and extra—cellular alternative carbon metabolism in Saccharomyces cerevisiae metabolite concentration measurements. Additionally, FDPA which reduces the metabolic cost of converting glucose to employs a set of metabolic objectives in order to rank the farnesyl pyrophosphate (FPP), the universal precursor for all thermodynamically feasible intracellular flux states based sesquiterpene isoprenoids. By partitioning glucose dissimi- on their performance against one (or more) of these objec- lation between parallel heterologous routes and by altering tives. We are therefore able to isolate internal flux states that the cofactor requirements of FPP biosynthesis, we dramati- are both thermodynamically consistent and consistent with cally improve internal pathway balance, reducing reliance the observed physiology. We demonstrate the utility of this

upon CO2-emitting and oxygen-consuming side-reactions to framework in studying central carbon metabolism of mam- maintain redox homeostasis and energy charge. The result is malian cells and we characterize the physiological differ- a 20% relative improvement in the theoretical yield of farne- ences between two distinct growth conditions. sene from glucose (g/g), and a 170% relative improvement in the theoretical yield of farnesene from oxygen (mol/mol). We 184. Trans-Regulatory Elements As Tools for Metabolic transplant the synthetic metabolic network into a strain of S. Engineering 1 2 1 cerevisiae that was previously engineered for commercial- Mark Politz* , Matthew F. Copeland , Charles Johnson and Brian F. Pfleger1 scale production of farnesene, a commercially important 1Chemical and Biological Engineering, University of Wiscon- molecule whose derivatives can be used in a wide variety of sin-Madison, Madison, WI applications ranging from fuels to novel performance materi- 2University of Wisconsin-Madison, Madison, WI als. In this strain background, we observe an 80% relative increase in the molar yield of farnesene from oxygen, which Metabolic engineering holds the potential to enable renew- translates to a proportional increase in farnesene productiv- able chemical and fuel production. In order to fulfill this ity in commercial fermentation conditions. This improvement potential, native metabolic pathways in relevant industrial in productivity enables microbial production of an anabolic host organisms must be optimized to channel flux towards secondary metabolite at a scale previously accessible only to desirable molecules. The complex task of optimizing native catabolites and primary metabolites. metabolism for chemical production would be facilitated by reliable trans-acting regulatory elements with customiz- 183. Systematic Characterization of Intracellular Meta- able target specificities, enabling engineers to probe chro- bolic States through Flux Directionality Profile Analysis mosomal gene expression levels without disrupting native Alexandros Kiparissides1,2, Joana Pinto Vieira*1,2 and Vassily Hatzimanikatis1,2 regulation. Fortunately, several trans-acting elements are 1Laboratory of Computational Systems Biotechnology, available to knock down the expression of any target gene EPFL, Lausanne, Switzerland of interest. These include Transcription Activator-Like Ef- 2Swiss Institute of Bioinformatics, Lausanne, Switzerland fectors (TALEs), the CRISPR/Cas system of Streptococcus pyogenes, and small RNAs (sRNAs). Here, we demonstrate Metabolic modeling has proven to be a valuable tool for the that TALEs may be effectively used to repress gene expres- investigation of cellular physiology in metabolic engineering sion in Escherichia coli. In addition, comparative studies studies. However, the sheer complexity of even the sim- performed to determine if any of the three aforementioned plest metabolic networks obstructs the exact determination systems are especially suited for metabolic engineering ap- of intracellular states, despite the ample supply of –omics plications are presented. data. Identifying the gamut of possible feasible states allows us to query for factors that are crucial in interpreting the

100 Poster Abstracts

185. Multiplex Amino Acid Metabolism Engineering for constituents of gasoline and jet fuel with ever increasing Increased Production of L-Ornithine in Yeast demand as alternative biofuels. However, the conventional 1,2 1 1 Jiufu G. Qin* , Yongjin J. Zhou , Anastasia Krivoruchko , n-alkanes production process based on expensive raw Verena Siewers1 and Jens Nielsen3 materials (coal, hydrogen and cobalt) increases the overall 1Department of Chemical and Biological Engineering, Chalmers University of Technology, Göteborg, Sweden production cost of n-alkanes. To meet the growing demand, 2State Key Laboratory of Food Science and Technology, efforts have been made to engineer microbial systems for Jiangnan University, Wuxi, China the economical production of n-alkanes. Nevertheless, the 3The Novo Nordisk Foundation Center for Biosustainability, microbial productions of n-alkanes are far below a com- Technical University of Denmark, Hørsholm, Denmark mercial threshold. Biologically, n-alkanes are produced from fatty acyl-ACPs with the help of acyl-ACP reductases Recent advances in metabolic engineering and synthetic (AAR) and aldehyde deformylating oxygenases (ADO). One biology enable the rational engineering of microbial amino of the major challenges in the biological n-alkanes produc- acid metabolism for efficient production of valuable amino tion process is a slow catalytic turnover rate of ADO. To acid-derived molecules. Several unique superiorities of increase the n-alkanes production, we controlled the spatial Baker’s yeast Saccharomyces cerevisiae make it an alterna- arrangement and stoichiometric ratio of enzymes. First, a tive host for production of valuable chemicals. However, chimeric protein of AAR and ADO was synthesized. Second, its complex regulation of metabolism and compartmental- the enzymes were arranged on a DNA scaffold with various ization of specific metabolic functions imposes additional ratios. As the result, production of n-alkanes was increased challenges for engineering its metabolism compared with (4.4-fold) by the chimeric fusion of ADO-AAR compared to a bacteria. Here, we demonstrate a multiplex and systematic control strain expressing wild type AAR and ADO. Further- engineering approach to thoroughly engineer the amino acid more, when the ratio of ADO to AAR was 3 to 1, n-alkanes metabolism of yeast. Our multiple engineering strategies production was increased (8.8-fold) compared to the control includes, besides traditional over-expression of pathway strain. Our results showed that the spatial organizations of enzymes, recruitment of subcellular trafficking and its en- enzymes using protein chimera and DNA scaffolds are ap- dogenous regulation, and we demonstrate that it is hereby plicable for establishing an efficient n-alkanes biosynthetic possible to overproduce L-ornithine, an intermediate of the pathway in Escherichia coli. arginine biosynthetic pathway. The final engineered strain could produce 1.46 g l-1 of L-ornithine with a specific yield 187. Tools for Advancing Genome Engineering on the of 78.27 mg (g DCW)-1 during a fed-batch fermentation, Protein, Pathway, and Genome Scale which is the highest reported production of any amino acid Emily Freed, Thomas J. Mansell, Gur Pines, T. Steele by S. cerevisiae. Our multiple engineering strategy demon- Reynolds* and Ryan T. Gill Chemical and Biological Engineering, University of strate the importance of systematic pathway optimization Colorado, Boulder, CO and show how recruitment of endogenous regulation can possible be used as a generalized approach for overproduc- Coordinating the action of biological parts (promoters, regu- ing amino acids by S. cerevisiae and possible even other lators, genes, etc.) to construct new genomes and biologi- eukaryotic microorganisms. cal systems with desirable phenotypes is one of the main challenges facing the field of metabolic engineering. Recent 186. Improved N-alkanes Production in Escherichia coli By Spatial Organization of Alkane Biosynthetic Pathway advancements in DNA synthesis, and improved understand- Enzymes ing of homologous recombination and the CRISPR-Cas sys- Ziaur Rahman*1, Le Minh Bui1, Bong Hyun Sung2, Almando tem enable new approaches that integrate rational design Geraldi1, Kyo Hun Kang1, Jun Hyoung Lee1 and Sun Chang and directed evolution to both create novel biological parts 1 Kim and inform how to assemble these parts into systems. We 1Biological Sciences, Korea Advanced Institute of Science present several genome engineering tools with applications and Technology, Daejeon, South Korea 2Biochemicals and Synthetic Biology Research center, Korea on the protein, pathway, and full genome scale. Research Institute of Bioscience and Biotechnology We have developed a method for protein engineering based Linear hydrocarbons, especially n-alkanes, are the major on the CRISPR-Cas RNA-guided nuclease system. We cou-

101 Poster Abstracts

ple desired point mutations within the open reading frame Federale de Lausanne (EPFL), Lausanne, Switzerland with a common mutation in the guide RNA recognition site, 2Swiss Institute of Bioinformatics, Lausanne, Switzerland 3 thus enabling a CRISPR-induced cell death of wild type Biochemistry, University of Geneva, Geneva, Switzerland cells and the survival of mutants. This approach enriches Lipids are major constituents of the cell. They are respon- significantly the mutated cells within the population without sible for major properties of the cellular membranes: hy- the need for an antibiotic selection marker. drophobicity, selective permeability and being the scaffold of signaling proteins. Many diseases are associated with We have applied this approach to protein engineering or- alterations in the lipid distribution in the cell and the compo- thogonal regulators, namely the tet repressor (TetR). Cou- sition of membrane domains. Metabolic syndrome, obesity, pled with a unique selection for orthogonal DNA and mRNA atherosclerosis, as well as Alzheimer’s, Huntington’s dis- binding, we present TetR mutants that allow increased eases and cancer, have an impact in the levels of lipids, with control of cellular circuits on the transcriptional and post- observed alterations in their concentrations compared to transcriptional levels. This method can potentially be scaled the healthy state. Applying computational techniques along to the pathway level as well. with systematic modeling of lipid metabolism can provide insights that can guide biomedical research and develop At the genome scale, we are working on improved meth- potential strategies for prevention and cure. ods to link genomic alterations to traits. This new method, TRMR2.0, allows the fine-tuning of the expression level of In the present study we developed a comprehensive model each gene in the E. coli genome in parallel. Molecular bar- of the sphingolipid biosynthesis in the yeast Saccharomyces coding allows rapid tracking of genes and expression levels cerevisiae. Sphingolipids are one of the four major lipid cat- conferring desirable phenotypes. Finally, we developed a egories, along with (glycero)phospholipids, sterols and fatty mathematical model of chromosomal segregation with the acids synthesized in the yeast S. cerevisiae. The importance of goal of optimizing genome engineering efforts to produce sphingolipids present in any higher eukaryote has been dem- more stable and reliable genotype alterations. onstrated in many recent studies. For this study, S. cerevisiae 188. Ultra-High-Throughput Screening of Enzyme has been chosen as a model organism due to its high homol- Libraries with Droplet-Based Microfluidics ogy of cellular processes with mammalian cells. The developed Philip Romero*, Tuan Tran and Adam R. Abate model will be an essential part towards the construction of a Department of Bioengineering and Therapeutic Sciences, detailed kinetic model of the whole lipidome of the cell. California Institute for Quantitative Biosciences, University of California, San Francisco, San Fransisco, CA We first constructed a stoichiometric model that contains all the currently known reactions for the biosynthesis of cerami- Directed evolution relies on the ability to screen large, des and complex sphingolipids in the yeast. Additionally, we diverse libraries of sequence variants. We have developed have accounted for all five reported hydroxylation states, a microfluidic platform for ultra-highthroughput screening of along with the reactions synthesizing the necessary precur- enzyme libraries. Individual enzyme variants are assayed in sor metabolites from other lipid pathways (i.e. palmitate- a monodisperse water-in-oil microemulsion. Aqueous drop- CoA from fatty acid synthesis and phosphatidylinositol from lets compartmentalize the enzymatic reaction and maintain the phospholipids metabolism). We next developed a kinetic the genotype-phenotype linkage. A quantitative fluores- model and we used a large number of lipidomic measure- cence detection system allows droplets to be analyzed ments of wild type yeast to consistently calibrate our model. and sorted at rates beyond 1 kHz. This screening platform Curation of the kinetic information of the model came from provides a throughput that is orders of magnitude beyond the comprehensive mining of references for operation of the traditional plate-based screens. enzymes as well as ranges of kinetic parameters from online databases. These datasets created the pool for a sampling 189. Kinetic Reconstruction and Analysis of Sphingolipid Metabolism technique that accounts for the uncertainty in the param- Georgios Savoglidis*1,2, Isabelle Riezman3, Aline X.S. eters of the model. This lead to a robust dynamic model, Santos3, Howard Riezman3 and Vassily Hatzimanikatis1,2 containing mass balances for all the components of the 1Chemistry and Chemical Engineering, Ecole Polytechnique 102 Poster Abstracts

biochemical network as well as terms that accounted for the functional expression of plant geraniol or limonene syn- dillution of these molecules in the cell due to growth. thases already led to the production of small amounts of the desired acids in shake flask cultures indicating exploitation We performed a thorough kinetic analysis of the system by of terpene precursors derived from the endogenous MEP examining the impact of different assumptions in enzyme pathway. Precursor supply was improved by expression of operation on the levels of ceramides and complex sphingo- the heterologous mevalonate (MVA) pathway, an elegant lipids (e.g. substrate competition, (un)competitive inhibition). concept of engineering the terpene biosynthesis in microbes By applying the principles of Metabolic Control Analysis first published by Martin et al. 20033. As donor strain of the (MCA) we were able to quantify the effect of enzyme activi- MVA genes, we chose Myxococcus xanthus for two rea- ties on the lipid profiles and we identified enzymes of the sons: the ease of the cloning procedure (5 of 6 genes are biochemical reaction network as efficient targets for meta- organized in one operon) and the fact that it is a prokaryot bolic engineering towards a desired state. We also applied which may facilitate functional expression in the prokaryotic a reverse engineering method and we were able to identify host. P. putida strains coexpressing Mentha spicata limo- enzyme perturbations responsible for various measured nene synthase and MVA pathway showed a 17-fold increase mutant states compared to the wild type cellular lipidomic in formation of perillic acid (0.68 mg/L). Strains coexpress- profile. Such an approach could lead to the identification ing Ocimum basilicum geraniol synthase and MVA pathway of genetic mutations by imploring the information containd showed a 13-fold increase in formation of geranic acid (16.8 within metabolomic measurements. mg/L); after doubling the concentration of the C source glycerol a product concentration of 36.1 mg/L was achieved Although we demonstrate the application of the method in in shake flasks. Further improvements of productivity and models of lipid metabolism it is possible to be applied to final product concentration can therefore be expected with biochemical systems of various parts of metabolism and an optimized fed-batch process on bioreactor scale. This is, sizes creating a modular platform for kinetic analysis of cel- to our best knowledge, the first example of de novo mono- lular operations. terpenoic acid production with an engineered microbe. We intend to further improve the product titers in the medium by 190. De Novo Production of Monoterpenoic Acids with pathway and bioprocess engineering to eventually take ad- Pseudomonas Putida vantage of the host’s pronounced monoterpene tolerance. Jens Schrader*, Markus Buchhaupt, Jia Mi, Patrice Lubuta Comparing growth of E. coli, S. cerevisiae and P. putida and Daniela Becher Biochemical Engineering, DECHEMA Research Institute, showed, that P. putida has a several times higher tolerance Frankfurt am Main, Germany towards limonene, perillic acid and geranic acid and an equal tolerance towards geraniol. We are also investigating Production of plant terpenes by engineered microbes has the cellular mechanisms to identify those responsible for the become a prime example of applied synthetic biology with monoterpene robustness of this strain. We assume the in- tremendous progress being made during the last decade. volvement of energy-dependent efflux pumps and a cellular Whereas sesquiterpene titers reported have already reached response by adapting the membrane composition4. g/L values in the bioreactor, efficient monoterpene pro- References: duction seems to be more difficult with conventional host [1] Mirata MA et al. (2009) Integrated bioprocess for the oxidation strains due to product toxicity. Hence, we set out to investi- of limonene to perillic acid with Pseudomonas putida DSM 12264, gate the potential of a solvent tolerant Pseudomonas putida Process Biochem 44: 764–771 strain for de novo monoterpene production. Our target [2] Yang T et al. (2011) Metabolic engineering of geranic acid in molecules are geranic acid and perillic acid, plant monoter- maize to achieve fungal resistance is compromised by novel glyco- penoic acids which show broad antimicrobial activity and sylation patterns, Metabolic Engineering 13(4):414-25 are of commercial interest, e.g. as natural preservatives for [3] Martin VJ et al. (2003) Engineering a mevalonate pathway cosmetics or antifungal agrochemicals1,2. in Escherichia coli for production of terpenoids, Nat Biotechnol 21(7):796-802. Since P. putida DSM12264 wild type is able to oxidize [4] Segura A et al. (2012) Solvent tolerance in Gram-negative bac- geraniol to geranic acid and limonene to perillic acid, the teria, Curr Opin Biotechnol 23(3):415-21

103 Poster Abstracts

191. A Computationally-Driven Metabolic Engineering due to its desirable fuel properties and its role as a platform Strategy to Increase Cellulose Production in Plants chemical. Currently, the global market of n-butanol has been 1 2 1 Jiun Yen , Glenda Gillaspy and Ryan S. Senger* estimated over $5 billion, with a predicted 4.7 % expansion 1Biological Systems Engineering, Virginia Tech per year (Mascal 2012). n-Butanol can be produced com- 2Biochemistry, Virginia Tech mercially from fossil fuels, through generally expensive and environmentally unfriendly routes. The biological production A novel approach in genome-scale metabolic flux modeling of n-butanol is traditionally through ABE (acetone-butanol- called Flux Balance Analysis with Flux Ratios (FBrAtio) has ethanol) process by Clostridia species. The limitation of been developed to derive “fine-tuned” metabolic engineer- this process includes lack of genetic tools, undesired ing strategies for the production of biofuels and chemicals. byproducts, and low alcohol tolerance of Clostridia spe- In general, the FBrAtio algorithm designs metabolic flux cies. To overcome these hurdles, various attempts have re-direction at critical metabolite “nodes” in the metabolic been made to transfer the Clostridial n-butanol pathway to network. Based on the optimum re-distribution at several more suitable industrial organisms including Escherichia critical nodes, metabolic engineering strategies are de- coli and Saccharomyces cerevisiae. This CoA-dependent signed using tools of (i) gene over-expression, (ii) partial pathway enables an impressive production titer (30 g/L) in knock-down by designer sRNA, and (iii) full gene knockout. E. coli (Shen et al. 2011). However, much lower productions In this research, FBrAtio was applied to Arabidopsis thaliana have been reported for S. cerevisiae, with reports of only 2.5 with the purpose of increasing cellulose content of plant mg/L (Steen et al. 2008) and 16.3 mg/L (Krivoruchko et al. biomass for use in consolidated bioprocesses. The FBrAtio 2013) from glucose. At the same time, n-butanol produc- approach returned a single non-intuitive gene over-expres- tion at grams per liter in E. coli by the amino-acid metabolic sion candidate to dramatically increase cellulose synthe- pathway (Atsumi et al. 2008) and the reversed β-oxidation sis. When implemented, record levels of cellulose (~100% pathway (Dellomonaco et al. 2011) were reported. increase over wild-type) were achieved in plants that grew

significantly larger (>20% increase) than wild-type plants Although the current production level of n-butanol in S. and had dramatically thicker stems (2-3 fold increase). cerevisiae is far less promising, there are advantages to These engineered plants represent a significant step forward utilize S. cerevisiae as an n-butanol producer, due to its high towards designer crops that can provide easily accessible n-butanol tolerance, phage resistance, well-established ge- and abundant cellulose for consolidated bioprocessing netic tools and compatibility to current industrial infrastruc- while growing on marginal lands, in low light, or in drought ture (Si et al. 2014). A recent paper suggested a metabolic conditions. In this presentation, a full description of the route for n-butanol production from l-glycine in S. cerevisiae FBrAtio algorithm will be presented along with instructions (Branduardi et al. 2013). The production relies on the addi- regarding how to use this approach to design multigenic tion of substrate, L-glycine. In addition, the pathway can- metabolic engineering strategies in any organism of interest. not avoid the production of isobutanol, and n-butanol and In addition, full details of the plant metabolic engineering isobutanol accumulate simultaneously. targets and process will be given along with a full characterization of the high cellulose producing plants and a way Here we report the discovery, characterization and engineer- forward for engineering additional useful traits in designer ing of an endogenous n-butanol pathway in S. cerevisiae. crops to be used for consolidated bioprocessing. The pathway was switched on to produce a large amount 192. Metabolic Engineering of an Endogenous Pathway of n-butanol from glucose (120 mg/L) by introduction of a for n-Butanol Production in Saccharomyces cerevisiae single gene deletion adh1Δ (Si et al. 2014). Little isobutanol Shuobo Shi*1, Tong Si2, Ee Lui Ang1 and Huimin Zhao2 (below 12 mg/L) was produced in this pathway. In addition 1Metabolic Engineering Research Laboratory, Institute of to the deletion of ADH1 for n-butanol production, we engi- Chemical Engineering and Sciences, Singapore, Singapore neered yeast to have an increased flux toward threonine, the 2Department of Chemical & Biomolecular Engineering, precursor metabolite for n-butanol biosynthesis. Elimination University of Illinois at Urbana Champaign, Urbana, IL of competing pathways could increase the n-butanol titer by up to 106%. The pathway downstream of threonine has Increasing concerns about depleting crude reserves have been over expressed in mitochondria or cytosol resulting in renewed interests in the biological production of n-butanol, 104 Poster Abstracts

a 87 % increase in the final titer when the pathway is in mi- because of the absence of the enzymatic reactions in cell. For tochondria and a 40 % increase when the same pathway is microbial productions of non-biological compounds synthe- in cytosol. Construction of a cimA mediated pathway further sized chemically from fossil materials, novel metabolic path- increased the n-butanol titer to 349 mg/L, which is the high- ways should be constructed in a cell. In this study, an in silico est n-butanol titer ever reported and represents a 21-fold tool was explored for searching for possibilities of enzymatic improvement compared to previous values reported in S. reactions. First, patterns of enzyme reactions registered in cerevisiaefrom glucose (Krivoruchko et al. 2013). A combi- KEGG or BRENDA databases were recategorized focused on nation of beneficial manipulations and strain engineering is each enzyme active center. Second, information of metabolites still in progress for further strain improvement. were transferred into vectors for calculations. Finally, in silico References: novel metabolic pathways for target chemicals were found using the two information. This search tool, combined with Atsumi S, Hanai T, Liao J. 2008. Non-fermentative pathways for synthesis of branched-chain higher alcohols as biofuels. Nature thermodynamics simulations and flux balance analysis (FBA) 451(7174):86-89. for the optimal metabolic pathway design in a cell, was applied Branduardi P, Longo V, Berterame N, Rossi G, Porro D. 2013. A for founding useful but non-biological chemicals such as engi- novel pathway to produce butanol and isobutanol in Saccharomy- neering plastics, rubber materials. ces cerevisiae. Biotechnol. Biofuels 6(1):68. Dellomonaco C, Clomburg JM, Miller EN, Gonzalez R. 2011. 194. Production of Fatty Acid Derived Biofuels in Engineered reversal of the beta-oxidation cycle for the synthesis of Saccharomyces cerevisiae fuels and chemicals. Nature 476:355–359. Verena Siewers* and Jens Nielsen Department of Chemical and Biological Engineering, Krivoruchko A, Serrano-Amatriain C, Chen Y, Siewers V, Nielsen J. Chalmers University of Technology, Gothenburg, Sweden 2013. Improving biobutanol production in engineered Saccharomy- ces cerevisiaeby manipulation of acetyl-CoA metabolism. J. Ind. Microbiol. Biotechnol.:1-6. Efforts to replace petroleum derived products include the development of sustainable processes to produce a variety Mascal M. 2012. Chemicals from biobutanol: technologies and of biofuels. While ethanol can be efficiently produced in yeast markets. Biofuels, Bioprod. Biorefin. 6(4):483-493. and used as a gasoline additive, its properties are not suitable Shen CR, Lan EI, Dekishima Y, Baez A, Cho KM, Liao JC. 2011. e.g. for diesel engines. Our aim is to generate yeast strains Driving forces enable high-titer anaerobic 1-Butanol synthesis in Escherichia coli. Appl. Environ. Microbiol. 77:2905-2915. optimized for the production of molecules that can be used as diesel fuels, especially fatty acid derived compounds. Si T, Luo Y, Xiao H, Zhao H. 2014. Utilizing an endogenous pathway for 1-butanol production in Saccharomyces cerevisiae. Metab. Eng. 22:60-68. Several efforts are combined to streamline yeast metabo- Steen E, Chan R, Prasad N, Myers S, Petzold C, Redding A, Ouel- lism and to generate a portfolio of platform strains for let M, Keasling J. 2008. Metabolic engineering of Saccharomyces different applications. These include (i) the screening of a cerevisiae for the production of n-butanol. Microb. Cell Fact. broad range of heterologous pathways for production of 7(1):36. the specific product class, (ii) optimization of acetyl-CoA as well as acyl-CoA precursor supply, (iii) inhibition of fatty acid 193. Development of an in silico Tool of Novel Metabolic catabolism and by-product formation, and (iv) construction Pathway Designs for Microbial Productions of strains producing fatty acids of specific chain length. We Tomokazu Shirai*1 and Akihiko Kondo2 are furthermore designing biosensors that allow for a fast 1Biomass Engineering Program, Riken, Kanagawa, Japan screening for strains with increased production of either the 2 Graduate School of Engineering, Department of Chemical final pathway products or intermediates. Science and Engineering, Kobe University, Kobe, Japan 195. Rapid One-Step Inactivation of Single or Multiple For countermeasures to problems of fossil material depletions Genes in Escherichia coli or concerns over global environmental problems, technologies Chan Woo Song1, 3*, Sang Yup Lee1, 2, 3 using non-fossil resources should be constructed in an urgent 1Metabolic and Biomolecular Engineering National Research manner. However, most of useful chemicals synthesized from Laboratory, Department of Chemical and Biomolecular Engi- neering (BK21 program), Center for Systems and Synthetic fossil material have not been produced by microorganisms Biotechnology, Institute for the BioCentury, Korea Advanced

105 Poster Abstracts

Institute of Science and Technology (KAIST), Republic of requiring lab tests that are prohibitively expensive in both Korea material and human resources for those in developing or 2 BioInformatics Research Center, KAIST, Republic of Korea remote areas. 3BioProcess Engineering Research Center, KAIST, Republic of Korea As obligate consumers of the same micronutrients, bacteria We developed an integrated helper plasmid-based gene possess cellular machinery to control intracellular micronu- manipulation system for more efficient and rapid engineering trient levels and have corresponding regulatory mechanisms of Escherichia coli. The integrated helper plasmid, pCW611, to respond to varying concentrations in their environment. contains two recombinases which are expressed in reverse We have developed whole-cell bacterial biosensors har- direction by two independent inducible systems. One is nessing these properties of bacteria for use in a diagnostic Red recombinase under the control of arabinose inducible test for blood micronutrient status. The tests are designed system to induce recombination event by using linear gene to be inexpensive in total cost of operation, requiring no knockout DNA fragment while the other is Cre recombinase complex equipment and minimal medical training to admin- which is controlled by IPTG inducible system to obtain ister and interpret. This would obviate the logistical problem marker-less mutant strains. The main advantage of this sys- of laboratory access and sample transport in remote and tem is that the time and effort required can be significantly low-resource environments, allowing on-site diagnosis of reduced because the iterative transformation of the helper micronutrient deficiencies in the populations most at risk. plasmid and curing steps are not required. We could delete one target gene in 3 days by using pCW611. To verify the To create this biosensor, we designed and implemented usefulness of this gene manipulation system, the deletion genetic circuitry to trigger specific changes in color in re- experiments were performed for knocking out four target sponse to defined micronutrient levels. In particular, we use genes individually (adhE, sfcA, frdABCD, and ackA) and pigments as readouts for the biosensor, not only for their two genes simultaneously for two cases (adhE-aspA and naked eye visibility but because with appropriate metabolic sfcA-aspA). Also, sequential deletion of four target genes engineering, color changes could be essentially switch- (fumB, iclR, fumA, and fumC) was successfully performed like, reducing ambiguity in the interpretation of test results. for the construction of fumaric acid producing strain. The Moreover, production of even a small amount of protein efficiencies of target gene replacement and removal of the could enable synthesis of sufficient pigment to be visible to resistance marker were nearly 50% and 100%, respectively. the naked eye, potentially decreasing deleterious effects on This rapid and efficient gene manipulation system suc- the cell. Here we present the results of our efforts to create cessfully developed and validated should be useful for the and tune this sensor using a variety of metabolic engineer- metabolic engineering of E. coli.(Development of systems ing strategies, including changes in promoters, ribosomal metabolic engineering platform technologies for biorefiner- binding sites, and transporters. ies; NRF-2012-C1AAA001-2012M1A2A2026556) funded by 197. The Role of Trehalose Metabolism in Saccharomy- the Ministry of Education, Science and Technology) ces cerevisiae from a Quantitative Approach Camilo A. Suarez-Mendez*1,2, Isabelle Duijnstee1, 196. Pigment-Based, Low-Cost, Portable Micronutrient J.J. Heijnen1,2 and S. Aljoscha Wahl1,2 Status Tests Using Engineered Bacteria 1Department of Biotechnology, Delft University of Mark P. Styczynski* Technology, Delft, Netherlands School of Chemical & Biomolecular Engineering, Georgia 2Kluyver Centre for Genomics of Industrial Fermentation, Institute of Technology, Atlanta, GA Delft, Netherlands

Micronutrient deficiencies are a significant healthcare con- INTRODUCTION cern across the globe. Significant even in some developed Saccharomyces cerevisiae utilizes two main mechanisms for nations, micronutrient deficiencies are more severe in the storage of glucose: via glycogen and via trehalose. Though developing world and locally in the wake of major disasters. both pathways are closely related through the metabolism These conditions, though easily treated, remain a problem of G6P and UDPG, the onset of accumulation of these stor- because they are often difficult to recognize and diagnose, age carbohydrates differs. It has been observed that this

106 Poster Abstracts

accumulation depends on growth conditions with glycogen famine consisted of an intermittent (block-wise) feeding accumulating mainly while the carbon source (i.e., glucose) regime in which the feeding pump was set on for 20s and is present in the medium. On the contrary, trehalose is usu- then switch off for 380s making up a cycle time of 400s. ally formed when the glucose is close to depletion. At low During this cycle time the residual glucose concentration growth rates S. cerevisiae accumulates high amounts of varied from high availability to almost complete depletion trehalose for two purposes: energy and carbon storage, and and vice versa. Highly reproducible cycles were obtained stabilization during stress conditions. by the block-wise feeding allowing for extensive sampling for determination of metabolite concentrations and label- Trehalose can be converted into glucose through enzy- ing state at different time-points during the cycle. The feast/ matic hydrolysis catalyzed by either Ath1p or Nth1p/Nth2p. famine setup may also be used for obtaining other types The former is known to operate on the outer cell surface of samples (e.g., proteomics) which in combination with (periplasm) converting trehalose extracellularly at acidic metabolite concentrations and labeling can be interpreted in pH, while the latter is located in the cytosol and functions the context of metabolic regulation and robustness at neutral pH. Furthermore, the acidic Ath1p trehalase has been also found in the vacuole. Due to a continuous RESULTS AND DISCUSSION turnover that vary rapidly as a response to environmental Clear and distinguishable dynamic responses to a pertur- changes and consumes ATP, it is expected that trehalose bation were observed for the different pathways analyzed exhibits a high impact on the central metabolism, especially (upper and lower glycolysis, pentose phosphate, tricarboxylic under dynamic conditions (e.g., at large industrial bioreac- acid and storage). In particular, 6PG and T6P exhibit delayed tors or stimulus response experiments at bench-top scale)1. responses compared to their precursor G6P suggesting the presence of regulatory mechanisms. Estimation of both, the Although extensive information has been published on steady-state and feast/famine dynamic fluxes showed that the role of trehalose in yeast metabolism, there is still few most of the carbon flux at the G6P node was directed to gly- quantitative knowledge with respect to the extent of such an colysis (about 80%), while about 10% was channeled to the effect. For instance, flux estimations based on 13C-labeling pentose phosphate pathway and the rest was invested in the experiments may be distorted due to the continuous cycling storage carbohydrate (trehalose and glycogen). of trehalose. In order to elucidate the impact of trehalose metabolism from a quantitative point of view, we aim to CONCLUSIONS quantify the dynamic response of the storage pools in rela- The aim of this work was to quantify up to what extent the tion with central metabolism under well-defined cultivation metabolism of storage carbohydrates interacts with the cen- conditions. Thus, stable (13C) isotope labeling is used in tral metabolism, and thus may alter the estimation of carbon combination with advance measurement techniques and fluxes. We determined that trehalose and glycogen can be modelling for non-stationary flux identification. recycled by diverting the carbon flux up to about 10-15% of the glucose uptake. The recycle of storage carbohydrates in- APPROACH volves not only the consumption of energy (i.e., ATP) but also In this study, a wild type strain and two mutant strains makes the estimation of carbon fluxes to be ill determined, deficient in trehalose metabolism were cultivated aerobically especially during dynamic conditions. Finally, introducing an under chemostat (D = 0.1 h-1) and feast/famine (dynamic extension to the model of the central metabolism to account cycles of high and low substrate availability) conditions. In for storage recycle, revealed a significant flux through the tre- addition U-13C-glucose was used as a tracer for later flux halose node. In addition, 13C-labeling data proves to be highly estimation by using measurements of mass isotopomers informative in order to quantify these intracellular cycles. and concentrations. Thus, 13C-flux estimation has been Reference performed at both steady and dynamic states. [1] J. H. van Heerden, M. T. Wortel, F. J. Bruggeman, J. J. Heijnen, Y. J.M. Bollen, R. Planqué, J. Hulshof, T. G. O’Toole, S. A. Wahl, B. During the metabolic dynamic experiment, the microor- Teusink., LOST IN TRANSITION: uncontrolled startup of glycolysis ganisms were exposed to cyclic regimes of high and low results in subpopulations of non-growing cells. Science (submit- substrate availability (feast/famine). In this case, the feast/ ted), 2014

107 Poster Abstracts

198. Characterization of Anaerobic Central Metabolism tude of linear biosynthetic pathways have been developed to Improve Succinate Production in Enterobacter and optimized. These existing pathways as well as new Aerogenes pathways can be introduced in parallel into a single host Yoshinori Tajima*1, Kenichi Kaida1, Atsushi Hayakawa1, to produce chemical compounds which microbes naturally Ryosuke Fudou2, Keita Fukui1, Yousuke Nishio1, Kenichi Hashiguchi3, Kazuhiko Matsui4, Yoshihiro Usuda3 and Koji produce in trace amounts or not at all. Since enzymes often Sode5 have promiscuous activity, by changing the flux toward 1Institute for Innovation, Ajinomoto Co., Inc. different precursors, the final product of a multi-part bio- 2Japan bioindustry association synthetic pathway can be altered in a combinatorial fashion. 3 Research Institute for Bioscience Product & Fine Here we explored the potential of multi-part synthesis in Chemicals, Ajinomoto Co., Inc. Escherichia coli by introducing parallel biosynthetic pathways. 4Business Strategy and Development Department, Ajinomoto Co. Systematic comparison of the components in each step of 5Graduate School of Engineering, Tokyo University of the biosynthetic pathway and careful matching of genes, Agriculture & Technology pathway, and chemical toxicity to the chosen host enabled specific and efficient production of chemical compounds. Succinate is a core biochemical building block, and optimizing production by microbial fermentation from biomass is a 200. Fatty-Acid Production in Yeast through Reversal of focus of basic and applied biotechnology research. Lower- the Beta-Oxidation Cycle Paulo G. Teixeira*1, Yongjin Zhou1, Verena Siewers1 and Jens ing pH in anaerobic succinate fermentation culture is a cost- Nielsen1,2 effective and environment-friendly approach to reducing the 1Department of Chemical and Biological Engineering, use of sub-raw materials such as alkali, which are needed Chalmers University of Technology, Göteborg, Sweden for neutralization. To perform succinate fermentation under 2The Novo Nordisk Foundation Center for Biosustainability, acidic (below pH 6.0) and anaerobic conditions, we selected Technical University of Denmark, Hørsholm, Denmark and characterized anaerobic metabolism in a newly isolated Enterobacter aerogenes strain that rapidly assimilates Microbial cell factories based on genetically engineered glucose at pH 5.0. Single-gene knockout studies showed yeast cells present a great opportunity for the sustainable that the ethanol synthesis pathway serves as the dominant production of both high-value and bulk chemicals and NADH re-oxidation pathway in this organism. We also gen- derivatives. Among those, renewable biofuel production is erated a ΔadhE/PCK strain to eliminate ethanol formation, regarded as a hotspot as an industrially relevant application. and introduced a heterogeneous carboxylation enzyme. The In this work, we present a metabolic engineering strategy strain produced succinate from glucose with a 60.5% yield for the production of short- and medium-chain fatty acids (grams of succinate produced per gram of glucose con- in S. cerevisiae that can be recognized as a reversal of the sumed) at pH 5.4 and anaerobic conditions. fatty-acid beta-oxidation cycle which could later be combined with a biofuel-producing strategy using fatty acids as 199. Expanding Biosynthetic Pathways Based on a precursor. Results from the expression of this pathway for Thermodynamic Preferences production of fatty alcohols will be presented. Gabriel Rodriguez, Yohei Tashiro* and Shota Atsumi Chemistry, UC Davis, Davis, CA 201. Increased 3-Hydroxypropionic Acid Production from Glycerol Based on the Genome-Scale Metabolic Whole-cell biocatalysts have the ability to catalyze reactions Simulation in Escherichia coli at ambient temperatures and in aqueous solutions. Inside a Kento Tokuyama*1, Satoshi Ohno1, Katsunori Yoshikawa1, 2,3 1 1,4 cell, various cofactors that contain stored energy drive reac- Takashi Hirasawa , Shotaro Tanaka , Chikara Furusawa and Hiroshi Shimizu3 tions forward. Metabolic flux toward a desired final product 1Department of Bioinformatic Engineering, Graduate School can drastically increased by coupling these cofactors even of Information Science and Technology, Osaka University, if the reaction is thermodynamically unfavorable at ambient Japan temperatures. Another advantage of whole-cell biocatalysts 2Department of Bioengineering, Tokyo Institute of is the ability to achieve multi-part syntheses, whereby mul- Technology, Japan 3 tiple intermediates are generated in parallel in the same host Graduate School of Information Science and Technology, Osaka University, Suita, Japan and are combined into a final product. In addition, a multi- 108 Poster Abstracts

4Quantitative Biology Center, RIKEN, Japan 5Graduate School of Biological Science, Nara Institute of Science and Technology 3-Hydroxypropionic acid (3HP) has been attracted attention because of its availability for precursor of various valuable Bacillus subtilis is an attractive microorganism for biochemicals such as acrylic acid and poly(3HP). Glycerol is production because the genome can be easily edited to one of the potential substrates for the bioproduction of 3HP give the desirable abilities. It was revealed that the genome since it is a surplus byproduct of growing biodiesel pro- reduction provides useful properties for enhancing and duction and it can be converted to 3HP by only two reac- maintaining the recombinant proteins productivity during tions. For improvement of the target production, in silico a stationary phase (Morimoto et al., 2008). Dipicolinic acid simulation using a genome-scale metabolic model with flux (DPA), major component of B. subtilis spore, is a useful balance analysis is a powerful tool to identify the candidate chemical as one of the precursors for bio-plastic. We con- genes for metabolic engineering. In this study, we tried to structed the DPA producing B. subtilis strains from wild-type increase 3HP production in Escherichia coli by modifica- 168 and its genome-reduced strain MGB874 by changing tion of the central metabolism based on the genome-scale the promoter of spoVFAB genes encoding DPA synthase metabolic simulation. for the expression. The DPA was mainly produced during stationary phase in both strains, and the yield and the pro- For 3HP production from glycerol in E. coli, following genes duction rate were improved by the genome reduction. In the were over-expressed in E. coli MG1655(DE3): dhaB and gdr- present study, we investigated the metabolic flows on the AB encoding glycerol dehydratase and glycerol dehydratase central carbon metabolism during the stationary phase, and reactivase derived from Klebsiella pnumoniae, respectively, performed the metabolic engineering for further enhancing and aldH encoding aldehyde dehydrogenase derived from E. the DPA productivity. coli (hereafter termed “3HP strain”). To identify the candidate genes whose deletion could improve 3HP production, in silico The DPA producing B. subtilis MGB874 strain was aerobi- knockout simulation using genome-scale metabolic model of cally cultured on synthetic media which containing glucose E. coli iAF1260 was performed. As the result, double knock- and glutamate as carbon sources. After the cell growth out of tpiA-zwf was predicted to enhance the 3HP yield on stopped, the 13C-labeled glucose was added in the culture, glycerol. In consistent with the simulation result, the tpiA-zwf and the 13C enrichments of intracellular metabolites were knockout strain constructed showed higher 3HP yield on measured by gas chromatography–mass spectrometry and the consumed glycerol (0.20 mol/mol) than that of the 3HP capillary electrophoresis time–of–flight mass spectrometry. strain (0.046 mol/mol). However, the tpiA-zwf knockout strain The flux distribution was optimized to minimize the differ- produced high yield of 1,3-propanediol, thus, yqhD encod- ence between the experimentally measured 13C-enriching an alcohol dehydrogenase related to the 1,3-propanediol ments of metabolites and those calculated from the flux biosynthesis was deleted in the tpiA-zwf knockout strain. The distribution. The estimated flux distribution revealed that 3HP yield of the tpiA-zwf-yqhD knockout strain increased the uptake glucose was mostly excreted as acetoin and the 7.4-fold of the 3HP strain (0.34 mol/mol). large part of DPA was derived from glutamate. We conducted the rational metabolic design based on the 13C-metabolic 203. Improving Dipicolinic Acid Production By flux distribution, and successfully improved the productivity Bacillus Subtilis during Stationary Phase Based on of DPA during the stationary phase. 13c-Metabolic Flux Analysis Yoshihiro Toya*1,2, Takashi Hirasawa1,2,3, Kenta Masuda2,4, 204. Construction of E. coli Reporter Strains for the Takuya Morimoto2,4, Yasushi Kageyama2,4, Katsuya Ozaki2,4, Study of Stress-Answer Regulator Proteins 1Graduate School of Information Science and Technology, Natalia Trachtmann* and Georg A Sprenger Osaka University, Suita, Japan Institute of Microbiology, University of Stuttgart, Stuttgart, 2Advanced Low Carbon Technology Research and Develop- Germany ment Program, Japan Science and Technology Agency Escherichia coli K12 is a model organism in microbiology (JST, ALCA) and genetics. It is currently also used for production of 3Department of Bioengineering, Tokyo Institute of Technol- ogy, Japan valuable compounds such as aminoacids, vitamins, or fine 4Biological Science Laboratories, Kao Corporation, Japan chemicals. During bioreactor cultivations, however, the cells

109 Poster Abstracts

often recognice varying conditions of aeration (aerobiosis, target products with different genetic modifications. An microaerophilic or anaerobic), of carbon source supply (glu- engineered host that is optimized to produce one target cose, glycerol, acetate), or nitrogen. To study the response product may not be suitable to function as an optimal host of E. colicells to these varying conditions, we have con- to efficiently produce other target compounds, which makes structed defined reporter strains. the conventional approach laborious and expensive. To First, knock-out strains for major regulator proteins of C-, N- address this bottleneck, we are developing a novel MOD- and O- regulation (GlnK, GlnB, ArcA, Fnr, OxyR, Mlc, Crp, CEL (Modular Cell) toolbox based on the metabolic network Crr, Cra, RpoS, RpoN)1 have been provided by the method modeling to design modular cells that can metabolically of Datsenko & Wanner2 and were characterized. Next, we couple with a diverse class of chemicals- and biofuels- compared various reporter genes (lacZ encoding ß-galacto- producing pathways as exchangeable production modules sidase, gus for a ß-glucuronidase3, gfp, or bfp for an oxy- to efficiently produce a diverse combinatorial set of chemi- gen-independent blue fluorescent protein from a metage- cals and biofuels. We will present the design, construction, nomic source4) for their applicability as C-; N-, O- monitors. and characterization of optimal modular E. coli cell factories We took care to have universal and stable reporter systems. assembled from an optimal modular cell and exchangeable Thus, apart from plasmid-borne reporter genes, we also are production modules in a rapid plug-and-play fashion to pro- working on the chromosomal integration to avoid genetic duce a diverse combinatorial spectrum of bioesters that can instabilities and the use of antibiotic resistance markers. To be used as fragrances, flavors, solvents, and biodiesels. assay the applicability of the reporter systems, we fused well-known promoter regions which are under the control of 206. Design and Construction of an Artificial Nonmeval- a given regulator (C-, N- or O-regulons) and integrated them onate Operon of Escherichia coli Kenji Tsuge*, Takashi Togashi, Masako Hasebe, Masaru as single copy into the chromosomal DNA of E.coliwild-type Tomita and Mitsuhiro Itaya strain W3110 and of the respective regulator mutant strains. Institute for Advanced Biosciences, Keio University, Tsuruoka, Japan Reporter strength was then measured by standard enzyme assays for the chromogenic substrates (ONPG for LacZ, So far it has been possible to construct bacterial genome PNPG for Gus) or for fluorescence (GFP, BFP) using a laser from chemically synthesis DNAs, but no example of de- fluorescence microscopy and fluorophotometric measure- signer genome emerged, due to non strategy for design ments. Results will be presented and discussed. genome. Toward creation of a novel genome, we should References: start to construct a readily simplified genome in terms of [1] A. Martinez-Antonio (2011) Network Biology 1(1):21-33 engineering that is far apart from the naturally occurring bacterial genome because it is too complicated. In a bac- [2] K.A. Datsenko, B.L. Wanner (2000) Proc.Natl.Acad.Sci. USA 97, 6640-45. terium with the simplified genome, it can be regarded as an assembly of metabolic pathways. To design each metabolic [3] C. Platteeuw al. (1994) Applied and Environmental Microbiology, 587-593 pathway, certain systems for balancing quantity of multiple kinds of enzymes are required. We have been proposed [4] Chang-Sun Hwang et al. (2012) Biochemical and Biophysical Research Communications Vol 419, 676-681 that polycistronic operon architecture that is well observed in bacterial genome contains hints for the genome designing. For a experimental demonstration for construction of a 205. Rational Design of Modular Cells for Efficient metabolic pathway by a designed polycistronic operon, 9 of Combinatorial Biosynthesis of Designer Bioesters Escherichia coli genes for nonmevalonate pathway, one of Cong T. Trinh*1, Yan Liu1 and Donovan Layton2 1Department of Chemical and Biomolecular Engineering, the essential metabolic pathways, were rendered. Accord- University of Tennessee, Knoxville, Knoxville, TN ing to the possible operon rule, the genes were connected 2University of Tennessee, Knoxville under a sole promoter from highest to lowest of the mRNA abundance in the wild type cell in order by OGAB assembly. In the conventional strain engineering approach, relevant This prototype operon plasmid was introduced into wild hosts such as Escherichia coli and Saccharomyces cereve- type E. coli cell to knock out relevant nonmevalonate genes siae are commonly engineered and optimized to produce in the genome, but two of the relevant genes weren’t able to

110 Poster Abstracts knock out relevant nonmevalonate genes in the genome, 208. Model Driven Mechanistic Analysis of Adaptive but two of the relevant genes weren’t able to remove from RNAP Mutations 1 1 1 the genome. In case of ispE, the ORF contains a promoter Jose Utrilla* , Edward O’Brien , Douglas McCloskey , Adam M. Feist1,2 and Bernard O. Palsson1,2 sequence for essential gene located downstream. In case of 1Department of Bioengineering, University of California, San dxr, expression level is lower than expected in the operon. Diego, La Jolla, CA By taking account of these facts, finally we could construct 2Novo Nordisk Foundation Center for Biosustainability, Tech- new operon that doesn’t require any of nonmevalonate nical University of Denmark, Lyngby, Denmark genes in the genome. Fitness increase by adaptive RNA polymerase mutations 207. Lactic Acid Production from Xylose By Engineered have been reported for different environments. The effects Saccharomyces cerevisiae of these mutations in molecular phenotype have been de- Timothy L. Turner*1, Guochang Zhang1, Vijay Subramaniam2, tailed but the mechanisms that underlie the network repro- David Steffen2 and Yong-Su Jin1 1Food Science and Human Nutrition, University of Illinois at gramming have not been decoded. In this work two RNAP Urbana-Champaign, Urbana, IL mutations obtained from a glucose minimal media Adaptive 2Molecular and Cellular Biology, University of Illinois at Laboratory Evolution experiment were reintroduced in wild Urbana-Champaign, Urbana, IL type Escherichia coli. Using genome wide data across different conditions and a genome scale model of metabolism As global climate change concerns escalate, development and gene expression it was possible to unravel the mecha- of carbon-neutral methods to produce industrially-relevant nism of network reprogramming by a single amino acid chemicals from non-food, non-fossil fuel sources has be- change in the transcription machinery. Our results show that come increasingly important. Lactic acid, primarily used for suboptimal phenotype is caused by the up regulation of the synthesis of the biodegradable polylactic acid (PLA) or as a non-growth functions, specifically during growth on batch feedstock for 3D printing, is industrially-produced by microbial minimal media. The down regulation of non-growth func- fermentation using glucose as a substrate. As an alternative tions by the introduced mutations reduces the proteome to glucose, xylose which is easily obtainable from lignocel- and the maintenance energy needs enabling higher cell lulosic biomass under mild pretreatment conditions without yield and growth rate. Structural and regulatory insights of excessive use of cellulase enzymes can be utilized for produc- mechanism of global network reprogramming are presented ing chemicals. In order to produce lactic acid from xylose, and their physiological implications for metabolic engineer- we introduced both a xylose-utilizing pathway from Schef- ing highlighted. fersomyces stipitis and lactate dehydrogenase from Rhizo- pus oryzae into Saccharomyces cerevisiae. It is well-known 209. Carbon Dioxide Fixation By Calvin-Cycle Enzymes that deletion of genes (PDC1, PDC5, and PDC6) coding for Improves Ethanol Yield in Yeast pyruvate decarboxylase was necessary for producing lactic Antonius J.A. van Maris* Department of Biotechnology, Delft University of Technology, acid without ethanol production from glucose. However, the Delft, The Netherlands xylose-utilizing engineered S. cerevisiae was able to produce lactic acid with trace amounts of ethanol production even Background: Redox-cofactor balancing constrains product though PDC1, PDC5, and PDC6 were not deleted. As a result, yields in anaerobic fermentation processes. This challenge the engineered strain produced lactic acid at high titers from is exemplified by the formation of glycerol as major by- xylose between pH 4.0 and 5.0, in either minimal or complex product in yeast-based bioethanol production, which is a media. Notably, the engineered S. cerevisiae produced negli- direct consequence of the need to reoxidize excess NADH gible amounts of xylitol and exhibited high yields of lactic acid and causes a loss of conversion efficiency. Enabling the use production from xylose (up to 0.7 g lactate/g xylose). These of CO as electron acceptor for NADH oxidation in hetero- results suggest that further studies involving scale-up of these 2 trophic microorganisms would increase product yields in results and improvement of tolerance will allow for usage of industrial biotechnology. renewable lignocellulosic feedstocks for efficient and sustainable production of lactic acid. Results: A hitherto unexplored strategy to address this redox challenge is the functional expression in yeast of en-

111 Poster Abstracts

zymes from autotrophs, thereby enabling the use of CO2 as pathway have met with only limited success. In part this electron acceptor for NADH reoxidation. Functional expres- reflects our lack of knowledge concerning the regulation sion of the Calvin cycle enzymes phosphoribulokinase (PRK) of MEP pathway. This is unfortunately, because the MEP and ribulose-1,5-bisphosphate carboxylase (Rubisco) in pathway seems energetically more efficient and balanced in Saccharomyces cerevisiae led to a 90% reduction of the by- the use of reduction equivalents. product glycerol and a 10% increase in ethanol production in sugar-limited chemostat cultures on a mixture of glucose A combinatorial approach of metabolic engineering and and galactose. Co-expression of the Escherichia coli chap- metabolic flux analysis was used for the production of the erones GroEL and GroES was key to successful expression economical relevant isoprenoids isoprene and farnesene in of CbbM, a form-II Rubisco from the chemolithoautotrophic E. coli. Metabolic flux analysis was used for the identifica- bacterium Thiobacillus denitrificans in yeast. tion of target genes, which influence the flux through the MEP pathway. These genes were then further assessed with Conclusions: Our results demonstrate functional expression metabolic control analysis for their quantitative properties. of Rubisco in a heterotrophic eukaryote and demonstrate Targeted proteomics enabled us to determine the enzyme

how incorporation of CO2 as a co-substrate in metabolic en- concentrations and pools of pathway intermediates were gineering of heterotrophic industrial microorganisms can be measured through LC-MS/MS experiments. This allowed us used to improve product yields. Rapid advances in molecular to determine control coefficients and kinetic values for the biology should allow for rapid insertion of this 4-gene expres- pathway and integrate them in computational models. The sion cassette in industrial yeast strains to improve production, acquired information was then used for rational engineering not only of 1st and 2nd generation ethanol production, but of the MEP pathway in E. coli. also of other renewable fuels or chemicals. 211. Comprehensive Evaluation of Two Genome-Scale 210. Rational Engineering of the Methylerythritol Phos- Metabolic Network Models of Scheffersomyces Stipitis phate Pathway through Metabolic Control Analysis Andrew Damiani1, Q. Peter He1 and Jin Wang*2 Daniel Volke*1, Benedikt Engels1, Louwrance Wright², 1Chemical Engineering, Tuskegee University, Tuskegee, AL Jonathan Gershenzon², Stefan Jennewein1 2Auburn University, Auburn, AL 1Fraunhofer Institute for Molecular Biology and Applied Ecology, Aachen, Germany Due to its important role in the production of lignocellulosic ²Max Planck Institute for Chemical Ecology, Jena, Germany ethanol, a second generation biofuel, Scheffersomyces stipitis has drawn tremendous research interest. This is Facing an increasing demand of sustainable energy and because S. stipitis has the highest native capacity to fer- chemicals, new strategies have to be explored to synthesis ment xylose into ethanol and how to efficiently utilize the these in low-cost and environmental friendly processes. pentose sugars in lignocellulosic hydrolysate remains one Isoprenoids are a promising source for a broad spectrum of of the major barriers in producing cost-effective lignocel- valuable molecules, including fine and bulk chemicals, fuels, lulosic ethanol through biological conversion. Although S. and pharmaceuticals. stipitis can produce ethanol at a yield close to the theoretical maximum under microaerobic condition, its xylose Isoprenoids are the largest as well as most diverse class of consumption rate is only half of that under fully respiratory chemical molecules synthesized by the natural world. All conditions. In addition, low ethanol tolerance and no growth isoprenoids are derived from two common precursors, iso- under anaerobic condition also limit its direct utilization in pentenyl phosphate and dimethylallyl phosphate, which are industrial applications. To address these limitations, at- synthesis via two distinct pathways. The long known meva- tempts have been made to improve xylose pick up rate of S. lonate (MVA) pathway uses acetyl-CoA as substrate, while stipitis under microaerobic condition, as well as introducing the recently discovered methylerythritol (MEP) pathway uses xylose fermentation pathways possessed by S. stipitis to glycerinaldehyde-3-phosphate and pyruvate. engineered Saccharomyces cerevisiae. However, these attempts have not been entirely successful: for engineered S. The MVA pathway has been engineered to produce large stipitis strains, they still suffer from low xylose consumption quantities of terpenoids, but similar attempts using the MEP rates, lower ethanol tolerance as well as incapable to grow

112 Poster Abstracts

anaerobically; for engineered S. cerevisiae mutants, in gen- network models in a systematic way. eral, the expression of target genes from S. stipitis has not Recently, two genome-scale models have been published: been very effective due to different regulatory mechanisms iSS8441 and iBB8142. iSS844 was constructed semi-au- in S. stipitis (Crabtree-negative) and S. cerevisiae (Crabtree- tomatically by integrating automatic reconstruction with S. positive), as well as the redox imbalance introduced by the cerevisiae as a reference framework and manual curation xylose fermentation pathway under microaerobic condition. and modification. iBB814 was constructed manually by following a published protocol for generating a high-quality Because of the complexity involved in the two yeast strains, genome-scale metabolic reconstruction. iSS844 includes a better understanding on cellular metabolism of S. stipitis 1332 reactions, 922 metabolites and 4 compartments; would be helpful for improving both S. stipitis and engi- iBB814 includes 1371 reactions, 644 metabolites and 3 neered S. cerevisiae. Specifically, genome-level understand- compartments. Both models have fair amount of experi- ing of xylose metabolism would provide valuable insights on mental validations and both performed well in matching identifying effective strategies for design of the engineered the model predictions with some experimental measure- mutant strains. The complete genome of S. stipitis has been ments. However, after careful examination, we found that sequenced, which provides the foundation for genome- both models have certain limitations. For example, neither scale metabolic network reconstruction. Because genome- of them predicts the production of xylitol under any condi- scale metabolic models connect the genotype with pheno- tion, aerobic, microaerobic or anaerobic. However, we know type, they provide a holistic view of the cellular metabolism. that xylitol is a key byproduct produced by S. Stipitis under Once validated, genome-scale models provide a platform to microaerobic condition due the redox imbalance caused by effectively interrogate cellular metabolism, such as charac- xylose fermentation step (XR and XDH). Therefore, being terizing metabolic resource allocation, predicting phenotype, able to predict xylitol production under microaerobic condi- designing experiments to verify model predictions, as well tion is very important for a model to identify candidates for as designing mutant strains with desired properties. More balancing cellular redox potential. importantly, genome-scale models allow systematic assessment of how a (genetic or environmental) perturbation would In this work, we conduct comprehensive evaluations of the affect the organism as a whole. two genome-scale models mentioned above, with the aim to identify a model that agrees better with existing bio- For metabolic network models, it is clear that the quality logical knowledge to serve as the basis for further model of the model determines the outcome of the application. improvement. The evaluation consists of three steps: first Therefore, it is critically important to determine how accurate is the manual comparison of the reactions included in both a metabolic network model is, particularly a genome-scale models; second is the conduction of in silico experiments to model, in describing the microorganism’s cellular metabolism. examine model’s global behavior through phenotype phase Currently, model validation is done mainly through wet lab plane analysis, and to compare the model predictions with experiments, i.e., comparing model predictions with experi- additional experimental results reported in literature; Final mental measurements. Due to the cost and other difficulties step is the application of the proposed system identification associated with getting intracellular measurements, most framework to extract biological knowledge from carefully experimental measurements are limited to cross membrane designed in silico experiments, and to examine whether the fluxes such as substrate pickup rates, production secretion extracted knowledge agrees with existing understandings. rates, and cell growth rate. However, due to the scale and complexity involved in genome-scale models, a good agree- From manual examination, we found that the two models ment between measured and computed cross-membrane have large discrepancies, as there are very limited number fluxes does not necessarily indicate that the model quality is of overlapping reactions from both models. In addition, we high. Therefore, new methods are needed to extract qualita- identified several key errors that each model contain. But tive biological knowledge embedded in the quantitative in overall, iBB814 agrees better with existing biological knowl- silico experiments based on the metabolic network models in edge on S. Stipitis. However, through additional experimen- order to evaluate the model quality. In this work, we present a tal validation, iSS884 performs noticeably better compared system identification based framework to examine metabolic to iSS814, even though the global examination through in

113 Poster Abstracts

silico experiments reveals some unexpected behavior of functions for their host, including anoxic respiration, deni- iSS884. Finally, with the system identification based frame- trification of waste products and production of secondary work, we obtain information on how the perturbation of metabolites that often attract pharmacological interest. increased oxygen pickup rate under microaerobic condition Indeed sponges and their associated bacteria are the most propagates through the whole network, and concluded that prolific producers of bioactive compounds in the marine en- iBB814 yielded biological information with better accuracy. vironment, with over 5000 compounds identified to date. Yet In addition, through the framework, a futile cycle was identi- despite this enormous potential, only one compound has fied in the nucleotides subsystem of iSS884, which was been approved for therapeutic use. The lack of commercial used to provide required energy (ATP supply) for cell growth. development of sponge-derived compounds is attributed When the futile cycle is eliminated from the model, the to a biomass supply problem: no reliable method to culture model fails to predict any cell growth. sponges exists, either at the scale of the whole organism or as a cell culture. In addition, the current lack of understand- In conclusion, iBB814 is the better model because it aligns ing of the sponge microbiome interaction limits the appli- better with existing biological knowledge. In addition, we cability of metabolic engineering approaches to microbial demonstrated that the proposed system identification based isolates or sponge cells for over-production of secondary framework is a powerful tool for extracting biological in- metabolites. formation embedded in the quantitative simulation results produced by the metabolic network models. This framework To better understand the metabolic processes underlying bridges the gap between the complicated numerical results sponge growth, we want to apply genome scale modeling generated from genome-scale models and qualitative biologi- and flux analysis to the marine demosponge, Amphim- cal knowledge that can be easily understood by biologists. edon queenslandica. As a first step, we have characterized References: the biochemical composition of A. queenslandica and the associated bacteria, essential information for metabolic 1. Caspeta, Luis, Saeed Shoaie, Rasmus Agren, Intawat Nookaew, and Jens Nielsen. “Genome-scale metabolic reconstructions of modeling. We have complemented these methods with Pichia stipitis and Pichia pastoris and in silico evaluation of their comprehensive, high resolution transcriptional profiling of potentials.” BMC systems biology 6, no. 1 (2012): 24. developmental changes in gene expression throughout the 2. Balagurunathan, Balaji, Sudhakar Jonnalagadda, Lily Tan, sponge life cycle. This integrative approach will give the first and Rajagopalan Srinivasan. “Reconstruction and analysis of a detailed insights into sponge biomass production and will genome-scale metabolic model for Scheffersomyces stipitis.” lay the basis for the metabolic reconstruction of a sponge Microb Cell Fact 11, no. 1 (2012): 27. / microbiome metamodel. This work will quantify carbon fluxes within a sponge holobiont for the first time. 212. Genome Scale Metabolic Modeling Reveals New Insights into Biomass Production in the Marine Sponge 213. Quorum-Sensing Linked RNAi for Dynamic Pathway Amphimedon Queenslandica Control in Saccharomyces cerevisiae 1 2 Jabin Watson* , Robin Palfreyman , Selene Fernandez-Val- Thomas C. Williams*1, Nils J. H. Averesch2, Nicolas 1 2 2 1 verde , Timothy Brennan , Lars K. Nielsen , Bernie Degnan , Lekieffre2, Gal Winter2, Claudia E. Vickers1, Lars K, Nielsen1 1 3 Sandie Degnan and Jens O. Krömer and Jens O. Krömer2 1 School of Biological Science, The University of Queensland, 1Australian Institute for Bioengineering and Nanotechnology, Brisbane, Australia The University of Queensland, St Lucia, Australia 2 Australian Institute for Bioengineering and Nanotechnology, 2Centre for Microbial Electrosynthesis (CEMES) within the The University of Queensland, Brisbane, Australia Advanced Water Management Centre (AWMC), The Univer- 3 Centre for Microbial Electrosynthesis (CEMES) within the sity of Queensland, St Lucia, Australia Advanced Water Management Centre (AWMC), The Univer- sity of Queensland, St Lucia, Australia Many metabolic engineering strategies involve the deletion or over-expression of genes which subsequently reduce or Marine sponges are among the simplest multicellular ani- eliminate biomass formation. However, in order to achieve mals, but under this simple façade exists complex meta- high compound titers a production strain must be able to bolic crosstalk between the host sponge and its symbiotic grow to a high population density by accumulating biomass. bacteria. Sponge bacterial symbionts perform diverse

114 Poster Abstracts

This means that the most productive genetic modifications maceuticals. Due to their sophisticated structures, chemical cannot be implemented unless population-growth and synthesis of these compounds is often difficult, necessitat- production phases are separated. To achieve this separa- ing biological production platforms. Metabolic engineering tion it is necessary to dynamically control the expression can be used to improve product yields in both whole plant of relevant genes so that a population can grow free of and cell culture systems. However, due to the complex metabolic burden to a high density before switching to interactions between plant primary and secondary me- production mode. We enabled dynamic regulation by using tabolism and within cooperating and competing secondary a synthetic quorum sensing circuit in S. cerevisiae which pathways, it is difficult to predict the effect of single gene utilises pheromone mediated cell-to-cell communication to overexpression or silencing on overall cellular metabo- autonomously trigger gene expression at a high population lism. To begin to understand these complex relationships, density. Dynamic silencing of gene expression was enabled we developed assays to map flux through key pathways by combining the quorum sensing circuit with an RNA in- to general classes of secondary products (i.e., phenolics, terference (RNAi) module which targets mRNA for degrada- flavonoids, alkaloids), as well as relevant endpoint products tion according to base-pair complementarity of expressed (lignin and taxanes). Using the Taxus suspension culture RNA hairpins. As a demonstration we used the quorum system that uniquely accumulates the anti-cancer drug sensing-RNAi circuit to control flux through the shikimate paclitaxel, we studied distribution of flux through secondary pathway for the production of para-hydroxybenzoic acid metabolic pathways upon induction with the abiotic elicitor (pHBA), a valuable aromatic chemical used in liquid crystal methyl jasmonate. Results demonstrate that individual cell polymers. Quorum sensing triggered a production phase lines respond variably to methyl jasmonate by upregulat- with the expression of a chorismate lyase for pHBA forma- ing different secondary metabolic pathways. In addition, tion (UBiC gene), feedback resistant 3-deoxy-D-arabino- we studied the heterogeneity in secondary metabolic flux heptulosonate-7-phosphate synthase (ARO4 gene), and distribution over multiple generations of growth, identify- transketolase (TKL1 gene) for increased precursor supply. ing key competing pathways for paclitaxel accumulation in This dynamic regulation also allowed the implementation of each cell line. These studies are amongst the first to study engineering strategies identified by elementary flux mode global secondary metabolism in a non-model plant spe- analysis as highly productive but inhibitory of biomass cies and provide valuable insight into the design of effective formation, after a population growth phase. To demonstrate metabolic engineering strategies to promote production of a this capacity we conditionally silenced the expression of particular class of secondary products. ARO7, TRP3, and ZWF1 genes which improve pHBA yield, but severely limit growth if deleted. We also silenced a 215. Optimization of Multi-Gene Biological Systems pyruvate kinase gene (CDC19) which when deleted with Using High-Throughput DNA Assembly, Sequencing, and Model-Guided Search Strategies traditional methods is not only inhibitory to growth, but is Lauren B. A. Woodruff*1,2, Tarjei Mikkelsen1, D. Benjamin lethal. This approach facilitated the attainment of the most Gordon1, Michael J. Smanski2, Christopher A. Voigt2 and highly productive flux modes after the population had grown Robert Nicol1 to a high density such that lethal gene silencing did not limit 1Broad Technology Labs, Broad Institute of MIT and biomass formation and therefore titer. Using simple shake- Harvard, Cambridge, MA 2 flask fermentations, this approach resulted in the highest Synthetic Biology Center, Department of Biological Engineering, Massachusetts Institute of Technology, recorded pHBA titer in yeast of 166 mg/L. Cambridge, MA

214. Balancing Flux through Secondary Metabolic Pathways in Plant Culture Systems We have developed a framework to modularize and re-opti- Sarah Wilson*1 and Susan Roberts2 mize massively multi-gene pathways or biological systems 1Chemical Engineering, University Of Massachusetts that increases the size and complexity of genetic systems Amherst, Amherst, MA that can be engineered. Engineering and optimizing the per- 2 University of Massachusetts, Amherst formance of large multi-gene systems requires combinatori- ally tuning gene expression through many design iterations. Plants are a valuable source of chemical compounds with Here, new high-throughput DNA assembly strategies were applications ranging from fragrances and flavors to phar- developed that leverage advances in DNA sequencing to 115 Poster Abstracts

speed this process and also enable statistical combinatorial However, the infrastructure to use natural gas directly (either design for targeted searches through the high-dimensional as CNG or LNG) has not been developed, and so natural space of possible genetic designs. We demonstrated these gas must first be converted to a liquid fuel. Chemical pro- strategies developed here by optimizing a pathway of 16 cesses for accomplishing this suffer from high capital costs genes. This 16-gene nitrogen fixation (nif) gene cluster is and low yields, thus there is significant interest in developing entirely modular and synthetic (“refactored”) with ~100 alternative processes for the high-efficiency conversion of parts. The synthetic nif pathway encodes 16 essential genes methane to a drop-in transportation fuel. Various microor- that have tightly coordinated action to convert atmospheric ganisms have evolved enzyme systems capable of metabo- nitrogen into ammonia, a critical process for global agri- lizing methane, thus with the application of metabolic engi- culture productivity. Using our novel assembly strategy, we neering technologies, it is conceivable that a strain capable constructed and sequenced a library of >55,000 unique nif of producing a liquid fuel from methane could be developed. transcriptional units (TUs). This is a 100-1,000-fold improve- Before embarking on such an undertaking, however, it is ment in library size over our and comparable previous critical to establish the inherent potential and limitations of studies. Each TU is composed of characterized genetic the chosen pathway, to maximize the likelihood of selecting parts (5-7 each), including one refactored nif gene. These a commercially viable process, and not just a novel scientific nif TU modules can be directly assembled into whole 16- investigation. To that end, we present here an analysis of gene, 100-part permuted cluster designs (~1045 unique the various biological routes for the conversion of methane cluster designs possible from this library). Additionally in this to a liquid fuel. We develop stoichiometric models for each study, we sequenced approximately 7.5 million nif TUs to scenario that allow calculation of the maximum potential characterize DNA assembly error modes and quantify their yields, and conclude with the implication of these results frequencies. Next, from the library of nif modules we built in developing an organism and pathway for the biological and tested systematically permuted refactored clusters to conversion of methane to a liquid fuel. investigate optimizing nitrogen fixation activity by tuning expression within the refactored cluster using character- 217. High-Efficiency Scarless Genetic Modification ized biological parts. Combinatorial design and statistical Method in Escherichia coli without Counterselection Junjie Yang*1, Bingbing Sun1, He Huang1, Yu Jiang1, Liuyang design of experiments were applied to constrain the part Diao2, Biao Chen1, Chongmao Xu1, Xin Wang1, Jinle Liu1, substitutions integrated into the cluster designs fabricated Weihong Jiang1 and Sheng Yang*1 and tested in each round. Together, these methods enable 1Key Laboratory of Synthetic Biology, Institute of Plant model-guided redesign and optimization of genetic system Physiology and Ecology, Shanghai Institutes for Biological performance with minimized DNA assembly required (~10- Sciences, Chinese Academy of Sciences, Shanghai, China 2 fold reduction in cloning). Looking ahead, we believe the CAS Key Laboratory of Synthetic Biology, Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological assembly and optimization strategies demonstrated here Sciences, Chinese Academy of Sciences, Shanghai, China will accelerate transferring complex functions to engineered systems and dramatically expand the complexity of biologi- Genetic modifications of bacteria chromosomes are impor- cal systems that can be written, engineered, and optimized. tant for both cognitive and applied research.

216. Evaluation of Biosynthetic Pathways for Conversion of Natural Gas to Liquid Fuels In this study, a new, two-plasmid method for unmarked Benjamin Woolston*, David Emerson and Greg genetic modification of E. coli was established. The method Stephanopoulos uses λ-Red recombination and I-SceI cleavage. E. coli Chemical Engineering, MIT, Cambridge, MA endogenous sequences were used as substrates for recombination and no additional segment was introduced into It has been estimated that the proven natural gas reserves in genome. The new method was successfully used for gene the U.S. could provide enough energy to meet the country’s deletion and insertion, seamless gene deletion, site-directed transportation energy demand for the next 50 years. This, mutagenesis, and gene replacement. The new method could coupled with the increasing price gap between natural gas be used to modify both nonessential and essential genes. and gasoline over recent years, makes natural-gas derived fuels an attractive option for use in the transportation sector.

116 Poster Abstracts

218. Metabolic Engineering of a Biosensor-Based Optimizing production of a specific chemical usually in- Screening Platform in Yeast volves increasing synthesis of the precursors of the chemi- 1 1 1,2,3,4,5 Jie Zhang* , Michael K. Jensen and Jay D. Keasling cal in bacteria. Pyruvate is a starting compound for syn- 1The Novo Nordisk Foundation Center for Sustainability, thesizing a variety of biofuels and chemicals. A high-yield Technical University of Denmark, Denmark 2Joint BioEnergy Institute, Emeryville, CA pyruvate producing strain has great potential for creating 3Synthetic Biology Division, Physical Biosciences Division, strains to produce a variety of valuable chemicals. Guided Lawrence Berkeley National Laboratory, Berkeley, CA by a genome-scale metabolic model of Escherichia coli, 4Department of Chemical & Biomolecular Engineering, Uni- we identified different strategies for enhancing the produc- versity of California, Berkeley, Berkeley, CA tion of pyruvate from glucose. The targeted gene deletions 5Department of Bioengineering, University of California, minimize acetyl-CoA production, undesired product (acetate Berkeley, Berkeley, CA and lactate) formation, and NAD(P)H formation. We con- Since its emergence in the 1990s, metabolic engineering structed a number of strains and six of them achieved yields has mainly focused on the optimization of certain products of more than 0.88 g pyruvate per g of glucose (90% theo- that can be converted from biomass feedstock and quickly retical yield) under aerobic conditions. evolved as an enabling technology for a biosustainable society. Besides, it largely expands the spectrum of prod- Pyruvate is a precursor for synthesis of ethanol. To produce ucts by introducing synthetic metabolic routes into a host ethanol, pyruvate formate-lyase (PflB) was deleted, and cell. Along with the recent advances in synthetic biology, it pyruvate decarboxylase (Pdc) and alcohol dehydrogenase becomes significantly easier and cheaper to design and as- II (AdhB) from Zymomonas mobilis were over-expressed in semble regulatory circuits or even complex networks using the engineered pyruvate strains. These genetically modified synthetic parts from a library. Given the diversity of these strains fermented glucose to ethanol with a yield of 0.42 g biochemical reactions, the increasing availability of genome ethanol per g of glucose (~80% of theoretical yield). sequences as well as the part libraries being constructed, the capacity of such a combinatorial approach for creat- In addition to producing native metabolites in E. coli, pyru- ing a collection of engineered strains is virtually unlimited. vate can be used to make non-native metabolites when the However, this imposes a bottleneck on the screening of necessary enzymes are expressed in E. coli to enable their the desired characteristics in these synthetic constructs. synthesis. For this purpose, we computationally identified Here we present our initial work on the development of a what non-native metabolites could be made from pyruvate biosensor-based screening for yeast strains with optimized and which exogenous reactions need to be added. The supplies of metabolites of interest. The target metabolite developed high-producing pyruvate strains can be subse- can be a general precursor like acetyl-CoA, or a cofactor quently adapted to generate strains capable of producing that is required for proper cell function and must be bal- other important chemicals. anced, such as NADPH. Considering that these key intracel- 220. Hy-Dynfba: A Software Platform to Build Large lular metabolites also participate in the production of many Models, Hybrid in Time Scale compounds with vast economic interest, a biosensor as Piotr Zakrzewski and Emrah Nikerel*1 such sees great potentials in the metabolic engineering of 1Department of Genetics and Bioinformatics, Bahcesehir yeast cell factories. University, Istanbul, Turkey

219. Metabolic Engineering for Production of Valuable Large scale dynamic simulation of (microbial) metabolism Chemicals Based on Escherichia coli Strains Designed is of great interest within systems biology and metabolic in silico engineering community, in several aspects, e.g. to obtain Xiaolin Zhang*1 and Jennifer L. Reed2,3 1Cellular and Molecular Biology Graduate Program, Univer- time traces of intermediates, to study regulation etc. This in sity of Wisconsin Madison, WI turn, is one of the big challenges since such an endeavor 2Department of Chemical and Biological Engineering, Uni- requires a large amount of detail about the metabolic reac- versity of Wisconsin-Madison, Madison, WI tion networks and is hampered by e.g. the presence of 3 DOE Great Lakes Bioenergy Research Center, University of multiple time-scales (from sub seconds e.g. covalent modifi- Wisconsin-Madison, Madison, WI cations, to hours e.g. protein synthesis), lack of knowledge

117 Poster Abstracts

in mechanisms of individual reactions, correlated kinetic the optimal combination of the three enzymes was selected parameters, lack of data to specify the parameters etc. for the maximal production of 3-HB. This is the first report of producing enantiomerically pure (S)-HB using S. cerevisiae. One of the approaches for genome scale modeling of microbial metabolism, namely flux balance analysis, enjoyed the 222. Characterization of High Ethanol Producing simplification on time scale differences, and has proven to Properties of Recombinant Saccharomyces cerevisiae ETS3 Transformed with a Mutated SPT15 Gene be very useful, supported by several applications in litera- Haeseong Park*1, Yeong-Je Seong1, Kyoung Heon Kim2 and ture. Kinetic modeling, in contrast, despite much greater Yong-Cheol Park1 information it provides and a longer research history, is at 1Department of Bio and Fermentation Convergence its infancy with respect to the size of the models, chiefly Technology, Kookmin university, Seoul, South Korea because of the above mentioned challenges. Modeling 2Department of Biotechnology, Korea University Graduate strategies that would be high in coverage, yet manageable School, Seoul, South Korea in complexity is therefore of immediate interest. Resultingly, Osmotic stress caused by high sugar and ethanol concentra- tools that would systematically integrate (preferably ge- tions is one of the major reasons for reduction of bioethanol nome scale) flux models with much smaller kinetic models, productivity during yeast fermentation. To overcome this whereby integrate multiple time scales is desired. hurdle, global transcription machinery engineering (gTME) by introducing SPT15 variants encoding S. cerevisiae TATA-bind- In this work, we present such a tool for simulation of large ing proteins was applied and S. cerevisiae ETS3 was selected scale systems, making use of available software, built upon to possess enhanced ethanol producing ability under high reliable stack of open source python libraries. Hy-DynFBA osmotic pressure. In this study, its fermentation properties software uses COBRApie as an extension to construct, were evaluated in batch and fed-batch cultures using various manipulate and import/export flux models. The software glucose concentration under different aeration conditions. A then allows selecting desired kinetics with corresponding batch culture of ETS3 with 300 g/L glucose at oxygen-limited parameters, for a subset of reactions, assigns convenience condition resulted in 22.2 g/L dry cell mass, 98.1 g/L ethanol kinetics by default if the mechanism of the selected reac- concentration and 1.82 g/L-h ethanol productivity which were tions are not known. The talk will introduce the concept, about 13–16% increases relative to those of the S. cerevi- theoretical background, equations and will further focus on siae BY4741 host strain. A micro-aerobic condition elevated implementations on modeling yeast (Saccharomyces cerevi- ethanol productivity to 2.49 g/L-h without significant changes siae) response to weak acid stress. of other fermentation parameters. In a fed-batch fermentation 221. Production of Enantiomerically Pure (S)-3- with 400 g/L glucose, ETS3 produced up to 151.7 g/L ethanol Hydroxybutyrate Using Metabolically Engineered with a maximum ethanol yield of 0.402 g/g and showed a Saccharomyces cerevisiae 23% increase in dry cell mass and 16% increases in ethanol Eun Ju Yun* and Kyoung Heon Kim concentration and productivity in comparison to the host Department of Biotechnology, Korea University Graduate strain. Considering the enhanced ethanol production perfor- School, Seoul, South Korea mances of ETS3 regardless of oxygen content, it is obvious that gTME is an effective tool to gain a robust S. cerevisiae (S)-3-Hydroxybutyrate can be used as a precursor to syn- strain against high osmotic pressure. thesize the biodegradable plastics, polyhydroxyalkalonoates (PHAs) and stereo-specific fine chemicals, such as antibiot- 223. Comparative Metabolomic Study of Anaerobic and ics, pheromone and drugs. To produce (S)-3-HB using Sac- Aerobic Processings of Metabolite Sample Preparation charomyces cerevisiae, the biosynthetic pathway consisting for Clostridium Acetobutylicum of three enzymes was introduced into the metabolic pathway Sooah Kim*1, Sang-Hyun Lee*1,2 and Kyoung Heon Kim1 1 of the yeast. To promote maximal activity, we compared the Department of Biotechnology, Korea University Graduate School, Seoul, South Korea enzymatic activities of each expressed proteins composing 2R&D Center, GS Caltex Corporation, Daejeon, South Korea the metabolic pathway of 3-HB synthesis. By comparing the in vitro enzymatic activities and in vivo assays by flask cul- Clostridium acetobutylicum, a Gram-positive strict anaer- tures of the recombinant strains with ethanol carbon source,

118 Poster Abstracts

obe, has been actively used for producing butanol that is the activity of the recombinant EG as a processive EG. The considered as an important chemical feedstock and biofuel. results from our study can be used for the industrial appli- Recently, metabolic engineering is vigorously exploited to cation of processive EGs, possibly to supplement cellulase improve the fermentation performance of C. acetobutylicum mixtures for producing fermentable sugar from lignocellu- for the production of butanol. Metabolomics is also used to lose feedstocks. elucidate the metabolic snapshot and signature of metabolically engineered strains, and these information can be used 225. Engineering of a protein translocation system in for further improvement of engineered strains. To obtain Rhodococcus jostii RHA1 for the secretion of ligninases Raphael Roccor* and Lindsay D. Eltis reliable metabolomic data, the reproducible, nonselec- Department of Microbiology and Immunology, The Univer- tive, reliable and efficient metabolome sample preparation sity of British Columbia, Vancouver, Canada is essential. Currently, the metabolite sample preparation method, which was originally developed for Escherichia coli, Lignin is an aromatic polymer in plant biomass that has has been used for C. acetobutylicum without any evaluation considerable potential as a sustainable source for fuels and under anaerobic conditions. The necessity of customization chemicals. One approach to transform lignin to target prod- of metabolite sample preparation depending on the speci- ucts is to develop bacterial biocatalyts. The soil bacterium ficity of microorganisms is already revealed by several other Rhodococcus jostii RHA1 (RHA1) is an attractive candidate studies. In this study, we systematically compared atmo- for developing lignin-transforming biocatalysts. A strain with spheric and anaerobic processings and evaluated extraction engineered catabolic pathways could be used in a bioreac- solvents for metabolome sample preparation methods for C. tor to convert lignin to high-value chemicals. To achieve acetobutylicum by focusing on the metabolomic discrimina- this, a strain which produces high amounts of extracellular tion of acidogenic and solventogenic phases. The metabo- lignin-modifying enzymes (ligninases) is required. Our aim lite analysis was performed by gas chromatography/time- is to engineer RHA1 to efficiently secrete ligninases using of-flight mass spectrometry. The results of this study could the twin-arginine translocation (Tat) system. The system has be used as a standard procedure for the metabolite sample been engineered in strains related to Rhodococcus using preparation of the metabolomics and metabolic engineering signal peptides that lead to high-level secretion. As a first of C. acetobutylicum. approach, we want to identify signal sequences that promote efficient secretion of ligninases inRHA1. We selected 224. Identification and Characterization of a Processive various sequences which were predicted as secretion signal Endoglucanase from a Marine Bacterium As a Means to Substitute Cellulbiose Hydrolases in RHA1 and based on studies in related bacteria. We tested Hee Jin Lee*, In Jung Kim, Hakjin Youn and the secretion qualities of the signal sequences on three Kyoung Heon Kim reporter proteins (green fluorescent protein, a phosphatase Department of Biotechnology, Korea University Graduate from RHA1, and the dye decolorizing peroxidase Dyp2). We School, Seoul, South Korea have begun to evaluate the secretion of the reporter enzymes by measuring their activities in the culture medium. Cellulases are categorized into cellobiose hydrolases Four of the five tested Tat signal peptides mediated protein (CBHs) and endoglucanases (EGs). CBHs cleave beta-1,4- secretion in RHA1 compared to the controls without a signal glycosidic linkages from the ends of cellulose chains thus peptide. Our preliminary data suggest that the effect of each producing large amounts of soluble sugar (such as cellobi- signal peptide is dependent on the Tat substrate. Using a ose) from cellluose. CBHs are prevalent in fungi but are rare Tat knockout mutant, we will evaluate whether the observed in bacteria. EGs cleave internally-locating beta-1,4-glyco- secretion is truly dependent on the Tat system. Furthermore, sidic linkages in cellulose chains. Interestingly, processive we will test the secretion of various ligninases, and attempt EGs not only possesses the activity of EGs but also those to maximize secretion by optimizing gene expression and of CBHs since they can cleave beta-1,4-glycosidic linkages by overexpression of the Tat system machinery. at the end sides of cellulose. As processive EGs can replace fungal CBHs, the importance of processive EGs from bacte- 226. Synthetic Design of Pathways and Organelles for rial origins in increasing. In this study, we have identified a Photosynthetic Terpene Production. processive EG from a marine bacterium, and characterized Yong Kyoung Kim*1, Hong Ma1, Yu Wang2, Sheba Goklany3,

119 Poster Abstracts

Eiji Takahashi4, Yi Cheng Liu1, Susie Dai5, Don Ort4, Joe we designed and implemented a Chloroplast-originated Chappell3, Xin-Guang Zhu2 and Joshua Yuan1 organelle to further increase terpene production by enhanc- 1 Plant Pathology and Microbiology, Texas A&M University, TX ing carbon sink. Redesigned lipid droplet forming proteins 2PICB, Chinese academy of science were used to generate the chloroplast originated droplet 3Pharmaceutical Sciences, University of Kentucky, Lexing- ton, KY organelle. The organelle design has led to an increased level 4Plant Biology, University of Illinois at Urbana-Champaign, IL of squalene production at 2500ug/g FW. The novel organelle 5Department of Veterinary Pathobiology, Texas A&M represents a unique approach to enhance the carbon sink University, TX for high value compounds or fuels. Overall, the two strategies enabled both the better integration of photosynthesis One of the fundamental challenges in renewable energy to terpene synthesis and the enhancement of carbon sink in is the efficient harnessing and transformation of sunlight terpene droplet. Combining the two technologies together energy for reducing inorganic carbon toward the production will lead to a ‘push and pull’ engineering strategy to maxi- of fuels and chemicals. Photosynthetically-driven terpene mize the photosynthetic terpene production. production represents one of the most straightforward and

energy-efficient route for converting sunlight and CO2 to 227. Production of Anteiso-Branched Fatty Acids in fuel molecules. Moreover, terpene has a diverse range of Escherichia coli, Next Generation Biofuels with industrial applications including special chemicals and nu- Improved Cold-Flow Properties Robert W. Hauschalter*1, Woncheol Kim1, Ted A. Chavkin2, traceuticals. We have designed and implemented synthetic Lionadi The2, Megan E. Garber2, Leonard Katz2 and Jay D. pathways and organelles to achieve record-level photo- Keasling1,2 synthetic production of squalene, a triterpene for biofuel, 1Joint BioEnergy Institute, Lawrence Berkeley National nutriceutical and other applications. First, the synthetic Laboratory, Emeryville, CA pathways were designed and implemented to redirect pho- 2QB3 Institute, University of California, Berkeley, Emeryville, torespiration by-product glycolate to pyruvate for terpene biosynthesis. The pathway design directly channeled a A major disadvantage of fuels derived from biological photosynthate into terpene production and has the potential sources is their undesirable physical properties such as to also reduce the energy consumed for recycling glycolate. high cloud and pour points and high viscosity. Here we Kinetics-based computational modeling was first carried out report the development of an Escherichia coli strain that to evaluate how carbon re-partitioning will impact photo- efficiently produces anteiso-branched fatty acids, which can synthesis and terpene production. The results indicated be converted into downstream products with lower cloud that high terpene level via C2 (glycolate) redirection can be and pour points compared to less heterogeneous mixtures achieved without significant reduction of photosynthesis produced via the native metabolism of the cell. This was rate. The implementation of two alternative pathways clearly achieved through the deletion of metA, tdh, ilvB, and ilvN indicated that redirecting glycolate to terpene biosynthe- and the overexpression of thrABC and ilvCD from E. coli, sis will lead to significant increase of terpene production. ilvA from Corynebacterium glutamicum, ilvGM from Salmo- In particular, the synergistic engineering of C2 redirection nella typhimurium, as well as bFabH2 and the bkd operon with terpene biosynthesis has led to the production of from Bacillus subtilis, which together promote the synthesis 2700ug/g FW squalene, which is over four-fold increase of of the 2-methylbutyryl-CoA and use this metabolite to prime the level achieved by engineering terpene pathway alone. fatty acid synthesis. When these genetic manipulations are The partial redirection led to a relatively less significant coupled with those that promote free fatty acid synthesis increase of squalene production at about 1200ug/g FW. and accumulation, 24% of the free fatty acids produced The further pathway channeling assay and photosynthesis in the engineered E.coli cells were anteiso-branched. This measurement all indicated that C2 redirection is functional work addresses a serious limitation that must be overcome and can lead to a high level of terpene production without in order to produce renewable biodiesel and oleochemicals comprising plant growth. The research thus established a that perform as well as their petroleum-based counterparts. novel approach to directly couple photosynthesis process with terpene production. Second, in addition to pathway design to couple photosynthesis with terpene synthesis,

120 Poster Abstracts

228. A Functional recT Gene for Recombineering of HA molecular weight obtained varies inversely with glucose Clostridium uptake rate (Sanghe et al., 2011). Hence, the same com- 1 1,2 1,2 1 Hongjun Dong* , Wenwen Tao , Fuyu Gong , Yin Li , Yan- bination of has-genes were expressed in ldh-mutant strain ping Zhang1* L. lactis NZ9020 (Bongers et al., 2003) procured from NIZO 1CAS Key Laboratory of Microbial Physiological and Metabolic Engineering, Institute of Microbiology, Chinese (Netherlands) in order to study the effect of flux diversion Academy of Sciences, Beijing, China from lactate production to HA and other metabolites. A dra- 2University of Chinese Academy of Sciences, Beijing, China matic improvement in HA yield and molecular weight was observed in recombinant cultures of L. lactis NZ9020 (ldh- Recombineering is an efficient genetic manipulation method mutant) compared to recombinant L. lactis NZ9000 cultures employing the mechanism of phagenic RecT-mediated under similar conditions. The L. lactis NZ9020 cultures also homologous recombination. To develop a recombineering produced substantial levels of ethanol, acetoin and formate, method for Clostridium, a putative recT gene (CPF0939) which were not observed in L. lactis NZ9000 cultures. The from C. perfringens genome was functionally verified in a production of these metabolites also resulted in a signifi- clostridial host C. acetobutylicum. We show that a short cant shift in the redox balance, which may contribute to synthetic oligonucleotide can be introduced into the target enhancement of fluxes in the biosynthetic pathway of the site for specific point mutation. This functional recT gene HA precursors, viz. UDP-glucuronic acid and UDP-N-acetyl- would therefore contribute to development of recombineer- glucosamine. This study will investigate the effect of flux ing tools for Clostridium. diversion from lactate towards formation of more reduced metabolites and the concomitant effect of flux variations 229. Enhancement of Hyaluronic Acid Molecular Weight in HA pathway. The study will also correlate the effect of By Re-Direction of Metabolic Fluxes in Engineered HA-precursor concentrations on the molecular weight of Lactococcus Lactis Cultures HA produced. Finally, we will investigate the effect of NAD+/ Mandeep Kaur* and Guhan Jayaraman Department of Biotechnology, Indian Institute of Technolo- NADH ratios on metabolic flux regulation in the pathways gy-Madras, Chennai, India leading to HA production and other metabolites in the engineered L. lactis cultures. The insights obtained from these Production of high-molecular weight hyaluronic acid (HA) is studies will be used to further re-engineer the metabolic a significant challenge in metabolically-engineered bacte- pathways as well as to develop bioprocess strategies for rial systems. Hyaluronic acid is a linear polymer made up high-MW HA production by recombinant L. lactis cultures. of alternating units of β-(1-4)-glucuronic acid and β-(1-3)- References: N-acetylglucosamine, whose precursors are synthesized in Bongers R.S., Hoefnagel M.H.N., Starrenburg M.J.C., Siemerink parallel branches of the HA biosynthetic pathway. This study M.A.J., Arends J.G.A., Hugenholtz J., Kleerebezem M. (2003) investigates the effect of metabolic flux distribution and “IS981-Mediated Adaptive Evolution Recovers Lactate Production redox balance on HA yield and molecular weight obtained by ldhB Transcription Activation in a Lactate Dehydrogenase-De- from metabolically-engineered Lactococcus lactis cultures. ficient Strain of Lactococcus lactis”, Journal of Bacteriology, 185, 4499-4507 The hyaluronic acid biosynthetic pathway was engineered in L. lactis using a combination of genes obtained from Prasad S.B., Jayaraman G., Ramachandran K.B. (2010) “Hyaluron- ic acid production is enhanced by the additional co-expression the has-operon of Streptococcus zooepidemicus. It was of UDP-glucose pyrophosphorylase in Lactococcus lactis”, Appl observed that an optimal combination of three has-genes Microbiol Biotechnol, 86, 273–283 is required for substantial HA production. Two different sets Prasad S.B., Ramachandran K.B., Jayaraman G. (2012) “Transcrip- of heterologous has-genes (hasABC and hasABD) were tion analysis of hyaluronan biosynthesis genes in Streptococcus expressed in nisin-inducible recombinant strain L. lactis zooepidemicus and metabolically engineered Lactococcus lactis”, NZ9000 (Prasad et al. 2010, 2012). This allowed for inde- Applied Microbiology and Biotechnology, 94, 1593-1607 pendent variation of fluxes in the two precursor pathways. Sanghe N., Prasad S.B., Ramachandran K.B., Jayaraman G. The yield and molecular weight of HA obtained from L. (2011), “Hyaluronic acid production in metabolically engineered lactis NZ9000 cultures was lower than HA produced by S. Lactococcus lactis NZ9000: Bioreactor studies and flux balance zooepidemicus cultures. It was observed that HA yield from analysis” (Oral Presentation), 1st European Congress of Applied Biotechnology, Sept 25-29, 2011, Berlin, Germany glucose substrate correlates inversely with lactate yield and

121 Poster Abstracts

230. High-Throughput Screening of Metabolite ate School of Engineering, Kobe University, Kobe, Japan Producers Using Synthetic Suicide Riboswitch in 5Graduate School of Engineering, Department of Chemical Saccharomyces cerevisiae Science and Engineering, Kobe University, Kobe, Japan Sang-Woo Lee* and Min-Kyu Oh Department of Chemical and Biological Engineering, Korea Computational metabolic design has become essential in University, Seoul, South Korea expanding metabolic pathway knowledge from an increas-

ing number of metabolic pathway data. In order to find po- Utilization of microorganisms has been drastically diversi- tential metabolic pathways of high value, it is necessary to fied since the invention of various methods for phenotypic expand knowledge from existing databases such as KEGG alteration such as the recombinant DNA technology and and BRENDA database by mining comprehensive chemical novel synthetic tools developed recently. Artificial devices and enzymatic information. Several computational methods such as synthetic riboswitch have shown potential for have been already reported for metabolic pathway design, distinct phenotypic perturbation by its synthetic nature. but most cases set a limit on the diversity of chemical and Riboswitch, a small regulatory element found in RNAs, was enzymatic information because of a trade-off between the conducted for reprogramming of microorganism to produce amount of information and the computational feasibility. We valuable metabolites in this work. Natural riboswitches can have developed an efficient method handling such data to sense intracellular concentration of metabolite and regulate design metabolic pathways including putative compounds the expression of gene in response to metabolite level. To and enzymatic reactions. mimicking this feature, we utilized a self-cleaving ribozyme glmS found in Gram-positive bacteria. glmS ribozyme We have taken advantage of chemical structures and cleaves its own transcript in response to intracellular glucos- structural differences that include sufficient knowledge for amine 6-phosphate(GlcN6P) concentration. glmS ribozyme reconstructing metabolic pathways. We first decomposed was integrated into 3’-untranslated region of FCY1, which chemical structure information into atom and bond types encodes cytosine deaminase, in Saccharomyces cerevisiae to represent chemical structures as feature vectors. As a to construct suicide riboswitch for evolutionary engieering. chemical information resource, a chemical database was The strain BY4742 FGU contains suicide riboswitch showed constructed from PubChem, one of the largest chemical reprogrammed metabolite level-dependent growth as we databases, to ensure the diversity of chemical information. expected. By using this riboswitch, we isolated a N-acetyl The enzymatic reactions derived from KEGG and BRENDA glucosamine(GlcNAc) producer strains by screening a mu- databases were subsequently defined by the differences tant library of glutamine-fructose-6-phosphate transaminase between substrate and product feature vectors, and linked (Gfa1p) and a library of haloacid dehalogenase-like phos- with enzymatic information. The simple method in rep- phatases (HAD phosphatases) from Escherichia coli which resenting chemical structures and reactions allows us to cannot be screened by conventional evolutionary engineer- design metabolic pathways on the basis of feature vectors. ing. Since the mechanisms used in this work are universal in microorganisms, our synthetic suicide riboswitch can be The design of metabolic pathways is started by finding applied to wide range of organisms and can aid efficient and combinations of reaction features that satisfy differences high-throughput screening of inconspicuous phenotypes. between two chemical features. The reaction features are then rearranged to yield chemical features in sequence, 231. A Computational Method to Construct an Extensive Metabolic Pathway Database which are used to assign compounds from our chemical Michihiro Araki1, Hiroki Makiguchi2, Teppei Ogawa2, Kohei database by similarity comparison. Compared with the Miyaoku*3, Takeshi Taniguchi*3, Robert Sidney Cox III4, previous method, our method significantly reduces the com- Masahiko Nakatsui4 and Akihiko Kondo5 putational time to find extensive metabolic pathways. The 1 Organization of Advanced Science and Technology, Kobe resulting metabolic pathways including putative compounds University, Kobe, Japan and enzymatic reactions are ranked on the basis of feasibil- 2Mitsui Knowledge Industry Co., Osaka, Japan 3Mitsubishi Chemical Group Science and Technology ity criteria using chemical similarity and stored in a pathway Research Center, Inc., Yokohama, Japan database. A web user interface is also developed to check 4Department of Chemical Science and Engineering, Gradu- pathway candidates by eye inspections.

122 Poster Abstracts

As a test case, a set of more than 7,000 alpha amino acid-like in the tolerant strain by whole genome sequencing. Moreover, compounds from the PubChem database is calculated to transcriptomic analysis identified that the expression levels find acceptable metabolic pathways for their synthesis from of the genes related the photosystem and phycobilisome glucose. We found putative metabolic pathways for 1987 in the tolerant strain were lower than the parent strain. This compounds, and checked to obtain some interesting path- suggests that the decrease of the amount of photon received ways out of them. Any chemical and reaction information can in the cell by repression of the light harvesting system is one be also applied to develop an extensive pathway database, possible mechanism of the high-light tolerance. which will increase a chance to find unknown metabolic path- 233. Transport Proteins for Itaconic Acid Production in ways with diverse chemical and enzymatic information. Aspergillus Niger Matthias G. Steiger*1,2,3, Peter J. Punt2,4, Arthur F.J. Ram2, Diethard Mattanovich1,4 and Michael Sauer1,4 232. Development and Analysis of High-Light Stress 1Austrian Centre of Industrial Biotechnology (ACIB GmbH), Tolerant Strain of Synechocystis Sp. PCC 6803 Vienna, Austria Katsunori Yoshikawa*1, Kenichi Ogawa1 and Hiroshi 2Institute of Biology, University Leiden, Leiden, Netherlands Shimizu2 3Department of Biotechnology, University of Natural Re- 1Department of Bioinformatic Engineering, Graduate School sources and Life Sciences, Vienna, Austria of Information Science and Technology, Osaka University, 4TNO Microbiology and Systems Biology, Zeist, Netherlands Japan 2Graduate School of Information Science and Technology, Osaka University, Suita, Japan Aspergillus niger is a well-established host organism for the production of carboxylic acids. Acids like citric, gluconic and Cyanobacteria adapt to the light intensity in the environ- oxalic acids can already be produced by A. niger and high ment by changing the photosynthetic apparatus. However, titers are obtained. The formation of carboxylic acids involves when the energy supply exceeds the capacity of the photo- the shuttling of intermediate metabolites between different in- synthesis under the high-light condition, the excess energy tracellular compartments and utilizes the different enzymatic becomes stress and decreases the cell growth. Adaptive capabilities of the respective compartment. The knowledge evolution is a powerful strategy to develop the stress toler- about the involved shuttling mechanisms and the localization ant strain by long-term cultivation under the target stress of the necessary enzymes is still fragmentary. conditions. In this study, to understand the high-light tolerance mechanism in cyanobacteria, we developed the high- In order to analyse the influence of the compartmental- light tolerant strain by adaptive evolution experiment, and ization on the organic acid production, we have chosen analyzed the tolerance mechanism by multi-omics analysis. itaconic acid as a target substance. Itaconic acid, which was selected by the US Department of Energy as one of the Synechocystis sp. PCC 6803 GT strain was used as the 12 building block chemicals for the industrial biotechnology, parental strain, and cells were cultivated with aeration of air is currently produced by A. terreus. Heterologous expres- and continuous light illumination using point source white sion of the A. terreus cadA gene also enables the forma- LED. To develop the high-light tolerant strain, the serial trans- tion of itaconic acid in A. niger although only low titers are fer experiment was performed under 7,000 μmol/m2/s light obtained. An increase of the productivity was obtained by intensity in which the growth rate of the parental strain was targeting the pathway to the mitochondria. Furthermore, it decreased. After 56 days of the serial transfer experiment, was shown that the heterologous expression of two trans- the high-light tolerant strain was obtained. The tolerant strain port proteins which are found in close proximity to the cadA could grow under 9,000 μmol/m2/s in which the growth of the gene in A. terreus, have a positive impact on the itaconic parental strain was severely inhibited. The high-light toler- acid formation. ance of the tolerant strain was not changed after 15 days of the serial transfer experiment using the tolerant strain under These two transport proteins are now characterized in more low-light condition (40 μmol/m2/s). This suggested that the detail making use of an inducible promoter system. Fur- high-light tolerance was not acclimation but probably caused thermore, their localization in the cell is determined. For this by mutations in the genome of the tolerant strain. Actually purpose the proteins are tagged with GFP and analysed by non-synonymous mutations in the two genes were identified fluorescence microscopy.

123 Poster Abstracts

234. A Strategy for Design, Redesign, and Optimization of Ethylene Production in E. coli Sean Lynch1,2, Carrie Eckert*2, 3, Jianping Yu2, Pin-Ching Maness2 and Ryan Gill2 1Biosciences, National Renewable Energy Laboratory, Golden, CO 2Chemical and Biological Engineering, University of Colorado, Boulder, CO 3Renewable and Sustainable Energy Institute, University of Colorado, Boulder, CO

In collaboration with the University of Colorado-Boulder and LBNL/University of California, Berkeley, we are developing a high-throughput platform approach for prediction and selection to systematically improve isobutanol and ethylene production in E. coli. This project will ultimately combine implementation and testing of predicted modifications based on pathway modeling followed by high-throughput screening of genome-scale (TRMR) and combinatorial pathway (MAGE) libraries to create strains with improved production. Current work at NREL is focused on enzyme and pathway optimization of a chassis ethylene production that will be subsequently utilized for TRMR/MAGE selection methods.

Ethylene is the most highly utilized organic compound used in the production of plastics and chemicals and can be utilized as a precursor for high-energy biofuels. Most ethylene is derived from fossil fuels by steam cracking, resulting in

the highest CO2 emissions in the chemical industry. Ethylene is produced in some bacteria and fungi via an ethylene- forming enzyme (Efe) that uses alpha-ketoglutarate (AKG) and arginine as substrates. Heterologous expression of efe alone is sufficient for ethylene production in a variety of organisms, but the reported productivity is low. In order to improve the often low titers of ethylene produced, an in- depth understanding of optimal expression levels, solubility/ stability of Efe, the Efe reaction mechanism, and interactions between the Efe reaction and other metabolic pathways are needed. Our current work is focused on investigation of optimal levels of Efe enzyme for maximal solubility and activity, testing substrate-feeding and/or targeted genetic modifications based on pathway modeling to improve flux to AKG and arginine, and development of high-throughput screening methods to select for increased production of key intermediates and/or ethylene to allow for selection from pooled libraries.

124