Appl Microbiol Biotechnol (2012) 93:2063–2073 DOI 10.1007/s00253-011-3725-1 APPLIED GENETICS AND MOLECULAR BIOTECHNOLOGY Computational identification of gene over-expression targets for metabolic engineering of taxadiene production Brett A. Boghigian & John Armando & Daniel Salas & Blaine A. Pfeifer Received: 4 August 2011 /Revised: 13 October 2011 /Accepted: 12 November 2011 /Published online: 30 November 2011 # Springer-Verlag 2011 Abstract Taxadiene is the first dedicated intermediate in compounds derive from plant sources, and many produc- the biosynthetic pathway of the anticancer compound tion methods have relied on adapting the plant hosts for Taxol. Recent studies have taken advantage of heterologous process development of the desired isoprenoid compound hosts to produce taxadiene and other isoprenoid com- (Lewis and Ausubel 2006; Wink 2010). Taxol has been a pounds, and such ventures now offer research opportunities prominent example with both semi-synthetic and cell that take advantage of the engineering tools associated with culture methods reliant upon plants for eventual production the surrogate host. In this study, metabolic engineering was of the final compound (Cragg et al. 1993). More recently, applied in the context of over-expression targets predicted heterologous biosynthesis of plant-derived natural products to improve taxadiene production. Identified targets included has emerged as an alternative to native plant production genes both within and outside of the isoprenoid precursor systems (Kirby and Keasling 2009; Leonard et al. 2009). In pathway. These targets were then tested for experimental this approach, the genetic material required for eventual over-expression in a heterologous Escherichia coli host product biosynthesis is transferred to a surrogate host. The designed to support isoprenoid biosynthesis. Results con- new host generally possesses some advantageous properties firmed the computationally predicted improvements and when compared to the original host, and these often include indicated a synergy between targets within the expected faster growth kinetics, more advanced engineering tools, isoprenoid precursor pathway and those outside this and a wider genetic and physiologic knowledge base pathway. The presented algorithm is broadly applicable to (Zhang et al. 2011). Recent high-profile examples of other host systems and/or product choices. heterologous biosynthesis include the production of arte- misinic acid through Saccharomyces cerevisiae (Ro et al. Keywords Taxol . Taxadiene . Taxadiene synthase . 2006) and the production of taxadien-5α-ol through Over-expression . E. coli . Heterologous biosynthesis . Escherichia coli (Ajikumar et al. 2010). These and Metabolic engineering numerous other examples support continued research into the prospect of using heterologous biosynthesis as an economically viable route to the commercial production of Introduction natural products (Kirby and Keasling 2009; Leonard et al. 2009). Taxol and other isoprenoids possess a wide range of As mentioned, the choice of heterologous biosynthesis therapeutic potential (Ajikumar et al. 2008). Many of these offers new engineering tools to aid in process development. This is especially important to improve the often low titers, yields, and specific productivities associated with heterol- : : : B. A. Boghigian J. Armando D. Salas B. A. Pfeifer (*) ogous production attempts (Boghigian and Pfeifer 2008). Department of Chemical and Biological Engineering, One of the most common approaches towards doing so is Science and Technology Center, Tufts University, the application of metabolic engineering in the form of 4 Colby Street, Medford, MA 02155, USA cellular-based modeling of metabolism (Boghigian et al. e-mail: [email protected] 2010b; Kim et al. 2008; Park et al. 2009). The underlying 2064 Appl Microbiol Biotechnol (2012) 93:2063–2073 goal is to improve carbon flow to a metabolite of interest. heterologous biosynthetic pathway, which converts the In the case of the heterologous Taxol system, a key universal isoprenoid precursors isopentenyl diphosphate milestone is to improve carbon flow to early intermediates (IPP) and dimethylallyl diphosphate (DMAPP) to taxa- thus far produced through E. coli. As seen in Fig. 1, the diene. The Fig. 1 depiction of the upstream and down- production of taxadiene (the first dedicated intermediate of stream pathways is a simplification of taxadiene the Taxol pathway) is produced by the combination of an biosynthesis within the complex, surrounding E. coli upstream native precursor pathway, termed the methylery- reaction network. The holistic view of the cell must be thritol phosphate (MEP) pathway (or the 1-deoxy-D- considered when optimizing the production of heterologous xylulose 5-phosphate (DXP) pathway), and a downstream taxadiene through metabolic engineering. O O O O P O- COO- OH O- OH O O O P O- OH O- HO O O P O- OH OH O- NH2 N HO O O O P O P O O N O OH OH O- O- - O NH2 O P O- OH OH N O O O O P O P O O N O OH OH O- O- OH OH O- O P O- O O P O O OH OH Fig. 1 The pathway from central metabolites pyruvate and glyceraldehyde-3-phosphate to taxadiene Appl Microbiol Biotechnol (2012) 93:2063–2073 2065 O O O P O P O- OH O- O- O O O O O P O P O- O P O P O- O- O- O- O- Fig. 1 (continued) While most previous metabolic engineering efforts have heterologous system currently available. The results support applied computational methods to model cellular metabo- the use of this approach to identify over-expression targets lism and predict gene deletions to improve specified capable of improving the titer of a desired product. Such an metabolite levels, this study presents the development and approach emphasizes the expanded repertoire of engineer- application of an algorithm for identifying over-expression ing options available with well-characterized heterologous targets to improve product titer. Specifically, a genome- systems and will be used together with accompanying scale metabolic network was assessed with over-expression genetic and process engineering tools to continually of genetic targets to improve taxadiene biosynthetic flux. improve titers of isoprenoid and other medicinally relevant Those targets were then tested experimentally through the natural products. 2066 Appl Microbiol Biotechnol (2012) 93:2063–2073 Materials and methods Over-expression target identification algorithm Model construction The over-expression algorithm involves: 1) imposing a taxadiene production flux (as determined experimentally), The E. coli genome-scale metabolic model iAF1260 was 2) solving a Flux Balance Analysis (FBA) problem used as the base model in this study. This model contains (Edwards et al. 2002, 2001; Edwards and Palsson 2000), 2,077 reactions, 1,039 metabolites, and 1,261 genes (Feist et 3) imposing an amplification in individual reaction fluxes al. 2007). Reactions then had to be added to account for (to simulate the effect of gene over-expression), 4) solving steps catalyzed by two heterologous enzymes introduced to a Minimization of Metabolic Adjustment (MoMA) problem produce taxadiene through E. coli. These reactions are: 1) a (Segre et al. 2002), and 5) identifying over-expressions that geranylgeranyl diphosphate synthase (GGPPS; to catalyze: led to a phenotype fraction value, fPH (the product of farnesyl diphosphate + isopentenyl diphosphate→ weighted and dimensionless biomass and taxadiene flux geranylgeranyl diphosphate + diphosphate), 2) a cyclizing values; Boghigian et al. 2010a), greater than unity (an taxadiene synthase (TXS; to catalyze: geranylgeranyl overview of this algorithm can be seen in Fig. 2). Steps 3 diphosphate→taxa-4,11-diene + diphosphate), and 3) a and 4 were iterated for every reaction within the network. taxadiene transport reaction (taxa-4,11-diene→[extracellular]) Calculations were made under conditions to simulate (Ajikumar et al. 2008, 2010). complex medium. Medium composition can be approximated Calculations were conducted in MATLAB® 7.4 (Math- by setting uptake rates of specific chemical components works Inc.; Natick, MA) utilizing the SBMLToolbox (version known to exist in the medium of interest. The “computational 2.0.2, http://sbml.org/software/sbmltoolbox/)(Keatingetal. complex medium” contained all twenty naturally occurring 2006;SchmidtandJirstrand2006) and the COBRA Toolbox amino acids (L-isomers). The lower bounds of the amino acid (version 1.3.3, http://gcrg.ucsd.edu/) (Becker et al. 2007). transport reactions were set to −0.1mmol/gDCW/h(a Optimization was undertaken using the CPLEX (version negative sign indicates metabolite uptake into the cell) 11.0) algorithm of the TOMLAB™ Optimization Environ- and were chosen based upon previous literature values ment (TOMLAB™/CPLEX) interfaced with the COBRA and because they satisfied the relative biomass differ- Toolbox and MATLAB® 7.4. ences experimentally observed between media (Oh et al. (1) (2) H H Add Heterologous Calculate Fluxes Base Model Components with FBA Iterate (5) (4) (3) 6 4 2 0 0 500 1000 Identify Over- Re-calculate Fluxes Amplify Reaction Producing Strains with MoMA Manually Fig. 2 An overview of the proposed algorithm for identifying over-expression targets to improve product titer Appl Microbiol Biotechnol (2012) 93:2063–2073 2067 2007; Selvarasu et al. 2009). Glycerol transport rates kanamycin, and 34 mg/l for chloramphenicol, and IPTG were set to −3.0 mmol/g DCW/h, as described previously was added at a concentration of 100 μM. (Boghigian et al. 2010a). Taxadiene quantification