Genome-scale metabolic network guided engineering of Streptomyces tsukubaensis for FK506 production improvement

Background FK506 is an important immunosuppressant, which can be produced by Streptomyces tsukubaensis. However, the production capacity of the strain is very low. Hereby, a computational guided engineering approach was proposed in order to improve the intracellular precursor and cofactor availability of FK506 in S. tsukubaensis. Results First, a genome-scale metabolic model of S. tsukubaensis was constructed based on its annotated genome and biochemical information. Subsequently, several potential genetic targets (knockout or overexpression) that guaranteed an improved yield of FK506 were identified by the recently developed methodology. To validate the model predictions, each target gene was manipulated in the parent strain D852, respectively. All the engineered strains showed a higher FK506 production, compared with D852. Furthermore, the combined effect of the genetic modifications was evaluated. Results showed that the strain HT-ΔGDH-DAZ with gdhA-deletion and dahp-, accA2-, zwf2-overexpression enhanced FK506 concentration up to 398.9 mg/L, compared with 143.5 mg/L of the parent strain D852. Finally, fed-batch fermentations of HT-ΔGDH-DAZ were carried out, which led to the FK506 production of 435.9 mg/L, 1.47-fold higher than the parent strain D852 (158.7 mg/L). Conclusions Results confirmed that the promising targets led to an increase in FK506 titer. The present work is the first attempt to engineer the primary precursor pathways to improve FK506 production in S. tsukubaensis with genome-scale metabolic network guided metabolic engineering. The relationship between model prediction and experimental results demonstrates the rationality and validity of this approach for target identification. This strategy can also be applied to the improvement of other important secondary metabolites.


Biomass composition
The macromolecular composition of the cell was partly measured and partly estimated from literature data. Dry weight was measured by washing cells with 0.9% NaCl solution and drying them at 80 °C on a pre-weighted filter until a steady weight was achieved. Biomass components contain protein, RNA, DNA, lipids, small molecules, cell wall components (peptidoglycan, carbohydrate and teichoic acid). Total protein was determined using the Bradford method [1], which was estimated based on a standard curve of bovine serum albumin (25-200 μg/mL). The protein content of samples for our experiments was within the range 40-50% of dry cell mass. Pellet hydrolysates were used to measure amino acid composition. The amino acid composition of the protein fraction has also been determined by Agilent HPLC systems (Agilent 1200, USA) with Agilent Zorbax Eclipse column (4.6 × 150 mm). No differences were observed in composition within the dilution rate range of 0.1-0.5 h -1 . Total carbohydrate was measured by the phenol/sulfuric acid method [2]. Cellular composition was estimated based on a standard curve of glucose (10-200 μg/mL). The carbohydrate content of samples was within the range 10-15% of dry cell mass, thus 1250 μg/mL of freeze-dried biomass was used in order to acquire a carbohydrate concentration that would be within the mid-standard range. Total DNA was determined by the diphenylamine method using a hot perchloric acid extraction of freshly-harvested biomass [3]. Samples were estimated based on a calf-thymus DNA standard solutions (100-500 μg/mL). Total RNA content was determined using an KOH/UV-based procedure based on calf liver RNA standard solutions (100-500 μg/mL) [4]. Ash composed of inorganic compounds and ions, as well as intracellular metabolites, were not taken into account in the model. The biomass equation consists of biomass components and the growth-associated ATP consumption.
Based on the data shown in Table S1   Energy requirement for polymerization (mmol ATP/g) = 40.04 The DNA composition was calculated based on the genome information of S.
tsukubaensis. GC content of S. tsukubaensis is 71.5%, which is then used to calculate the ratio of the nucleic acids in the DNA. The polymerization energy was assumed to be the same as in E. coli [7]. Energy requirement for polymerization (mmol ATP/g) = 4.4 The RNA composition was determined from genomic data assuming the following composition: 5% mRNA, 75% rRNA and 20% tRNA [5]. The polymerization energy was assumed to be the same as E. coli [7]. Energy requirement for polymerization (mmol ATP/g) = 1.25 The composition of phospholipids was assumed to be the same as E. coli.
Biosynthesis of phospholipids and fatty acids components are included in the reaction set.   Energy requirement for polymerization (mmol ATP/g) = 1.24 a Data was taken from S. coelicolor composition [5].
It was assumed that the selected small molecules were equally presented (w/w) in the pool. Energy requirement for polymerization (mmol ATP/g) = 5.026 a Glycine and diaminopimelic acid were required for peptidoglycan polymerization with molecule of water. Other components were required without molecule of water. b Data was taken from average molecular composition of peptidoglycan from S. coelicolor [5].  as molecular formula [5]. b Data was taken from S. coelicolor composition [5].