Zhifeng Liu1,2, Jianlin Xu1,2,3, Zhanguang Feng1,2, Yong Wang1. 1. CAS-Key Laboratory of Synthetic Biology, CAS Centre for Excellence in Molecular Plant Sciences, Institute of Plant Physiology and Ecology, Chinese Academy of Sciences, Shanghai, 200032, China. 2. University of Chinese Academy of Sciences, Beijing, 100039, China. 3. State Key Laboratory of Bioreactor Engineering, East China University of Science and Technology, Shanghai, 200237, China.
Abstract
The insufficient supply of sugar units is the key limitation for the biosynthesis of glycosylated products. The unusual sugar TDP-l-mycarose is initially attached to the C3 of the polyketide erythronolide B, resulting in 3-O-α-mycarosylerythronolide B (MEB). Here, we present the de novo biosynthesis of MEB in Escherichia coli and improve its production using multi-strategy metabolic engineering. Firstly, by blocking precursor glucose-1-phosphate competing pathways, the MEB titer of triple knockout strain QC13 was significantly enhanced to 41.2 mg/L, 9.8-fold to that produced by parental strain BAP230. Subsequently, the MEB production was further increased to 48.3 mg/L through overexpression of rfbA and rfbB. Moreover, the CRISPRi was implemented to promote the TDP-l-mycarose biosynthesis via repressing the glycolysis and TDP-l-rhamnose pathway. Our study paves the way for efficient production of erythromycins in E. coli and provides a promising platform that can be applied for biosynthesis of other glycosylated products with unusual sugars.
The insufficient supply of sugar units is the key limitation for the biosynthesis of glycosylated products. The unusual sugar TDP-l-mycarose is initially attached to the C3 of the polyketide erythronolide B, resulting in 3-O-α-mycarosylerythronolide B (MEB). Here, we present the de novo biosynthesis of MEB in Escherichia coli and improve its production using multi-strategy metabolic engineering. Firstly, by blocking precursor glucose-1-phosphate competing pathways, the MEB titer of triple knockout strain QC13 was significantly enhanced to 41.2 mg/L, 9.8-fold to that produced by parental strain BAP230. Subsequently, the MEB production was further increased to 48.3 mg/L through overexpression of rfbA and rfbB. Moreover, the CRISPRi was implemented to promote the TDP-l-mycarose biosynthesis via repressing the glycolysis and TDP-l-rhamnose pathway. Our study paves the way for efficient production of erythromycins in E. coli and provides a promising platform that can be applied for biosynthesis of other glycosylated products with unusual sugars.
Glycosylated natural products have diverse bioactivity and improved properties, which are closely associated with the structure and attachment pattern of the sugar units [1,2]. It has been well explored that the sugar moiety of many therapeutic agents participates in the interaction between the drug and cellular target which enables effective drug targeting and improved pharmacological properties [[3], [4], [5], [6], [7]].Typically, the sugar units were originated from nucleoside diphosphate (NDP)-sugars, which can be categorized into distinct groups according to the complexity of biosynthetic pathway, namely common sugars and unusual sugars. Common sugars such as UDP-glucose and TDP-glucose are directly synthesized from sugar-1-phosphate (mainly glucose-1-phosphate, G1P) under the catalysis of nucleotidylyltransferase. Highly modified unusual sugars are generated through multi-step decorations including the epimerization, deoxygenation, ketoreduction and C-, N-, or O-methylations of common precursor TDP-4-keto-6-deoxy-d-glucose (TKDG), the immediate product of TDP-glucose catalyzed by TDP-4,6-dehydratase, which contribute to the formation of structurally diverse natural glycosylated compounds. In recent years, pathway modification and/or combinatorial biosynthesis has emerged as a promising strategy to generate various unusual sugars and novel glycosylated compounds with enhanced activities [[8], [9], [10]]. For example, by the inactivation of native gene dnrV and the expression of heterologous gene avrB or eryBIV, the TDP-l-daunosamine biosynthetic pathway of Streptomyces peucetius was modified to produce TDP-4′-epi-l-daunosamine that was the epimer at C4′ hydroxyl group of TDP-l-daunosamine, and the resulting glycosylated product epirubicin (4′-epidoxorubicin) possessed more effective antitumor property [11]. Besides, a series of undescribed deoxysugars were synthesized and attached to the anthracycline aglycones, resulting in the generation of unusual anthracycline analogues [12].Nevertheless, the biosynthesis of complex glycosylated products is usually hampered attributable to the insufficient availability of sugar units in the heterologous host [[13], [14], [15]]. Several metabolic engineering approaches including overexpression of endogenous genes [16], deletion of competitive pathways [17], the introduction of heterologous pathway genes [18], and reconstruction of independent pathways from other carbon resources [19,20] have been widely employed to enhance the production of glycosides conjugated with UDP-glucose. Though attempts have been made to increase the supply of precursor G1P via engineering the metabolic pathway of streptomyces [[21], [22], [23]], metabolic engineering Escherichia coli to enhance the unusual sugar pool has been scarcely reported.Erythromycin A, the 14-membered macrocyclic lactone attached with two unusual sugar units TDP-l-mycarose and TDP-d-desosamine that are the fundamental determinants for its antibacterial activity, is the representative member of polyketides used as an antibacterial drug in clinical treatment. Despite the fact that the production of polyketide skeleton erythronolide B (EB) reached 180 mg/L in E. coli [24], the yield of downstream erythromycins was extremely low. For example, the heterologous expression of 17 genes encoding two unusual monosaccharide biosynthetic enzymes enabled the 6-dEB-producing E. coli to generate erythromycins C and D with titers of 0.4 and 0.5 mg/L, respectively [25]. When introducing the whole erythromycin biosynthesis genes into E. coli and performing the two-step fermentation, the yield of the final product erythromycin A just reached 0.6 mg/L [26]. Additionally, Jiang et al. reconstructed the plasmids that contained the entire erythromycin biosynthetic gene cluster, resulting in the production of 1.2 mg/L erythromycin A [27]. The possible reason that the titer of erythromycins in E. coli is low might be attributed to the insufficient supply of two unusual sugars. It has been demonstrated that deletion of three genes (WecE, VioA, and RmlC) in the TKDG consumed pathway, the genetically modified strains strengthened the biosynthesis of TDP-l-mycarose and TDP-d-desosamine and was able to transform 6-dEB into erythromycin D, 60-fold to that of the original strain [28]. While the reported biocatalysts of 6-dEB necessitate further improvement to support the high production of erythromycin, there have been rare efforts to investigate and optimize the de novo microbial biosynthesis of 3-O-α-mycarosylerythronolide B (MEB), the first glycosylated intermediator of erythromycin.In this study, we establish the de novo MEB biosynthesis in E. coli and present a more comprehensive approach to promote the yield of MEB (Fig. 1). To achieve this, the bypass pathway of G1P was initially blocked (strategy 1). Then, key enzymes of the metabolic pathway were overexpressed to facilitate the biosynthesis of TKDG (strategy 2). Next, genes rmlC and rfbD were repressed to redirect the metabolic flux toward the TDP-l-mycarose via CRISPRi (strategy 3). Furthermore, repression of the glycolysis pathway was applied to enhance the endogenous TDP-l-mycarose pool and facilitate MEB concentration (strategy 4).
Fig. 1
Metabolic strategies for improving the 3-O-α-mycarosylerythronolide B (MEB) production in E. coli. The scheme represents key metabolic pathways, metabolites and genes involved in the biosynthesis of MEB. Orange-coloured arrows indicate the main metabolic pathways of MEB; black-coloured arrows indicate the native metabolic pathways in E. coli; green-coloured arrows indicate the overexpression of the heterologous TDP-l-mycarose from Saccharopolyspora erythraea; the blue rectangle with dash line indicate the CRISPRi system based on l-arabinose. the pink circles indicate deletion of the genes; the green up arrows indicate the overexpression of genes; 6-dEB, 6-deoxyerythronolide B; EB, erythronolide B; MEB, 3-O-α-mycarosylerythronolide B; PPP, pentose phosphate pathway; G6P, glucose-6-phosphate; F6P, fructose-6-phosphate; PEP, phosphoenolpyruvate; PYR, pyruvate; G1P, glucose-1-phosphate; TDPG, TDP-glucose; TKDG, TDP-4-keto-6-deoxy-d-glucose; TDR, TDP-4-dehydro-l-rhamnose; TR, TDP-l-rhamnose; TKDA, TDP-4-oxo-2,6-dideoxy-d-allose; TMKDG, TDP-3-methyl-4-oxo-2,6-dideoxy-d-glucose; PKS, polyketide synthetase; eryF, C6-hydroxylase; eryBV, l-mycarosyltransferase; zwf, glucose-6-phosphate dehydrogenase; pgi, phosphoglucose isomerase; pykA, pyruvate kinase II; pykF, pyruvate kinase I; pgm, phosphoglucomutase; yihX, α-d-glucose-1-phosphate phosphatase; rfbA, glucose-1-phosphate thymidylyltransferase; rfbB, dTDP-glucose-4,6-dehydratase; rmlC, TDP-4-dehydrorhamnose-3,5-epimerase; rfbD, TDP-l-rhamnose synthase.
Metabolic strategies for improving the 3-O-α-mycarosylerythronolide B (MEB) production in E. coli. The scheme represents key metabolic pathways, metabolites and genes involved in the biosynthesis of MEB. Orange-coloured arrows indicate the main metabolic pathways of MEB; black-coloured arrows indicate the native metabolic pathways in E. coli; green-coloured arrows indicate the overexpression of the heterologous TDP-l-mycarose from Saccharopolyspora erythraea; the blue rectangle with dash line indicate the CRISPRi system based on l-arabinose. the pink circles indicate deletion of the genes; the green up arrows indicate the overexpression of genes; 6-dEB, 6-deoxyerythronolide B; EB, erythronolide B; MEB, 3-O-α-mycarosylerythronolide B; PPP, pentose phosphate pathway; G6P, glucose-6-phosphate; F6P, fructose-6-phosphate; PEP, phosphoenolpyruvate; PYR, pyruvate; G1P, glucose-1-phosphate; TDPG, TDP-glucose; TKDG, TDP-4-keto-6-deoxy-d-glucose; TDR, TDP-4-dehydro-l-rhamnose; TR, TDP-l-rhamnose; TKDA, TDP-4-oxo-2,6-dideoxy-d-allose; TMKDG, TDP-3-methyl-4-oxo-2,6-dideoxy-d-glucose; PKS, polyketide synthetase; eryF, C6-hydroxylase; eryBV, l-mycarosyltransferase; zwf, glucose-6-phosphate dehydrogenase; pgi, phosphoglucose isomerase; pykA, pyruvate kinase II; pykF, pyruvate kinase I; pgm, phosphoglucomutase; yihX, α-d-glucose-1-phosphate phosphatase; rfbA, glucose-1-phosphate thymidylyltransferase; rfbB, dTDP-glucose-4,6-dehydratase; rmlC, TDP-4-dehydrorhamnose-3,5-epimerase; rfbD, TDP-l-rhamnose synthase.
Materials and methods
Strains, plasmids and chemicals
E. coli DH10B was used for plasmid construction and the previously reported BAP1 [29] was used for the biosynthesis of polyketides EB and MEB. The compatible vectors pET21c and pCDFDuet-1 (Novagen, Germany) were used to express the heterologous gene of the TDP-l-mycarose. 4-(2-hydroxyethyl)-1-piperazi-neethanesuffonic acid (HEPES) used in fermentation was bought from Sangon (Shanghai). Authentic chemical standards 6-dEB, EB were prepared by our group. All restriction enzymes and DNA ligases were bought from NEB (New England Biolabs, USA).
Heterologous TDP-l-mycarose pathway construction and assembly
The plasmids related to TDP-l-mycarose were listed in Table 1. The TDP-l-mycarose pathway genes from Saccharopolyspora erythraea were previously constructed and stored by our lab. Genes AeeryBIII and AeeryBVII from Aeromicrobium erythreum were synthesized by Tongyong (Anhui, China) with codon optimization for E. coli (Table S1) and cloned into pET21c yielding plasmid pZF90 and pZF91. Gene AeeryBIII was then inserted into the pZF91 between the Spe I/Sac I to result in pZF92. The chaperone GroESL coding sequence was obtained through PCR with the primers 224_F/R (Table S2) and the template pZF223, and then inserted into Spe I/Sac I site of pZF84 to generate pZF93.The pZF94 was generated by inserting the DNA fragment containing gene eryBII into pJF26 between Spe I/Sac I site. SaeryBVI and SaeryBII expression cassette was obtained by digestion pZF94 with restriction enzyme Xba I and Sac I, and then inserted into pZF93 between Spe I and Sac I to give pZF95. Subsequently, SaeryBVII_SaeryBIII and AeeryBVII_AeeryBIII expression cassettes were obtained by digesting pJF33 and pZF92 with restriction enzyme Xba I and Sac I, respectively, and then inserted into pZF95 between Spe I and Sac I to generate pZF227 and pZF228. Finally, the DNA fragment containing genes SaeryBIV, SaeryBV and ermE was obtained after the digestion of pJF37 with restriction enzyme Xba I and Sac I, and then constructed into pZF227 and pZF228, respectively, creating the corresponding pZF229 and pZF230 (Fig. S3).
CRISPR/Cas9-mediated knockout of chromosomal genes
The knockout of chromosomal genes in E. coli BAP1 was conducted by CRISPR/Cas9 system [30]. The sgRNA plasmid pZF9 was obtained from pCB003 by inverse PCR utilizing primer pairs pCB003_N20_pgi_F/R and its sequence was confirmed by sequencing. Similarly, other plasmids pZF10 and pZF11 were constructed with primers pCB003_N20_zwf_F/R and pCB003_N20_yihX_F/R, respectively (Table S2). The upstream (h1) and downstream (h2) homologous arms of the target genes (pgi, zwf, yihX) with the length of about 300-bp were separately amplified and then generated the donor DNA fragments by overlap PCR (Table S2). The PCR products were purified by gel extraction before electroporation. For the electroporation, 100 μL of E. coli BAP1 competent cells harboring pCB006 were prepared and mixed with 1000 ng donor DNA and 200 ng sgRNA plasmid. Electroporator (Bio-Rad, USA) was used for electroporation (1 cm cuvette, 1.8 kV). Cells were resuspended in 1 mL Luria Broth (LB) medium and recovered at 30 °C for 2 h before being plated onto LB agar containing kanamycin (50 mg/L) and spectinomycin (50 mg/L). The recombinant colonies were verified by colony PCR and DNA sequencing after incubating at 30 °C overnight. The individual colony edited successfully was inoculated into 2 mL of LB medium containing kanamycin (50 mg/L) and IPTG (0.5 mM) to cure the sgRNA plasmid, and the pCB006 could be eliminated when cell cultures were cultivated at 42 °C for 12 h.
Creation of plasmids for genes rfbA and rfbB overexpression
The rfbA and rfbB genes were amplified from the genomic DNA of E. coli BAP1 with primer pairs 225_rfbA_F/R, 225_rfbB_F/R (Table S2) and then fused by overlap PCR to generate rfbAB. The purified rfbAB fragments were assembled with pET21c treated with Nde I/Hind III using ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China), resulting in pZF225. The rfbAB expression cassette was amplified from template pZF225 with primer pairs FseI_rfbB/PacI_rfbA and then inserted into pZF230 under Fse I/Pac I sites to generate pZF234.
Construction of the CRISPRi-mediated system
For the construction of the l-arabinose-based CRISPRi system, the sgRNA cassette sequences were amplified using primers 236_vector_F/R (Table S2) from pZF208. The Para was amplified from pACYC-dCas9-Ter using primers 236_ara_F/R. These two PCR products were combined to make pZF236 using ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). For the silencing of different endogenous genes (such as rmlC, rfbD, pykA, and pykF), the 20-bp guide sequences were obtained using predictions from ATUM's gRNA design tool (https://www.atum.bio/) and then inserted into pZF236 by PCR using primers 237_F/R, 238_F/R, 243_F/R, and 244_F/R (Table S2), generating four sgRNA plasmids pZF237, pZF238, pZF243, and pZF244. To produce vector pZF246 which express multiple guides and dCas9 under the control of individual Para, the pZF243 and pZF244 were used as templates to obtain the fragments which contain Bsa I restriction site using primers 246_rmlC_F/R and 246_rfbD_F/R. The Bsa I sites of pACYC-dCas9-Ter backbone were introduced by PCR using primers ACYC_F/R. Then, all fragments were digested with Bsa I (NEB, USA), and ligated to yield pZF246. The pZF237 and pZF238 were used as templates to obtain the fragments which contain Bsa I restriction site using primers 239_pykA_F/R and 239_pykF_F/R, then all fragments were digested with Bsa I (NEB, USA) and ligated to yield pZF239.
Media and culture conditions
Fermentation medium, LB (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) supplemented with 15 g/L glycerol, 100 mM 4-(2-hydroxyethyl)-1-piperazi-neethanesuffonic acid (HEPES), was adjusted to pH 7.6 by NaOH before autoclaving and used to compare the production of engineering strains. For the biosynthesis of MEB, 100 μL of seed inoculum was inoculated into a 100 mL flask containing 10 mL fermentation medium supplemented with appropriate antibiotics (100 mg/L ampicillin, 50 mg/L kanamycin, 50 mg/L spectinomycin, and 34 mg/L chloramphenicol) and grown at 37 °C. Isopropyl β-d-thiogalactopyranoside (IPTG) and sodium propionate were added at a final concentration of 0.5 mM and 5 mM when OD600 reached 0.4. l-arabinose at a final concentration of 10 mM was added to induce the CRISPRi system when OD600 = 2. Cell cultures were subsequently incubated at 22 °C for 7 days.
HPLC and LC-MS/MS analytic methods
All samples were analyzed by high-performance liquid chromatography (HPLC) on an Ultimate 3000 HPLC system (ThermoFisher Scientific) with ELSD detector (Alltech U3000, Agilent) and a SilGreen ODS column (φ 4.6 × 250 mm, S-5 μm, Greenherbs, Beijing, China) maintained at 30 °C. Compounds were separated by acetonitrile (solvent A) and water (containing 50 mM ammonium formate, solvent B) at a flow rate of 1.0 mL/min under the following conditions: 0 min:100% B; 0–30 min: linear-gradient increase to 95% A in 5% B; 30–31 min: linear-gradient increase to100% B; 31–35 min: 100% B. The data shown in this study were generated from three independent experiments.LC-MS/MS was performed on Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (Thermo Scientific, MA, U.S.A.) equipped with Infinity Lab Poroshell 120 SB-AQ C18 column (φ 3.0 × 100 mm, 2.7 μm, Agilent, U.S.A.). The mobile phase was acetonitrile (A) and H2O with 0.1% formic acid (B). A linear gradient was set as follows: 5–95% solvent A for 10 min; 95% solvent A for 1min; 95-5% solvent A for 5 min. The flow rate was 0.4 mL/min, and the injection volume was 2 μL. The mass acquisition was performed in positive ionization mode with a full scan (100–1000).
Purification and quantification of 3-O-α-mycarosylerythronolide B
To obtain high purity MEB, collected samples were first separated by column chromatography over SiliaSphere C18 (50 μm, Silicycle, QuébecK, QC, Canada), and then purified by semi-preparative HPLC (Dionex UltiMate 3000 Semi-Preparative HPLC Systems, Thermo Scientific, MA, U.S.A.) with 40% acetonitrile in water (flow rate of 10 mL/min, detected at 205 nm) and a SilGreen ODS C18 column (φ 20 × 250 mm, 5 μm, Greenherbs Co., Ltd., Beijing, China). 1H and 13C and 2D NMR spectra of MEB were recorded by an Avance DRX 400 (500 MHz for 1H, 125 MHz for 13C) spectrometer (Bruker, Germany). The standard curve of MEB was generated by calculating the peak area of MEB.
Results and discussion
Establishing the biosynthesis of 3-O-α-mycarosylerythronolide B in E. coli
To rebuild the biosynthesis of 3-O-α-mycarosylerythronolide B (MEB) in E. coli, the previously reported BAP1 harboring pBP130 and pBP144 was used as the starting strain (WT) to produce 6-dEB. Subsequently, we performed the expression analysis of the pathway genes of TDP-l-mycarose. As shown in Fig. S1, except for EryBIV and ErmE, the other six proteins could not be expressed. It has been demonstrated that the chaperones GroEL/GroES achieved the highest titer for MEB through improving the protein solubility when the TDP-l-mycarose operon was coexpressed with chaperones GroEL/GroES, GroEL/GroES/TIG, GrpE/DnaJ/DnaK, and GroEL/GroES/GrpE/DnaJ/DnaK, respectively [25]. Therefore, plasmid pZF93 was constructed by incorporating the chaperones GroEL/GroES into the previous hydroxylase SaEryF-expressing plasmid pZF84. All the genes involved in the biosynthesis and transfer of TDP-l-mycarose were cloned from S. erythraea and were assembled into plasmid pZF93 to generate pZF229. The recombinant strain BAP229 (transforming pZF229 into strain WT) was cultivated in shake flask for 168 h, together with strain BAP93 (introducing pZF93 into strain WT) being used as a control for TDP-l-mycarose biosynthesis. The precursor EB (2) was detected in the fermentation media of strain BAP93, whereas newly appeared compounds 3 (Rt = 18.2 min) and 4 (Rt = 18.8 min) were observed in the fermentation broth of BAP229 (Fig. 2a). The HPLC-MS/MS analysis indicated that compounds 3 and 4 might be the isomerized MEB derivatives lacking the C3′ methyl group according to the identical characteristic ion peaks of [M + H–H2O]+ (m/z = 515.3232), which was 14 mass units less than that of MEB (m/z = 529.3389) (Fig. 2b and c). A careful comparison of 1H and 13C NMR data of compounds 3 and 4 with those of 2 (Figs. S2–S7) suggested that 3 and 4 were the glycosylated derivatives of 2. This deduction was supported by 1H–1H COSY cross-peaks of H1′/H2′/H3′/H4′/H5′/H6′ (δH 4.99/1.64/3.78/2.97/3.88/1.26 in 3, and 4.98/1.95/3.96/3.26/4.15/1.28 in 4) and the HMBC correlations from H1′ to C3 (δC 88.78 in 3, and 88.36 in 4) and C3′ (δC 69.39 in 3, and 67.94 in 4), H3′ to C2′ (δC 39.30 in 3, and 36.39 in 4), C4′ (δC 78.69 in 3, and 73.51 in 4) and C5′ (δC 70.20 in 3, and 67.70 in 4) as well as H6′ to C4′ and C5′ (Figs. S8–S15). The large coupling constants (J = 9.6 Hz) of H4′, H5′ and H6′ in 3 indicated that the substituents of C3′-C5′ are equatorial, while the small values of 3JH1′,H2′a (3.0 Hz) and 3JH2′a,H3′ (3.0 Hz) in 4 suggested that the substituents of C1′ and C3′ are axial. Therefore, compounds 3 and 4 were assigned as 3-O-(2″,6″-dideoxy-α-l-arabino-hexopyranosyl) erythronolide B and 3-O-(2″,6″-dideoxy-α-l-ribo-hexopyranosyl) erythronolide B, respectively, which were reported previously in the mutant S. erythraea [31].
Fig. 2
The biosynthesis of 3-O-α-mycarosylerythronolide B (MEB) in E. coli. a, The HPLC analysis of the fermentation product of strains BAP229 (BAP1 harboring pBP130, pBP144, and pZF229) and BAP230 (BAP1 harboring pBP130, pBP144, and pZF230). ST-6-dEB, 6-dEB standard; ST-EB, EB standard; 1, 6-dEB; 2, EB; 3, 3-O-(2″,6″-dideoxy-α-l-arabino-hexopyranosyl) erythronolide B; 4: 3-O-(2″,6″-dideoxy-α-l-ribo-hexopyranosyl) erythronolide B; 5: MEB; b, The LC-MS/MS fragments of compound 3. c, The LC-MS/MS fragments of compound 4. d, The LC-MS/MS fragments of compound 5.
The biosynthesis of 3-O-α-mycarosylerythronolide B (MEB) in E. coli. a, The HPLC analysis of the fermentation product of strains BAP229 (BAP1 harboring pBP130, pBP144, and pZF229) and BAP230 (BAP1 harboring pBP130, pBP144, and pZF230). ST-6-dEB, 6-dEB standard; ST-EB, EB standard; 1, 6-dEB; 2, EB; 3, 3-O-(2″,6″-dideoxy-α-l-arabino-hexopyranosyl) erythronolide B; 4: 3-O-(2″,6″-dideoxy-α-l-ribo-hexopyranosyl) erythronolide B; 5: MEB; b, The LC-MS/MS fragments of compound 3. c, The LC-MS/MS fragments of compound 4. d, The LC-MS/MS fragments of compound 5.We speculated that the generation of MEB derivatives instead of MEB might be attributed to the low enzymatic activity of SaeryBVII (TDP-4-keto-2,6-dideoxyhexose 3-C-methyltransferase) and SaeryBIII (TDP-4-deoxyglucose 3,5-epimerase). To verify this hypothesis, an assessment of protein expression of the pathway enzymes was conducted in E. coli BAP1 with the aid of chaperones GroEL/GroES. As we anticipated, with the exception of SaeryBVII and SaeryBIII, all TDP-l-mycarose pathway enzymes could be highly expressed (Fig. S16), which indicated that the soluble expression of eryBVII and eryBIII exerted an important effect on the biosynthesis of MEB.To facilitate the creation of TDP-l-mycarose, the homologous AeeryBVII and AeeryBIII originated from A. erythreum were selected and synthesized with optimized codons. Protein expression analysis showed that both AeeryBVII and AeeryBIII gave distinct expression bands under the same cultivation condition (Fig. S17), which were utilized to replace the SaeryBVII and SaeryBIII of plasmid pZF229, resulting in plasmid pZF230 (Fig. S18). HPLC analysis clearly revealed the presence of 6-dEB (1) and a new compound 5 (Rt = 20.4 min) in the fermentation media of strain BAP230 generated by introducing pZF230 into strain WT (Fig. 2a). Compound 5 was subsequently identified as MEB by the characteristic ion peaks m/z 529.3389 observed in the HPLC-MS/MS mass profile and NMR spectra (Fig. 2d, S19 and S20). This result indicated the feasibility to achieve the biosynthesis of MEB in E. coli by combining pathway enzymes with effective expression. Eventually, the MEB concentration of BAP230 was quantified to be 4.2 mg/L on the basis of the established standard curve of MEB (Fig. S21).
Enhancing the 3-O-α-mycarosylerythronolide B production via pathway disruption
Given that the unusual sugar TDP-l-mycarose is the crucial biosynthetic bottleneck of MEB [28], it is essential to improve the yield of MEB by enhancing the intracellular TDP-l-mycarose pool. Glucose-6-phosphate (G6P) is the common precursor of glycolysis, pentose phosphate pathway and TDP-l-mycarose biosynthetic pathway. To provide more G6P for the synthesis of TDP-l-mycarose, glycolysis and pentose phosphate pathway that consume G6P need to be blocked. It has been demonstrated that deletion of genes pgi (encoding phosphoglucose isomerase) and zwf (encoding glucose-6-phosphate dehydrogenase) could improve the level of UDP-glucose and elevate the production of the corresponding glycosylated products [[32], [33], [34]]. Moreover, the G1P hydrolase encoded by gene yihX was capable of selectively hydrolyzing G1P, which is the intermediator of TDP-l-mycarose biosynthesis [35]. To reinforce the TDP-l-mycarose biosynthetic pathway and increase the MEB production, we individually knocked out genes pgi, zwf, and yihX in E. coli BAP1, generating strains ZF1(BAP1Δpgi), ZF2(BAP1Δzwf), and ZF3(BAP1ΔyihX) (Fig. 3a). Based on this, three plasmids pBP130, pBP144, and pZF230 responsible for the biosynthesis of MEB were introduced into ZF1, ZF2, and ZF3, respectively, yielding strains QC1, QC2, and QC3. QC2 and QC3 exhibited no growth defect compared with the parent strain BAP230 (OD600 = 13), while QC1 (OD600 = 8.7) was impaired to a certain extent. The parental strain BAP230 could produce 8.6 mg/L 6-dEB and 4.2 mg/L MEB with no detectable accumulation of EB, while the engineered QC2 and QC3 produced 12.3 mg/L and 10.7 mg/L MEB, a 2.9-fold and 2.5-fold to that produced by BAP230. Interestingly, the titer of MEB (5.1 mg/L) in QC1 was lower than QC2 and QC3, but it enabled the EB production to 13.4 mg/L EB (Fig. 3b). These results indicated that the single-gene knockout of the bypass pathway contributed to enhancing MEB production. Thus, we move on to investigate whether the combinatorial deletion of two genes could further elevate the MEB production. Three engineered strains QC7, QC8, and QC9 were subsequently constructed by introducing pBP130, pBP144, and pZF230 into strains ZF7 (BAP1ΔpgiΔzwf), ZF8 (BAP1ΔzwfΔyihX), and ZF9 (BAP1ΔpgiΔyihX), respectively. As shown in Fig. 3b, the cell growth of QC7 (OD600 = 7.9) and QC9 (OD600 = 8.3) was similar to that of strain QC1, while QC8 (OD600 = 13.1) showed no growth difference with wild-type strain BAP230. The 6-dEB and MEB concentrations in strain QC7 were 24.7 mg/L and 35.1 mg/L, increasing significantly by 190% and 740% compared to strain BAP230, respectively. The strain QC8 generated 12.9 mg/L 6-dEB and 14.0 mg/L MEB. Compared with strain BAP230, the titers of 6-dEB and MEB have no significant change accompanied with the accumulation of EB in strain QC9, suggesting that engineered E. coli drive more metabolic flux towards EB biosynthesis and the availability of endogenous TDP-l-mycarose was limited. Hence, we attempted to disrupt genes pgi, zwf, and yihX simultaneously and construct the recombinant strain ZF13 (BAP1ΔpgiΔzwfΔyihX) (Fig. S22). The corresponding fermentation strain QC13 afforded the highest MEB production with a titer of 41.2 mg/L, which is a 9.8-fold increase to that produced by BAP230. In addition, QC13 also achieved the highest production of 24.5 mg/L 6-dEB, which indicated the efficient downstream pathways utilizing 6-dEB as building unit or biosynthetic precursor might lead to the improved metabolic flux toward 6-dEB. These results suggested that increasing the carbon flux at the G1P node by blocking the competing pathway could drive the biosynthesis of TDP-l-mycarose and MEB. Therefore, strain ZF13 was chosen for further engineering.
Fig. 3
Effects of the gene knockout on the cell growth and MEB production. a, The strategy 1 to enhance the MEB production by genes deletion. b, The 6-dEB, EB, and MEB production and OD600 of recombinant strains BAP230 (BAP1 harboring pBP130, pBP144, and pZF230), QC1 (ZF1 harboring pBP130, pBP144, and pZF230), QC2 (ZF2 harboring pBP130, pBP144, and pZF230), QC3 (ZF3 harboring pBP130, pBP144, and pZF230), QC7 (ZF7 harboring pBP130, pBP144, and pZF230), QC8 (ZF8 harboring pBP130, pBP144, and pZF230), QC9 (ZF9 harboring pBP130, pBP144, and pZF230) and QC13 (ZF13 harboring pBP130, pBP144, and pZF230).
Effects of the gene knockout on the cell growth and MEB production. a, The strategy 1 to enhance the MEB production by genes deletion. b, The 6-dEB, EB, and MEB production and OD600 of recombinant strains BAP230 (BAP1 harboring pBP130, pBP144, and pZF230), QC1 (ZF1 harboring pBP130, pBP144, and pZF230), QC2 (ZF2 harboring pBP130, pBP144, and pZF230), QC3 (ZF3 harboring pBP130, pBP144, and pZF230), QC7 (ZF7 harboring pBP130, pBP144, and pZF230), QC8 (ZF8 harboring pBP130, pBP144, and pZF230), QC9 (ZF9 harboring pBP130, pBP144, and pZF230) and QC13 (ZF13 harboring pBP130, pBP144, and pZF230).
Overexpression of rfbA and rfbB for 3-O-α-mycarosylerythronolide B production
Glucose-1-phosphate thymidylyltransferase (rfbA) and TDP-glucose-4,6-dehydratase (rfbB) are capable of converting G1P to TKDG, a crucial intermediate of TDP-l-mycarose (Fig. 4a). To further increase the MEB titer, we sought to reinforce the TDP-l-mycarose biosynthetic pathway via overexpression of rfbA and rfbB. Accordingly, the rfbA_rfbB module was cloned from E. coli and incorporated on the pZF230 as an independent operon, yielding plasmid pZF234 (Fig. 4b). The shake flask fermentation results showed that the MEB production of strain QC234 (ZF13 harboring pBP130, pBP144, pZF234) was slightly increased and reached a maximum of 48.3 mg/L, a 17% increase relative to strain QC13 (41.2 mg/L). Intriguingly, strain QC234 produced 9.6 mg/L of 6-dEB, which is comparable to that of QC13 (Fig. 4c).
Fig. 4
The effects of overexpressing rfbA and rfbB on MEB production. a, The strategy 2 to reinforce the MEB production. b, Schematic diagram of pZF234. c, Cell growth and 6-dEB and MEB production of QC13 (ZF13 harboring pBP130, pBP144, and pZF230) and QC234 (ZF13 harboring pBP130, pBP144, and pZF234).
The effects of overexpressing rfbA and rfbB on MEB production. a, The strategy 2 to reinforce the MEB production. b, Schematic diagram of pZF234. c, Cell growth and 6-dEB and MEB production of QC13 (ZF13 harboring pBP130, pBP144, and pZF230) and QC234 (ZF13 harboring pBP130, pBP144, and pZF234).To verify whether the increase in MEB yield was due to the enhancement of TDP-l-mycarose, we measured the cellular concentration of TDP-l-mycarose in the engineered strains that exclusively synthesize the sugar skeleton. As shown in Fig. S23, the triple knockout strain sZF13 (pZF230) that was engineered to enhance the supply of G1P was capable of producing 107.2 mg/L TDP-l-mycarose, a 12.9-fold to that of the control strain BAP1 (pZF230) (8.3 mg/L), while the sZF13 (pZF234) which was created to enhance the supply of G1P and further convert the precursor G1P into the key intermediator TKDG achieved the highest titer of TDP-l-mycarose of 143.3 mg/L, a 16.3-fold increase to that produced by BAP1 (pZF230).
Regulating the metabolic pathway of E. coli with CRISPRi
Considering that genes overexpression involved in TDP-l-mycarose pathway led to a modest increase in MEB production, which might be due to the leakage of TKDG caused by TDP-l-rhamnose synthesis and the shortage of intracellular G1P, we next aim to improve MEB titer by addressing these problems. TDP-4-dehydrorhamnose-3,5-epimerase (rmlC) and TDP-l-rhamnose synthase (rfbD) were reported to catalyze TKDG to form TDP-l-rhamnose that played pivotal roles in membrane synthesis and cellular function [36]. To mitigate the leakage of TKDG, we implemented the CRISPRi system mediated by Para promoter in the established strain QC234 to downregulate the expression of these two endogenous genes rmlC and rfbD (Fig. 5a). Consequently, pZF246 was constructed by inserting two sgRNA cassettes targeting rmlC and rfbD into the dCas9-expressing plasmid pACYC-dCas9-Ter (Fig. S24a). Introduction of the pACYC-dCas9-Ter and pZF246 into the efficient MEB producer QC234 resulted in recombinant strain DTAC and DT246, respectively. Unexpectedly, the repression of TDP-l-rhamnose biosynthesis in strain DT246 (OD600 = 2.8) resulted in striking growth defect compared with strain QC234 (OD600 = 10.1) (Fig. 5b), whereas strain DTAC (OD600 = 8.0) showed slightly impaired growth. The reason that the titers of MEB in strains DTAC (44.4 mg/L) and DT246 (46.9 mg/L) were decreased could be ascribed to the compromised cell growth. Despite the application of CRISPRi exhibited no beneficial effects on MEB production, the MEB concentration per OD600 of DT246 was 3.5-fold to that of QC234.
Fig. 5
Effects of CRISPRi targeting to rmlC and rfbD on MEB production. a, The strategy 3 to repress the TDP-l-rhamnose biosynthesis. b, The cell growth and production of 6-dEB, EB, and MEB of recombinant strains QC234 (ZF13 harboring pBP130, pBP144, and pZF234), DTAC (ZF13 harboring pBP130, pBP144, pZF234, and pACYC-dCas9-Ter), and DT246 (ZF13 harboring pBP130, pBP144, pZF234, and pZF246).
Effects of CRISPRi targeting to rmlC and rfbD on MEB production. a, The strategy 3 to repress the TDP-l-rhamnose biosynthesis. b, The cell growth and production of 6-dEB, EB, and MEB of recombinant strains QC234 (ZF13 harboring pBP130, pBP144, and pZF234), DTAC (ZF13 harboring pBP130, pBP144, pZF234, and pACYC-dCas9-Ter), and DT246 (ZF13 harboring pBP130, pBP144, pZF234, and pZF246).Pyruvate kinases II (pykA) and I (pykF) are indispensable enzymes in the glycolysis pathway which have been widely investigated and engineered to rewire the carbon metabolism and facilitate the generation of nucleotide-activated sugar donor [34,37]. To further promote the formation of G1P, we thus reconstructed the CRISPRi system to reduce the expression of pykA and pykF by replacing the sgRNAs of pZF246, yielding pZF239 (Fig. S24b), which was transformed into strain QC234 to obtain DT239 (Fig. 6a). Although a slightly recovered cell growth was observed in DT239 (OD600 = 3.7) in comparison with DT246, simultaneous inhibition of pykA and pykF failed to accomplish the improvement of MEB production in strain DT239 (48.2 mg/L), nearly equal to the titer of QC234 (48.3 mg/L) (Fig. 6b). This might largely be attributable to the metabolite burden caused by the expression of multiple pathway genes and transcriptional regulators. In spite of the fact that the implementation of the CRISPRi resulted in no remarkable increase in the concentration of MEB, the higher biomass specific rate of MEB still demonstrated the functionality of CRISPRi system in driving carbon flux from G1P to TDP-l-mycarose.
Fig. 6
Effects of CRISPRi targeting to pykA and pykF on MEB production. a, The strategy 4 to repress the glycolysis pathway. b, The cell growth and production of 6-dEB, EB, and MEB of recombinant strains QC234 (ZF13 harboring pBP130, pBP144, and pZF234), DTAC (ZF13 harboring pBP130, pBP144, pZF234, and pACYC-dCas9-Ter), and DT239 (ZF13 harboring pBP130, pBP144, pZF234, and pZF239).
Effects of CRISPRi targeting to pykA and pykF on MEB production. a, The strategy 4 to repress the glycolysis pathway. b, The cell growth and production of 6-dEB, EB, and MEB of recombinant strains QC234 (ZF13 harboring pBP130, pBP144, and pZF234), DTAC (ZF13 harboring pBP130, pBP144, pZF234, and pACYC-dCas9-Ter), and DT239 (ZF13 harboring pBP130, pBP144, pZF234, and pZF239).
Conclusions
In summary, the multi-level metabolic engineering approach including gene disruption, gene overexpression and CRISPRi was successively performed to achieve high-level TDP-l-mycarose and boost MEB production. Notably, the strain QC234 producing 48.3 mg/L MEB and 9.6 mg/L 6-dEB, was constructed by deletion of pgi, zwf, and yihX and overexpression of rfbA and rfbB. The CRISPRi system was employed to repress bypass pathways that consume precursors, leading to a 250% increase in the titer of MEB per OD600 in DT246 compared with QC234. This study lays the foundation for de novo biosynthesis of erythromycin and other glycosylated products decorated by unusual sugar.
Authors: Peng Xu; Sridhar Ranganathan; Zachary L Fowler; Costas D Maranas; Mattheos A G Koffas Journal: Metab Eng Date: 2011-07-13 Impact factor: 9.783
Authors: Lei Fang; Guojian Zhang; Omar El-Halfawy; Max Simon; Eric D Brown; Blaine A Pfeifer Journal: Biotechnol Bioeng Date: 2018-09-15 Impact factor: 4.530
Authors: Salvador Peirú; Hugo G Menzella; Eduardo Rodríguez; John Carney; Hugo Gramajo Journal: Appl Environ Microbiol Date: 2005-05 Impact factor: 4.792
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6