Enhancing flavonoid production by systematically tuning the central metabolic pathways based on a CRISPR interference system in Escherichia coli

1Malonyl-coenzyme A (malonyl-CoA) is an important precursor metabolite involved in the biosynthesis of many important chemicals, such as flavanones, polyketides and fatty acids. However, in most microorganisms commonly used for the heterologous production of these chemicals, such as Escherichia coli and Saccharomyces cerevisiae, the endogenous central metabolism predominates and strongly competes for carbon sources and energy during malonyl-CoA synthesis, leaving only small amounts available for the production of recombinant products. Nevertheless, previous studies have made some exciting gains in improving the intracellular synthesis of malonyl-CoA through overexpression or deletion of specific genes.

2For example, overexpression of four acetyl-CoA carboxylase (ACC) subunits and biotin ligase from Photorhabdus luminescens or the malonate assimilation pathway from Rhizobium trifolii have significantly improved the availability of malonyl-CoA. However, overexpression of these genes would impose metabolic burdens on the cells and often requires supplementation of biotin or malonate in the medium, which is commercially unfavorable. Simultaneous deletion of the genes sdhA (succinate dehydrogenase), adhE (acetaldehyde dehydrogenase), brnQ (branched chain amino acid transporter) and citE (citrate lyase) or fumC (fumarate hydratase) and sucC (succinyl-CoA synthetase) also enhanced the malonyl-CoA concentration. While numerous metabolic engineering targets, such as genes essential for growth, are yet to be explored due to the disadvantages of conventional gene-knockout strategies. Furthermore, there has been no effort to evaluate the combinatorial effects of various distinct genetic interventions including those involving genes essential for growth.

3The use of RNA-mediated regulatory mechanisms for fine flux control has been well documented. Due to the inefficiency of RNA interference (RNAi) in bacteria, the clustered regularly interspaced short palindromic repeats interference (CRISPRi) system offers a potential alternative for targeted gene regulation in bacteria, as it uses small base-pairing RNAs to regulate gene expression in a sequence-specific manner. Compared with other sRNA strategies, the CRISPRi system can easily tune gene expression levels by directing sgRNA to different regions of the non-template DNA strand of target genes. It is believed that this RNA-guided genome regulation can identify unprecedented genetic perturbations and explore the combinatorial effects of multiple genetic manipulations.

4Flavonoids are valuable natural products widely used in human health and nutrition due to their biochemical properties, which include antiviral, anti-obesity and anti-cancer activities. (2S)-Naringenin is a common precursor of most flavonoids and has become a potential candidate for treating numerous human maladies. Formation of (2S)-naringenin from L-tyrosine occurs through the action of tyrosine ammonia lyase (TAL), 4-coumarate:CoA ligase (4CL), chalcone synthase (CHS) and chalcone isomerase (CHI). Every flavanone compound generated in E. coli requires 3 mol of malonyl-CoA, which imposes a significant metabolic burden on the host strain. Previous approaches to redirecting endogenous malonyl-CoA into heterologous pathways have relied heavily on overexpression or deletion of particular pathways.

5Here, we explored the impact of fine-tuning the central metabolic pathways by using a CRISPRi system to enhance heterologous pathway productivity, using the production of (2S)-naringenin as a model system. Genes involved in central metabolic pathways were repressed by the CRISPRi system to identify individual target genes that could increase the intracellular malonyl-CoA level by over 223%. Furthermore, the efficiency of repression of these target genes was tuned to balance malonyl-CoA generation and biomass accumulation (the final OD600 decreased by less than 10%). Finally, multiple gene repressing was performed to achieve a high yield of (2S)-naringenin (421.6 mg/L). This strategy enhances the malonyl-CoA concentration without the need to add substrates for malonyl-CoA generation, which potentially provides an economically sustainable process for the efficient production of other plant natural compounds.

6To implement the CRISPRi platform in E. coli, a catalytically dead Cas9 mutant (dCas9) derived from Streptococcus pyogenes cas9 gene, which acts as an RNA-guided DNA-binding complex, was expressed under T7 promoter. The sgRNA molecule, which consists of four domains (a Trc promoter, a 20-nucleotide (nt) complementary region for specific DNA binding, a 42-nt dCas9-binding hairpin and a 40-nt transcription terminator derived from S. pyogenes), was coexpressed with dCas9 (Fig. 1).

7Acetyl-CoA serves as the first flux control point for flavonoid biosynthesis, whereas malonyl-CoA serves as a starting point for the synthesis of flavonoids, which are only consumed for synthesizing fatty acids. To channel carbon flows toward acetyl-CoA, the availability of acetyl-CoA precursors such as pyruvate needs to be increased and hence, zwf (glucose-6-phosphate 1-dehydrogenase), pgl (6-phosphogluconolactonase), tpiA (triosephosphate isomerase), ppsA (phosphoenolpyruvate synthase), eno (phosphopyruvate hydratase), glyA (serine hydroxymethyltransferase) and fold (bifunctional 5,10-methylene-tetrahydrofolate dehydrogenase/5,10-methylene-tetrahydrofolate cyclohydrolase) were chosen as target genes. Consumption of acetyl-CoA in other central metabolic pathways, such as the TCA cycle and glycolysis, needs to be reduced and hence, mdh (malate dehydrogenase), fumC, sdhABCD (succinate dehydrogenase), sucCD (succinyl-CoA synthetase), sucA (2-oxoglutarate dehydrogenase), sucB (dihydrolipoamide acetyltransferase), acnAB (aconitate hydratase), citE (citrate lyase), gdhA (glutamate dehydrogenase), gltp (glutamate/aspartate:proton symporter) and adhE (acetaldehyde dehydrogenase) were also chosen (Fig. 2). Fatty acid biosynthesis related genes, such as fabH (3-oxoacyl-acyl carrier protein synthase III), fabB (3-oxoacyl-acyl carrier protein synthase I), fabF (3-oxoacyl-acyl carrier protein synthase II), fabG (3-oxoacyl-acyl carrier protein reductase), fabA (3-hydroxydecanoyl-acyl carrier protein dehydratase), fabI (enoyl-acyl carrier protein reductase) and fabD (malonyl-CoA-acyl carrier protein transacylase) were chosen as target genes (Fig. 2) and were also inhibited to prevent the diversion of malonyl-CoA to fatty acid synthesis.

8The efficacy of silencing is inversely correlated with the distance of the target gene from the translation start codon. A sequence of designed sgRNA firstly targeted the non-template DNA strand of a specific gene after the first protospacer adjacent motif (PAM) sequence (NGG) at the open reading frame (ORF). To identify potential genes related to the enhanced supply of intracellular malonyl-CoA and acetyl-CoA, thus confirm the efficiency of CRISPRi based technique, the changing pattern of intracellular malonyl-CoA and acetyl-CoA concentration were measured after silencing each target gene.

9The results showed that sgRNAs targeting ppsA, eno, glyA, adhE, mdh, fumC, sdhABCD, sucC and citE produced a dramatic increase in acetyl-CoA concentration (over 180%), while sgRNAs targeting ppsA, eno, adhE, mdh, fumC, sdhA, sucC and citE produced a dramatic increase in malonyl-CoA concentration (over 223%). It was also found that sgRNAs targeting fabH, fabB, fabF and fabI produced a simultaneous increase in acetyl-CoA and malonyl-CoA concentration (over 244%) (Fig. 3). Hence, ppsA, eno, adhE, mdh, fumC, sdhA, sucC, citE, fabH, fabB, fabF and fabI were chosen as target genes.

10Acetyl-CoA and malonyl-CoA are important metabolic intermediates involved in central metabolic pathways, such as glycolysis, the TCA cycle and fatty acid biosynthesis. Reducing the acetyl-CoA or malonyl-CoA pool available to these pathways would decrease cell growth. To examine whether fine-tuning target gene expression levels using the CRISPRi system could balance the metabolic flux between cell growth and malonyl-CoA accumulation, three sgRNA variants with different efficiencies for repressing every target gene were generated.

11To achieve a high, medium or low repressing efficacy toward each target gene, sgRNA was designed to bind the non-template DNA strand of the target gene at the initial, middle or terminal region. High repressing efficacy toward fabF increased the malonyl-CoA concentration by 433.3% and decreased the final biomass by 9% (Fig. 4). Medium repressing efficacy toward sucC and fumC increased the malonyl-CoA concentration by 222.4% and 166.8% and decreased the final biomass by 9.1% and 7.9%, respectively. Low repressing efficacy toward eno, adhE, mdh and fabB increased the malonyl-CoA concentration by 77.8%, 222.2%, 244.4% and 111.1% and decreased the final biomass by 6.4%, 7.3%, 9.9% and 8.2%, respectively (Fig. 4). SgRNAs targeting other genes resulted in dramatic decreases in the final biomass (by over 54.5%) (Fig. 4). Hence, eno, adhE, mdh, fabB, fabF, sucC and fumC were chosen as the optimal target genes.

12Based on a previous study, the control strain overproducing (2S)-naringenin from L-tyrosine was constructed. The plasmids pCDF-Trc-TAL-Trc-4CL and pET-CHS-CHI were transformed into E. coli BL21 (DE3), which yielded a production titer of 50.5 mg/L. Appropriate repressing efficacy toward a target gene that could increase the malonyl-CoA concentration without significantly altering the final biomass was ranked as a beneficial genetic perturbation. The impacts of these independent genetic interventions, namely low repressing efficacy toward eno, adhE, mdh and fabB, medium repressing efficacy toward sucC and fumC and high repressing efficacy toward fabF, on (2S)-naringenin production were analyzed. Single perturbations of eno, adhE, mdh, fabB, fabF, sucC and fumC increased production by up to 38.6%, 74.3%, 86.1%, 54.5%, 135.6%, 78.2% and 96.1%, respectively, and perturbation of fabF resulted in the highest (2S)-naringenin production (119.6 mg/L) (Fig. 5).

13The impacts of different synthetic sgRNA combinations on (2S)-naringenin production were evaluated. Firstly, the effect of two genetic perturbations on (2S)-naringenin production was investigated. Anti-fabF sgRNA in combination with anti-eno, anti-adhE, anti-fabB, anti-sucC or anti-fumC sgRNA but not anti-mdh produced a higher (2S)-naringenin titer than single anti-fabF sgRNA. The latter result is probably due to an imbalance between cell growth and product formation because it was observed that the strain anti-fabF/mdh resulted in the lowest cell growth (Fig. 5). Among these sgRNA combinations, anti-fabF/fumC sgRNA produced the highest (2S)-naringenin titer (220.1 mg/L).

14Secondly, the effect of anti-fabF/fumC sgRNA in combination with anti-eno, anti-adhE, anti-fabB or anti-sucC sgRNA was investigated. Each sgRNA combination produced a higher (2S)-naringenin titer than anti-fabF/fumC, and anti-fabF/fumC/fabB produced the highest (2S)-naringenin titer (298.8 mg/L) (Fig. 5). Similarly, anti-fabF/fumC/fabB sgRNA in combination with anti-eno, anti-adhE or anti-sucC sgRNA produced a higher (2S)-naringenin titer than anti-fabF/fumC/fabB sgRNA, and anti-fabF/fumC/fabB/sucC sgRNA produced the highest (2S)-naringenin titer (398.3 mg/L). Anti-fabF/fumC/fabB/sucC sgRNA in combination with anti-eno, or anti-adhE sgRNA produced a higher (2S)-naringenin titer than anti-fabF/fumC/fabB/sucC sgRNA, and anti-fabF/fumC/fabB/sucC/adhE sgRNA produced the highest (2S)-naringenin titer (421.6 mg/L). However, anti-fabF/fumC/fabB/sucC/adhE/eno sgRNA produced a lower (2S)-naringenin titer (384.5 mg/L) than anti-fabF/fumC/fabB/sucC/adhE sgRNA (Fig. 5).

15Identification of potential genetic targets and optimization of their expression are essential for the efficient production of desired metabolites. However, the number of genes that can be manipulated through the traditional gene disruption approach is limited, especially when target genes are critical for growth. There is an urgent requirement for a tool that enables genome-scale identification of suitable gene targets for rational engineering. In this study, the CRISPRi system was employed to efficiently regulate the expression of targeted genes involved in central metabolic pathways, such as glycolysis, the TCA cycle or fatty acid synthesis (Fig. 3). The strategy reported here is advantageous compared to conventional gene-knockout strategies for large-scale screening and manipulating genes essential for growth.

16There has been increasing interest in the use of RNA-mediated regulatory mechanisms for target gene regulation. Previously, antisense RNA was employed to simultaneously repress fabB and fabF to enhance the intracellular malonyl-CoA concentration. However, numerous metabolic engineering targets are yet to be explored and there has been no effort to evaluate the combinatorial effects of various distinct genetic interventions. Here, genes involved in central metabolic pathways were explored to identify target genes. Furthermore, combinatorial effects of various genetic interventions were evaluated to obtain the best combination.

17Previous studies have made significant gains in improving the intracellular malonyl-CoA concentration and achieved a high yield of flavonoids through: (1) the ACC route, which requires coordinated overexpression of four acetyl-CoA carboxylase (ACC) subunits from Photorhabdus luminescens and biotin ligase from the same species; (2) the malonate assimilation pathway, which requires the malonate assimilation pathway from Rhizobium trifolii. However, overexpression of these genes would impose a metabolic burden on cells and often requires supplementation of biotin or malonate in the medium, which is commercially unfavorable. In this study, metabolic circuits were rewired by multiple and simultaneous perturbations, through reducing flux to undesirable by-products and enhancing precursor and cofactor flux, the target compound (2S)-naringenin production without adding biotin or malonate was maximizing.

18In many cases, simultaneous modification of the expression levels of many genes is necessary to improve the production of target molecules. However, previous studies have found that the impact of multiple genetic interventions on production is not additive. Simple stacking of genetic interventions led to only limited increases in recombinant production. Here, by coupling genetic modification to cell growth, it was found that, except for repressing mdh which resulted in the lowest cell growth, the combined effects of other genetic modifications on production were additive (Fig. 5). This demonstrated that coupling genetic modification to cell growth when scoring the fitness of a genotype is critical to identifying beneficial interventions.

19Although malonyl-CoA is the limiting factor for flavonoid production, many studies have demonstrated that balance in the synthetic pathway is vital for productivity. Accumulation of intermediate metabolites may result in suboptimal product titers, as further increases in these intermediates cannot be accommodated by the capacity of the downstream pathway. Here, it was found that combining all of the positive genetic interventions resulted in a decrease in (2S)-naringenin production compared to other combinations (Fig. 5). Therefore, systematically repressing target genes to approach the limit of malonyl-CoA saturation generates the optimal result.

20In conclusion, this CRISPRi system is advantageous over traditional gene-knockout strategies because of its ability to manipulate genes essential for growth. Furthermore, because of its high specificity when regulating a single gene or multiple genes, this RNA-guided system is more productive than antisense RNA strategies. Moreover, the CRISPRi strategy, which couples genetic modification to cell growth to identify beneficial interventions and systematically combines genetic interventions to approach the limit of malonyl-CoA saturation, could be employed for the efficient production of many other malonyl-CoA derived compounds, such as the flavonoids, fatty acids and polyketides.

21E. coli JM109 was used for plasmid propagation. E. coli BL21 (DE3) was used to express sgRNA and produce flavonoids. The compatible vector set (pETDuet-1, pCDFDuet-1, pCOLADuet-1, and pACYCDuet-1) was used to express multiple genes in one strain (Novagen, Darmstadt, Germany). Luria broth (LB) (10 g/L tryptone, 5 g/L yeast extract and 10 g/L NaCl) was used for strain construction. MOPS minimal medium supplemented with 5 g/L glucose and an additional 4 g/L NH4Cl was used for sgRNA expression and flavonoid production. Ampicillin (100 μg/mL), kanamycin (40 μg/mL), chloramphenicol (20 μg/mL), and streptomycin (40 μg/mL) was added when required. Cell growth was monitored by measuring the absorbance at 600 nm (OD600) with a UV/Vis spectrophotometer (UVmini-1240, Shimadzu, Kyoto, Japan). The primers and plasmids used in this study are listed in Tables 1, 2 and 3.

22A catalytically dead Cas9 mutant (dCas9) was codon-optimized and synthesized by GenScript (Nanjing, China) (see Supplementary Sequences for the dCas9 sequence). pACYC-dCas9 was constructed by digesting dCas9 from pUC57-dCas9 (GenScript, Nanjing, China) into NcoI/HindIII sites of pACYCDuet-1. The sgRNA chimera, which consists of four domains (a Trc-inducible promoter, a 20-nucleotide (nt) complementary region for specific DNA binding, a 42-nt dCas9-binding hairpin and a 40-nt transcription terminator derived from Streptococcus pyogenes) was synthesized by GenScript (Nanjing, China) and inserted into pfoI/BamHI sites of pCOLADuet-1 (see Supplementary Sequences for sgRNA chimera sequence). This resulted in the plasmid pCOLA-fabF(high), which was the template for PCR-based mutagenesis. PCR-based site-directed mutagenesis (TaKaRa MutanBEST Kit, Takara Biotechnology, Dalian, China) with the corresponding primer pairs was used to generate sgRNA cassettes with new 20-bp complementary regions. Oligonucleotides used to generate sgRNA cassettes and the resultant sgRNA expression vectors are listed in Table 1.

23To repress multiple genes, different sgRNA sequences including the Trc promoter, 20-bp complementary region, 42-bp dCas9-binding hairpin and 40-bp transcription terminator were cloned into the same plasmid. The primer pair Pf_sgRNA(BamHI)/Pr_sgRNA(EcoRI) was used to amplify anti-adhE, anti-eno, anti-mdh, anti-fabB, anti-fumC and anti-sucC sgRNA sequences into the BamHI/EcoRI sites of pCOLA-fabF(high). This resulted in the plasmids pCOLA-fabF(high)/adhE(low), pCOLA-fabF(high)/eno(low), pCOLA-fabF(high)/mdh(low), pCOLA-fabF(high)/fabB(low), pCOLA-fabF(high)/fumC(medium) and pCOLA-fabF(high)/sucC(medium). The primer pair Pf_sgRNA(EcoRI)/Pr_sgRNA(HindIII) was used to amplify anti-adhE, anti-eno, anti-fabB and anti-sucC sgRNA sequences into the EcoRI/HindIII sites of pCOLA-fabF(high)/fumC(medium). This resulted in the plasmids pCOLA-fabF(high)/fumC(medium)/adhE(low), pCOLA-fabF(high)/fumC(medium)/eno(low), pCOLA-fabF(high)/fumC(medium)/fabB(low) and pCOLA-fabF(high)/fumC(medium)/sucC(medium). The primer pair Pf_sgRNA(HindIII)/Pr_sgRNA(NdeI) was used to amplify anti-adhE, anti-eno and anti-sucC sgRNA sequences into the HindIII/NdeI sites of pCOLA-fabF(high)/fumC(medium)/fabB(low). This resulted in the plasmids pCOLA-fabF(high)/fumC(medium)/fabB(low)/adhE(low), pCOLA-fabF(high)/fumC(medium)/fabB(low)/eno(low) and pCOLA-fabF(high)/fumC(medium)/fabB(low)/sucC(medium). The primer pair Pf_sgRNA(NdeI)/Pr_sgRNA(BglII) was used to amplify anti-adhE and anti-eno sgRNA sequences into the NdeI/BglII sites of pCOLA-fabF(high)/fumC(medium)/fabB(low)/sucC(medium). This resulted in the plasmids pCOLA-fabF(high)/fumC(medium)/fabB(low)/sucC(medium)/adhE(low) and pCOLA-fabF(high)/fumC(medium)/fabB(low)/sucC(medium)/eno(low). The primer pair Pf_sgRNA(BglII)/Pr_sgRNA(KpnI) was used to amplify anti-eno sgRNA sequences into the BglII/KpnI sites of pCOLA-fabF(high)/fumC(medium)/fabB(low)/sucC(medium)/adhE(low). This resulted in the plasmid pCOLA-fabF(high)/fumC(medium)/fabB(low)/sucC(medium)/adhE(low)/eno(low).

24To investigate the silencing effects of various sgRNA species, cells were cultured in 25 mL of MOPS medium at 37 °C with 220 rpm orbital shaking at a starting OD600 of 0.1. After the OD600 reached 1.65, an additional 25 mL of fresh MOPS medium was added. sgRNA expression was induced with 0.5 mM IPTG. Cultures were subsequently incubated at 30 °C. 1 mL of cell culture was harvested at the mid-log phase of growth to quantify the intracellular concentrations of malonyl-CoA and acetyl-CoA. Final OD600 values of the cultures were measured after a total fermentation time of 48 h.

25For (2S)-naringenin production, cells were first cultured in 25 mL of MOPS medium at 37 °C with 220 rpm orbital shaking at a starting OD600 of 0.1. After an OD600 of 1.65 had been reached, an additional 25 mL of fresh MOPS medium was added. Cultures were subsequently incubated at 30 °C for (2S)-naringenin production. IPTG and L-tyrosine were provided at final concentrations of 1 mM and 3 mM, respectively. After a total fermentation time of 48 h, 1 mL of cell culture was collected to measure the concentrations of (2S)-naringenin and p-coumaric acid. Final OD600 values of the cultures were also measured after a total fermentation time of 48 h.

26For each experiment, triplicate cultures were measured, and their deviation is represented by the error bar. To analyze (2S)-naringenin and p-coumaric acid production, E. coli cells were separated through centrifugation (5000 g, 15 min, 4 °C). 1 mL of supernatant was extracted with an equal volume of ethyl acetate. After vortexing and centrifugation (5000 g, 15 min, 4 °C), the top organic layer was separated and evaporated to dryness, and the remaining residue was resolubilized with 1 mL of methanol. Samples were analyzed by high-performance liquid chromatography (HPLC), using an Agilent 1100 series instrument and a reverse-phase Gemini NX-C18 column (5 × 110 mm) maintained at 25 °C. (2S)-Naringenin and p-coumaric acid were separated by elution with an acetonitrile/water gradient at a flow rate of 1.0 mL/min under the following conditions: 10 to 40% acetonitrile (vol/vol) for 10 min, 40% acetonitrile (vol/vol) for 5 min, 40 to 10% acetonitrile (vol/vol) for 2 min.

27To quantify the malonyl-CoA concentration, an aliquot of 1 mL of cell culture was removed and chilled immediately on ice and centrifuged at 5000 g and 4 °C for 10 min. The cell pellet was resuspended in 1 mL of 6% perchloric acid to facilitate cell lysis. The lysed cell suspension was then neutralized with 0.3 mL of 3 M potassium carbonate while vortexing to neutralize the acid. The solution was centrifuged to pellet the cell debris. The supernatant was collected and stored chilled until the malonyl-CoA content was analyzed. To determine the dry cell weight, 2 mL of the same culture were filtered through a 0.45-μm cellulose membrane (Sangon Biotech, Shanghai, China), followed by washing with distilled water and drying in a conventional oven. Dry cell weight was represented by the weight difference between empty membranes and those with cell residues.

28Malonyl-CoA and acetyl-CoA from the cell cultures were identified by the area of major mass spectra signals ([M–H]−). A liquid chromatography-mass spectrophotometer (Shimadzu, Kyoto, Japan) equipped with an electrospray ionization (ESI) source and a reverse-phase Gemini NX-C18 column (5 × 110 mm) was used. A Shim-Pack VP-ODS 150L × 2.0 HPLC column (Shimadzu, Kyoto, Japan) was used to perform HPLC separation. Samples of malonyl-CoA and acetyl-CoA were eluted at a flow rate of 0.3 mL/min with a gradient of 15 mM ammonium formate (A) and 10% 10 mM ammonium acetate in methanol (v/v) (B) as follows: at 0 min, 2% B; 10 min, 60% B; 20 min, 76% B; 25 min, 2% B. The MS/MS system was operated in negative ion mode using optimized conditions: detector voltage, 1.60 Kv; nebulizing gas (N2) flow, 1.5 L/min; drying gas (N2) flow, 200 kPa; ion accumulation time, 30 ms; scan range m/z, 100–1000 for MS1, 100–500 for MS2.

29How to cite this article: Wu, J. et al. Enhancing flavonoid production by systematically tuning the central metabolic pathways based on a CRISPR interference system in Escherichia coli. Sci. Rep. 5, 13477; doi: 10.1038/srep13477 (2015).