A Synthetic Approach for Biosynthesis of Miquelianin and Scutellarin A in Escherichia coli

Grapevine (Vitis vinifera) glycucuronosyltransferase (VvGT5) specifically catalyzes flavonol-3-O-glucuronosylation and the blue flowers of Veronica persica (Lamiales, Scrophulariaceae) uridine diphosphate (UDP)-dependent glycosyltransferase (UGT88D8) as flavonoid 7-O-specific glucuronosyltransferases, were chosen, codon optimized, and employed to synthesize the high valued flavonoids glucuronoids, miquelianin and scutellarin A in Escherichia coli. A single vector system was constructed to overexpress entire UDP-glucuronic acid biosynthesis pathway genes, along with a glucokinase gene in Escherichia coli BL21 (DE3). The newly generated E. coli BL21 (DE3) piBR181-glk.pgm2.galU.ugd.UGT88D8 strain produced 12 mg/L (28 μmol/L) of scutellarin A from apigenin, representing only 14% of maximum conversion percentage. Similarly, the strain E. coli BL21 (DE3) piBR181-glk.pgm2.galU.ugd.VvGT5 produced 30 mg/L (62 μmol/L) of miquelianin, representing a 31% conversion of quercetin. This production profile is a good starting point for further host engineering, and for production of respective compounds.


Introduction
Flavonoids have always been recognized as a promising factor in human health, as they refer to various biological responses, such as an inherent ability to modify the body's reaction to allergens, viruses, and carcinogens. Experimental pieces of evidences have always been positive in terms of their activities as anti-allergens, anti-inflammatory, anti-microbial, anti-cancer, etc. [1][2][3]. Flavonoids cover a large group of plant secondary metabolites with a profound research interest in various biological and therapeutic applications, since its genetic and biosynthetic investigations have been well-documented over the decades [4]. These aromatic molecules are derived from p-coumaroyl-CoA and malonyl-CoA. They cover a diverse class of natural products, including; the chalcones, flavones, isoflavones, flavonols, flavandiols, anthocyanins, and condensed tannins [5,6]. Flavonols (quercetin, kaempferol, myricetin, isorhamnetin) represent their diversity, and stem from the 3-OH phenolic group. They are present in a wide variety of fruits and vegetables (apples, berries, onions, broccoli, grapes, flowers and bark), including botanicals [6]. They are well-known for their various health-promoting effects, which include antioxidant, anti-inflammatory, anti-angiogenic, anti-proliferative, and neuro-pharmacological properties [1][2][3]. However, flavones (apigenin, luteolin, chrysin, baicalein) have recently been generating enormous public and scientific interest due to their beneficial effects against atherosclerosis, osteoporosis, diabetes mellitus, and certain cancers [7][8][9]. Interestingly, both classes of flavonoids act as inhibitors of cytochrome (CYP) P450 enzymes, such as CYP2C9 and CYP3A4, enzymes that help in the metabolism of xenobiotics in the human body [10,11].
Thus, polyphenols are worth intaking naturally as they have substantial health benefits; however, there are limitations in direct consumption. The limitations could be rooted in their water solubility, limited bioavailability or xenobiotic neutralizations, which challenges the necessity of their modifications by glycosylation/hydroxylation (to enhance solubility), and methylation (to increase lipophilicity, cell recognition) [12][13][14]. Nevertheless, conjugation of carbohydrate appendages in such promising metabolites has always improved pharmaceutical values [15]. The glucuronic acid itself is often linked to the xenobiotic metabolism in the human body, and the substances resulting from glucuronidation are typically much more water-soluble [16]. In plants and animals, glucuronic acid is considered to be a precursor of vitamin C [17]. Thus, producing plant polyphenols conjugated with sugars such as glucuronic acid could help to develop such molecules as prospective food additives.
Biological synthesis of regio-and stereo-selective phytochemicals using an industrially important Escherichia coli system has been widely studied in current research trends [14]. Most of the bacteria, including E. coli, can produce nucleotide sugars, which are building blocks of bacterial polysaccharides and have been well-studied for various reasons [14,18]. These activated nucleotide diphosphate sugars are utilized as sugar donors by glycosyltransferase (UGTs), which are central players in decorating these valuable natural plant products with sugar moieties [19]. Although sugars like UDP-glucose, UDP-galactose, and UDP-glucuronic acid are common nucleotide diphosphate sugars present in E. coli, engineering or overexpression of the pathway-specific genes increase their availability for the biotransformation.
The current research employs Vitis vinifera glucuronosyltransferase; VvGT5 catalyzes the 3-O-specific glucuronic acid conjugation, and flavonoid 7-O-specific glucuronosyltransferase, UGT88D8 from the blue flowers of Veronica persica to produce miquelianin and scutellarin A in E. coli. Both plant UGTs were codon optimized, synthesized, and functionally expressed in E. coli. Moreover, both genes were incorporated in single vector multi-monocistronic systems, overexpressing UDP-glucoronic pathway genes to develop two independent systems for regiospecific glycosylation.

Bacterial Strains and Culture Conditions
E. coli XL-1 Blue (Stratagene, CA, USA) was used for plasmid cloning and propagation, whereas E. coli BL21 (DE3) was used for expression of the proteins and flavonoid biotransformation. All strains, vectors, and plasmids used in this study are listed in Table 1 and the construction of the mono-cistronic vector system is described in (Figure 1). Cloning and manipulations of DNA were performed according to the standard procedures [20]. A previously constructed expression vector piBR181 [21] was used for cloning and expression. The pGEM-T ® easy vector system (Promega, Madison, WI, USA) was used as a vector for cloning polymerase chain reaction (PCR) fragments and sequencing. E. coli BL21 (DE3) ∆pgi∆zwf ∆ushA mutant harboring the UDP-glucose biosynthetic pathway was used as a final strain for biotransformation. E. coli strains were routinely cultured in Luria-Bertani (LB) broth or on LB agar plates, supplemented with the appropriate antibiotics (ampicillin 100 µg/mL, kanamycin 50 µg/mL) when necessary. Authentic apigenin and quercetin were purchased from Sigma-Aldrich (St. Louis, MO, USA), and the rest of the chemicals used in this study were high-grade and purchased from commercially-available sources.

Cloning and Expression of Recombinant Proteins
Restriction enzymes, LaTaq polymerase, and T4 DNA ligase were purchased from Takara Bio Inc. (Shiga, Japan). UDP-glycosyltransferases, VvGT5 (grapevine; Vitis vinifera UDP-glucuronic acid: flavonol 3-O-glucuronosyltransferase) and UGT88D8 (the blue flowers of Veronica persica poiret; Lamiales, Scrophulariaceae structurally flavonoid 7-O-specific glucuronosyltransferases (F7GAT)) were codon optimized, synthesized, cloned and expressed in restriction site dependent multi-monocistronic vector, piBR181. Oligonucleotide primers were synthesized from Genotech (South Korea). All PCR size and primers used in this study are described in Table 2. To increase the pool of UDPglucuronic acid, its pathway specific genes were individually cloned and assembled together in the

Cloning and Expression of Recombinant Proteins
Restriction enzymes, LaTaq polymerase, and T4 DNA ligase were purchased from Takara Bio Inc. (Shiga, Japan). UDP-glycosyltransferases, VvGT5 (grapevine; Vitis vinifera UDP-glucuronic acid: flavonol 3-O-glucuronosyltransferase) and UGT88D8 (the blue flowers of Veronica persica poiret; Lamiales, Scrophulariaceae structurally flavonoid 7-O-specific glucuronosyltransferases (F7GAT)) were codon optimized, synthesized, cloned and expressed in restriction site dependent multi-mono-cistronic vector, piBR181. Oligonucleotide primers were synthesized from Genotech (South Korea). All PCR size and primers used in this study are described in Table 2. To increase the pool of UDP-glucuronic acid, its pathway specific genes were individually cloned and assembled together in the piBR181 vector. The UDP-glucuronic acid sugar pathway genes glucokinase (glk: from Zymomonas mobilisi, GenBank ID: AE008692.2), a vector pVWEx1-glf -glk provided by Dr. Stephanie Bringer-Meyer (Institut für Bio-und Geowissenschaften, IBG-1: Biotechnologie, Forschungszentrum Jülich, Jülich, Germany), phosphoglucomutase (pgm2; from Bacillus licheniformis DSM 13, GenBank accession No. YP_006712377.1), glucose 1-phosphate uridylyltransferase (galU; from E. coli K-12, GenBank accession No. CP001509.3), and UDP-glucose 6-dehydrogenase (ugd; GenBank accession No. YP_490271.1 from E. coli BL21(DE3) were selected for overexpression. All these genes were amplified from the genomic DNA of respective strains using a pair of primer that had XbaI/HindIII restriction sites, as shown in Table 1. Finally, the sugar cassette was assembled with codon optimized regiospecific glycurunosyltransferases (UGT88D8 or VvGT5) separately, as the physical scheme explains in Figure 1. These two plasmids were used for enhanced biosynthesis of two molecules. In all cases, construction of each recombinant and assembly of each pathway gene were verified with restriction digestion, PCR, and direct nucleotide sequencing of PCR-amplified products of the respective genes in the recombinant plasmid. Each recombinant plasmid containing UDP-glucuronic acid sugar pathway genes was transformed into E. coli BL21 (DE3) with the heat shock transformation technique. Later, the restriction-digestion-confirmed single transformant was cultured for further studies. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis was performed to check the expression of each gene and was compared with the standard protein marker. To check the expression of each gene, we followed the same methodology as described previously [22] with the same concentration of IPTG. However, the protein expression was carried out at 15 • C. Table 2. List of bacterial strains and vector maps used in this study.

Strains/Plasmids Description Source/Reference
This study

Construction of Sugar Cassettes
All genes cloned from different organism sources were restriction digested with XbaI/HindIII after routine manipulation in pGEM-T Easy vector, and individually cloned into the expression vector piBR181 at the restriction site of SpeI/HindIII. SpeI and XbaI are cohesive enzymes as they generate similar over-hangs. Thus, ligation of these two fragments is possible, but after ligation a 'scar' is formed, which replaces the restriction sites. In the vector backbone, BamHI/XhoI site remains, which can be applied for further cloning of other genetic parts digested with the BglII and XhoI enzymes (includes a set; an RBS, a T7 promoter, a target gene, and a transcription terminator). Similarly, BglII and BamHI also generate cohesive ends and forms a scar after ligation. This means, in each insertion of a circuit part, there is the formation of a scar. This restricts further cuts by the same restriction endonucleases, leaving a new cloning site at BamHI/XhoI in the vector. Likewise, the cloning of each sugar pathway gene, including UGTs, completed the two cassettes of glucuronosylation system.

Whole-Cell Biotransformation and Flavonoid Extraction
E. coli strains were pre-cultured in 3 mL of LB liquid medium with antibiotics and incubated at 37 • C in 200 rpm overnight. Five-hundred microliters of inoculum was transferred into 50 mL of Luria-Bertani (LB) liquid medium with appropriate antibiotics and cultured at 37 • C until the optical density at OD 600nm reached 0.5-0.7. The culture was induced by adding isopropyl-β-Dthiogalactopyranoside (IPTG) with a final concentration of 0.5 mmol/L and incubated at 15 • C for protein expression, and to increase cell biomass. The following day, each culture was supplemented with specific flavonoids; apigenin in E. coli harboring UGT88D8 to produce scutellarin A, and quercetin in E. coli harboring VvGT5 to produce miquelianin. Then, the flasks were incubated at 15 • C for 60-72 h for biotransformation. Each culture was extracted with an equal volume of ethyl acetate. The organic layer was collected and concentrated to dryness through the evaporation of excess solvent, followed by the addition of methanol to dissolve remaining products. The extracts were subjected to high performance liquid chromatography (HPLC) and liquid chromatography quadrupole time-of-flight-electrospray ionization mass spectrometry (LC-QTOF-ESI/MS) analyses.
Following the confirmation of products, we performed the bio-catalysis reaction with the strains containing sugar cassettes, including UGTs. Equal quantities (200 µL) having the same cell density (0.7) inoculums were transferred into 50 mL E. coli flasks containing LB medium. Identical conditions were applied for IPTG induction, incubation temperature, and shaking of the culture. E. coli BL21 (DE3) ∆pgi∆zwf ∆ushA piBR181-glk.pgm2.galU.ugd.UGT88D8 was fed with apigenin while E. coli BL21 (DE3) ∆pgi∆zwf ∆ushA piBR181-glk.pgm2.galU.ugd.VvGT5 was fed with quercetin. One milliliter sample was drawn from each culture at 12 h time intervals until 72 h. The samples were centrifuged, and supernatants were extracted with a double volume of ethyl acetate, followed by drying the samples. Then, the samples were dissolved in 100 µL methanol. These reaction mixtures were analyzed by HPLC-PDA and LC-MS for confirmation of glucuronic acid conjugated derivatives of quercetin and apigenin.

Selection of Plant UGTs and Their Activities
Flavonoids are an important group of plant secondary metabolites and are usually present in the human diet in the form of glycosides. Glycosylation typically occurs in the final step of flavonoid biosynthesis, which is catalyzed by several UGTs displaying high substrate stereoand regio-selectivity [24]. In this context, although plants could be considered to have multiple glycosyltransferases enzymes capable of conjugating diverse sugars to different classes of plant molecules, we cannot escape from the difficulties using them in microbial cell factory, as most of the eukaryotic enzymes are difficult to express functionally. Previously reported regiospecific and well-characterized plant glucuronosyltransferases (UGT88D8 and VvGT5) [25,26] were selected to synthesize valuable glucuronic acid conjugated flavonoids in an industrially important bacterial strain E. coli. These UGTs were codon optimized prior to synthesize based on the E. coli codon usage to facilitate functional expression and used in biotransformation. According to a previous report, UGT88D8 favored the substrate apigenin catalyzing glucuronidation at 7-OH position where as VvGT5 favored quercetin and conjugate glucuronic acid at 3-OH position. Thus, we tested these GTs with respective substrates. The HPLC-PDA and LC-MS analyses could easily detect the generation of apigenin 7-O-glucuronoid (scutellarin A) and quercetin 3-O-glucuronoid (miquelianin). Furthermore, we tested similar class of flavonoids (baicalein, luteolin, crysin) in vivo bio-transformation catalyzed by UGT88D8 and VvGT5 with myricetin, kampferol, fisetin, as substrates. When carrying out analyses in HPLC-PDA and LC-MS, we were unable to detect product peak(s) and unable to detect in mass (data not shown). This also supported that the aforementioned UGTs are regio-, stereo-as well as a substrate specific.

Cloning and Expression of NDP-Sugar Pathway Genes and Assembly in a Single Vector
The UDP-glucuronic acid pathway genes were responsible for increasing the desired sugar pool in the cell. Glucokinase (Glk) catalyzes to form glucose 6-phosphate (G-6-P) from primary carbon source D-glucose, phosphoglucomutase (Pgm2) was responsible for converting the G-6-P into glucose 1-phosphate (G-1-P), which was further converted into UDP-glucose by the enzyme glucose 1-phosphate uridylyltransferase (GalU). Finally, UDP-glucose 6-dehydrogenase (Ugd) catalyzes to convert UDP-glucose into UDP-glucuronic acid (Figure 1). These genes were individually cloned into the piBR181 vector and expressed as explained in materials and methods. Expression of soluble proteins was examined using SDS-PAGE analyses, comparing with the standard sized protein marker (data not shown). Ultimately, these genes were assembled into a single vector piBR181 and the recombinant plasmid was later transformed into E. coli hosts for the biotransformation process.
After successful expression of each protein in soluble form, these genes, including UGT88D8 and VvGT5, were cloned to construct single recombinant plasmids ( Figure 2). Under the hypothesis that overexpression of each glycolysis gene should enhance the nucleotide diphosphate (NDP)-sugar pool in the cytosol, this vector was transformed into a bacterial platform for the biocatalysis reaction.

Synthesis of Glucuronoids in Recombinant Strains
The codon optimized UGTs from the respective source were initially screened by feeding in recombinant strains. E. coli BL21 (DE3)/piBR181-UGT88D8 and E. coli BL21 (DE3)/piBR181-VvGT5 were grown in LB medium, induced and incubated at 15 °C for 20 h to allow functional expression of the protein, as was explained in Materials and Methods. A total of 200 µM of apigenin and quercetin dissolved in dimethyl sulfoxide (DMSO) were added into the respective culture medium. After 48 h of incubation, 500 µL culture broth was taken and mixed with double volume ethyl acetate

Synthesis of Glucuronoids in Recombinant Strains
The codon optimized UGTs from the respective source were initially screened by feeding in recombinant strains. E. coli BL21 (DE3)/piBR181-UGT88D8 and E. coli BL21 (DE3)/piBR181-VvGT5 were grown in LB medium, induced and incubated at 15 • C for 20 h to allow functional expression of the protein, as was explained in Materials and Methods. A total of 200 µM of apigenin and quercetin dissolved in dimethyl sulfoxide (DMSO) were added into the respective culture medium. After 48 h of incubation, 500 µL culture broth was taken and mixed with double volume ethyl acetate and vigorously vortexed followed by centrifugation to separate the organic and aqueous layer. The dried supernatant organic layer was dissolved in 100 µL of methanol and subjected to HPLC analysis. New peak(s) were detected at retention time (t R )~17.5 min before the standard apigenin t R~2 0.2 min. in the case of UGT88D8 at UV absorbance of 300 nm. While in VvGT5 culture sample, new peak(s) at t R~1 7.5 were detected as compared to the standard quercetin t R~1 9 min. These samples were further analyzed by high resolution mass analyses to confirm if the generated peaks were respective glucuronoids.  (Figure 3). The conversion of apigenin to scutellarin A was~6% while that of quercetin to miquelianin was~13%. The conversion was unexpectedly low. Thus, we presumed that the low conversion was due to the low cytosolic pool of UDP-glucuronic acid.  (Figure 3). The conversion of apigenin to scutellarin A was ~6% while that of quercetin to miquelianin was ~13%. The conversion was unexpectedly low. Thus, we presumed that the low conversion was due to the low cytosolic pool of UDP-glucuronic acid.

Enhanced Synthesis of Miquelianin and Scutellarin A
After a successful confirmation of target products through spectroscopic analyses, we used other recombinants generated for this study. Two single recombinant vectors generated after assembly of UDP-glucuronic acid biosynthesis pathway genes and regiospecific UGTs were compared for increased production of respective products. Based on previous experiences, we hypothesized that the production of glycosides should increase upon overexpression of UDP-glucuronic acid pathway specific genes, along with UGT88D8 and VvGT5. As in previous experiments, 50 mL cultures were grown and allowed the functional expression of protein for 20 h at 15 °C. For biotransformation reaction, the same 200 µmol/L concentration of substrates apigenin and quercetin were supplemented in respective recombinant strains. The cultures were harvested in 48 h incubation and analysed by HPLC-PDA. The production of each glycoside was increased approximately twice. The conversion of apigenin to scutellarin A was ~14% (28 µmol/L; 12 mg/L) and that of quercetin to miquelianin was ~31% (62 µmol/L; 30 mg/L). However, the increment was not satisfactory. We optimized the experiments with the addition of glucose as a carbon source. The production did not change significantly, although the cell growth was reasonably high (Figure 4).

Enhanced Synthesis of Miquelianin and Scutellarin A
After a successful confirmation of target products through spectroscopic analyses, we used other recombinants generated for this study. Two single recombinant vectors generated after assembly of UDP-glucuronic acid biosynthesis pathway genes and regiospecific UGTs were compared for increased production of respective products. Based on previous experiences, we hypothesized that the production of glycosides should increase upon overexpression of UDP-glucuronic acid pathway specific genes, along with UGT88D8 and VvGT5. As in previous experiments, 50 mL cultures were grown and allowed the functional expression of protein for 20 h at 15 • C. For biotransformation reaction, the same 200 µmol/L concentration of substrates apigenin and quercetin were supplemented in respective recombinant strains. The cultures were harvested in 48 h incubation and analysed by HPLC-PDA. The production of each glycoside was increased approximately twice. The conversion of apigenin to scutellarin A was~14% (28 µmol/L; 12 mg/L) and that of quercetin to miquelianin was~31% (62 µmol/L; 30 mg/L). However, the increment was not satisfactory. We optimized the experiments with the addition of glucose as a carbon source. The production did not change significantly, although the cell growth was reasonably high (Figure 4). Several plant metabolites, including flavonoids and their glycoside derivatives, have been produced from engineered E. coli cells as a microbial cell factory. Such microbial production has attracted extensive interest in the development of a novel system for the modification and efficient production of highly useful value-added compounds [14]. To produce glycosylated flavonoids from engineered E. coli, the pool of activated NDP-sugars, the donor substrates of UGTs and limiting factor in the cell cytosol is enhanced by introducing heterologous genes encoding the target sugar pathway genes, or by blocking the indigenous competing genes [14]. Using this approach, several flavonoid glycosides are produced from E. coli. For example, 3-O-rhamosyl quercetin, 3-O-rhamnosyl kaempferol, and 3-O-allosyl quercetin [27], astragalin [28], quercetin-3-O-4-deoxy-4-formamido-Larabinose [29], quercetin 3-O-N-acetylglucosamine [30], quercetin-3-O-xyloside [31] and quercetin 3-O-arabinoside [32], flavonol 3-O-rhamnosides [33,34], flavone 6-C-glucosides [35] are some representative flavonoid glycosides produced from E. coli. Using a similar approach, several polyketide glycosides have been produced from engineered microbial hosts [36]. Most of these systems used multiple vectors for overexpression of genes, which required the addition of multiple antibiotics to the culture broth for selection pressure. Thus, to reduce the antibiotic stress on the growing cell, we used a single vector system in which multiple genes were assembled for biosynthesis of two bioactive molecules; miquelianin and scutellarin A. Previously, miquelianin was produced using an approach similar to the approach that produced engineered E. coli cell [37]. However, this is the first report to produce scutellarin A from E. coli system. The production of miquelianin and scutellarin A from the systems developed in this study is relatively low compared to previous studies that reported biosynthesis of flavonoid glycosides [14]. However, this study opens possible paths for further harnessing of systems and synthetic biology tools to engineer central carbon metabolic pathways, to dam up the flow of UDP-glucuronic acid, and to develop an efficient system to produce glucuronic acid conjugated flavonoids.

Conclusions
Two different recombinant plasmids were developed and successively incorporated into the industrially important bacterium E. coli. The recombinant strains were used to synthesize glucuronoids of valuable dietary flavonoids. We successfully produced scutellarin A and miquelianin in engineered E. coli strains. A further experiment was carried out to construct a vector by assembling entire sugar biosynthesis pathway genes, aiming to increase the pool of UDPglucuronic acid in the cytosol considering that pool of UDP-glucuronic acid is a limiting factor for lower production of respective compounds. However, when we used the final recombinant strains for biotransformation of quercetin and apigenin, we were unable to achieve high yield from the culture medium. Thus, there are two possible reasons for low production of target compounds: (1) Several plant metabolites, including flavonoids and their glycoside derivatives, have been produced from engineered E. coli cells as a microbial cell factory. Such microbial production has attracted extensive interest in the development of a novel system for the modification and efficient production of highly useful value-added compounds [14]. To produce glycosylated flavonoids from engineered E. coli, the pool of activated NDP-sugars, the donor substrates of UGTs and limiting factor in the cell cytosol is enhanced by introducing heterologous genes encoding the target sugar pathway genes, or by blocking the indigenous competing genes [14]. Using this approach, several flavonoid glycosides are produced from E. coli. For example, 3-O-rhamosyl quercetin, 3-O-rhamnosyl kaempferol, and 3-O-allosyl quercetin [27], astragalin [28], quercetin-3-O-4-deoxy-4-formamido-L-arabinose [29], quercetin 3-O-N-acetylglucosamine [30], quercetin-3-O xyloside [31] and quercetin 3-O-arabinoside [32], flavonol 3-O-rhamnosides [33,34], flavone 6-C-glucosides [35] are some representative flavonoid glycosides produced from E. coli. Using a similar approach, several polyketide glycosides have been produced from engineered microbial hosts [36]. Most of these systems used multiple vectors for overexpression of genes, which required the addition of multiple antibiotics to the culture broth for selection pressure. Thus, to reduce the antibiotic stress on the growing cell, we used a single vector system in which multiple genes were assembled for biosynthesis of two bioactive molecules; miquelianin and scutellarin A. Previously, miquelianin was produced using an approach similar to the approach that produced engineered E. coli cell [37]. However, this is the first report to produce scutellarin A from E. coli system. The production of miquelianin and scutellarin A from the systems developed in this study is relatively low compared to previous studies that reported biosynthesis of flavonoid glycosides [14]. However, this study opens possible paths for further harnessing of systems and synthetic biology tools to engineer central carbon metabolic pathways, to dam up the flow of UDP-glucuronic acid, and to develop an efficient system to produce glucuronic acid conjugated flavonoids.

Conclusions
Two different recombinant plasmids were developed and successively incorporated into the industrially important bacterium E. coli. The recombinant strains were used to synthesize glucuronoids of valuable dietary flavonoids. We successfully produced scutellarin A and miquelianin in engineered E. coli strains. A further experiment was carried out to construct a vector by assembling entire sugar biosynthesis pathway genes, aiming to increase the pool of UDP-glucuronic acid in the cytosol considering that pool of UDP-glucuronic acid is a limiting factor for lower production of respective compounds. However, when we used the final recombinant strains for biotransformation of quercetin and apigenin, we were unable to achieve high yield from the culture medium. Thus, there are two possible reasons for low production of target compounds: (1) the catalytic activity of synthesized UGTs (UGT88D8 and VvGT5) were not sufficient enough to conjugate sugar moiety, or (2) the downstream pathway genes of UDP-glucuronic acid are actively utilizing UDP-glucuronic acid, thus limiting the pool of UDP-glucuronic acid for UGTs. Importantly, further strategic engineering of host bacterium or troubleshooting the problems could achieve an appreciable yield of desire glycosides.
Author Contributions: R.P.P. and P.P. conceived the idea. H.Y.J., P.P. and R.P.P. performed experiments, T.H.T.N. and P.B. assisted in recombinant construction. P.P., R.P.P. and J.K.S. wrote the manuscript. R.P.P. and H.Y.J. contributed equally to this work.

Conflicts of Interest:
The authors declare no conflicts of interest.