Chromosome engineering of Escherichia coli for constitutive production of salvianic acid A

Background Salvianic acid A (SAA), a valuable natural product from herbal plant Salvia miltiorrhiza, exhibits excellent antioxidant activities on food industries and efficacious therapeutic potential on cardiovascular diseases. Recently, production of SAA in engineered Escherichia coli was established via the artificial biosynthetic pathway of SAA on the multiple plasmids in our previous work. However, the plasmid-mediated system required to supplement expensive inducers and antibiotics during the fermentation process, restricting scale-up production of SAA. Microbial cell factory would be an attractive approach for constitutive production of SAA by chromosome engineering. Results The limited enzymatic reactions in SAA biosynthetic pathway from glucose were grouped into three modules, which were sequentially integrated into chromosome of engineered E. coli by λ Red homologous recombination method. With starting strain E. coli BAK5, in which the ptsG, pykF, pykA, pheA and tyrR genes were previously deleted, chassis strain BAK11 was constructed for constitutive production of precursor l-tyrosine by replacing the 17.7-kb mao-paa cluster with module 1 (PlacUV5-aroG fbr-tyrA fbr-aroE) and the lacI gene with module 2 (Ptrc-glk-tktA-ppsA). The synthetic 5tacs promoter demonstrated the optimal strength to drive the expression of hpaBC-d-ldh Y52A in module 3, which then was inserted at the position between nupG and speC on the chromosome of strain BAK11. The final strain BKD13 produced 5.6 g/L of SAA by fed-batch fermentation in 60 h from glucose without any antibiotics and inducers supplemented. Conclusions The plasmid-free and inducer-free strain for SAA production was developed by targeted integration of the constitutive expression of SAA biosynthetic genes into E. coli chromosome. Our work provides the industrial potential for constitutive production of SAA by the indel microbial cell factory and also sets an example of further producing other valuable natural and unnatural products. Electronic supplementary material The online version of this article (doi:10.1186/s12934-017-0700-2) contains supplementary material, which is available to authorized users.


Background
Salvianic acid A (SAA, 3-(3′,4′-dihydroxyphenyl)-2-hydroxypropanoic acid), also called danshensu, is the major bioactive ingredient of traditional Chinese herb plant Salvia miltiorrhiza (danshen) which is widely used for the prevention and treatment of vascular diseases in clinic [1,2]. SAA is well-known for its distinguished antioxidant capacity to scavenge the superoxide anion radicals and free hydroxyl radicals, which is even higher than vitamin C [3]. In recent years, SAA has attracted considerable attentions due to its various pharmacological activities, including inhibition of platelet activation and arterial thrombosis [4], alleviation of alcohol-induced acute liver damage [5] and myocardial ischemia injury [6]. Moreover, SAA derivatives, salvianolic acids B and A, and rosmarinic acid, show promising application in medicines and food industries. Salvianolic acid B has already been used to alleviate angina pectoris and treat coronary heart diseases in clinic [7]. Salvianolic acid A has been approved by China Food and Drug Administration (CFDA) into phase I clinical trial. Rosmarinic acid could be used for food preservation as a natural antioxidant to substitute synthetic antioxidant like butylated hydroxytoluene [8].
Notably, conjugates of SAA with cysteine show better vascular-protective effect than SAA [9].
Synthetic biology and metabolic engineering are promising strategies to deal with those obstacles by reconstructing inducer-free and plasmid-free strains. The constitutive promoters, the core elements for metabolic engineering, have been paid more attention to allowing inducer-free and continuous gene expression as well as low production cost [15]. Native constitutive promoters have been widely employed for producing bio-based chemicals [16][17][18] and natural products [19][20][21][22] by tuning target gene expression in engineered yeasts. In E. coli, the conventional T7 promoter has always been used for heterologous protein expression by additional inducer IPTG, but the constitutive promoters showed better performance for production of natural and non-natural products, which are adapted to the growth of host cells. The native constitutive gap promoter was more suitable for the heterologous expression of the 4 cl and sts genes in E. coli than the inducible T7 promoter, resulting in high production of resveratrol [23]. The P L derivative promoters could constitutively drive target gene expression in E. coli to maximize the desired phenotypes and benefit for effective production of lycopene [24], 2,3-butanediol [25] and human growth hormone receptor antagonist [26]. In addition, chromosomal integration of the target gene would eliminate the use of plasmids in host cells and relieve the metabolic burden. Excellent examples were high-yielding production of artemisinic acid, a precursor of artemisinin, and lycopene in baking yeast [27,28]. The plasmid-free E. coli strains could stablely produce astaxanthin [29], shikimic acid [30] and β-carotene [31].
In this study, E. coli BAK5 [12] was used as the start strain. Limited enzymatic reactions of SAA synthetic pathway from glucose in engineered E. coli were grouped into three modules as in previous work [12]. Modules 1 and 2 were integrated into the chromosome of strain BAK5, and strain BAK11 for overproducing l-tyrosine without any plasmids was constructed (Fig. 1b). The optimal promoter for driving module 3 was screened out by fine tuning constitutive expression, and integrated into chromosome of strain BAK11, a plasmid-free and inducer-free E. coli strain BKD13 for constitutive production of SAA was constructed (Fig. 1b). Finally, fed-batch fermentation was taken out and engineered strain BKD13 produced 5.6 g/L of SAA in 60 h. To our knowledge, this is the first report describing SAA production by engineered E. coli without addition of any inducers and antibiotics.

Bacterial strains, plasmids and reagents
The bacterial plasmids and strains used in this study were listed in Tables 1 and 2, respectively. All primers used in this study were summarized in Additional file 1: Table  S1). E. coli DH5α was employed for all gene cloning work and E. coli BW25113 derivatives were used to construct SAA producing strains. The primers and long DNA fragments were synthesized by GENEWIZ (Suzhou, China). Methanol and acetate (HPLC grade) were purchased from Concord Tech (China). All other chemicals and reagents used in the experiment were of analytical grade commercially available. SAA (98% purity) was purchased from Xi'an Honson Biotechnology Company (China). l-tyrosine (99.9% purity) and l-dopa (99.9% purity) were purchased from Dingguo Biotech (China). DNA Polymerase of Phanta Super Fidelity and Taq for PCR were purchased from Vazyme (Nanjing, China). T4 DNA ligase and restriction endonucleases were purchased from Thermo Scientific (Beijing, China). The plasmids were constructed by ligating the PCR products and plasmids, which were all digested by the same restriction endonucleases. Successful recombinant plasmids were confirmed by colony PCR and DNA sequencing.

Construction of SAA-producing expression vectors with constitutive promoters
Four constitutive promoters, P BBa-J23100 , P tac , P 5tacs and P rrnB P1 , were used to construct expression vectors. Three fragments (F1, F2 and F3) containing multiple cloning sites, BBa-B0015 terminator were designed and synthesized to simplify the vector construction (Additional file 1: Table S2). In order to achieve the maximum translation level, the synthetic 5′-untranslated region (5′-UTR) sequences of the hpaBC and d-ldh Y52A genes were predicted by the UTR Designer (http://sbi.postech. ac.kr/utr_designer) and designed in primers hpaBC F and d-ldh Y52A F directly. Plasmid pACYCDuet-1 was digested with ApaI and XhoI, and the fragment containing the p15A ori and Cm R was served as the skeleton vector pZL.
To construct plasmids pZL3, 4, 5 and 6 for bicistron expression of the hpaBC and d-ldh Y52A genes, pZL ligated with fragment F1 containing two BBa-J23100 promoters and fragment F2 containing two tac promoters at ApaI/ XhoI, respectively, generating pZL1 and pZL2. Then pZL1 and pZL2 ligated with the hpaBC gene at HindIII/SpeI and the d-ldh Y52A gene at EcoRI/BamHI, which were amplified from plasmid pYBD4 with primers hpaBC F/R and d-ldh Y52A F/R, respectively, generating pZL3 and pZL4. The P BBa-J23100 -hpaBC of pZL3 was replaced with P tac -hpaBC of pZL4 at ApaI/SpeI to generate pZL5. The P BBa-J23100 -d-ldh Y52A of pZL3 was replaced with P tac -d-ldh Y52A of pZL4 at SpeI/BamHI to generate pZL6.
To construct plasmids pZL9 and pZL10 for expressing the hpaBC and d-ldh Y52A genes as one operon, pZL ligated with F3 containing the 5tacs promoter digested with ApaI/XhoI, generating pZL7. The rrnB P1 promoter cloned from BAK5 genome with primers rrnB F/R was inserted into pZL7 digested with ApaI/HindIII, generating plasmid pZL8. Then pZL7 and pZL8 ligated with the hpaBC and d-ldh Y52A genes, generating pZL9 and 10, respectively.

Modular integration into chromosome
The integration strains were derived from strain BAK5 by using the λ Red homologous recombination method [32]. Module 1 replaced the mao-paa cluster, and module 2 replaced the lacI gene. Module 3 was integrated at the locus between nupG and speC. The three recombinated fragments for targeted integration of corresponding modules (ydbL-P lacUV5 -aroG fbr -tyrA fbr -aroE-Chl-ydbA for module 1, lacZ-P trc -ppsA-tktA-glk-Chl-mhpR for module 2, nupG-P 5tacs -hpaBC-d-ldh Y52A -Chl-speC for module 3) were constructed by overlapping extension PCR. Here, construction process for module 1 integrating fragment as example was demonstrated in Additional file 1: Figure S1. Module 1 (P lacUV5 -aroG fbr -tyrA fbr -aroE) was amplified from pYBT5 employing primers M1 F/R. The ydbL fragment (500 bp upstream) and the ydbA fragment (500 bp downstream) as homologous arms were cloned from BAK5 genome with primers ydbL F/R and ydbA F/R, respectively. Plasmid pKD3 was used as a template to clone the chloramphenicol resistance cassette employing primers Chl F/R. The ydbL fragment and module 1 were assembled into the ydbL-M1 fragment by overlapping extension PCR, and the chloramphenicol resistance cassette and the speC fragment were assembled into the Chl-ydbA fragment. The final fragment ydbL-M1-Chl-ydbA was generated by combining fragments ydbL-M1 with Chl-ydbA, and then electrotransformed into strain BAK5 which contained plasmid pKD46. The positive clone was confirmed by PCR. Afterwards, the chloramphenicol resistance was eliminated with the help of plasmid pCP20, which was further verified by PCR, and the final strain BAK10 was obtained. The same procedure was performed for the integration of module 2 and module 3, respectively.

Fermentation media and cultivation conditions
Escherichia coli cells were cultivated in Luria broth (LB) for strain maintenance and seed preparation. Modified MOPS (morpholinepropanesulfonic acid) medium containing 1× MOPS minimal salt [33], yeast extract (1 g/L) and glucose (5 g/L) was used for the production of l-tyrosine. The cultivation was conducted at 37 °C and 220 rpm with 50 mL of modified MOPS medium in 250 mL of shake flasks for 24 h. YM9 medium contained glucose (5 g/L), Na 2 HPO 4 (6 g/L), KH 2 PO 4 (3 g/L), NH 4 Cl (1 g/L), NaCl (0.5 g/L), CaCl 2 (17 mg/L), MgSO 4 (58 mg/L) and yeast extract (1 g/L) was used for the production of SAA. The cultivation was conducted at 30 °C and 220 rpm with 50 mL of YM9 medium in 250 mL of shake flasks for 24 h. Appropriate antibiotics were added in medium when necessary: ampicillin (50 μg/mL) and chloramphenicol (20 μg/mL).
For bioreactor fermentation, seed culture (~400 mL) was inoculated into a 5 L fermenter (Bailun, Shanghai) containing 2.1 L fermentation medium, generating an initial OD 600 of ~0.6. The fermentation medium contained glucose (7.5 g/L), Na 2 HPO 4 (6.8 g/L), KH 2 PO 4 (8.5 g/L), NH 4 Cl (3 g/L), NaCl (0.5 g/L), CaCl 2 ·2H 2 O (0.07 g/L), MgSO 4 ·7H 2 O (1 g/L), yeast extract (5 g/L). The pH was maintained at 7.0 by automatic addition of 10 M NaOH solution. The fermentation was performed at 30 °C with a 2.5 L/min airflow, and the dissolved oxygen (DO) level was controlled at 30% (v/v) by changing the agitation speed from 300 to 600 rpm automatically. The feeding solution contained 500 g/L glucose and 60 g/L yeast extract. Samples of fermentation broth were periodically withdrawn for analysis. The fermentation experiments were carried out in triplicates.

Analytical methods
Cell growth was determined by measuring the optical density (OD 600 ) using a TU-1810 spectrophotometer. The concentration of residual glucose was quantified by a biosensor SBA-90 (Biology Institute of Shandong Academy of Sciences, China). To measure l-tyrosine, the sample was prepared as previously described [12]. To measure SAA and l-dopa, the broth sample was centrifuged directly and the supernatant was filtered through 0.22 μm syringe filter. All the metabolites were analyzed using Agilent 1200 HPLC system (LabAlliance Corp, USA) equipped with a C18 column (250 mm × 4.6 mm, 5 μm, Bonna-Agela, China) and a DAD detector (Agilent). The column temperature was set at 25 °C. 10 μL of sample was injected to the HPLC system for analysis. SAA, l-tyrosine and l-dopa were quantified at 281 nm. The mobile phase was methanol-water-acetate (20:80:0.1, v/v/v) and the flow rate was set at 1 mL/min. Identification and quantitation of compounds were verified by comparison of retention time and using a standard curve, which the R 2 coefficient was higher than 0.99.

Results and discussion
Construction of plasmid-free chassis strain for constitutive production of precursor l-tyrosine l-tyrosine is precursor for SAA biosynthesis (Fig. 1a). Accordingly, sufficient supplement of l-tyrosine facilitates the production of SAA. The biosynthetic pathway of l-tyrosine is tightly regulated by l-tyrosine feedback and transcription repression [34]. A variety of metabolic engineering approaches have been use to improve l-tyrosine production by deleting repressing gene tyrR and overexpressing feedback-resistant genes aroG fbr and tyrA fbr , and other genes of limited steps on plasmids [35,36]. Plasmid-mediated l-tyrosine producer strains needed the addition of corresponding antibiotics and IPTG to control gene overexpression of interest [37][38][39].
Although feedback-resistant genes aroG fbr and tyrA fbr were integrated into the tyrR locus of E. coli chromosome, inducer IPTG was required for l-tyrosine fermentation as the T7 promotor was used [40].
In previous work, we constructed an engineered l-tyrosine overproducing E. coli strain BAK5 with a plasmid overexpressing module 1 and module 2. Module 1 containing aroG fbr -tyrA fbr -aroE and module 2 consisting of ppsA-tktA-glk were driven by the lacUV5 and trc promoters (Fig. 1b), respectively. In order to eliminate the drawbacks of plasmid-mediated system, module 1 and module 2 were integrated into the chromosome of strain BAK5 for constitutive production of precursor l-tyrosine. The mao-paa cluster involved in the degradation of aromatic acids [41] was chosen as the target site to integrate module 1. The fragment ydbL-P lacUV5 -aroG fbr -tyrA fbr -aroE-Chl-ydbA was assembled by overlapping extension PCR, and replaced the long mao-paa cluster (~17.7 kb) on chromosome of strain BAK5 by λ Red homologous recombination method [32] as described in section of "Methods". Successful replacement of the maopaa cluster with module 1 was verified by the colony PCR, and strain BAK10 was generated (Fig. 2a).
In order to test integrative effect of module 1 on the production of l-tyrosine, the shake fermentation was carried out. As shown in Fig. 2b, l-tyrosine titer of strain BAK10 was 320.6 mg/L after 24 h of cultivation without IPTG induced, increasing 31.8% than that of strain BAK5. When inducer IPTG (0.1 M) was added into fermentation medium, the production of l-tyrosine had a 22.1% increment, indicating that the expression of module 1 was partially repressed by the negative regulatory gene lacI due to the existence of lacO in lacUV5 promoter region of module 1, we speculated that deleting lacI gene would further improve the production of l-tyrosine as well as eliminate the supplementation of inducer IPTG. Thus, we replaced lacI gene with module 2, and generated strain BAK11 (Fig. 2c). The l-tyrosine titer of BAK11 was 401.5 mg/L without the induction of IPTG (Fig. 2d), increasing 65.1 and 25.2% than those of strains BAK5 and BAK10, respectively. We also investigated whether IPTG could still regulate l-tyrosine production in strain BAK11. After induction with additional IPTG, l-tyrosine production of strain BAK11 was slightly decreased, along with the inhibition of cell growth (Fig. 2d), which might be resulted from the cellular toxicity of inducer IPTG [42]. Deletion of lacI gene completely eliminated the repression of LacI, and strain BAK11 constitutively produced precursor l-tyrosine without IPTG induction.

Screening promoter for constitutive expression of hpaBC and d-ldh Y52A to produce SAA
In the downstream of artificial biosynthetic pathway of SAA, module 3 contains the hpaBC and d-ldh Y52A genes, which code 4-hydroxyphenylacetate 3-hydroxylase and d-lactate dehydrogenase, respectively, and catalyze the reactions of 4HPP to SAA via two putative routes (Fig. 1a). To construct inducer-free SAA producing strain, promoters for the constitutive expression of the hpaBC and d-ldh Y52A genes were considered. The constitutive promoter BBa-J23100 has the strongest strength among BBa series of promoters in the Anderson promoter library (Registry of Standard Biological Parts, http:// parts.igem.org), and has been used to metabolically engineer E. coli for the constitutive production of l-tyrosine [43], butanol [44] and cadaverine [45]. In addition, hybrid tac promoter, derived from the trp and lacUV5 promoters, is well-known for the 2-7 folds higher strength than the parent promoters [46]. The tac promoter has been popularly employed in metabolic engineering [47,48]. Thus, two strong constitutive promoters BBa-J23100 and tac were chosen for expression of the hpaBC and d-ldh Y52A genes as bicistron. Moreover, according to the N-terminal coding sequences of the hpaBC and d-ldh Y52A genes, 25-bp sequences of 5′-UTR were designed to achieve the the maximum translation level using UTR Designer [49]. The BBa-J23100 and tac promoters followed by the 5′-UTR were put in the front of the hpaBC and d-ldh Y52A genes, and four expression plasmids were generated (Additional file 1: Figure S2), after being introduced into strain BAK11, respectively, strains BKD7, BKD8, BKD9 and BKD10 were constructed.
To test expression efficiency of the hpaBC and d-ldh Y52A genes under the control of the BBa-J23100 and tac promoters, the production of SAA in fermentation broth was analyzed by HPLC (Additional file 1: Figure  S3). As shown in Fig. 3a, among four expression patterns, the BBa-J23100 promoter in strain BKD7 gave the lowest titer of SAA (260.4 mg/L), while the tac promoter in strain BKD10 made the highest titer of SAA (409.5 mg/L), the combinations of the BBa-J23100 and tac promoters led to the moderate production of SAA. Notably, the broth turned brown or dark after 30 h fermentation. We suspected that the intermediate l-dopa could be accumulated from l-tyrosine and further converted into melanin by HpaBC [50]. As shown in Fig. 3a, l-dopa and l-tyrosine were detected in broth of four strains BKD7, BKD8, BKD9 and BKD10, and the higher accumulation of l-dopa and l-tyrosine, the lower production of SAA, indicating that the expression of the hpaBC and d-ldh Y52A genes driven by either the tac or BBa-J23100 promoter was inefficient for SAA production.
The rrnB P1 promoter was considered as a super constitutive promoter composing of a core promoter, a cisacting DNA sequence and a trans-acting transcription factor-binding site, and has a major role in high-level synthesis of rRNA during exponential growth of E. coli cells [51]. Although the tac promoter is stronger than the BBa-J23100 promoter, one copy tac promoter might not be enough to express target genes for SAA production (Fig. 3a). The tandem repetitive promoter was more powerful than unrepetitive one for gene expression [52]. Five repetitive tac core promoter had suitable strength for transcription control and allowed high production of polyhydroxybutyrate in E. coli [53]. Here, the synthetic 5 × tac (designated as 5tacs) and rnnB P1 promoters were employed to drive expression of the hpaBC-d-ldh Y52A as monocistron operon (Additional file 1: Figure S2). As shown in Fig. 3b, strain BKD12 (with P 5tacs ) produced 492.2 mg/L of SAA, 1.2-fold and 1.5fold higher than strains BKD10 and BKD11 (with P rrnB P1 , 198.5 mg/L), respectively. We noticed that broth of strain BKD11 turned brown at 27 h fermentation, while broth of strain BKD12 did not. Interestingly, intermediates l-tyrosine (76.1 mg/L) and l-dopa (24.5 mg/L) were detected in broth of strain BKD11, but not in broth of strain BKD12, consistent without the observation of brown color. It indicated that the hpaBC-d-ldh Y52A under the control of the 5tacs promoter could completely convert l-tyrosine to SAA.

Chromosomal integration of module 3 for plasmid-free production of SAA
In order to construct plasmid-free strain, chromosomal integration of module 3 was further adopted. The expression cassette of P 5tacs -hpaBC-d-ldh Y52A in module 3 was integrated at the position between nupG and speC, which was demonstrated to be the most transcriptionally active for the expression of inserted gene [54], and then plasmid-free strain BKD13 was constructed (Fig. 4a).
To estimate the constitutive production behavior of engineered strain BKD13, preliminary fermentation was conducted in shake flasks with different concentrations of the initial glucose. As shown in Fig. 4b, 86.2 mg/L of SAA was obtained with 5 g/L of glucose, and 45.2 mg/L of l-tyrosine was accumulated in broth (Fig. 4b). With 7.5 g/L of the initial glucose, the SAA titer was increased to 120.5 mg/L and l-tyrosine titer was decreased to 21.3 mg/L, while SAA was decreased and l-tyrosine was increased with higher initial glucose (10 g/L) (Fig. 4b).
However, no l-dopa was detected in broth.
In order to testify the potential of strain BKD13 for SAA production, the fed-batch fermentation of SAA was also carried out in shake flasks. As shown in Fig. 4c, with feeding glucose, the cell growth of strain BKD13 entered the stationary phase at 36 h and the SAA production was gradually accumulated to 620.15 mg/L at 48 h. Additionally, there were minor amounts of l-tyrosine (6.5 mg/L) and l-dopa (2.9 mg/L) in fermentation broth. The results indicated that the chromosomal integration of module 3 was efficient to produce SAA.

Fed-batch fermentation for constitutive production of SAA in 5 L bioreactor
To evaluate the performance of plasmid-free and inducer-free SAA producing strain BKD13, the fed-batch fermentation was further carried out in a 5 L bioreactor. Based on concentration of residual glucose, feeding solution was added into the bioreactor to maintain it lower than 1.0 g/L. As shown in Fig. 5, during the fermentation process, the consumption of glucose was used for cell growth as well as for SAA biosynthesis, thus, the production of SAA coupled with the formation of biomass [55]. The final titer of 5.6 g/L SAA with the maximal biomass at OD 600 ~ 86 was achieved at 60 h. Small amount of precursor l-tyrosine accumulated in the early stage of the fermentation and completely converted into SAA after 51 h, while tiny amount of l-dopa (5.5 mg/L) was detected only before 36 h, and the fermentation broth had never become brown or dark. These results indicated that the plasmid-free and inducer-free strain BKD13 could constitutively produce SAA from glucose. We previously constructed a plasmid-mediated strain BKD5, which produced 7.1 g/L of SAA and accumulated 53.8 mg/L of l-tyrosine in 70 h [12]. Compared to strain BKD5, strain BKD13 seems more efficient for the conversion of l-tyrosine precursor to SAA since no l-tyrosine was detected at the end of fermentation. However, cell growth of strain BKD13 was faster than that of strain BKD5 (OD 600 ~ 4.0), which might lead to lower titer of SAA. Additionally, one copy of SAA synthetic genes in the chromosome of strain BKD13 might not be sufficient to utilize the central carbon metabolites to synthesize SAA. Amplification of integrated SAA synthetic genes [56] and optimization of fermentation process would further improve SAA production of strain BKD13.

Conclusions
In this paper, we developed a constitutive SAA-overproducing E. coli strain by chromosomal engineering to overcome the drawbacks of plasmid-mediated expression systems. In order to streamline metabolic flux towards precursor l-tyrosine from glucose and eliminate the repression regulation of LacI, module 1 (P lacUV5 -aroG fbr -tyrA fbr -aroE) replaced mao-paa cluster of 17.7 kb and module 2 (P trc -ppsA-tktA-glk) replaced lacI gene in strain BAK5 by targeted integration and deletion. Furthermore, we screened the synthetic promoter 5tacs driving the expression of module 3 (hpaBC-d-ldh Y52A ) in one operon, which was further inserted at the position between nupG and speC. The plasmid-free and inducerfree E. coli strain BKD13 produced 5.6 g/L SAA from glucose at 60 h in a 5L fed-batch fermentation. Our work showed the opportunities for industrial production of SAA and its derivatives, like rosmarinic acid and salvianolic acids, as a good example without additional antibiotics nor inducers for engineered microbial fermentation.