Comparison of glucose, acetate and ethanol as carbon resource for production of acetyl-CoA derived chemicals in engineered Escherichia coli CURRENT

Background: Acetyl-CoA is a fundamental metabolite in Escherichia coli , and also a precursor for biosynthesis of chemicals and materials suitable for multiple applications. The acetyl-CoA synthesis route from glucose presents low atomic economy due to the release of CO 2 in pyruvate decarboxylation reaction. Because ethanol and acetate, both ordinary and inexpensive chemicals, can be converted into acetyl-CoA directly, they could be alternative substrates for production of acetyl-CoA derivatives. Results: In this study, the bifunctional reductase AdhE mutant (A267T/E568K), which converts ethanol into acetyl-CoA, was used to enable E. coli to grow on ethanol, and AMP-forming acetyl-CoA synthetase ACS was employed to enhance the ability of E. coli to utilize acetate. Several products derived from acetyl-CoA, including polyhydroxybutyrate, 3-hydroxypropionate, and phloroglucinol, were produced from glucose, ethanol, and acetate, respectively, by engineered E. coli strains. Compared with glucose and acetate, the strains grown on ethanol presented the highest production and yield of carbon source, and metabolome analysis revealed the reasons of high yield from ethanol. Conclusions: The conversion of ethanol into acetyl-CoA presents high atomic economy along with generation of reducing power, and the yield of target chemical from ethanol is much higher than those from glucose and acetate. All these results suggested that ethanol could be a suitable carbon source for production of acetyl-CoA derived bioproducts. µL pre-cooled methanol/acetonitrile v/v), and mixed After sonication 20 min on the mix was at -20 °C for 1 h to precipitate proteins. The mix was centrifuged for 15 min (13000 rpm, 4 °C) and supernatant was dried by a vacuum drying system. A targeted metabolic analysis was performed using an LC-MS/MS system. The dried metabolites were dissolved in 100 µL of acetonitrile/ H 2 O (1:1, v/v) and centrifuged (14000 × g) for 20 min. Electrospray ionization was conducted with an Agilent 1290 Infinity chromatography system and AB Sciex QTRAP 5500 mass spectrometer. NH 4 COOH 4 (10 mM) and acetonitrile were used as mobile phases A and B, respectively. The sample was placed in a 4℃ automatic sampler with a column temperature of 45℃, a flow rate of 300 µL/min and an injection volume of 2 µL. A binary solvent gradient was used as follows: A, NH 4 COOH; B, 0–18 min at 90–40% acetonitrile; 18-18.1 min at 40–90% acetonitrile; and 18.1–23 min at 90% acetonitrile. The LC-MS/MS was operated in the negative mode under the following conditions:

The most common metabolic route for acetyl-CoA synthesis in E. coli is the glycolysis pathway coupled with decarboxylation of pyruvate by pyruvate dehydrogenase [10]. Through this pathway, each mol of glucose is converted into 2 mol of acetyl-CoA with generation of 4 mol of NADH, 2 mol of ATP and 2 mol of CO 2 (Fig. 1). The release of CO 2 lowers the atomic economy of targeted chemical biosynthetic pathway, leading to decrease of theoretical production yield, titer and productivity [11].
To improve the atomic economy of biosynthesis process, fatty acids were tested as an alternative carbon source [7]. Catabolized through β-oxidation pathways, fatty acids are transformed into acetyl-CoA with 100% carbon recovery in addition to production of reducing power. Although the production and yield of target chemical were increased significantly [7], the water-insoluble nature of fatty acids makes it difficult to handle and monitor the fermentation process. Several synthetic pathways for acetyl-CoA from one-carbon substrates have been constructed in E. coli [12,13], but their efficiency remains far below what is required for effective production of chemicals and materials.
Acetate and ethanol are both ordinary and inexpensive commodity chemicals, which can be directly converted into acetyl-CoA by microorganism, and could be considered as alternative substrate for production of acetyl-CoA with high atomic economy, and then production of acetyl-CoA derived chemicals. In E. coli, acetyl-CoA is produced from acetate via two different pathways, AMP-forming acetyl-CoA synthetase (ACS) catalyzing the higher-efficient pathway and phsphotransacetylase/acetate kinase (Pta-AckA) catalyzing the lower-efficient pathway [14]. In both pathways, ATP is consumed for the production of acetyl-CoA from acetate, and no reducing power is produced during this process (Fig. 1), indicating that extra acetate is demanded to generate ATP and reducing power when acetate is used as sole carbon source.
E. coli possesses a bifunctional reductase AdhE, which catalyzes the two-step reduction of acetyl-CoA to acetaldehyde and then to ethanol during anaerobic growth [15]. Although both reactions catalyzed by AdhE are reversible, E. coli cannot grow aerobically on ethanol as the sole carbon source due to the insufficiency of adhE expression [16]

Results And Discussion
Tolerance of E. coli to ethanol and acetate Ethanol usually shows antimicrobial activity at high concentration, and acetate represses the growth rate and maximum cell density of E. coli [23]. To figure out appropriate concentrations of ethanol and acetate as carbon source, they were added into minimal medium containing 10 g/L glucose at a series of concentrations from 1 g/L to 20 g/L, respectively. As shown in Fig. 2, addition of ethanol slightly slowed the growth, but did not affect maximum cell density of E. coli W3110 wild-type strain, as well as acetate at concentrations of 1 g/L and 2 g/L. However, acetate at concentrations of 5-15 g/L severely retarded E. coli growth, and 20 g/L acetate almost inhibited bacterial growth. As E. coli can catabolize acetate naturally, the maximum cell density with addition of 5-15 g/L acetate was higher than that at control conditions. Therefore, the concentration of ethanol was fixed at 10 g/L, and 2 g/L acetate was supplied at the beginning of cultivation with follow-up addition when exhausted.
Construction of E. coli strain growing on ethanol As mentioned above, two systems, E. coli AdhE mutant and P. aeruginosa ExaABC, can support bacterial growth on ethanol as sole carbon source (Fig. 1). As the route employing AdhE mutant has some advantages like shorter pathway, more NADH production and no energy requirement, it was chosen to construct engineered E. coli strain which can grow using ethanol as sole carbon source. The gene encoding AdhE A267T/E568K mutant was cloned into the vector pACYCDuet-1 under the IPTGinducible promoter P lac1−6 [24], and transformed into E. coli W3110 strain, along with empty vector, to generate strains Q3092 and Q3352, respectively. When grown in LB broth and induced by IPTG, AdhE mutant protein was observed as distinct band with the expected molecular weight on SDS-PAGE ( Fig. 3A). Then, these two strains were inoculated into minimal medium supplemented with 10 g/L ethanol, but neither could grow under this condition, implying that overexpression of heterologous gene at beginning of cultivation might impair the cell growth seriously. So, 1 g/L glucose was added into the medium, and transcription of adhE mut gene was initiated at an OD 600 of 0.6. After 57 h cultivation in shaking flask, expression of AdhE mutant significantly increased the cell density, up to 6.7-time higher than the strain carrying empty vector (Fig. 3B), suggesting that appropriate expression of AdhE double mutant is efficient to support E. coli growth on ethanol.

Production Of PHB From Ethanol
PHB is the most representative biodegradable and biocompatible thermoplastic, and suitable for applications in packaging, medicine, pharmacy, and food industries [25]. It is synthesized from its precursor acetyl-CoA by three enzymes encoded by phaCAB operon of Ralstonia eutropha (formerly Alcaligenes eutrophus) (Fig. 4A) [26].
In this study, PHB was used as example to evaluate the feasibility of ethanol as carbon source for production of acetyl-CoA derived bioproducts, and the plasmid pBAD-Ae-pha carrying phaCAB operon [27] was transformed into strains Q3352 and Q3092 to generate Q3095 and Q3094, respectively.
Then, these two strains were grown in minimal medium containing 1 g/L glucose and 10 g/L ethanol, and expression of adhE mut and phaCAB was initiated by addition of IPTG and l-arabinose, respectively.
During the cultivation process, the OD 600 and residual ethanol concentration in medium were measured.

6
As shown in Fig. 4B, the final OD 600 of Q3094 strain was 19.77 ± 0.36, while the control strain Q3095 only reached an OD 600 of 3.65 ± 0.07. Growth of Q3094 strain depleted all ethanol in the medium, however the ethanol concentration in Q3095 culture decreased slightly, similar with that of medium without inoculation but incubated at 37 ℃ (Fig. 4C). After 54 h cultivation, the strain Q3094 accumulated 3.12 ± 0.15 g/L PHB representing 48.7% of cell dry weight (CDW), while the control strain Q3095 produced 0.40 ± 0.11 g/L CDW and trace amount of PHB (Fig. 4D). All these results demonstrated that E. coli with overexpression of AdhE mut acquired the ability to produce value-added bioproducts using ethanol as carbon source.
Comparison of glucose, ethanol and acetate as carbon source for PHB production Combining the reactions in Fig. 1A and Fig. 4A, the overall theoretical stoichiometry of PHB synthesis from various carbon source is deduced (Fig. 5A). It shows that the PHB production from both glucose and ethanol is coupled with generation of energy and reducing power, but extra acetate is oxidized to meet the demand of energy and reducing power in pathway from acetate to PHB. The theoretical yield of PHB from ethanol is 0.935 g/g, much higher than its theoretical yields from the other two feedstocks. So, it is hypothesized that ethanol could be an ideal carbon source for PHB production in engineered bacteria.
To verify this hypothesis, comparison of glucose, ethanol and acetate as carbon source for PHB production was conducted. The gene acs was cloned into the vector pACYCDuet-1 under the IPTGinducible promoter P lac1−6 , and co-transformed with pBAD-Ae-pha into E. coli W3110 to generate strain Q3140. Then Q3095, Q3094 and Q3140 were grown on glucose, ethanol and acetate, respectively, and the cell growth, consumption of carbon source, PHB production, and intracellular metabolites were monitored during the cultivation process.
After 54 h cultivation, the strain Q3094 presented the highest cell density, followed by strain Q3095 and Q3140 (Fig. 5B). At D1 and D2 stages shown in Fig. 5B, the consumption of carbon sources was measured. The strain Q3095 almost exhausted 10 g/L glucose in 24 h, while 4.81 ± 0.17 g/L ethanol and 1.29 ± 0.01 g/L acetate were consumed by strain Q3094 and Q3140, respectively. At the end of cultivation, the strain Q3094 consumed 9.68 ± 0.02 g/L ethanol, and 4.85 ± 0.13 g/L acetate was catabolized by strain Q3140 (Fig. 5C). The PHB production of Q3094 is 3.12 ± 0.18 g/L, and the yield of PHB from ethanol is 0.32 ± 0.02 g/g ethanol. Both PHB production and yield were much higher than those from glucose and acetate (Fig. 5D). These results suggested that ethanol have some advantages as carbon source for PHB production than glucose and acetate, such as high cell density, high production and yield.

Differences Between The Metabolism Of Glucose, Ethanol And Acetate
To figure out the differences between the metabolism of glucose, ethanol and acetate, the targeted metabolome analysis was performed. Relative amount of acetyl-CoA, by-products, and main metabolites in central carbon metabolism pathways was determined at D1 and D2 stages ( At D1 stage, Q3095 strain grown on glucose produced the most acetyl-CoA, which was 8.73 times and 6.91 times higher than strains Q3094 and Q3140, respectively, and the similar results were observed in supply of energy and reducing power (Fig. 6). It is probably because that glucose is the favorite carbon source of E. coli [28], and was rapidly metabolized. This was also confirmed by the highest consumption rate of glucose (Fig. 5C). At D2 stage, acetyl-CoA and NADH were undetectable in Q3095 strain as glucose had been depleted.
For metabolites in glycolysis pathway, Q3095 strain also showed the highest abundance at D1 stage.
In particular, the pyruvate amount in Q3095 strain was more than 20 times higher than those in the other two strains (Fig. 6). It was very reasonable because that glucose was metabolized mainly through glycolysis pathway, and in Q3094 and Q3140 strains these metabolites have to be produced from ethanol and acetate through gluconeogenesis as building blocks for synthesis of components for construction of cell architecture. At D2 stage, there was not significant difference between three strains.
In contrast, the relative amount of TCA cycle intermediates in Q3140 strain was significantly higher than those in the other two strains at both D1 and D2 stages (Fig. 6). This phenomenon proved our hypothesis that extra acetate was oxidized through TCA cycle to produce ATP and NADH. 8 Remarkably, two by-products, lactate and acetate, were produced in Q3095 strain in large quantities at D1 stage, and the accumulation of acidic products could inhibit further growth of Q3095 strain. The lactate production of Q3095 strain was 229.22 times and 48.59 times as much as those of Q3094 strain and Q3140 strain, respectively. The accumulation of lactate was consistent with the high contents of pyruvate and NADH in Q3095 strain. As is known, glucose is catabolized through glycolysis pathway to pyruvate, which is converted into acetyl-CoA under catalysis of pyruvate dehydrogenase, with reduction of NAD + to NADH in both glycolysis pathway and pyruvate dehydrogenation. When NADH could not be oxidized through respiratory chain sufficiently, the buildup of NADH rapidly inactivated the pyruvate dehydrogenase [29], leading to accumulation of pyruvate in cells, and the recycling of NAD + must be achieved through the reduction of some metabolites.
Given the circumstances, a large quantity of lactate was produced by lactate dehydrogenase LdhA via reduction of pyruvate with consumption of NADH molecule [30].
Furthermore, Q3095 strain accumulated acetate 10.33-fold higher than Q3094 strain at D1 stage. In addition to be oxidized through TCA cycle, acetyl-CoA could be converted into acetate by PTA and AckA, in coupling with generation of one ATP molecule [31]. It was demonstrated that the majority of the acetyl-CoA is converted into acetate through the PTA-AckA pathway, and only a minority is metabolized via TCA cycle to generate NADH and CO 2 when E. coli grows on glucose aerobically [32].
Therefore, acetate accumulated in medium in large amount. Acetate overflow is caused by an imbalance between the pathways of glycolysis and TCA cycle in rapidly growing cells, and severely decreases the yield of target chemicals from glucose [33,34].
When the glucose supply is depleted, lactate and acetate can be taken back into the cells and respired. So, the amount of lactate and acetate dramatically decreased at D2 stage. Based on the above, the growth of E. coli on glucose was classically a diauxie, consuming glucose in the first half and consuming lactate and acetate in the second half. It was reported that E. coli cells grown on glucose produce acetate and consume it after glucose exhaustion, but do not grow on acetate due to the decoupling of acetate anabolism and acetate catabolism, and the growth restores only after 9 prolonged exposure to acetate [35]. Here, it is believed that this glucose-lactate/acetate transition would delay the growth of E. coli, and further affect the production of target product.
Another thing that deserves special attention was that the levels of cyclic AMP (cAMP) in Q3095 and Q3094 strains were essentially the same, much lower than that in Q3140 strain grown on acetate at D1 stage. The secondary messenger cAMP is synthesized from ATP, catalyzed by the adenylyl cyclase Cya whose activity is controlled by glucose availability [36]. When the preferred carbon source glucose is absent, the intracellular cAMP concentration increases ~ 10-fold, converting the cAMP receptor protein CRP into an active form to activate a number of genes for utilization of carbon sources other than glucose [37]. The low cAMP level in Q3094 strain grown on ethanol indicated that ethanol could be one of preferred carbon sources of E. coli.

Production Of Other Acetyl-CoA Derivatives From Ethanol
To test whether ethanol is suitable for production of other acetyl-CoA derived chemicals, the biosynthetic pathways for PG and 3HP from glucose, ethanol and acetate was constructed in E. coli, respectively, and comparison of those three carbon sources was carried out. 3HP and PG are both derived from the intermediate malonyl-CoA, which is produced from acetyl-CoA under catalysis of acetyl-CoA carboxylase ACC. Then malonyl-CoA is reduced to 3HP by malonyl-CoA reductase MCR, or catalyzed by polyketide synthase PhlD to form PG (Fig. 7A). The theoretical yields of 3HP and PG from various carbon sources were calculated, and those from ethanol were the highest (Fig. 7A).
Then, the 3HP-and PG-producing E. coli strains from glucose, ethanol and acetate were constructed, respectively, and were inoculated into minimal medium with corresponding carbon sources. After cultivation in shaking flasks, Q3398 strain grown on ethanol produced 0.50 ± 0.01 g/L 3HP, and Q3433 strain grown on ethanol produced 0.38 ± 0.02 g/L PG, much higher than strains grown on the other two carbon sources. These results demonstrate that ethanol can be used as carbon source for production of acetyl-CoA derived chemicals besides PHB.
Ethanol is a suitable carbon source for production of acetyl-CoA derivatives In this study, E. coli strain growing on ethanol was constructed, and the bioconversion of ethanol into a series of bioproducts including PHB, 3HP and PG was achieved in recombinant E. coli strains. Furthermore, metabolome analysis was carried out to discover the differences between metabolism of glucose, ethanol and acetate in engineered E. coli strains. All these results suggested that ethanol may be a suitable carbon source for production of acetyl-CoA derived bioproducts.
Firstly, ethanol is an ordinary and inexpensive commodity chemical. Besides fermentation of lignocellulosic biomass [38], ethanol can also be produced from coal by chemical methods. There are three ethanol-producing routes from coal with syngas [39], acetate [40], and dimethyl ether [41] as intermediate, respectively, which were all applied in industrial scale. The capacity of ethanol production from coal has exceeded 2 million tons per year in China. Secondly, the conversion of ethanol into acetyl-CoA presents high atomic economy, in addition to the generation of NADH.
However, CO 2 is released in glucose catabolism and ATP is consumed to synthesis acetyl-CoA from acetate ( Fig. 1). Therefore, the theoretical yield of target chemical from ethanol is much higher than those from glucose and acetate (Fig. 5A, Fig. 7A and 7B). Furthermore, ethanol has a higher energy density. If oxidized completely to CO 2 and H 2 O in bacteria, 0.326 mol of ATP is generated from 1 g ethanol, while 0.178 and 0.133 mol of ATP can be produced from 1 g glucose and acetate, respectively. Moreover, E. coli is easy to reach a balance between growth and production when grown on ethanol. As glucose is the favorite carbon source of E. coli, it was rapidly metabolized in the first half of cultivation (Fig. 5C) to produce a large amount of lactate and acetate (Fig. 6), leading to a biphasic growth which may delay the further bacterial growth and lower the production of target chemical. The assimilation of ethanol was neither as fast as glucose to accumulate by-products, nor as slow as acetate to retard the bacterial growth, helping bacteria reach a balance between growth and production. Additionally, the nature of ethanol, such as low toxicity and high water-solubility, makes it friendly to the bacterial cultivation process.
Although our study showed the feasibility of ethanol to support E. coli growth and bioproduction, there are still some problems remaining. The ethanol utilizing gene adhE mut was carried by a plasmid vector, leading to addition of antibiotic into medium and increasing of production cost. Besides that, the strain performance may be affected by the strain instability due to plasmid loss. This problem can 11 be dissolved by the integration of adhE mut gene into bacterial chromosome. Furthermore, the PHB yield was 0.32 ± 0.02 g/g ethanol in our engineered strain, representing only 34% of the theoretical limit (Fig. 4), and the productions of 3HP and PG from ethanol were both much lower than previous reports from other carbon sources probably due to the lack of overexpression of acetyl-CoA carboxylase [6,7]. It is necessary to carry out further development to achieve a higher production and yield from ethanol.

Conclusion
In summary, the bifunctional dehydrogenase AdhE A267T/E568K mutant, which converts ethanol into acetyl-CoA, was introduced into E. coli strain, conferring E. coli the growth capability on ethanol.
Several products derived from acetyl-CoA, such as PHB, 3HP and PG, were produced from ethanol by recombinant E. coli strains. Compared with glucose and acetate, the strains grown on ethanol presented the highest production and yield, and metabolome analysis revealed the reasons of high yield from ethanol. All these results demonstrate that ethanol is a putative carbon source for production of acetyl-CoA derived bioproducts.

Ethics approval and consent to participate
Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
All data generated during the present study are included in this published article.
This study was financially supported by NSFC (31722001 and 31670089), Natural Science Foundation of Shandong Province (JQ201707), National Defense Science and Technology Innovation Zone Foundation of China, and Defense Industrial Technology Development Program (JCKY2018130B005).

Authors' contributions
GZ and MX convinced and designed the study. SS, YD, and ML performed the strain construction, fermentation, and metabolome analysis. SS, LM, MX, and GZ analyzed the data. SS and GZ wrote the manuscript. All authors read and approved the final manuscript.