Renewable synthesis of n-butyraldehyde from glucose by engineered Escherichia coli

Background n-Butyraldehyde is a high-production volume chemical produced exclusively from hydroformylation of propylene. It is a versatile chemical used in the synthesis of diverse C4–C8 alcohols, carboxylic acids, esters, and amines. Its high demand and broad applications make it an ideal chemical to be produced from biomass. Results An Escherichia coli strain was engineered to produce n-butyraldehyde directly from glucose by expressing a modified Clostridium CoA-dependent n-butanol production pathway with mono-functional Coenzyme A-acylating aldehyde dehydrogenase (Aldh) instead of the natural bifunctional aldehyde/alcohol dehydrogenase. Aldh from Clostridium beijerinckii outperformed the other tested homologues. However, the presence of native alcohol dehydrogenase led to spontaneous conversion of n-butyraldehyde to n-butanol. This problem was addressed by knocking out native E. coli alcohol dehydrogenases, significantly improving the butyraldehyde-to-butanol ratio. This ratio was further increased reducing media complexity from Terrific broth to M9 media containing 2% yeast extract. To increase production titer, in situ liquid–liquid extraction using dodecane and oleyl alcohol was investigated. Results showed oleyl alcohol as a better extractant, increasing the titer of n-butyraldehyde produced to 630 mg/L. Conclusion This study demonstrated n-butyraldehyde production from glucose. Through sequential strain and condition optimizations, butyraldehyde-to-butanol ratio was improved significantly compared to the parent strain. Results from this work may serve as a basis for further development of renewable n-butyraldehyde production. Electronic supplementary material The online version of this article (10.1186/s13068-017-0978-7) contains supplementary material, which is available to authorized users.


Background
Commodity chemical industry relies almost entirely on non-renewable petroleum and other fossil fuel resources, leading to environmental pollutions and the inevitable depletion crisis. As a potential solution, renewable chemical production through microbial conversion of biomass is an attractive direction for sustainability. However, chemicals natural to biological production are limited in quantity and type. Therefore, synthetic metabolic pathways are designed and engineered into microorganisms for increasing both the amounts and types of chemicals accessible to bio-based conversion. n-Butyraldehyde is a large volume chemical produced exclusively from hydroformylation of propylene ( Fig. 1a) with annual production of 7 million tons and estimated growth of 2-3% per year [1]. As a reactive chemical, n-butyraldehyde is a versatile intermediate for the synthesis of various C4 and C8 alcohols, carboxylic acids, amines, and esters. In particular, 2-ethylhexanol has been a large volume derivative from n-butyraldehyde for its use in synthesizing phthalate plasticizers. Furthermore, n-butyraldehyde is also the precursor to polyvinyl butyral, a common polymer used for laminated glass in automotive and architectural industries. Recently, n-butyraldehyde has also been shown to be a biological precursor for bio-propane production [2]. These widespread applications of n-butyraldehyde make it an ideal chemical to be produced renewably.
Biological production of aldehydes is limited due to toxicity and reactivity. While few aldehyde products have been produced by engineered microbes [3][4][5][6], the biochemical repertoire for aldehydes needs to be expanded to support the effort in sustainability. n-Butyraldehyde has been previously reported in a mutant strain of Clostridium acetobutylicum [7] lacking alcohol dehydrogenase, capable of secreting up to 1.6 g/L of n-butyraldehyde. However, Clostridia are more difficult to work with than other well-characterized microorganism such as Escherichia coli and Saccharomyces cerevisiae due to their complex physiology and metabolism, as well as having less developed genetic manipulation tools. In addition, facultative anaerobes such as E. coli are often preferred for bio-based chemical productions because they grow rapidly during aerobic cultivation and conserves carbon for production under anaerobic conditions, increasing product yield due to elimination of respiration. Therefore, commercial interests are in engineering E. coli for n-butyraldehyde production [8]. However, due to the presence of numerous native alcohol dehydrogenases (Adh) in E. coli, n-butyraldehyde is spontaneously converted to n-butanol, thereby lowering the yield of aldehyde. This same behavior was observed in isobutyraldehyde production in E. coli [4]. Through knocking out several endogenous genes coding for Adh in E. coli, isobutyraldehyde production was significantly improved to roughly 2.5 g/L in test tubes and up to 35 g/L with gas stripping as in situ product removal. Inspired by the E. coli isobutyraldehyde production, here we also deleted adh genes and showed significant improvement in n-butyraldehyde production. Furthermore, in the process of constructing a n-butyraldehyde production pathway, we identified an alternative and better CoA-acylating aldehyde dehydrogenase than what has been previously reported. Lastly, instead of using gas stripping for in situ product removal as has been demonstrated for isobutyraldehyde, we tested in situ removal through organic overlay for liquid-liquid extraction and showed that oleyl alcohol is a suitable extractant for n-butyraldehyde production. The results obtained from this study provided a method for renewable synthesis of n-butyraldehyde with significantly reduced butanol co-production.

Strains and plasmids construction
Strains and plasmids used in this study are listed in the Additional file 1: Table S1. Primer sequences used are Fig. 1 a Chemical synthesis of n-butyraldehyde from petrochemical feedstock. Propylene and syngas are reacted under high temperature and pressure to form n-butyraldehyde. Commercially important downstream products are shown as representative applications. b Metabolic pathway for n-butyraldehyde biosynthesis from glucose. Six genes are overexpressed to produce n-butyraldehyde. Native adh genes coding for alcohol dehydrogenases are knocked out to prevent excessive reduction of n-butyraldehyde. Fdh formate dehydrogenase; AtoB acetyl-CoA acetyltransferase; Hbd 3-hydroxybutyryl-CoA dehydrogenase; Crt crotonase; Ter trans-enoyl-CoA reductase; Aldh CoA-acylating aldehyde dehydrogenase listed in Additional file 1: Table S1. Strains ELeco1 to KS8 were constructed from JCL299 [9] by sequential deletion of aldehyde reductase genes. All gene deletions were carried out using P1 transduction [10] with Keio collection [11] as donor strains. Kanamycin resistance marker was removed via FLP-mediated recombination. The successful gene deletions were subsequently verified by PCR (Additional file 1: Figure S1). All plasmids in this study were constructed using Gibson assembly [12]. Plasmids pKU48, pKU49, pKU50, and pKU51 were constructed by replacing adhE2 in pRW13 with aldh (CB), aldh (CB(mut)), aldh (CS), and aldh (CS(N1-4)), respectively. Briefly, a fragment containing plasmid vector, atoB, crt, and hbd was amplified using primers KU115 and KU116 using pRW13 as a template. This fragment was assembled with individual aldh fragments amplified by the specified primers in Additional file 1: Table S1 using the genomic DNA of the corresponding source organism. aldh (CB(mut)) gene was cloned from a cloning vector containing the gene in our lab collection. For its sequence, see Additional file 1. Plasmids were then verified through sequencing.

n-Butyraldehyde production
1% (v/v) of overnight cultures in LB was used to inoculate 3 mL of production media (TB or M9 with varying concentration of yeast extract) containing 20 g/L glucose with appropriate antibiotics in test tubes. When the cultures reached OD 600 of 0.4-0.6, they were switched to anaerobic by transferring them into a 10-mL BD vacutainer tube. Head space was then purged with anaerobic gas (95% N 2 , 5% H 2 ). Cultures were then sampled at specified times for optical density measurement and product quantification.

In situ removal of n-butyraldehyde by liquid-liquid extraction
1% (v/v) of overnight cultures were inoculated into 20 mL TB supplemented with 20 g/L glucose in 250-mL baffle flasks. When the culture OD600 reached 0.4-0.6, and 10 mL or 20 mL of either dodecane or oleyl alcohol was added as extractant to the culture. Subsequently, the cultures were switched to anaerobic by placing them in BD GasPak Anaerobe Gas Generating Pouch System with Indicator. When taking samples, anaerobic pouch was opened inside an anaerobic chamber to maintain the anaerobic environment.

Product quantification
Culture samples (1 mL) were centrifuged at 10,000×g for 5 min. The supernatants were then collected for product analysis. When analyzing product content in extractant, extractant was diluted using ethyl acetate. Aldehyde and alcohol concentrations in both medium and extractant were quantified by a Shimadzu GC-2010 gas chromatography (GC) equipped with a barrier ionization discharge (BID) detector. The separation of compounds was performed by SH-Rtx-wax GC column (30 m, 0.32 mm i.d., 0.50-μm-thick film). GC oven temperature was initially held at 40 °C for 2 min and increased with a gradient of 5 °C/min until 80 °C followed by a gradient of 12 °C/min until 120 °C. Then the temperature continues to rise with a gradient of 20 °C/min until 230 °C and held for 2 min. Helium was used as the carrier gas. The injector was maintained at 220 °C, and the detector was maintained at 230 °C. 1 μL of samples was injected in split injection mode (1:15 split ratio) using 2-methyl-1-pentanol or 1-pentanol as the internal standard. Glucose consumption was determined by subtracting the glucose concentration in samples from the concentration in original medium. Glucose concentration was measured using Agilent 1260 HPLC equipped with a refractive index detector. The injection volume used was 20 μL. The mobile phase consisted of 5 mM H 2 SO 4 with a linear flow rate of 0.6 mL/min. Separation of metabolites was done by Agilent HiPlex-H (700 × 7.7 mm) organic acid analysis column maintained at 65 °C. A Bio-Rad Micro-Guard Cation H guard column (30 × 4.6 mm) was connected in front of the analysis column. Glucose was monitored by refractive index detector. Concentration of glucose in the collected samples was determined by standard curve constructed from HPLC analysis of standard glucose solutions.

Partition coefficient determination for n-butyraldehyde in dodecane and oleyl alcohol
To measure the partition coefficient of n-butyraldehyde for dodecane and oleyl alcohol, 800 μL of different concentrations of n-butyraldehyde (0.01, 0.02, 0.05, 0.1, 0.2% in water) were mixed with the same volume organic extractant in glass GC vials. The mixtures were sealed and mixed by vortex for 1 min followed by a 36-h incubation in 37 °C. After incubation, the concentrations of n-butyraldehyde in each phase were determined by GC as described in "Methods" section. The partition coefficients were calculated using the equation below:

Selection of CoA-acylating aldehyde dehydrogenase for n-butyraldehyde production n-Butyraldehyde is an intermediate in the Clostridium
CoA-dependent n-butanol production pathway [9,13,14] (Fig. 1b). However, bifunctional aldehyde/alcohol dehydrogenase AdhE2 catalyzes the direct two-step conversion of butyryl-CoA to n-butanol, bypassing n-butyraldehyde as a product. To avoid conversion of n-butyraldehyde to n-butanol, we first replaced AdhE2 with CoA-acylating aldehyde dehydrogenase (Aldh), catalyzing only the conversion of butyryl-CoA to n-butyraldehyde. Some Clostridia such as Clostridium beijerinckii contain individual Aldh and Adh instead of a bifunctional enzyme for butanol production. Alternatively, Aldh is found in the degradation pathways for ethanolamine and 1,3-propanediol [15,16]. However, the Aldh from ethanolamine and 1,3-propanediol utilization operons are not specific for butyryl-CoA reduction and have been previously shown to produce ethanol when expressed in E. coli [17]. Therefore, for the present study, we chose to work with Clostridium Aldh. Based on the sequence of aldh from C. beijerinckii [18], we selected two additional homologues from C. saccharolyticum and C. saccharoperbutylacetonicum, as well as a mutant aldh from C. beijerinckii which we isolated previously in our lab (see Additional file 1 for its sequence). These four aldh genes were individually cloned into synthetic operons with the genes necessary to convert acetyl-CoA to butyryl-CoA (Fig. 1b). These synthetic operons were driven by native E. coli promoter of ack and adhE genes, P ack and P adhE , respectively, which have been previously shown to produce higher titers of butanol compared to using IPTGinducible P LlacO1 promoter [19]. Here the P ack and P adhE were defined to include the ribosomal binding site and 5′ untranslated region upstream of their corresponding genes. atoB, aldh, crt, and hbd were cloned as one operon on a colE1 origin plasmid under the control of P ack . ter and fdh were individually expressed on colA and pSC101 origin plasmids, respectively, under the control of P adhE [19]. These plasmids were transformed into E. coli strain JCL299 which was previously shown to efficiently produce n-butanol and has ldhA, adhE, frdBC, and pta knocked out [9]. Having the mixed acid fermentation pathways knocked out, JCL299 efficiently channels acetyl-CoA and NADH for the synthesis of n-butyraldehyde. As expected, due to the presence of endogenous alcohol dehydrogenases, the resulting strains showed minimal production of n-butyraldehyde (Fig. 2a). Majority of the fermentation products was n-butanol across the strains expressing the four different aldh genes. Chromosomal yqhD, coding for NADPH-dependent alcohol dehydrogenase, is known to reduce aldehydes to their corresponding alcohols and highly active as a detoxification mechanism [20][21][22]. Therefore, we knocked out yqhD in JCL299, yielding strain ELeco1. Expressing the n-butyraldehyde pathway in strain ELeco1, n-butyraldehyde production was observed. The best strain ELeco1/ pKU48/pRW18/pRW22 produced 0.16 g/L of n-butyraldehyde (Fig. 2b). However, the aldehyde-to-alcohol ratio was 0.39. This low aldehyde-to-alcohol ratio indicates the presence of other active native Adh capable of reducing n-butyraldehyde.

Improving aldehyde-to-alcohol ratio by knocking out native alcohol dehydrogenases
To decrease n-butanol formation and increase aldehydeto-alcohol ratio, we knocked out the genes coding for other native Adh. Based on previous work for isobutyraldehyde production [23], we deleted eight adh genes that are likely to contribute to n-butyraldehyde reduction: yjgB, fucO, eutG, ybbO, adhP, gldA, yahK, and yghA. These genes were selected because their knockouts led to higher production titers of isobutyraldehyde. We sequentially knocked out each of these adh genes using P1 phage transduction with the Keio collection. Since all of these adh genes have been shown to be effective for increasing isobutyraldehyde production, they were knocked out without specific order. The results of n-butyraldehyde production titers and the aldehyde-to-alcohol ratio from these mutant strains are shown in Fig. 3. We noted that the titers achieved by the ELeco1/pKU48/ pRW18/pRW22 are comparable to those achieved by the published patent application from Easel Biotechnologies [8], which used a strain with identical genotype. However, because the n-butanol titers were not reported in their work, the butyraldehyde-to-butanol ratios cannot be compared. Nevertheless, here we took the strain design further and showed that additional knockouts of native adh genes led to significant reduction in butanol. The final strain KS8/pKU48/pRW18/pRW22 reached an aldehyde-to-alcohol ratio of 3.1, representing an eightfold improvement compared to strain ELeco1/pKU48/ pRW18/pRW22. Here we noted that strain KS7 harboring the same plasmids reached slightly higher aldehyde-toalcohol ratio of 3.4. However, the difference in aldehydeto-alcohol ratios was within error range and insignificant. Therefore, strain KS8 was used for downstream experiments. Among the eight additional adh that we knocked out, eutG, ybbO, and yghA showed no effect for increasing aldehyde-to-alcohol ratio. All other adh knock out contributed positively towards increasing aldehydeto-alcohol ratio, indicating their native expression and corresponding enzymes' capability of reducing n-butyraldehyde. While aldehyde-to-alcohol ratio increased with each adh gene knock out, the titer of n-butyraldehyde did not significantly increase. Correspondingly, glucose consumption by strains with each additional adh gene knock out also decreased (Fig. 3). These results indicate that carbon flux was reduced. Analysis of the thermodynamics of each step revealed that butyryl-CoA reduction to n-butyraldehyde is thermodynamically unfavorable with ΔG′° of 7.7 kJ/mol (calculated using eQuilibrator [24]), which may lead to inefficient conversion of butyryl-CoA to n-butyraldehyde, particularly after n-butyraldehyde concentration reached a certain threshold. It is likely that as n-butyraldehyde production is slowed, glucose metabolism also slows due to inability for NADH recycling. Since the native E. coli fermentation genes (adhE, frdBC, and ldhA) have been knocked out in strain KS8, n-butyraldehyde production becomes the only fermentative pathway available to recycle NADH back to NAD + . When n-butyraldehyde biosynthesis is slowed down, less NAD + is available for use in glycolysis. As a result, glucose consumption rate decreases. Therefore, we next investigated the effect of in situ removal on n-butyraldehyde production titer.

Improving n-butyraldehyde titer by in situ product removal
Here we chose to use in situ liquid-liquid extraction for n-butyraldehyde removal using organic overlay. Dodecane and oleyl alcohol were selected as extractants because of their general application and non-toxicity to microbial cultures [25,26]. We determined the partition coefficient of n-butyraldehyde in water and the two organic solvents by measuring the ratio of n-butyraldehyde appearing in both aqueous and organic phase (Additional file 1: Figure S2 Table 1. Triangles in the table indicate gene knock out. Error bars represent standard deviation of three experiments 0.764 for dodecane and oleyl alcohol, respectively. This result indicated that oleyl alcohol may be a more suitable extractant for n-butyraldehyde as higher partition coefficient indicates the higher ratio of n-butyraldehyde found in the organic layer. These two extractants are only mildly toxic to E. coli as the growth of the n-butyraldehyde producing strain cultivated in the presence of either dodecane or oleyl alcohol were only slightly lowered compared to the control without any extractant (Fig. 4a), indicating the suitability of these solvents for in situ extraction. As results shown in Fig. 4b, n-butyraldehyde titer significantly improved in the presence of extractant. Consistently for both dodecane and oleyl alcohol, using 1 volume of extractant outperforms that using 0.5 volume. The best condition using oleyl alcohol with a 1:1 extractant-to-culture volume ratio produced over 0.6 g/L of n-butyraldehyde, representing a near three-fold improvement over no extractant. As expected, oleyl alcohol outperformed dodecane as extractant for n-butyraldehyde (Fig. 4c), consistent with the partition coefficients.

Effect of reducing media complexity on n-butyraldehyde production
Next, we evaluated the effect of yeast extract and tryptone concentration on n-butyraldehyde production. Using terrific broth (TB) is typically not commercially viable due to its expensive cost. Furthermore, aldehydes are reactive and can spontaneously form Schiff base with amines. Since TB contains high amounts of yeast extract and tryptone, aldehydes may likely be spontaneously reacted with the amino groups present on amino acids and oligopeptides in TB. Using M9 media with glucose as the base, we supplement 0 to 2% yeast extract or tryptone to determine an optimum level. Results are summarized in Fig. 5. Figure 5a shows the effect of yeast extract concentration on the production of n-butyraldehyde after 24 h of anaerobic incubation. n-Butyraldehyde titer was not significantly sensitive to yeast extract concentration between 0.125 and 2%. However, if no yeast extract was added, only minimal amount of n-butyraldehyde was observed, indicating the importance of complex nitrogen source. Interestingly, those using M9 with yeast extract had a higher aldehyde-to-alcohol ratio than that using TB due to higher levels of butanol produced. 48-h postanaerobic switch (Fig. 5b) showed slight decrease in n-butyraldehyde and increase in n-butanol titer, indicating functional alcohol dehydrogenase actively converting n-butyraldehyde to n-butanol. n-Butyraldehyde production was more sensitive to tryptone concentration than that of yeast extract as cultures containing 0.125 and 0.25% tryptone showed lower n-butyraldehyde titer compared to the corresponding concentrations of yeast extract (Fig. 5c, d). Increasing tryptone concentration led to increased n-butanol production, indicating that tryptone contributed towards the lowered aldehyde-to-alcohol ratio for using TB as production media. By comparing the components of yeast extract and tryptone from the manufacturers' manual, we noticed that tryptone has higher percentage of larger molecules with molecular weight in the range of than 500-2000 Da, indicating a larger amount of oligopeptides. On the other hand, yeast extract contains mostly smaller molecules with molecular weight less than 250 Da. It is possible that this discrepancy led to different expression patterns which may include nonspecific native alcohol dehydrogenases capable of reducing n-butyraldehyde. Nonetheless, the exact mechanism to why tryptone causes increase in n-butanol production is unclear.

Conclusions
This study demonstrated n-butyraldehyde production from glucose using engineered E. coli. We showed that aldh gene from C. beijerinckii outperformed the other aldh genes tested in achieving highest butyraldehyde-tobutanol ratio. Subsequent knockouts of endogenous adh genes including yqhD, yjgB, fucO, adhP, gldA, and yahK, in situ product removal by oleyl alcohol, and medium optimization using M9 2% glucose with 1-2% yeast extract significantly improved both the n-butyraldehyde titer (from 10 mg/L to 630 mg/L) and butyraldehydeto-butanol ratio. Compared to E. coli glucose-based n-butanol production (with titer up to 15 g/L in test tubes), n-butyraldehyde production using similar strain and pathway resulted in significantly lower titer. It is possible to achieve renewable n-butyraldehyde production via bio-butanol followed by chemical conversion. The chemical conversion of n-butanol to n-butyraldehyde is possible using Cu [27,28]-or Pt [29]-based catalysis. However, the Cu-based catalysis requires high temperature of 500 to 800 K. While the Pt-based catalysis can produce n-butyraldehyde from n-butanol at lower temperatures, leaching of the expensive Pt-based catalyst increases cost of the overall process. Therefore, sugarbased direct production of n-butyraldehyde remains an attractive potential direction. In order for it to become industrially viable in the future, further optimization of genetic expression, media, and product removal techniques is necessary.

Additional file
Additional file 1. Supporting information.