Coproduction of Acetaldehyde and Hydrogen during Glucose Fermentation by Escherichia coli

Abstract

Escherichia coli K-12 strain MG1655 was engineered to coproduce acetaldehyde and hydrogen during glucose fermentation by the use of exogenous acetyl-coenzyme A (acetyl-CoA) reductase (for the conversion of acetyl-CoA to acetaldehyde) and the native formate hydrogen lyase. A putative acetaldehyde dehydrogenase/acetyl-CoA reductase from Salmonella enterica (SeEutE) was cloned, produced at high levels, and purified by nickel affinity chromatography. In vitro assays showed that this enzyme had both acetaldehyde dehydrogenase activity (68.07 ± 1.63 μmol min⁻¹ mg⁻¹) and the desired acetyl-CoA reductase activity (49.23 ± 2.88 μmol min⁻¹ mg⁻¹). The eutE gene was engineered into an E. coli mutant lacking native glucose fermentation pathways (ΔadhE, ΔackA-pta, ΔldhA, and ΔfrdC). The engineered strain (ZH88) produced 4.91 ± 0.29 mM acetaldehyde while consuming 11.05 mM glucose but also produced 6.44 ± 0.26 mM ethanol. Studies showed that ethanol was produced by an unknown alcohol dehydrogenase(s) that converted the acetaldehyde produced by SeEutE to ethanol. Allyl alcohol was used to select for mutants with reduced alcohol dehydrogenase activity. Three allyl alcohol-resistant mutants were isolated; all produced more acetaldehyde and less ethanol than ZH88. It was also found that modifying the growth medium by adding 1 g of yeast extract/liter and lowering the pH to 6.0 further increased the coproduction of acetaldehyde and hydrogen. Under optimal conditions, strain ZH136 converted glucose to acetaldehyde and hydrogen in a 1:1 ratio with a specific acetaldehyde production rate of 0.68 ± 0.20 g h⁻¹ g⁻¹ dry cell weight and at 86% of the maximum theoretical yield. This specific production rate is the highest reported thus far and is promising for industrial application. The possibility of a more efficient “no-distill” ethanol fermentation procedure based on the coproduction of acetaldehyde and hydrogen is discussed.

INTRODUCTION

Acetaldehyde is an important compound that is widely used in the food and chemical industries. It is one of the ingredients that impart the typical yogurt flavor and is used as an additive in a large variety of dairy products, beverages, and other food products to enhance the fruit flavor and overall taste (8). Acetaldehyde is also of value in organic synthesis and is commonly used as an electrophile in condensation reactions such as 1,2 additions and aldol additions (32). The global production of acetaldehyde in 2003 was over 10⁶ tons (14). Currently, the majority of industrial acetaldehyde is produced by the oxidation of ethylene via the Wacker-Hoechst process (39). However, because ethylene is derived from petroleum, which is a nonrenewable resource, alternative sources of acetaldehyde are desirable.

Prior studies reported the production of acetaldehyde by bioconversion of ethanol by the use of alcohol oxidase (AOX) (19, 23, 24, 27). Kierstan used purified AOX from Candida boidinii to convert ethanol to acetaldehyde, which was evaporated out of the reactor (19). Other research groups developed analogous processes using purified enzymes or whole cells expressing AOX (23, 24, 27). Previous research also investigated the use of bacteria to convert glucose to acetaldehyde. Zymomonas mobilis alcohol dehydrogenase (Adh) mutants were used to produce acetaldehyde from glucose via the pyruvate decarboxylase (Pdc) reaction (38). In processes employing Lactococcus lactis, acetaldehyde was produced using strains engineered to overproduce Z. mobilis Pdc and L. lactis NADH oxidase. In a Streptococcus thermophilus study, the glyA gene was used to produce acetaldehyde from threonine, but the yield was typically below 1 mM (10). Although progress has been made, as yet, the production of acetaldehyde from renewable carbon is not commercially viable. Improved yield and specific productivity are needed, and an anaerobic process using a single organism would also have economic advantages.

In this study, we engineered Escherichia coli for coproduction of acetaldehyde and H₂ during glucose fermentation. This was done by deleting the native fermentation pathways of E. coli (frdC, ldhA, ackA-pta, and adhE) and introducing an exogenous acetyl-coenzyme A (CoA) reductase/acetaldehyde dehydrogenase (ACR/ALDH) (Fig. 1). This approach allowed the production of acetaldehyde with the highest yield and productivity reported to date. In addition, the process is anaerobic and produces H₂ as a valuable coproduct.

Fig. 1.

Engineered fermentation pathway for the coproduction of acetaldehyde and H₂ by Escherichia coli. Glucose is taken up by the phosphotransferase system (PTS), forming glucose-6-P and one molecule of pyruvate. A second molecule of pyruvate is formed by the pyruvate kinase (PK) reaction. Subsequently, 2 pyruvate molecules are converted to 2 acetyl-CoA molecules. The pathways shown in gray, which are used for the mixed acid fermentation, were knocked out. The eutE gene of Salmonella enterica encodes an acetyl-CoA reductase (ACR) that converts acetyl-CoA to acetaldehyde. The net result of the engineered pathway is glucose + 2 ADP + 2 Pi → 2 acetaldehyde + 2 H₂ + 2 CO₂ + 2 ATP.

MATERIALS AND METHODS

Chemicals and reagents.

Restriction enzymes, Taq polymerase, and T4 ligase were purchased from New England BioLabs Inc. (Beverly, MA). Acetyl-CoA, MBTH (3-methyl-2-benzothiazolinone hydrazone), and antibiotics were from the Sigma-Aldrich Corporation (St. Louis, MO). Other chemicals and reagents were purchased from Fisher Scientific (Pittsburgh, PA).

General protein methods.

Denaturing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed using Bio-Rad Redigels or TGX precast gels and Bio-Rad Mini-Protean II or Tetra electrophoresis cells according to the manufacturer's instructions. Following gel electrophoresis, Coomassie brilliant blue R-250 was used to stain proteins. The protein concentration of solutions was determined using Bio-Rad protein assay reagent (Bio-Rad).

General molecular methods.

Agarose gel electrophoresis was performed as previously described (29). Plasmid DNA was purified by the alkaline lysis procedure (29) or by using Qiagen products (Qiagen, Chatsworth, CA) according to the manufacturer's instructions. Following restriction digestion or PCR amplification, DNA was purified using Qiagen PCR purification or gel extraction kits. Restriction digests were carried out using standard protocols (29). For ligation of DNA fragments, T4 DNA ligase was used according to the manufacturer's directions (New England BioLabs). Electroporation was carried out as previously described using a Bio-Rad GenePulser (5).

Bacterial strains and culture conditions.

The bacterial strains used in this study are listed in Table 1. The rich medium used was modified lysogeny broth, which is sold as Luria-Bertani (LB)/Lennox medium (Difco, Detroit, MI) (4, 21). The minimal medium used was no-carbon-E (NCE) (3, 35). Antibiotics were used at the following concentrations: kanamycin (Kan) at 25 mg liter⁻¹, ampicillin (Amp) at 100 mg liter⁻¹, and chloramphenicol (Cm) at 20 mg liter⁻¹.

Table 1.

Bacterial strains used in this study

Strain	Genotype	Reference or source
BE400	E. coli strain K-12 substrain MG1655	Laboratory collection
BE100	E. coli BL21(DE3) RIL (Cmr) (for expression of rare RIL codons [from Stratagene])	Laboratory collection
ZH99	BE100, pBE522 (Kanr, Cmr)	Laboratory collection
ZH100	BE100, pBE522-eutE (Kanr, Cmr)	This study
ZH103	BE100, pBE522-His6-SeEutE	This study
ZH39	BE400, pBE522 (Kanr)	This study
ZH40	BE400, pBE522-eutE (Kanr)	This study
NA19	BE400, adhE::FRT	This study
ZH58	BE400, adhE::FRT, pLAC22 (Ampr)	This study
ZH54	BE400, adhE::FRT, pLAC22-adhE (Ampr)	This study
ZH97	BE400, adhE::FRT, pBE522 (Kanr)	This study
ZH98	BE400, adhE::FRT, pBE522-eutE (Kanr)	This study
ZH44	BE400, pta::FRT	This study
ZH57	BE400, pta::FRT, pBE522 (Kanr)	This study
ZH55	BE400, pta::FRT, pBE522-pta (Kanr)	This study
ZH94	BE400, pta::FRT, pBE522-eutE (Kanr)	This study
ZH46	BE400, ackA::FRT	This study
ZH64	BE400, ackA::FRT, pLAC22 (Ampr, Tetr)	This study
ZH65	BE400, ackA::FRT, pLAC22-ackA (Ampr)	This study
ZH95	BE400, ackA::FRT, pBE522 (Kanr)	This study
ZH96	BE400, ackA::FRT, pBE522-eutE (Kanr)	This study
ZH28	BE400, ldhA::FRT	This study
ZH41	BE400, ldhA::FRT, pBE522 (Kanr)	This study
ZH132	BE400, ldhA::FRT, pBE522-ldhA (Kanr)	This study
ZH42	BE400, ldhA::FRT, pBE522-eutE (Kanr)	This study
ZH71	BE400, frdC::FRT	This study
ZH91	BE400, frdC::FRT, pLAC22 (Ampr, Tetr)	This study
ZH90	BE400, frdC::FRT, pLAC22-frdC (Ampr)	This study
ZH109	BE400, frdC::FRT, pBE522 (Kanr)	This study
ZH108	BE400, frdC::FRT, pBE522-eutE (Kanr)	This study
ZH30	BE400, adhE::FRT, ackA-pta::FRT	This study
NA20	BE400, adhE::FRT, ldhA::FRT	Laboratory collection
ZH4	BE400, adhE::FRT, ldhA::FRT, pBE522 (Kanr)	This study
ZH3	BE400, adhE::FRT, ldhA::FRT, pBE522-eutE (Kanr)	This study
ZH111	BE400, adhE::FRT, ackA-pta::FRT, pBE522 (Kanr)	This study
ZH110	BE400, adhE::FRT, ackA-pta::FRT, pBE522-eutE (Kanr)	This study
ZH87	BE400, adhE::FRT, ldhA::FRT, ackA-pta::FRT, frdC::FRT, pBE522 (Kanr)	This study
ZH88	BE400, adhE::FRT, ldhA::FRT, ackA-pta::FRT, frdC::FRT, pBE522-eutE (Kanr)	This study
ZH134	BE400, adhE::FRT, ldhA::FRT, ackA-pta::FRT, frdC::FRT, pBE522-eutE (Kanr), allyl alcohol resistant, isolate 1	This study
ZH135	BE400, adhE::FRT, ldhA::FRT, ackA-pta::FRT, frdC::FRT, pBE522-eutE (Kanr), allyl alcohol resistant, isolate 2	This study
ZH136	BE400, adhE::FRT, ldhA::FRT, ackA-pta::FRT, frdC::FRT, pBE522-eutE (Kanr), allyl alcohol resistant, isolate 3	This study

Growth of strains for analysis of fermentation products.

Strains were streaked from frozen stocks to LB agar containing appropriate antibiotics. A single colony was used to inoculate 2 ml of LB medium with antibiotic(s), and cultures were incubated overnight at 37°C. A 1-ml volume of this culture was centrifuged at 10,000 × g, and cells were resuspended in 1 ml of NCE glucose minimal medium (0.4% [wt/vol] glucose, NCE, 1 mM magnesium sulfate, 1 mM IPTG [isopropyl-β-d-thiogalactopyranoside], 10 μM sodium molybdate, 3.6 μM ferrous citrate, 1 μM sodium selenate, 1 μM nickel chloride). Cell suspensions were used to inoculate 6 ml of NCE glucose minimal medium at an initial optical density at 600 nm (OD₆₀₀) of 0.2 into 18-by-150-mm serum vials (Bellco Glass, Vineland, NJ). Vials were sealed with rubber stoppers and aluminum crimp seals (Wheaton) inside an anaerobic chamber (Coy Laboratory Products Inc., Grass Lake, MI). When appropriate, the pH was lowered by addition of 1 M phosphoric acid. All cell cultures made for fermentation product analysis were grown anaerobically at 37°C in an I2400 incubator shaker (New Brunswick Scientific) at 275 rpm for 24 h, unless otherwise stated. After appropriate incubation, growth was determined by measuring cell density at 600 nm using a Spectronic D20+ spectrophotometer (Thermo Scientific), and fermentation products were measured using high-performance liquid chromatography (HPLC) and chemical methods as described below.

HPLC analysis of fermentation products.

After 24 h of incubation, cells were spun down and the supernatant was filtered through a 0.22-μm-pore-size Millex syringe filter with a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA). The filtrate was analyzed using a Bio-Rad Aminex HPX-87H column (300 by 7.8 mm) with a Varian ProStar HPLC system that included a 230 solvent delivery module, a 430 autosampler, a 325 UV-visible light (UV-Vis) detector, a 355 refractive index detector, and a MetaTherm column heater (Varian, Palo Alto, CA). The column was heated to 50°C, and the reaction mixture was eluted at 0.3 ml min⁻¹ with 5 mM H₂SO₄ (isocratic). Lactate, succinate, acetate, and formate were quantitated by monitoring absorbance of 210 nm and comparing peak areas to a standard curve. Ethanol was similarly quantitated using the refractive index detector. Standard solutions were prepared using 5 mM sulfuric acid as a solvent.

Chemical quantitation of glucose and acetaldehyde.

Glucose levels were determined using a glucose (GO) assay kit (Sigma-Aldrich Corp.). Acetaldehyde levels were determined using the MBTH assay as described with the following modifications (34). Assay mixtures contained 186 μl of water, 143 μl of 0.1% MBTH, 286 μl of 100 mM potassium citrate at pH 3.6, and 100 μl of appropriately diluted sample and were incubated at 37°C for 15 min prior to measurement of absorbance at 305 nm.

Hydrogen measurement.

Hydrogen was measured by injecting 50 μl of headspace gas from a sealed culture tube into a Varian CP-3800 gas chromatograph equipped with a CP-Molsieve 5A column and a thermal conductivity detector. The temperatures of the injector, the column, and the detector were maintained at 200°C, 40°C, and 220°C, respectively. The carrier gas was nitrogen (flow rate, 29 ml min⁻¹). Under these conditions, the retention time of hydrogen was 0.67 min.

Construction of deletion mutants.

The maternal parent of the deletion mutants used in these studies was E. coli K-12 MG1655. Single-gene knockout mutants were from the Keio collection and were purchased from the Genome Analysis Project in Japan (1). P1 transduction was used to move specific deletions from the Keio collection mutants into E. coli K-12 MG1655, selecting for kanamycin resistance. Transductants were colony purified and then transformed with pCP20, which expresses the Flp recombinase, to remove the kanamycin resistance gene as previously described (13). Multiple chromosomal deletions were made by repeating P1 transduction and Flp recombination with additional mutants from the Keio collection.

Construction of plasmids for protein production and complementation.

To construct strains for high-level protein production, the genes encoding SeEutE and His₆-SeEutE were cloned via PCR (17) into a T7 expression plasmid. The template was chromosomal DNA from Salmonella enterica serovar Typhimurium LT2, and the primers were primer pair eutEBgl-f and eutEHinDIII-r and primer pair eutEHTN-f and eutEHTN-r (see Table S1 in the supplemental material). The restriction sites used for cloning were BglII and HindIII, and the vector was pBE522. Vector pBE522 was constructed from pET41-a (EMD Chemicals, Gibbstown, NJ) by replacing the DNA between its SphI and NdeI sites with the following linker: GCATGCAATTAATACGACTCACTATAGGGGAATTGACAATTAGTTAACTATTTGTTATAATGTATTCCGGGGAATTGTGAGCGGATAACAATTCCCCTCTAGAAATAATTTTGTTTAACTTTAAGAAGG AAGATCTCATATG. The linker, which was made by annealing synthetic oligonucleotides as previously described (17), provides both T7 and E. coli consensus promoters that are regulated by IPTG, making it useful for protein production using both wild-type E. coli and DE3 lysogens.

The vectors used for complementation studies were pBE522 for ldhA and pta and pLAC22 (36) for adhE, ackA, and frdC. For both vectors, protein production is regulated by IPTG. The adhE, pta, ackA, ldhA, and frdC genes were each cloned into these vectors via PCR using genomic DNA of E. coli K-12 MG1655 as the template. The pta gene was amplified with forward primer ptaBam-f and reverse primer ptaNco-r (Table S1), digested with BamHI and NcoI, and ligated into pBE522 cut with BglII and NcoI. The adhE gene was amplified with primers adhEBam-f and adhESph-r (Table S1), digested by BamHI and SphI, and ligated into LAC22 cut with BglII and SphI. The ackA, frdC, and ldhA genes were amplified with primer pair ackABglII-f and ackAHinDIII-r, primer pair frdCBglII-f and frdCHindIII-r, and primer pair ldhABglII-f and ldhAHindIII-r (Table S1), digested by BglII and HindIII, and ligated into similarly cut vector. The DNA sequences of all clones were verified prior to further studies.

Preparation of cell extracts.

To obtain cell extracts of the SeEutE production strain, −80°C stocks were streaked onto LB agar plates containing kanamycin and incubated at 37°C overnight. A single colony was used to inoculate 2 ml of LB-kanamycin medium in a polystyrene tube (Evergreen Scientific, Los Angeles, CA) (17 by 100 mm). After overnight incubation at 37°C and shaking at 275 rpm, 1 ml of culture was used to inoculate 40 ml of LB medium containing kanamycin. IPTG was added to achieve a concentration of 1 mM when the cell density reached 0.6. The culture was then incubated overnight. Cells were harvested by centrifugation at 10,000 × g using an Avanti J-25 centrifuge and a JA-17 rotor (Beckman Coulter, Brea, CA). Cell pellets were lysed with bacterial protein extraction reagent (BPER) (Thermo Scientific), and the lysate was centrifuged to separate the supernatant from the pellet according to the manufacturer's instructions.

To obtain extracts from cells grown anaerobically on glucose, a fresh colony from LB-kanamycin plates was used to inoculate 2 ml of LB-kanamycin broth in a polystyrene tube (Evergreen Scientific, Los Angeles, CA) (17 by 100 mm). After overnight incubation at 37°C and shaking at 275 rpm, 2 ml of the culture was spun down, resuspended in 2 ml of NCE glucose minimal medium, and used to inoculate 400 ml of NCE glucose minimal medium in a 1-liter flask (Bellco Glass, Vineland, NJ). The flask was sealed with a rubber stopper wrapped with Parafilm inside an anaerobic chamber. The flask was taken out of the chamber and incubated in a 37°C incubation shaker for 24 h. The cells were harvested as described above.

Purification of His₆-SeEutE.

An overnight culture of the His₆-SeEutE production strain (ZH103) was prepared from a single colony as described above and used to inoculate 40 ml of LB-kanamycin medium at a 1:50 dilution in a 125-ml Pyrex flask (Corning Inc., Corning, NY). This culture was grown at 37°C, and when the OD at 600 nm reached 0.6, IPTG was added to achieve a concentration of 1 mM. After induction, cells were grown at 30°C overnight and pelleted by centrifugation at 10,000 × g for 10 min at 4°C. The supernatant was removed, and 1.5 ml of BPER was added to the pellet. This mixture was gently shaken for 10 min at room temperature and centrifuged at 15,000 × g for 15 min. The supernatant was removed, and 1.5 ml of BPER was used to resuspend the pellet. Freshly prepared lysozyme was added to the resuspended pellet at a final concentration of 200 μg ml⁻¹, and the suspension was incubated at room temperature for 5 min. Next, 4.5 ml of BPER diluted 1:10 in 20 mM Tris-HCl (pH 7.5) was added and the sample was mixed well by pipetting up and down and placed on ice for 30 min. This mixture was centrifuged at 35,000 × g for 30 min, and the supernatant was passed through a 0.45-μm-pore-size syringe filter. The filtered sample was loaded onto a 5-ml nickel-nitrilotriacetic acid (Ni-NTA) column (Qiagen) that had been previously equilibrated with 15 ml of 20 mM Tris-HCl (pH 8.0), 300 mM NaCl, and 20 mM imidazole. The column was washed with 7.5 ml of equilibration buffer and 100 mM imidazole and eluted with 3.75 ml (5 aliquots of 0.75 ml) of a similar buffer containing 300 mM imidazole that was collected separately.

Aldehyde dehydrogenase (ALDH) and acetyl-CoA reductase (ACR) enzyme assays.

Assays were performed by monitoring A₃₄₀ using a Cary 50 Bio UV-Vis spectrophotometer (Varian, Inc., Palo Alto, CA). The assay buffer was 35 mM potassium phosphate (pH 8.0) supplemented with 50 mM KCl. When desired, the pH was adjusted by adding 1 M phosphoric acid. All reagents were made using assay buffer as the solvent. The assay volume was 1 ml, and reagents were used at the following concentrations: for NADH, 400 μM; for acetyl-CoA, 500 μM; for dithiothreitol (DTT), 1 mM; for NAD⁺, 500 μM; for coenzyme A, 500 μM; and for acetaldehyde, 2 mM. Quantitation was done using ε₃₄₀ = NADH 6.22 mM⁻¹·cm⁻¹.

Allyl alcohol selection.

A fresh colony of ZH88 was used to inoculate 2 ml of LB medium with kanamycin into a polystyrene tube (Evergreen Scientific, Los Angeles, CA) (17 by 100 mm). A 100-μl volume of overnight culture was spread on an LB agar plate containing 25 g of kanamycin liter⁻¹, 0.4% (wt/vol) glucose, and 0.2 mM, 2 mM, 10 mM, 20 mM, 50 mM, or 100 mM allyl alcohol. The plates were incubated at 37°C in an anaerobic chamber. Three colonies isolated on the plate with 100 mM allyl alcohol were colony purified on similar medium and then on an LB agar plate with kanamycin.

Dry cell weight measurement and determination of specific production rate.

Cells were grown anaerobically in 400 ml of NCE glucose minimal medium with 1 g of yeast extract/liter at pH 6.0. At the desired time points, 40 ml of cell culture was removed, spun down, and washed once with 40 ml of double-distilled water. The cell pellet was transferred to a dry, preweighed microcentrifuge tube and then dried overnight in an oven at 200°C and weighed. The dry cell weight was determined at measured OD values to calculate a standard curve. The specific rate of production of acetaldehyde was calculated as follows: Δ acetaldehyde (in grams/liter)/Δ dry cell weight (in grams/liter) × growth rate (h⁻¹). The growth rate was calculated as follows: (log₁₀ t2 − log₁₀ t1)/(t2 − t1) × 2.303. Measurements were done in the time period from 0 to 8 h.

RESULTS

The S. enterica EutE protein catalyzes the conversion of acetyl-CoA to acetaldehyde.

The purpose of this study was to engineer E. coli to coproduce acetaldehyde and H₂ from glucose (a renewable carbon source). Our approach required an enzyme that catalyzes the following reaction: acetyl-CoA + NADH + H⁺ → acetaldehyde + HS-CoA + NAD⁺ (Fig. 1). In this report, we refer to the forward reaction as written above as ACR (acetyl-CoA reductase) and the reverse reaction as ALDH (acetaldehyde dehydrogenase). Based on sequence analyses and genetic tests, the S. enterica EutE enzyme (SeEutE) was previously proposed to be a CoA-dependent ALDH used for B₁₂-dependent ethanolamine degradation (33); however, its enzymatic activity and reversibility were not previously reported. To test the activity of SeEutE, a recombinant enzyme was produced using an E. coli T7 expression strain (ZH100). SDS-PAGE showed that this strain produced relatively large amounts of protein near the molecular mass of SeEutE (49 kDa) whereas the control strain containing the expression plasmid without the insertion (ZH99) produced little protein near 49 kDa (Fig. 2a). Enzyme assays showed that crude cell extract from the SeEutE production strain had 14.27 ± 1.84 μmol min⁻¹ mg⁻¹ ACR activity and 33.62 ± 3.39 μmol min⁻¹ mg⁻¹ ALDH activity. In contrast, control extracts lacked detectable activity for either reaction. These results provided biochemical evidence that SeEutE is a reversible enzyme that has both ALDH and ACR activity.

Fig. 2.

Production and purification of SeEutE and His₆-SeEutE. (a) SDS-PAGE analysis of SeEutE production. Lane M, molecular mass marker; lane 1, crude lysate of SeEutE production strain ZH100; lane 2, supernatant of ZH100; lane 3, pellet of ZH100; lane 4, crude lysate of control strain ZH99; lane 5, supernatant of ZH99; lane 6, pellet of ZH99. A 5-μg volume of protein was loaded in each lane. (b) SDS-PAGE analysis of His₆-SeEutE purification. Lane M, molecular mass marker; lane 1, crude lysate; lane 2, supernatant from the crude lysate; lane 3, pellet from the crude lysate; lane 4, solubilized pellet, filtered through a 45-μm-pore-size membrane; lane 5, flow through; lane 6, 100 mM imidazole wash; lane 7, 300 mM imidazole eluant. A 10% acrylamide gel was used, and proteins were stained with Coomassie. A 1-μg volume of protein was loaded in lane 7, and 5 μg of protein was loaded in each of the other sample lanes.

We also tested the solubility of recombinant SeEutE by centrifuging cell extract from the expression strain at 10,000 × g and measuring the ACR/ALDH activity in the pellet and supernatant fractions. Greater than 95% of the total activity was in the pellet fraction. This was somewhat unexpected, since insoluble fractions usually contain inactive misfolded protein. In this case, however, the insoluble fraction had high activity, indicating that the expression strain formed aggregates consisting of properly folded active SeEutE. A similar result was reported for the PduO protein, whose commonality with EutE is that both are associated with bacterial microcompartments (16). Aggregation may be a general property of microcompartment proteins.

Purification of His-tagged EutE.

The studies described above were done with native recombinant SeEutE. To facilitate purification, we constructed an E. coli strain (ZH103) to produce SeEutE with an N-terminal 6×His tag (His₆-SeEutE). Crude extracts from this strain had 7.89 ± 1.23 μmol min⁻¹ mg⁻¹ ACR activity, most (∼90%) of which was in the insoluble fraction, in similarity to what was observed for non-His-tagged enzyme (described above). The insoluble fraction was solubilized with BPER (a proprietary detergent mixture), and His₆-SeEutE was purified by Ni-NTA affinity chromatography. SDS-PAGE results indicated that the His₆-SeEutE was about 90% pure following Ni-NTA chromatography (Fig. 2). The specific ACR activity of the purified enzyme was 49.23 ± 2.88 μmol min⁻¹ mg⁻¹, and a 41% yield was obtained. For the same purified enzyme, the specific ALDH activity was 68.07 ± 1.63 μmol min⁻¹ mg⁻¹, and the yield was 34%. We note that when the pellet containing His₆-SeEutE was solubilized, >100% total activity was recovered (47.67 ± 0.70 μmol min⁻¹) from the crude lysate (47.01 ± 4.23 μmol min⁻¹). This suggests that aggregation may have slightly inhibited the enzyme activity.

Reaction requirements of the SeEutE enzyme.

Using SeEutE purified by Ni-NTA chromatography as described above, we examined its reaction requirements. No ACR activity was detected when NADH or acetyl-CoA was omitted from the reaction. No ALDH activity was detected when acetaldehyde or NAD⁺ and HS-CoA was omitted from assay mixtures. In addition, no alcohol dehydrogenase activity was detected in assays containing ethanol and NAD(P)⁺ or acetaldehyde and NAD(P)H. Thus, assays with purified enzyme provided direct experimental evidence that SeEutE represents a reversible ALDH/ACR process with relatively high specific activity.

Cofactor preference and kinetic constants of His₆-SeEutE.

Using purified SeEutE, we tested its preference for NAD⁺/NADP⁺. The ACR activity of SeEutE was 6% with NADPH compared to 100% with NADH. The ALDH activity was 3% with NADP⁺ compared to 100% with NAD⁺. Thus, SeEutE prefers NAD⁺/NADH and has low activity with NADP⁺/NADPH. Using purified enzyme, we also determined the kinetic constants for the ACR reaction. The K_m for acetyl-CoA was 23.4 ± 6.8 μM, with a V_max of 69.91 ± 2.20 μmol min⁻¹ mg⁻¹.

Effect of pH on His₆-SeEutE activity.

The effect of pH on the activity of His₆-SeEutE was examined (see Fig. S1 in the supplemental material). We tested pH values from 6 to 8 in increments of 0.5 units. Maximal ACR activity was at pH 6.5, and >60% activity was retained at pH 7.5. The ALDH activity of His₆-SeEutE was maximal at pH 7.5, and >50% activity was retained at pH 6.5. These results are similar to those of prior studies that found optimal ACR activities at pH 6.8 to 7.0 and optimal ALDH activities at pH 8.0 to 9.1 (25, 28, 42). Of the most significance for this study is the fact that SeEutE has high activity at physiological pH (7 to 7.5), which is important for high acetaldehyde production during fermentation.

Use of SeEutE acetyl-CoA reductase for the production of acetaldehyde from glucose.

E. coli grows anaerobically on glucose by mixed acid fermentation. The major end products of this process are ethanol, acetate, lactate, succinate, formate, H₂, and CO₂ (Fig. 1) (11, 43). As a first step toward the coproduction of acetaldehyde and H₂ during glucose fermentation, an eutE expression plasmid (pBE522-eutE) was transformed into wild-type E. coli MG1655. This was expected to redirect some of the acetyl-CoA produced during glucose degradation to acetaldehyde. However, when E. coli producing SeEutE (strain ZH40) was grown anaerobically on glucose for 24 h, no acetaldehyde production was detected (Table 2). Enzyme assays showed that cell extracts from ZH40 grown anaerobically on glucose had 15.74 ± 0.06 μmol min⁻¹ mg⁻¹ ACR activity, verifying that SeEutE was expressed under these growth conditions. Consequently, we inferred that the native E. coli fermentation enzymes that also use acetyl-CoA as a substrate (AdhE and Pta-Ack) outcompeted EutE for acetyl-CoA, inhibiting acetaldehyde formation, or that the acetaldehyde made by EutE was converted to ethanol by the bifunctional acetaldehyde dehydrogenase/alcohol dehydrogenase AdhE. To test these possibilities and increase acetaldehyde production, the eutE gene was expressed in genetic backgrounds carrying deletions of the adhE, ackA, pta, ldhA, and frdC genes as well as combinations of these mutations.

Table 2.

Cell growth and metabolite profile of various strains^a

Strain	Genotype							Cell growth		Glucose consumption (mM)	Coproduction of H2 and CH3CHO (mM)		Metabolite concn (mM)
	ldhA	frdC	adhE	ackA	pta	pBE522	pBE522-eutE	Doubling time (h)	OD600 at endpoint		Hydrogenb	Acetaldehyde	Succinate	Lactate	Formate	Acetate	Ethanol
Wild-type strains
BE400	+	+	+	+	+	−	−	4.26 ± 0.19	1.07 ± 0.04	22.24	22.50 ± 2.62	<0.1	3.80 ± 0.06	0.32 ± 0.01	0.53 ± 0.13	15.80 ± 0.87	15.87 ± 0.88
ZH39	+	+	+	+	+	+	−	3.32 ± 0.20	0.92 ± 0.06	22.24	24.17 ± 1.38	<0.1	3.92 ± 0.21	0.24 ± 0.01	0.16 ± 0.02	15.94 ± 1.02	16.38 ± 0.59
ZH40	+	+	+	+	+	−	+	5.33 ± 0.85	1.01 ± 0.02	22.24	24.77 ± 0.40	<0.1	2.98 ± 0.13	0.56 ± 0.04	0.12 ± 0.03	13.79 ± 1.15	17.79 ± 0.02
Single mutants
NA19	+	+	−	+	+	−	−	PGd	0.22 ± 0.06	1.23	NDc	<0.1	0.07 ± 0.01	0.48 ± 0.10	0.26 ± 0.05	0.78 ± 0.34	ND
ZH97	+	+	−	+	+	+	−	PG	0.15 ± 0.01	1.19	ND	<0.1	0.08 ± 0.03	0.60 ± 0.01	ND	1.02 ± 0.05	ND
ZH98	+	+	−	+	+	−	+	PG	0.29 ± 0.02	5.32	ND	0.51 ± 0.06	0.94 ± 0.04	2.77 ± 0.21	3.46 ± 0.23	2.70 ± 0.09	4.12 ± 0.83
ZH44	+	+	+	+	−	−	−	4.53 ± 0.57	0.75 ± 0.01	22.24	3.55 ± 0.44	0.13 ± 0.02	3.44 ± 0.05	30.96 ± 0.06	ND	0.85 ± 0.03	2.65 ± 0.17
ZH57	+	+	+	+	−	+	−	5.51 ± 0.71	0.70 ± 0.01	22.24	3.93 ± 0.19	0.15 ± 0.05	4.52 ± 0.01	29.84 ± 0.27	ND	0.74 ± 0.01	3.00 ± 0.12
ZH94	+	+	+	+	−	−	+	6.32 ± 0.65	0.82 ± 0.01	17.70	11.27 ± 0.66	0.68 ± 0.15	3.72 ± 0.09	11.92 ± 0.23	1.77 ± 0.29	0.60 ± 0.02	10.02 ± 0.87
ZH46	+	+	+	−	+	−	−	7.01 ± 0.39	0.78 ± 0.01	22.24	7.41 ± 0.63	0.25 ± 0.06	6.12 ± 0.17	23.14 ± 0.19	ND	3.85 ± 0.05	3.98 ± 0.45
ZH95	+	+	+	−	+	+	−	6.59 ± 0.07	0.73 ± 0.01	22.24	7.43 ± 0.39	0.28 ± 0.02	6.59 ± 0.06	22.77 ± 0.16	ND	3.99 ± 0.05	4.51 ± 0.28
ZH96	+	+	+	−	+	−	+	7.64 ± 0.55	0.54 ± 0.03	15.56	6.70 ± 0.87	0.91 ± 0.40	3.32 ± 0.24	6.71 ± 0.48	2.91 ± 0.64	1.21 ± 0.12	7.71 ± 0.69
ZH71	+	−	+	+	+	−	−	4.10 ± 0.34	1.02 ± 0.06	22.24	25.36 ± 0.67	<0.1	ND	0.33 ± 0.03	0.21 ± 0.21	16.15 ± 0.05	17.71 ± 1.21
ZH109	+	−	+	+	+	+	−	7.33 ± 0.19	0.96 ± 0.00	22.24	26.99 ± 0.12	<0.1	0.64 ± 0.05	0.27 ± 0.00	ND	17.04 ± 0.02	19.66 ± 0.95
ZH108	+	−	+	+	+	−	+	15.66 ± 0.47	1.09 ± 0.02	22.24	25.41 ± 0.12	<0.1	ND	0.83 ± 0.06	1.17 ± 0.07	12.70 ± 0.18	21.72 ± 0.20
ZH28	−	+	+	+	+	−	−	4.42 ± 0.17	1.09 ± 0.03	22.24	24.97 ± 0.88	<0.1	3.94 ± 0.08	ND	0.68 ± 0.06	16.51 ± 0.02	17.12 ± 0.19
ZH41	−	+	+	+	+	+	−	2.88 ± 0.56	1.00 ± 0.01	22.24	25.15 ± 0.66	<0.1	4.27 ± 0.04	ND	0.45 ± 0.03	16.77 ± 0.05	17.74 ± 0.20
ZH42	−	+	+	+	+	−	+	3.65 ± 0.76	1.06 ± 0.01	22.24	23.54 ± 0.54	<0.1	3.82 ± 0.04	ND	0.59 ± 0.04	14.22 ± 0.11	21.37 ± 0.07
Multiple mutants
ZH30	+	+	−	−	−	−	−	7.84 ± 0.56	0.68 ± 0.00	22.24	ND	0.21 ± 0.06	3.64 ± 0.07	36.92 ± 0.05	ND	0.64 ± 0.02	ND
ZH111	+	+	−	−	−	+	−	3.17 ± 0.29	0.62 ± 0.02	22.24	ND	0.16 ± 0.01	4.31 ± 0.01	35.92 ± 0.06	ND	0.59 ± 0.02	ND
ZH110	+	+	−	−	−	−	+	5.40 ± 0.64	0.65 ± 0.04	12.03	5.66 ± 0.68	1.28 ± 0.43	2.88 ± 0.05	9.48 ± 0.11	4.58 ± 0.04	0.48 ± 0.02	7.72 ± 0.15
NA20	−	+	−	+	+	−	−	PG	0.18 ± 0.07	ND	ND	0.21 ± 0.01	0.07 ± 0.01	ND	0.28 ± 0.06	0.74 ± 0.08	ND
ZH4	−	+	−	+	+	+	−	PG	0.16 ± 0.01	1.51	ND	0.21 ± 0.00	0.16 ± 0.01	ND	0.22 ± 0.15	0.80 ± 0.09	ND
ZH3	−	+	−	+	+	−	+	7.95 ± 0.97	0.23 ± 0.00	5.58	ND	1.15 ± 0.48	0.99 ± 0.04	ND	5.61 ± 0.08	3.18 ± 0.04	2.97 ± 0.03
ZH84	−	−	−	−	−	−	−	PG	0.21 ± 0.04	0.05	ND	<0.1	ND	ND	0.24 ± 0.06	0.21 ± 0.02	ND
ZH87	−	−	−	−	−	+	−	PG	0.17 ± 0.00	1.23	ND	<0.1	0.06 ± 0.04	ND	ND	0.28 ± 0.02	ND
ZH88	−	−	−	−	−	−	+	8.36 ± 1.07	0.46 ± 0.03	11.05	3.13 ± 0.65	4.91 ± 0.29	0.54 ± 0.03	ND	9.23 ± 0.11	ND	6.44 ± 0.26

^aResults were obtained after 24 h of anaerobic growth on glucose minimal medium.

^bThe concentration of H₂ was calculated as the amount of hydrogen (in moles) in the headspace divided by 6 ml, which was the volume of liquid culture used. This allowed direct comparison of the H₂ concentrations with the concentrations of fermentation products.

^cND, not detected.

^dPG, poor growth.

Construction of mutants and complementation tests.

To block possible competing fermentation pathways and, we hoped, direct more carbon flux toward acetaldehyde production by EutE, we constructed single-gene knockouts of E. coli MG1655 in which one of the native fermentation pathways was deleted. The adhE, ackA, pta, ldhA, and frdC deletions used in this study were from deletion mutants in the Keio collection (1). These mutations were constructed by a PCR-based method that replaced nearly the entire coding sequence of a target gene with a kanamycin resistance marker that can be removed using flp recombinase (13). Each single mutation was moved from the Keio collection into E. coli MG1655 separately by P1 transduction (22). The insertion site of the kanamycin marker was verified by PCR, and HPLC analyses were used to confirm that each single mutation altered the glucose fermentation profile of E. coli as expected (see Table S2 in the supplemental material). In addition, complementation tests were performed to show that mutant phenotypes were corrected by expression of the corresponding minimal clone from pLAC22 or pBE522 (see Table S2 in the supplemental material). Strains with multiple mutations were constructed by P1 transduction using deletion mutations characterized as described above.

Effect of an adhE mutation on acetaldehyde production by EutE during glucose fermentation.

The AdhE enzyme of E. coli converts acetyl-CoA to ethanol. Hence, this enzyme might compete with SeEutE for substrate (acetyl-CoA) or convert acetaldehyde to ethanol, limiting acetaldehyde production. To test these possibilities, an adhE mutant was transformed with pBE522 with and without the eutE gene and the resulting strains were cultured under standard conditions: NCE glucose minimal medium at 37°C for 24 h (Table 2). The AdhE mutant grew poorly and consumed little glucose. Prior work showed that this result is due to a redox imbalance (12, 15). Expression of the eutE gene in the adhE background partially restored glucose consumption and resulted in the production of about 0.5 mM acetaldehyde. In contrast, no acetaldehyde was produced by the control strain lacking the eutE gene. Thus, elimination of AdhE allowed acetaldehyde production via EutE. We think the most likely explanation is that deletion of adhE reduced competition for acetyl-CoA and forced the conversion of acetyl-CoA to acetaldehyde as a means of regenerating NAD⁺ from NADH.

It was also observed that the adhE mutant did not produce ethanol, which was as expected. However, production of EutE in the adhE deletion background resulted in substantial ethanol production (about 4 mM). This result cannot be easily attributed to the AdhE whose encoding gene was deleted. Neither can ethanol production be attributed to EutE, since in vitro studies did not detect any alcohol dehydrogenase activity associated with this enzyme. We think that the most likely source of the ethanol was an unknown alcohol dehydrogenase that uses the acetaldehyde produced by SeEutE as its substrate.

Effect of a single ackA or pta mutation on acetaldehyde production by SeEutE during glucose fermentation.

AckA and Pta convert acetyl-CoA to acetate during glucose fermentation by E. coli (Fig. 1). The K_m of the Pta enzyme for acetyl-CoA (44.9 ± 4.1 μM) (9) is in the same range as that of the SeEutE enzyme (23.4 ± 6.8 μM). Hence, Pta and AckA might also compete with EutE for acetyl-CoA. Therefore, we evaluated the effects of ackA and pta deletion mutations on acetaldehyde production. The single ackA or pta mutants fermented glucose reasonably well and produced much less acetate than the wild type, as expected (7) (Table 2). They also produced much less ethanol but substantially more lactate. Hence, the ackA and pta single mutants grew primarily by as a result of lactate fermentation. Expression of eutE in strains carrying an ackA single mutation and strains carrying a pta single mutation allowed the coproduction of acetaldehyde and H₂, with acetaldehyde reaching levels of 0.91 ± 0.40 and 0.68 ± 0.15 mM, respectively (Table 2). This suggests that the AckA-Pta pathway competes with EutE for acetyl-CoA, inhibiting acetaldehyde production. Hence, deletion of this pathway increased acetaldehyde production.

The results also showed that expression of eutE by strains that had an ackA or pta deletion inhibited glucose consumption, reduced lactate formation, and increased ethanol production. As described above, the increased ethanol production in strains producing SeEutE most likely resulted from the conversion of acetaldehyde to ethanol by an unknown endogenous alcohol dehydrogenase. Although the reason is uncertain, the reduced rates of growth and glucose consumption may have been due to redox imbalances caused by excess ethanol production, as previously observed (11).

Effect of ldhA and frdC deletions on acetaldehyde production.

Lactate and succinate are also end products of glucose fermentation by E. coli. These compounds are derived from phosphoenolpyruvate or pyruvate and hence might indirectly reduce acetyl-CoA pools, reducing acetaldehyde production. However, ldhA and frdC single deletions did not appreciably affect acetaldehyde formation (Table 2).

Acetaldehyde production via SeEutE in an ackA, pta, and adhE triple mutant.

The results described above showed that adhE, ackA, and pta single mutations resulted in increased acetaldehyde production by strains producing SeEutE, perhaps by reducing competition for acetyl-CoA. Next, we tested the effect of an ackA, pta, and adhE triple mutation. HPLC analyses showed that the major glucose fermentation products formed by the triple mutant were lactate (about 37 mM) and (secondarily) succinate (about 3.6 mM). Production via SeEutE in this mutant increased the acetaldehyde level to 1.28 ± 0.43 mM, which is higher than the level observed for any of the single mutants. Production via SeEutE also resulted in increased ethanol production (7.72 ± 0.15 mM) and decreased lactate formation. Hence, overall EutE production in the ackA, pta, and adhE triple mutant resulted in diversion of carbon from lactate to acetaldehyde and ethanol.

Acetaldehyde production via SeEutE in an ackA, pta, adhE, ldhA, and frdC quintuple mutant.

Up to this time point, the adhE, ackA, and pta triple mutant produced the highest level of acetaldehyde (1.28 ± 0.43 mM) during glucose fermentation. The mutant also produced substantial amounts of lactate, succinate, and ethanol, which may have limited acetaldehyde production. Therefore, we tested whether acetaldehyde production could be increased by moving frdC and ldhA deletions into strains that also carried ackA, pta, and adhE mutations. The resulting quintuple mutant (adhE, ackA, pta, ldhA, and frdC) was blocked in all major glucose fermentation pathways. This mutant did not ferment glucose appreciably without SeEutE production (Table 2). However, with SeEutE production, it grew with a doubling time of about 8.4 h and produced about 5 mM acetaldehyde. It is likely that SeEutE restored growth by allowing regeneration of NAD⁺ from NADH. Ethanol (6.44 ± 0.26 mM) was again observed when SeEutE was produced, even though this strain has an adhE deletion.

Hydrogen production during acetaldehyde formation.

The main goal of this work was to coproduce acetaldehyde and H₂. Therefore, we also monitored H₂ production during glucose fermentation by mutant strains (Table 2). The strains that produced the highest levels of acetaldehyde were the adhE, ackA, and pta triple mutant (1.28 ± 0.43 mM) and the adhE, ackA, pta, ldhA, and frdC quintuple mutant (4.91 ± 0.29 mM). These strains also produced 5.66 ± 0.68 mM and 3.13 ± 0.65 mM H₂, respectively. Thus, coproduction of H₂ and acetaldehyde was observed. However, in general, acetaldehyde-producing strains also produced significant amounts of formate. During glucose fermentation, H₂ and CO₂ are produced from formate by formate hydrogen lyase (2). Thus, under the conditions used, the production of H₂ from formate was incomplete.

Elimination of ethanol dehydrogenase activity by allyl alcohol selection.

During the course of these studies, we observed that strains expressing eutE produced ethanol even though the native alcohol dehydrogenase (AdhE) had been deleted. We attributed this to unknown E. coli alcohol dehydrogenase(s) acting on acetaldehyde produced by SeEutE. To eliminate this alcohol dehydrogenase activity, an allyl alcohol selection of ZH88 was performed (Adh enzymes convert allyl alcohol to acrolein, which is toxic [26]). Three allyl alcohol-resistant mutants were identified, and glucose fermentation by the mutants was examined (Table 3). All three mutants (ZH134, ZH135, and ZH136) produced reduced levels of ethanol and higher levels of acetaldehyde. The highest level of acetaldehyde obtained was 8.56 ± 0.56 mM. Hence, the allyl alcohol-resistant mutants showed increased acetaldehyde and substantially reduced ethanol production. We also attempted to eliminate the remaining alcohol dehydrogenase activity. Further selection using up to 500 mM allyl alcohol failed to yield additional mutants. We also tested a strain with a deletion of the adhP alcohol dehydrogenase (31), but this deletion did not affect ethanol production significantly during glucose fermentation by E. coli ZH88.

Table 3.

Metabolite production by 3 allyl alcohol-resistant mutants and their ZH88 parent strain^a

Strain	Acetaldehyde production (mM)	Ethanol production (mM)	Glucose consumed (mM/%)	OD600 at 24 h (anaerobic conditions)	Yieldb (%)
ZH134	8.56 ± 0.56	3.24 ± 0.54	9.95/45	0.33	43
ZH135	6.69 ± 0.50	2.48 ± 0.03	7.80/35	0.27	43
ZH136	6.98 ± 0.18	1.60 ± 0.10	6.74/30	0.26	52
ZH88	4.91 ± 0.29	6.44 ± 0.26	11.05/50	0.46	22

^aMetabolite measurements were made after 24 h growth on glucose minimal medium under anaerobic conditions at 37°C. Strains ZH134, ZH135, and ZH136 are ally alcohol-resistant mutants derived from ZH88.

^bThe yield was calculated based on the equation showing that one molecule of glucose is converted to two molecules of acetaldehyde.

Effect of yeast extract and pH on coproduction of acetaldehyde and hydrogen.

Although the 3 allyl alcohol-resistant mutants described above (ZH134 to ZH1136) produced higher acetaldehyde and lower ethanol levels, they grew slower and consumed less glucose than ZH88 and the conversion of formate to H₂ and CO₂ was incomplete under the conditions used (Table 3). To improve the coproduction of acetaldehyde and hydrogen, the strain with the highest acetaldehyde production (ZH134) was cultured under various conditions. Addition of 1 g of yeast extract/liter to the standard glucose minimal medium at pH 7.0 allowed ZH134 to utilize 50% more glucose while producing more hydrogen and acetaldehyde with no significant increase in ethanol production (Fig. 3 a). However, the hydrogen production was significantly below the expected 1:1 ratio with acetaldehyde. Prior studies showed that the conversion of formate to hydrogen and CO₂ by E. coli is enhanced at slightly acidic pH (30). Lowering the pH of the growth media from 7.0 to 6.0 substantially (3.6-fold) increased hydrogen production by ZH134, with only small changes in acetaldehyde (+0.93 mM) and ethanol (+1.83 mM) production and glucose consumption (−1.28 mM) (Fig. 3a). In summary, at pH 6.0 and with 1 g of yeast extract/liter added to the growth medium, ZH134 produced 17.60 ± 0.70 mM acetaldehyde and 17.19 ± 1.73 mM hydrogen, thus realizing the coproduction of acetaldehyde and hydrogen at a nearly 1:1 molar ratio and a 64% theoretical yield relative to the amount of glucose consumed (13.67 mM). Other than acetaldehyde, the only organic coproduct detected was ethanol, which was formed at 5.59 ± 0.22 mM. Acetaldehyde and ethanol combined were formed at an 85% theoretical yield.

Fig. 3.

Coproduction of acetaldehyde and hydrogen under selected growth conditions. (a) Effect of yeast extract (YE) (1 g/liter) and pH. The strain used was ZH134. The growth medium used was NCE glucose minimal medium at the indicated pH with and without yeast extract. (b) Comparison of ZH134 and ZH136 growing anaerobically in NCE glucose minimal medium with 1 g of yeast extract/liter at pH 6.0. ZH134 and ZH136 are both quintuple mutants (Table 1) that lack the native E. coli fermentation pathways and that produce SeEutE acetyl-CoA reductase from an expression plasmid. Both strains were obtained from allyl alcohol selection and have reduced alcohol dehydrogenase activity.

Coproduction of acetaldehyde and H₂ by ZH136.

Using optimized growth conditions, we tested a second isolate obtained from the allyl alcohol selection (ZH136) for coproduction of acetaldehyde and hydrogen. ZH136 was promising, because it produced relatively a high level of acetaldehyde and the lowest level of ethanol among the three mutants tested (Table 3). After 24 h of fermentation, ZH136 produced 15.78 ± 2.21 mM acetaldehyde, 15.97 ± 0.78 mM hydrogen, and 2.21 ± 0.23 mM ethanol while consuming 9.18 mM glucose. Overall, hydrogen was produced in at a nearly 1:1 molar ratio with acetaldehyde, which was formed at an 86% theoretical yield relative to the amount of glucose consumed (9.18 mM). Ethanol which was formed at 2.21 ± 0.23 mM, and acetaldehyde and ethanol combined were formed at a 98% theoretical yield.

Specific productivity for ZH136.

To determine specific productivity, acetaldehyde formation was measured over time (from 0 to 8 h) using medium that contained 1 g of yeast extract/liter at pH 6. Cell dry weight was determined as described in Materials and Methods. In three trials with strain ZH136, the specific productivity of acetaldehyde formation was 0.68 ± 0.20 g h⁻¹ g⁻¹ dry cell weight.

DISCUSSION

The goal of this study was to engineer E. coli to ferment glucose to produce 2 acetaldehyde + 2 H₂ + 2 CO₂ (Fig. 1). This process is redox balanced, generates 2 net ATP molecules per glucose molecule, and is energetically feasible: ΔG^o′ = −151 kJ/mol. Our approach required an enzyme that reduces acetyl-CoA to acetaldehyde. We used the SeEutE enzyme for this purpose. The specific activity of purified His₆-SeEutE in the direction of acetaldehyde formation was 49.23 ± 2.88 μmol min⁻¹ mg⁻¹. This gives a k_cat value of 40 s⁻¹ (57 s⁻¹ at pH 7.0), which corresponds to the production of 184 g of acetaldehyde per gram of SeEutE per hour. Thus, the activity of the SeEutE enzyme is suitable for commercial use.

When SeEutE was expressed in E. coli, maximum acetaldehyde production occurred in strains whose native fermentation pathways were eliminated by genetic deletion and in which alcohol dehydrogenase activity was reduced by allyl alcohol selection. Deletion of the native fermentation pathways helped reroute carbon to acetaldehyde. The allyl alcohol selection was necessary, because an unknown alcohol dehydrogenase converted the acetaldehyde produced by SeEutE to ethanol, reducing the yield. While the allyl alcohol selection substantially reduced ethanol formation, it did not eliminate it completely. We attempted to further reduce the alcohol dehydrogenase activity by repeating the allyl alcohol selection and by deletion of the adhP alcohol dehydrogenase gene (31) without success.

The coproduction of acetaldehyde and H₂, was further improved by adjusting the growth conditions. The best medium tested was NCE glucose minimal medium supplemented with 1 g of yeast extract/liter at pH 6.0. With this medium, ZH136 produced 15.78 ± 2.21 mM acetaldehyde and 15.97 ± 0.78 mM H₂ with good specific productivity (0.68 ± 0.20 g acetaldehyde h⁻¹ g⁻¹ dry cell weight) and theoretical yield (86%). To our knowledge, this is the best theoretical yield reported for the biological conversion of glucose to acetaldehyde. In studies of Zymomonas mobilis and Lactococcus lactis, reported yields were 40% and 47%, respectively (6, 38). In prior studies, the specific productivity levels of acetaldehyde formation were not reported (6, 38), but the specific productivity obtained here was comparable to that for the best ethanologenic E. coli strain in minimal medium, which was 0.47 to 0.52 g h⁻¹ g⁻¹ dry cell weight (44). In this study, optimal acetaldehyde production required a relatively small amount of yeast extract (1 g liter⁻¹). This medium would increase production costs compared to minimal medium, but it might be possible to eliminate this requirement by further genetic engineering, as was previously done for ethanol production by E. coli (44). We also note that, under the conditions used, cell growth was slow and glucose use was incomplete (about 41%), which is disadvantageous. However, this may have been due to product toxicity. Prior studies showed that acetaldehyde is toxic to bacteria (6, 38), and in preliminary studies we found that removal of acetaldehyde with an N₂ stream restored normal growth and allowed complete consumption of glucose by our best production strains (not shown). Thus, yields and rates might improve with continuous acetaldehyde removal (we did not measure acetaldehyde production during preliminary N₂ stripping studies), which should be achievable given its high volatility (boiling point = 20.2°C). In addition, the continuous removal and collection of acetaldehyde would obviate the high volumetric productivity and eliminate the high energy cost of distillation (6, 23, 24, 37, 38).

Importantly, the conversion of glucose to 2 acetaldehyde + 2 H₂ + 2 CO₂ by E. coli as described here has some intrinsic advantages compared to previously reported methods for renewable acetaldehyde production. Optimal bioconversion of ethanol to acetaldehyde (which occurred at a rate of about 0.6 g h⁻¹ g⁻¹ dry cell weight) was performed under conditions of 100% O₂ and 100 kPa pressure to ensure an adequate supply of O₂ for AOX (24). In addition, the overall bioconversion process currently requires two different organisms, one for ethanol production and a second for bioconversion of ethanol to acetaldehyde (23, 24). In studies of glucose conversion to acetaldehyde by bacteria (Z. mobilis and L. lactis), Pdc was used to convert pyruvate to acetaldehyde and CO₂, resulting in 1 excess NADH molecule produced per acetaldehyde molecule (6, 38). In a study using Z. mobilis, O₂ was used to maintain redox balance during acetaldehyde production via Pdc (38), but O₂ addition is expensive on an industrial scale and it can activate the tricarboxylic acid (TCA) cycle, resulting in reduced product yield due to loss of carbon as CO₂. In an L. lactis study, it was also possible to produce acetaldehyde via Pdc under anaerobic conditions, but the manner in which redox balance was maintained was not determined (6). This is potentially problematic, since the formation of reduced organic coproducts would substantially reduce carbon yield. In the approach described here (Fig. 1), pyruvate is converted to acetyl-CoA and formate. The reduction of acetyl-CoA to acetaldehyde allows regeneration of NAD⁺ to achieve redox balance, and formate is converted to H₂ + CO₂. Oxygen addition is unnecessary to achieve redox balance, and a valuable coproduct (H₂) is produced. To our knowledge, there are no prior reports of similar studies in which H₂ was coproduced with acetaldehyde.

In this study, we used an ALDH/ACR enzyme to convert an acyl-CoA compound to an aldehyde as a terminal product. To our knowledge, this is the first time an aldehyde has been produced in the manner. This approach may have applicability to the production of longer-chain aldehydes from their corresponding CoA derivatives that might be produced by fermentation or derived from fatty acid biosynthesis. This would require specific longer-chain acyl-CoA reductases that do not use acetyl-CoA as a substrate.

The coproduction of acetaldehyde and H₂ raises the possibility of “no-distill” ethanol fermentation that has the potential to reduce the energy costs and water usage at ethanol biorefineries. Acetaldehyde has a boiling point of 20.2°C and hence is readily evaporable under fermentation conditions (37°C) (14). Simultaneous acetaldehyde removal and recovery have been previously reported (19, 38). In our study, acetaldehyde and H₂ could be removed from the fermentation together by gas stripping and converted to ethanol by chemical catalysis that would bypass distillation. Currently, most fuel ethanol is purified by distillation, which requires large amounts of water and energy (40). Water used for cooling distillation towers usually accounts for over 50% of total water consumption, and the energy needed for heating distillation columns consumes about 75% of total process heat (18, 20, 41). However, potential savings would need to be weighed against the cost of chemical conversion of acetaldehyde and H₂ to ethanol.