Converting Carbon Dioxide to Butyrate with an Engineered Strain of Clostridium ljungdahlii

As previously reported (46, 47), wild-type cells grew with either H₂, CO, or fructose as the electron donor, with acetate as the primary end product (Fig. 1 and 2). Growth rates and yields were highest with fructose, intermediate with CO, and lowest with H₂.

In order to redirect carbon and electron flow to the production of butyrate, the following genes from Clostridium acetobutylicum were introduced into C. ljungdahlii: thl, which encodes thiolase for the synthesis of acetoacetyl-coenzyme A (CoA) from acetyl-CoA; hbd, which encodes 3-hydroxybutyryl-CoA dehydrogenase for the synthesis of 3-hydroxybutyryl-CoA from acetoacetyl-CoA; crt, which encodes crotonase for the synthesis of crotonyl-CoA from 3-hydroxybutyryl-CoA; bcd, which encodes butyryl-CoA dehydrogenase for the synthesis of butyryl-CoA from crotonyl-CoA; etfA and etfB, which encode electron transfer flavoproteins that form an enzyme complex with butyryl-CoA dehydrogenase for the synthesis of butyryl-CoA; ptb, which encodes phosphotransbutyrylase for the synthesis of butyryl phosphate from butyryl-CoA; and buk, which encodes butyrate kinase for synthesis of butyrate from butyryl phosphate (Fig. 1). The genes thl, crt, bcd, etfA, etfB, and hbd were on one plasmid, and the genes ptb and buk were on a second plasmid (see Fig. S1 and S2 in the supplemental material). The expression of all eight genes was under the control of the putative promoter of the pta gene from C. ljungdahlii, which is considered to be highly expressed under a diversity of conditions. This strain was designated B1.

Both acetate synthesis via the Pta/Ack pathway and butyrate synthesis via the Ptb/Buk pathway result in ATP production (Fig. 1). The conversion of acetyl-CoA to acetate is expected to yield 1 ATP via substrate-level phosphorylation, whereas only 0.5 ATP is expected to be derived from each acetyl-CoA converted to butyrate (Fig. 1). However, the conversion of crotonyl-CoA to butyryl-CoA catalyzed by the complex of Bcd and EtfAB is considered to be coupled with energy generation via the Rnf complex and ATP synthase (Fig. 1) to yield ca. 0.5 ATP (48, 49), but the stoichiometry of the proton gradient and ATP generation has not been clarified for C. ljungdahlii. Butyrate synthesis from acetyl-CoA requires additional reducing equivalents in the form of NADH for Hbd to reduce acetoacetyl-CoA to 3-hydroxybutyryl-CoA (50) and for Bcd to reduce crotonyl-CoA to butyryl-CoA (51).

When grown with H₂ as the electron donor, strain B1 produced butyrate as well as acetate and ethanol over time (Fig. 3 and 4). Acetate remained the primary product, with minor production of ethanol, but the butyrate produced (8.5 mM) accounted for 13% of the carbon and 16% of the electrons appearing in acetate. More butyrate was produced with CO as the electron donor than with H₂ (Fig. 3 and 4), and the proportion of carbon and electrons in butyrate compared to that in acetate, 25% and 29%, respectively, was also greater. Butyrate production was lowest during growth on fructose, with only slightly more butyrate than ethanol produced (Fig. 3 and 4). When grown with H₂ or CO as the electron donor, strain B1 had longer lag periods and lower growth rates than the wild-type strain, but the final cell yields were comparable to those of the wild-type strain with all three substrates.

Proteomic analysis demonstrated that with the exception of crotonase (Crt), all of the heterologous enzymes were expressed at levels comparable to those for highly expressed native proteins, such as phosphotransacetylase (Pta) and acetate kinase (Ack) (Fig. 5A). However, the transcript abundance of crt was comparable to that of other heterologous genes (Fig. 5B). These results suggested that the low abundance of the Crt protein resulted from inefficient translation of crt transcripts. The distance between the putative ribosome binding site (RBS) and translation initiation codon of crt was shorter than those for the other heterologous genes (Fig. 6A). Therefore, a crt sequence in which the length between the putative RBS and translation initiation codon was increased and substituted for the original crt sequence in constructing strain B2. Western blot analysis revealed that strain B2 produced substantially more Crt than strain B1 (Fig. 6B). Strain B2 produced more butyrate with all three electron donors (Fig. 4 and 7), suggesting that the low level of Crt was a bottleneck in butyrate production in strain B1.

Chromosomal integration of the genes for butyrate synthesis may be preferable to maintaining the genes on plasmids, to avoid the need to add antibiotics to retain plasmids and to reduce the cellular metabolic burden (52, 53). A fragment containing the genes ptb and buk was integrated at the adhE1 locus on the chromosome with double-crossover homologous recombination, but a longer fragment containing thl-crt-bcd-etfB-etfA-hbd could not be integrated in this manner (data not shown). However, the DNA fragment of thl-crt-bcd-etfB-etfA-hbd-ptb-buk (crt with the modified RBS), under the control of the putative pta promoter region, was successfully integrated into the putative pta promoter region via single-crossover homologous recombination (see Fig. S3 in the supplemental material).

Compared with strain B2, which had the same genes expressed on plasmids, strain B3 grew faster but with a lower final biomass yield when H₂ was the electron donor, (Fig. 8). A final butyrate yield by strain B3 was slightly lower than that by strain B2, but butyrate production compared to acetate production was improved (Fig. 4). In contrast, strain B3 grew faster than strain B2 with a similar final biomass yield on CO, and butyrate production was lower than that of strain B2. With fructose as the substrate, strain B3 grew faster than strain B2, but yields of biomass, acetate, and butyrate were comparable. In addition to acetate and butyrate, strain B3 produced ethanol at lower levels under all growth conditions (Fig. 8).

In an attempt to further enhance the diversion of carbon and electron flux from acetate production to butyrate synthesis, Pta, which is thought to catalyze the first step in the conversion of acetyl-CoA to acetate (Fig. 1), was disrupted with a single-crossover homologous recombination, which simultaneously integrated the butyrate pathway genes (see Fig. S4 in the supplemental material). The Cre-lox system, which allows reuse of an antibiotic resistance gene (54), was applied to C. ljungdahlii strain engineering (see Fig. S5, S6, and S7). This strain was designated B4.

Strain B4 grew slower than strain B3 with H₂ as the electron donor but with a comparable final cell yield (Fig. 9). Surprisingly, strain B4 produced acetate at amounts comparable to those for strain B3 (Fig. 4), despite the fact that the expected primary route for acetate synthesis was disrupted. Strain B4 produced slightly more butyrate and less acetate than strain B3 during growth with CO or fructose (Fig. 4 and 9), but acetate production remained substantial (Fig. 4 and 9), suggesting that there are one or more unknown pathways for acetate synthesis in C. ljungdahlii.

In an attempt to divert the NAD(P)H being consumed for ethanol production toward the production of butyrate, the ethanol pathway in strain B4 was inactivated by disrupting the adhE1 gene, which was shown (18) to encode the major bifunctional aldehyde/alcohol dehydrogenase for ethanol production, via single-crossover homologous recombination (Fig. 1; see also Fig. S8 in the supplemental material), creating strain B5. Strain B5 produced significantly less ethanol than strain B4, slightly improving butyrate production during growth on CO or fructose but not H₂ (Fig. 4 and 10).

It was considered that a CoA transferase might convert butyrate back to butyryl-CoA with simultaneous production of acetate or other fatty acids (Fig. 1). In an attempt to avoid this possibility, the gene Clju_c39430, which encodes a homolog of CoA transferases, was disrupted via single-crossover homologous recombination (see Fig. S8 in the supplemental material) in strain B4, yielding strain B6. Acetate continued to be produced in strain B6, but this strain was the best butyrate-producing strain in terms of butyrate yield under all three growth conditions tested (Fig. 4 and 11). Carbon and electron yields in butyrate were 42% and 48% (H₂), 68% and 73% (CO), and 71% and 75% (fructose), respectively.

The results demonstrate that it was possible through genetic manipulation to redirect carbon and electron flow to the production of butyrate from carbon dioxide in C. ljungdahlii. However, further genetic manipulation will be required before the goal of producing butyrate as the sole product of metabolism will be achieved.

The continued production of acetate when pta is disrupted was previously observed in C. acetobutylicum, which has a fermentative metabolism (55). A potential alternative route for acetate production in C. ljungdahlii, even during autotrophic growth, is conversion of acetyl-CoA to acetate via an acetaldehyde intermediate (15). This could yield energy to support cell growth and maintenance through the Rnf complex and ATP synthase. Conversion of acetyl-CoA to acetaldehyde and acetaldehyde to acetate is thought to proceed with oxidation of NADH to NAD⁺ and reduction of ferredoxin, respectively (Fig. 1) (15). Proton gradients generated with the reactions coupled with the reduction of NAD⁺ and the oxidation of ferredoxin via the Rnf complex result in ATP synthesis by ATP synthase (48, 49). Therefore, net ATP synthesis may be possible via this route of acetate production. ATP generated in this manner would yield ca. 0.5 ATP, which would be comparable to ATP generation via butyrate production (Fig. 1).

A previous study demonstrated that deleting adhE1, which encodes a bifunctional aldehyde/alcohol dehydrogenase, prevented ethanol formation from acetyl-CoA, presumably by eliminating acetaldehyde formation (18). However, strain B5, in which both the pta gene and the adhE1 gene were inactivated, still produced acetate. There are other homologs for aldehyde dehydrogenase in the C. ljungdahlii genome (Clju_c11960, Clju_c39730, and Clju_c39840). Therefore, deletion of one or more of these genes may be required in order to completely eliminate acetate production. Deletion of many genes in the same strain is currently not possible because there are only three antibiotic resistance markers that have been identified, severely limiting the number of genes that can be deleted if they are not colocalized on the chromosome. We adopted the Cre-lox system in order to permit simultaneous disruption of three genes, but this method leaves the recombination site on the chromosome, and thus they can be used only once in a strain. Negative- or counterselection markers, such as pyrF (56), galK (57), and mazF (58), which can be used repeatedly and have been applied to other Clostridium species, would be useful, but attempts to apply them to C. ljungdahlii have not yet been successful (unpublished data).

Another possibility for the unexpected acetate production is that the enzyme phosphotransbutyrylase (Ptb) introduced in the synthetic butyrate pathway may also act on acetyl-CoA, as previously described for the Ptb purified from C. acetobutylicum (59). The specificity for acetyl-CoA was only 1.6% of that for butyryl-CoA, but even low reactivity with acetyl-CoA, which is likely to be a more abundant metabolite, could be significant. Acetate production via this pathway could be favored because it would yield more ATP and have lower reducing equivalent demands than the desired production of butyrate. When the appropriate tools are available, deleting ack may eliminate this possibility because acetate kinase is essential for this potential pathway. Substituting bukII, which encodes another butyrate kinase that is more specific for butyrate synthesis (37), for buk in the ptb-buk operon might also improve butyrate yields.

These and previous (16, 18) findings that carbon and electron flow associated with the Wood-Ljungdahl pathway in C. ljungdahlii can be genetically redirected suggest that C. ljungdahlii may be an effective chassis for the production of organic commodities from carbon monoxide-containing waste gases (11, 15) or from carbon dioxide via electrosynthesis (9). Further development of the genetic tools needed to accomplish this goal is under way.

C. ljungdahlii DSM 13528 (ATCC 55383) from the German Collection of Microorganisms and Cell Cultures was the parent strain for butyrate production engineering. When fructose served as the carbon and energy source, cultures were grown in PETC 1754 medium (100% CO₂ in the headspace) with 28 mM fructose as described previously (18). For growth with CO as the electron donor, the fructose was omitted from the PETC 1754 medium and CO was added to the headspace at 20 lb/in². When H₂ served as the electron donor, the cells were grown in DSMZ medium 879 with the fructose omitted and with a headspace of H₂-CO₂ (80%/20%). H₂-CO₂ (80%/20%) was further added to the headspace at 20 lb/in². DSMZ medium 879 was used to maintain the CO₂ partial pressure since the original contains N₂-CO₂ (80%/20%) in the headspace. Cultures were grown in anaerobic pressure tubes (27 ml) containing either 10 ml (fructose grown) or 5 ml (CO or H₂ grown) of medium. Cultures grown on CO or H₂ were shaken at 100 rpm.

Escherichia coli NEB 5-alpha and NEB Express (New England Biolabs) were used for plasmid preparation and grown as instructed.

C. acetobutylicum genes necessary for synthesis of butyrate from acetyl-CoA (Fig. 1) were amplified by PCR. The amplified genes were thl (NCBI GenBank; CA_C2873), crt (NCBI GenBank; CA_C2712), bcd (NCBI GenBank; CA_C2711), etfB (NCBI GenBank, CA_C2710), etfA (NCBI GenBank; CA_C2709), hbd (NCBI GenBank; CA_C2708), buk (NCBI GenBank; CA_C3075), and ptb (NCBI GenBank; CA_C3076). Primers used for the PCR are listed in Table S1 in the supplemental material. PCR products were cloned in pCR-Blunt II-TOPO (Life Technologies). The NdeI-PstI DNA fragment containing etfA was cloned at the NdeI (in etfB) and PstI sites in pCR-Blunt II-TOPO containing bcd and etfB. The KpnI-XhoI DNA fragment containing thl was cloned at the KpnI and XhoI sites in pBluescript II KS(−) (Stratagene). Then, the XhoI-EcoRI DNA fragment containing crt was cloned at the XhoI and EcoRI sites in pBluescript II KS(−) containing thl. The EcoRI-PstI DNA fragment containing bcd, etfB, and etfA was cloned at the EcoRI and PstI sites in pBluescript II KS(−) containing thl and crt. The PstI-BamHI DNA fragment containing hbd was cloned at the PstI and BamHI sites in pBluescript II KS(−) containing thl, crt, bcd, etfB, and etfA (see Fig. S1A). The KpnI-BamHI DNA fragment containing thl, crt, bcd, etfB, etfA, and hbd was cloned at the KpnI and BamHI sites in an expression vector, pJe-p (see Fig. S2). The BamHI-SalI DNA fragment containing ptb and buk was cloned at the BamHI and SalI sites in pM6-p (see Fig. S2). pJe-p and pM6-p were constructed by placing the putative promoter for the pta gene from C. ljungdahlii in pJe (16) and pMTL82151 (a gift from N. P. Minton) (60), respectively. pJe is a derivative of pCL2 (18) and contains an erythromycin resistance gene instead of the chloramphenicol resistance gene. The putative pta promoter was prepared by PCR using primers listed in Table S1. PCR products were cloned in pCR-Blunt II-TOPO. The EcoRI-KpnI DNA fragment containing the putative pta promoter was cloned at the EcoRI-KpnI sites in these vectors. Transformation of these plasmids was conducted as described previously (18).

The BamHI DNA fragment from pCR-Blunt II-TOPO containing ptb and buk as described above (a BamHI site is located downstream of buk in the vector) was cloned at the BamHI site of pBluescript II KS(−) containing thl, crt, bcd, etfB, etfA, and hbd (see Fig. S1B in the supplemental material). The crt gene with the modified RBS was used. The SalI site remains at the 3′ end of buk upstream of the BamHI site. A resultant plasmid containing ptb and buk in the same direction with thl, crt, bcd, etfB, etfA, and hbd was used for subsequent plasmid preparation. The ermC gene was amplified by PCR with primers listed in Table S1 and pCL1 (18). PCR products were cloned in pCR-Blunt II-TOPO. The SpeI-NotI fragment containing ermC was cloned at the SpeI and NotI sites in pBluescript II KS(−) containing thl, crt, bcd, etfB, etfA, hbd, ptb, and buk (see Fig. S2). The resultant plasmid was used to integrate the butyrate pathway genes at the pta promoter region on the chromosome (see Fig. S2).

A loxP cassette was constructed as follows. Oligonucleotides loxP1-T and lox-P1-B (see Table S1 in the supplemental material) were phosphorylated, annealed, and cloned at the EcoRI site in the plasmid pNEB193 (New England Biolabs). A resultant plasmid with the loxP1 sequence in the proper orientation was used for cloning the phosphorylated and annealed oligonucleotides loxP2-T and loxP2-B at the HindIII site. A resultant plasmid with the loxP2 sequence in the proper orientation was used for the subsequent cloning. The catP gene was amplified by PCR with primers listed in Table S1 and pMTL82151 (60). PCR products were cloned in pCR-Blunt II-TOPO. The BglII-SacII DNA fragment of the catP gene was cloned at the BglII and SacII sites in pNEB containing loxP1 and loxP2. Then, the ermC gene was amplified by PCR with primers listed in Table S1 and pCL1 (18). PCR products were cloned in pCR-Blunt II-TOPO. The SphI-HindIII DNA fragment of ermC was cloned at the SphI and HindIII sites in pNEB containing loxP1, loxP2, and catP (see Fig. S4).

A cre gene with codon usage optimized for Clostridium was synthesized by GenScript (see Fig. S5 in the supplemental material). The KpnI-BamHI DNA fragment of cre was cloned at the KpnI and BamHI sites in an expression vector, pJKe-p. For the construction of pJKe-p, a kanamycin resistance gene (kan) with codon usage optimized for Clostridium was synthesized by GenScript (see Fig. S6). The kan gene was fused with the putative promoter region of ermB (PermB) from pQexp (61). PermB-kan was amplified by PCR with primers listed in Table S1. PCR products were cloned in pCR-Blunt II-TOPO. The ClaI-HpaI DNA fragment of PermB-kan replaced the ClaI-HpaI DNA fragment including catP in pJIR750ai (Sigma). pJKe-p was constructed by placing the pta promoter at the EcoRI-KpnI sites in the resultant plasmid, pJKe, as described above.

The KpnI-SalI DNA fragment containing thl, crt, bcd, etfB, etfA, hbd, ptb and buk from the pBluescript II KS(−) containing thl, crt, bcd, etfB, etfA, hbd, ptb, and buk as described above was cloned at the KpnI-SalI sites in pNEB containing loxP1, loxP2, catP, and ermC as described above. The internal region of the pta-coding sequence (pta-int) was amplified by PCR with primers listed in Table S1 in the supplemental material. PCR products were cloned in pCR-Blunt II-TOPO. The SpeI-KpnI DNA fragment of pta-int was cloned at the SpeI and KpnI sites in the pNEB containing loxP1, loxP2, catP, ermC, and the butyrate pathway genes (see Fig. S4). The resultant plasmid was used to integrate the butyrate pathway genes at the internal region of the pta-coding sequence on the chromosome, which resulted in disruption of pta (see Fig. S4).

After integration of the above plasmid on the chromosome DNA, the vector region and catP were excised by Cre (see Fig. S4 and S7 in the supplemental material). Competent cells of this strain were prepared as described previously (18) and transformed with the plasmid containing cre with the following modification. After electroporation and subsequent outgrowth in the presence of erythromycin, kanamycin was added to the outgrowth culture at the final concentration of 20 µg/ml and further incubated overnight to facilitate recombination at the loxP sites by Cre. The outgrowth cultures were then plated normally in the presence of erythromycin and kanamycin. Absence of the vector region and catP and presence of the butyrate genes were verified by PCR analysis (see Fig. S7).

The adhE1 gene was disrupted by single-crossover homologous recombination at the internal region of the adhE1-coding sequence (see Fig. S8 in the supplemental material). The internal region of the adhE1-coding sequence (adhE1-int) was amplified by PCR using primers listed in Table S1. PCR products were cloned in pCR-Blunt II-TOPO. The SpeI-KpnI DNA fragment of adhE1-int was cloned at the SpeI and KpnI sites in the pNEB containing catP. The resultant plasmid was used to disrupt adhE1 on the chromosome of strain B4, in which pta was inactivated by inserting the butyrate pathway genes at the internal region of the pta-coding sequence on the chromosome.

The ctf gene (Clju_c39430) was disrupted as described for adhE1 (see Fig. S8 in the supplemental material). The internal region of the ctf-coding sequence (ctf-int) was amplified by PCR with primers listed in Table S1. PCR products were cloned in pCR-Blunt II-TOPO. The SpeI-KpnI DNA fragment of ctf-int was cloned at the SpeI and KpnI sites in pNEB containing catP. The resultant plasmid was used to disrupt ctf on the chromosome of strain B4.

Protein expression profiles of C. ljungdahlii wild-type and B1 strains grown on fructose were analyzed at the Proteomics and Mass Spectrometry Facility, University of Massachusetts Medical School. Cells were harvested from cultures at a log phase. Each protein sample (5 µg) was run in to a gel. The gel slice was cut out, washed thoroughly, and digested with trypsin. Each digested sample was analyzed in triplicate by using nano-liquid chromatography–tandem mass spectrometry (nano-LC-MS/MS).

Total RNA was prepared from the fructose-grown B1 strain by using the RiboPure-Bacteria kit (Life Technologies) as instructed. cDNA was prepared with random nonamers by using the Enhanced Avian RT First Strand synthesis kit (Sigma) as instructed. PCR was conducted with primers listed in Table S1 in the supplemental material as described previously (62).

Cell extracts were prepared from the fructose-grown wild-type, B1, and B2 strains by disruption with sonication. Cell extracts were loaded on an SDS-PAGE gel. Western blot analysis was carried out with an antibody against the peptide DISEMKEMNTIERG, from amino acid residues 67 to 80 of the Crt protein. The peptide and the antibody were prepared by GenScript.

Growth of cells was monitored by measuring the optical density at 600 nm (OD₆₀₀). Acetate and butyrate were measured by high-performance liquid chromatography (9), and ethanol was measured by gas chromatography (63).