|Home | About | Journals | Submit | Contact Us | Français|
A pathway toward isobutanol production previously constructed in Escherichia coli involves 2-ketoacid decarboxylase (Kdc) from Lactococcus lactis that decarboxylates 2-ketoisovalerate (KIV) to isobutyraldehyde. Here, we showed that a strain lacking Kdc is still capable of producing isobutanol. We found that acetolactate synthase from Bacillus subtilis (AlsS), which originally catalyzes the condensation of two molecules of pyruvate to form 2-acetolactate, is able to catalyze the decarboxylation of KIV like Kdc both in vivo and in vitro. Mutational studies revealed that the replacement of Q487 with amino acids with small side chains (Ala, Ser, and Gly) diminished only the decarboxylase activity but maintained the synthase activity.
We have previously shown that 2-keto acids generated from amino acid biosynthesis can serve as precursors for the Ehrlich degradation pathway (15) to higher alcohols (3). In order to produce isobutanol, the valine biosynthesis pathway was used to generate 2-ketoisovalerate (KIV), the precursor to valine, which was then converted to isobutanol via a decarboxylation and reduction step (Fig. (Fig.1A).1A). The entire pathway to isobutanol from glucose is shown in Fig. Fig.1A.1A. To produce isobutanol, we overexpressed five genes, alsS (Bacillus subtilis), ilvC (Escherichia coli), ilvD (E. coli), kdc (Lactococcus lactis), and ADH2 (Saccharomyces cerevisiae) (Fig. (Fig.1A).1A). This E. coli strain produced 6.8 g/liter isobutanol in 24 h (Fig. (Fig.1B)1B) and more than 20 g/liter in 112 h (3). More recently, we have found that an alcohol dehydrogenase (Adh) encoded by yqhD on the E. coli genome can convert isobutyraldehyde to isobutanol efficiently (5) (Fig. (Fig.1B1B).
One key reaction in the production of isobutanol is the conversion of KIV to isobutyraldehyde catalyzed by 2-ketoacid decarboxylase (Kdc) (Fig. (Fig.1C).1C). Since E. coli does not have Kdc, kdc from L. lactis was overexpressed. Kdc is a nonoxidative thiamine PPi (TPP)-dependent enzyme and is relatively rare in bacteria, being more frequently found in plants, yeasts, and fungi (8, 19). Several enzymes with Kdc activity have been found, including pyruvate decarboxylase, phenylpyruvate decarboxylase (18), branched-chain Kdc (8, 19), 2-ketoglutarate decarboxylase (10, 17, 20), and indole-3-pyruvate decarboxylase (13).
In this work, unexpectedly, we find that Kdc is nonessential for E. coli to produce isobutanol (Fig. (Fig.1).1). An E. coli strain overexpressing only alsS (from B. subtilis), ilvC, and ilvD (both from E. coli) is still able to produce isobutanol. Since E. coli is not a natural producer of isobutanol, it cannot be detected from the culture media in any unmodified strain. We identify that AlsS from B. subtilis, which was introduced in E. coli for acetolactate synthesis (Als), catalyzes the decarboxylation of 2-ketoisovalerate like Kdc both in vivo and in vitro. AlsS is part of the acetoin synthesis pathway and catalyzes the aldo condensation of two molecules of pyruvate to 2-acetolactate (Als activity) (Fig. (Fig.1C)1C) (11). The overall reaction catalyzed by AlsS is irreversible because of CO2 evolution. The first step in catalysis is the ionized thiazolium ring of TPP reacting with the first pyruvate, followed by decarboxylation. This intermediate then reacts with the second pyruvate. Deprotonation followed by C-C bond breakage produces 2-acetolactate. In this work, mutational approaches were used to assess the importance of Q487 in the Kdc activity of AlsS.
Restriction enzymes and Antarctic phosphatase were from New England Biolabs (Ipswich, MA). The rapid DNA ligation kit was from Roche (Mannheim, Germany). KOD DNA polymerase was from EMD Chemicals (San Diego, CA). Oligonucleotides were from Operon (Huntsville, AL).
A list of many of the strains, plasmids, and oligonucleotides used is given in Table Table1.1. JCL16 (2) is BW25113 (rrnBT14 ΔlacZWJ16 hsdR514 ΔaraBADAH33 ΔrhaBADLD78) (7) with F′ transduced from XL-1 Blue to supply lacIq. JCL260 is JCL16 ΔadhE Δfnr-ldhA ΔfrdBC ΔpflB Δpta. The ilvC gene was inactivated by P1 transduction with JW3747 (6).
To clone alsS, pSA69 (3) was digested with AatII and SalI. A shorter fragment was purified and cloned into plasmid pCS27 (16) cut with the same enzymes, creating pZL8. Both alsS single-site mutations (Q487N and Q487A) were introduced using PCR-directed mutagenesis. To introduce the mutation into alsS, pSA69 was used as a PCR template with A306 and A124 (Q487N) and A307 and A124 (Q487A). The beginning of the alsS gene located on pSA69 was also amplified from the AatII site upstream of the ribosome binding site to the 1,458th base in the alsS gene, using primers A300 and A305. These two fragments were then joined by splice overlap extension. The products were digested with AatII and SalI and cloned into pSA69 cut with the same enzyme, creating pSA163 and pSA164.
For protein overexpression and purification, the wild type and alsS variants were amplified with primers A297 and A298. PCR products were digested with BamHI and SalI and cloned into pETDuet-1 (Novagen, Madison, WI) cut with the same enzymes (Table (Table22).
M9 medium (64 g Na2HPO4·7H2O, 15 g KH2PO4, 2.5 g NaCl, 5 g NH4Cl, 2 mM MgSO4, 0.1 mM CaCl2, 10 mg thiamine per liter water) containing 36 g/liter glucose, 5 g/liter yeast extract, 30 μg/ml kanamycin, and 1 ml/liter of trace metal mix A5 [2.86 g H3BO3, 1.81 g MnCl2·4H2O, 0.222 g ZnSO4·7H2O, 0.39 g Na2MoO4·2H2O, 0.079 g CuSO4·5H2O, 49.4 mg Co(NO3)2·6H2O per liter water] was used for cell growth. Preculture in test tubes containing 3 ml of medium was performed at 37°C overnight on a rotary shaker (250 rpm). Overnight culture was diluted 1:100 into 20 ml of fresh medium in a 250-ml screw cap conical flask. Cells were grown at 37°C for 3 h, followed by adding 0.1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). Production was performed under microaerobic conditions at 30°C on a rotary shaker (250 rpm) for 24 h. Isobutanol was quantified by a gas chromatography-flame ionization detector as previously described (3). Secreted pyruvate was quantified by a high-performance liquid chromatography as previously described (2).
The wild type and AlsS variants were synthesized from a His-tag plasmid in E. coli strain BL21 Star (DE3) (Invitrogen, Carlsbad, CA) followed by purification with Ni-nitrilotriacetic acid (NTA) spin columns (Qiagen, Valencia, CA). Protein concentrations were determined by the Bradford assay (Bio-Rad, Hercules, CA).
An enzyme assay for Als activity of AlsS was carried out in 1 ml of morpholinepropanesulfonic acid (MOPS) buffer (pH 7.0) containing 80 nM AlsS, 100 mM MOPS (pH 7.0), 1 mM MgCl2, 0.1 mM TPP, 10 mM acetate, and various concentrations of pyruvate at 37°C for 10 min. The reaction was terminated by acidification of the solution with 0.1 ml of 50% H2SO4. The mixture was incubated for an additional 25 min at 37°C to allow for the acid hydrolysis of the acetolactate to acetoin. Acetoin formation was measured as described previously (11). One unit of enzyme activity was defined as the amount of enzyme that converts 1 μmol of substrate into product in 1 minute under these conditions. The Km values for pyruvate and the Vmax were extrapolated after nonlinear regression of the experimental points with the Gauss-Newton method using Matlab.
The enzyme assay for Kdc activity of AlsS was carried out in reaction mixtures containing 80 nM AlsS, 100 mM MOPS (pH 7.0), 1 mM MgCl2, 0.1 mM TPP, 10 mM acetate, and various concentrations of KIV at 37°C for 1 h. The production of isobutyraldehyde was confirmed to be linear over 1 h. Isobutyraldehyde was measured by a gas chromatography-flame ionization detector as previously described (3). One unit of enzyme activity was defined as the amount of enzyme that converts 1 μmol of substrate into product in 1 minute under these conditions. The Km values for KIV and the Vmax were extrapolated after nonlinear regression of the experimental points with the Gauss-Newton method using Matlab.
Since E. coli does not have any Kdc, we first hypothesized that pyruvate dehydrogenase (PDH) or 2-ketoglutarate dehydrogenase (KGDH) of E. coli could catalyze the conversion of KIV to isobutyryl coenzyme A, which is followed by the conversion of isobutyryl coenzyme A to isobutyraldehyde and then isobutanol by aldehyde and alcohol dehydrogenases. To test these possibilities, we deleted aceE and sucA, which encode subunits of the PDH and KGDH complexes, respectively. However, this double knockout strain with overexpression of alsS, ilvC, and ilvD was still capable of producing isobutanol (data not shown), indicating that neither PDH nor KGDH catalyzes the reaction in the Kdc-independent isobutanol production.
To determine essential components for the Kdc-independent isobutanol production, we measured isobutanol production from the strain overexpressing different combinations of alsS, ilvC, and ilvD (Fig. (Fig.2).2). The strain overexpressing alsS alone produced isobutanol, but the strain overexpressing ilvC and ilvD did not (Fig. (Fig.2A).2A). When ilvC and ilvD were overexpressed with alsS, isobutanol production increased nearly ninefold (Fig. (Fig.2A).2A). Because the only known activity of AlsS is acetolactate synthase, it is unclear how the strain could produce isobutanol only with alsS overexpression. As a control experiment, ilvI and ilvH (E. coli), which encodes an acetohydroxy acid synthase (Ahas) instead of AlsS (B. subtilis), were overexpressed. The strain overexpressing ilvI and ilvH (E. coli) did not produce isobutanol. Increasing AlsS levels in E. coli led to a parallel increase in the formation of acetoin, which is the product of spontaneous decarboxylation of 2-acetolactate (1). To test whether some enzymes in E. coli could utilize acetoin as a substrate for isobutanol production, acetoin was fed to the E. coli culture. Neither isobutyraldehyde nor isobutanol was detected from this culture, indicating that acetoin was not a precursor of isobutanol in this pathway (data not shown). To confirm that the Kdc-independent pathway used the same route as the Kdc-dependent pathway, the ilvC gene on the genome was deleted (Fig. (Fig.2A).2A). The deletion of ilvC abolished isobutanol production, indicating that this Kdc-independent pathway utilized KIV as a precursor (Fig. (Fig.2A2A).
We then supplied KIV to the medium to assess the capability to utilize KIV for isobutanol production (Fig. (Fig.2B).2B). The strain without AlsS did not produce isobutanol in the presence of 6 g/liter of KIV (Fig. (Fig.2B),2B), but addition of KIV to strains overexpressing alsS led to isobutanol production in the wild-type and ΔilvC backgrounds (Fig. (Fig.2B).2B). These tests revealed that Kdc-independent isobutanol production requires overexpression of alsS and a high concentration of KIV. Feeding of KIV to the strain overexpressing alsS, ilvC, and ilvD did not change the production of isobutanol (Fig. (Fig.2C),2C), presumably because the concentration of KIV may already saturate the enzyme which utilizes KIV for isobutanol production or the efficiency of KIV uptake may decrease (Fig. (Fig.2C2C).
After exhausting all other possibilities, we hypothesized that AlsS could catalyze the decarboxylation of KIV and give isobutyraldehyde without the nucleophilic attack of the second pyruvate in the presence of a high concentration of KIV and a low concentration of pyruvate. Because the overexpression of alsS, ilvC, and ilvD significantly decreases the secretion of pyruvate to below a detection limit of 0.1 mM (the host strain without these plasmids secretes ~7 mM pyruvate), it is possible that under this condition, KIV reacts with TPP and undergoes decarboxylation, and then escapes by giving isobutyraldehyde before undergoing the carboligation.
Using the crystal structure of Klebsiella pneumoniae AlsS, Pang et al. (14) showed that the extended side chain of Gln483 causes steric hindrance with the larger substrate, 2-ketobutyrate, which explains why AlsS reacts very poorly with a larger 2-keto acid as the second substrate (9, 14). Gln483 in K. pneumoniae AlsS has been shown to be in close proximity to the first pyruvate and also involved in second-substrate specificity (14), suggesting that Gln483 may play a role in the release of aldehydes. The residue corresponding to Gln483 in K. pneumoniae AlsS is Gln487 in B. subtilis AlsS. To test whether Gln487 could play a role in decarboxylation, we replaced Gln487 with various other amino acids.
To assay the Kdc activity (Fig. (Fig.1C,1C, bottom) of AlsS, His-tagged wild-type AlsS was expressed from a His-tag plasmid and purified as described in Materials and Methods. The Kdc activity of the His-tagged wild-type AlsS was 5.5 μmol·min−1·mg−1, while isobutyraldehyde production was not detected from a negative control experiment without AlsS. Although the activity was weak, AlsS surely showed the decarboxylase activity toward KIV in vitro.
The kinetic parameters were measured for AlsS variants (Table (Table2).2). The kcat/Km values for pyruvate of Q487 variants with small residues (Q487A, Q487G, and Q487S) were similar to that of the wild type, while the kcat/Km values for KIV of these variants decreased dramatically (Table (Table2).2). Q487L and Q487I replacements impaired Als activity (Table (Table2).2). However, the kcat/Km values for KIV of these variants were similar to that of the wild type. The wild type and all variants showed extremely high Km values for KIV (Table (Table2),2), which may explain why an increase of the flux toward KIV is required for the decarboxylase activity of AlsS. Further analysis is required to explain the relationship between Q487 and decarboxylase activity.
To test these AlsS variants' capability to produce isobutanol, which requires both Als (Fig. (Fig.1C,1C, top) and Kdc (Fig. (Fig.1C,1C, bottom) activities, these alsS variants were overexpressed with ilvC and ilvD. Isobutanol production with AlsS (Q487N) was similar to the production achieved using wild-type AlsS (Fig. (Fig.3A),3A), presumably because the side chain of Asn has an amine group, like Gln. However, the replacement of Q487 with valine, alanine, glycine, serine, leucine, and isoleucine nearly abolished isobutanol production (Fig. (Fig.3A).3A). According to the results of the performed enzyme assays (Table (Table2),2), the replacement of Q487 with glycine and serine maintained Als activity and decreased Kdc activity. The ratios of Km for KIV to Km for pyruvate of Q487G and Q487S were 109 ± 30 and 140 ± 79, respectively, while the ratio for the wild-type enzyme was 20 ± 2.9. The strains with either Q487G or Q487S cannot produce isobutanol, presumably because the Kdc activity of Q487G and Q487S could not compete with the Als activity. Enzyme assays showed that Q487L and Q487I replacements impaired Als activity. We showed that increased flux toward KIV was important for isobutanol production when using AlsS for Kdc activity (Fig. 2A and B), but the Km values for KIV of Q487L and Q487I were 342 ± 45 mM and 323 ± 26 mM, respectively (Table (Table2),2), which were extremely high. Because these Km values are extremely high, the strains with these replacements could not produce isobutanol, presumably because the intracellular concentration of KIV would not be high enough for the Kdc activity. No replacements were found that could increase isobutanol production.
To distinguish between Als and Kdc activities in these variants, we tested the growth rate of an E. coli strain KS145 (ΔilvI ΔilvB) expressing various Q487 variants in a minimum glucose medium supplemented with l-isoleucine. AlsS is a distant homologue of Ahas, which is responsible for both Als (Fig. (Fig.1C,1C, top) and 2-aceto-2-hydroxy butyrate synthase (Ahbs) (Fig. (Fig.1C,1C, middle) activities in the branched chain amino acid biosynthesis. The KS145 strain does not have Als and Ahbs activities (Fig. (Fig.1C);1C); thus, the specific growth rate in the minimal medium with l-isoleucine reflects the Als activity. Figure Figure3B3B shows that all of the Q487 variants retain significant Als activity. Considering that most of the Q487 variants did not generate isobutanol (Fig. (Fig.3A),3A), we conclude that AlsS is indeed responsible for the Kdc activity observed in isobutanol synthesis, and that Q487 is important for this activity. KS145 cells expressing Q487L or Q487I showed slow growth with the l-isoleucine supplement (Fig. (Fig.3B),3B), indicating that these replacements would reduce Als activity. These results were consistent with the results of enzyme assays (Table (Table2).2). In the structure model of K. pneumoniae AlsS, the C-1 carbonyl oxygen of the modeled second pyruvate is hydrogen-bonded to the side chain of Gln483 (14). Thus, the nonpolar side chain of isoleucine and leucine in the 487th residue would reduce the binding affinity of the second pyruvate to the site. No growth phenotype was observed in any strain while grown on minimum glucose medium supplemented with l-valine, l-leucine, and l-isoleucine (data not shown).
AlsS reacts very poorly with the larger substrate, 2-ketobutyrate (9). If these replacements change the second substrate specificity by removing steric hindrance, the AlsS variants would gain the Ahbs activity so that the KS145 cells expressing the variants could grow in minimal medium with l-valine and l-leucine supplementation. As predicted, KS145 cells expressing the wild-type AlsS were unable to grow without l-isoleucine (Fig. (Fig.3C),3C), indicating that the wild-type AlsS was not capable of catalyzing Ahbs reaction, which is consistent with previous studies (12). Interestingly, the AlsS variants which contain small residues (Ala, Gly, and Ser) at the 487th residue rescued the growth of KS145 under the same conditions (Fig. (Fig.3C).3C). This result suggests that the replacement of Q487 with small side chain amino acids would make the substrate binding site larger so that the variants are able to react with 2-ketobutyrate as the second substrate (Ahbs activity) (Fig. (Fig.1C,1C, middle).
Our evidence shows that Kdc is not essential for isobutanol production and that AlsS, previously known only to have Als activity, can catalyze the decarboxylation of KIV. The use of mutational studies allowed us to identify that Q487 is important for Kdc activity. For further analysis, mutational studies can be expanded to other residues in the catalytic center of AlsS. These studies will give us implications to distinguish important residues for Als and Kdc activities. Protein structural analysis, such as X-ray crystallography, can also be applied to increase our understanding of the catalytic mechanism of Kdc activity of AlsS.
This work was partially supported by the UCLA-DOE Institute for Genomics and Proteomics.
We are grateful to anonymous reviewers for helpful comments on the manuscript, and to Steven Clarke, Kendall Houk, and members of the Liao laboratory for helpful discussions.
Published ahead of print on 14 August 2009.