Branched-chain 2-keto acid decarboxylases derived from Psychrobacter

Authors

  • Jiashi Wei,

    1. Department of Bioproducts and Biosystems Engineering, University of Minnesota, St. Paul, MN, USA
    2. Biotechnology Institute, University of Minnesota, St. Paul, MN, USA
    Search for more papers by this author
  • Jacobe G. Timler,

    1. Department of Bioproducts and Biosystems Engineering, University of Minnesota, St. Paul, MN, USA
    Search for more papers by this author
  • Carolann M. Knutson,

    1. Department of Bioproducts and Biosystems Engineering, University of Minnesota, St. Paul, MN, USA
    Search for more papers by this author
  • Brett M. Barney

    Corresponding author
    1. Biotechnology Institute, University of Minnesota, St. Paul, MN, USA
    • Department of Bioproducts and Biosystems Engineering, University of Minnesota, St. Paul, MN, USA
    Search for more papers by this author

Correspondence: Brett M. Barney, Department of Bioproducts and Biosystems Engineering, University of Minnesota, 1390 Eckles Avenue, St. Paul, MN 55108-6130, USA. Tel.: 612 626 8751; fax: 612 625 6286; e-mail: bbarney@umn.edu

Abstract

The conversion of branched-chain amino acids to branched-chain acids or alcohols is an important aspect of flavor in the food industry and is dependent on the Ehrlich pathway found in certain lactic acid bacteria. A key enzyme in the pathway, the 2-keto acid decarboxylase (KDC), is also of interest in biotechnology applications to produce small branched-chain alcohols that might serve as improved biofuels or other commodity feedstocks. This enzyme has been extensively studied in the model bacterium Lactococcus lactis, but is also found in other bacteria and higher organisms. In this report, distinct homologs of the L. lactis KDC originally annotated as pyruvate decarboxylases from Psychrobacter cryohalolentis K5 and P. arcticus 273-4 were cloned and characterized, confirming a related activity toward specific branched-chain 2-keto acids derived from branched-chain amino acids. Further, KDC activity was confirmed in intact cells and cell-free extracts of P. cryohalolentis K5 grown on both rich and defined media, indicating that the Ehrlich pathway may also be utilized in some psychrotrophs and psychrophiles. A comparison of the similarities and differences in the P. cryohalolentis K5 and P. arcticus 273-4 KDC activities to other bacterial KDCs is presented.

Introduction

The conversion of 2-keto acids (α-keto acids) to aldehydes and carbon dioxide is important in the food industry and more recently has become of interest in the production of branched-chain alcohols for biofuel and commodity feedstock production (De la Plaza et al., 2004; Smit et al., 2005; Atsumi et al., 2008). In lactic acid bacteria, branched-chain amino acids (leucine, isoleucine, and valine) are converted to the corresponding 2-keto acids by aminotransferases and are subsequently decarboxylated by 2-keto acid decarboxylases (KDCs), yielding a branched-chain aldehyde (Fig. 1). Further metabolism of the aldehyde results in the production of either acids or alcohols from various cellular dehydrogenases, all of which are important in flavor development (Ardo, 2006). Two KDCs have been identified from Lactococcus lactis with a high level of conservation (almost 90 percent amino acid sequence identity) between them and were named α-ketoisovalerate decarboxylase (Kivd, gene accession number AJ746364) or keto acid decarboxylase (KdcA, gene accession number AY548760) (De la Plaza et al., 2004; Smit et al., 2005). The substrate ranges and kinetic characteristics of these enzymes have been reported and show some contrast between them (De la Plaza et al., 2004; Smit et al., 2005; Yep et al., 2006), and a structure has been published for KdcA (Berthold et al., 2007).

Figure 1.

Enzyme reactions converting branched-chain amino acids to the intermediate 2-keto acids and then to medium branched-chain alcohols. The above pathway is shown starting from the substrate leucine, but would also occur with valine and isoleucine.

Psychrobacter is a genus of psychrophilic and psychrotrophic bacteria often found in extremely cold habitats, such as Antarctic ice, soils, and sediments, and have been found in the deep sea. Psychrobacter cryohalolentis K5 was initially isolated from samples of Siberian permafrost from a frozen salt lens, grows between −10 and 30 °C with an optimum growth temperature of 22 °C and is capable of growing on citrate, lactate or acetate as simple carbon compounds (Bakermans et al., 2006). Psychrophiles and psychrotrophs are of interest, as their enzymes may be beneficial for various biotechnological and industrial applications where function at lower temperature is important (Drienovska et al., 2012; Novototskaya-Vlasova et al., 2012). Some Psychrobacter are also reported to accumulate various high-value lipids such as wax esters when grown on simple carbon compounds (Barney et al., 2012).

A primary amino acid blast search (Altschul et al., 1990) of the P. cryohalolentis K5 genome revealed a protein (accession number YP_580229) with modest similarity (c. 40 percent identity and 60 percent similarity) to both KdcA and Kivd from L. lactis. A similar protein was found in P. arctius 273-4 (accession number YP_264145). Further analysis using the structure of KdcA (Berthold et al., 2007) with a substrate analog shows that the conservation of residues around the active site and thiamine pyrophosphate (TPP) cofactor is much higher, indicating that these proteins, annotated in the published genomes (NCBI reference sequence NC_007969.1 and NC_007204.1) as pyruvate decarboxylases, may be branched-chain KDCs. In this work, we aimed to clone, purify, and characterize these putative KDC enzymes from these Psychrobacter strains and determine their activity toward several branched-chain 2-keto acids and pyruvate. We also evaluated features of a coupled spectrophotometric assay and the utilization of a medium branched-chain alcohol dehydrogenase (MADH) cloned from Marinobacter aquaeolei VT8 to substitute for conventional alcohol dehydrogenase (ADH) in these assays.

Materials and methods

Reagents and media

Bacteria were obtained from the American Type Cultures Collection (ATCC) or James Tiedje. Chemicals were obtained from Fisher Scientific (Pittsburgh, PA) and Sigma-Aldrich Chemical Company (St. Louis, MO). Psychrobacter cryohalolentis K5 was grown on the following minimal media: 5 grams NaCl, 10 grams sodium acetate, 50 mg CaCl2·2H2O, 200 mg MgSO4·7H2O, 500 mg K2HPO4, 10 mg FeSO4·7H2O, 1.5 grams Na2SO4, and 270 mg (NH4)2SO4, brought to 1.0 L of distilled water and adjusted to pH 7.2. Psychrobacter and Escherichia coli strains were also grown on Miller's lysogeny broth (LB).

Cloning of genes

A gene for the putative KDC (gene product accession number YP_580229, P. cryohalolentis K5 KDC) from P. cryohalolentis K5 was cloned from genomic DNA using the primers BBP471 (5′GACTAAGCTT CCATGGAAGT CAACAATATA CCATCGCTGA TTATTTGTTT G3′ with HindIII and NcoI sites) and BBP472 (5′CTGGTCTAGA TTAATCATTC TTAGGCTTCA GCGCGGCACT GATATCTG3′ with XbaI site underlined) using the failsafe PCR kit (Epicenter, Madison, WI). The PCR product was purified and digested with HindIII and XbaI and then ligated into a pUC19 derivative plasmid digested with the same enzymes. The entire insert was sequenced and then shuttled to a pET-derived vector (pPCRKDC3) for expression, using the NcoI site to clone into an in-frame polyhistidine tag on the N-terminus. The gene for a second putative KDC (gene product accession number YP_264145, P. arcticus 273-4 KDC) from P. arcticus 273-4 was cloned from genomic DNA using the primers BBP1575 (5′NNNCCATGGC AGTCAACAAT ATACCATCGC TGATTATTTG TTTG3′ with NcoI site underlined) and BBP1576 (5′NNNGAATTCT TAATCAGCTT TTGGCTTAAG CGCAGCGC3′ with EcoRI site underlined) and cloned into an in-frame polyhistidine tag on the N-terminus as described above to construct the pET-derived vector (pPCRKDC17). The gene for a putative MADH (gene product accession YP_958650, MADH) from M. aquaeolei VT8 was cloned from genomic DNA using the primers BBP544 (5′GACACCATGG AGCCAATACC AAAGCATACG CCGC 3′ with NcoI site underlined) and BBP545 (5′GACAAAGCTT CTAGCCTTCT TTCAACGTTT TCATATCAAT GACGAAG3′ with HindIII site underlined) and cloned into an in-frame polyhistidine tag on the N-terminus as described above to construct the pET-derived vector (pPCRADH2). A pET-derived vector harboring a polyhistidine-tagged version of the gene encoding the KDC from L. lactis KdcA was kindly provided by Michael McLeish (University of Michigan) (Yep et al., 2006).

Expression and isolation of purified protein

Each enzyme was expressed separately by transforming the completed plasmids into E. coli BL21 and growing in 1 liter cultures of LB media at 30 °C until an OD at 600 nm of 0.6 was reached and then adding IPTG (50 mg L−1) to induce the expression of protein for 2 h. Cells were isolated by centrifugation at 7000 g for 8 min. Cells were broken by resuspending the pellet in 30 mLs of extraction buffer (50 mM sodium phosphate, pH 7.0, and 300 mM NaCl) and then adding 10 mg of chicken egg white lysozyme and 30 mg of CHAPS detergent along with several flakes of DNAse. Cells were further disrupted by sonication three times for 30 s on ice and then centrifuged at 12 000 g for 8 min to remove cell debris. Polyhistidine-tagged protein was isolated by passing over a nickel-charged metal chelating column (5 mL bed volume, GE Healthcare, Piscataway, NJ) and then washing the column with the extraction buffer supplemented with 50 mM imidazole. Protein was eluted in the same buffer with 500 mM imidazole and then pooled and loaded onto a G-10 Sephadex desalting column (20 mL bed volume, Sigma-Aldrich) equilibrated with extraction buffer. Protein purity for each of the enzymes was analyzed by SDS-PAGE and found to be homogeneous. The proteins were flash-frozen in liquid nitrogen for long-term storage. In certain growths, the cofactors TPP or zinc were added to determine whether this improved the activity of isolated enzymes.

Enzyme activity assays for MADH

Standard enzymatic reactions for ADH were followed spectrophotometrically in quartz cuvettes with a 1-cm pathlength using 1 mL of 50 mM sodium phosphate buffer, pH 7.0, supplemented with 160 μM NADPH and varying amounts of substrate aldehyde. Assays were performed at room temperature (22 °C) unless specified. Following a 60-second period to confirm minimal background absorbance change, reactions were initiated by the addition of a stock solution of MADH. A Varian Bio 50 UV/Vis Spectrophotometer was used to monitor the absorbance at 340 nm. The amount of substrate consumed was assumed to be equal to the amount of NADPH consumed per min using the extinction coefficient for NADPH of 6220 M−1 cm−1 at 340 nm. The results were expressed as μmol of product produced per min per mg of protein. Substrates tested include butanal, isobutanal, acetaldehyde, 3-methylbutanal, 2-methylbutanal, and phenylacetaldehyde. Results obtained for each substrate were fit to Michaelis–Menten kinetics to calculate values of Km and Vmax. A unit of activity is defined as the amount of enzyme required to convert 1 μmol of aldehyde to the corresponding alcohol in 1 min. Results presented represent a single data set, but agreed well with several rounds of replicates in all cases (n ≥ 3).

Coupled enzyme activity assays for KDC

Standard enzymatic reactions were followed spectrophotometrically in quartz cuvettes using a coupled assay based on the reaction described above for the MADH assays, with the following alterations. Reactions used 1 mL of 50 mM MES and 50 mM MOPS buffer, pH 6.5, containing 3 mM of MgSO4 and 0.5 mM TPP and supplemented with 160 μM NADPH and varying amounts of substrate 2-keto acid. All assays also included 1 unit of M. aquaeolei VT8 MADH. Following a 60-s period to confirm minimal background absorbance change, reactions were initiated by the addition of a stock solution of KDC. The success of this assay relied on the lower Km values for each of the generated substrates of the MADH for this reaction, and the excess quantity of MADH included in the coupled assay. The results are expressed in μmol of product produced per min per mg of protein. Substrates tested include pyruvate, phenylpyruvate, α-ketoisocaproate (2-keto-4-methylpentanoate), α-ketomethylvalerate (2-keto-3-methylpentanoate), and α-ketoisovalerate (2-keto-3-methylbutyrate). Results obtained for each substrate were fit to a derivation of the Michaelis–Menten kinetics equation to compensate for cooperativity and calculate values of K0.5, Vmax, and n (Eqn (1)). Results presented represent a single data set, but agreed well with several rounds of replicates in all cases (n ≥ 3). Fits were done iteratively using the software program igor pro (Wavemetrics, Lake Oswego, OR).

display math(1)

pH and temperature effect on KDC activity

The effect of pH on the activity of KDC was investigated in a multiple buffer system containing 50 mM each of MES, MOPS, TES, and TAPS with 3 mM MgSO4 and 0.5 mM TPP. The buffer system was then adjusted over a range of pH 5 to pH 9 with HCl or NaOH and supplemented with 160 μM NADPH. The effect of temperature on the enzyme activity was investigated in the standard buffer as described above. A Varian SPV-1X0 Cary single cell Peltier accessory was used to control the incubation temperature from 4 to 45 °C, and temperatures inside the cuvette were confirmed with an external thermocouple. Excess MADH was tested to confirm sufficient activity under all pH and temperature conditions. The activity assay was performed by the same spectrophotometric method as described above. Where shown, statistics represent the standard deviation for n ≥ 3.

Studies of KDC activities in P. cryohalolentis K5

Cells of P. cryohalolentis K5 were grown in the minimal media described above. After 3 days of growth at 30 °C and 180 r.p.m. on a rotary shaker, cells were collected by centrifugation at 7000 g and washed with buffer (50 mM phosphate, pH 7.0). Aliquots of the cells were sonicated to disrupt the cells, and the supernatant was collected, representing the soluble cell fraction. This sample was then utilized to measure KDC activity with addition of NADPH and MADH using the assay conditions described above, except that reaction products were analyzed by gas chromatography (GC) utilizing a flame ionization detector or mass spectrometer. The protocol for analysis utilized a wax column (HP-FFAP, 25 M, 0.2 mm ID, 0.33 μm d.f., Agilent Technologies, Santa Clara, CA) and a GC-2010 gas chromatograph (Shimadzu Scientific Instruments, Columbia, MD) equipped with a PTV injector. The program utilized a PTV temperature profile of 60 °C for 1 min, with a 10 °C min−1 temperature ramp to 240 °C for 7 min, while the oven temperature profile was 60 °C for 5 min, with a 10 °C min−1 temperature ramp to 220 °C for 5 min. A carrier gas of helium was utilized running at a linear velocity of 30 cm s−1. Further analysis for the identification of specific compounds was carried out using a QP2010 GC/MS (Shimadzu Scientific Instruments, Columbia, MD) equipped with a split/splitless injector. Prior to introduction to the GC, all samples were first filtered through an Amicon Ultra Centrifugal Filter (Ultracel 3K, EMD Millipore, Billerica, MA) to separate the analytes from the remaining protein fraction.

Whole-cell assays were also performed using the same media as described above, except that (NH4)2SO4 was substituted with equivalent molar quantities of various amino acids (leucine, valine, and isoleucine). Supernatants from the growths were collected following centrifugation at maximum speed (approximately 20 000 g) on a tabletop microfuge, and the supernatant was further analyzed by GC as described above.

Results and discussion

Enzyme isolation

The genes coding for two putative branched-chain 2-keto acid decarboxylases (KDCs) from P. cryohalolentis K5 and P. arcticus 273-4 (accession numbers YP_580229 and YP_264145 respectively for the protein products) were cloned into E. coli for heterologous expression with a polyhistidine tag on the N-terminus. The recombinant proteins were expressed from pET-derived vectors and resulted in a highly pure single band for each enzyme as analyzed by SDS-PAGE following purification. A similar approach was taken with a putative MADH from M. aquaeolei VT8, again resulting in a highly purified single band. The KdcA enzyme was purified following the same approach using a plasmid kindly provided by Michael McLeish (Yep et al., 2006). The purified proteins for each of the preparations are shown in Fig. 2. The KDCs from P. cryohalolentis K5 and P. arcticus 273-4 are currently annotated in the published genome sequences (NCBI reference sequence NC_007969.1 and NC_007204.1) as pyruvate decarboxylases. These proteins share some homology in protein primary sequence (c. 40% identity and 60% similarity over a region of 553 residues, see Supporting Information, Data S1) to the established branched-chain KDCs from L. lactis (accession numbers AJ746364 and AY548760), which have a biotechnological interest in the production of branched-chain alcohols or commodity chemicals and are important for the production of flavor compounds in cheese production (Ardo, 2006). While the similarity between the P. cryohalolentis K5 sequence and these two protein sequences is modest, the P. cryohalolentis K5 enzyme also shares significant similarity (> 45% identity) with other enzymes from a range of additional organisms that are annotated as either pyruvate decarboxylases, indolepyruvate decarboxylases or other keto acid decarboxylases, including an additional KDC characterized from M. tuberculosis, which was originally annotated as a pyruvate decarboxylase or indolepyruvate decarboxylase (Werther et al., 2008).

Figure 2.

SDS-PAGE gel of KDC and MADH proteins purified as part of this work. A protein standard (New England Biolabs, Ipswich, MA) is included for reference (lanes 1 and 6 with standard proteins labeled in kDa). A prominent band is seen for each protein corresponding to the expected size, indicating that each protein was purified to homogeneity. Proteins are Lactococcus lactis KdcA (lane 2), Psychrobacter cryohalolentis K5 KDC (lane 3), P. arcticus 273-4 KDC (lane 4), and Marinobacter aquaeolei VT8 MADH (lane 5).

MADH assay

Similar to approaches utilized by others to characterize KDC enzymes (Yep et al., 2006), we selected a coupled spectrophotometric assay using the MADH from M. aquaeolei VT8, a bacterial strain that produces a range of interesting high-value lipids, and is a model bacterium in our laboratory (Barney et al., 2012). The MADH from M. aquaeolei VT8 (NCBI accession YP_958650) was characterized kinetically and the results fit to the Michaelis–Menten equation. MADH reduced a variety of aldehydes as shown in Table 1. The best substrate for the enzyme was 3-methylbutanal and showed similar levels of activity with other aldehydes including isobutanal and n-butanal, supporting the classification of this enzyme as a MADH that acts on small (acetaldehyde), medium (butanal), aromatic (phenylacetaldehyde), and branched-chain (isobutanal, 4-methylbutanal, and 3-methylbutanal) aldehydes. The MADH was selected here based on similarity found (28% identical and 43% similarity over a region of 246 residues) between this enzyme and the primary sequence of a MADH (NCBI accession ABC02081) identified in the melon plant Cucumis melo (Manriquez et al., 2006). Similar to what was found in the C. melo MADH, the MADH from M. aquaeolei VT8 showed a high level of activity with acetaldehyde and a comparable Km for acetaldehyde (0.3 mM found here versus 0.25 mM in C. melo). The activity was also similar when comparing larger aldehydes such as n-butanal, although the Km values for these substrates were not reported for C. melo (Manriquez et al., 2006). M. aquaeolei VT8 MADH is easily expressed and purified while having ideal kinetic characteristics for further conversion of the branched-chain aldehyde products of KDC (Table 1). Based on the low Km values and high enzyme activity with specific aldehyde substrates, it was determined that the MADH from M. aquaeolei VT8 could be substituted for commercially isolated ADH in a coupled enzyme assay with KDCs from P. cryohalolentis K5, P. arcticus 273-4, and L. lactis.

Table 1. Kinetic parameters for Marinobacter aquaeolei VT8 MADH
Aldehyde substrateaKm (μM)kcat (s−1)kcat/Km (mM−1 s−1)
  1. a

    Results for kcat were derived from Vmax measurements under saturated conditions. Kinetic measurements were repeated several times to confirm reproducibility, although the results shown are derived from the fit to a single data set.

3-methylbutanal20201000
Butanal1712705
Phenylacetaldehyde6533508
2-methylbutanal3517485
Isobutanal9318194
Acetaldehyde3001756

Coupled assays for KDC activity

Using the coupled assay with KDC and MADH (Fig. 1), it was possible to determine kinetic parameters for the P. cryohalolentis K5 and P. arcticus 273-4 KDC enzymes (Table 2). This data had to be fit to a derivation of the Hill equation (Eqn (1)) as these KDC enzymes exhibit cooperativity. P. cryohalolentis K5 KDC had the highest kcat/K0.5 value when using 2-keto-4-methylpentanoate (α-ketoisocaproate). These results contrast with Kivd and KdcA from L. lactis, which had higher activities with 2-keto-3-methylbutyrate (α-ketoisovalerate) or 2-keto-3-methylpentanoate and reported Hill coefficients indicating these were noncooperative (De la Plaza et al., 2004; Smit et al., 2005). The allosteric features found for P. cryohalolentis K5 and P. arcticus 273-4 KDCs are similar to what is reported for M. tuberculosis KDC, although the substrate profiles vary considerably (Werther et al., 2008). Kivd from L. lactis was found to have a Vmax of c. 118 μmol min−1 (mg protein)−1 and a Km of 1.9 mM with 2-keto-3-methylbutyrate (De la Plaza et al., 2004). Comparisons made here with the Psychrobacter KDCs and KdcA indicate similar levels of activity with all three enzymes outside of the difference in cooperativity and compare well with the results reported by Yep et al. (Yep et al., 2006). Levels of specific activity found with pyruvate and indole pyruvate were extremely low (about 0.2% and 0.5% of what was found for 2-keto-3-methylpentanoate respectively) in the Psychrobacter KDCs, which is more similar to what was found with Kivd (De la Plaza et al., 2004) and differentiates the results from the higher levels of activity found with the M. tuberculosis KDC for these additional substrates (Werther et al., 2008). Although the substrate specificity of the enzyme seems to support classification as an α-ketoisocaproate decarboxylase (KicD), based on the general broad substrate ranges found in this expanding class of branched-chain KDCs, we favor the general terminology as KDC.

Table 2. Kinetic parameters for KDCs
2-keto acid substrateaK0.5 (mM)kcat (s−1)kcat/K0.5 (mM−1 s−1) n
  1. a

    Results for kcat were derived from Vmax measurements under saturated conditions. Kinetic measurements were repeated several times to confirm reproducibility, although the results shown are derived from a single data set. Coupled assays were performed as described in the Materials and Methods using an excess of MADH at room temperature (22 °C) and are reported for the KDC enzyme.

P. cryohalolentis K5 KDC
2-keto-4-methylpentanoate0.138.565.32.00
2-keto-3-methylpentanoate2.12712.71.88
Phenylpyruvate0.231.87.81.51
2-keto-3-methylbutyrate5.3336.21.96
P. arcticus 273-4 KDC
2-keto-4-methylpentanoate0.57141.55
2-keto-3-methylpentanoate4.3235.31.75
Phenylpyruvate0.41.74.11.48
2-keto-3-methylbutyrate11.5292.51.44
L. lactis KdcA
2-keto-3-methylpentanoate1.42517.41.00

Features for assay optimization

Several reports have characterized the kinetic features of branched-chain KDCs. A comparison of the methods used by different laboratories to measure activities reveals stark differences in these approaches. For example, several laboratories included TPP, or (thiamine diphosphate) and magnesium salts in assay buffers, while others did not. Selection of buffer systems ranged from MES to phosphate. Analytical techniques to measure substrate or product have utilized derivatization and chromatographic separation versus coupled spectrophotometric assays. We favored the use of coupled spectrophotometric assays for this work, as they provide real-time data that can be significantly more informative than fixed time point assays. We also found that the three KDC enzymes (L. lactis KdcA, P. cryohalolentis K5 KDC, and P. arcticus 273-4 KDC) each exhibited suboptimal levels of activity in the absence of TPP. L. lactis KdcA was practically inactive in buffer devoid of TPP following isolation, even if cells expressing KdcA were grown in the presence of additional TPP. P. arcticus 273-4 KDC showed much higher activity from preparations of the enzyme when supplemented with TPP during expression, while P. cryohalolentis K5 KDC had similar activity when grown on standard rich media preparations such as LB Broth without the inclusion of added TPP. However, even for P. cryohalolentis K5 KDC, lack of TPP in the assay media resulted in differences in kinetic parameters (generally increased K0.5 and lower specific activities). This indicates that certain KDC enzymes might have better utility and stability based on a higher assimilation or affinity for the TPP cofactor.

Cooperativity in KDC enzymes

Reports of related branched-chain KDCs have illustrated significant differences in the substrate profiles and level of substrate activation in these isolated enzymes (De la Plaza et al., 2004; Smit et al., 2005; Yep et al., 2006; Werther et al., 2008). However, as mentioned above, various methods and buffer systems have been utilized in these different characterizations (De la Plaza et al., 2004; Smit et al., 2005; Yep et al., 2006; Werther et al., 2008). As the results here indicate that the P. cryohalolentis K5 and P. arcticus 273-4 KDCs are cooperative enzymes, we also tested the L. lactis KdcA, which was reported to show no evidence of substrate activation (Yep et al., 2006). Each of the three KDCs isolated were run under similar temperature and buffer conditions with the same substrate (2-keto-3-methylpentanoate) and are plotted together (Fig. 3) to illustrate that this is a clear distinction between the L. lactis KdcA and Psychrobacter KDCs. These results illustrate that P. cryohalolentis K5 and P. arcticus 273-4 KDCs share specific features with both the L. lactis KdcA and Kivd enzymes and also with the M. tuberculosis KDC (De la Plaza et al., 2004; Smit et al., 2005; Yep et al., 2006; Werther et al., 2008). While the P. cryohalolentis K5 and P. arcticus 273-4 KDCs shared a similar cooperativity with M. tuberculosis KDC, many other features related to substrate specificity were far more similar to the KDCs from L. lactis. Particularly, the substrate profiles were much closer to L. lactis, with little activity found for either pyruvate or indolepyruvate, while M. tuberculosis KDC was found to have a significant activity with indolepyruvate (Werther et al., 2008). The activities measured here for Psychrobacter KDCs with pyruvate and indolepyruvate were even lower than activities found when KdcA (from L. lactis) was tested as a comparison with indolepyruvate and pyruvate. These features indicate that the Pyschrobacter KDCs may represent a different and intermediate class of KDC from both L. lactis and M. tuberculosis and further indicates that there may be a much broader range of these enzymes from a number of other bacteria. This also supports reclassification of these Psychrobacter enzymes as branched-chain KDCs and not pyruvate decarboxylases, as they are currently annotated.

Figure 3.

Data used to calculate kinetic parameters and the resulting curve fits for the various KDC enzymes presented in Table 2 with the substrate 2-keto-3-methylpentanoate. Results are shown for Psychrobacter cryohalolentis K5 KDC (♢), P. arcticus 273-4 KDC (○), and Lactococcus lactis KdcA (▵).

Enzyme pH and temperature profiles

A comparison of the pH and temperature profiles with the three different KDC enzymes tested here revealed that each enzyme had broad activity between pH 6.0 and 7.0 and dropped off slowly at higher and lower pH values, similar to what has been reported previously for KdcA and Kivd (De la Plaza et al., 2004; Smit et al., 2005). Temperature profiles for the upper range (20–50 °C) have been reported previously for Kivd (De la Plaza et al., 2004), but did not extend to lower temperatures (4–20 °C). Here, three KDC enzymes were tested from 4 to 45 °C (Fig. 4). Interestingly, all three enzymes showed a similar pattern of decreasing activity with lower temperature, including the KdcA enzyme from L. lactis, and the only stark differences were seen at higher temperatures. P. arcticus 273-4 KDC showed maximum activity at 30 °C and then dropped steadily above that temperature. P. cryohalolentis K5 KDC showed maximum activity at 35 °C, lost activity at 40 °C and then exhibited instability at 45 °C, where the enzyme dramatically lost activity during the real-time assay. The results for L. lactic KdcA were very similar to what was reported for Kivd, with continued increasing activity up to 45 °C. All of the enzymes remained active at 4 °C (Fig. 4).

Figure 4.

Enzyme activity at varying temperatures. Assays were run with 10 mM 2-keto-3-methylpentanoate as described in the methods. The Psychrobacter cryohalolentisK5 KDC rate at 45 °C was difficult to fit to a linear rate equation as the activity dropped rapidly during the real-time assay, indicating instability at this temperature. All other samples and temperatures fit well to a linear equation revealing peak activity for P. arcticus 273-4 KDC at 30 °C, peak activity for P. cryohalolentis K5 KDC rate at 35 °C, while Lactococcus lactis KdcA activity was still increasing at 45 °C.

In vivo KDC activity in P. cryohalolentis K5

Analysis of the P. cryohalolentis K5 soluble protein fraction following cell disruption revealed that P. cryohalolentis K5 produced the expected alcohols when provided excess MADH and NADPH and the corresponding 2-keto acids. Analysis by GC yielded the expected alcohols corresponding to the three 2-keto acids that were found to have significant activity with the heterologously expressed KDC from P. cryohalolentis K5 in Table 2. Further, incubation of intact cells with specific 2-keto acids also resulted in the production of the expected corresponding alcohol (without the inclusion of MADH and NADPH), indicating that P. cryohalolentis K5 exhibits KDC activity in vivo, which would be expected if it were expressing active KDC and ADH enzymes.

Growth experiments where the ammonium sulfate from the defined media was exchanged with an equimolar concentration of one of the three amino acids (leucine, valine, or isoleucine) leading to 2-keto acids used in the assays described above resulted in an increased lag phase for all three amino acids, and a significant decrease in cell density found for valine and isoleucine (c. 20% of the values obtained for ammonium sulfate), while leucine was able to support a similar cell density to that found with ammonium sulfate. Analysis of the spent media following the growth also showed positive evidence of the expected alcohols. These results indicate that P. cryohalolentis K5 can naturally produce these medium branched-chain alcohols when grown on the specific precursor amino acids, which is similar to what is found for L. lactis (Smit et al., 2005). Further confirmation that observed KDC activity in P. cryohalolentis K5 is the result of the enzyme characterized here will require gene deletion studies in the native strain. However, although others have reported the potential for genetic manipulation in P. arcticus 273-4 (Bakermans et al., 2009), which we have similarly been able to successfully transform, the transformation of P. cryohalolentis K5 has not yet been successful to date. Efforts to delete this gene in P. arcticus 273-4 and compare in vivo activities may be pursued in future studies.

Conclusions

These results indicate that KDC enzymes capable of producing branched-chain aldehydes may be more common in a broader range of bacteria, with a larger range of kinetic features than what has been characterized previously. Growth of P. cryohalolentis K5 on minimal media with leucine, valine, or isoleucine as the sole nitrogen source resulted in the production of specific alcohols, indicating that the Ehrlich pathway exists in this species to convert these amino acids to the corresponding alcohols following transfer of the amino group and conversion by a functional KDC. It is unclear why P. cryohalolentis K5 would produce a KDC enzyme with this activity. It is further unclear why some organisms have evolved enzymes that exhibit varying levels of cooperativity. Based on these findings, it is expected that many additional KDC enzymes with activity toward branched-chain 2-keto acids derived from specific amino acids will be identified in the future.

Acknowledgements

This work is supported by a grant from the National Science Foundation (Award Number 0968781) to B.M.B and the Microbial Engineering Program through the Biotechnology Institute at the University of Minnesota for partial support of J.W. We thank Bradley Wahlen for assistance in early cloning efforts. We thank James Tiedje for providing the P. cryohalolentis K5 and P. arcticus 273-4 strains used in this work. We thank Michael McLeish for providing the KdcA expressing plasmid and Justin Peterson and Rachel Cordes for assisting in early isolation and preliminary characterization studies with the KDC.

Ancillary