Efficient PCR-Based Amplification of Diverse Alcohol Dehydrogenase Genes from Metagenomes for Improving Biocatalysis: Screening of Gene-Specific Amplicons from Metagenomes

Abstract

Screening of gene-specific amplicons from metagenomes (S-GAM) has tremendous biotechnological potential. We used this approach to isolate alcohol dehydrogenase (adh) genes from metagenomes based on the Leifsonia species adh gene (lsadh), the enzyme product of which can produce various chiral alcohols. A primer combination was synthesized by reference to homologs of lsadh, and PCR was used to amplify nearly full-length adh genes from metagenomic DNAs. All adh preparations were fused with lsadh at the terminal region and used to construct Escherichia coli plasmid libraries. Of the approximately 2,000 colonies obtained, 1,200 clones were identified as adh positive (∼60%). Finally, 40 adh genes, Hladh-001 to Hladh-040 (for homologous Leifsonia adh), were identified from 223 clones with high efficiency, which were randomly sequenced from the 1,200 clones. The Hladh genes obtained via this approach encoded a wide variety of amino acid sequences (8 to 99%). After screening, the enzymes obtained (HLADH-012 and HLADH-021) were confirmed to be superior to LSADH in some respects for the production of anti-Prelog chiral alcohols.

INTRODUCTION

Metagenomics is an emerging and powerful tool for the isolation of genes, the enzyme products of which have industrial applications (1–6). Screening of metagenomic libraries to find novel enzymes or enzymes homologous to those described previously has been used successfully to isolate lipases (7, 8), amylases (9), amidases (10), oxidoreductases (11), dehydratases (12), cytochrome P450 (13), styrene monooxygenase (14), and other enzymes (1–6). However, most previous approaches featured metagenomic DNA extraction and Escherichia coli library construction, followed by sequence- or function/molecule-based screens of the library. Such approaches are very time-consuming and inefficient, especially in terms of detection; much of the DNA sequenced and analyzed is irrelevant, and target genes may be expressed ambiguously in E. coli host cells. PCR amplification of truncated genes from metagenomes would facilitate the identification of genes encoding superior enzymes and yield homologous gene sets that could be used for DNA shuffling (15). Although previous studies based on PCR-mediated methods that utilize primers designed from inner conserved sequences have been conducted for biocatalysts, including lipase (8), cytochrome P450 (13), 2,5-diketo-d-gluconic acid reductase (16), Pseudomonas alcohol dehydrogenase (ADH) (17), and other biocatalysts (3, 4, 6), the methods are not very efficient in many cases and often fail to produce complete functional genes.

Enantioselective organic synthesis is useful for producing chiral synthones for the preparation of fine chemicals, including pharmaceuticals and agricultural chemicals. The asymmetric reduction of ketones is one of the most promising approaches, because no substrate is lost, in contrast to when racemic separation is performed. Chiral metal complexes, such as BINAP-Ru, have been used successfully as chemocatalysts in a number of cases of enantioselective synthesis (18). However, biologically based methods using enzymes or whole-cell systems offer several advantages over the BINAP process for industrial applications, including improved material handling and lower costs for the preparation of catalysts (19–21).

Previously, we reported an efficient method for producing both enantiomers of chiral alcohols by asymmetric hydrogen-transfer bioreduction of ketones in a 2-propanol (IPA)–water medium using E. coli biocatalysts expressing a mutated form of phenylacetaldehyde reductase (PAR) (22, 23) and Leifsonia ADH (LSADH) (24, 25). However, PAR and LSADH do not fully possess the required substrate specificity or stereospecificity; for example, LSADH does not accept methyl benzoylformate, 2-acetylpyridine, or 3-quinuclidinone as a substrate (26, 27). Thus, we sought to clone genes encoding enzymes with properties distinct from those of LSADH. Moreover, dehydrogenases, such as LSADH, that yield anti-Prelog chiral alcohols [e.g., (R)-1-phenylethanol from acetophenone] are rare in nature (26, 27); however, the pharmaceutical industry demands this type of enzyme (28, 29).

In this paper, we describe the construction of a metagenomic library of enzyme genes that focused on the lsadh gene. Our approach, which involved PCR amplification of nearly full-length genes from metagenomes fused with the terminal region of an lsadh-expressing vector, enabled the isolation of many novel and diverse adh genes and lsadh homologs. This highly efficient approach of screening of gene-specific amplicons from metagenomes (S-GAM) shows tremendous biotechnological potential for obtaining gene resources from metagenomes. We also present the potential use of novel enzymes as biocatalysts for converting ketones to various anti-Prelog chiral alcohols at high production levels.

MATERIALS AND METHODS

Metagenome preparation.

Metagenomic DNA was extracted from 20 environmental samples, including various soils collected from farms and paddy fields, gardens at independent sites in Japan, and farm (35 to 45°C) and bark (50 to 80°C) composts in Toyama, Japan, using an ISOIL for bead beating kit (Nippon Gene, Tokyo, Japan) without further purification. Bark compost samples in fermentation at approximately 50 to 80°C were generously supplied by a compost-producing company (Hokuriku Port Service, Toyama, Japan). Successful extraction of DNA from the soil and compost samples was confirmed using agarose gel electrophoresis; these DNA samples served as templates for PCR.

Primers, PCR conditions, and cloning of adh genes.

Standard techniques were used for DNA manipulation (30). E. coli JM109 cells were used to host adh genes fused with the pKELA-del plasmid. This vector was derived from pKELA (27), which expresses the lsadh gene of pKK233-3, by deletion of part of the lsadh gene (100 bp) with XhoI, and then PCR was performed to introduce fusion sites to both 5′ ends using the following primers: F-vec-1, 5′-ACCGCCCAGTGACCGGGCTGCAGGT-3′, and R-vec-1, 5′-ACGATCGCGGACCGGTCGGCGACGT-3′ (underlined sequences indicate fusion sites). PCR was performed using KOD FX Neo DNA polymerase (Toyobo, Osaka, Japan). The reaction mixture contained 10 μl 2× buffer for the KOD FX Neo kit, 2 nmol of each deoxynucleoside triphosphate (dNTP), 8 pmol of each primer, and 0.4 U DNA polymerase in a total volume of 20 μl. PCR commenced at 94°C for 2 min, followed by 30 cycles at 98°C for 10 s, 60°C for 30 s, and 68°C for 3 min, and then the sample was kept at 68°C for 5 min. The linearized vector contained 15 nucleotides at both 5′ ends for fusion with the adh gene amplified directly by PCR using metagenomic DNA. The adh gene of this vector is under the control of the tac promoter. DNA sequences were determined for both strands using a capillary DNA sequencer (ABI PRISM 310; Applied Biosystems Life Technologies, Carlsbad, CA).

PCR was performed under optimized conditions using KOD FX Neo DNA polymerase or Phusion hot start flex DNA polymerase (New England BioLabs, Tokyo, Japan) to obtain amplicons from the metagenomes. The reaction conditions were based on those suggested by the manufacturer, except that the reaction mixture contained an approximately 10-fold higher concentration of each primer for KOD FX Neo DNA polymerase and 5% (vol/vol) dimethyl sulfoxide for Phusion hot start flex DNA polymerase; both polymerases offer high fidelity and robust performance. Hot-start and step-down PCR protocols were used. Each reaction mixture contained 10 μl 2× buffer for the KOD FX Neo kit, 2 nmol of each dNTP, 80 pmol of each primer, 10 to 50 ng metagenomic DNA, and 0.4 U DNA polymerase in a total volume of 20 μl. In the case of Phusion hot-start flex DNA polymerase, the reaction mixture contained 4 μl 5× Phusion GC buffer, 2 nmol of each dNTP, 8 pmol of each primer, 10 to 50 ng metagenomic DNA, 1 μl dimethyl sulfoxide, and 0.4 U DNA polymerase in a total volume of 20 μl. PCR commenced at 94°C for 2 min, followed by a step-down protocol of 5 cycles at 98°C for 10 s, 74°C for 1 min, 5 cycles at 98°C for 10 s, 72°C for 1 min, 5 cycles at 98°C for 10 s, 70°C for 1 min, 5 cycles at 98°C for 10 s, and 68°C for 1 min, and finally the sample was kept at 68°C for 7 min. Good amplification of the target genes was obtained from all samples using both sets of primers, F-1/R-1 and F-2/R-2 (Table 1). To accelerate the fusion reaction with the plasmid vector, we generally used the F-2/R-2 primer set for the subsequent experiments.

TABLE 1.

Primers used for gene-specific amplification of lsadh from metagenomes

Primer (nucleotide length)	Sequencea (5′–3′)
F-1 (25)	CGGTCCGCGATCCTVACMGGMGSMG
F-2 (28)	GACCGGTCCGCGATCGTVACMGGMGSMG
F-3 (28)	GATCGTTCAGCAATCGTVACMGGMGSMG
F-4 (35)	CGGTCCGCGATCGTGACCGGMGSCGGSTCSGGSAT
F-5 (35)	GCCGACCGGTCCGCGATMGTGACCGGMGSCGGSTC
R-1 (25)	TCACTGGGCGGTRTABCCRCCRTCB
R-2 (28)	CGGTCACTGGGCGGTRTABCCRCCRTCB
R-3 (29)	CGGTCACTGGGCGGTRTABCCRCCRTCBA
R-4 (28)	CGGTTATTGAGCAGTRTABCCRCCRTCB
R-5 (29)	CGGTTATTGAGCAGTRTABCCRCCRTCBA
R-6 (35)	CACTGCTCGGTGTAGCCGCCRTCSACCAGRTGRTA
R-7 (35)	CGGTCACTGCGCGGTSTAGCCGCCRTCSACCAGRT

^aV, G, A, or C; M, A or C; S, G or C; R, A or G; B, G, T, or C. No amplification was observed for the following primer combinations: F-3/R-3, F-3/R-4, F-3/R-5, F-4/R-6, F-4/R-7, F-5/R-6, and F-5/R-7. Combinations other than F-1/R-1 and F-2/R-2 were not tested.

Amplified fragments were separated by agarose gel electrophoresis, purified using the Wizard Plus SV Minipreps DNA purification system (Promega, Fitchburg, WI), and fused between the same sites of pKELA-del with an In-Fusion HD cloning kit (Clontech, Mountain View, CA). The reaction mixture consisted of 6 μl pKELA-del (ca. 20 ng/μl), 2 μl amplified DNA fragments (5 to 20 ng/μl), and 2 μl In-Fusion enzyme in a total volume of 10 μl, and the mixture was incubated at 50°C for 15 min. The plasmids obtained were electroporated into E. coli JM109. Clones were grown at 37°C on agar plates containing Luria-Bertani (LB) medium (1% [wt/vol] tryptone, 0.5% [wt/vol] yeast extract, and 1.0% [wt/vol] NaCl; pH 7.0) with 100 μg/ml ampicillin.

Screening for ADH activity in E. coli clones.

Screening for ADH activity in E. coli was performed for about 2,000 E. coli clones in 96-well plates by spectrophotometric measurement of the activity of 1,1-dichloroacetone and phenyl trifluoromethyl ketone (PTK). E. coli cells were cultured at 37°C overnight with shaking in LB liquid medium (0.5 ml) containing 100 μg/ml ampicillin and 0.1 mM isopropyl-β-thiogalactopyranoside (IPTG) in 96-well deep plates. IPTG was added to the culture medium to confirm expression, although the enzyme expression of pKELA-del-HLADH was leaky and barely controlled by IPTG. The cells were collected by centrifugation after cultivation, rinsed in 100 mM potassium phosphate buffer (KPB) (pH 7.0), and disrupted in 0.1 ml BugBuster master mix (Novagen, Merck Biosciences Japan, Tokyo, Japan) for 20 min at room temperature. Cell debris was removed by centrifugation, and the supernatant was used as a crude enzyme solution. ADH activity was measured using a microplate reader at 340 nm and 25°C; the reaction mixture consisted of 100 mM KPB (pH 7.0), 0.3 μmol NADH, 50 μmol substrate, and 10 μl crude enzyme in a total volume of 1.0 ml.

Purification of recombinant HLADH-012 and HLADH-021.

HLADH-012 and HLADH-021 were purified at 0 to 4°C in 20 mM KPB (pH 7.0), unless otherwise stated. For HLADH-012, washed recombinant E. coli cells from 200 ml culture broth (37°C for 17 h in a shake flask) were suspended in 30 ml buffer and disrupted using an ultrasonic oscillator (Ultra Sonic Disrupter UD-200; Tomy Corp., Tokyo, Japan) for 150 s (5 disruption sequences of 30 s followed by a 60-s interval for cooling). After centrifugation (10 000 × g, 10 min), the cell extract was mixed with ammonium sulfate up to 40% saturation and maintained overnight under gentle stirring. The precipitate was removed by centrifugation (10,000 × g, 10 min), and the supernatant was recovered. The solution was applied to a Toyopearl Butyl-650M column (2.5 by 13 cm; Tosoh, Tokyo, Japan) equilibrated with buffer containing 1.0 M ammonium sulfate. The enzyme was eluted at 1 ml/min with a linear 1.0 to 0 M ammonium sulfate gradient in the same buffer. Fractions exhibiting high levels of enzyme activity (27 to 30 min) were collected and dialyzed against the buffer. The enzyme preparation then was loaded onto a Resource Q column (0.64 by 3 cm; GE Healthcare Japan, Tokyo, Japan) equilibrated with the buffer described above and connected to the AKTA purifier system (GE Healthcare Japan). The enzyme was eluted using a linear 0 to 0.5 M NaCl gradient in the same buffer at a flow rate of 0.3 ml/min. Fractions exhibiting high levels of enzyme activity (43 to 46 min) were collected, desalted, and concentrated using a Centriprep YM-30 centrifugal filter unit (EMD Millipore, Billerica, MA). The enzyme solution obtained was the purified preparation used for enzyme characterization.

HLADH-021 was purified from 100 ml culture broth in the same way as HLADH-012, and after ammonium sulfate fractionation up to 40%, the supernatant solution was similarly applied to a Toyopearl Butyl-650M column. Fractions exhibiting high levels of enzyme activity (43 to 47 min) were collected and desalted using a Centriprep YM-30 filter unit. The resulting enzyme solution was applied to a DEAE-Toyopearl column (2.5 by 7 cm) equilibrated with the buffer, and the enzyme was eluted using a linear 0 to 0.5 M NaCl gradient in the same buffer at a flow rate of 1.0 ml/min. Fractions exhibiting high levels of enzyme activity (31 to 33 min) were collected and concentrated using a Centriprep YM-30 filter unit. The enzyme solution obtained was the purified preparation used for enzyme characterization.

Analysis of enzymatic properties.

General ADH activity was assayed spectrophotometrically at 25°C by measuring the decrease in the absorbance of NADH at 340 nm (ε = 6,220 M⁻¹ · cm⁻¹). Each reaction mixture contained 10 μmol PTK or other substrates, 0.3 μmol NADH, 1 mmol KPB (pH 7.0), and 10 μl enzyme solution in a total volume of 1.0 ml. The oxidation activity of ADH also was measured at 340 nm in reaction mixtures with a total volume of 1.0 ml containing 10 μmol 2-propanol as the substrate, 1.0 μmol NAD⁺, 1 mmol KPB (pH 7.0), and 10 μl enzyme solution. One unit of enzyme was defined as the amount that converted 1 μmol NADH or NAD⁺ in 1 min under these conditions. The kinetic parameters of HLADH-012 and HLADH-021 were calculated from a Lineweaver-Burk plot. Protein concentration was estimated by measuring the absorbance of protein-containing solutions at 280 nm or by the method of Bradford (31) using bovine serum albumin as a standard (Bio-Rad protein assay kit; Bio-Rad Laboratories, Hercules, CA).

SDS-PAGE was performed using 12% (wt/vol) polyacrylamide slab gels and the Tris-glycine buffer system of Laemmli (32). The molecular mass of the enzyme subunit was determined from the relative mobility of standard proteins.

The molecular mass of the native recombinant enzyme was determined using an LC-10 high-performance liquid chromatography (HPLC) system (Shimadzu, Kyoto, Japan) equipped with a TSK-GEL G3000SW_XL column (Tosho). Ten microliters of the obtained sample solution was separated by using a mobile phase containing 20 mM KPB and 200 mM NaCl (pH 7.0). The flow rate was 1.0 ml/min, and the column temperature was kept at 25°C. The protein was monitored at 280 nm, and molecular mass was estimated from the retention times of authentic molecular weight markers (Oriental Yeast Co., Ltd., Tokyo, Japan).

Analysis of enzymatic products.

For the analysis of enzymatic products, the reaction mixture contained 50 μmol KPB (100 μmol for 1-Boc-pyrrolidinone, 200 μmol for 3-quinuclidinone) (pH 7.0), 50 mg substrate, 1 μmol NAD⁺, 5% (vol/vol) 2-propanol, and recombinant E. coli cells collected from a 10-ml culture broth. The resting cell reaction proceeded for 24 h at 25°C with shaking at 2,500 rpm using a microtube shaker (M-BR-022UP; Taitec, Saitama, Japan). The mixture next was extracted twice with ethyl acetate (or, in the case of 3-quinuclidinol, 1-butanol at an alkaline pH of 12, achieved with 6 N NaOH) (33), and the combined extracts were dried using anhydrous Na₂SO₄ before analysis. The absolute configuration and enantiomeric purity of the product was evaluated using the peak areas of alcohol products visible on GC or HPLC compared with authentic compounds by following previously reported procedures (23, 25, 33).

Nucleotide sequence accession numbers.

The nucleotide sequences of the metagenomic adh genes determined here, Hladh-001 to Hladh-040 (for homologous Leifsonia adh), homologs of lsadh from L. poae NBRC103069 (lpadh), and L. naganoensis NBRC103131 (lnadh), were submitted to the DNA Data Bank of Japan under the accession numbers AB916600 to AB916639, AB917070, and AB917071, respectively.

RESULTS AND DISCUSSION

Metagenome preparation, design of primers, and amplification of adh genes.

The average weight of DNA obtained from 0.5 g (wet weight) of the soil samples and those of the farm and bark composts used in this study was 1.8 ± 1.3 (standard deviations [SD]) and 2.7 ± 2.3 μg, respectively; thus, the SD were very high because yields varied greatly depending on the nature of the soil. We isolated metagenomic DNA from 20 soil and compost samples, and these served as templates for the amplification of homologous lsadh genes.

Some of the primer sets were designed with reference to the N- and C-terminal region sequences of lsadh and its related genes (Table 1), which belong to the short-chain dehydrogenase/reductase (SDR) family. SDRs are proteins of approximately 250 to 300 amino acid residues, have a wide variety of substrate specificities (34), and are useful biocatalysts for producing chiral alcohols from various ketones (35). Alignment analysis of some ADHs belonging to the SDR family at the terminal regions showed that they shared 37 to 58% identity with LSADH (Fig. 1), which clearly indicated the well-conserved regions among them. GXXXGXG can interact with coenzyme NAD(P)H in the near N-terminal region and VDGGXTA in the C-terminal region, which consists of a β-sheet secondary structure in SDRs. Primers of ∼30 bp were designed based on these sequences and contained the regions for fusion with the expression vector pKELA-del (Fig. 1). A combination of the primers F-2 and R-2 was used to amplify nearly full-length genes from the metagenomes after they were fine-tuned. Adjustments were made for the regeneration of codon usage, to primer length and melting point (T_m) for the prevention of primer duplex formation, to primer set combinations, and to optimize other PCR conditions.

FIG 1.

Primer design (A) and schematic procedure of the S-GAM technique (B). Primers were designed based on the alignment of LSADH and putative SDRs from some microorganisms constructed by CLUSTALW. Identical residues at the terminal regions are shown by white letters on a black background. The conserved amino acid sequences surrounded by a dotted square were used as reference points in the design of each primer set.

We used hot-start and step-down PCR protocols to avoid nonspecific amplification and to support the sufficient amplification of DNA using KOD FX Neo DNA polymerase or Phusion hot-start flex DNA polymerase. Both polymerases offer high fidelity and robust performance. It was important to optimize and fine-tune the PCR conditions, including the choice of DNA polymerase, primer concentration, and so on, to obtain successful amplification of the genes from the metagenomes. We observed 100% amplification of target genes from 12 general soil metagenomic samples isolated from farms and paddy fields or garden soil samples collected in Japan; in addition, target genes from 3 genomes isolated from the genus Leifsonia (L. poae NBRC103069, L. naganoensis NBRC103131, and L. aquatica NBRC15710) were amplified using KOD FX Neo DNA polymerase and from 8 samples of farm and bark composts in fermentation at 35 to 80°C, as well as target genes isolated from 4 genomes (L. aurea NBRC104579, L. ginsenji NBRC104580, L. pindariensis JCM15132, and L. shinshuensis NBRC103132), were amplified using Phusion hot start flex DNA polymerase. Under the optimized PCR conditions, we successfully amplified genes from all 20 independent environmental metagenomes and 7 Leifsonia sp. genomes. The results showed that once the amplification conditions for PCR were determined, the target genes could be obtained easily from various metagenomes. Thus, this simple PCR-based approach has tremendous biotechnological potential for obtaining useful or novel enzyme genes from metagenomes.

Screening and analyses of adh genes from the metagenomic library.

PCR-amplified genes were fused with the original lsadh gene at both terminal regions in pKELA-del, an expression vector of the homologous lsadh genes in E. coli constructed from pKELA (27), as described in Materials and Methods. They then were transferred into E. coli cells and expressed. This simple technique prevented the ambiguous expression of the target genes and enabled the high-throughput screening of enzyme activity. Approximately 2,000 colonies obtained from 20 soil and compost metagenomes were measured directly using their enzymatic activity for two substrates, 1,1-dichloroacetone and PTK; 1,1-dichloroacetone is a small-sized molecule and a good substrate for many ADHs, whereas PTK is a medium-sized molecule and an original substrate of LSADH. Approximately 1,200 (60%) of the 2,000 colonies were positive in a spectrophotometric assay using 1,1-dichloroacetone or PTK as the substrate. Thus, a library containing homologous genes of lsadh was constructed efficiently by gene-specific PCR amplification of metagenomes and fusion with a suitable expression vector.

To facilitate the analysis of the adh genes from the library, 223 of the 1,200 clones (∼20%) were selected randomly and sequenced. Genes were given the descriptor Hladh (homologous Leifsonia adh) to indicate their similarity to the lsadh gene. After eliminating duplicate genes in terms of amino acid sequence, 40 different genes were isolated from the 223 clones: 29 from bark compost in fermentation (50 to 80°C), 7 from farm compost (35 to 45°C), and 4 from general soil samples. Notably, the amino acid sequences encoded by the Hladh genes obtained via this metagenomic approach varied greatly, especially those isolated from the bark compost samples. The results are summarized in Table 2. Amino acid sequence identity was compared to that of lsadh, except at the chimeric regions (amino acids 1 to 12 and 249 to 251 in LSADH consisted of 251 amino acid residues), by BLASTP analysis (36). The isolated genes were divisible into at least 5 groups: Hladh-001 to Hladh-011, obtained from the general soil and farm compost (35 to 45°C) samples, with 98 to 99% amino acid sequence identity with lsadh; Hladh-014, Hladh-015, and Hladh-016, from bark compost (50 to 80°C), 73 to 75%; Hladh-012, Hladh-013, Hladh-017 to Hladh-025, and Hladh-034 to Hladh-038, from bark compost (50 to 80°C), 50 to 63%; Hladh-026 to Hladh-032, from bark compost (50 to 80°C), 36 to 44%; and Hladh-033, Hladh-039, and Hladh-040, from bark compost (50 to 80°C), <17%. Table 3 shows the results of the BLASTP analyses of the isolated adh gene products with known SDRs. HLADH-012, HLADH-013, HLADH-017 to HLADH-026, and HLADH-031 to HLADH-040 shared <65% sequence identity with known SDRs, including putative ones; therefore, we deduced that they were novel functional ADHs. Interestingly, 45 of the 61 LSADH homologs isolated from the general soil and farm compost samples completely matched the amino acid sequence of LSADH (Leifsonia sp. strain S749, a styrene-tolerant strain isolated from soil in Japan) (24) or LNADH (L. naganoensis, isolated from soil in Japan) (37) (Table 2), suggesting that the genes of both strains either fit the primer set very well or are major Leifsonia habitants in soil environments in Japan. The results also indicated that the origin of the metagenome is the key to obtaining novel ADH genes. We presumed that metagenomic DNA isolated from general soil environments includes DNA from dominant microorganisms, including the genus Leifsonia, which could prevent the amplification of diverse adh genes, whereas special or extreme environments (such as bark compost fermented at high temperature) could lead to the amplification of diverse target genes. The fact that HLADH-027 to HLADH-030 were very similar to the SDR from Sphaerobacter thermophilus, a thermophilic bacterium, supports this theory. Figure 2 displays the phylogenetic analysis of the gene products with known SDRs. Alignment analysis of the HLADHs and LSADH is shown in Fig. 3. These data showed that our approach could cover a wide range of SDR genes, including lsadh and its related homologs. Of course, our library did not necessarily guarantee the isolation of complete adh genes even if they were functional, because the obtained genes were chimeric with lsadh at the terminal regions; thus, they were artificial. However, this gene-specific amplification approach using metagenomes offers an excellent opportunity to identify novel subfamilies of enzyme genes that were previously unknown and uncharacterized. In our approach, the GAM technique attained a high efficiency to obtain target genes; 1,200 clones in 2,018 colonies (∼60%) were adh positive. On the contrary, previous reports with function-based screening indicate a low hit rate (<1.2%) for the identification of target activity from a metagenomic library (3, 4). Thus, the efficiency of our approach to find target activity in an Hladh library is much higher than that in previous studies. Recently, some sequence homology-based metagenomic approaches have been used to elucidate the abundance of novel tfdA-like (dioxygenase) genes (38) and aromatic dioxygenase genes (39) in soil and dmdA (dimethylsulfoniopropionate demethylase) genes in marine environments (40) by using suitable primer sets for GAM. Similarly, the high efficiency of our approach is due to the excellent PCR amplification of target genes from a suitable metagenome-containing sample, such as bark compost fermented at a high temperature. The adoption of a fusion technique with a high-level expression vector for the target gene (pKELA-del for lsadh), which can avoid the ambiguous expression of the gene, should also contribute to our high efficiency. Our approach is based on sequence homology, but it is incorporated with the technique of function-based screening. Using the data presented in Table 2, we estimate that 200 or more different adh genes could be discovered if all 1,200-positive clones were sequenced.

TABLE 2.

Numbers of adh amplicons from different metagenomes and their gene analysis data

Origin of metagenome (no. of samples)	No. of genes sequenced/no. of colonies	No. of genes different from lsadh/no. of duplicates
Soil (12)	27/185	4/23
Farm compost (3)	34/495	7/27
Bark compost (5)	162/1,338	29/133
Total (20)	223/2,018	40/183

TABLE 3.

BLASTP analysis of isolated Hladh genes from metagenomes

Clone no.	Description of known enzyme	Amino acid identity to known enzyme (%)	Amino acid identity to LSADH (%)
001–011	Short-chain alcohol dehydrogenase (Leifsonia sp. strain S749)	98–99	98–99
014–016	Short-chain alcohol dehydrogenase (Leifsonia sp. strain S749)	73–75	73–75
012	2,5-Dichloro-2,5-cyclohexadiene-1,4-diol dehydrogenase (Paenibacillus sp. strain HGF7)	52	52
013	SDR (Truepera radiovictrix DSM 17093)	55–64	55–63
017–019
021–022
034–038
020	SDR (Pedobacter sp. BAL39)	55	55
023	Oxidoreductase, SDR family protein (deltaproteobacterium NaphS2)	49	50
024, 025	SDR (Chelativorans sp. BNC1)	51–52	50–52
026	Putative oxidoreductase, SDR family (Variovorax paradoxus B4)	55	40
027–030	SDR (Sphaerobacter thermophilus DSM 20745)	91–96	44
031	SDR (Desulfatibacillum alkenivorans AK-01)	57	36
032	SDR (Thermobaculum terrenum ATCC BAA-798)	53	43
033, 039, 040	Molybdopterin molybdochelatase (Desulfomonile tiedjei DSM 6799)	48	8–17

FIG 2.

Phylogenetic analysis of HLADHs (boldface red letters), including the chimeric parts of LSADH with known SDRs from various microorganisms with their accession numbers. The phylogenetic tree was constructed using Kimura's method with CLUSTALW and sequences in the DNA Data Bank of Japan. HBADH-1 and HPADH-24 are gene products isolated from metagenomes (17). The scale bar represents the calculated nucleotide substitution ratio.

FIG 3.

Amino acid sequence alignment of HLADHs with LSADH and LPADH. LSADH, accession no. AB213459; LPADH, AB917070; HLADH-012, AB916616; HLADH-014, AB916611; HLADH-021, AB916614; and HLADH-027, AB916626. Alignment was performed using CLUSTALW software. Identical residues are shown by white letters on a black background. A putative coenzyme-binding region is shown inside a dotted square. The amino acid residues considered important for enzymatic activity are marked with an asterisk. The coenzyme-interacting Asp39 residue for NAD⁺/NADH-dependent enzymes is boxed by a solid line. The closed arrows indicate the chimeric regions of LSADH, and the open arrows indicate the binding sites of the primers.

Screening of suitable ADHs for biocatalysis from the isolated adh genes.

As confirmed by the primary screening, all isolated HLADHs were active with 1,1-dichloroacetone, and most HLADHs, including HLADH-001 to HLADH-025, HLADH-27 to HLADH-30, and HLADH-34 to HLADH-38, also were active with PTK. Therefore, the activities of HLADHs for some ketones, including 1-Boc-pyrrolidinone, 2′-chloroacetophenone, 3-quinuclidinone, and ethyl benzoylformate, which cannot be converted easily to anti-Prelog chiral alcohols with high yield or enantioselectivity by general ADHs, including LSADH, were evaluated for their enzymatic function. PTK and acetophenone were used as positive controls, and crude enzymes obtained from the E. coli libraries were used for screening. HLADH-012, HLADH-013, and HLADH-021 showed relatively high levels of activity toward these ketones compared to LSADH, and HLADH-021 possessed wide substrate specificity and demonstrated activity toward all ketones tested (Fig. 4A). Moreover, the results indicated there were large differences in the enzymatic properties of each enzyme, such as activity and substrate specificity, suggesting that examination of the amplicons from the metagenomic E. coli library can be used to obtain an enzyme catalyst imparting the desired substrate specificity.

FIG 4.

Substrate specificity (A) and organic solvent tolerance (B) of HLADHs. Crude enzyme solutions prepared from recombinant E. coli cultures were used for each assay. The bar indicates SD from 3 measurements. (A) All substrate concentrations were 10 mM, and activity for PKT was defined as 100%. (B) Crude enzymes were treated with each organic solvent (50% [vol/vol]) and were shaken for 30 min at room temperature before the remaining activity was measured.

The high tolerance of enzymes to organic solvents is a desirable function when they are to be applied to organic synthesis as biocatalysts under harsh conditions. From this point of view, the crude enzymes obtained from the libraries were subjected to tolerance testing with 3 organic solvents possessing different logP_ow ratios (41), in which P_ow is defined as the ratio of the equilibrium concentrations of a dissolved substance in an n-octanol and water two-phase system: n-octane (logP_ow, 4.5), butyl acetate (1.7), and 2-methyltetrahydrofuran (1.0). The crude enzyme solution was treated with 50% (vol/vol) organic solvent and shaking for 30 min at room temperature, and the remaining activity was measured. HLADH-012 and HLADH-037 were highly tolerant of all organic solvents tested (Fig. 4B). Conversely, LSADH and LPADH readily lost their activity. Generally, 2-methyltetrahydrofuran greatly affects enzymes because of its low logP_ow, that is, its high polarity; however, HLADH-012 showed relatively high tolerance to this solvent. The strong inhibitory effect of n-octane on most of the enzymes tested could not be explained despite its medium-range logP_ow. Purified HLADH-012 and HLADH-021 were later subjected to the same test and almost the same results were obtained, which confirmed that these enzymes possess tolerance in the absence of concomitant proteins in a crude solution.

We successfully obtained these superior enzymes from our limited library; HLADH-021 possessed wide substrate specificity, and HLADH-021/-037 demonstrated organic solvent tolerance. The results clearly indicated that the S-GAM technique is very efficient and useful for obtaining a target enzyme from a metagenome.

Purification and characterization of HLADH-012 and HLADH-021.

Preliminary experiments showed that both enzymes were inactive when 6×His tag was added to their N termini. Therefore, recombinant HLADH-012 and HLADH-021 were purified without the addition of a specific tag. HLADH-012 was purified 11.8-fold, to homogeneity, from the cell extract of a 200-ml culture of recombinant E. coli cells (Table 4). Purified HLADH-012 produced 9.0 U/mg of protein when PTK was the substrate. HLADH-021 was purified 13.0-fold, to homogeneity, from the cell extract of a 100-ml culture by sequential column chromatographic steps. Purified HLADH-021 showed relatively high activity, producing 122.3 U/mg of protein when PTK was the substrate. The purity of both enzymes was evaluated by SDS-PAGE (see Fig. S1 in the supplemental material) and analytical HPLC using a TSK gel 3000SW_XL column. Both enzyme preparations appeared to be almost pure.

TABLE 4.

Purification of recombinant HLADH-012 and HLADH-021

Step	Total protein (mg)	Total activity (U)	Sp act (U/mg)	Yield (%)	Purification (fold)
HLADH-012 (200-ml culture)
Cell extract	165.2	125.0	0.76	100	1
Butyl-Toyopearl	2.14	7.1	3.3	5.7	4.5
Resource Q	0.20	1.8	9.0	1.4	11.8
HLADH-021 (100-ml culture)
Cell extract	67.0	630.1	9.40	100	1
Butyl-Toyopearl	2.05	45.3	22.1	7.2	2.3
DEAE-Toyopearl	0.13	15.9	122.3	2.5	13.0

Analytical HPLC yielded molecular masses of 121 kDa for HLADH-012 and 120 kDa for HLADH-021. The theoretical subunit molecular mass was 26,191 Da for HLADH-012 and 26,291 Da for HLADH-021. Thus, like LSADH, both enzymes were tetramers of identical subunits.

The K_m values of both enzymes in reductive reactions with PTK and NADH at pH 7.0 were calculated by Lineweaver-Burk plots at 0.89 ± 0.14 mM and 0.14 ± 0.002 mM for HLADH-012 and 1.4 ± 0.3 mM and 0.027 ± 0.009 mM for HLADH-021, respectively. Neither enzyme utilized NAPDH as a coenzyme. In oxidative reactions, the K_m values of both enzymes with 2-propanol and NAD⁺ at pH 7.0 were calculated as 73.2 ± 7.8 mM and 0.77 ± 0.09 mM for HLADH-012 and 1.9 ± 0.3 mM and 0.87 ± 0.09 mM for HLADH-021. Notably, the high k_cat value (345 ± 49 s⁻¹) of HLADH-021 for PTK indicated the potential of this enzyme as a biocatalyst.

The effect of pH on the activity of both enzymes is shown in Fig. S2 in the supplemental material. HLADH-012 showed maximum activity at pH 6.0 in the reductive reaction and at pH 8.0 in the oxidative reaction. The pH profile of HLADH-021 was similar to that of HLADH-012; however, HLADH-012 showed higher thermal stability than HLADH-021. The enzymatic properties of HLADH-012 and HLADH-021 are summarized in Table 3.

Production of anti-Prelog chiral alcohols using E. coli whole-cell system expressing HLADH-012 or HLADH-021.

HLADH-012 and HLADH-021 were evaluated as potential biocatalysts. Self-regeneration of NADH with 2-propanol as a hydrogen donor, coenzyme and substrate specificity, activity and stereoselectivity, and robustness during the reaction conditions were assessed (Tables 5 and 6). In addition, an E. coli whole-cell system with 5% (vol/vol) 2-propanol as a hydrogen donor was used to produce anti-Prelog chiral alcohols from some of the ketones, and the enantiomeric excess (e.e.) of the alcohols produced was determined (Table 6). It was confirmed that the production levels of chiral alcohols were relatively high for HLADH-021, although the reactions were not optimized. Moreover, the e.e. of the chiral alcohols produced was excellent for products from 2′-chloroacetophenone, 3-quinuclidinone, and ethyl benzoylformate. Interestingly, the stereoselectivity of HLADH-021 toward 1-Boc-3-pyrrolidinone and ethyl benzoylformate was contrary to that of LSADH. As expected, the performance of HLADH-012 in the production of various chiral alcohols was not outstanding, because it was selected for its high tolerance to organic solvents as an indicator.

TABLE 5.

Properties of purified HLADH-012, HLADH-021, and LSADH

Enzyme	Molecular massb (kDa)	Subunit structureb (kDa)	pIb	Coenzyme	Kmc (mM)				kcatd (s−1)		Optimum pH (reduction/oxidation)	Thermal stabilitye (°C)
					PTK (reduction)	NADH	2-Propanol (oxidation)	NAD+	PTK	2-Propanol
HLADH-012	105	Homotetramer (26.2)	4.89	NAD+/NADH	0.89 ± 0.14	0.14 ± 0.002	73.2 ± 7.8	0.77 ± 0.09	21 ± 0.3	35 ± 4	6.0/8.0	<55
HLADH-021	105	Homotetramer (26.2)	4.75	NAD+/NADH	1.4 ± 0.3	0.027 ± 0.009	1.9 ± 0.3	0.87 ± 0.09	345 ± 49	47 ± 6	6.0/8.0	<40
LSADHa	100	Homotetramer (25.0)	4.58	NAD+/NADH	13.6	0.048	57.5	0.12	127	43	6.0/9.5

^aData are from references 21, 22, and 24.

^bTheoretical calculation derived from study of amino acid sequences.

^cThe errors indicate the SD from 3 measurements.

^dValue for 1 mol enzyme consisting of 4 subunits.

^eTemperature indicating more than 70% of the original activity after treatment at each temperature for 30 min at pH 7.0.

TABLE 6.

Biocatalytic properties of HLADH-012 and HLADH-021 compared to those of LSADH for producing anti-Prelog chiral alcohols

^aData from references 23 and 24.

^bND, not determined.

^cThe errors indicate SD from 3 measurements.

These results suggested that the S-GAM technique is quite efficient and useful for identifying suitable biocatalysts with superior functions. It would be easy to obtain additional novel adh genes by continuing this approach for the metagenomes of bark compost. Thus, a suitable GAM technique offers the opportunity to make a useful collection of desirable genes that exist in nature. Of course, the S-GAM technique could be applied to other enzyme genes and is a promising approach for enzyme engineering and biocatalysis.