The Combination of Functional Metagenomics and an Oil-Fed Enrichment Strategy Revealed the Phylogenetic Diversity of Lipolytic Bacteria Overlooked by the Cultivation-Based Method

Metagenomic screening and conventional cultivation have been used to exploit microbial lipolytic enzymes in nature. We used an indigenous forest soil (NS) and oil-fed enriched soil (OS) as microbial and genetic resources. Thirty-four strains (17 each) of lipolytic bacteria were isolated from the NS and OS microcosms. These isolates were classified into the (sub)phyla Betaproteobacteria, Gammaproteobacteria, Firmicutes, and Actinobacteria, all of which are known to be the main microbial resources of commercially available lipolytic enzymes. Seven and 39 lipolytic enzymes were successfully retrieved from the metagenomic libraries of the NS and OS microcosms, respectively. The screening efficiency (a ratio of positive lipolytic clones to the total number of environmental clones) was markedly higher in the OS microcosm than in the NS microcosm. Moreover, metagenomic clones encoding the lipolytic enzymes associated with Alphaproteobacteria, Deltaproteobacteria, Acidobacteria, Armatimonadetes, and Planctomycetes and hitherto-uncultivated microbes were recovered from these libraries. The results of the present study indicate that functional metagenomics can be effectively used to capture as yet undiscovered lipolytic enzymes that have eluded the cultivation-based method, and these combined approaches may be able to provide an overview of lipolytic organisms potentially present in nature.

Lipolytic enzymes, including esterases (EC 3.1.1.1) and lipases (EC 3.1.1.3), are used in food, chemical, and pharmaceutical industries as well as waste treatment and biofuel production (19,20,22,24). Commercially available bacterial lipolytic enzymes are mainly derived from culturable strains within the phyla Proteobacteria, Firmicutes, and Actinobacteria. For example, lipolytic enzymes purified from the genera Burkholderia and Pseudomonas are widely used as detergent additives and in organic synthesis (19,22,25), and members of the genera Bacillus, Chromobacterium, Staphylococcus, and Streptomyces are listed as popular lipolytic enzyme-producing bacteria (19,20,24). The cultivation-based screening method has been employed as one of the main strategies to obtain such microbial resources of not only lipolytic enzymes, but also other biocatalysts from microbial ecosystems (42,52). However, significant proportions of microorganisms in various environment samples (e.g., soil, sediment, sea water, fresh water, and sludge) have yet to be cultivated and remain functionally unknown (41,44), which implies that there is room for the expansion of microbial resources to obtain better lipolytic enzymes produced by uncultivated microorganisms.
In addition, although the functional metagenomic approach is a powerful screening tool for novel biocatalysts, a large number of metagenomic clones are generally needed to obtain gene fragments encoding target enzymes. To the best of our knowledge, the maximum screening efficiency of lipolytic enzymes (i.e., the ratio of positive lipolytic clones to total metagenomic clones) was shown to be 0.35% in a soil metagenomic library containing 34,560 fosmid clones (15); however, most attempts have reported markedly lower efficiencies (<0.1%) (6,21,27,28,30,35,36,48,53,59). Therefore, a specific biocatalyst-targeted enrichment strategy was applied to improve this efficiency rate (8,12,13,18,33,59). Olive oil (8), mineral oil (12), and crude oil (18) have been used as substrates for the metagenomic screening of lipolytic enzymes. In these studies, two phylogenetically novel lipolytic genes were successfully obtained from metagenomic libraries derived from olive oil-enriched soil and water samples (8). However, no comparative study has examined the frequency of positive clones and phylogenetic diversity of the screened lipolytic enzymes between enriched and non-enriched environmental samples.
In this study, we used comprehensive approaches including both functional metagenomics and a cultivation-dependent method combined with oil-fed enrichment to uncover the entire microbial resources of lipolytic enzymes. Forest soil microbial communities serve important ecological functions, including carbon, nitrogen, and phosphorous cycles, plant diversity, and soil fertility (10,58). Natural (unmanaged) forest soils are considered to contain an arsenal of microbial genetic resources including lipolytic enzymes (27,38). We here used natural forest soil as a representative community and employed four different procedures: (i) functional metagenomic screening from natural forest soil; (ii) cultivation-based screening from natural forest soil; (iii) functional metagenomic screening from oil-enriched soil; (iv) cultivation-based screening from oil-enriched soil.

Materials and Methods
Strains, plasmids, and growth conditions E. coli strain EPI300 (Epicentre, Madison, WI, USA) and the Copy Control pCC1FOS vector (Epicentre) were used to construct metagenomic libraries. The growth conditions of E. coli strains were described previously (32).

Soil sample and enrichment
Soil samples were collected in May 2007 from a forest located in Sapporo, Hokkaido, Japan (43°04' N, 141°30' E). The vegetation type was a cool temperate deciduous broadleaf forest (Quercus crispula Blume, Trillium smallii, Anemone flaccida, Sasa spp., and ferns of the Dryopteridaceae were observed). Pebbles and plant roots were removed from the samples. The moisture content and pH of the soil samples were measured using a previously described method (40). Non-enriched soil samples (hereafter "NS") were immediately subjected to cell counting, DNA extraction, and lipolytic activity measurements. To prepare the oil-enriched soil sample (hereafter "OS"), 500 g (wet wt) of soil was placed into a 2-L conical glass flask with a silicon cap, and incubated at 30°C for 2 months without continuous aeration and shaking. A total of 2.5 mL of olive oil (Wako Pure Chemical Industries, Osaka, Japan) was added to the soil and mixing by a spoon every second week. The moisture content and pH were maintained at the initial states during the enrichment by adding distilled water and sodium bicarbonate, respectively.

Lipolytic activity of soil samples
The lipolytic activities of NS and OS were measured by copper soap colorimetry with small modifications (49). Briefly, 1 g (wet wt) of soil was mixed with 1.25 mL of 2% polyvinyl alcohol:olive oil (3:1, v/v) emulsified solution and 1.25 mL of 200 mM potassium phosphate buffer (pH 6.5). The reaction proceeded at 37°C for 20 min and was stopped by the addition of 5 mL of 200 mM potassium phosphate buffer and 5 mL of acetone. After 2.5 mL of isooctane was added, the reaction mixture was vortexed and boiled for 1 min. The upper isooctane phase containing free fatty acids was collected and mixed with 1 mL copper reagent (49). The resultant mixture was vortexed for 1.5 min and the absorbance of the isooctane phase was measured at 715 nm. One unit of lipase activity was defined as 1 µM free fatty acid released for 1 min. Oleic acid was used as the standard of free fatty acids.

Cell counting and isolation of major lipolytic bacteria
In cell counting and the isolation of lipolytic bacteria, 1 g (wet wt) of a soil sample was sonicated and diluted serially with phosphate-buffered saline (PBS), as described previously (39). Direct total counts of bacteria were measured by the epifluorescence microscopic method described by Yoshida et al. (63). Plate counts of culturable lipolytic bacteria were performed using tributyrin agar plates: per L, 5 g (NH 4 ) 2 SO 4 ; 1 g KH 2 PO 4 ; 1 g K 2 HPO 4 ; 0.1 g MgSO 4 7H 2 O; 5 g tributyrin; 1 g arabic gum (pH 7.0 by NaOH). We initially attempted to prepare the olive oil-containing agar plate for the isolation of lipolytic bacteria and selection of the E. coli transformant; however, it was difficult to prepare a homogeneously diffused olive oil agar plate. Due to this obstacle, a tributyrincontaining agar plate was used for the isolation of lipolytic bacteria from NS and OS samples (and also for the selection of the E. coli transformant as described below). The inoculated plates were incubated at 30°C for 2 months before counting colony-forming units (CFU). In order to isolate the main lipolytic bacteria in NS and OS communities, single halo-forming colonies were picked up randomly from the "countable plates," which were inoculated with 100 µL of 10 −3 -10 −4 diluted samples and contained 3-10 colonies with clear single halos, and subjected to a standard purification procedure with the repeated streaking of tributyrin agar plates. Phylogenetic analysis of the isolates was performed by partial (ca. 500 bp) 16S rRNA gene sequencing according to a previously described method (40).

Metagenomic screening of the lipolytic fosmid clone
DNA extraction, purification, manipulation, and metagenomic library construction were performed using the methods described by Kimura et al. (32). To select lipolytic clones from the constructed fosmid libraries, metagenomic clones in E. coli EPI300 were cultured on LB agar supplemented with 12.5 µg mL −1 chloramphenicol for 3 d at 37°C. The number of clones per plate was adjusted to approximately 1,000 by the dilution of library stocks, and the plates were incubated overnight at 37°C and stored as a "master plate". The colonies growing on the master plates were replicated on a LB agar plate supplemented with 0.5% tributyrin, 0.1% arabic gum, and chloramphenicol (7). The replicated plates were incubated for 3 d at 37°C, and the metagenomic clones that produced single halos were isolated. The average insert sizes of cloned lipolytic DNA were estimated by BamH1 digested fragment analyses of 16 randomly selected clones (32).

In vitro mutagenesis and sequencing of metagenomic clones
In vitro mutagenesis of the selected lipolytic clones was performed by using the EZ::TN KAN-2 transposon kit (Epicentre) according to a previously described method (32). Briefly, positive E. coli EPI300 cells were incubated in 10 mL LB broth supplemented with 12.5 µg mL −1 chloramphenicol at 37°C for 16 h. After incubation, the cells were collected by centrifugation (6,300×g, at 4°C for 5 min). Fosmid DNA was purified from the cells with a QIAprep Spin Miniprep Kit (Qiagen, Hilden, Germany). Regarding transposon-based mutagenesis, 2 µL fosmid DNA (100 ng µL −1 ) was mixed with 2 µL EZ-Tn5 Transposon containing a kanamycinresistant gene (0.002 pmol µL −1 ), 0.5 µL 10× Reaction Buffer, and 0.5 µL EZ-Tn5 Transposase (1U µL −1 ), and then incubated at 37°C for 2 h. The reaction was stopped by adding 1 µL 10× Stop Solution, and incubated at 60°C for 10 min. The mutated fosmid DNA was mixed with E. coli EPI300 cells in 0.2 cm-gap Gene Pulser/ MicroPulser Cuvettes (Bio-Rad, Hercules, CA, USA), and electro-porated at 2.0 kV with a Gene Pulser (Bio-Rad). After electroporation, the cells were suspended in 815 µL of LB broth and allowed to grow at 37°C for 1 h. A portion of the cell suspension (ca. 100 µL) was cultured on the LB agar plate supplemented with 12.5 µg mL −1 chloramphenicol and 50 µg mL −1 kanamycin at 37°C for 1 d. The colonies growing on the plates were replicated on the LB agar plate supplemented with 0.5% tributyrin, 0.1% arabic gum, 12.5 µg mL −1 chloramphenicol, and 50 µg mL −1 kanamycin. E. coli transformants without clear halos represented on LB-tributyrin plate were isolated, and fosmid DNA was purified with a QIAprep Spin Miniprep Kit (Qiagen). The transposon insertion sites were sequenced with the primers KAN-2 FR-1 and RP-1 (Epicentre), a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Carlsbad, CA, USA), and a 3130xl Genetic Analyzer (Applied Biosystems). DNA sequences were analyzed with BLAST and ORF finder programs provided by the National Center for Biotechnology Information (51). The alignment of the translated amino acid sequences and construction of a distance matrix tree on the basis of the neighbor-joining method (50) were performed with the MEGA software (55). According to previous studies on the phylogeny of lipolytic enzymes, the phylogenetic affiliations of the lipolytic enzymes were relatively consistent with the 16S rRNAbased phylogeny (2,8,24,27,38). Therefore, we compared the phylogenetic diversity of lipolytic bacteria based on the 16S rRNA gene sequences of the culturable isolates and amino acid sequences of the metagenomic clones.

Nucleotide sequence accession numbers
The nucleotide sequences of the lipolytic enzyme genes and 16S rRNA genes presented in this study have been deposited under DDBJ/EMBL/GenBank accession numbers AB811356 to AB811435.

Effect of the oil-fed enrichment strategy on cell counting and soil lipase activity
The moisture content and pH of NS was 7.1 and 60%, respectively. During the enrichment period, moisture content and pH values were maintained at levels of 6.5-7.5 and 55-65%, respectively. The plate counts of lipolytic bacteria in NS and OS accounted for 0.031% and 0.57% of the direct total counts, respectively, indicating that the number of lipolytic bacteria was 18-fold higher in the OS microcosm than in NS microcosm (Table 1). Soil lipolytic activity also increased 4.6-fold after the oil-fed enrichment strategy (Table 1), which demonstrated that lipolytic microorganisms were enriched in the OS microcosm. The bacterial community structures of NS and OS were analyzed by 16S rRNA gene cloning (see Supplemental information). The results obtained showed that the clonal sequences of both microcosms were assigned to eight bacterial phyla (i.e., Proteobacteria, Actinobacteria, Acidobacteria, Bacteroidetes, Verrucomicrobia, Planctomycetes, Chloroflexi, and Nitrospirae) and hitherto-uncultured bacterial groups (i.e., candidate phyla SPAM and WS3) ( Table S1). The Chao1 diversity index based on phylotypes having >97% nucleotide sequence identity was reduced to one-quarter after the oil-fed enrichment.

Phylogenetic affiliations of culturable lipolytic isolates
We isolated the respective 17 strains from the tributyrin agar plates used for plate counting of lipolytic bacteria in the NS and OS samples, and these were identified by partial 16S rRNA gene sequencing ( Table 2). All of the isolates were identified as members of the (sub)phyla Betaproteobacteria, Gammaproteobacteria, Firmicutes, and Actinobacteria. The majority of genera to which the isolates from the NS were assigned were associated with Bacillus (7 out of 17 strains), Lysinibacillus (4 out of 17 strains), and Pseudomonas (4 out of 17 strains), and those associated with the genera Burkholderia (1 out of 17 strains) and Rhodococcus (1 out of 17 strains) were found to be minor constituents. After the enrichment procedure, the lipolytic bacteria assigned to the genera Burkholderia (12 out of 17 strains) and Bacillus (3 out of 17 strains) dominated in the OS microcosm, except for one strain of Mitsuaria (1 out of 17 strains) and one strain of Curtobacterium (1 out of 17 strains) ( Table 2). These results indicated that the oil-fed enrichment strategy selected particular lipolytic bacteria and excluded other indigenous culturable lipolytic bacteria.

Screening of lipolytic clones from the metagenomic library
Two fosmid libraries were constructed from the total genomic DNA extracted from NS and OS. The average insert sizes of lipolytic clones obtained from NS and OS were estimated to be 28.1 and 28.6 kbp, amounting to approximately 2.3 and 3.7 Gbp, respectively (Table 1). Among the clones retrieved from NS and OS, 7 and 49 clones showed a clear halo on the LB-tributyrin agar plate. A 4.3-fold increase was observed in the frequency (a ratio of positive lipolytic clones to the total number of clones) after the oil-fed enrichment procedure, suggesting that the oil-fed enrichment strategy was effective at capturing lipolytic metagenomic clones. A total of 56 clones were subjected to subcloning and screening for lipolytic activity in the secondary libraries. All of the 7 clones retrieved from NS were successfully subcloned and sequenced. Of the 49 clones from the OS microcosm library, 39 fosmid clones whose lipolytic activity was confirmed were successfully obtained (Table 3).

Phylogenetic affiliations of lipolytic enzymes from metagenomic clones
The deduced amino acid sequences of all lipolytic enzymes from the NS and OS metagenomic libraries had high identities with those of the conserved domains of α/β hydrolase, β-lactamase-type hydrolase, and lipase/esterase ( Table 3). The majority of lipolytic enzymes from NS was assigned to Alphaproteobacteria, Planctomycetes, and Acidobacteria, and was associated with the families IV, V, and VIII (Fig.  S1). On the other hand, those from the OS library were mainly assigned to Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, Deltaproteobacteria, Acidobacteria, Armatimonadetes, and Actinobacteria, and were associated with the families II, IV, V, and VIII, and the feruloyl esterase group (Fig. S1). The 16S rRNA gene clones classified into these taxa were detected in both the NS and OS communities, except for the members of Armatimonadetes (Table S1). In addition to the previously known taxa, the functional metagenomic approach enabled us to capture lipolytic enzymes associated with uncultured microorganisms ( Table  3). The phylogenetic tree based on amino acid sequences indicated that clones NSm06 and NSm07 obtained from NS were clustered into lipase family V, and had a similar identity to the soil-derived metagenomic clone pUlp286 (53.8% and 50.5% amino acid identities, respectively) (Fig. 1). Within the OS library, a total of 14 clones showed the highest identities to the lipolytic enzymes of uncultured microbes (Table  3), and were divided into three clusters: lipase family IV, the feruloyl esterase group, and a novel cluster (Fig. 1). Twelve clones (OSm27-35 and OSm37-39) were assigned to lipase family IV, and all of these clones had significant identities (61.7%-84.8% amino acid identities) to the lipolytic genes obtained from soil and marine sediment samples. The OSm36 clone had 45.1% amino acid identity with the feruloyl esterase gene of the uncultured bacterium FeKT1 obtained from the marine sediment metagenomic library. In addition, the OSm26clone showed 50.0% amino acid identity to the lipolytic gene found in the groundwater sample.

Discussion
As reported here, our comparative analyses with two screening approaches (i.e., functional metagenomics and a cultivation-based method) combined with the oil-fed enrichment strategy revealed the entire picture of the phylogenetic diversity of lipolytic bacteria in the forest soil environment, as summarized in a Venn diagram (Fig. 2). One of the most important results was that the functional metagenomic approach detected significant amounts of lipolytic enzymes associated with the phylogenetic groups that were absent in the catalogs of culturable isolates in both NS and OS microcosms; i.e., the (sub)phyla Alphaproteobacteria, Deltaproteobacteria, Acidobacteria, Armatimonadetes, and Planctomycetes. Within the class Alphaproteobacteria, 2 and 12 lipolytic enzymes were identified in NS and OS, respectively. One metagenomic clone associated with Deltaproteobacteria and Armatimonadetes was retrieved from the OS library only. Within the phylum Acidobacteria, 1 and 3 lipolytic enzymes were retrieved from NS and OS, respectively. Two lipolytic enzymes showed the highest identity to those of the phylum Planctomycetes in the NS library only. Although several lipolytic enzymes associated with Alphaproteobacteria, Deltaproteobacteria, Acidobacteria, and Planctomycetes have been identified in soil metagenomic libraries (6,12,16,35,38), to the best of our knowledge, there have been no commercially available lipolytic enzymes produced by the culturable strains of these taxa (19,22). Another important result in our study is that both NS and OS libraries contained a significant proportion of lipolytic enzymes associated with uncultured microorganisms. Based on the phylogenetic analysis, such lipolytic enzymes showed significant identity to the metagenomic clones obtained from soil (28,36,38,46,53), marine sediment (23,26), tidal flat sediment (34), and groundwater (62). These results strongly suggest that functional metagenomics combined with the oil-fed enrichment strategy contributed to uncover the phylogenetic diversity of lipolytic bacteria that had been overlooked by the cultivation-based screening technique. All strains of culturable lipolytic isolates were phylogenetically assigned to three phyla and were further identified as seven genera: Burkholderia, Mitsuaria, and Pseudomonas of the phylum Proteobacteria; Bacillus and Lysinibacillus of the phylum Firmicutes; and Rhodococcus and Curtobacterium of the phylum Actinobacteria. To date, many lipolytic enzymes have been purified from members of the strains of Bacillus, Burkholderia, and Pseudomonas (19,20,24,57). The closest relatives of the isolates classified into the other four genera were also shown to be lipolytic enzymeproducing bacteria (1, 3, 4, 14, 29, 54). Thus, all culturable  (2,8,23,35,39): II, IV, V, and VIII, number of the family; FE, feruloyl esterase group; NC, novel cluster (shown in Fig. 1 and Fig. S1). lipolytic strains were phylogenetically associated with previously characterized lipolytic bacteria. Further studies are needed to compare the biochemical functions (e.g., activity and substrate specificity) of lipolytic enzymes obtained from the culturable isolates and metagenomic libraries.
We did not detect any lipolytic enzymes associated with the phylum Firmicutes from either the NS or OS metagenomic library in spite of their presence in the culturable lipolytic bacterial isolates. In addition, the 16S rRNA gene clones classified into this phylum were also undetectable (Table S1). These results indicate that culture conditions needed to isolate the lipolytic bacteria used in this study were suitable for the proliferation of Firmicutes, especially the genera Bacillus and Lysinibacillus. This may also indicate that the metagenomic screening procedure used in this study had potential biases against members of the phylum Firmicutes. To overcome this, there is still room for further improvements in the functional metagenomic approach. For example, the host strain may be an important factor in this respect. Previous metagenomic studies reported the use of Bacillus subtilis (5), Pseudomonas putida (43), Rhizobium leguminosarum (61), and Streptomyces lividans (11) as host strains as an alternative to E. coli to explore novel biocatalysts. Further studies using a functional metagenomic approach with multiple host strains may provide a deeper insight into the phylogenetic diversity of lipolytic microbes in forest soil ecosystems.