OUP user menu

Chitinase genes revealed and compared in bacterial isolates, DNA extracts and a metagenomic library from a phytopathogen-suppressive soil

Karin Hjort, Maria Bergström, Modupe F. Adesina, Janet K. Jansson, Kornelia Smalla, Sara Sjöling
DOI: http://dx.doi.org/10.1111/j.1574-6941.2009.00801.x 197-207 First published online: 24 December 2009

Abstract

Soil that is suppressive to disease caused by fungal pathogens is an interesting source to target for novel chitinases that might be contributing towards disease suppression. In this study, we screened for chitinase genes, in a phytopathogen-suppressive soil in three ways: (1) from a metagenomic library constructed from microbial cells extracted from soil, (2) from directly extracted DNA and (3) from bacterial isolates with antifungal and chitinase activities. Terminal restriction fragment length polymorphism (T-RFLP) of chitinase genes revealed differences in amplified chitinase genes from the metagenomic library and the directly extracted DNA, but approximately 40% of the identified chitinase terminal restriction fragments (TRFs) were found in both sources. All of the chitinase TRFs from the isolates were matched to TRFs in the directly extracted DNA and the metagenomic library. The most abundant chitinase TRF in the soil DNA and the metagenomic library corresponded to the TRF103 of the isolate Streptomyces mutomycini and/or Streptomyces clavifer. There were good matches between T-RFLP profiles of chitinase gene fragments obtained from different sources of DNA. However, there were also differences in both the chitinase and the 16S rRNA gene T-RFLP patterns depending on the source of DNA, emphasizing the lack of complete coverage of the gene diversity by any of the approaches used.

Keywords
  • metagenomic library
  • chitinase
  • terminal restriction fragment length polymorphism (T-RFLP)
  • Streptomycetes
  • suppressive soil

Introduction

Exploitation of the previously hidden members of the soil microbiota is a focus of current research interest in the hunt for novel bioactive molecules beneficial in medicine, industry and agriculture; for reviews, see (Handelsman, 2004; Sjöling et al., 2007). For example, some soils are naturally suppressive towards plant diseases and microorganisms in these soils are often proposed to be the cause of suppressiveness (Borneman & Becker, 2007; Steinberg et al., 2007). Therefore, there is considerable commercial and research interest in isolation of the microorganisms or the bioactive compounds that might contribute to disease suppression. Biological control methods have been recommended to replace chemical control methods because these are more economical and environmentally sustainable (Herrera-Estrella & Chet, 1999; Fravel, 2005). One example of a biological mechanism for suppression of fungal pathogens in suppressive soils is that of microbial chitinase activity (Chernin et al., 1997; Downing & Thomson, 2000; Kobayashi et al., 2002), and chitinases (or chitinase-producing microorganisms) have a potential application for biocontrol of plant diseases.

Chitinases belong to the group of glycosyl hydrolases, either family 18 or 19. Family 18 is further subdivided into A, B or C based on the amino acid sequence similarities of the catalytic domains (Henrissat & Bairoch, 1993; Karlsson & Stenlid, 2009). Chitinases hydrolyse chitin, which is otherwise rather resistant to degradation, to enable utilization of the end products as an energy, carbon and/or nitrogen source (Gooday, 1990; Williamson et al., 2000; Lindahl & Finlay, 2006). This is an important step in the biogeochemical cycling of carbon and nitrogen in the environment. In soil, chitin is widely distributed within insect bodies and fungal cell walls (Gooday, 1990). Conventional molecular screening approaches have identified chitinase genes within aquatic (Ramaiah et al., 2000; Hobel et al., 2005) and soil environments (Williamson et al., 2000; Metcalfe et al., 2002; LeCleir et al., 2004; Uchiyama & Watanabe, 2006). However, only a few studies have used a metagenomic approach to identify chitinase genes (Cottrell et al., 1999; LeCleir et al., 2007).

Metagenomics offers access to functional genes in uncultured representatives of the microbiota and has previously facilitated the characterization of large genomic regions or even complete genomes of uncultured bacteria (Rondon et al., 2000; Gillespie et al., 2002; Tringe et al., 2005) and access to novel bioactive products (Hårdeman & Sjöling, 2007; Sjöling et al., 2007). Soil metagenomics typically involves the isolation and purification of high-molecular-weight (HMW) DNA, followed by cloning into a library and sequencing, or alternatively, direct sequencing using second-generation sequencing platforms. A clone library has the advantages of facilitating functional, expression-based screening and sequencing of long contigs (Rondon et al., 2000; Sjöling et al., 2007).

We have previously (Hjort et al., 2007) used the molecular fingerprinting techniques, terminal restriction fragment length polymorphism (T-RFLP) and denaturing gradient gel electrophoresis to study changes in bacterial communities in response to chitin amendment in a soil reported to be suppressive towards clubroot disease caused by Plasmodiophora (Worku & Gerhardson, 1996). We found that after chitin addition to the soil, the relative abundances of known chitin-degrading genera such as Oerskovia, Kitasatospora and Streptomyces species increased dramatically and became dominant both in the total and in the actively growing bacterial community. Also, a number of isolates with antifungal and chitinase activity were obtained from this soil (Adesina et al., 2007).

The aim of this study was to screen for chitinase genes in the suppressive soil using a combination of molecular approaches. To begin with, we searched for chitinase genes in the bacterial isolates obtained previously from the soil with known chitinase and antifungal activities (Adesina et al., 2007). We also used T-RFLP (Liu et al., 1997; Braker et al., 2001) to screen DNA directly extracted from the same soil. Finally, we prepared a fosmid library and screened the resulting soil metagenome for chitinase genes. This study is the first to compare these different sources of DNA from the same soil. We found surprisingly good agreement between the different sources of material for the dominant chitinase genes detected, but some differences were also found, indicating that specific biases need to be taken into account for each method. These results should lay the groundwork for making informed decisions about the appropriate source material to use in other studies that aim to screen for specific functional genes in environmental samples.

Materials and methods

Soil sampling

The soil (clay loam, pH 6.9, and an organic carbon content of 1.48%) was sampled from a field at the Swedish University of Agricultural Sciences in Uppsala, Sweden, in October 2003 and July 2004 as described previously (Hjort et al., 2007). The field was previously characterized as suppressive to clubroot disease of cabbage (Worku & Gerhardson, 1996). Twenty soil core samples from the top 20 cm were randomly collected from four sites (each 5 m × 5 m) using a core sampling device. All 20 core samples from an individual site were mixed to one composite sample, sieved through a 4-mm mesh and HMW DNA was immediately extracted from the soil as described below.

Metagenomic library construction

The metagenomic library was constructed in fosmids using a modification of the procedure described by Hårdeman & Sjöling (2007). The cells were extracted from 100 g of freshly collected soil as described previously (Gabor et al., 2003) with the following modifications. The soil was mixed with 100 mL 0.5% pyrophosphate buffer, pH 8.0, in a Waring blender (Robert Bosch GmbH, Germany) three times for 30 s, followed by incubation at 4 °C for 30 min. Soil particles were sedimented by low-speed centrifugation at 500 g for 20 min, at 10 °C. The supernatant containing the cells was collected and transferred to a different tube. The soil pellet was resuspended and mixed in 50 mL hexadecyltrimethylammonium bromide (CTAB) buffer, pH 8.5 [100 mM EDTA, 100 mM Tris-HCl, 0.1% sodium dodecyl sulphate (SDS), 1% CTAB], using the blender for 30 s, followed by an additional centrifugation at 500 g for 20 min at 10 °C. This step was repeated. Finally, all the supernatants containing microbial cells extracted from the soil were pooled and cells were collected by centrifugation at 10 000 g for 30 min at 10 °C. The cells were resuspended in 2 mL TE buffer (10 mM Tris-HCl and 1 mM EDTA, pH 8.0). Twelve microlitre proteinase K (20 mg mL−1) and 120 μL 10% SDS were added to lyse the cells during incubation at 37 °C for 60 min. This was followed by the addition of 400 μL 5 M NaCl and 320 μL (10% CTAB, 0.7 M NaCl) and incubation at 65 °C for 10 min. The DNA was recovered by gentle phenol/chloroform/isoamylalcohol extraction (25 : 24 : 1) and precipitated by a 1 : 10 vol. of 3 M sodium acetate and 2.5 vol. of ethanol.

HMW DNA was separated on a 1% low-melting-point agarose gel (GE Healthcare, Sweden) by pulsed-field gel electrophoresis, 20 V × 40 s, at 6 V cm−1, 5–15 s switch at 14 °C for 18 h (CHEF-DR II, Bio-Rad Laboratories, UK). DNA fragments ranging from 25 to 300 kb were excised from the gel and extracted using β-agarase I according to the manufacturer's instructions (New England Biolabs, Ipswich, MA) to avoid shearing. The DNA was gently precipitated with ethanol as above. Approximately 500 ng of DNA was cloned into a CopyControl Fosmid vector (Epicentre, Madison, WI) according to the manufacturer's instructions. Fosmid clones were picked into 96-well microtitre plates and grown in Luria–Bertani medium, supplemented with 12.5 μg mL−1 chloramphenicol and 7% glycerol, overnight at 37 °C. The 7800 clones of the original library, stored at −80 °C, were pooled into one sample, which was used for chitinase screening and 16S rRNA gene analysis.

The average insert size was analysed by randomly selecting 20 clones where the vector was isolated by means of standard alkaline lysis and plasmid mini preparation (Qiagen, Hilden, Germany) and the insert sizes were determined by NotI digestion (Fermentas, Ontario, Canada). The sizes of the inserts were estimated from 1% agarose gels.

Subsequent PCR screenings of the library for the presence of chitinase and 16S rRNA genes were made after extraction of vector DNA from the pooled fosmid library using the plasmid midi prep kit following the manufacturer's instructions (Qiagen).

Direct soil DNA extraction

Triplicate soil DNA samples were directly extracted from 400 mg frozen soil (collected October 2003) by bead beating using the FastPrep for soil kit, Bio101 (Qbiogene Carlsbad, CA), and a FastPrep bead-beating machine (BIO101, Qbiogene) according to the manufacturer's instructions. The extracted DNA had a lower average molecular weight (<20 kb) compared with the HMW DNA prepared for the metagenomic library.

DNA extraction from isolates

Genomic DNA was extracted from 18 bacterial isolates obtained previously from the same soil batch as that used for DNA extractions described above, with demonstrated chitin-degrading capacities based on an agar plate assay (Adesina et al., 2007). The isolates were also previously demonstrated to have antifungal activity towards Rhizoctonia solani and/or Fusarium oxysporum (Adesina et al., 2007). Cells were lysed with 0.1 mm silica beads (Biospec Production Inc., Bartlesville, OK) and two executive bead-beating steps at a speed of 5.5 m s−1 for 45 s each in the Fast prep bead-beating machine, and DNA was extracted using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI), according to the manufacturer's instructions, except for the additional lysis step.

T-RFLP of 16S rRNA and chitinase genes

Partial 16S rRNA genes were amplified in triplicate from DNA (pooled metagenomic library and directly extracted from soil) using bacterial forward primer fD1-FAM (5′-AGAGTTTGATCMTGGCTCAG-3′) 5′ end labelled with 6-FAM (phosphoramidite fluorochrome 5-carboxy-fluorescein) and reverse primer 926r (5′-CCGTCAATTCCTT TRAGTTT-3′) (Weisburg et al., 1991; Muyzer et al., 1995), as described in Edlund et al. (2006). All primers were synthesized by Invitrogen (Carlsbad, CA).

Partial family 18 chitinase genes were amplified in triplicate from each DNA sample of the same source as above and in single amplifications from DNA of bacterial isolates using forward primer ChiA_F2 (5′-CGT GGA CAT CGA CTG GGA RTW YCC-3′) 5′ end labelled with 5′6-FAM and reverse primer ChiA_R2 (5′-CCC AGG CGC CGT AGA RRT CRT ARS WCA-3′) (Hobel et al., 2005). The PCR reactions were set up according to Hobel et al. (2004) with a few modifications: primer concentration 20 pmol, Taq polymerase 2.5 U (GE Healthcare) (Hobel et al., 2004) and the annealing temperature was increased to 47 from 42 °C.

For the analysis of bacterial community structures, duplicate PCR reactions were amplified and pooled from each of the triplicate DNA extracts from a composite soil sample of four sampled sites. The duplicate amplicons were pooled, digested in parallel with HaeIII, HhaI and MspI (GE Healthcare) and analysed by T-RFLP (Cybergen, Huddinge, Sweden) as described by Edlund et al. (2006). The relative abundance of each terminal restriction fragment (TRF) was determined by dividing the area of the electropherogram fluorescent signal for the peak of interest by the total fluorescent signal area of peaks within the following threshold values: lower threshold, 60 bp; upper threshold, 500 bp; and a fluorescent threshold of 50. The TRF value corresponding to Escherichia coli (TRF 498, using Msp1) was excluded because E. coli was the host for the fosmid vector used, and the relative abundances of the remaining peaks were then recalculated for both the soil and the metagenomic library. TRFs were only included in the analyses if they were present in at least two of the three replicates.

For the analysis of chitinase genes, triplicate PCR products were digested in parallel with HaeIII, HhaI and MspI and analysed by T-RFLP (Uppsala Genome Centre, Uppsala, Sweden) as described by Hjort et al. (2007). Threshold values: lower threshold, 60 bp; upper threshold, 245 bp; and a fluorescent threshold of 50 were applied. For assignment of possible chitinase genes, data from all three restriction enzyme digests were combined. The sizes of TRFs from T-RFLP analysis of chitinase genes of 18 bacterial isolates (antifungal and chitin degrading) from the suppressive soil described in the following section were used as references for comparison of TRFs in the T-RFLP analysis of chitinase genes, in DNA from directly extracted soil and DNA from pooled metagenomic library. TRFs were only included in the analyses if they were present in at least two of the three replicates.

Sequencing of 16S rRNA and chitinase genes

16S rRNA genes were amplified from the bacterial isolates with the forward primer 27f (5′-AGA GTT TGA TCM TGG CTC AG-3′) and the reverse primer 1492r (5′-GGY TWC CTT GTT ACG ACT T-3′) using the same PCR conditions as those for T-RFLP according to Hjort et al. (2007). The PCR products were sequenced with the 27f, 1492r and the 1378R (5′-CGG TGT GTA CAA GGC CCG GGA ACG-3′) primers. The partial chitinase gene was amplified with the same primer set as that mentioned above for chitinase T-RFLP, except that the forward primer was unlabelled. The chitinase PCR products from all three sources (directly extracted from soil, pooled metagenomic library and bacterial isolates) were separately purified (according to the manufacturer's instructions, Qiagen), ligated into a pCR 4-TOPO vector (Invitrogen) and transformed into competent cells (TOP10 Chemically Competent E. coli) as described by the manufacturer (Invitrogen). All sequencing was performed at the Uppsala Genome Centre. The cloned partial chitinase genes were sequenced using the T7 primer (5′-TAATACGACTCACTATAGGG-3′). Sequence identities of 16S rRNA and chitinase genes were determined with blast searches in GenBank (NCBI database). The chitinase sequences were aligned with clustalw and a neighbour-joining, best tree was constructed using the software macvector (http://www.macvector.com/index.html).

Results

Soil microbiome metagenomic library

The soil microbiome metagenomic library comprised 7800 fosmid clones with insert sizes ranging between 20 and 40 kb, with an average insert length of 30 kb covering an estimated 230 Mbp, calculated from the average insert sizes of the clones. It has been estimated that 1 g of soil may contain 4000 (Torsvik et al., 1990) to 50 000 species (Roesch et al., 2007). Given an average genome size of at least 3.8 Mb (calculated from 220 fully sequenced bacterial genomes randomly selected from the Genomes OnLine Database), our metagenomic library could cover approximately 0.1–1.5% of the diversity in a typical soil.

16S rRNA gene sequences from bacterial isolates

16S rRNA genes (approximately 1310–1440 nt) of 18 bacterial isolates from the suppressive soil, which were previously shown to have the combined features of antifungal activity and the ability to degrade chitin (Adesina et al., 2007), were PCR amplified and sequenced (Table 1, Fig. 1). Sequence alignment to known sequences in GenBank showed that 11 of the isolates had closest identities to Streptomyces spp. and nine of these isolates (Nos IX–XVII) matched most closely to Streptomyces clavifer and/or Streptomyces mutomycini. The rest of the isolates showed closest matches to Pseudomonas (Nos III and IV), Stenotrophomonas (Nos V, VI and VII), Bacillus pumilus (Nos I) and Brevibacterium antarcticum (No. II).

View this table:
1

16S rRNA and chitinase genes from bacterial isolates with their respective similarities (nucleotide) to sequences in GenBank and TRF sizes of their chitinase genes

IsolateClosest identity 16S rRNA gene(% identity, over nt)Closest identity chitinase gene(% identity, over nt)TRF (HhaI) chitinase gene
IBacillus pumilus (DQ523500)100%, 1434Uncultured bacterium gene for chitinase (AB361986)85%, 166No fragment
IIBrevibacterium antarcticum (AJ577724)99%, 1405Uncultured organism clone ChiCSR29 (AY699345)88%, 23875
IIIPseudomonas sp. (AY599710)100%, 1414No chitinase gene detected with PCR
IVPseudomonas sp. (DQ279324)99%, 1404Uncultured organism clone ChiCSR29 (AY699345)88%, 23775
VStenotrophomonas sp. (AY131216)99%, 1422Myxococcus xanthus USC7-1p (AY033407)86%, 239181
VIStenotrophomonas maltophilia (AJ293470)99%, 1431Myxococcus xanthus USC7-1p (AY033407)97%, 220181
VIIStenotrophomonas maltophilia (AJ293470)99%, 1439Myxococcus xanthus USC7-1p (AY033407)97%, 219181
VIIIStreptomyces viridochromogenes (AB184088)100%, 1353Uncultured bacterium clone control1 (AF455091)87%, 196103
IXStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1353Uncultured bacterium clone control1 (AF455091)88%, 200103
XStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1354Uncultured bacterium clone control1 (AF455091)87%, 208103
XIStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1378Uncultured bacterium clone control1 (AF455091)87%, 16285
XIIStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1354Uncultured bacterium gene for chitinase (AB361986)85%, 198No fragment
XIIIStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1345Uncultured bacterium clone control1 (AF455091)87%, 172103
XIVStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1353Streptomyces sp. An26 (AJ968655)89%, 196No fragment
XVStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1353Uncultured bacterium clone control1 (AF455091)85%, 236103
XVIStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1353Uncultured bacterium clone control1 (AF455091)86%, 199103
XVIIStreptomyces clavifer/S. mutomycini (DQ026670, AB249951, AJ781357)100%, 1377Uncultured bacterium clone control1 (AF455091)85%, 175103
XVIIIStreptomyces viridochromogenes (AB184088)100%, 1352Uncultured bacterium clone control1 (AF455091)86%, 175103
1

Neighbour-joining tree based on clustalw sequence alignment of the partial chitinase genes from clone library analyses of the soil, the pooled metagenomic library and from bacterial isolates. Clones from the pooled soil metagenomic library (blue) and clones from directly extracted soil DNA (red).

Chitinase gene sequences from isolates, the pooled metagenomic library and directly extracted soil DNA

From all isolates, except the strain identified as Pseudomonas sp. (Table 1; No. III), a chitinase-specific PCR product could be amplified using specific bacterial chitinase primers. Most of the amplified PCR products were approximately 240-bp long, except for amplicons of 277 bp from two isolates with closest 16S rRNA gene identities to B. antarcticum (No. II) and Pseudomonas sp. (No. IV). Not unexpectedly, all chitinase gene sequences from Streptomycetes isolates showed closest identities to chitinase genes from Streptomycetes. Phylogenetic analyses showed that chitinase genes from the S. mutomycini and/or S. clavifer isolates clustered together with three different sequences in GenBank (uncultured bacterium clone control1, uncultured bacterium gene for chitinase and Streptomyces sp. An26) of chitinase genes (Fig. 1). All three Stenotrophomonas isolates (Nos V, VI and VII) contained chitinase gene fragments with closest identities to a Myxococcus xanthus chitinase gene (Table 1). However, the chitinase genes from these isolates were also very similar (85%, 95% and 96%) to a chitinase gene from a Stenotrophomonas maltophilia strain.

Clone libraries of chitinase gene fragments (240 bp) were also constructed from amplified PCR fragments of directly extracted soil DNA and pooled metagenomic library DNA. Sequences from both the clone library from directly extracted soil DNA (35 sequences) and the pooled soil metagenomic library (29 sequences) showed similarities to a diversity of chitinase genes when aligned with known sequences in GenBank. None of the sequenced chitinase gene fragments were identical on a nucleotide level to each other or to any of the isolates.

Phylogenetic analysis (Fig. 1) showed that one sequence cluster contained only chitinase sequences from the metagenomic DNA library and sequences from isolates V, VI and VII (Stenotrophomonas spp.), all similar to a chitinase gene of M. xanthus USC7 (AY033407), and contained no soil-derived chitinase genes. By contrast, another cluster contained only sequences from the directly extracted soil DNA and sequences from isolates I (Bacillus) and XII (Streptomyces) that showed the highest similarities to a chitinase gene previously sequenced from an uncultured bacterium (AB361986) amplified from arable soil DNA (Terahara et al., 2009). However, a third cluster contained chitinase gene sequences from the metagenomic library, the directly extracted soil DNA, the most dominant bacterial isolate S. clavifer and/or S. mutomycini and from two isolates of Streptomyces viridochromogenes. These sequences were similar (81–88% identity in nucleotide sequence) to the chitinase gene described as originating from an uncultured bacterium clone control1 (AF455091), initially detected by molecular analysis of a chitinolytic bacterial community in chitin-containing bags buried in grassland sites (Metcalfe et al., 2002). The distribution of the sequences within this latter cluster was relatively even and the cluster also contained chitinase sequences of the most dominant bacterial isolates, S. clavifer and/or S. mutomycini (Table 1). A fourth cluster was smaller and contained chitinase sequences from both directly extracted soil DNA and the metagenomic library with the highest similarities (82–84% nucleotide identity) to a chitinase gene described as an uncultured bacterial clone sludge C95 (AF484821) different from the chitinolytic community as that mentioned above (Metcalfe et al., 2002).

T-RFLP of 16S rRNA and chitinase genes

A rapid screening of chitinase genes in the pooled metagenomic library (HMW DNA) and the directly extracted soil DNA was performed by T-RFLP analysis. The results showed a difference in TRF profiles between the DNA directly extracted from soil and the pooled metagenomic library DNA, with an average of 42% shared TRFs between the T-RFLP profiles for all three enzymes (HhaI, HaeIII and MspI; not shown) (Table 2 and Fig. 2).

View this table:
2

16S rRNA and chitinase gene T-RFLP analyses of bacterial isolates, DNA from the metagenomic library and from directly extracted soil

16S rRNA geneChitinase gene
MspIHhaIHaeIIIMspIHhaIHaeIII
Soil metagenomic library (DNA of pooled clones)982434534542
Number of isolates represented in the metagenomic library18/1818/1818/186/6*13/1312/12
Soil DNA (directly extracted)936282547758
Number of isolates represented in the soil DNA (directly extracted)15/1815/1815/186/6*11/1312/12
Common TRFs between metagenomic and direct extracted DNA321120183120
33%23%34%34%51%40%
  • Number of TRFs detected using different restriction enzymes. Number of isolates with their 16S rRNA or chitinase gene TRF size represented in the T-RFLP analysis of metagenomic library and directly extracted soil.*The total number of isolates was six due to two isolates without detected chitinase genes and 10 isolates of the remaining were without TRF fragments in the detectable range of 60–240 nt.

  • The total number of isolates was 13 due to two isolates without detected chitinase genes and three isolates of the remaining were without TRF fragments in the detectable range of 60–240 nt.

  • The total number of isolates was 12 due to two isolates without detectable chitinase genes and four isolates of the remaining were without TRF fragments in the detectable range of 60–240 nt.

2

Chitinase genes detected by T-RFLP (HhaI). DNA from direct extracted soil (red) and the pooled metagenomic library (blue). The graph represents the average relative abundance from three replicates for each TRF.

The most dominant TRF identified in the T-RFLP profiles of both directly extracted soil DNA and the pooled metagenomic library was TRF103 using HhaI (Fig. 2). A comparison of the T-RFLP profiles (HhaI) from the chitinase genes of the antifungal isolates showed that TRF103 corresponded to chitinase genes from the most common isolates from the suppressive soil (S. mutomycini and/or S. clavifer), which were previously demonstrated to have both antifungal and chitin-degrading activity (Fig. 1[link], Table 1). However, this pattern was not consistent for all nine of these isolates (Table 1; Nos IX–XVII): two of the isolates did not have any detectable TRFs (Nos XII and XIV) and No. XI had a shorter TRF (TRF85).

Other TRFs could be matched to some additional isolates. For example, TRF181 could be matched to chitinase genes from three isolates with closest identities to a Stenotrophomonas sp. (Nos V–VII). In addition, TRF75 could be matched to the chitinase genes of the Pseudomonas sp. (No. IV) and the B. antarcticum (No. II) isolates and these sequences were nearly identical (over 99% identical).

T-RFLP was also used to analyse the dominant 16S rRNA genes in the pooled fosmid library (Figs 3 and 4). The results were compared with analyses of the bacterial community structure in directly extracted soil DNA and with the 16S rRNA gene sequences of the 18 isolates from the soil. The dominant 16S rRNA gene TRFs of the pooled metagenome indicated the presence of common soil bacteria such as Bacillus, Paenibacillus, Nitrosomonas, Rhizobium and Clostridium (Fig. 3). Representative TRFs of all of the bacterial isolates (based on in vitro digestion of their cloned 16S rRNA genes) could be identified in both sources of DNA (pooled fosmid library and directly extracted soil DNA), with the exception of TRFs for the Stenotrophomonas isolates, which were not detected in the directly extracted soil DNA. Although there were differences in many of the 16S rRNA gene TRFs detected in the DNA from the pooled fosmid library compared with the directly extracted DNA (Fig. 3b), approximately 30% of the TRFs were detected in both sources of DNA (Table 2, Fig. 4). This finding was enforced by the high agreement (high reproducibility) in the T-RFLP results obtained from replicate DNA samples that were obtained using both approaches. We could also conclude that the expected high level of E. coli contamination (host cell for the metagenomic library), although present, did not interfere with the analysis after subtraction (Figs 3and 4).

3

(a) 16S rRNA gene analysed by T-RFLP (MspI). TRFs from 16S rRNA genes amplified from DNA directly extracted from soil (red) compared with DNA from the pooled metagenomic library (blue). (b) The TRF values corresponding to Escherichia coli (host) were excluded and the relative abundance was recalculated. The graphs represent the average relative abundance from three replicates for each TRF.

4

16S rRNA gene analysed by T-RFLP (MspI). Comparison showing only those TRFs that were identified in both DNA sources; DNA from directly extracted soil (red) and extracted from the pooled metagenomic library (blue) without Escherichia coli (host) TRFs. The graph represents the average relative abundance from three replicates for each TRF.

Nucleotide sequence accession numbers

The 16S rRNA gene fragment sequence data were submitted to GenBank under accession numbers EU864323EU864340 and the chitinase gene fragment data under accession numbers EU864341EU864421.

Discussion

In this study, we used a combination of approaches to screen for chitinase genes in a Swedish soil that was previously characterized to be suppressive to phytopathogens. We compared the results obtained from a metagenomic DNA library with those obtained from direct extraction of DNA from soil. In addition, we investigated a number of isolates obtained previously from the same suppressive soil that were demonstrated to be antagonistic to phytopathogens and to have chitinase activity on agar plates (Adesina et al., 2007). The different DNA sources (pooled fosmid library, directly extracted soil DNA and bacterial isolate DNA) were screened for chitinase genes and 16S rRNA genes by T-RFLP and cloning and sequencing. To our knowledge, our study is the first to use such a comprehensive set of comparisons to assess a specific function in soil.

Previously, Ikeda et al. (2007) used T-RFLP and clone library analysis to assess chitinase genes in bulk and rhizosphere soil from a maize field. They found novel groups of bacterial chitinase genes and large differences in chitinase gene diversity between the bulk and the rhizosphere soil. Metagenomics has previously been used to identify chitinase genes in aquatic environments (Cottrell et al., 1999; LeCleir et al., 2007). Cottrell and colleagues found that culture-dependent methods were in line with metagenomic estimations of bacterial communities capable of chitin degradation. This is in line with our results in soil, where the chitinase genes of the isolates were well distributed among the clusters of sequences from both metagenomic and directly extracted soil DNA. Also, all of the isolate's 16S rRNA gene sequences corresponded to TRFs and were either represented in directly extracted soil DNA or in the pooled fosmid DNA.

Nearly all of the chitin-degrading isolates belonged to known genera with chitinase-producing capacity, such as Streptomyces (Joo, 2005), Stenotrophomonas (Zhang et al., 2001), Pseudomonas (Kitamura & Kamei, 2003) and Bacillus (Watanabe et al., 1990). The most common chitinase-producing isolates (Adesina et al., 2007) corresponded to S. mutomycini and/or S. clavifer and these bacteria also contained the most dominant chitinase gene variant (Fig. 2; TRF103). In a previous T-RFLP analysis of 16S rRNA genes from the same suppressive soil, we found that representatives of Pseudomonas and Streptomyces increased significantly in abundance after chitin was added to the soil (Hjort et al., 2007) and S. mutomycini and/or S. clavifer were predicted to be the dominant species in both the total and the active bacterial communities after chitin addition. In the present study, we also found that TRF159 that correlates to an S. mutomycini and/or the S. clavifer 16S rRNA gene was present in and highly abundant in the soil. Taken together, these combined results strongly suggest that the S. mutomycini and/or S. clavifer chitinase and 16S rRNA genes that we detected using molecular approaches correspond to some of the Streptomyces spp. isolates that we obtained from the same soil. These isolates, therefore, were most likely responsible for chitinase production in the suppressive soil and they may have potential for biocontrol of some soil-borne fungal diseases.

Previous studies have shown that the soil we studied here contains bacteria that have the dual effect of growth inhibition of Rhizoctonia and production of chitinolytic activity (Adesina et al., 2007). In addition, the same soil was previously classified as suppressive to clubroot disease caused by Plasmodiophora. Both the cell wall of Plasmodiophora spores and mycelia of Rhizoctonia contain chitin (Bartnicki-Garcia, 1968; Moxham & Buczacki, 1983), suggesting that chitinase activity would be a relevant tool in the antagonistic arsenal used against these phytopathogens. However, abiotic or other unknown biological factors could also be the cause for the suppressiveness.

Interestingly, the clone library analyses showed that some chitinase gene groups were specifically detected in different sources of DNA. For example, chitinase genes from two of our Stenotrophomonas sp. isolates were only detected in the metagenomic library, whereas another group of chitinases was only detected in the directly extracted soil DNA. However, all bacterial isolates, except the Stenotrophomonas sp. isolates that were only detected in the metagenomic library, were represented in T-RFLP profiles from both sources of DNA.

The species prediction based on TRF length is not conclusive because more than one species can have the same TRF length, although three different restriction enzymes were used in this study to increase the predictive power of the analysis. In addition, previous studies have shown that the T-RFLP technique does not detect all 16S rRNA genes present in a complex sample, but identifies the most dominating populations, limiting the detection of rare populations (Engebretson & Moyer, 2003; Bankhead et al., 2004; Benítez et al., 2007). However, the T-RFLP method is highly reproducible and we have previously observed that this soil has a very similar T-RFLP temporal profile of 16S rRNA genes over different seasons (Hjort et al., 2007; K. Hjort, unpublished data).

Optimally, both sources (metagenome and directly extracted soil DNA) should contain the same chitinase and 16S rRNA gene profiles for the same soil samples. However, the DNA preparation procedure differed for these two approaches: i.e. harsh but efficient direct extract of DNA vs. a gentle HMW extraction from extracted microbial cells for the metagenomic library construction. Also, there is more loss of DNA during preparation of the metagenomic library compared with directly extracted DNA. In addition, the efficiency of cloning of different sources of DNA, the ability of the vector and the host to stably replicate the foreign DNA (Riesenfeld et al., 2004; Hårdeman & Sjöling, 2007) or the potential toxicity of a cloned insert encoding molecules harmful to the host, if expressed, may be some of the factors contributing to the differences in the composition of the DNA cloned into the fosmid library compared with the directly extracted DNA. Undoubtedly, we were primarily limited by low coverage with all the sampling methods used and have only screened a small fraction of the diversity of the soil community.

Regardless of these technical limitations, we demonstrated for the first time an impressive agreement between three very different screening techniques, all of which pointed towards specific Streptomyces species that could play a role in suppression of fungi by chitinase production in soil. At the same time, due to different biases in the methods used, we found different clusters of chitinase genes that were represented depending on the approach used. Therefore, we can conclude that the combination of targeted molecular approaches increases the information obtained and the reliability of the data. These results should lay the groundwork for making informed decisions about the appropriate source material to use in other studies that aim to screen for specific functional genes in environmental samples.

Acknowledgements

This study was funded by the EU Metacontrol project (QLK 3-2002-2068) and in part by the EU Metaexplore project (KBBE-222625), the US Department of Energy Contract DE-AC02-05CH11231 with Lawrence Berkeley National Laboratory and the Baltic Sea Foundation.

Footnotes

  • Editor: Philippe Lemanceau

References

View Abstract