Terminal restriction fragment length polymorphism analysis of ribosomal RNA genes to assess changes in fungal community structure in soils

Abstract

Monitoring the structure and dynamics of fungal communities in soils under agricultural and environmental disturbances is currently a challenge. In this study, a terminal restriction fragment length polymorphism (T-RFLP) fingerprinting method was developed for the rapid comparison of fungal community structures. The terminal restriction fragment polymorphism of different regions of the small-subunit (SSU) ribosomal RNA (rRNA) gene was simulated by sequence comparison using 10 restriction enzymes, and analyzed among three different soils using fungal-specific primers. Polymerase chain reaction amplification of the 3′ end of the SSU rRNA gene with the primer nu-SSU-0817-5′ and with the fluorescently labelled primer nu-SSU-1536-3′, and digestion of the amplicons with AluI and MboI were found to be optimal and were used in a standardized T-RFLP procedure. Both the number and the intensity of terminal restriction fragments detected by capillary gel electrophoresis were integrated in correspondence analyses. Three soils with contrasting physicochemical properties were differentiated according to the structure of their fungal communities. Assessment of the impact on the fungal community structure of the amendment of two soils with compost or manure confirmed the reproducibility and the sensitivity of the method. Shifts in the community structure were detected between non-amended and amended soil samples. In both soils, the shift differed with the organic amendment applied. In addition, the fungal community structures of the two soils were affected in a different way by the same organic amendment. The fingerprinting method provides a rapid tool to investigate the effect of various perturbations on the fungal communities in soils.

Microbial community, Fungus, Soil, Fingerprinting, Terminal restriction fragment length polymorphism, Small-subunit ribosomal DNA

1 Introduction

Microbial processes in soil depend on the microflora composition and are influenced by the soil environment [1]. Various agricultural practices have been promoted to improve soil conditions for plant growth. These practices include chemical and mechanical inputs such as the use of mineral fertilizers, pesticides and tillage, or more environmentally sound practices such as rotational crop schemes or the use of organic amendments. Particularly, soil amendments with compost are being used more frequently both because of their agronomic importance and because they allow the recycling of diverse biowastes. Soil management practices not only influence plant growth, but can also modify the microbial community composition and affect microbial processes through changes in the soil structure or the availability of gas, water and nutrients. However, agricultural practices are generally evaluated on the basis of agronomical parameters, while their impact on the structure of the soil microflora has been largely neglected because of the absence of relevant tools to characterize the microbial variables.

During the last decade, the development of culture-independent molecular approaches has facilitated the monitoring of complex microbial communities in soil ecosystems. Initially, such approaches focused on the bacterial component of the soil-borne microflora. They mainly rely on the direct extraction of microbial DNA from soil, the selective amplification by polymerase chain reaction (PCR) of ribosomal RNA (rRNA) genes used as a marker of the community of interest, and the resolution of differences in sequence or size among the amplified fragments using various fingerprinting methods [2–6]. Among them, terminal restriction fragment length polymorphism (T-RFLP) was shown to be a powerful method for bacterial community analysis in environmental samples [4,7–9]. This technique involves the use of a fluorescently labelled oligonucleotide primer for PCR amplification of rRNA gene fragments and the digestion of the PCR products with one or more restriction enzymes, generating labelled terminal restriction fragments (TRFs) of different lengths according to the DNA sequence of the bacteria present in the soil community. The TRFs are separated by high-resolution electrophoresis on automated DNA sequencers, allowing the simultaneous characterization of microbial communities in numerous environmental samples.

More recently, efforts have been made to develop similar approaches in order to characterize fungal communities in complex environments. However, the development of direct molecular approaches for fungal communities has suffered from the lack of PCR primers that are both consensual and specific for fungi. This limitation has been circumvented by using a nested PCR approach or by restricting the analysis to part of the fungal community with group-specific primers [10–13]. In these studies, both the small-subunit (SSU) ribosomal DNA (rDNA) and the internal transcribed spacer (ITS) region of the rRNA genes have been investigated. Lord et al. [14] compared the suitability of both rDNA regions for T-RFLP analysis of fungal communities. Although the authors found a greater fungal diversity from the ITS region than from the SSU rDNA, they also underlined the lack of specificity of the primers used to amplify fungal SSU rDNA. The use of primers that are too broad may lead to the amplification of SSU rDNA sequences originating mostly from other soil organisms than fungi and to an underrepresentation of the fungi in the community profile. In addition, investigations of fungal community fingerprints based on SSU rDNA polymorphism performed up to now have mainly focused on the 5′ region of the SSU rDNA. However, other variable regions of the SSU rDNA may be useful to characterize fungal communities. Indeed, a recent investigation of the fungal community composition in plant roots based on the variations of the SSU rDNA sequences has revealed an unexpected diversity, with all known fungal phyla represented [15]. In this way, fungal-specific primers designed to amplify different variable regions of the SSU rDNA [16] might be good candidates to develop a fingerprinting method addressing the fungal community structure in complex ecosystems.

The objective of this study was to develop a T-RFLP approach based on the SSU rDNA sequence variations in order to rapidly profile fungal communities in soils. This approach relies on the use of PCR primers targeting different variable regions of the SSU rDNA designed to specifically amplify fungal rRNA genes from soil [16]. T-RFLP analysis was simulated with fungal SSU rDNA sequences in order to select the most polymorphic combination of PCR primers and restriction enzymes. The method was used to assess the fungal community structure of three different soils with contrasting physicochemical properties, and to detect potential changes in these communities as the result of the introduction of two different organic amendments in soil.

2 Materials and methods

2.1 Soils and treatments

Three soils with different physicochemical properties were used (Table 1). In addition, the soils of Dijon (D) and Ouroux (O) were amended with 5% compost or 5% manure. Compost derived from spent mushroom substrate (France Champignon, Saumur, France) was used fresh. Manure derived from composted dairy cattle was dried and ground by the manufacturer (SERP, Dijon, France). Unamended soil samples and soil samples amended with compost (D-Compost and O-Compost) or manure (D-Manure and O-Manure) were mixed with a 3D rotary shaker and incubated in 30-l containers for 1 month at 18°C. Aliquots of soils were passed through a 200-μm sieve and stored at −80°C until thawing immediately prior to DNA extraction.

1

Properties of soils and organic amendments used in this study

Soil sampling location or organic amendment	Soil texture	Mineral fraction in soils <2 mm (%)		pH	Organic matter (%)	N (%)
		Sand	Silt	Clay
Dijon	Loamy silt	20.8	44.8	37.4	7.1	3.25	0.15
Ouroux	Sandy	86	5.8	8.1	4.8	0.77	0.05
Châteaurenard	Loamy	23.8	49.6	26.6	7.9	5.14	0.28
Spent mushroom compost					7.8	48.4	2.20
Dairy cattle manure					7.4	53.3	2.95

2.2 DNA extraction from soils

The procedure used for soil DNA extraction was adapted from the method described by Martin-Laurent et al. [17]. The procedure was optimized on the basis of the suitability of the extracted DNA for PCR amplification with fungal SSU rDNA primers [16] from both original and amended soils. One gram of soil was added to 2 g of 106-μm-diameter glass beads, eight 2-mm-diameter glass beads and 4 ml of lysis buffer containing 50 mM Tris–HCl (pH 8.0), 20 mM EDTA (pH 8.0), 100 mM NaCl and 1% (w/v) sodium dodecyl sulfate. Samples were shaken for 30 s at 1600 rpm in a bead-beater (Mikro-Dismembrator S, B. Braun Biotech International, Melsungen, Germany) and incubated for 20 min at 70°C with mixing every 5 min. Samples were centrifuged at 14 000×g for 1 min at 4°C. The supernatant was recovered in a 2-ml microcentrifuge tube and incubated for 10 min on ice with 0.1 volume of 5 M potassium acetate. After centrifugation at 14 000×g for 10 min at 4°C, the nucleic acids in the collected supernatants were precipitated with one volume of ice-cold isopropanol for 15 min at −20°C. The precipitate was pelleted by centrifugation at 15 000×g for 20 min at 4°C, washed with 70% (v/v) ice-cold ethanol, air-dried and dissolved in 50 μl of TE buffer (50 mM Tris–HCl pH 8.0 and 1 mM EDTA). The extracts of soil nucleic acids were purified using a polyvinylpolypyrrolidone (PVPP) spin column to remove co-extracted humic acids [18], followed by a Sepharose 4B spin column. The two purification steps were miniaturized using Micro Bio-Spin columns (Bio-Rad, Marnes La Coquette, France) adapted to microcentrifuge tubes. For the preparation of PVPP columns, Micro Bio-Spin columns were filled with 100 mg of PVPP (Sigma-Aldrich, Saint Quentin Fallavier, France). The spin columns were washed twice by the addition of 400 μl of sterile water and centrifugation at 1000×g for 2 min at 10°C. The columns were plugged at the bottom and 400 μl of sterile water was added. For the preparation of Sepharose 4B spin columns, Micro Bio-Spin columns were filled with 1 ml of Sepharose 4B (Sigma-Aldrich). The liquid was allowed to flow out from the column for 5 min before centrifugation at 1000×g for 2 min at 4°C. After addition of 400 μl of TE buffer, the columns were centrifuged again. The columns were plugged at the bottom and 400 μl of TE buffer was added. Both PVPP spin columns and Sepharose 4B spin columns could be stored at 4°C for several weeks. Before use, the PVPP spin column was centrifuged at 1000×g for 2 min at 10°C. 50 μl of crude DNA extract was loaded slowly onto the top center of the PVPP spin column and the purified extract was collected after 5 min of incubation of the column on ice, followed by 2 min of centrifugation at 1000×g at 10°C. Then, the DNA extract was purified on Sepharose 4B spin column, using the same procedure except that the centrifugation steps were performed at 4°C. DNA extractions were performed in triplicate from independent soil samples.

Of the purified DNA extracts, 5 μl was resolved by electrophoresis in a 0.8% agarose gel in Tris-acetate-EDTA (TAE) buffer, together with dilutions of calf thymus DNA (Bio-Rad). Gels were stained with ethidium bromide, photographed under a camera and the staining intensities were measured with Bio-1D⁺⁺ software (Vilber-Lourmat, Marne La Vallée, France). The DNA concentrations in the DNA extracts were calculated using a standard curve of 10–100 ng of calf thymus DNA versus intensity.

2.3 PCR conditions

The primers nu-SSU-0817-5′ (TTAGCATGGAATAATRRAATAGGA), nu-SSU-1196-3′ (TCTGGACCTGGTGAGTTTCC) and nu-SSU-1536-3′ (ATTGCAATGCYCTATCCCCA) [16] were used for the direct amplification of fungal SSU rDNA from soil. Two primer sets were used in order to compare the number of TRFs from three positions of the SSU rDNA. The first primer set corresponded to nu-SSU-0817-5′ and nu-SSU-1536-3′, which gives a PCR product of approximately 750 bp of the SSU rDNA. Two different PCR reactions were conducted in which one of the primers was labelled at the 5′ end with the fluorescent dye D2 or D3 (Beckman Coulter, Fullerton, CA, USA) (D2-nu-SSU-0817-5′ or D3-nu-SSU-1536-3′). For the second primer set, the reverse primer nu-SSU-1196-3′ was 5′ end-labelled with the fluorescent dye D3 and combined with primer NS1 [19] to amplify a 1200-bp region of the SSU rDNA. The labelled and unlabelled primers were synthesized by Proligo (Paris, France) and MWG Biotech (Courtaboeuf, France), respectively. PCR amplifications were performed in a final volume of 25 μl by mixing 20 ng of soil DNA with 0.25 μM of each primer, 200 μM each of dATP, dCTP, dGTP and dTTP, 1 U of Taq DNA polymerase (Q-BIOgene, Evry, France), 500 ng of T4 gene 32 protein (Q-BIOgene), and PCR reaction buffer (10 mM Tris–HCl [pH 9.0] at 25°C, 50 mM KCl, 1.5 mM MgCl₂, 0.1% Triton X-100, 0.2 mg ml⁻¹ bovine serum albumin). DNA amplifications were performed in a Mastercycler (Eppendorf, Hamburg, Germany) with an initial denaturation of 3 min at 94°C followed by 35 cycles of denaturation (1 min at 94°C), primer annealing (1 min at 56°C) and extension (1 min at 72°C), and a final extension of 10 min at 72°C. Aliquots of 2 μl of PCR products were checked by electrophoresis in 2% agarose gels and stained with ethidium bromide.

2.4 TRF lengths deduced from SSU rDNA sequence analysis

We examined a total of 120 SSU rDNA sequences of representatives of all four major phyla of fungi from GenBank (Ascomycota, Basidiomycota, Zygomycota and Chytridomycota). The theoretical lengths of TRFs from the positions matched by the three PCR primers nu-SSU-0817-5′, nu-SSU-1196-3′ and nu-SSU-1536-3′[16] were predicted for 10 tetrameric restriction endonucleases: AluI (AG′CT), BstUI (CG′CG), DdeI (C′TNAG), HaeIII (GG′CC), HhaI (GCG′C), HinfI (G′ANTC), MboI (′GATC), MspI (C′CGG), RsaI (GT′AC) and TaqI (T′CGA). Sequences containing undetermined nucleotides were not taken into account in order to avoid artificial overestimation of the potential number of TRFs.

2.5 T-RFLP analysis

The PCR products of SSU rDNA were purified in order to remove residual primers, and quantified. Preliminary experiments had shown that the digestion of non-purified PCR products generated inconsistent T-RFLP profiles for some samples, in which residual fluorescent primers could interfere with a correct separation of the TRFs by the automated sequencer. Preliminary experiments had also shown that standardization of the quantity of purified PCR products used for restriction digestions contributed to the reproducibility of the method, since it generated TRF profiles with similar amounts of total fluorescence and similar numbers of peaks between replicates.

PCR products were purified using the MinElute PCR purification kit (Qiagen, Courtaboeuf, France) according to the instructions of the manufacturer, with two final elutions of the PCR products in 2×10 μl. Purified PCR products were quantified by comparison with known quantities of the molecular mass marker Smart Ladder (Eurogentec, Seraing, Belgium) in 2% agarose gels. Gels were stained with ethidium bromide and the staining intensities of the bands were measured with Bio-1D⁺⁺ software as above. 50 ng of purified PCR products was digested with 5 U of restriction enzymes in a final volume of 100 μl for 3 h at 37°C. Restriction digests were precipitated with 2 μl of 2.5 mg ml⁻¹ glycogen (Q-BIOgene), 10 μl of 3 M sodium acetate (pH 5.2) and 250 μl of ice-cold ethanol, centrifuged for 15 min at 12 000×g at 4°C, rinsed twice with 200 μl of ice-cold 70% ethanol and resuspended in 39.5 μl of Sample Loading Solution (Beckman Coulter). Samples were stored at −20°C until analysis. Samples were mixed with 0.5 μl of Size Standard-600 (Beckman Coulter) and loaded onto a capillary electrophoresis sequencer CEQ™ 2000XL (Beckman Coulter). Analyses were run with the Frag-4-45s method, including a denaturation of 2 min at 90°C, an injection at 2000 V during 45 s and a separation at 4800 V during 60 min. After electrophoresis, the length and the signal intensity of the fluorescently labelled TRFs were automatically calculated by comparison with the size standard, using the CEQ 2000XL DNA analysis system version 4.3.9. Fragments between 60 and 640 bp corresponding to the size range of the standard were considered. Preliminary analyses had shown that most TRFs differed by less than 1 bp between replicates; however, in rare cases, the difference was slightly higher. Thus, TRFs differing by more than 1.25 bp were further considered to be different. The comparison of the TRF sizes between samples was automated by assigning them to discrete categories using the program Lis [20]. The communities, characterized by the number of fragments and their intensity measured by the height of the peaks, were compared by correspondence analysis using the computer program NTSYS [21]. This ordination method summarizes multivariate data to a few variables or dimensions and tries to arrange the communities in a two-dimensional diagram based on their scores on the two first dimensions. Finally, pairwise Euclidean distances between communities were calculated from their coordinates on the five first dimensions.

3 Results

3.1 T-RFLP analysis of the different soils

The theoretical number of different TRFs generated with 10 different restriction enzymes was predicted from sequence analysis at three positions within SSU rDNA, corresponding to the primers nu-SSU-0817-5′, nu-SSU-1196-3′ and nu-SSU-1536-3′ (Table 2). The three combinations of labelled primer and most polymorphic restriction enzyme were used in T-RFLP analysis of DNA isolated from three different soils. With the primer pair NS1/D3-nu-SSU-1196-3′ and the enzyme HaeIII, the analysis produced a mean number of 15 TRFs per soil sample and a total number of 25 different TRFs among the three soils. With the primer pair nu-SSU-0817-5′/nu-SSU-1536-3′, a mean number of 20 distinct TRFs per soil sample was detected from either the 5′ end of the PCR product digested with MspI or the 3′ end of the PCR product digested with MboI. With this primer pair, total numbers of 30 and 34 different TRFs were recorded among the three soils from the 5′ end and the 3′ end of the PCR product, respectively. The polymorphism revealed by the method was increased by using a combination of two restriction enzymes (AluI and MboI) to digest the 3′ end-labelled PCR product nu-SSU-0817-5′/nu-SSU-1536-3′. In this case, 39 distinct TRFs were detected among the three soils.

2

Potential number of different TRF lengths generated with 10 restriction enzymes for 120 fungal sequences and for three different regions of the SSU rRNA gene

Primer	Region analyzed	Number of different TRFs with restriction enzyme
		AluI	BstUI	DdeI	HaeIII	HhaI	HinfI	MboI	MspI	RsaI	TaqI
NS1/nu-SSU-1196-3′	3′ end	47	45	49	53	42	5	27	48	43	33
nu-SSU-0817-5′/nu-SSU-1536-3′	5′ end	18	29	23	21	34	26	20	41	25	15
nu-SSU-0817-5′/nu-SSU-1536-3′	3′ end	35	14	19	18	14	31	36	22	17	34

The fungal community structures of the three soils were compared using correspondence analysis of the TRF profiles obtained from the 3′ end of the PCR product nu-SSU-0817-5′/D3-nu-SSU-1536-3′ after digestion with AluI and MboI (Fig. 1). In this analysis, both the number and the intensity of the different TRFs were considered to provide an ordination of the communities which were plotted in a two-dimensional diagram. The first dimension explained 39% of the variability observed in the data and the second dimension explained 32% of this variability. The TRF profiles derived from three completely independent analyses of the same soil clustered tightly together, showing a good reproducibility of the method, whereas the fungal communities of different soils were clearly separated. Clustering was confirmed by the comparison of intra- and inter-soil Euclidean distances, calculated from the coordinates of the samples on the five first dimensions, which represented 95% of the variability observed in the data. Indeed, pairwise comparisons of the communities revealed mean distances of 0.43, 0.41 and 0.82 between the three soil replicates of Dijon, Ouroux, and Châteaurenard, respectively. On the other hand, the mean pairwise distance between communities of two different soils was 1.27.

1

Correspondence analysis of T-RFLP data sets from three soils: Dijon (D), Ouroux (O) and Châteaurenard (C). For each soil, 1, 2, and 3 represent three independent replicates. Dim, dimension.

3.2 Comparison of non-amended and amended soils

Potential changes in the fungal community structure in response to the introduction of manure or compost in the soils of Dijon and Ouroux were evaluated by analyzing their TRF profiles, obtained from the 3′ end of the PCR products nu-SSU-0817-5′/D3-nu-SSU-1536-3′ after digestion with AluI and MboI. The application of organic amendments induced considerable changes in the TRF patterns of both soils, as illustrated in Fig. 2 with the soil samples from Ouroux. These changes corresponded mostly to an increased or decreased abundance of existing peaks in the original soil samples, while the total number of peaks per TRF pattern remained similar between original and amended soil samples. The changes induced by compost and manure in the Ouroux soil were compared using correspondence analysis (Fig. 3). The first dimension explained 43% of the variability observed in the data and the second dimension explained 22% of this variability. The variability between TRF fingerprints was again smaller for replicate community DNA of the same treatment than for different treatments. The comparative analysis revealed an evident shift in the community structures between non-amended and amended soils. In addition, the fungal community structures from the O-Compost and the O-Manure soils were differentiated from each other, indicating that the modifications in the community structures were related to the type of organic material applied. Finally, the shifts of O-Manure and O-Compost were mainly explained by the first and the second dimension, respectively, indicating that manure produced a stronger alteration of the fungal community structure than compost. Similar results were obtained for the Dijon soils amended with compost and manure.

2

T-RFLP profiles of fungal communities generated from SSU rDNA amplified from soil DNA using the primer nu-SSU-0817-5′ and the fluorescently labelled primer nu-SSU-1536-3′, and digested with AluI and MboI. A: Ouroux soil. B: Ouroux soil amended with manure. x-axis, fragment size in bp.

3

Correspondence analysis of T-RFLP data sets from non-amended and amended soil samples. For each soil, 1, 2, and 3 represent three independent replicates. Dim, dimension.

The magnitudes of changes induced by the manure in both soils were compared using correspondence analysis (Fig. 4). The first dimension explained 48% of the variability observed in the data and the second dimension explained 16% of this variability. Again, a shift was detected in the community structures between non-amended and amended soils. However, this effect was more pronounced in the Ouroux soil than in the Dijon soil. Both the first and the second dimension, originating from the multivariate analysis, explained the variations induced by the organic amendment in the Ouroux soil. On the other hand, the impact of the manure on fungal community structure from the Dijon soil was mostly explained by the second dimension. Finally, the fungal community structures from both amended soils were differentiated from each other, indicating that the fungal communities were affected in both soils by the organic amendment, but in a different way.

4

Correspondence analysis of T-RFLP data sets from non-amended and amended soil samples. For each soil, 1, 2, and 3 represent three independent replicates. Dim, dimension.

4 Discussion

Monitoring the structure and dynamics of soil-borne microbial communities in relation with soil management practices requires high-throughput and reproducible methods allowing rapid comparative analyses of large numbers of environmental samples. With this aim in view, we have developed a T-RFLP fingerprinting procedure based on sequence variations in SSU rDNA in order to address the fungal communities in soils and to investigate shifts due to organic amendments. To our knowledge, this is the first report using this approach in comparative analyses of fungal communities under agricultural perturbations. A polymorphic combination of labelled primer and restriction enzymes was selected from sequence comparisons and used to produce community fingerprints from the 3′ end of the fungal SSU rRNA gene. Applied to different soil samples, the procedure was shown to be reproducible and sensitive enough to allow both the discrimination of soils according to their fungal communities and the detection of shifts in the fungal community structures in response to stress.

Until recently, the main problem in addressing the fungal community structure in soil, using culture-independent molecular approaches, was the lack of adequate PCR primers. In our study, the fingerprinting approach was based on the specific amplification of fungal SSU rDNA from soil, using the primers nu-SSU-0817-5′ and nu-SSU-1536-3′[16]. Anderson et al. [22] have recently confirmed the adequacy of these primers for fungal community analysis by analyzing clone libraries from a soil. Such primers allowed the fungal component of the soil microflora to be specifically targeted in a single PCR step. Compared to fingerprinting methods based on a nested PCR strategy, a direct amplification of the target organisms offers advantages, in terms of both rapidity and accuracy. Indeed, the potential biases that could be introduced because of preferential amplifications of certain templates from mixed communities could be further increased if using two successive rounds of PCR amplification [23].

Community fingerprints were compared integrating both the number and the intensity of TRFs. The T-RFLP method can give a semi-quantitative view of the community analyzed by displaying the relative abundance of the community members represented in the PCR mixture [8,9]. Comparative analysis of the community structures was performed using correspondence analysis, which has already proved useful in assessing changes in bacterial communities in relation with environmental factors [8]. The evaluation of the variability observed within and between soils revealed comparable community structures for each group of three totally independent replicates derived from separate DNA extractions, PCR amplifications and restriction hydrolyses. Different parameters are likely to contribute to the reproducibility of a fingerprinting method, among which the representativeness, the purity and the standardization of the quantity of soil DNA used appear essential [24,25]. In our study, the clustering of the community profiles obtained from independent triplicates suggests that these criteria were achieved. The optimization of the preparation of the T-RFLP samples through purification and the use of standardized quantities of PCR products was also thought to contribute to the reproducibility of the fingerprinting method.

Each soil was characterized by a specific fingerprint, showing the potential of the T-RFLP procedure for discriminating the structures of fungal communities in soils. The sensitivity of the method was confirmed when revealing the impact of two different organic amendments on the structure of the soil communities. The introduction of manure and compost into soils clearly produced a shift in the fungal community structures. In addition, the fingerprinting method revealed differential responses of the fungal community, depending on the soil and on the type of organic amendment. The differentiation of the community structures of two different soils amended with the same organic material suggested that the shifts detected were related to changes of the original community structures under the influence of the amendment used, rather than to a direct input of microorganisms through the organic amendment.

The shifts detected in the community structure following a given perturbation may be further investigated by identifying the fungal responders to the perturbation applied. We used our simulation of restriction digestions on fungal sequences to investigate the possibility of direct taxonomic assignment of characteristic TRFs based on their length. This analysis revealed that many TRFs may be assigned to different fungal species or even different genera. This result was already expected from the comparison of the theoretical numbers of different TRFs presented in Table 2, which revealed a maximum number of 53 putative TRFs among 120 fungal species, indicating that different species may correspond to the same TRF. An alternative for the taxonomic interpretation of T-RFLP patterns would be to use the cloning strategies of TRFs recently proposed for bacterial community analyses [26,27]. Indeed, fungi corresponding to characteristic SSU rDNA sequences retrieved from soil samples should be identifiable at least at a high taxonomic level [22]. This approach may be further complemented, using group-specific primers in case key members of the fungal community are to be more accurately identified.

The characterization of the microbial community structure corresponds to the first step towards understanding soil functioning under the influence of different management strategies or different environmental conditions. The T-RFLP procedure described should facilitate the more systematic integration of fungal components in studies of microbial diversity in natural environments. It is expected to find widespread applications in studies when a large number of samples has to be processed, in order to rapidly monitor the structure and dynamics of indigenous fungal communities over time or under the influence of human activities or environmental disturbances.

Acknowledgements

This work was partly supported by the EU project RECOVEG: QLTR-01458. We thank J. Borneman for providing an alignment of fungal SSU rDNA. We also thank G. Recorbet and K. Klein for critical evaluations and suggestions and for English correction of the manuscript.