Introduction

Some heavy-metal elements (e.g., Cu, Fe, and Zn) are essential micronutrients that are required for a wide variety of cellular processes, but may become cytotoxic at elevated concentrations. Various other metal elements (e.g., Hg and Pb) have not been shown to be essential for living organisms and can be toxic even at very low concentrations. Hence, soils that contain high concentrations of heavy metals, whether from natural origin or from anthropogenic activity, may pose a considerable challenge to exposed biota, and this selective pressure may lead to the evolution of tolerant ecotypes. The ectomycorrhizal basidiomycete Suillus luteus (L.:Fr.) is a common root symbiont found in pioneer conditions and has been shown to alleviate heavy-metal toxicity for its host plant in metal-contaminated environments. In vitro dose–response experiments have revealed the adaptive nature of the heavy-metal tolerance in this fungal species: populations inhabiting heavy-metal contaminated soils characteristically displayed a higher level of tolerance to those metals that were enriched in the soil of origin when compared with populations from noncontaminated soils (Colpaert et al. 2000, 2004; Adriaensen et al. 2005). Although the exact mechanism remains unclear, this tolerance trait is believed to be due to adaptations of mechanisms that are involved in the general homeostasis of and the constitutive tolerance to essential metal elements. Possible strategies for preventing the build-up of toxic heavy-metal concentrations at sensitive sites within the cell include reduced uptake or enhanced efflux pumping of heavy metal ions, extracellular chelation, and cell-wall binding, intracellular chelation, and compartmentation by transport into the vacuole or other cell organelles. Additionally, heat-shock proteins and metallothioneins may be involved in the repair of stress-damaged proteins, and antioxidative detoxification mechanisms may reduce the accumulation of reactive-oxygen species initiated by metal ions (Hall 2002). In S. luteus, Zn exclusion mechanisms are most likely involved in the naturally selected adaptive Zn resistance (Colpaert et al. 2005).

In the present study, complementary DNA-amplified fragment length polymorphism (cDNA–AFLP) was performed to obtain a first insight into genes that are differentially expressed in Zn-tolerant and sensitive S. luteus isolates, both in the presence and absence of high external zinc concentrations, and that may be involved in stress relief and tolerance. cDNA–AFLP is an efficient method for the isolation and identification of differentially expressed genes, which gives reliable results when confirmed by RNA gel blot analysis (Bachem et al. 1996, 1998). It is a genome-wide expression analysis tool that does not require prior sequence information and, therefore, constitutes a useful tool for gene discovery (Ditt et al. 2001). In comparison with other techniques used for this purpose (e.g., suppression subtractive hybridization), cDNA–AFLP shows a very high specificity and allows distinguishing between genes with a high degree of sequence homology (Breyne and Zabeau 2001). The obtained information is useful for a better understanding of the mechanisms of heavy-metal toxicity and tolerance in ectomycorrhizal fungi, and because these fungi protect their host against toxic effects of heavy metals, it may eventually contribute to the development of more efficient phytoremediation strategies of polluted soils (Adriaensen et al. 2005, 2006).

Material and methods

Fungal material

Two isolates of S. luteus were used in this study: one Zn-tolerant isolate (UH–Slu–Lm8) obtained from a heavy metal polluted site in Lommel (Belgium) and one Zn-sensitive isolate (UH–Slu–P13) obtained from a nonpolluted site in Paal (Belgium). In a previous study, these isolates have been shown to reside at opposite extreme ends of the tolerance spectrum (Colpaert et al. 2004). The in vitro EC50 value for biomass production on solid Fries medium was 3.1 mM Zn for the UH–Slu–P13 isolate. The Zn-tolerant UH–Slu–Lm8 from the Lommel population had an in vitro EC50 value of 27.0 mM Zn.

For the present investigation, both isolates were cultured on the same solid modified Fries medium (Colpaert et al. 2004). The basic solution contained 28 mM glucose, 5.4 mM ammonium tartrate, 1.5 mM KH2PO4, 0.4 mM MgSO4·7H2O, 0.3 mM NaCl, 0.2 mM CaCl2·2H2O, 56 μM myo-inositol, 4 μM FeCl3·6H2O, 6 μM MnSO4·H2O, 0.8 μM CuSO4·5H2O, 0.8 μM nicotinamide, 0.7 μM p-aminobenzoic acid, 0.5 μM pyridoxine, 0.3 μM riboflavin, 0.3 μM thiamine, 0.2 μM Ca-pantothenate, 0.1 μM biotin, and 0.8% agar. Zinc was added to the nutrient medium as ZnSO4·7H2O at three different concentrations of 0.02, 1.5, and 3 mM Zn2+ for the different treatments. The pH of the final medium was adjusted to 4.5.

Uniform inocula (0.5-cm2 plugs of fungal mycelium) were prepared from 1-week-old cultures of both isolates by preincubation of various plugs of mycelium on cellophane-covered agar plates with basic medium for 2 or 3 days. Single inocula were transferred to the test plates covered with cellophane and were allowed to grow in the dark for 7 days at 23°. Three replicates were made for each treatment.

RNA extraction and cDNA synthesis

Cultured mycelial tissue from the three replicates was pooled and ground in liquid nitrogen. Total RNA was extracted from 60-mg ground tissue using the RNeasy plant mini-kit (Qiagen, France) and poly(A)+ RNA was isolated from 250-μg samples of total RNA using Oligotex columns (Qiagen), both according to the manufacturer’s instructions. Double-stranded cDNA was synthesized starting from 0.5–1 μg of poly(A)+ RNA using the Smart™ cDNA library construction kit (Clontech, USA) following the manufacturer’s instructions and ligation-dependent polymerase chain reaction (LD-PCR) products were purified by a QIAquick PCR purification kit (Qiagen).

cDNA–AFLP analysis

cDNA–AFLP analysis was performed following Bachem et al. (1996). In brief, approximately 0.5 μg of double-stranded cDNA was digested with the restriction enzymes BstYI and MseI in a two-step reaction. After adaptor ligation, the preamplifications were performed with the MseI and BstYI primers without selective nucleotides (BstYI primers had C as 3′ nucleotide). From a tenfold dilution of preamplification product, 5 μl was used for the subsequent selective amplification using MseI and 33P-labelled BstYI primers with two selective nucleotides (all possible combinations of BstYI primers + CN and MseI primers + NN were used). Amplification products were separated on 5% denaturing polyacrylamide gels using the SequiGen system (BioRad, USA). Vacuum-dried gels were exposed to Fuji super RX medical X-ray film (Fujifilm Medical Systems, Benelux NV, Belgium) and scanned with a Storm 860 PhosphoImager (Molecular Dynamics, USA).

Characterization of cDNA–AFLP fragments

Bands corresponding to transcripts of which the steady-state expression levels differed between both isolates in at least one of the three zinc treatments were cut out from gel, and the DNA fragments were extracted according to Frost and Guggenheim (1999). The gel fragments were rehydrated in 100 μl of 2× GoTaq PCR buffer (100 mM KCl, 20 mM Tris–HCl at pH 9.0 and 25°C, 3 mM MgCl2, and 0.2% Triton® X-100, Promega Benelux, The Netherlands) for 10 min at room temperature. The remaining buffer was removed, 100 μl of fresh buffer was added, and the samples were incubated at 94°C for 90 min.

The eluted DNA was purified using a Purelink™ 96 PCR purification kit (Invitrogen, USA) and reamplified under the same conditions as used for selective amplification. The amplified fragments were ligated into pCR®2.1-TOPO® vectors (Invitrogen) according to the manufacturer’s instructions, and the plasmids were transformed into Escherichia coli, TOP10 strain (Invitrogen). Plasmid inserts from four positive clones were PCR amplified in 20-μl reactions containing 0.5 μM M13f-primer, 0.5 μM M13r-primer, 1.5 U Taq DNA polymerase (Amersham Pharmacia Biotech, USA), 10 mM Tris–HCl at pH 9, 15 mM MgCl2, 50 mM KCl, and 2 mM of all four deoxyribonucleotide triphosphates (dNTPs). After an initial cell lysis at 95° for 5 min, PCR was performed with the following temperature profile: 35 cycles of 95° for 20 s, 50° for 20 s, and 72° for 1.5 min. The PCR products were diluted 20 times, and 1 μl of each was used in a separate 20-μl sequencing reaction with the M13f-primer using ABI BigDye® terminator reaction mix (Applied Biosystems, USA), following the manufacturer’s instructions. The sequences were run on an ABI PRISM® 3100-avant genetic analyzer. The local basic local alignment search tool (BLAST) function in the BioEdit software package (Hall 1999) was used to identify redundant transcript-derived fragments (TDFs), and database searches were performed using the BLAST network service (NCBI, National Center for Biotechnology Service; http://www.ncbi.nlm.nih.gov/BLAST). The functions of function-known genes by BLASTX and TBLASTX sequence alignments (Altschul et al. 1997) with the nonredundant (nr) databases were classified according to their putative function as described in the Saccharomyces cerevisiae functional catalogue (http://mips.gsf.de/genre/proj/yeast/).

RT-PCR analysis

To verify cDNA–AFLP expression patterns, RT-PCR analysis of selected differentially regulated genes and a nonregulated control gene was performed using cDNA from the original S. luteus isolates as template. Based on the sequence of the recovered TDFs, gene-specific primers were designed using primer3 software (Rozen and Skaletsky 1996) and tested for PCR amplification of a fragment of the same size with equal efficiency in both isolates using genomic DNA as template and the same PCR conditions as described below. Confirmation of the cDNA–AFLP expression patterns was performed with equal amounts of template cDNA (approximately 25 ng) in 20-μl reactions containing 0.5 μM of each primer (see Table 1 for primer combinations), 1 U Taq DNA polymerase (Amersham Pharmacia Biotech), 10 mM Tris–HCl at pH 9, 15 mM MgCl2, 50 mM KCl, and 200 μM of all four dNTPs. The temperature profile of the PCR was the following: initial denaturation at 95° for 2 min, 30 cycles of 95° for 30 s, primer-specific annealing temperature for 30 s and 72° for 1 min, followed by a final extension at 72° for 10 min. Amplified PCR products (10 μl) were electrophoresed on a 2% (w/v) agarose gel and visualized by ethidium bromide staining.

Table 1 Primer combinations used for RT-PCR verification of differentially expressed cDNA–AFLP fragments in a Zn-tolerant and a Zn-sensitive Suillus luteus isolate
Full size table

Results

cDNA was synthesized from RNA extracted from a Zn-tolerant and a Zn-sensitive S. luteus isolate, cultured on solid medium with three different concentrations of zinc. A clear difference in the response to elevated zinc concentrations was apparent between the two isolates, as growth rate of the nontolerant isolate was significantly more restricted at zinc concentrations of 1.5 and 3 mM when compared with the growth rate of the tolerant isolate. However, both isolates maintained a continuous growth rate at the applied zinc concentrations, and total RNA quantification indicated that the amounts of RNA did not vary significantly between the control and the zinc-treated mycelia, which indicates that the zinc treatments were not lethal. The fact that only a limited number of genes were affected in the Zn-treated UH–Slu–P13 isolate also suggests a relatively mild level of the Zn stress.

cDNA–AFLP expression patterns were obtained by selective amplification using 64 of the 128 possible primer combinations (BstYI primers + CN, MseI primers + NN), and 59 of these revealed TDFs with differential expression patterns (Fig. 1). In total, 3,793 TDFs were screened, and 213 (6%) were found to be expressed at different levels in both isolates in at least one of the three zinc treatments. Most of these TDFs were exclusively and constitutively expressed either in the tolerant isolate or in the sensitive isolate (70 TDFs in the tolerant isolate and 67 in the sensitive isolate). Two of the TDFs that were exclusively expressed in the tolerant isolate showed reduced expression levels at elevated zinc concentrations, and four of the TDFs that were exclusively expressed in the nontolerant isolate showed increased expression levels at higher zinc concentrations. The remaining TDFs were expressed in the tolerant and the nontolerant isolate, either constitutively in both but with different expression levels (38 TDFs) or modulated in response to an elevated zinc concentration (32 TDFs). In the Zn-tolerant isolate, expression levels of four TDFs increased at elevated zinc concentrations and expression levels of ten TDFs decreased, whereas in the sensitive isolate, expression levels of seven TDFs increased and expression levels of three TDFs decreased.

Fig. 1
figure 1

Part of an autoradiograph obtained by cDNA–AFLP analysis of a Zn-tolerant and a Zn-sensitive S. luteus isolate. The primer combinations that were used to generate the gene expression profiles are indicated above the lanes; A tolerant isolate grown on Fries medium with 0.02 mM Zn2+, B sensitive isolate grown on Fries medium with 0.02 mM Zn2+, C tolerant isolate grown on Fries medium with 1.5 mM Zn2+, D sensitive isolate grown on Fries medium with 1.5 mM Zn2+, E tolerant isolate grown on Fries medium with 3 mM Zn2+, F sensitive isolate grown on Fries medium with 3 mM Zn2+, arrows indicate TDFs with differential expression patterns

Full size image

One hundred ninety-seven TDFs (92%) were successfully recovered from the gel matrix and PCR amplified. The recovered fragments, ranging in length between 50 and 450 bp, were cloned before sequencing to prevent problems associated with direct sequencing of PCR products (Durrant et al. 2000; Ditt et al. 2001), and correct sequences were obtained for 156 TDFs. Due to the presence of comigrating fragments, the correct sequences of 41 TDFs could not be identified unambiguously.

Based on pairwise comparisons among the 156 sequences of recovered fragments, 144 TDFs were found to represent potentially different genes (Table 2; GenBank accession numbers AM085154–AM085297). Redundancy was due to the presence of highly homologous sequences, resulting from mispriming during PCR amplification or representing either different alleles from the same gene or different instances of multicopy genes (Bachem et al. 1996).

Table 2 Sequence homology, expression patterns, and functional classification of differentially expressed transcript-derived fragments (TDFs) obtained by cDNA–AFLP of a Zn-tolerant and a Zn-sensitive Suillus luteus isolate cultured on solid media with different zinc concentrations; the expression levels per gene are described as highest, reduced, or none
Full size table

Using a threshold value of 1E-05 for the E value returned by the BLASTX or the TBLASTX search algorithm, about 25% of the nonredundant TDFs showed homology with genes of known function, 4% showed homology with function-unknown genes, and 71% showed no homology (Table 2). In most cases, highest homology scores corresponded to sequences from fungal species, in particular Cryptococcus neoformans (20 sequences). The species showing highest homologies with S. luteus translated gene sequences included the closely related species S. pictus, but high homology scores were also obtained for sequences from bacterial, animal, and plant species. Some TDFs showed homology with different parts of the same gene and are either derived from the same transcript or represent different alleles or different instances of a multicopy gene: TDFs 27 and 47, homologous with a gene encoding for glutathione S-transferase, TDFs 28 and 133, homologous with an ARF small monomeric GTPase gene, and TDFs 81 and 100, homologous with a gene encoding for arginine–transfer RNA (tRNA) ligase. Of the 36 nonredundant function-known genes, 15 are involved in cellular metabolism and energy production, 6 in cell rescue/defense and 6 in protein synthesis. Other genes have functions in cellular transport (four), signal transduction (two), transcription (two), and cellular organization (one).

To validate the cDNA–AFLP expression patterns, eight differentially expressed TDFs were selected for RT-PCR analysis using specific primer sequences and cDNA from the original isolates as template. The comparisons showed that six of the eight cDNAs examined had the same expression profiles as revealed by cDNA–AFLP (Fig. 2). A different expression pattern was obtained for TDFs 89 and 134. In case of TDF 134, the homologous band was absent in the transcript profile of the nontolerant isolate, whereas RT-PCR showed the corresponding gene to be expressed. This apparent absence of a homologous band in the transcript profile could be due to sequence divergence of the corresponding genes in both isolates. In case of TDF 89, RT-PCR did not confirm the observed presence of a band in the tolerant isolate treated with 3 mM Zn2+, which may indicate a failed RT-PCR or the presence of nonhomologous fragments of the same size. However, in general, cDNA–AFLP is shown to be a reliable method for identifying differentially expressed genes in S. luteus exposed to heavy-metal stress.

Fig. 2
figure 2

Comparison of expression patterns of eight differentially expressed genes and one constitutively expressed control gene (TDF C) obtained by cDNA–AFLP and RT-PCR; A/C/E transcript profile of Zn-tolerant isolate treated with 0.02, 1.5, and 3 mM Zn2+, respectively; B/D/F transcript profile of Zn-sensitive isolate treated with 0.02, 1.5, and 3 mM Zn2+, respectively

Full size image

Discussion

Estimations suggest that the genome of filamentous fungi, including ectomycorrhizal fungi, usually varies between 20 and 40 Mb, with a complement of about 8,000 genes (Kupfer et al. 1997; Le Quéré et al. 2002). Using this estimation for the number of genes in S. luteus, the present set of 140 nonredundant differentially expressed TDFs corresponds to approximately 2% of the total expected complement of genes. A total of 42 TDFs (30%) corresponded to genes homologous to GenBank entries, including genes of known function as well as hypothetical proteins. The remaining TDFs (70%) did not have significant matches within the GenBank databases and may represent undescribed fungal genes, although this could also be partly due to the fragments being derived from either the 5′ or the 3′ untranslated messenger RNA (mRNA) regions. This relatively high proportion of TDFs showing no homology to GenBank entries is compatible with the range of 50–65% reported for expressed sequence tags (ESTs) in other filamentous fungi (Lee et al. 2002; Ospina-Giraldo et al. 2000; Skinner et al. 2001) and might reflect the low number of database entries describing sequences of fungal origin.

In concordance with the high levels of genomic divergence found between S. luteus isolates (Muller et al. 2004), considerable variation in steady-state transcript profiles was revealed between the tolerant and the nontolerant S. luteus isolate, both following treatments with physiological external zinc concentrations and increased concentrations. Although, in several cases, the differential expression pattern was shown to be caused by sequence divergence of the corresponding transcripts, RT-PCR analysis confirmed that the differential TDF profiles were mainly due to differential gene expression in both isolates. Adaptive evolution of heavy-metal tolerance by natural selection may explain this difference in gene expression to a certain extent, but other causes, including genetic hitchhiking and natural regulatory variation, may be equally important. In genomic regions with low recombination rates, positive selection of alleles responsible for the tolerance trait and modifiers that enhance the tolerance may be associated with a reduction of the genetic variation around the selected sites (Maynard Smith and Haigh 1974), and this signature of selection may be important at the genome-wide level and more common than usually accepted (Quesada et al. 2003). Besides the variation due to this genetic hitchhiking effect, part of the observed differential expression in both S. luteus isolates will be due to variation in gene expression levels present in natural populations. Townsend et al. (2003) reported extensive natural regulatory variation in Saccharomyces cerevisiae and showed that the majority of these differences in gene expression are below the twofold level. Although most of the differences in expression profiles between both S. luteus isolates were due to TDFs exclusively observed in either one of the isolates and could, therefore, be expected not to result from natural regulatory variation, a larger number of isolates should be screened to assess the importance of natural regulatory variation in S. luteus.

Relatively few genes were found to be responsive to the zinc treatments at the transcript level, but an apparent difference in the TDF profiles of these genes may reflect a difference in response to elevated external zinc concentrations between the tolerant and the nontolerant isolate. In the Zn-tolerant isolate, a modulation of the TDF profiles was, in general, only observed at 3 mM zinc, whereas in the Zn-sensitive isolate, a response was already observed at 1.5-mM external zinc. Both isolates also differed in the set of genes responsive to zinc treatment and in the way that TDF expression levels were affected, a decrease being the main effect in the tolerant isolate and an increase in the nontolerant isolate. In the nontolerant isolate, elevated external zinc concentrations increased the TDF expression levels of ketol-acid reductoisomerase and translation elongation factor 1 respectively involved in amino acid biosynthesis and protein synthesis. Increased transcript levels have been shown for ketol-acid reductoisomerase after heat shock and for translation elongation factor 1 in response to heavy-metal-induced stress, but the functional significance of this increase remains unclear (Joseph et al. 2002; Rosen et al. 2002). In the tolerant isolate, differentially expressed genes of which the TDF expression levels were down regulated at the 3-mM zinc concentration encoded for homologues of ketol-acid reductoisomerase, thioredoxin peroxidase, manganese superoxide-dismutase, and translation elongation factor 1, as well as homologues of Myc1 and a 60S ribosomal protein, the latter respectively involved in transcription and protein synthesis.

Among the identified genes, five showed significant similarity with genes encoding proteins involved in cell rescue and defense: glutathione S-transferase (TDFs 27/47), metallothionein (TDF 89), thioredoxin peroxidase (TDF 128), heat-shock protein HSP60 (TDF 134), and manganese superoxide-dismutase (TDF 147). Three of these proteins are known to be responsive to oxidative stress: thioredoxin peroxidase, which reduces organic hydroxyperoxides and H2O2 to harmless products using thioredoxin as an electron donor (Godon et al. 1998; Lee et al. 1999); glutathione S-transferase, which metabolizes toxic products of lipid peroxidation and DNA damage caused by reactive oxygen species and detoxifies H2O2 by oxidizing reduced glutathione (Marrs 1996; Roxas et al. 2000); and manganese superoxide-dismutase, which catalyzes the conversion of superoxide radicals (\( \cdot O^{ - }_{2} \)) to hydrogen peroxide and oxygen (Jacob et al. 2001). Although zinc is not a redox-active metal, exposure to zinc is connected with the accumulation of reactive oxygen species because of its capability to displace bounded iron, a redox-active metal, from DNA and proteins, and because of its stimulatory effect on ferrireductase activity, which may also generate superoxide radicals (Fujs et al. 2005). Nevertheless, an increase of the TDF expression levels of the three genes involved in the protection against oxidative stress was not observed in either S. luteus isolate after zinc treatment. On the contrary, a reduction of the expression level was observed in the tolerant isolate for manganese superoxide-dismutase and thioredoxin peroxidase.

An increase in TDF expression levels of heat-shock protein HSP60 or metallothionein after zinc treatment was not observed either. Heat-shock proteins are known to be induced under a variety of stress factors that cause protein unfolding, misfolding, or aggregation (e.g., heat, UV radiation, and heavy-metal ions), and are capable of re-establishing the balance between protein synthesis, assembly, and degradation. HSP60 is believed to protect enzymes possessing Fe/S clusters, such as aconitase and succinate dehydrogenase, from oxidative inactivation and release of their iron, thus, preventing additional oxidative stress (Cabiscol et al. 2002). Metallothioneins are small proteins containing cysteine clusters that have a high affinity for heavy metals, in particular copper, cadmium, and zinc. Their gene expression is induced by a great number of stimuli, among which are heavy-metal load, viral infection, and UV radiation, and they have been implicated in heavy-metal detoxification, metal homeostasis, and radical scavenging (Heuchel et al. 1994).

Vallino et al. (2005) reported a similar observation in the ericoid mycorrhizal fungus Oidiodendron maius, where EST analysis revealed no transcriptional response of stress-related genes after zinc treatment, and proposed this to be due to the applied zinc concentrations not being high enough to induce severe toxic effects. However, growth of the nontolerant S. luteus isolate was clearly inhibited by both zinc treatments, thus, rendering their nontoxicity unlikely. Possibly, other defense systems dominate the stress response, or posttranslational mechanisms control these defense systems. Jacob et al. (2001) showed that the transcript level of manganese superoxide-dismutase was not altered by cadmium exposure of Paxillus involutus isolates and postulated this result to be due to posttranslational mechanisms that activate a pre-existing pool of manganese superoxide-dismutase in response to cadmium-induced stress.

In conclusion, the results from this study proved cDNA–AFLP to be an efficient and high fidelity method to screen the effects of heavy-metal-induced stress on the steady-state transcript profiles of S. luteus. One hundred forty-four nonredundant and differentially expressed TDFs were isolated, and RT-PCR analysis confirmed their differential expression to be mainly due to differences in gene expression levels. Although a clear difference at the transcript level in response to the zinc treatments was found between the tolerant and the sensitive isolate, the relationship between the affected genes and the metal-induced stress was not always clear. In some cases, however, the difference in expression profiles between both isolates indicated a possible role in the cellular defense against heavy metals of the genes involved, including genes encoding for a putative metal transporter, a hydrophobin (Pawlik-Skowroñńska et al. 2002), and several proteins involved in the ubiquitin-dependent degradation of proteins (Chondrogianni et al. 2005), but further experiments are necessary to confirm their roles. These data provide a basis for the use of S. luteus as a model system for molecular genetic studies of heavy-metal tolerance in ectomycorrhizal fungi and may aid researchers studying heavy-metal tolerance in other filamentous fungi. It will now be necessary to characterize the involvement of the identified genes in heavy metal tolerance and to establish their precise role.