INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Aspergillus Section Flavi has attracted worldwide attention for economic and public health reasons because of its industrial use and toxigenic potential. This Section consists of two groups of species; one includes Aspergillus flavus, A. parasiticus, and A. nomius, all of which produce aflatoxin and cause serious problems worldwide in agricultural commodities (4). The other group includes the non-aflatoxin-producing species A. oryzae, A. sojae, and A. tamarii, which have been used for production of traditional fermented foods in Asia. Identification of the species has relied mainly upon the morphological characters used as taxonomic criteria (23). It is, however, difficult to identify Section Flavi species because of morphological divergence among isolates of the same species (15).

Recently, molecular genetic techniques have been used as tools with which to study the phylogeny and classification of Aspergillus Section Flavi. Egel et al. (6) reported that restriction enzyme analysis of a Taka-amylase A gene can divide the species of Section Flavi into several groups and detect intraspecific variations within A. nomius, A. tamarii, and A. flavus strains. On the basis of the sequence data of several protein-coding genes, Geiser et al. (10) failed to find any evidence that A. flavus and A. oryzae are independent species and suggested that A. oryzae has evolved by domestication from A. flavus. Other genetic attempts, by using Southern hybridization analyses, have been made to classify the Section Flavi strains with other genes as targets, such as restriction fragment length polymorphisms of nuclear and mitochondrial DNAs (21, 22), and sequence divergence of aflR genes involved in the aflatoxin pathway (3). The above-described methods have all provided useful information on the phylogenetic relationships among species of Aspergillus Section Flavi. However, there may be disadvantages in these genetic techniques in that they are time-consuming and complicated due to nucleotide sequencing, the emergence of many bands, isotope use, and cumbersome preparation of DNA.

Internal transcribed spacer (ITS) regions ITS1 and ITS2, which evolve relatively faster than the coding regions of the rRNA gene, are present in tandemly repeated units and easily isolated by PCR amplification (12). Hence, such characters are useful for the classification of closely related organisms (9, 11, 24, 25). By using ITS regions (ITS1-5.8S-ITS2), we previously found that single-strand conformation polymorphism (SSCP) analysis assisted in the morphological identification of Section Flavi strains (17). This analysis effectively differentiates the species of Section Flavi but is applicable to only the strains of Section Flavi for the following reasons. The PCR products amplified with general primers are not section specific; the mobility of single-stranded DNA cannot be estimated because DNA mobility in gel is dependent on its secondary structure and its interaction with the gel matrix. Since the specificity is limited, some other fungi might show mobility indistinguishable from that of Section Flavi species.

Heteroduplex analysis allows genetic screening of unknown DNA fragments in a sequence-dependent manner. Briefly, similar but not identical DNA fragments obtained form heteroduplexes by their denaturation and reannealing. Unpaired and mismatched nucleotides affect the DNA structure and lower the electrophoretic mobility of the heteroduplexes by sequence divergence (5). This analytical method has been used in medical genetics (8) and to clarify the evolution of organisms including viruses (5, 16) and bacteria (19).

The purpose of the present study was to establish heteroduplex analysis for not only routine identification of Aspergillus Section Flavi species without using morphological techniques but also sensitive detection of genetic diversity within species. By forming interspecific heteroduplexes, heteroduplex panel analysis (HPA) markedly improved the discriminating power of sequence variability to as little as 1 bp.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Fungal strains. A total of 94 Aspergillus Section Flavi strains, including 30 A. flavus, 13 A. oryzae, 16 A. parasiticus, 14 A. sojae, 10 A. tamarii, and 11 A. nomius strains, were used in this study. The origins, designations, and aflatoxin production of 68 of the 94 strains were described previously (17), and only the strain names are listed in Table 1. The 28 additional Section Flavi strains, 12 non-Section Flavi strains, and 5 Penicillim strains used are listed in Table 2. All strains were maintained on potato dextrose agar slants.

DNA extraction and PCR amplification. DNA extraction and PCR amplification analysis were performed as described previously (17). PCR conditions were slightly modified. After addition of 7% dimethyl sulfoxide to a reaction mixture (25 µl), PCR amplification was performed in a DNA thermal cycler (model 2400; Perkin-Elmer Cetus Corp., Norwalk, Conn.) under the following conditions: an initial denaturation at 94°C for 5 min, followed by 7 cycles of 94°C for 1 min, 40°C for 1 min, and 72°C for 1 min; 25 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min; and a final extension at 72°C for 5 min. PCR amplification achieved higher yields by addition of 7% dimethyl sulfoxide to the reaction mixture.

HPA. Reference fragments for HPA were prepared by PCR amplification of the ITS regions of four strains: A. flavus P 39-1, A. parasiticus IFO 4082, A. tamarii A0754, and A. nomius ATCC 15546. All strains were previously characterized by PCR-SSCP analysis (17). Three microliters of the tested fragments, amplified from a total of 108 strains used, was mixed with an equal volume of each reference fragment and 1 µl of gel loading buffer (FMC BioProducts, Rockland, Maine). To form heteroduplex molecules, the mixtures were denatured at 94°C for 5 min, followed by slow cooling to 20°C for 10 min in a model PJ2000 DNA thermal cycler (Perkin-Elmer Cetus Corp.). The resulting heteroduplexes were applied to a hydrolink mutation detection enhancement (MDE) gel (FMC BioProducts) in accordance with the manufacturer's instructions. Electrophoresis was performed in 0.6× TBE buffer (53.4 mM Tris-acetate and 1.2 mM EDTA) at 300 V for 2.5 h at a constant temperature of 20°C in a minislab gel apparatus (gel size, 90 by 80 by 0.75 mm; AE-6410; ATTO, Co., Tokyo, Japan). The gels stained with ethidium bromide were scanned with a video capture apparatus (AE 6915, Printgraph; ATTO, Co.), and then images (NIH image, version 1.62; http://rsb.info.nih.gov/nih-image/) were stored on a Macintosh computer.

Nucleotide sequencing. After digestion of the amplified DNA (approximately 600 bp) with EcoRI, the resulting two fragments (approximately 300 bp) treated with a DNA-blunting kit (Takara Shuzo Co. Ltd., Kyoto, Japan) were ligated into the HincII and EcoRI sites of the pUC18 vector. The ligated DNA was transformed into Escherichia coli XL1 Blue. DNA double digested with EcoRI and PstI was electrophoresed on agarose gel to confirm that the cloned inserts were approximately 300 bp. At least three clones containing each digested fragment were selected based on the presence or absence of the EcoO109I site, which was identified by restriction fragment length polymorphism analysis, as reported previously (17). The selected clones were purified with Q Sepharose fast flow (Pharmacia LKB Biotechnology, Uppsala, Sweden). Double-stranded sequencing of each clone was performed with an ALF autoread sequencer (Pharmacia LKB Biotechnology) by using an autoread sequencing kit (Pharmacia LKB Biotechnology). Sequence information was used only when two or more independent clones showed the same sequences.

Phylogenetic analysis. The DNA sequence data obtained in this study were aligned with the Clustal W program, version 1.6 (13), followed by manual adjustment. The PHYLIP 3.5 package (7) was used for phylogenetic analysis. A distance matrix was generated by the two-parameter method of Kimura (14) in the DNADIST program. Phylogenetic trees were then constructed by the neighbor-joining (NJ) method (NEIGHBOR program) and the maximum-likelihood (ML) method (DNAML program). SEQBOOT was used to generate 100 bootstrapped data sets.

Nucleotide sequence accession numbers. The ITS1-5.8S-ITS2 sequence data of 10 strains reported in the present study have been submitted to the DDBJ database and assigned accession no. D84353 to D84358 and AB000533 to AB000536, inclusive.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification by HPA panels. Of 42 HPA panels that were generated from 29 A. flavus and 13 A. oryzae strains of various origins, 31 panels (20 of A. flavus and 11 of A. oryzae) matched the F-1 standard panel, while the other strains showed a new HPA pattern. The panels of nine A. flavus strains (RIB 1410, ATCC 28539, OPS 375, OPS 398, OPS 402, OPS 406, OPS 421, OPS 422, and OPS 423) and two A. oryzae strains (IFO 5375 and RIB 1039) showed identical HPA patterns and were designated type F-2. Strains of type F-2 formed a single band with the reference strain of A. flavus (lanes f in panel F-2 of Fig. 1), as did those of type F-1. In pairing with other reference strains, however, strains of type F-2 displayed heteroduplex bands with apparently lower mobility than those of strains of type F-1 (Fig. 1). An additional character used to discriminate type F-2 from type F-1 was the formation of overlapping heteroduplex bands with the A. parasiticus reference strain (lanes p in panel F-2 of Fig. 1). The panels of all 15 A. parasiticus and 14 A. sojae strains matched the P-1 standard panel. All nine panels of A. tamarii strains also matched the T-1 standard panel. Of the 10 HPA panels of A. nomius strains, 5 matched the N-1 standard panel while the other strains yielded 3 different panels (N-2, N-3, and TN-1 in Fig. 1). The new panels of NRRL 6552 and IMI 358751 (designated N-2 and N-3, respectively) were identical to those of IMI 358750 and IMI 358752, respectively. These two types were differentiated from type N-1 in that types N-1 and N-2 showed overlapping heteroduplex bands in lanes t and f, respectively. Type N-3 showed two distinct heteroduplex bands in lanes f, p, and t. Despite the variability shown in A. nomius HPA panels, all nine strains but IMI 358749 (TN-4 type) formed a single band with the reference strain of A. nomius. One strain, IMI 358749, displayed unique HPA panels designated type TN-1. IMI 358749 formed heteroduplex bands with mobility shifts even when reannealed with the other A. nomius strains (data not shown).

View larger version (47K): [in this window] [in a new window]   FIG. 1.   HPA panels of Aspergillus Section Flavi strains obtained by using ITS-region DNAs amplified by PCR. The left panels of reference strains P-39-1, IFO4082, A0754, and ATCC 15546 are the standard panels of F-1, P-1, T-1, and N-1, respectively. On the right are the test panels of 20 strains that are indicated above the panels. F-2, N-2, N-3, and TN-1 are additional new panels of heteroduplex types. Heteroduplex bands in lanes f, p, t, and n were formed with reference strains of A. flavus, A. parasiticus, A. tamarii, and A. nomius, respectively.

View larger version (59K): [in this window] [in a new window]   FIG. 2.   HPA panels of A. niger IFO 8340, A. fumigatus OPS 43, and Penicillium expansum IFO 8800 obtained by using ITS region DNAs amplified by PCR. Heteroduplex bands in lanes f, p, t, and n were formed with reference strains of A. flavus, A. parasiticus, A. tamarii, and A. nomius, respectively.

Nucleotide sequence analysis. To verify the findings of HPA analysis, the PCR products sequenced from 24 strains were as follows: four references used in this study and an additional 13 strains of standard types, A. flavus strains of type F-1 (RIB 1427 and S-21), A. oryzae strains of type F-1 (IFO 30113 and HIC 6154), A. parasiticus strains of type P-1 (P-12-2 and IFO 4082), A. sojae strains of type P-1 (IFO 4200 and IFO 5241), A. tamarii strains of type T-1 (RIB 3005 and SP-27-3), A. nomius strains of type N-1 (OPS 372, OPS 387, and OPS 401); seven strains of new types identified by HPA, A. oryzae IFO 5375 (type F-2), A. flavus ATCC 28539 (type F-2), and A. nomius strains of type N-2 (NRRL 6552 and IMI 358750), of type N-3 (IMI 358751 and IMI 358752), and of type TN-1 (IMI 358749). The ITS regions amplified by PCR ranged from 594 to 598 bp. The nucleotide sequences of ITS1 (Fig. 3A) and ITS2 (Fig. 3B) were aligned for comparison with A. flavus P-39-1 (type F-1). The coding regions, the 3' end of the 18S genes (11 bp), the 5.8S genes (157 bp), and the 5' end of the 28S genes (38 bp), were identical among all of the strains examined (data not shown). ITS1 and ITS2 sequences of A. flavus strains of type F-1 and A. parasiticus strains of type P-1 were all identical to those of A. oryzae strains of type F-1 and A. sojae strains of type P-1, respectively.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   Nucleotide sequence alignments for ITS1 (A) and ITS2 (B) of rRNA genes from Aspergillus Section Flavi strains. F-1, A. flavus P-39-1; F-2, A. oryzae IFO 5375; P-1, A. parasiticus IFO 4082; T-1, A. tamarii A0754; N-1, A. nomius ATCC 15546; N-2, A. nomius NRRL 6552; N-3, A. nomius IMI 358751; TN-1, A. nomius IMI 358749. Dots indicate identity to the sequence of F-1. Dashes indicate alignment gaps (insertion or deletion differences). Boxes A to E indicate variable regions.

Sequence diversity and relative mobilities of heteroduplex bands. The ITS regions of eight types had interspecific diversities ranging from 5 to 21 bp (Table 3), which were detected as mobility shifts of heteroduplex bands on an MDE gel (Fig. 1). Intraspecific diversity was shown in A. nomius (1 to 8 bp) and A. flavus/A. oryzae (only 1 bp) but not in A. parasiticus/A. sojae or A. tamarii. Type TN-1 A. nomius had the highest intraspecific diversity (8 bp) and the lowest interspecific diversity from A. tamarii (5 bp). Except for type TN-1, intraspecific diversity (1 to 3 bp) did not display any heteroduplex bands with mobility shifts (lane f of F-2 and lanes n of N-2 to N-3 in Fig. 1), while other interspecific heteroduplexes (lanes p, t, and n of F-2 and lanes f, p, and t of N-2 and N-3 in Fig. 1) displayed mobility shifts apparently different from those of the corresponding standard panels.

View larger version (16K): [in this window] [in a new window]   FIG. 4.   Relative mobilities of heteroduplexes and pairwise nucleotide diversity of ITS regions among the eight heteroduplex types shown in Table 2. They were calculated as the average distance of migration of the two heteroduplex bands divided by the distance of migration of the homoduplex bands. The values were then plotted against nucleotide diversity. Differences in relative mobility shown in Fig. 1 between F-1 () and F-2 () are indicated by the arrows: p, A. parasiticus; t, A. tamarii; n, A. nomius.

Phylogenetic analysis. From the data on the ITS1 and ITS2 sequences from the six species, a phylogenetic tree was constructed by the NJ or ML method to show the genetic relationships among Aspergillus Section Flavi species. The sequence data of ITS1 and ITS2 from A. wentii (EMBL database accession no. U03522 and U03523, respectively) served as an outgroup. The rooted NJ tree (Fig. 5) was topologically identical to the ML tree (data not shown) in major respects. Bootstrap analysis strongly supported an A. flavus/A. oryzae-A. parasiticus/A. sojae clade (NJ, 98%; ML, 98%), an A. flavus/A. oryzae clade (NJ, 100%; ML, 100%), and an A. nomius clade, except for type TN-1 (NJ, 95%; ML, 94%). The A. tamarii-A. flavus/A. oryzae-A. parasiticus/A. sojae and A. nomius type TN-1-A. nomius type N-1 to N-3 clades had low bootstrap values in NJ and ML trees. When IMI 358749 (type TN-1) was excluded from the analysis, the bootstrap support for the A. tamarii-A. flavus/A. oryzae-A. parasiticus/A. sojae clade increased from 60 to 83% in the ML tree but not in the NJ tree. In order of genetic proximity to A. flavus, A. parasiticus was most closely related, followed by A. tamarii and finally A. nomius.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   NJ phylogenetic analysis of eight heteroduplex types of Aspergillus Section Flavi based on an ITS region nucleotide sequence comparison. Codes in parentheses correspond to those presented in Fig. 3. Numbers indicate bootstrap frequencies (percent) based on NEIGHBOR and DNAML analyses (in parentheses), respectively. The bootstrap value of 60% was increased to 83% when IMI 358749 (type TN-1) was excluded from the ML analysis. The scale indicates 1.0% divergence.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The HPA presented herein is a novel system that enables accurate identification of the species of Aspergillus Section Flavi and subdivision based on highly sensitive discrimination of sequence variability. Gaps lead to the formation of a bulge in the heteroduplex, bend the DNA axis, and affect the mobility shift much more than do simple substitutions (bubble) (2). However, none of the intraspecific heteroduplexes that contained even a gap or two substitutions displayed any mobility shift under the present HPA conditions in a minislab MDE gel with a shorter distance (ca. 8 cm). These disadvantageous conditions contributed to identification of the strains to the species level. The formation of interspecific heteroduplexes between strains of different species significantly improved the ability of HPA to detect a single-base substitution or a single-base gap. The highest sensitivity must have been due to the bulge in heteroduplexes resulting from the gap variation in ITS sequences because of the regions not coding proteins. For example, F-1 and F-2 interspecific heteroduplexes formed with P-1 had a marked difference in mobility shifts, although the sequence difference between F-1 and F-2 was only a single-base (T) gap. This can be explained by the increase in bulge size, 2 bp (AT) to 3 bp (ATT), by only one additional gap. It has been suggested that the magnitude of DNA bending depends on the bulge size (2, 20), and the ATT bulge disrupted the flanking of TA and GC base pairs and affected the AC bubble near the bulge (1). From these presumptions, the conformational change of DNA in the heteroduplexes formed between P-1 and F-2 was much greater than that between P-1 and F-1. This might explain the easy detection of the 1-bp diversity between types F-1 and F-2. One- or two-bp diversity among A. nomius types was detected by smaller mobility shifts of interspecific heteroduplexes. In such cases as the interspecific heteroduplexes formed with type T-1, interspecific heteroduplexes of N-1, N-2, and N-3 were discriminated from one another, although each intraspecific heteroduplex was not. The largest difference between inter- and intraspecific heteroduplexes was the presence in the former of a larger number of concurrent mismatches containing bubbles and bulges at common positions in the ITS regions. This suggests that such mismatches act as a cofactor to enhance minor conformational differences among three heteroduplexes introduced by base effects of the bulge and the bubble.

Two heteroduplex types, F-1 and F-2, created by HPA were common in the A. flavus and A. oryzae strains tested, and thus, there was no association with aflatoxin production. Such intraspecific diversity and the conspecificity found in A. parasiticus and A. sojae were confirmed by subsequent nucleotide sequencing. Early genetic approaches (6, 10) with several protein-coding genes as targets have failed to identify substantial interspecific diversity between A. flavus and A. oryzae. These genetic findings have therefore provided no justification for maintaining the industrial fungi A. oryzae and A. sojae as individual species even if there are subtle morphological differences (15). If the industry requires discrimination of each pair, additional tests, such as random amplification of polymorphic DNA (26), will have to be done.

Greater intraspecific diversity of A. nomius, 1 to 3 bp, was accurately detected by HPA, and it created three heteroduplex types in the 10 strains of A. nomius, except IMI35849, examined in the present study. The diversity is also evident in Taka amylase A gene analysis (6), by which seven of the A. nomius strains, except IMI35849, were divided into three groups. These findings were in agreement with the DNA complementarity studies of Kurtzman et al. (18), who found relatively lower homology within A. nomius strains (85 to 100%).

IMI 35849 (type TN-1), which HPA failed to identify, also showed an unusual SSCP pattern and morphological atypicalness (4), as reported previously (17). The heteroduplexes formed with A. tamarii displayed the fastest migration bands (relative mobility, 0.96). Such a high value indicated a closer relationship between the two strains, since the relative mobilities detected in the present study were correlated with nucleotide diversity (Fig. 4). Such information was consistent with the phylogenetic tree illustrated in Fig. 5, in which IMI 358749 was most closely related to A. tamarii and formed an independent clade separated from that of the A. nomius group (95 or 94% bootstrap support). Based on these genetic and morphological characters, it is suggested that IMI 358749 is an intermediate strain between A. tamarii and A. nomius.

Cross-genus or -species formation of heteroduplexes was observed in HPA of Penicillium and non-Section Flavi Aspergillus strains. Such fungi, however, could easily be identified as a species different from Aspergillus Section Flavi (relative mobility, 0.67 to 0.96) by their apparently lower mobility (relative mobility below 0.52) and poor formation of heteroduplex bands. This was due to the lower DNA homology of the ITS regions between the fungi tested and the Section Flavi strains that served as references and which indicates that our HPA is more specific to section Flavi than is the SSCP method (17). HPA also identified several industrial strains as A. oryzae or A. sojae that had lost the ability to form conidia because of repeated transfers for many years. The discriminating power of HPA suggests that it allows direct identification of the Section Flavi isolates without morphological observation. This may also be the case with nonliving fungi. It may also be possible to detect cryptic species in morphologically atypical isolates of Section Flavi.

ITS sequences are generally known to reflect phylogenetic relationships very well. Indeed, there were disagreements between the ITS-based phylogenetic relationships and traditional taxonomic arrangements in some fungi such as Fusarium (25), Trichoderma (11), and Trichophyton (24). The HPA typing of ITS regions we developed coincided well with the taxonomy of Section Flavi. In addition, the time to identification from a small amount of fungal mycelia could be 1 day, including DNA extraction, amplification, and electrophoresis, whereas morphological methods require at least a week for cultivation. HPA must therefore be applicable and practical as an alternative to the conventional method (23).

In conclusion, ITS1 and ITS2 genes contain sufficient interspecific and minimal intraspecific sequence diversity, allowing identification of the species of Aspergillus Section Flavi by HPA. This method is rapid and sensitive and requires minimal technological expertise, in contrast to conventional morphological methods. HPA typing also coincided well with the taxonomy of the Section Flavi and it must therefore be applicable as an alternative to the traditional method.