INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pseudallescheria boydii (Shear) McGinnis et al. [anamorph, Scedosporium apiospermum (Sacc.) Sacc.] is one of the emerging agents of opportunistic mycoses in humans. Twelve case reports for hospitalized patients suffering from myeloid leukemia or subjected to immunosuppressive therapy were published in 1999 alone. Infections are difficult to treat because of its resistance to common antifungal drugs such as amphotericin B. The diagnostic problems with this species are considerable. In tissue P. boydii is indistinguishable from other filamentous fungi (20), and Cryptococcus antigen testing may lead to false-positive reactions (23). Systemic infections may easily be misidentified as aspergillosis, leading to inappropriate therapy. This is one of the reasons that the mortality rate due to invasive pseudallescheriasis is high.

Judging from the number of patient isolates received by the Centraalbureau voor Schimmelcultures (CBS) Identification Department, there may still be a considerable underdetection of P. boydii (9), probably due to the species' clinical and morphological diversity. In the past, etiologic agents were frequently introduced as new taxa, as their identity with existing species was not recognized. Still, the literature contains a number of "ghost taxa" in such divergent genera as Acremonium, Actinomyces, Cephalosporium, Graphium, Madurella, and Verticillium that may be identical to P. boydii (8).

The anamorph of the species was introduced more than 100 years ago by Harz and Bezold (see references 27 and 28) as an agent of human otitis. Since the 1920s it became known as one of the major agents of mycetoma (26) and other subcutaneous infections. During the last few decades, rhinopharyngeal and pulmonary colonization was reported in leukemic, cystic fibrosis, and otherwise impaired patients (35). In disseminated cases the species is significant because of its neurotropism, with marked predilection for the cerebrospinal fluid (12). Such infections were sometimes acquired after near-drowning events in polluted waters (24).

Much of the observed morphological variability can be ascribed to variable abundance of (syn)ana- and teleomorphs. This was confirmed by the investigation of characters independent of morphology, such as nutritional physiology (10) and 18S ribosomal DNA (rDNA) sequencing (16). Wedde et al. (36), using rDNA ITS2 sequencing, were able to select primers for species recognition. However, infraspecific molecular variability of the internal transcribed spacer (ITS) rDNA of P. boydii seems to be larger than that of species such as Trichophyton rubrum (Castell.) Sab. (14), Hortaea werneckii (Horta) Nishimura et Miyaji (38), or Cladophialophora bantiana (Sacc.) de Hoog et al. (13). Previously, nuclear DNA (nDNA) homology studies had indicated the existence of three infraspecific groups in P. boydii (10, 15), and the clinical pictures and environmental sources of isolation of these three groups broadly matched. Bell (2) had found differences in virulence between a strain from patients with subcutaneous infections and one from the environment. Thus, P. boydii may comprise entities with different pathogenicities.

In the study described in the present paper, the molecular variability of P. boydii was evaluated. This should help to establish a clear species concept which is required for reliable molecular identification of this opportunist in the routine laboratory. To this aim we compared the published nDNA-DNA homology data with results of small subunit (SSU) restriction analysis and sequencing and of fingerprinting by PCR (primer M13). The strains of P. boydii compared included isolates that originated from a single source (patient or environmental), as well as a worldwide selection of strains collected over a period of 70 years (Table 1). Some closely related taxa are included for comparison.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

DNA extraction. The methods applied for DNA extraction were described previously (13).

Restriction analysis. Amplification of the 18S rDNA was performed with primers Oli4 and NS24 (31). The resulting amplicons were digested with the restriction endonucleases HaeIII, HinfI, DdeI, RsaI, TaqI, MspI, and HhaI (Amersham/Pharmacia Biotech) under the conditions recommended by the manufacturer. The corresponding products were electrophoresed in 1.5% agarose gels at 150 V for 2 to 3 h. The fragments were analyzed by computer-aided image analysis (ImageMaster; Pharmacia Inc.) and were verified by comparison with the expected patterns on the basis of the sequences of the same or related strains.

Sequence determination and analysis. The ribosomal ITS region was amplified with primers V9G and LS266 (13). Both strands were sequenced with the internal primers ITS2, ITS3, ITS4, and ITS5 and the Big Dye terminator cycle sequencing kit (PE Applied Biosystems, Warrington, United Kingdom), as recommended by the manufacturer, combined with an ABI automatic DNA sequencer. Sequences were sampled with the Seqman package (DNAStar Inc., Madison, Wis.) and were aligned with BioNumerics software (Applied Maths, Kortrijk, Belgium). The distance tree was constructed with the neighbor-joining algorithm with the Kimura correction in the Treecon package (32). The robustness of the branches was assessed by bootstrap analysis with 100 replicates. The topology of the tree was verified with several algorithms including the parsimony algorithm.

PCR fingerprinting. The core sequence of phage M13 was used as a single primer in the PCR experiments. Amplification reactions were performed as described by Weising et al. (37) in an Amplitron II thermocycler. For each reaction 10 ng of DNA was used. Forty PCR cycles were programmed as follows: 20 s at 93°C, 60 s at 50°C, and 20 s at 72°C, with elongation at 72°C for 6 min and chilling to 4°C. The PCR products were electrophoresed on 1.5% agarose gels at 100 V for 4 to 6 h. The gels were stained with ethidium bromide and photographed under UV light. The DNA fragment profiles were analyzed with the help of Gelcompar software (Applied Maths).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Restriction analysis. Data obtained by restriction fragment length polymorphism (RFLP) analysis of the 18S rDNA gene are summarized in Table 2. Amplicons varied in length because of the occurrence of inserts of about 380, 470, 920, or 1,500 bp. Only restriction patterns that resulted from the digestion of 18S rDNA genes of the same length are comparable. Twenty-two strains had no inserts. Ten such strains, identified as P. boydii, yielded identical patterns when the same enzymes were used; the exception was RKI 866/94 digested with MspI. The patterns of strains of nDNA reassociation groups 1 to 3 were identical. The patterns of type strains of Pseudallescheria africana (von Arx et G. Franz) McGinnis et al., Pseudallescheria angusta (Malloch et Cain) McGinnis et al., Pseudallescheria fusoidea (von Arx et al.) McGinnis et al., and Pseudallescheria ellipsoidea (von Arx et Fassatiová) McGinnis et al. were identical to those of P. boydii. In contrast, strain CBS 100.26, maintained in the CBS culture collection as P. boydii, was deviant with all enzymes used. Graphium tectonae C. Booth and Pseudallescheria fimeti (von Arx et al.) McGinnis et al. deviated from P. boydii in HaeIII restriction patterns. Four strains of Scedosporium prolificans (Hennebert et Desai) Guého et de Hoog had the same profiles only with MspI and TaqI; they were identical to each other except for the HinfI pattern for CBS 467.74. Petriella guttulata Barron et Cain and Petriella lindforsii Curzi showed significant differences with most enzymes.

Sequencing. For the distance trees presented in Fig. 1 and 2 we included ITS1 and ITS2 sequences published by Wedde et al. (36) and Lennon et al. (18). Sequences could not always be aligned with confidence, particularly the sequence of P. boydii CBS 100.26 (Acladium castellanii Pinoy), which was excluded from further analysis, and a group around Petriella setifera, which could be only partially aligned, as indicated in Fig. 1. Separate partial and complete ITS domain analyses were performed with and without this group. Table 3 summarizes the numbers of base substitutions in the ITS domains by taking P. boydii CBS 330.93 as a reference. Within the entire group compared, a maximum of 58 positions in ITS1 were variable, 79 positions in ITS2 were variable, and 1 position in the 5.8S rDNA gene was variable.

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Phylogenetic tree of species of the family Microascaceae studied. The tree was constructed on the basis of confidently aligned positions of the rDNA ITS domain. The tree was generated with the Treecon package with the neighbor-joining algorithm and Kimura correction. G. calicioides was used as the outgroup. The tree was subjected to 100 bootstrap replications; only values >90 are shown. G, Graphium; P, Pseudallescheria; Pe, Petriella; R, Rhinocladium; S, Scedosporium; (T), ex type strain; 1 to 3, nDNA homology groups (15).

View larger version (31K): [in this window] [in a new window]   FIG. 2.   Phylogenetic tree of P. boydii, constructed as described for the tree in Fig. 1. S. prolificans CBS 114.90 was used as the outgroup.

Fingerprinting with M13. Examples of the banding patterns resulting from PCR fingerprinting are shown in Table 4. Profiles were found to be highly heterogeneous. Almost every strain had its own pattern; very few bands could be matched with confidence. All tests were repeated several times and proved to be reproducible. The fingerprints of the type strains of P. africana (CBS 311.72) and P. ellipsoidea (CBS 418.73), having somewhat deviant ITS sequences, proved to be identical to each other. Clinical strains IP 1698.87 and IP 1945.90 were also identical. The two strains from a single patient, CBS 329.93 and CBS 330.93, had the same patterns (data not shown), whereas strain CBS 101720, from the site of the accident but with another ITS sequence, proved to have different patterns.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

P. boydii is a cleistothecial ascomycete that belongs to the family Microascaceae. This corresponds to 18S rDNA sequencing data (16), which also indicated that Microascus cirrosus Curzi is a close relative. In a similar study, Okada et al. (21) found three Graphium species among the members of the family Microascaceae. From our more detailed ITS data, G. putredinis was found to be close to G. tectonae, while Graphium penicillioides Corda took an isolated position amid the Petriella species.

Phenetically the genera Petriella and Pseudallescheria are distinct by having ostiolate versus nonostiolate ascomata, respectively (33). Our ITS1 and ITS2 distance tree (Fig. 1) does not show a clear bipartition. Most of the nonostiolate species are found in the upper part of the tree, and the ostiolate species are found in the lower branch of the tree. Issakainen et al. (16) supposed that S. prolificans is close to P. setifera, but in our study S. prolificans proved to be clearly different. No teleomorph has yet been found for S. prolificans.

von Arx et al. (33) distinguished seven species in the genus Pseudallescheria. Of these, only P. fimeti was found to have clearly different ITS sequences. P. angusta, P. ellipsoidea, and P. fusoidea were found within the P. boydii main group and must be regarded as synonymous names, as already supposed on the basis of morphological studies by McGinnis et al. (19) and on the basis of the ITS2 sequences by Wedde et al. (36). The species have been distinguished primarily on the basis of sizes of the cleistothecia and ascospores, but ranges of variability were strongly overlapping. The positions of P. desertorum and P. africana remain ambiguous, since their sequences differed from that of the type strain of P. boydii, CBS 101.22, by 12.8 and 6.8%, respectively.

The patterns of the ITSs of strain CBS 100.26, included in the CBS List of Cultures as P. boydii, obtained by RFLP analysis with SSU were markedly different from those of the remaining species; the ITSs could not be aligned. Earlier, it was found to have a clearly different mole percent G+C DNA content (15). The strain is of the Acladium castellanii type and was described from a patient with subcutaneous, suppurative mycosis (5). The fungus was described as having yellowish cultures which became blackish on some media; conidiogenous cells were aggregated in fascicles and produced apical clusters of truncate conidia (4). Butler (4) supposed that it was related to Sporothrix schenckii Hektoen et Perkins, while de Hoog (6) suggested an affinity to Raffaelea on the basis of its truncate sympodial conidia. Both suggestions were refuted on the basis of ITS (data not shown) and 18S rDNA sequence data (M. Blackwell, personal communication; compare these suggestions with those in reference 17), respectively. The taxonomic position of A. castellanii remains uncertain.

Rhinocladium lesnei (34) was found to be identical to G. putredinis and represents an anamorph taxon clearly apart from P. boydii. Morphologically and culturally, the two strains analyzed, CBS 108.10 and CBS 102083, were very different. CBS 108.10 was whitish, showing a Scedosporium anamorph only. However, in the original publication of Vuillemin (34), fasciculate conidiophores and denticulate conidiogenous cells are depicted. Thus, the identity of the two taxa is highly probable.

The genetic variability within P. boydii is considerable. Guého and de Hoog (15) found three infraspecific ecological and clinical groups on the basis of nDNA-DNA reassociation experiments. Reassociation between these groups was consistently about 50%, whereas within the groups the values were >80%. Similar reassociation groups in Galactomyces geotrichum Redhead et Malloch have recently been recognized as separate taxonomic entities (29). Contrary to these findings, two groups in Cladophialophora were judged to belong to different species, despite nDNA-DNA homology values of >80% (13). In bacteria, homology values are more precise than sequence data (30), but this is particularly due to the use of SSU sequences. In fungi, DNA homology values compared to ITS sequence data indicate a level of diversity which is similar or somewhat lower. The P. boydii reassociation groups are not supported statistically in the ITS tree (Fig. 1). It is remarkable, however, that all strains of reassociation group 2 except strain CBS 499.90 contain an intron in the SSU rDNA gene and can be found in a single branch in the ITS tree.

Judging from the data obtained by molecular fingerprinting with M13, the variability within P. boydii is considerable. Nearly all strains proved to be different from each other; the exceptions were three pairs (CBS 329.93 and CBS 330.93, CBS 101721 and CBS 101723, and IP 1698.87 and IP 1945.90). The ITS sequences of these pairs were also identical, while for two of the three pairs the sources of isolation were remote from each other. This finding indicates the existence of widespread genotypes. Due to the molecular variability of P. boydii, identification based on species-specific primers or RFLP analysis should be interpreted with care in routine diagnostics.

In the case of strains CBS 329.93, CBS 330.93, and CBS 101720, molecular fingerprinting with M13 proved to be an appropriate method for characterization of individual strains, since identical fingerprinting patterns were generated for the strains isolated from a single patient. Data obtained by molecular biological analysis for all strains from the same nonclinical isolation site (strains CBS 499.90 and CBS 101719 and strains CBS 101725 and CBS 101726) were different (Fig. 1 and 2; Tables 2 to 4). For one pair of strains (CBS 101725 and CBS 101726), the strains were different as determined by culture, and their ITS sequences deviated to such an extent that they may be two different species (Fig. 1). These findings are in agreement with those of April et al. (1), who found P. boydii strains from closely similar sites to be highly variable in cultural and physiological parameters.

Apparently, environments which are suitable for the growth of P. boydii are consistently inhabited by different populations. The small amount of correspondence between fingerprinting bands indicates that variation is continually generated, probably by meiotic recombination. However, the variability of the ITS sequences of P. boydii strains exceeds that of intermating populations. This may be a reason why such a broad spectrum of clinical pictures caused by P. boydii occurs, and a tendency to neurotropic colonization has evolved. The distance between CBS 101725 and CBS 101726 from a single sample is particularly illustrative. It is unclear why different populations have been maintained in a single environment and isolates with similar genotypes have been isolated from locations that are separated by significant geographic distances. A population genetic study is required to obtain more insight into speciation and evolution of P. boydii. M13 fingerprints, ITS sequences, and RFLP patterns proved to be more conserved in S. prolificans. Wedde et al. (36) and San Millán et al. (25) showed that variability in ITS sequences and randomly amplified polymorphic DNA analysis data, respectively, for this anamorphic taxon seem low. These results indicate that the life cycle of this species might be reduced to clonal reproduction. In contrast to P. boydii, a teleomorph name that indicates a sexual mode of reproduction is not known for S. prolificans.

Clinical strains are distributed over the entire tree. Infections caused by P. boydii-related species other than the opportunistic species P. boydii and S. prolificans are a case of lobomycosis caused by Petriella setifera in a dolphin (11) (another case of infection in a dolphin was attributed to P. boydii [see also reference 22]) and a human skin infection due to R. lesnei (34). Within P. boydii, the amount of variability of clinical strains is comparable to that of the environmental strains of that species (Fig. 1). This suggests that no particular selection by the host occurs and, thus, that all environmental strains have equal pathogenic potential. The natural ecological niche of P. boydii remains unknown, but nutrient-rich, poorly aerated environments have been described as the ecological niche of P. boydii (10). Ecological data should be taken into account in clinical practice, as they differ considerably between P. boydii and Aspergillus fumigatus. P. boydii has often been isolated from polluted ponds frequented by waterbirds; the fungus possibly occurs as a commensal organism in their intestinal tracts. It may then be expected that P. boydii has a significant potential to be an opportunistic pathogen. Given the fact that the species is particularly controlled by unspecific defense systems (3), primary pathogenicity seems unlikely.