点击显示 收起
Laboratorio de Pesquisas Clínicas, Centro de Pesquisas Rene Rachou, FIOCRUZ, Brasil, and Museum National d'Histoire Naturelle, Paris, France
Departamento de Física, Universidade Federal de Minas Gerais, Minas Gerais, Brazil
Museum National d'Histoire Naturelle, Paris, France
Centraalbureau voor Schimmelcultures, Utrecht, The Netherlands, and Institute of Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, The Netherlands
ABSTRACT
Sequences of rRNA gene internal transcribed spacer (ITS) of a standard set of black yeast-like fungal pathogens were compared using two methods: local and global alignments. The latter is based on DNA-walk divergence analysis. This method has become recently available as an algorithm (DNAWD program) which converts sequences into three-dimensional walks. The walks are compared with, or fit to, each other generating global alignments. The DNA-walk geometry defines a proper metric used to create a distance matrix appropriated for phylogenetic reconstruction. In this work, the analyses were carried out for species currently classified in Capronia, Cladophialophora, Exophiala, Fonsecaea, Phialophora, and Ramichloridium. Main groups were verified by small-subunit rRNA gene data. DNAWD applied to ITS2 alone enabled species recognition as well as phylogenetic reconstruction reflecting clades discriminated in small-subunit rRNA gene phylogeny, which was not possible with any other algorithm using local alignment for the same data set. It is concluded that DNAWD provides rapid insight into broader relationships between groups using genes that otherwise would be hardly usable for this purpose.
INTRODUCTION
Black yeast-like fungal pathogens are gaining importance because of their increasing incidence in often life-threatening infections in immunocompromised patients, sometimes also in otherwise healthy persons (18). The taxonomy and identification of these fungi is difficult due to a lack of phenetic characters and high degree of morphological plasticity. With the advent of molecular taxonomy, numerous cryptic taxa are being recognized in Capronia (24), Phialophora (3, 4), Cladophialophora (17), Exophiala (27, 5), and Ramichloridium and Rhinocladiella (5).
Several comparative molecular techniques have been applied to display polymorphism among strains of black yeasts. Mitochondrial DNA (13) or mitochondrial cytochrome oxidase have been used for diversity at the population level. Chitin synthase (12), in contrast, has insufficient resolution to recognize species but rather displays phylogenetic relationships between species aggregates. The newly described species mentioned above were invariably supported by sequence data of the internal transcribed spacer (ITS) rRNA gene region. Marked ITS differences can sometimes be found in groups that are otherwise monomorphic, such as Ochroconis (G. S. de Hoog and H.-J. Choi, unpublished data). However, the taxonomic value of rRNA gene ITS has been questioned, as it does not provide a sufficient level of resolution in Trichoderma (15), Alternaria (11, 19), and other genera. It is apparent that ITS divergence rates cannot be used as a gold standard for taxonomic differences all over the fungal kingdom, but particular levels of divergence have been reached in some groups, among which are the black yeasts and their allies. In a number of cases (4, 5, 27) the entities found were found to coincide with phenetic characters that had been neglected or overlooked in earlier taxonomic systems, underlining the taxonomic predictivity of ITS characters.
ITS divergence between black yeast species is marked, species being clearly separated by at least 1% sequence diversity (3). A major drawback of this ITS diversity in black yeasts is the large degree of divergence when the entire ascomycete family Herpotrichiellaceae, to which they belong through Capronia teleomorphs (24), is compared. Such comparisons may be necessary, because morphology does not provide a priori clues to phylogenetic positions (9). Over larger phylogenetic distances, large parts of the ITS region may not be alignable with confidence (20) and therefore have been excluded from the analysis.
Recently, the algorithm DNA-Walk Divergence (DNAWD) has become available (14). The algorithm's robustness relies on global alignment of complete sequences, which resolves the arbitrariness often associated with the precise identification of indel mutation sites, while the integral nature of the DNA-walk still allows for the sensitivity in their occurrence. This allows for comparison of entire ITS sequences including a broad taxonomic distance range. It can be shown that a properly defined DNA-walk is the exact equivalent to the sequence composition, and therefore mutations and dislocations produce divergences in the walk geometry which can be measured. In this paper the DNAWD program is applied to the phylogenetic reconstruction of black yeast strains, exploring total ITS domains of the rRNA gene. A small-subunit (SSU) rRNA gene tree is used to confirm grouping as given by ITS trees.
MATERIALS AND METHODS
Species. A model set of strains of selected species was compared using ITS regions of the rRNA gene (Table 1). Species included were selected in such a way that both large and small phylogenetic differences, within and between genera, were available in the data set. The identity of the species was verified using a large ITS and phenetic database available at the Centraalbureau voor Schimmelcultures, Utrecht, which contains numerous strains of all existing and recently described species, based on ex-type strains. Each species was also represented in a tree based on SSU rRNA gene sequences.
Sequence verification. The edited sequences all have been deposited in GenBank. The basic structure of the tree based on ITS and generated with conventional methods was verified by near-complete SSU rRNA gene sequencing of a smaller selection of strains of the same species (Table 1) plus 70 additional members of the family Herpotrichiellaceae. SSU sequences were aligned and analyzed in an aligned database containing about 3,000 fungal sequences in the ARB package developed by W. Ludwig (www.mikro.biologie.tu-muenchen.de/pub/ARB). The SSU tree was made with the Parsimony/ML algorithm with 100 bootstrap replications; Phaeomoniella chlamydospora (CBS 101359) was taken as outgroup.
ITS sequences were aligned using the BioNumerics package (Applied Maths, Kortrijk, Belgium). Due to gaps necessary for alignment, the ITS 1 domain spanned 258 positions (real lengths, 197 to 215 bp), the 5.8S gene 157 positions, and the ITS2 domain 224 positions (real lengths, 171 to 211 bp). Positions 97 to 155 and 184 to 208 (ITS1) and 456 to 484 and 541 to 631 (ITS2) could not confidently be aligned and were excluded from most of the comparisons. The same alignment was used for a comparison with different packages applying different algorithms: Phylip package (v.3.572c) (6), according to maximum likelihood; BioNumerics according to Ward's averaging (27), without Indels, using Kimura's two-parameter model, distance estimation through neighbor joining using the Treecon package with Kimura's two-parameter model (25), and Paup (23) according to parsimony. In all comparisons, insertions and deletions (indels) were not taken into account. Exophiala pisciphila (AF050273) was taken as the outgroup.
Homology analysis by DNA-walk divergence. DNA-walks are defined by incrementing walk steps for each nucleotide in the sequence (for example, a positive step for purines and negative for pyrimidines). The DNA-walk divergence method (14) makes simultaneous three-dimensional walk comparisons (representing three composition skews), AG-TC, AC-TG, and AT-CG for each pair of sequences. One sequence slides against the other until the minimum squared walk difference is found, corresponding to a global alignment. This is then taken as a measure of their distance since statistically independent mutations and indels increase the mean square walk differences linearly. The resulting distance matrices are then fed to the Kitsch program of the Phylip package (v. 3.572c) (6), which generates trees with contemporary leaves.
The analyses were carried out for both the ITS1 and ITS2 regions separately, defined within the consensus limits ATCATTA to CAACAAC and TGTTGGA to AGGTTGA, respectively (Fig. 1a). The choice of the intervals was based on apparent homogeneity of sequence dissimilarities among species and can be appreciated in the DNA-walk graph of Fig. 1b. The lengths and average dissimilarities of the sections analyzed are also given in Fig. 1a.
RESULTS
All individual species were distinguishable with ITS sequence data with any of the algorithms applied (Fig. 2, 3, 4, and 5). Using the complete ITS region, a similar unresolved structure with nine clusters (A to I) was found when neighbor joining, parsimony, and Ward's algorithms were used (Fig. 2). The mutual distances of these groups were inferred from an SSU rRNA gene tree (Fig. 3) of the herpotrichiellaceous black yeasts, containing about 1,600 positions in 70 strains. Cluster numbering (1 to 5) in this tree was derived from a similar SSU tree published by Haase et al. (9), groups having Exophiala dermatitidis (in one) Cladophialophora bantiana (in two), Exophiala spinifera (in three), Exophiala nigra (in four), and Coniosporium perforans (in five) as core taxa. ITS groups A to C on the one hand and F plus H on the other appeared to be mutually related as they clustered within single SSU groups, while distances between remaining groups were significantly larger.
With DNA-walk, the trees based on the complete ITS region (Fig. 4) as well as that of the ITS2 domain alone (Fig. 5) recognized all species correctly. The analysis of the ITS domain data resulted in slightly different trees when either the complete ITS region, or ITS1 or ITS2 alone were used. In the DNA-walk tree based the complete ITS domain (Fig. 4) all groups, including A to C and F and H, were clearly separated. The structure of the tree could not be linked to the distances as provided in the SSU tree (Fig. 3). In the tree based on ITS2 (Fig. 5) Fonsecaea (C), Phialophora and Cladophialophora carrionii (A), and C. bantiana (B) were all found in an area of the tree where mutual distances between taxa were moderate, while groups D (E. dermatitidis complex), E, G, and I were more distantly related to other members of the Herpotrichiellaceae and mostly also to each other.
DISCUSSION
Sequencing of the rRNA genes is indispensable for species recognition in black yeasts and their allies that are causative agents of important human diseases such as chromoblastomycosis and phaeohyphomycosis (1, 2, 8, 21). Several species are known to have reproducible intraspecific polymorphisms in the ITS domain enabling the recognition of variants differing in virulence that are easily detected in an aligned ITS sequence base (16). On the other hand, considerable interspecific polymorphisms may occur between morphologically identical species (5), as phenetically similar structures can be found all over the phylogenetic tree of the Herpotrichiellaceae (9). Sequences may show too much divergence to allow reliable alignment comparison, limiting identification of clinical samples to local BLAST searches. nBlast in GenBank for most species is insufficient, since the taxonomic representation is as yet fragmentary. As yet no molecular marker other than ITS is available with optimal display of differences between species. For this reason the best approach is application of a comparative method that is based on ITS but is independent of alignment bias.
Our selection of species and strains analyzed is more or less representative for the hyper-variability of the ITS domains of black yeasts and allies. With growing distances between strains, alignments quickly can no longer be done with confidence. Meaningful comparison to reveal evolutionary relationships between some of the species is then only possible when small selections of the ITS domains are taken into account, as mentioned in Materials and Methods. Phylogenetic interpretation is strongly hampered, as trees lack any resolved substructure, being based on too low number of mutations in the small portions that are still comparable. Due to the large distance between most of the groups, the clusters were robust and remained strictly identical with any algorithm applied, particularly in the lack of a hierarchical structure of main branches (Fig. 2).
A substructure, i.e., with separate species in a more or less hierarchical ordering, was revealed with ITS in the Phialophora/Cladophialophora carrionii cluster (group A) only. When the neighbor joining algorithm was applied to the complete ITS domain, which could be confidently aligned in the mentioned cluster, C. carrionii was found paraphyletically to Phialophora. This matches phenetic data because members of the clade share potential production of phialides and are agents of human chromoblastomycosis. These characters are also found in Fonsecaea (group C). In the SSU tree of the Herpotrichiellaceae published by Haase et al. (9) Fonsecaea has members of Cladophialophora other than C. carrionii as sister group species (cluster 2 of Haase et al.) (9). Cladophialophora species other than C. carrionii cause phaeohyphomycoses rather than chromoblastomycoses.
In our updated SSU tree of the same family (Fig. 3) members of Fonsecaea, Phialophora, and Cladophialophora (A to C) all are united in a single clade (2), showing phylogenetic coherence of the Herpotrichiellaceae with high degrees of virulence. In the DNA-walk tree based on ITS2 (Fig. 5) Fonsecaea, Phialophora, and Cladophialophora are all found in a main subdivision of the tree (II) where distances between species are smaller than the isolated, more distantly related taxa opposite of the dotted line (I) in Fig. 5. The radial scale measures twice the percent dissimilarity in composition. Branches crossing the dotted circle (phenon circle) group about 13 species-aggregates with dissimilarities below 6.3%.
With DNA-walking, the analysis of ITS domain data resulted in slightly different trees when either the complete ITS region (Fig. 4) or ITS2 alone (Fig. 5) was used. All trees recognized all species correctly. The question of whether the structure is phylogenetically interpretable can be verified by a comparison with SSU rRNA gene data of the same fungi. The most comprehensive phylogenetic overview of the family is that of Haase et al. (9). These authors noted that the tree was poorly resolved but recognized five approximate clades, with Exophiala dermatitidis, Cladophialophora bantiana, E. spinifera, E. nigra, and Coniosporium perforans as core species, as listed above. Ecologically these groups seem to be meaningful. Clade 1 is a thermophilic yeast group around the neurotrope E. dermatitidis (10), clade 2 is a group with virulent species with accent on species causing chromoblastomycosis and brain disease (7), clade 3 is the E. spinifera/E. jeanselmei complex (5), clade 4 is a meso- to psychrophilic group of Exophiala species in showers and ocean waters (9), and clade 5 contains meristematic species inhabiting rocks (22). This ecological support for the groups suggests that this reflects the phylogeny of these groups optimally.
In the SSU tree based on our expanded and updated data set (Fig. 3) some further, ecologically defined clades can be located. ITS groups F and G are psychrophilic species similar to the fish pathogen E. pisciphila, although unexpectedly E. salmonis is found in E, which is a cluster of mutually distantly related environmental species. In the SSU tree these taxa are mutually remote (Fig. 3). Exophiala oligosperma is a close relative of E. jeanselmei (5) and indeed is found together with this species in ITS cluster C (Fig. 5), matching the E. spinifera complex SSU group 3 (Fig. 3). Some dispersed strains previously listed as "E. spinifera" (Fig. 3) are now known to represent individual taxa, such as IFM 14855, which has been described as E. nishimurae (26).
With the DNA-walk of the entire ITS domain, no coherent substructure was recognizable. SSU-clade 2/ITS-clades A to C were found at the longest mutual distances (Fig. 4). In contrast, a major subdivision (I and II) was consistently recognized based on ITS2 alone, with E. pisciphila CBS 464.81, representing a distantly related, undescribed species, in an external position (G). Main branches I and II contained five to nine subclusters representing the individual species analyzed (Fig. 5). C. carrionii and Phialophora (2/A) were found to be interrelated, with Fonsecaea (2/C) branching off at the base; all are agents of chromoblastomycosis. Cladophialophora bantiana (2/B) was found as well-individualized clusters paraphyletically to psychrophilic Exophiala species (F).
In conclusion, the DNA-walk showed consistent results in the recognition of individual species, whether these consisted of several strains or a single strain. This was achieved with the complete ITS domain as well as with rRNA gene spacers 1 and 2 separately. In addition, ITS2 alone also enabled phylogenetic reconstruction to some extent, which was not possible with any other algorithm using the same data set. DNA-walk divergence proved to be a powerful tool for the analysis of ribosomal genes and their evolutionary interpretation. The gold standard for phylogenetic interpretations is the small or large rRNA gene subunits, which in general contain more information. However, in the case of the black yeasts and their relatives, these genes often contain degrees of variability too small to resolve individual species (9). DNA-walk divergence bridges the gap between current methods of phylogenetic reconstruction. Another straightforward application of this technique could be the molecular identification of the agents of phaeohyphomycosis and chromoblastomycosis isolated in the clinical laboratory.
ACKNOWLEDGMENTS
Rachel B. Caligorne and Pedro Licinio acknowledge support by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnologico, Brazil).
Bert Gerrits van den Ende and Adile Cegcel are thanked for technical assistance.
REFERENCES
Caligiorne, R. B., M. A. Resende, E. Dias-Neto, S. C. Oliveira, and V. Azevedo. 1999. Dematiaceous fungal pathogens: analysis of ribosomal DNA gene polymorphism by polymerase chain reaction-restriction fragment length polymorphism. Mycoses 42:609-614.
De Hoog, G. S., L. Sigler, W. A. Untereiner, E. Gueho, J. M. Uijthof, and K. J. Chung. 1994. Changing taxonomic concepts and their impact on nomenclatural stability. J. Med. Vet. Mycol. 32:113-122.
De Hoog, G. S., X. O. Weenink, and A. H. G. Gerrits van den Ende. 1999. Taxonomy of the Phialophora verrucosa complex with the description of four new species. Stud. Mycol. 43:107-142.
De Hoog, G. S., P. Mayser, G. Haase, R. Horre, and A. M. Horrevorts. 2000. A new species of Phialophora from human skin. Mycoses 43:409-416.
De Hoog, G. S., V. Vicente, R. B. Caligiorne, S. Kantarcioglu, K. Tintelnot, A. H. G. Gerrits van den Ende, and G. Haase. 2003. Species diversity and polymorphism in the Exophiala spinifera clade containing opportunistic black yeast-like fungi. J. Clin. Microbiol. 41:4767-4778.
Felsenstein, J. 1996. Phylip (Phylogeny Inference Package), v. 3.572. Department of Genetics, University of Washington, Seattle, WA.
Gerrits van den Ende, A. H. G., and G. S. De Hoog. 1999. Variability and molecular diagnostics of the neurotropic species Cladophialophora bantiana. Stud. Mycol. 43:151-162.
Haase, G., L. Sonntag, Y. Van de Peer, J. M Uijthof, A. Podbielski, and B. Krick, 1995. Phylogenetic analysis of ten black yeast species using nuclear small subunit rRNA gene sequences. Antonie Leeuwenhoek 68:19-33.
Haase, G., L. Sonntag, B. Krick, and G. S. De Hoog, 1999. Phylogenetic inference by SSU gene analysis of members of the Herpotrichiellaceae with special reference to human pathogenic species. Stud. Mycol. 43:80-97.
Hiruma, M., A. Kawada, H. Ohata, Y. Ohnishi, H. Takahashi, M. Yamazaki, A. Ishibashi, K. Hatsuse, M. Kakihara, and M. Yoshida. 1993. Systemic phaeohyphomycosis caused by Exophiala dermatitidis. Mycoses. 36:1-7.
Kang, J. C., P. W. Crous, G. R. Mchau, M. Serdani, and S. M. Song. 2002. Phylogenetic analysis of Alternaria spp. associated with apple core rot and citrus black rot in South Africa. Mycol. Res. 106:1151-1162.
Karuppayil, S. M., M. Peng, L. Mendoza, T. A. Levins, and P. J. Szaniszlo. 1996. Identification of the conserved coding sequences of three chitin synthase genes in Fonsecaea pedrosoi. J. Med. Vet. Mycol. 34:117-125.
Kawasaki, M., M. Aoki, H. Ishizaki, et al. 1999. Molecular epidemiology of Fonsecaea pedrosoi using mitochondrial DNA analysis. Med. Mycol. 37:435-440.
Licinio, P., and R. B. Caligiorne. 2004. Inference of phylogenetic distances from DNA-walk divergences. Physica A 341:471-481.
Lieckfeldt, E., and K. A. Seifert. 2000. An evaluation of the use of ITS sequences in the taxonomy of the Hypocreales. Stud. Mycol. 45:35-44.
Matos, T., G. Haase, A. H. G. Gerrits van den Ende, and G. S. De Hoog, 2003. Molecular diversity of oligotrophic and neurotropic members of the black yeast genus Exophiala, with accent on E. dermatitidis. Antonie Leeuwenhoek 83:293-303.
McGinnis, M. R., S. M. Lemon, H. D, Walker. G. S. De Hoog, and G. Haase. 1999. Fatal cerebritis caused by a new species of Cladophialophora. Stud. Mycol. 43:166-171.
Schell, W. A. 1995. New aspects of emerging fungal pathogens: a multifaceted challenge. Clin. Lab. Med. 15:365-387.
Serdani, M., J. C. Kang, B. Andersen, and P. W. Crous. 2002. Characterization of Alternaria species-groups associated with core rot of apples in South Africa. Mycol. Res. 106:561-569.
Simmons, M. P., and J. V. Freudenstein. 2003. The effects of increasing genetic distance on alignment of, and tree construction from rDNA internal transcribed spacer sequences. Mol. Phyl. Evol. 26:444-451.
Spatafora, J. W., T. G. Mitchell, and R. Vilgalys. 1995. Analysis of genes coding for small-subunit rRNA sequences in studying phylogenetics of dematiaceous fungal pathogens. J. Clin. Microbiol. 33:1322-1326.
Sterflinger, K., G. S. De Hoog, and G. Haase. 1999. Phylogeny and ecology of meristematic ascomycetes. Stud. Mycol. 43:5-22.
Swofford, D. L. 2002. Paup. phylogenetic analysis using parsimony. version 4. Sinauer Associates, Sunderland, Mass.
Untereiner, W. A. 2000. Capronia and its anamorphs: exploring the value of morphological and molecular characters in the systematics of the Herpotrichiellaceae. Stud. Mycol. 45:141-149.
Van de Peer, Y., and R. De Wachter. 1994. Treecon for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment. Comput. Appl. Biosci. 10:569-570.
Van Ooyen, A. 2002. Theoretical aspects of pattern analysis, p. 31-45. In L. Dijkshoorn et al. (ed.), New approaches for the generation and analysis of microbial typing data. Elsevier, Amsterdam, The Netherlands.
Vitale, R. G., and G. S. De Hoog. 2002. Molecular diversity, new species and antifungal susceptibilities in the Exophiala spinifera clade. Med. Mycol. 40:545-556.
Wang, L., K. Yokoyama, M. Miyaji, and K. Nishimura. 2001. Identification, classification, and phylogeny of the pathogenic species Exophiala jeanselmei and related species by mitochondrial cytochrome b gene analysis. J. Clin. Microbiol. 39:4462-4467.