RESEARCH ARTICLE Re-examining the phylogeny of clinically relevant Candida species and allied genera based on multigene analyses Clement K.M. Tsui1, Heide-Marie Daniel2, Vincent Robert3 & Wieland Meyer1,4 1Molecular Mycology Research Laboratory, Center for Infectious Diseases and Microbiology, Westmead Hospital, Westmead Millennium Institute, Westmead, NSW, Australia; 2BCCM/MUCL culture collection, Universite´ Catholique de Louvain, Unite´ de Microbiologie, Louvain-la-Neuve, Belgium; 3Centraalbureau voor Schimmelcultures, Utrecht, The Netherlands; and 4Western Clinical School, Department of Medicine, The University of Sydney, Sydney, Australia Correspondence: Wieland Meyer, Molecular Abstract Mycology Research Laboratory, Center for Infectious Diseases and Microbiology, Yeasts of the artificial genus Candida include plant endophytes, insect symbionts, Westmead Hospital, Westmead Millennium and opportunistic human pathogens. Phylogenies based on rRNA gene and actin Institute, Westmead, NSW 2145, Australia. sequences confirmed that the genus is not monophyletic, and the relationships Tel.: 161 2 98456895; fax: 161 2 98915317; among Candida species and allied teleomorph genera are not clearly resolved. e-mail: [email protected] Protein-coding genes have been useful to resolve taxonomic positions among a broad range of fungi. Over 70 taxa of the genus Candida and its allied sexually Received 10 August 2007; revised 18 November reproducing genera were therefore selected, and their phylogenetic relationships 2007; accepted 20 November 2007. were investigated using nuclear sequences of the largest subunit and second largest First published online 31 January 2008. subunit of RNA polymerase II gene, actin, the second subunit of the mitochondrial DOI:10.1111/j.1567-1364.2007.00342.x cytochrome oxidase gene, and D1/D2 LSU rRNA gene. The DNA sequences were analysed by maximum parsimony and Bayesian inference, resulting in the Editor: Teun Boekhout recognition of six major phylogenetic groups (A–F). Group A contains six facultative pathogenic Candida species, which seem to have derived from Keywords nonpathogenic species, while Group B contains species of Clavispora, Metschniko- Ascomycetous yeasts; Candida ; wia, and Pichia guilliermondii. Species of Debaryomyces form an independent Debaryomyces ; Pichia ; Saccharomyces ; group C that is related to groups A and B. Pichia fermentans and other phylogeny. environmental species are concentrated in Group D. Group E, containing Pichia anomala, may be a sibling to group F, which is represented by the Saccharomyces species complex. place the clinically relevant species in their evolutionary Introduction context. Members of the polyphyletic genus Candida are ubiquitous The genus Candida comprises a wide range of phylogen- in numerous natural and artificial habitats (Kurtzman & etically unrelated anamorphic fungi. These species were Fell, 1998). These single-celled fungi have been the subjects classified into one artificial genus due to the lack of of clinical research because many Candida species are distinctive characters that would allow their assignment to a facultative pathogens in humans and have become pressing more natural genus (Meyer et al., 1998). Indeed, for many concerns in public health with the increase of immunosup- Candida species, connections with known teleomorphic pressed patients. Other members of the genus Candida or ascomycete genera have been established using a single-gene their teleomorph counterparts have an ancient history in phylogeny (Kurtzman & Robnett, 1998). However, taxono- food production and some are currently used in indus- mists have restrained from the erection of new genera for trial processes, being listed as ‘Generally Regarded As Safe’ ascomycetous yeasts in many cases because of their unre- (GRAS) organisms by the FDA of the USA (http:// solved phylogeny. Therefore, the current classification is in www.gpoaccess.gov/cfr/index.html). Because of these close large part artificial and unsatisfactory. Multiple gene phylo- links, a phylogenetic analysis needs to integrate species of genies have started to establish new genera in phylogeneti- clinical, applied and environmental relevance. Here, species cally well-circumscribed groups (Kurtzman & Robnett, 2003; of these different orientations were included with the aim to Suh et al., 2006). To revise additional well-circumscribed FEMS Yeast Res 8 (2008) 651–659 c 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved 652 C.K.M. Tsui et al. groups, a greater resolution at the backbone of the phyloge- analysed with the goal of dividing Candida species into netic tree is necessary. A more natural classification system well-supported clades and consolidating the phylogenetic should be able to accommodate novel species in their positions of the taxa with an unresolved status. appropriate genera and to predict significant characteristics within genera or larger groups. Materials and methods A better knowledge on the evolutionary relationships of Candida species is vital to understand the ecology, clinical Taxon sampling, DNA extraction, and PCR relevance, and diagnosis of these yeasts. The sequences of the amplification D1/D2 region of the large subunit (LSU) of the ribosomal Seventy isolates including numerous type strains were gene complex (rRNA gene) have been used widely in yeast studied (Table 1). The taxon sampling represented a large phylogeny. But D1/D2 LSU sequences alone were not able diversity of Saccharomycetales and included two potentially to resolve the inter-generic relationships (Kurtzman & basal taxa: Yarrowia lipolytica and Zygoascus hellenicus. Robnett, 1997, 1998). They were also unable to distinguish These taxa represented distinct evolutionary lineages by the a number of closely related species such as Pichia guillier- composition of their cell wall polysaccharides (Weijman & mondii and Pichia caribbica, Candida mucifera, and Stepha- Golubev, 1988; Phaff, 1998; Suzuki, 2002) and 18S rRNA noascus ciferrii, as well as Kluyveromyces marxianus and gene analysis (Prillinger et al., 2002). All cultures are Kluyveromyces lactis (Daniel & Meyer, 2003). Previous maintained in the Australian Medical Fungal Culture research showed that actin gene sequences offered a better Collection at the Westmead Hospital/University of Sydney. phylogenetic resolution among Candida species. Four major Cultures were grown and DNA was extracted as described well-supported groups were established (Daniel et al., 2001). previously (Meyer et al., 1997). The extracted DNA was However, in a follow-up study, most branches in the tree stored at À 20 1C until future use. Fragments of the RPB1 of an expanded taxon data set lacked significant statistical and RPB2 genes were amplified using the primer sequences support (Daniel, 2003). Because the results of single gene published in Diezmann et al. (2004). A fragment of the analyses do not provide sufficient information to resolve the COX2 gene was amplified with the primers published by phylogeny of the ascomycetous yeasts and because single Kurtzman & Robnett (2003) or primers CO2F1 (50-GGT gene trees could be misleading (Rokas et al., 2003), multi- ATW CAK GAA TTA TAY GA-30) and CO2R1 (50-CAS ATT gene analyses have been conducted to clarify the phyloge- CWA RTT TKG TWG GCA T-30) designed during the netic relationships of the genus Candida and its allied genera current study. The PCR was performed in volumes of 25 mL and to improve the branch resolutions (Kurtzman & Rob- containing 75 mM Tris-HCl, 1.5 mM MgCl , 7.5 pmol of nett, 2003; Diezmann et al., 2004). These investigations, 2 each primer, 1 U Taq (AmpliTaq, Applied Biosystems), and however, sampled only single families or were restricted to 20 ng of genomic DNA. The reactions were carried out in a the common clinically relevant species. Suh et al. (2006) PTC-100 Thermal Cycler (MJ Research Inc., Watertown, have defined the lack of protein-coding gene sequences for MA) using the following programme: 5-min denaturation at several major branches of yeasts. 95 1C, and 30–35 cycles of: 95 1C for 30 s, 54 1C for 30 s In the present study, additional sequence data were (or 55 1C for RPB2), 72 1C for 45 s, a final extension at 72 1C generated for the expanded taxon sampling set (Daniel, for 7 min, and a 4 1C soak. The PCR product was then 2003), and sequences for four of the nine clades that were purified using the GFX PCR DNA and Gel Band purification without alternative data to rRNA gene sequences were kit (Amersham Biosciences, Buckinghamshire, UK) and provided (Suh et al., 2006). In previous studies, the sequenced using the ABI PRISMs BigDyes Terminator occurrence of genome duplication in the evolution of the v3.1 Cycle Sequencing Kit (PE Applied Biosystems, Foster Saccharomyces species and Candida glabrata was revealed City, CA) with the amplification primers in both forward (e.g. Dujon, 2006). According to current knowledge, this and reverse directions. duplication concerned a subgroup of the well-circumscribed Saccharomyces clade, and potential gene paralogues would Phylogenetic analyses not influence the basal branching of the phylogenetic tree, which was the subject of the current study. With this fact in Sequences were edited in SEQUENCHER 4.1 (Gene codes mind we carefully selected genes that are commonly used to corporation), and their identity was verified by BLAST study fungal phylogeny, using primers that do not amplify searches (with significant scores with an e-value o 0.001)