Polyphasic taxonomy of the basidiomycetous yeast genus Rhodotorula: Rh. glutinis sensu stricto and Rh. dairenensis comb. nov.

Abstract

The phenotypic and genetic heterogeneity of the basidiomycetous yeast species Rhodotorula glutinis was investigated in a group of 109 isolates. A polyphasic taxonomic approach was followed which included PCR fingerprinting, determination of sexual compatibility, 26S and ITS rDNA sequence analysis, DNA–DNA reassociation experiments and reassessment of micromorphological and physiological attributes. The relationships with species of the teleomorphic genus Rhodosporidium were studied and isolates previously identified as Rh. glutinis were found to belong to Rhodosporidium babjevae, Rhodosporidium diobovatum and Rhodosporidium sphaerocarpum. Other isolates included in the study were found to belong to Rh. glutinis var. dairenensis, which is elevated to the species level, or to undescribed species. The concept of Rh. glutinis sensu stricto is proposed due to the close phenetic and phylogenetic proximity detected for Rh. glutinis, Rhodotorula graminis and R. babjevae.

Basidiomycetous yeast, Polyphasic taxonomy, Rhodotorula glutinis, Rhodotorula dairenensis

1 Introduction

Rhodotorula glutinis (Fresenius) Harrison is a pink-colored anamorphic urediniomycetous yeast and the type species of the genus Rhodotorula Harrison. According to Barnett et al. [1] and Fell and Statzell-Tallman [2] this species has a world-wide distribution and has been isolated from different environments and substrates. Several studies have indicated considerable phenotypic and genetic heterogeneity [3–8], which suggests that the present concept of Rh. glutinis might encompass more than one species. Moreover, the large number of variable responses to physiological tests (47 of the 80 tests listed by Barnett et al. [1] are recorded as variable) results in a vague nutritional delimitation. In a recent taxonomic revision, Fell and Statzell-Tallman [2] recognized two varieties, viz. Rh. glutinis var. glutinis and Rh. glutinis var. dairenensis Hasegawa and Banno, based on a few discrepant physiological tests and on relevant differences in the D1/D2 region of the 26S rDNA.

The relationship of Rh. glutinis with the teleomorphic genus Rhodosporidium Banno is not clear. Several species of Rhodosporidium like Rhodosporidium azoricum Sampaio and Gadanho, Rhodosporidium babjevae Golubev, Rhodosporidium diobovatum Newell and Hunter, Rhodosporidium kratochvilovae Hamamoto, Sugiyama and Komagata, Rhodosporidium paludigenum Fell and Statzell-Tallman, Rhodosporidium sphaerocarpum Newell and Fell and Rhodosporidium toruloides Banno have distinct morphological and molecular traits, but cannot be clearly distinguished from Rh. glutinis on the basis of physiological characteristics. In addition, no teleomorphic stage is known for the type strains of the two varieties of Rh. glutinis, since sexual compatibility could not be found with any of the mating strains of the species of Rhodosporidium mentioned above.

We have recently addressed the issue of the heterogeneity of Rh. glutinis and its relationship with the different Rhodosporidium species. Several isolates formerly identified as Rh. glutinis were found to represent mating strains of R. kratochvilovae[9], a species originally described as homothallic [10]. Molecular fingerprinting methods allowed us to investigate in detail isolates that fit the physiological concept of Rh. glutinis and to detect sexually compatible strains that were found to represent an undescribed species, R. azoricum[11]. In this report we present our results on the molecular typing of a large set of strains, previously identified as Rh. glutinis or as related species using conventional methods, viz. Rhodotorula araucariae Grinbergs and Yarrow, Rhodotorula graminis di Menna and Rhodotorula mucilaginosa (Jörgensen) Harrison. This group included strains isolated and characterized in our laboratory as well as reference strains from other culture collections. For comparison, all presently-known species of Rhodosporidium were included in our study. A polyphasic approach was followed and the results obtained with the technique of micro/minisatellite-primed polymerase chain reaction (MSP-PCR) were confirmed by means of sexual compatibility tests, rDNA sequence analyses and DNA–DNA reassociation experiments. The results indicated that a group of closely related taxa could be interpreted as the Rh. glutinis sensu stricto complex. In addition, we propose the elevation of the variety dairenensis to the rank of species.

2 Materials and methods

2.1 Cultures

A total of 109 strains of Rhodotorula and Rhodosporidium species were investigated. The list of cultures used in this study, and relevant information, is presented in Table 1.

1

List of strains used in this study and relevant information related to them

Original species designation	Species designation after this study	Strain no.a	Sexualityb	Origin – culture collectionc	Origin – substrate and locality
R. azoricum Sampaio and Gadanho		IGC 5062T	MT A1		soil, Azores, Portugal
R. azoricum		IGC 5084	MT A2		soil, Azores, Portugal
R. babjevae Golubev		IGC 5168T	MT A1	VKM Y-2275T	herbaceous plant, Moscow, Russia
R. babjevae		IGC 5169	MT A2	VKM Y-2276	herbaceous plant, Moscow, Russia
Rh. glutinis (Fresenius) Harrison var. glutinis	anamorph of R. babjevae	IGC 4258			polluted river water, Oeiras, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4348		FO 34575	unknown
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4651			leaf, Estoril, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4706			reed, Oeiras, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4731			grasshopper, Sintra, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4777			flower, Oeiras, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4780			blackberries, Sesimbra, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4781			dead leaves, Oeiras, Portugal
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4915			pond water, Silwood, UK
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 4957			grapes, Sintra, Portugal
Rh. glutinis var. glutinisd	anamorph of R. babjevae	IGC 5590		CBS 322	water supply of brewery, Germany
Rh. glutinis var. glutinise	anamorph of R. babjevae	IGC 5592		CBS 347	air, Japan
Rh. glutinis var. glutinisf	anamorph of R. babjevae	IGC 5593		CBS 2366	air, Japan
Rh. glutinis var. glutinis	anamorph of R. babjevae	IGC 5598		IAM 12961	core samples from stratigraphic drillings, Japan
Rh. glutinis var. glutinisg	anamorph of R. babjevae	IGC 5610		CBS 6020	soil from drill core, Japan
R. diobovatum Newell and Hunter		IGC 4364T	MT α	CBS 6085T	seawater, USA
R. diobovatum		IGC 4640	MT a	CBS 6084	seawater, USA
Rh. glutinis var. glutinis	R. diobovatum	IGC 2920	MT a		seawater, Scripps pier, USA
Rh. glutinis var. glutinis	R. diobovatum	IGC 2922	MT a		seawater, California, USA
Rh. glutinis var. glutinis	R. diobovatum	IGC 4914	MT a		rotten wood, Arrábida, Portugal
Rh. glutinis var. glutinish	R. diobovatum	IGC 5607	MT α	CBS 324	substrate unknown, Portugal
R. fluviale Fell, Kurtzman, Tallman and Buck		IGC 4701T	SF	CBS 6568T	brackish water, USA
R. kratochvilovae Hamamoto, Sugiyama and Komagata		IGC 4583T	SF	IAM 13072T	unknown
R. kratochvilovae		IGC 4776	MT A2		leaf, Oeiras, Portugal
R. kratochvilovae		IGC 4778	MT A2		pine cone, Oeiras, Portugal
R. kratochvilovae		IGC 4783	MT A2		polypore fungus, Arrábida, Portugal
R. kratochvilovae		IGC 4785	MT A2		plant litter, Arrábida, Portugal
R. kratochvilovae		IGC 4793	SF, MT A1		grasshopper, Sintra, Portugal
R. kratochvilovae		IGC 4818	MT A1		wood chips, Setúbal, Portugal
R. kratochvilovae		IGC 4819	MT A2		moss, Oeiras, Portugal
R. kratochvilovae		IGC 4883	MT A2	JCM 3772	unknown
R. kratochvilovae		IGC 5085	MT A2		soil, Azores, Portugal
R. kratochvilovae		IGC 5244	MT A1		culture contaminant, Oeiras, Portugal
R. kratochvilovae		IGC 5579	SF	JCM 8172	unknown
R. kratochvilovae		IGC 5068	MT A2		soil, Azores, Portugal
Anamorph of R. kratochvilovae		IGC 4787			seawater, Arrábida, Portugal
Anamorph of R. kratochvilovae		IGC 5580		JCM 8173	unknown
Anamorph of R. kratochvilovaei		IGC 5591		CBS 332	sputum
Anamorph of R. kratochvilovaej		IGC 5606		CBS 321	air, Japan
R. lusitaniae Fonseca and Sampaio		IGC 4641T	SF		woodland soil, Lisbon, Portugal
R. lusitaniae		IGC 4599	SF		dry leaf, Arrábida, Portugal
R. lusitaniae		IGC 4642	SF		plant litter, Arrábida, Portugal
R. paludigenum Fell and Statzell-Tallman		IGC 4495T	MT A2	CBS 6566T	Juncus roemerianus marsh, USA
R. paludigenum		IGC 4496	MT A1	CBS 6567	Rhizophora mangle marsh, USA
R. paludigenum		IGC 5613	MT A2	CBS 4477	water, USA
R. paludigenum		IGC 5614	MT A1	CBS 6565	R. mangle swamp, USA
Rh. graminis di Menna	anamorph of R. paludigenum	IGC 5611		CBS 5016	seawater, Atlantic Ocean
R. sphaerocarpum Newell and Fell		IGC 4104T	MT a	CBS 5939T	water, Antarctica
R. sphaerocarpum		IGC 4105	MT α	CBS 5940	seawater, Antarctica
R. sphaerocarpum		IGC 4106	SF	CBS 5941	seawater, Antarctica
R. sphaerocarpum		IGC 5414	SF		salt farm, Slovenia
R. sphaerocarpumk		IGC 5595	MT a	CBS 6985	salt farm, Japan
Rh. graminis	R. sphaerocarpum	IGC 2988	MT a		unknown
Rh. glutinis var. glutinis	R. sphaerocarpum	IGC 5074	SF		beach sand, Azores, Portugal
Rh. glutinis var. glutinis	R. sphaerocarpum	IGC 5105	MT a		salt farm, Alcochete, Portugal
Rh. glutinis var. glutinis	R. sphaerocarpum	IGC 5106	MT a		salt farm, Alcochete, Portugal
Rh. glutinis var. glutinis	R. sphaerocarpum	IGC 5107	MT α		salt farm, Alcochete, Portugal
Rh. glutinis var. glutinis	R. sphaerocarpum	IGC 5108	MT α		salt farm, Alcochete, Portugal
R. toruloides Banno		IGC 5615T	SFl	CBS 6016T
R. toruloides		IGC 4350	SF	RB 659	unknown
R. toruloidesm		IGC 4416	MT α	CBS 14	wood pulp, Sweden
R. toruloidesn		IGC 4417	MT a	CBS 5745	soil, Japan
R. toruloides		IGC 4661	MT a		woodland soil, Alentejo, Portugal
R. toruloides		IGC 4786	MT a		dry leaf, Rio de Janeiro, Brazil
R. toruloides		IGC 4943	MT a		salt farm, Alcochete, Portugal
R. toruloides		IGC 5081	MT a		soil, Azores, Portugal
R. toruloides		IGC 5082	MT α		soil, Azores, Portugal
R. toruloides		IGC 5109	MT a		salt farm, Alcochete, Portugal
Rh. araucariae Grinbergs and Yarrow		IGC 4625T		CBS 6031T	rotting bark of Araucaria araucana, Chile
Rh. araucariae	Rhodotorula sp.	IGC 4691			wood, Sesimbra, Portugal
Rh. araucariae	Rhodotorula sp.	IGC 4724			soil, Alentejo, Portugal
Rh. glutinis var. glutinis	Rh. glutinis	IGC 4177T		CBS 20T	air
Rh. glutinis var. glutinis	Rh. glutinis	IGC 4107		ML 2248	seawater, Florida, USA
Rh. glutinis var. glutinis	Rh. glutinis	IGC 5599		IAM 12962	unknown
Rh. glutinis var. glutiniso	Rh. glutinis	IGC 5609		CBS 2367	fruit of Phylodendron sp., Italy
Rh. glutinis var. glutinis	Rhodotorula sp.	IGC 4782			wood from Quercus suber, Portugal
Rh. glutinis var. glutinis	Rhodotorula sp.	IGC 4826			bark from Pinus pinaster, Sesimbra, Portugal
Rh. glutinis var. glutinis	Rhodotorula sp.	IGC 4884		JCM 3773	unknown
Rh. glutinis var. glutinis	Rhodotorula sp.	IGC 5380			soil, Azores, Portugal
Rh. glutinis var. glutinis	Rhodotorula sp.	IGC 5597		JCM 8164	unknown
Rh. glutinis var. glutinis	Rhodotorula sp.	IGC 5600		IAM 12263	unknown
Rh. glutinis var. dairenensis Hasegawa and Banno	Rh. dairenensis (Hasegawa and Banno) Fell, Sampaio and Gadanho comb. nov.	IGC 4885LT		JCM 3774LT	air, locality unknown
Rh. glutinis var. dairenensis	Rh. dairenensis	IGC 4784			plant litter, Arrábida, Portugal
Rh. glutinis var. dairenensis	Rh. dairenensis	IGC 4897			flower of Canna indica, Azores, Portugal
Rh. glutinis var. glutinis	Rh. dairenensis	IGC 4615		VKM Y-1133	water, river Ili, Kazakhstan
Rh. glutinis var. glutinis	Rh. dairenensis	IGC 4944			salt farm, Alcochete, Portugal
Rh. glutinis var. glutinis	Rh. dairenensis	IGC 5601		JCM 8168	core samples from stratigraphic drillings, Japan
Rh. mucilaginosa (Jörgensen) Harrison	Rh. dairenensis	IGC 5602		CBS 7294	core samples from stratigraphic drillings, Japan
Rh. glutinis var. glutinisp	Rh. dairenensis	IGC 5608		CBS 2202	soil, USA
Rh. graminis		IGC 4842T		CBS 2826T	grass, New Zealand
Rh. graminis		IGC 5594		CBS 3043	leaf of Citrus sp., Indonesia
Rh. graminisq	anamorph of S. ruineniae Holzschu, Tredick and Phaff	IGC 5596		CBS 5624	air, Japan
Rh. mucilaginosa		IGC 5166T		CBS 316T	unknown
Rh. mucilaginosar		IGC 4616		VKM Y-253	unknowns
Rh. mucilaginosa		IGC 4707			Chrysalis, Oeiras, Portugal
Rh. mucilaginosa		IGC 4758			marsh water, Setúbal, Portugal
Rh. mucilaginosa		IGC 4759			marsh water, Setúbal, Portugal
Rh. mucilaginosa		IGC 4760			paper mill effluent, Setúbal, Portugal
Rh. mucilaginosa		IGC 4772			polluted river water, Oeiras, Portugal
Rh. mucilaginosa		IGC 4791		CBS 17	unknown
Rh. mucilaginosa		IGC 5180			culture contaminant, Oeiras, Portugal
Rhodotorula sp.		IGC 4820			plant litter, Oeiras, Portugal
Rhodotorula sp.		IGC 4821			humus, Arrábida, Portugal
Rhodotorula sp.		IGC 4824			rotten wood, Gerês, Portugal
Rhodotorula sp.		IGC 5612		CBS 5143	bronchial secretion, Netherlands

^aIGC, Portuguese Yeast Culture Collection, FCT-UNL, Portugal.

^bMT, mating type; SF, self-fertile.

^cCBS, Centraalbureau voor Schimmelcultures, Yeast Division, Utrecht, The Netherlands; IAM, Institute of Molecular and Cellular Biosciences, University of Tokyo, Japan; JCM, Japan Collection of Microorganisms, RIKEN, Japan; ML, Marine Laboratory, Rosenstiel School of Marine and Atmospheric Science, University of Miami, USA; VKM, All-Russian Collection of Microorganisms, Moscow, Russia; FO, RB, Personal collections of Drs. F. Oberwinkler and R.J. Bandoni, respectively.

^dReceived at CBS as T. rubra Schimon.

^eOriginally identified as Rhodotorula suganii (Okonuki) Lodder.

^fT of Torulopsis saitoi Ciferri and Redaelli; T of T. rubra Saito; T of Rh. rubra (Saito) Harrison.

^gT of Rhodotorula terrea Sugiyama and Goto.

^hT of Rh. glutinis (Fresenius) Harrison var. lusitanica Marcilla, Feduchy and Gomes.

ⁱT of Torulopsis bronchialis Ciferri and Redaelli.

^jT of Torula rufula Saito.

^kT of Rh. glutinis (Fresenius) Harrison var. salinaria Hirosawa and Takada.

^lMycelial conjugate of IFO 0559 and IFO 0880 (IFO, Institute for Fermentation, Osaka, Japan).

^mT of Rhodotorula gracilis Rennerfelt.

ⁿT of Rh. glutinis (Fresenius) Harrison var. rufusa Iizuka and Goto.

^oT of Mycotorula rosea-corallina Scaramella.

^pT of Torulopsis minuta (Saito) Ciferri and Redaelli var. americana Ciferri.

^qT of P. rosa Nishiwaki.

^rAuthentic strain of Mycotorula cisnerosi nom. nud.

^sDerived from CBS 2387, isolated by Radaelli (Italy).

2.2 Morphological characterization and studies of sexual compatibility

For microscopy, cultures were grown on MYP agar (malt extract 0.7% w/v, yeast extract 0.05% w/v, soytone 0.25% w/v and agar 1.5% w/v) at room temperature (20–23°C) and studied with an Olympus BX50 microscope, using phase contrast optics. For determination of sexual compatibility, pairs of 2–4 days old cultures were crossed on corn meal agar (CMA), incubated at room temperature and examined for production of mycelium and teliospores after 1 week. Teliospore germination was obtained by transferring to water agar (1.5% w/v), 4–6 weeks old teliospores soaked in demineralized water for 2–3 weeks. Observations were made 2–3 days after transfer.

2.3 Physiological characterization

Physiological and biochemical characterization was performed according to the techniques described by Yarrow [12]. Additional assimilation tests were performed using aldaric acids and aromatic compounds, as described by Fonseca [13] and Sampaio [14], respectively. The complete data matrix is available upon request.

2.4 MSP-PCR, rDNA sequence analysis and DNA–DNA reassociation experiments

For PCR fingerprinting, the DNA extraction protocol, PCR and electrophoresis conditions and gel image analysis procedures were those reported by Sampaio et al. [9]. The primers used were the core sequence of the phage M13 and two synthetic oligonucleotides, (GTG)₅ and (GAG)₅. DNA banding patterns were analyzed using the GelCompar software package, version 4.1 [15]. Similarities among isolates were estimated using the Dice coefficient and clustering was based on the UPGMA method.

For rDNA sequence analysis, total DNA was extracted using the protocol of Sampaio et al. [9] and amplified using primers ITS5 and LR6. Cycle sequencing of the 600–650 bp region at the 5′-end of the 26S rDNA D1/D2 domain employed forward primer NL1 (5′-GCA TAT CAA TAA GCG GAG GAA AAG) and reverse primer NL4 (5′-GGT CCG TGT TTC AAG ACG G). The ITS region was sequenced using the forward primer ITS1 (5′-TCC GTA GGT GAA CCT GCG G) and the reverse primer ITS4 (5′-TCC TCC GCT TAT TGA TAT GC). Sequences were obtained with an Amersham Pharmacia ALF express II automated sequencer using standard protocols. Alignments were made with MegAlign (DNAStar) and visually corrected. Phylogenetic trees were computed with PAUP version 4.0 using the neighbor-joining method [16]. Distances between sequences were calculated using Kimura's two-parameter model [17]. Bootstrap analysis [18] was based on 1000 random resamplings.

For DNA–DNA reassociation experiments, a Gilford Response UV-VIS spectrophotometer was used and the procedures of Sampaio et al. [9] were employed.

3 Results and discussion

3.1 PCR fingerprinting of Rhodotorula and Rhodosporidium

The 109 strains investigated were initially characterized by MSP-PCR fingerprinting using primers (GTG)₅ and M13. The dendrogram obtained from the combined numerical analysis of the fingerprints is shown in Fig. 1. In general, a good interspecific discrimination was obtained since each type strain gave a different banding pattern. Several strains originally identified as Rh. glutinis var. glutinis produced PCR fingerprints more similar to those obtained for the reference strains of other species, thus confirming the already reported heterogeneity of Rh. glutinis. We defined 11 clusters, which grouped the vast majority of the investigated strains (Fig. 1). Instead of setting a threshold similarity value for cluster definition, our clusters were established by combining the numerical analysis shown in the dendrogram with a visual comparison of the fingerprints. Therefore, the numbered clusters correspond to groups of isolates showing visually indistinguishable or highly similar profiles. Fifteen strains were not assigned to any of the 11 clusters.

1

MSP-PCR dendrogram and fingerprints of Rhodotorula and Rhodosporidium strains generated with primers (GTG)₅ and M13 (Dice coefficient, UPGMA clustering method, r=co-phenetic correlation coefficient).

The nine species of the genus Rhodosporidium were clearly discriminated by PCR fingerprinting. Clusters were formed for all the Rhodosporidium species represented by more than one strain, the exception being Rhodosporidium fluviale of which a single strain is known. Some strains originally identified as Rh. glutinis showed DNA banding patterns similar to those of certain Rhodosporidium species. This occurred in cluster 1 (R. diobovatum), cluster 2 (R. babjevae), cluster 3 (R. sphaerocarpum) and cluster 4 (R. kratochvilovae, see Sampaio et al. [9]). In the case of clusters 1, 3 and 4, the correct identity of the ‘Rh. glutinis’ strains was determined in mating experiments involving strains representing the different mating types of the presumed Rhodosporidium species. The isolates originally identified as Rh. glutinis, but falling in a Rhodosporidium cluster, proved to be mating strains of Rhodosporidium spp. (Table 1 and Sampaio et al. [9]). The exception was IGC 4915 (cluster 1) that did not behave as a mating strain of R. diobovatum.

A considerable number of strains that displayed the physiological attributes of Rh. glutinis had patterns showing similarity with R. babjevae and were included in cluster 2. The type strains of Rh. glutinis var. glutinis and of Rh. graminis also exhibited the same type of profiles. The closeness between these three species was also observed in the analysis of the D1/D2 region of the 26S rDNA [9,20]. The PCR fingerprints obtained with the (GTG)₅ primer for the strains of clusters 1 and 2 presented a low number of clear amplicons (Fig. 1). However, the absence of clear bands is also useful for characterization since, as it can be observed in Fig. 1, the fingerprints of strains in clusters 1 and 2 are easily discriminated from those of the remaining isolates. Due to the difficulties in resolving the relationships of strains belonging to clusters 1 and 2, we decided to perform additional MSP-PCR studies on all members of those two clusters. These results will be presented in the next section.

The strains of R. toruloides showed two main types of DNA banding profiles and were grouped in two clusters (clusters 5 and 6). Cluster 5 included IGC 5615^T which is a combination of two sexually compatible strains, IGC 4350 which is self-fertile, two strains of mating type α (IGC 4416 and IGC 5082) and one strain of mating type a (IGC 5081). Cluster 6 included only strains of mating type a. These results are in accordance with the analyses of Hamamoto et al. [4] who have reported intraspecific heterogeneity in R. toruloides. In an investigation of the relative mobility of selected enzymes those authors grouped 23 strains of R. toruloides in two clusters. As in our study, one cluster contained only strains belonging to mating type a, whereas the other cluster included strains from both mating types, as well as self-fertile strains.

Four strains were placed between clusters 5 and 6. They do not belong to R. toruloides and will be discussed later. The remaining Rhodosporidium species, R. paludigenum (cluster 7), R. azoricum (cluster 8) and Rhodosporidium lusitaniae (cluster 9) did not include strains previously identified as Rh. glutinis.

Among the anamorphic species, Rh. araucariae and Rh. graminis presented also interesting taxonomic problems. The identification of some isolates as members of those two species based on physiological attributes was not supported by MSP-PCR. With the latter approach we observed that the three Rh. araucariae strains exhibited distinct DNA banding patterns. Among the five Rh. graminis isolates, IGC 5594 seems to be closely related to the type strain of the species (cluster 2), IGC 2988 belongs to the R. sphaerocarpum cluster (cluster 3) and is in fact a mating strain of this species (Table 1), and IGC 5596 and IGC 5611 were placed in the vicinity of cluster 8. All but one of the 10 Rh. mucilaginosa strains showed similar profiles and were grouped in a single cluster (cluster 11). The exception (IGC 5602) exhibited the typical DNA banding pattern of Rh. glutinis var. dairenensis (cluster 10). The dairenensis cluster included also four strains previously identified as Rh. glutinis var. glutinis (IGC 4615, IGC 4944, IGC 5601 and IGC 5608). The two varieties of Rh. glutinis presented completely distinct profiles. As was also observed by several authors [6,20], the variety dairenensis (cluster 10) seems to be more closely related to Rh. mucilaginosa (cluster 11) than to the variety glutinis (cluster 2).

3.2 PCR fingerprinting of the Rh. glutinis species complex

In the MSP-PCR studies with primers M13 and (GTG)₅, the type strains of Rh. glutinis var. glutinis, Rh. graminis and R. babjevae gave similar fingerprints which were included in a single cluster. In order to disclose the relationships between these species and also to improve the characterization of the remaining strains in cluster 2, we performed an additional MSP-PCR analysis using the primer (GAC)₅ as shown in Fig. 2. The members of cluster 1 (R. diobovatum) were also included since their profiles did not differ markedly from those of the species mentioned above. R. diobovatum was clearly differentiated from the remaining species. IGC 4915, the only strain of cluster 1 (Fig. 1) that did not display sexual compatibility with strains of this species, presented a (GAC)₅-primed profile similar to those recorded for R. babjevae (Fig. 2). In spite of the close resemblance of the remaining profiles, four fingerprints (IGC 4107, IGC 5599, IGC 5609, and IGC 4826) appeared to be more related to the one obtained for the type strain of Rh. glutinis var. glutinis. The other profiles seemed to be more similar to those of the type strains of R. babjevae and Rh. graminis.

2

MSP-PCR fingerprints of the taxa belonging to the Rh. glutinis species complex generated with the primer (GAG)₅. M, molecular size marker (λ DNA cleaved with HindIII and ΦX174 DNA cleaved with HaeIII).

3.3 Sequence analyses

The results obtained with the MSP-PCR approach were evaluated by means of sequence analysis of the D1/D2 region of 26S rDNA from selected isolates (Fig. 3). The delimitation of Rh. glutinis var. glutinis obtained with primer (GAC)₅ was confirmed for three strains (IGC 4107, IGC 5599 and IGC 5609). The fourth strain, IGC 4826, seems to represent an undescribed species due to relevant nucleotide differences (12 discrepant bases towards Rh. glutinis var. glutinis).

3

Phylogenetic tree of Rh. glutinis and related species of the genera Rhodosporidium and Rhodotorula, obtained by neighbor-joining analysis of the D1/D2 region of the 26S rDNA. S. dacryoides was included to root the tree. Percentage bootstrap values of 1000 replicates are given at each node. GenBank accession numbers are indicated after strain designation (names in boldface correspond to sequences determined in this study; T, type strain).

Among the strains originally identified as Rh. graminis and included in our study, only IGC 5594 seems to belong to Rh. graminis (Figs. 1, 2 and 3). As already mentioned, IGC 2988 belongs to R. sphaerocarpum (Fig. 1). IGC 5596 has a D1/D2 sequence identical to the sequence of Sporidiobolus ruineniae Holzschu, Tredick and Phaff, although ballistoconidia have not been reported for this isolate, which was originally described as the type strain of Pichia rosa Nishiwaki. Finally, IGC 5611 seems to belong to R. paludigenum, although confirmation through mating was not possible because even the reported mating strains of R. paludigenum failed to react in several essays performed during this study.

The two R. babjevae mating strains showed distinct D1/D2 sequences differing in one nucleotide. The D1/D2 sequence of R. babjevae IGC 5169 is identical to the sequences of Rh. graminis (type strain group) and to the sequences of several anamorphic strains belonging to the babjevae group. Strain IGC 4258 exhibited an additional nucleotide difference towards the other strains of this group. R. diobovatum presents a more divergent D1/D2 sequence with three nucleotide differences for the remaining taxa in the Rh. glutinis species complex (Fig. 3).

The sequence of IGC 4724 was found to be identical with the sequence of IGC 4824 (Fig. 3). The possible conspecificity of these two isolates and their relationship with Rh. araucariae will be addressed in another publication dealing with the Rh. glutinis sensu lato group. A possible third member of the Rh. araucariae species complex is represented by strains IGC 4820 and IGC 4821. Another undescribed Rhodotorula species seems to be represented by strain IGC 4691, phenotypically similar to Rh. araucariae.

A clade composed of possibly three undescribed Rhodotorula species contained the closest relatives of R. sphaerocarpum (Fig. 3). It includes IGC 4782 and IGC 5597, with identical sequences and already found similar in Fig. 1; IGC 4884 and IGC 5600, also with identical sequences and also closely related in Fig. 1; and finally IGC 5380, which was placed close to cluster 8 in Fig. 1.

The strain IGC 5612 was found to be unrelated to Rh. glutinis and to Rhodosporidium and its closest relative is Sakaguchia dacryoides (Fell, Hunter and Tallman) Yamada, Maeda and Mikata.

For the Rh. glutinis species complex we performed also sequence analysis of the total ITS region of the rDNA (Fig. 4) due to its adequacy to resolve close relationships between taxa [20]. The results derived from the analysis of the D1/D2 region were supported by ITS sequence data except for IGC 5609. This strain was found to have one base difference towards the remaining strains of Rh. glutinis and no base differences towards the two Rh. graminis strains.

4

Phylogenetic tree of Rh. glutinis and related species of the genera Rhodosporidium and Rhodotorula, obtained by neighbor-joining analysis of the ITS region of the rDNA. Sporidiobolus salmonicolor was included to root the tree. Percentage bootstrap values of 1000 replicates are given at each node. GenBank accession numbers are indicated after strain designation (all sequences were determined in the present study; T, type strain).

The two mating strains of R. babjevae had two base differences in the ITS region and the discrepancies corresponded to those reported for the D1/D2 domain, i.e. the sequence of IGC 5168^T was unique and the sequence of IGC 5169 matched those of several anamorphic strains, including IGC 4915 and IGC 4258.

3.4 DNA–DNA reassociation experiments

In order to access the degree of relatedness between Rh. glutinis, Rh. graminis, the mating strains of R. babjevae and the remaining anamorphic strains of the babjevae group, we performed a series of DNA–DNA reassociation experiments (Fig. 5). High reassociation values were observed between strain IGC 5169 of R. babjevae (mating type A2) and several anamorphic strains of this species (Fig. 5), whereas intermediate reassociation values were obtained between IGC 5168 and IGC 5169, the two mating strains of R. babjevae. In spite of these intermediate values, we were able to observe a mating reaction between the two strains, leading to the production of teliospores and their germination by basidia and basidiospores, as already reported in the original description of this species [21]. Intermediate reassociation values were also recorded in experiments involving strains from the two groups of R. babjevae and the other two groups, viz. Rh. glutinis and Rh. graminis, whereas high homology values were obtained in all experiments involving strains of the same group (Fig. 5). It should be mentioned that intermediate reassociation values had been already observed in experiments involving the type strains of Rh. glutinis var. glutinis and Rh. graminis[22]. DNA–DNA reassociation results confirmed the close relationship between R. babjevae, Rh. glutinis and Rh. graminis and therefore we propose the designation Rh. glutinis sensu stricto for this group. The closest relative of Rh. glutinis sensu stricto is R. diobovatum. DNA–DNA reassociation values between R. diobovatum and the taxa of the sensu stricto group ranged between 5 and 15%.

5

Results of DNA–DNA reassociation experiments involving strains of the Rh. glutinis sensu stricto group (white boxes). Values in the shadowed boxes correspond to the number of different bases in the D1/D2 region of the 26S rDNA and those in the black boxes correspond to the number of different bases in the ITS region.

3.5 Species re-delimitation and taxonomic changes

Based on the results available so far, the circumscription of the species of Rh. glutinis sensu stricto is updated. As depicted in Fig. 1 we have identified additional strains of Rh. glutinis var. dairenensis and it seems justified, based on the MSP-PCR fingerprints, sequence data and nutritional information, to propose here its elevation to the rank of species.

3.5.1 R. babjevae Golubev

Yeast cells are normally ovoidal, single or in pairs (Fig. 6). In some cases (e.g. IGC 5169) short chains are formed. Cells of IGC 5590 are elongated and deviate from the typical cell shape or R. babjevae. This strain presents an additional unusual characteristic since it assimilates d-glucuronate, a peculiarity not found in any other member of the genus or in Rh. glutinis sensu stricto. Teliospore production and germination were also investigated (Fig. 7). In accordance with the original description [21] we observed the production of very short basidia, having two basidial compartments and producing oval basidiospores (Fig. 7). Mating experiments were performed between the mating strains of R. babjevae and the 15 remaining strains of the babjevae group depicted in Table 1, but sexual compatibility was not observed. However, it should be noted that even between the two R. babjevae mating types it is difficult to obtain the sexual state since conjugation is sporadic. Because IGC 5168 and IGC 5169, the two mating strains of R. babjevae investigated, are sexually compatible, we treat them as members of a single species, giving therefore more weight to this trait than to the intermediate DNA–DNA reassociation values detected in the present study. The asexual strains that exhibited high molecular similarity with IGC 5169 are considered to belong to the same taxon but are designated as anamorphic strains of R. babjevae. Relevant physiological traits of R. babjevae are presented in Table 2. The geographic distribution of R. babjevae seems to be vast since the studied isolates were from Russia, Portugal, UK, Germany and Japan. This species seems to be abundant in nature and it is possible that the majority of strains identified in the past as Rh. glutinis var. glutinis belonged in fact to R. babjevae. Unpublished data from our laboratory indicate that this species is the prevalent taxon of Rh. glutinis sensu stricto in Portugal. Most of the isolates are from terrestrial environments, but recently this species was also found in marine samples (Gadanho, unpublished).

6

Line drawings of cells of selected strains of R. babjevae (A), Rh. glutinis (B), Rh. graminis (C) and Rh. dairenensis (D). Cultures were grown on MYP agar and observed after 2–4 days. Bar=10 μm (same for all illustrations).

7

Line drawings of teliospores, basidia and basidiospores of R. babjevae IGC 5168^T×IGC 5169. The teliospore on the left side is not germinated. Note that the basidia of the other three teliospores have two compartments. Bar=10 μm.

2

Physiological discrimination between selected species of Rhodotorula and Rhodosporidium

L-Rha.

L-Ara.

Malt.

Melz.

Gall.

35°C

DL-Lact.

L-Tar.

Sacc.

Vit.

Nitra.

Nitri.

Rh. glutinis 4

−

+

+

+

−

−

−

+

−

+

+

+

Rh. graminis 2

+

+

−

−

−

−

−

−

+

+

+

+

R. babjevae 17

+

−

+

+

−

−

+ (11)

+ (14)

+ (15)

+

+

+

R. diobovatum 5

−

+

+

+

+

+

−

− (4)

+ (4)

+

+

+

Rh. mucilaginosa 9

− (8)

+

+

+

−

+ (6)

+ (8)

−

−

−

−

−

Rh. dairenensis 8

−

+

+

+

−

+

+

−

−

−

+

+

Rh. araucariae 1

−

−

−

−

+

−

+

−

−

+

+

+

+, positive result (including delayed and weak results); −, negative result. The number of strains investigated is indicated after the species name and numbers between brackets indicate the number of strains that gave the specified result. Abbreviations: L-Rha., l-rhamnose; L-Ara., l-arabinose; Malt., maltose; Melz., melezitose; Gall., gallic acid; DL-Lact., dl-lactate; L-Tar., l-tartaric acid; Sacc., saccharic acid; Vit., growth without vitamins; Nitra., nitrate; Nitri., nitrite.

3.5.2 Rh. glutinis (Fresenius) Harrison

We have detected and characterized four strains belonging to this species. Their micromorphology is depicted in Fig. 6 and important nutritional characteristics are shown in Table 2. Whilst the cells of IGC 4177^T and IGC 4107 are ovoidal, fusiform or cylindrical, IGC 5599 and IGC 5609 produce only subglobose to ovoidal cells. The combined results of four physiological tests (l-rhamnose, l-arabinose, maltose and melezitose) allow the discrimination of this species from the other two taxa of the Rh. glutinis sensu stricto group (Table 2). Only the origin of three isolates is known (air, fruit in Italy, and seawater off Florida, USA) and therefore presently the ecology of this species cannot be understood. Mating experiments involving the four strains gave negative results.

3.5.3 Rh. graminis di Menna

Two strains of this species were characterized in the present study. Their cells are ovoidal or elongate (Fig. 6). Relevant physiological tests allowing discrimination from related species are presented in Table 2. Both strains were isolated from plant material, one in New Zealand and the other in Indonesia. Mating experiments were negative.

3.5.4 Rhodotorula dairenensis comb. nov. (Hasegawa and Banno) Fell, Sampaio and Gadanho

Basionym: Rh. glutinis (Fresenius) Harrison var. dairenensis (Hasegawa and Banno [24]).

Eight strains of this species were investigated in the present study (see cluster 10 of Fig. 1 and Table 1). The relevant morphological and physiological characteristics of Rh. dairenensis are depicted in Fig. 6 and Table 2, respectively. The cells of the various strains investigated were similar, globose to slightly subglobose, and measured (2) 3–4×3–5 (6) μm. Saito [23] isolated for the first time a strain of this species and designated it Torula rubra var. α. Hasegawa and Banno [24] renamed the type culture as Rh. glutinis var. dairenensis. This variety differed from the variety glutinis because nitrate was assimilated weakly. Moreover, growth without vitamins was absent. The two varieties of Rh. glutinis were recognized by Phaff and Ahearn in the second edition of the monograph ‘The Yeasts, a Taxonomic Study’, published in 1970. In the following edition, published in 1984, the variety dairenensis was regarded as a synonym of Rh. glutinis[25], probably because the physiological differences were not considered to be relevant. However, the variety dairenensis was reinstated in the fourth edition of this treatise [2]. Based on rDNA sequence analysis, the closest relative of Rh. dairenensis is Rh. mucilaginosa (Jörgensen) Harrison. We performed DNA–DNA reassociation experiments between the type strains of the two species, which gave low values (10–20% homology). These values are in accordance with the four base differences observed in the D1/D2 region of the rDNA between the two species, but contradict previously published data of Hamamoto et al. [6] who recorded 66% DNA homology in experiments involving the strain YK 116 (=IGC 4885, type strain of Rh. dairenensis) and YK 1036 (=ATCC 9449=CBS 17), the neotype of Rhodotorula rubra (Demme) Lodder and a synonym of Rh. mucilaginosa. Nutritionally, Rh. dairenensis is distinguished from Rh. mucilaginosa based on nitrate and nitrite utilization, and from Rh. glutinis on the basis of the ability to grow without vitamins and at 35°C. The ecology of Rh. dairenensis is presently not clear. Two strains were isolated from aquatic environments, one strain was isolated from air, and five isolates originated from plant material and soil. This species was found in Portugal, Kazakhstan, USA and Japan. Mating experiments involving the eight strains gave negative results.

3.6 Final remarks

In yeast systematics the implementation of DNA characterization methods is challenging the previously existing classification, since unsuspected relationships among taxa have been detected. In the present study we addressed the issue of the delimitation of Rh. glutinis and its relationships with related taxa. In the past, several studies had already revealed the heterogeneity of this species but no taxonomic revision was carried out. Most molecular studies on yeast systematics deal with sequence analyses performed on type strains. It was our purpose to further investigate the reported heterogeneity of Rh. glutinis by studying a large set of strains. To achieve this, sequence analysis is not recommendable since it is time-consuming and has an elevated cost. PCR fingerprinting with micro/minisatellite primers proved to be a suitable method for a quick and reliable typing of large numbers of strains. This method allowed us to obtain, in most cases, species-specific profiles and to allocate several strains, physiologically identified as Rh. glutinis var. glutinis, to different Rhodosporidium species. MSP-PCR is especially relevant as a first approach for the preliminary analysis of large groups of strains, to be then further characterized by sequence analysis performed on selected isolates. Inconclusive results can then be clarified by DNA–DNA reassociation studies and mating experiments. With respect to traditional taxonomic methods, and in spite of current criticisms, we consider that physiological markers, when employed after adequate species delimitation, still provide useful taxonomic information.

Acknowledgements

M. Gadanho was supported by a grant SFRH/BD/1170/2000. The authors are indebted to Prof. Isabel Spencer-Martins for critical reading of the manuscript and to Drs. David Yarrow and Vincent Robert from CBS for sending valuable cultures.