On the Evolutionary History of Uleiella chilensis, a Smut Fungus Parasite of Araucaria araucana in South America: Uleiellales ord. nov. in Ustilaginomycetes

The evolutionary history, divergence times and phylogenetic relationships of Uleiella chilensis (Ustilaginomycotina, smut fungi) associated with Araucaria araucana were analysed. DNA sequences from multiple gene regions and morphology were analysed and compared to other members of the Basidiomycota to determine the phylogenetic placement of smut fungi on gymnosperms. Divergence time estimates indicate that the majority of smut fungal orders diversified during the Triassic–Jurassic period. However, the origin and relationships of several orders remain uncertain. The most recent common ancestor between Uleiella chilensis and Violaceomyces palustris has been dated to the Lower Cretaceous. Comparisons of divergence time estimates between smut fungi and host plants lead to the hypothesis that the early Ustilaginomycotina had a saprobic lifestyle. As there are only two extant species of Araucaria in South America, each hosting a unique Uleiella species, we suggest that either coevolution or a host shift followed by allopatric speciation are the most likely explanations for the current geographic restriction of Uleiella and its low diversity. Phylogenetic and age estimation analyses, ecology, the unusual life-cycle and the peculiar combination of septal and haustorial characteristics support Uleiella chilensis as a distinct lineage among the Ustilaginomycotina. Here, we describe a new ustilaginomycetous order, the Uleiellales to accommodate Uleiella. Within the Ustilaginomycetes, Uleiellales are sister taxon to the Violaceomycetales.


Introduction
With more than 1500 known species, smut fungi (Ustilaginomycotina) represent a highly diverse group of plant parasites [1]. Teliospore-forming species predominantly parasitize non-woody herbs (typically grasses, Poaceae), whereas those without teliospores prefer trees or shrubs [1,2]. A few species parasitize ferns [3] or conifers [1]. Some species with yeast or yeast-like growth, or with dimorphic life cycles are saprobic [4] or parasitic on animals [5,6]. Hypotheses on the evolution of smuts have focused either on their origin as parasites of the ancestors of monocot families or an earlier origin, followed by diversification on grass-like monocots [4]. The geographic distribution of plant-parasitizing Ustilaginomycotina has either been interpreted as the result of (i) habitat specializations rather than host preferences [1,7]; (ii) host jumps to closely or distantly related plant species [2,4]; or (iii) cospeciation events [8][9][10]. For instance, Uleiella is a unique ustilaginomycotinous genus occurring on gymnosperms restricted to the genus Araucaria in South America. There are two species, Uleiella chilensis on female cones of Araucaria araucana (Fig 1) in Chile and Argentina and U. paradoxa on male cones of A. angustifolia in Brazil [11]. The relationship between Uleiella and Araucaria provides a model to explore the origin and evolution of parasitism. Currently, species of Araucaria have disjunct distributions in the Southern Hemisphere, although this genus was widely distributed in both hemispheres during the Mesozoic around 251 to 65 mya [12][13][14][15]. Divergence time estimates from the study by Kranitz et al. [16] indicated that the stem origin of Araucaria was in the Early Cretaceous to Paleocene (~138-60 mya) and that of the Araucariaceae in the Permian-Triassic (~284-202 mya). However, fossils of Araucaria were dated as far back as the Lower Jurassic (~200 to 176 mya) [17]. It is unclear (i) whether the parasitic association between Uleiella and Araucaria is the result of co-evolution or a host jump; (ii) whether Uleiella predates its host plant; and (iii) whether Uleiella is ancestral to the smut fungi. Traditionally, the taxonomic position of the genus Uleiella within the Ustilaginomycotina has been uncertain. Limited ultrastructural data has placed it tentatively in the Ustilaginales [1,2].
In the present study, the evolutionary history of the gymnosperm smut fungus Uleiella chilensis on Araucaria araucana is inferred by comparison of subcellular and cellular features with those of related taxa and analysis of nuclear DNA sequences. The evolutionary age of Uleiella chilensis and its phylogenetic position within the Ustilaginomycotina are resolved in this study. To address the issue against an absence of fossil records for the Ustilaginomycotina, we assembled a dataset comprising 18S, 28S and rpb1 sequences from a representative sampling of Basidiomycota that was calibrated at two nodes, including the fossil ancestors of the orders Boletales and Agaricales. The results of this dataset were used as the basis for a secondary calibration of Ustilaginomycotina using a dataset that included 18S, ITS, 28S, rpb2 and EF1α sequences. To address the second question, we used the same datasets complemented with newly generated subcellular and cellular data to infer the phylogenetic placement of Uleiella chilensis within the Ustilaginomycotina.

Ethics statement
The fungal species used in this study were not protected and specimens were traded according to standard international herbaria policy and loan regulations. Additionally, the sampling sites of recently collected specimens were not protected and specific permits for sampling were not required. . Two DNA sequence datasets were compiled (i) to estimate the relative age of the Ustilaginomycotina within Basidiomycota (dataset 1), and (ii) to estimate the relative ages of Uleiella chilensis and major lineages within the Ustilaginomycotina (dataset 2). Dataset 1 was sampled from representatives in the main clades of the Agaricomycotina, Pucciniomycotina and Ustilaginomycotina, plus three ascomycetes as outgroup, including sequences from the 18S, 28S and rpb1 (exons B-C) genes for all 86 species. Dataset 2 was sampled from representatives of all ustilaginomyceteous fungi and Colacogloea peniophorae (Pucciniomycotina) as outgroup for which at least four of 18S, ITS, 28S, rpb2 and EF1α gene sequences were available from GenBank (24 species). For the species included in datasets 1 and 2, see S1 and S2 Tables, respectively. DNA extraction, PCR, sequencing and sequence editing Fungal genomic DNA was isolated using the InnuPREP Plant DNA Kit (Analytik Jena, Jena, Germany) following the standard protocol. For each sample, fungal material was ground in liquid nitrogen with a plastic pestle, suspended in 400 μL of extraction buffer and incubated for 1 hour at 50°C. PCR primers RPB1-A and RPB1-C were used to target domains A-C of rpb1 following the protocol of Matheny et al. [18]. For rpb2, regions 5-11 were amplified with the primer combinations RPB2-5F/RPB2-11bR [19], RPB2-5F/bRPB2-7.1 [20] and bRPB2-6F/ RPB2-11bR [19,20] and PCR conditions as described in [20]. The EF1α gene was amplified using the primer combinations EF-526F/EF-2218R, EF-526F/EF-ir and Ef-df/EF-2218R (S. Rehner, http://aftol.org/pdfs/EF1primer.pdf) using the protocol of Rehner & Buckley [21]. The internal transcribed spacer (ITS) region of the rDNA including the 5.8S rDNA and the 5'-end of the nuclear large subunit ribosomal DNA (LSU) were amplified using the primer pairs ITS1F/NL4 [22] and LR0R/LR9 (R. Vilgalys lab, http://biology.duke.edu/fungi/mycolab/ primers.htm; [23]) following the protocol of Riess et al. [24]. Positive PCRs were purified using ExoSap-IT 1 (USB Corporation, Cleveland, OH, USA) diluted 1:6. Sequencing of rpb1, rpb2, EF1α and ITS + LSU was carried out using the amplification primers and additional primers as described for rpb1 [25], rpb2 [26] and LR3R [23] and LR6 [27] for LSU. Cycle sequencing was accomplished using the BigDye Terminator v. 3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA). Sequencing reactions were run through an ABI Prism 3130xl Genetic Analyzer (Applied Biosystems). Sequence chromatograms were checked for accuracy and edited using Sequencher v. 4.9 (Gene Codes Corporation, Ann Arbor, MI, USA). DNA sequences obtained directly from herbarium specimens were compared to the sequences obtained from cultures (see below). The GenBank (http://www.ncbi.nlm.nih.gov) accession numbers for Uleiella chilensis sequences are KF061293 (rDNA), KF061319 (rpb1), KF061318 (rpb2) and KP413031 (EF1α).

Alignments and phylogenetic reconstructions
Sequences of the small (18S) and large subunit (28S) ribosomal DNA of dataset 1 were aligned independently using MAFFT v. 6.935b [28], applying the E-INS-i method [29]. Both alignments were automatically trimmed if more than 60% of all sequences exhibited gaps [30]. The nuclear DNA sequences from the rpb1 gene were aligned using DIALIGN-TX [31] and, in the case of rpb1, split into two exons. Highly divergent portions and alignment flanks were excluded using trimAl. Subsequently, the amino acid sequences translated from the DNA sequences were subjected to visual adjustments using Se-Al v. 2.0a11 [32]. Autapomorphic insertions and non-coding segments were removed from each alignment. Finally, all three gene alignments were concatenated (the resulting alignment length was 4398 bp, S1 Data). Ribosomal DNA (18S, ITS, 28S) and protein coding (rpb2, EF1α) sequences in dataset 2 were aligned and concatenated in the same way as described for dataset 1 (the final alignment length was 5767 bp, S2 Data).
For both datasets, maximum likelihood (ML) trees were computed using RAxML v. 8.1.3 [33] with 1000 bootstrap replicates (bootstrap support, BS) [34]. We used the general timereversible (GTR) substitution model and the CAT approximation to account for heterogeneity along different evolutionary branches. Additional posterior probability nodal support values were determined in a Bayesian phylogenetic Markov chain Monte Carlo (MCMC) search using MrBayes v. 3.2.2 [35] under the GTR model with a gamma-distributed rate variation. Each search comprised two runs of four chains each for 5 × 10 6 generations sampled every 100 generations with the first 2.5 × 10 5 generations discarded as burn-in.

Divergence time estimations
To estimate divergence times of Uleiella chilensis, we used the ML trees as starting trees for a MCMC-based time estimation in BEAST v. 1.8.1 [36]. We transformed branch lengths to ages, calibrating two nodes by using fossils [37,38] for the Basidiomycota (dataset 1) and used the time estimations for the Ustilaginomycotina within this dataset as a time constraint for the analysis of dataset 2. Calibrated clades were monophyletic in the starting trees and constrained as such in BEAST. For both datasets, 50 million generations were evaluated, sampling trees every 1000 generations with a burn-in of 10%. For more information on BEAST settings and the priors used, see S3 Data (Basidiomycota) and S4 Data (Ustilaginomycotina). After checking for convergence with Tracer v. 1.6 [39], consensus trees were calculated and the age estimations plus the highest density probabilities (HDPs) and posterior probabilities for all nodes were reported. We calibrated two nodes in the Basidiomycota dataset using fossil data: (i) the Suillinae (Suillus pictus, Chroogomphus rutilus and Gomphidius roseus) were calibrated based on a fossil of a Pinaceae-associated suilloid ectomycorrhiza (~50 mya) [37]; (ii) the Tricholomatoid clade (represented by Panellus stipticus and Pleurotopsis longinqua) was calibrated using a fossil Archaeomarasmius leggetti (~90 mya) from mid-Cretaceous amber [38]. Subsequently, we used the age estimation of the split between Ustanciosporium standleyanum and Schizonella melanogramma as well as the root node of the Ustilaginomycotina from the Basidiomycota dataset as secondary calibration points in dataset 2. The 95% HDP range was taken as a prior and the starting tree was calibrated with the mean age estimation as proposed by Forest [40].

Light and transmission electron microscopy
For the study of germination and the subsequent culture, spores were spread thinly on water agar (WA) and on malt-yeast-peptone agar (MYP) in Petri dishes kept at room temperature. As soon as germlings were produced, a suitable piece of medium (about 10 mm square) was cut out, transferred to a slide and covered with a cover glass. A small droplet of lactophenol with cotton blue, added to the side of the square of medium, fixed and stained the germlings. A culture was deposited in the Herbarium Tubingense culture collection (TUB F 4418). For light microscopy, living material was examined with an Axioplan microscope (Carl Zeiss) using bright field, phase contrast or Nomarski interference contrast optics. For transmission electron microscopy, fungal material from plant infected tissues and cultures were fixed overnight with 2% glutaraldehyde in 0.1 M of a sodium cacodylate buffer (pH 7.2) at room temperature. Following six transfers in 0.1 M of a sodium cacodylate buffer, samples were postfixed in 1% osmium tetroxide in the same buffer for 1 h in the dark, washed in distilled water and stained with 1% aqueous uranyl acetate for 1 h in the dark. After five washes in distilled water, samples were dehydrated in acetone, using 10-min changes at 25%, 50%, 70% and 95% and three times in 100% acetone. Samples were embedded in Spurr's plastic and sectioned with a diamond knife. Serial sections were mounted on formvar-coated single-slot copper grids, stained with lead citrate at room temperature for 5 min, washed with distilled water and studied with a EM 109 transmission electron microscope (Car Zeiss, Oberkochen, Germany) at 80 kV. For teliospore terminology see [41].

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi and plants and hence the new names contained in the electronic publication of a PLOS article are effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies.
In addition, new names contained in this work have been submitted to MycoBank from where they will be made available to the Global Names Index. The unique MycoBank number can be resolved and the associated information viewed through any standard web browser by appending the MycoBank number contained in this publication to the prefix http://www. mycobank.org/MB. The online version of this work is archived and available from the following digital repositories: PubMed Central, LOCKSS.
The monophyly of the Ustilaginomycetes and Exobasidiomycetes including the moniliellomyceous Moniliella acetoabutans was only supported by Bayesian inference. Malasseziomycetes represented by the yeast Malassezia furfur had an isolated and unresolved position (Fig  2). Several deep relationships between Ustilaginomycotina orders were only resolved with significant support values from Bayesian analyses (Fig 2). Uleiella chilensis was nested with Violaceomyces palustris with significantly high BS (99%) and PP (1.00) support values and a sister to Ustilaginales and Urocystidales (Ustilaginomycetes), supported with a BS value of less than 70 and PP 1.00 (Fig 2). ML and Bayesian analyses of D1/D2 LSU rDNA sequences from a wide taxonomic and phylogenetic spectrum of Ustilaginomycotina recovered congruent phylogenetic trees with relatively lower BS (72%) support and strong PP (95%) support for the monophyly of a group containing Uleiella chilensis as sister taxon to the yeasts Violaceomyces palustris (Violaceomycetales) and Tilletiopsis sp. (DQ404470) (data not shown).

Sporulation in host tissue and on artificial media
Uleiella chilensis sporulated exclusively on the surface of the host tissue (Figs 1 and 3a). Teliospores were produced singly and their wall consisted of an electron-opaque exosporium with reticulate ornamentation, occasionally embedded and partly covered by remnants of the sheath and the wall of the sporogenous hypha and an electron-transparent more or less two-lamellate endosporium (Fig 3b). During sporogenesis, the teliospores became multi-celled by septation (Fig 3c-3f). Subsequently, the teliospore segments appeared rounded giving the impression of "endospores" (Figs 3e and 4). Soral hyphae were thick-walled and filled the intercellular spaces. Aseptate haustoria arose from intercellular hyphal cells that contacted host cells. Haustoria were not constricted at the penetration point and extended only a short distance into the host cell. Haustoria terminated in the host cell and were surrounded by an electron-opaque matrix (Fig 5a). In the initial stages of interaction, a matrix began to develop and appeared on the host side between the host cell wall and the host cytoplasm. Occasionally, the matrix was arranged in two or more layers, separated from each other by a secondary layer of host origin (Fig 5a). Septa in soral hyphae were rare and thick-walled. Mature septa in soral hyphae and in cultural hyphae were poreless. Central swellings of variable size with an interrupted or branched electron-transparent middle layer or a plasmodesma-like perforation were usually present (Fig 5b).
Teliospore cells germinated after one day on WA and MYP with hyphae on which monokaryotic conidia arose asynchronously (Fig 4). Usually, conidia were produced on short germination tubes. The conidiogenous hyphae became zigzag in profile. Septa and branches developed so that the resulting cultures consisted of more or less pseudohyphae on which masses of conidia arose. On MYP (but not on WA), the formation of conidia stopped after a month, followed by the formation of thick-walled cells in chains. Simultaneously, the colour of the cultures changed from white to dark olivaceous green. Yeasts and ballistoconidia were absent.

Discussion
The results of the present study estimated smut fungi to have originated in the Ordovician period (~450 mya), which is in agreement with previous studies [42,43]. Our results indicate that most orders within the Ustilaginomycotina diverged during the Triassic-Jurassic period (Fig 2). Therefore, it appears that the origins of the ancestral lineages (the crown node) of smut fungi date back before the radiation of angiosperms and coincide with the major expansion of gymnosperms [44]. Considering the age estimates that place the origin (the stem node) of smut fungi before the diversification of vascular plants, as well as the lack of resolution in the basal nodes, we propose the following evolutionary history for the Ustilaginomycotina: Early smut fungi must have been living as saprobic organisms as no host plants would have existed at this time. This is also supported by the widely distributed yeast genus Malassezia, which diverged early and which is known to occur both as a saprophyte and as a parasite on animals [45]. It therefore seems appropriate to assume that the most recent common ancestor of Uleiellales and Violaceomycetales also had a saprobic lifestyle. The Violaceomycetales subsequently are specialised as endophytes of Pteridophyta or stayed saprobic [46], whereas the Uleiellales changed their lifestyle and became obligate parasites of the Araucaria lineage. In the genus Uleiella, only two extant species are known, U. chilensis and U. paradoxa. Interestingly, the respective hosts, Araucaria araucana and A. angustifolia, are closely related and are the only representatives of the genus in South America. The fact that Uleiella is restricted to South America while Araucaria species occur disjunct across the whole Southern Hemisphere might be resolved when we assume that the transition in the lifestyle of the Uleiellales happened after the separation of continents (this is also supported by the lower bound of the age estimate for U. chilensis and V. palustris). Furthermore, as both species of South American Araucaria host a unique species of Uleiella, we propose that it was either coevolution between this branch of Araucaria and the genus Uleiella, or that there was a host shift followed by allopatric speciation as the most likely scenarios explaining the evolutionary history. This evolutionary hypothesis is in agreement with the coevolutionary dynamics between hosts and parasites as postulated by [47].
After the emergence of the angiosperms, the orders in the Ustilaginomycotina that became associated with them underwent rapid diversification. The low diversity of the parasitic smut fungi on gymnosperms may be explained by low diversification and high extinction rates, as well as the geographic isolation of their hosts.
In our multi-gene analysis (Fig 2), Uleiella chilensis and Violaceomyces palustris clustered together forming a sister clade to the Urocystidales and Ustilaginales (Ustilaginomycetes). The relationship between Uleiella and the Urocystidales [1,2] is supported by the morphology of teliospores, which have a similar appearance compared to those of Mundkurella [48]. The lack of pores at the hyphal septa of Uleiella chilensis supports a closer relationship to the families Mycosyringaceae and Glomosporiaceae, which belong to the Urocystidales and almost all Ustilaginales (with the exception of Melanotaeniaceae). The presence of enlarged interaction zones [2] supports the placement in the class Ustilaginomycetes sensu Begerow et al. [49].
Interestingly, the asynchronous development of the conidia in Uleiella chilensis indicates that they do not represent basidiospores and, consequently, the teliospore germlings do not represent basidia. However, it is known that in many smut fungi, teliospore germination often depends on environmental conditions, ranging from true holobasidia to septate hyphae that sometimes bear conidia [50,51]. Possibly, teliospore germination in Uleiella chilensis represents an atypical germination resulting from an adaptation to extreme environmental factors. Araucaria araucana, the host of Uleiella chilensis, occurs in sites that experience extreme temperature and humidity conditions in the Andes and the Chilean coastal range. In order to better understand whether teliospore germination effectively represents an adaptive mechanism, further field investigations are needed. The ecology, the deep genetic divergence and the presence of haustoria and poreless septa characterize Uleiella as a unique evolutionary lineage within Ustilaginomycotina for which we describe a new order.   [52] and class Ustilaginomycetes [53] parasitizing gymnosperms, having haustoria and poreless septa. Equivalent to Uleiellaceae [54].
The newly described order is phylogenetically closely related to the order Violaceomycetales, but differs considerably in its ecology. Violaceomycetales includes a single species, Violaceomyces palustris that apparently occurs endophytically associated with Salvinia ferns from invaded mostly aquatic habitats in Louisiana, USA [46]. As Violaceomyces palustris and also Tilletiopsis sp. (DQ404470) are known only from their yeast phases and other cellular or subcellular features are unknown it is difficult to carry out morphological comparisons with Uleiella chilensis.
Supporting Information S1 Data. A concatenated alignment of dataset 1 containing the 18S, 28S and RPB1 sequences used for estimating the age of the Basidiomycota (4398 bp in length). For Gen-Bank accession numbers, see S1 Table. (NEXUS) S2 Data. A concatenated alignment of Dataset 2 containing the 18S, ITS, 28S, rpb2 and EF1α sequences used for estimating the age of the Ustilaginomycotina (5767 bp in length). For GenBank accession numbers, see S2 Table. (NEXUS) S3 Data. This file contains information about the priors and parameters used in BEAST to obtain the age estimates of the Basidiomycota (dataset 1). For GenBank accession numbers, see S1 Table. (XML) S4 Data. This file contains information about the priors and parameters used in BEAST to obtain the age estimates of the Ustilaginomycotina (dataset 2). For GenBank accession numbers, see S2 Table. (XML) S1 Fig. Chronogram of Basidiomycota evolution inferred from concatenated 18S, 28S and RPB1 sequences. Numbers on branches before slashes are ML bootstrap support values ( 70); numbers on branches after slashes are estimates for a posteriori probabilities ( 0.90). The ascomycetes Candida albicans, Taphrina deformans and Saccharomyces cerevisae were used as outgroup. The lines in bold indicate a maximum support of 100/1.00. The age estimation values (in million years ago, mya) are given for each node. The age estimation mean is followed by the 95% highest density probability (HDP) range in square brackets. Arrows indicate the nodes used for the secondary calibration (dataset 2). (PDF) S1