Diversity of Moesziomyces (Ustilaginales, Ustilaginomycotina) on Echinochloa and Leersia (Poaceae)

Abstract A combined ecological, morphological, and molecular approach was used to examine 26 herbarium specimens and eight strains of Moesziomyces. The phylogenetic analysis resolved eight well-supported clades, of which three contained type specimens of known species of Moesziomyces. One clade contained two specimens that produced a teleomorph in the flowers of Echinochloakimberleyensis in Australia. The name Moesziomyceskimberleyensis is proposed for this smut fungus. Another clade contained specimens that produced sori in the flowers of Leersiahexandra. The name Thecaphoraglobuligera (now Moesziomycesglobuligerus) is available for this species, which is lectotypified. The teleomorph of Moesziomycesantarcticus, previously known only from Japan, is found for the first time in China, on Echinochloacrus-galli.

The teleomorphs of Ustilaginaceae are mostly host specific (Stoll et al. 2003(Stoll et al. , 2005Skibbe et al. 2010;McTaggart et al. 2012;Li et al. 2017aLi et al. , 2017b. Given that species of Moesziomyces have been reported from seven different genera of grasses (Echinochloa, Leersia, Panicum, Paspalum, Pennisetum, Polytrias, and Uranthoecium), it is likely that additional species remain to be discovered. The aim of this study was to build on the work of Kruse et al. (2017) by examining specimens of Moesziomyces held in herbaria BRIP (Queensland Plant Pathology Herbarium), HMAS (Herbarium Mycologicum Academiae Sinicae), and HUV (Herbarium Ustilaginales Vánky, now deposited in BRIP), as well as eight yeast strains deposited in LC Culture Collection (personal culture collection held in the laboratory of Dr Lei Cai).

Specimen examination
Specimens borrowed from several herbaria were examined by light microscopy (Table  1) by mounting the spores in lactic acid (100% v/v). Teliospore measurements were expressed as ranges (min-) mean-standard deviation-mean + standard deviation (-max) (n = 50). Images were captured by using a Nikon DS-Fi1 camera attached to a Nikon Eclipse 80i microscope with Nomarski differential interference contrast. Helicon Focus ver. 4.46.1 (Helicon Soft Ltd) was used to combine images to increase depth of field. Nomenclatural novelties and descriptions were registered in MycoBank (http:// www.MycoBank.org).

DNA extraction, PCR amplification and sequencing
Sori were carefully removed from herbarium specimens, up to 149 years old, with a fine needle, sterilized by dipping in 75% ethanol for 30 s, air-dried on sterilized filter paper, and deposited in cell lysis solution (CTAB). Pure yeast colonies grown on yeast extract peptone dextrose (YPD) plates were transferred to cell lysis solution directly. Genomic DNA was extracted following the protocol of Cubero et al. (1999). Fragments of internal transcribed spacer rDNA were amplified by PCR with primers M-ITS1/ITS4 (White et al. 1990;Stoll et al. 2003).
PCR amplifications were carried out in 25 μl reactions containing 1 μl of genomic DNA template, 9.5 μl distilled water, 12.5 μl of 2 X Taq Plus Master Mix (Nanjing Vazyme Biotech Co. Ltd, Nanjing, China) and 1 μl of each primer (10 μM). Amplification reactions were run as follows: initial denaturation of 95 °C for 5 min followed by 35 cycles at 95 °C for 30 s, 45 s at 58 °C (annealing temperature) and 1 min at 72 °C with a final extension of 10 min at 72 °C. PCR products were sent to Tianyihuiyuan (Beijing, China) for sequencing with the forward and reverse primers indicated above. AB1 sequence traces were assembled with Sequencher version 5 (Genecodes, Ann Arbor, USA).

Phylogenetic analyses
The sequences included in this study (Tables 1, 2) were aligned online with MAFFT (https://mafft.cbrc.jp/alignment/server/index.html) using auto strategy, and observed in MEGA 5 (Katoh and Toh 2008). Phylogenetic analyses were based on both maximum likelihood (ML) and Bayesian Inference (BI). RAxML (Stamatakis 2006) and PhyML 3.0 (Guindon et al. 2010) were used for ML analyses. GTRGAMMA was specified as the model of evolution in both programs. The RAxML analyses were run with a rapid Bootstrap analysis (command -f a) using a random starting tree and 1 000 ML bootstrap replicates. The PhyML analyses were implemented using the ATGC bioinformatics platform (available at: http://www.atgcmontpellier.fr/phyml/), with six substitution type and SPR tree improvement, and support obtained from an approximate likelihood ratio test (Anisimova et al. 2011).
For BI, MrBayes was used with a Markov Chain Monte Carlo algorithm incorporating four runs, each consisting of four chains, until the standard deviation of split frequencies was reached. The cold chain was heated at a temperature of 0.25. Substitution model parameters were sampled every 50 generations and trees were saved every 5000 generations. Convergence of the Bayesian analysis was confirmed using AWTY (Nylander et al. 2008) (available at: http://ceb.csit.fsu.edu/awty/). A user-defined tree obtained from the PhyML analyses was used as a starting point for all the Bayesian analyses, which helped to improve convergence of the four runs.

Results
The ITS dataset comprised the newly sequenced Moesziomyces specimens and strains (Table 1) Table 2) and Triodiomyces altilis and Ustilago echinata as the outgroup based on the phylogenetic analyses of Wang et al. (2015). The topology of the ML and BI analyses ( Fig. 1) were congruent. The phylogenetic

Discussion
The phylogenetic analyses in this study supported the host specificity of the teleomorphic stage of six species of Moesziomyces, specifically, M. antarcticus on Echinochloa crus-galli, M. bullatus on E. crus-galli and E. muricata, M. globuligerus on Leersia hexandra, M. kimberleyensis on E. kimberleyensis, M. penicillariae on Pennisetum glaucum, and M. verrucosus on Paspalum distichum. The teleomorph of M. eriocauli may be specific to Eriocaulon spp., although this cannot be ascertained from the sequence data of one specimen. Specimens that have been assigned to M. bullatus were not well resolved and formed a number of smaller clades with varying degrees of support (Fig. 1). The M. bullatus clade contained several anamorphic yeasts isolated from diverse habitats (Wang et al. 2015;Kruse et al. 2017), including leaves of Digitaria sp., Pennisetum sp., and Setaria faberii. This shows that the anamorphs of Moesziomyces are widespread in the environment as saprobes.
The anamorphs of Moesziomyces, together with most members of the Ustilaginales, have a dimorphic lifecycle comprised of a parasitic dikaryotic phase characterized by teliospores, together with a saprobic yeast-like haploid phase (Brefeld 1883;de Bary 1884;Sampson 1939;Begerow et al. 2014). The teliospores are generally thick-walled and darkened, which protects against desiccation and UV radiation, thereby facilitating survival and long-distance dispersal (Piepenbring et al. 1998). The basidiospores are usually thin-walled, hyaline, and survive as free-living saprobic yeasts that may occur on a vast diversity of substrates (Wang et al. 2015;Kruse et al. 2017;Tanaka et al. 2019). There is genomic evidence that some saprobic ustilaginalean yeasts, e.g. M. antarcticus, Kalmanozyma brasiliensis (= P. brasiliensis), Pseudozyma hubeiensis, and the yeast stage of M. bullatus (= P. aphidis), have retained the capacity to produce effector proteins, which hints at the possibility that undiscovered plant pathogenic stages may exist for these fungi (Sharma et al. 2018). Indeed, a teleomorph for M. antarcticus (=P. antarctica) was recently reported for the first time on Echinochloa crus-galli (Tanaka et al. 2019). Further collections are needed to resolve the ecological relationships and elucidate the life cycles of the ustilaginalean fungi and their hosts.