Cladosporium cladosporioides and Cladosporium pseudocladosporioides as potential new fungal antagonists of Puccinia horiana Henn., the causal agent of chrysanthemum white rust

Puccinia horiana Hennings, the causal agent of chrysanthemum white rust, is a worldwide quarantine organism and one of the most important fungal pathogens of Chrysanthemum × morifolium cultivars, which are used for cut flowers and as potted plants in commercial production regions of the world. It was previously reported to be controlled by Lecanicillium lecanii, Cladosporium sphaerospermum, C. uredinicola and Aphanocladium album, due to their antagonistic and hyperparasitic effects. We report novel antagonist species on Puccinia horiana. Fungi isolated from rust pustules in a commercial greenhouse from Villa Guerrero, México, were identified as Cladosporium cladosporioides and Cladosporium pseudocladosporioides based upon molecular analysis and morphological characters. The antagonism of C. cladosporioides and C. pseudocladosporioides on chrysanthemum white rust was studied using light and electron microscopy in vitro at the host/parasite interface. Cladosporium cladosporioides and C. pseudocladosporioides grew towards the white rust teliospores and colonized the sporogenous cells, but no direct penetration of teliospores was observed; however, the structure and cytoplasm of teliospores were altered. The two Cladosporium spp. were able to grow on media containing laminarin, but not when chitin was used as the sole carbon source; these results suggest that they are able to produce glucanases. Results from the study indicate that both Cladosporium species had potential as biological control agents of chrysanthemum white rust.

Introduction chrysanthemum white rust in México. Also, this study did not involve endangered or protected species.
Gray dusty mycelia were removed from P. horiana teliospores with a sterilized needle and transferred to synthetic PDA medium (Bioxon, Mexico). After incubation at 24˚C for 5 days, a spore suspension was prepared with sterilized water and transferred to water-agar medium. After incubation for 48h at 24˚C, single-spore mycelia were picked off and transferred to PDA medium to obtain pure cultures.
Morphological characteristics were determined following the standardized methodology of Schubert et al. [31] for identification of the Cladosporium species. Colonies grown, for 5 days at 24˚C in the dark, on synthetic-nutrient-limited media (SNA) plates, were used for morphological and morphometric observations of conidia, ramoconidia and conidiophores, using a photomicroscope Provis AX70 (Olympus, USA). Colony characteristics were determined after growing on PDA (Synthetic Potato-Dextrose-Agar; Bioxon, Mexico), MEA (Malt-Extract Agar; Bioxon, Mexico) and OA (Oat Agar), for 14 days at 24˚C in the dark.
Sequence analysis. The three gene sequences from each antagonist, Cladosporium cladosporioides and C. pseudocladosporioides, were aligned with the sequences available in GenBank (NCBI, USA). Our sequences were manually edited by CLCbio (Qiagen, USA). Sequence data obtained from Bensh et al. [18,35] were used as reference data for the alignments (Table 1). Multiple alignments were performed by ClustalW software and best nucleotide model determinate by jModelTest v. 2.1.7 [36] using BIC criteria for each locus and then incorporating it in the analysis. A Bayesian phylogenetic inference tree was generated based on data from each partition sequence of the three genes on BEAST v.1.8.1 [37] and Markov Chain Monte Carlo analysis, from four chains started from random tree topology and taken over 80 000 000 generations. Three independent runs were combined by LogCombiner v.1.8.1. Trees were saved each 1 000 generations, resulting in 80 001 saved trees. Using Tracer v.1.6, burn-in was set at 15 000 000 generations, after which the likelihood values were stationary. The coalescent algorithm with GTR+G+I substitution model and a lognormal uncorrelated relaxed clock was selected for the data. Maximum clade credibility tree was visualized by Fig Tree v.  the stability and robustness of each species, Neighboring-Joining analysis was performed for each data partition, using MEGA 6.0 [38] and 1000 replications using bootstrap. The ITS region has limited resolution for species in Cladosporium, therefore results for the ACT and EF-1α regions were used for comparison of clade stability (S1 Fig).

Antagonism assay
For the antagonism test, leaves infected (severity up 30%) and non-infected by chrysanthemum white rust, were collected from 20 different 30-day-old Chrysanthemum × morifolium cv. Polaris plants in a commercial greenhouse at Texcoco, México. The leaves were disinfested by immersion in sodium hypochlorite 3% for 3 min and then triple washed with sterilized water. There were five treatments: 1) antagonist 1 vs P. horiana (conidia were applied on pustules); 2) antagonist 2 vs P. horiana; 3) antagonist 1, conidia applied on healthy chrysanthemum leaves; 4) antagonist 2, conidia applied on healthy chrysanthemum leaves; and 5) control, P. horiana infected leaves treated with sterile water. Ten leaves per treatment were put into a humid chamber, each leaf representinged a repetition. A spore suspension (2×10 5 conidia mL -1 ) of each antagonist was sprayed onto pustules on diseased leaves and onto healthy chrysanthemum leaves. All treatments were incubated at 24˚C and 12 h light/dark. When signs of antagonists appeared, a sample of the fungus was cultured on synthetic PDA (Bioxon, Mexico) to confirm that it was the fungus originally inoculated.
After 96 h of incubation, antagonism percentages, measured as the proportion of pustules of P. horiana colonized by C. cladosporioides and C. pseudocladosporioides, were recorded. Differences in percentage were statistically tested by one-way ANOVA. To meet ANOVA assumptions, normal distribution was assessed by a Shapiro-Wilk test [P>0.05] and homogeneity of variance was evaluated by Levene's test [P>0.05]. The differences among treatments were tested by post hoc Ryan-Einot-Gabriel-Welch based on an F test (REGW-F; P = 0.05). All statistical analyses were carried out using SPSS Statistics 21.0.
Microscope observations. Leaves collected from Chrysanthemum × morifolium, infected and non-infected with chrysanthemum white rust, were treated under the same conditions and subjected to the same treatments as in the antagonism assay described above. After 36 h of incubation, 0.5 × 0.5 mm samples of the leaves were fixed in glutaraldehyde/paraformaldehyde 3:1 (in 0.2 M phosphate buffer, pH 6.8) overnight. Fixed leaf samples were washed four times with phosphate buffer for 15 min each, dehydrated through an ethanol series (30-100%, 1 h each) and infiltrated with LR-White [40]. Sections of 1μm thickness were cut on an ultramicrotome Reichert Jung Ultra E (USA) and stained with toluidine blue 1%. For SEM, pustules were dehydrated in a CO 2 vacuum, mounted on carbon tape and coated with gold. The samples were observed on a JEOL JSM6360LV low vacuum SEM (JEOL, USA).
Glucanase and chitinase production. Glucanase production of both isolates was evaluated on growth medium consisting of an agar synthetic medium (NaNO 3 , 0.2; KH 2 PO 4 , 0.1; MgSO 4 7H 2 O, 0.05; KCl, 0.05; agar 15 g/L and deionized water) supplemented with laminarin 1% (laminarin from Laminaria digitata, Sigma Aldrich, México) as a sole carbon source. Chitinase production was evaluated on the same synthetic basal medium but supplemented with colloidal chitin 1% rather than laminarin. Colloidal chitin was prepared using 10 g of purified crab shell chitin (Sigma Aldrich, México) suspended in 100 mL concentrated HCl for 2.5 h at 4˚C and then was washed with cold deionized water and NaOH overnight at 4˚C, followed by re-washing with cold deionized water at approximately pH 7.0. Mycelia discs (4 mm) were removed from PDA purified cultures with a sterilized needle and transferred to specific carbon-source media. After incubation at 24˚C for 15 days, growth was evaluated either by measurement of the diameter of the developing colonies in comparison with a negative control (basal medium without carbon source), or by comparison with a positive control (basal medium supplemented with D-glucose 1%) in threefold replication.

Morphological determination of Cladosporium spp.
Our collections from chrysanthemum plants yielded two different strains of fungus resembling the Cladosporium genus (Cladosporiaceae, Capnodiales) [18]. Based on the morphological characteristics, one of the isolated fungi was identified as Cladosporium cladosporioides (Fresen.) GA de Vries, and the second isolated fungus was identified as Cladosporium pseudocladosporioides Bensh, Braun & Crous (Fig 1). Their morphological characteristics are summarized in Table 2.
Sequence analysis of C. cladosporioides and C. pseudocladosporioides,

ITS, ACT & EF-1α
The analysis of the genomic sequences of internal transcribed spacer (ITS) region, actin (ACT) and translation elongation factor-1α (EF-1α) supported the morphological identification of the fungal isolates as C. pseudocladosporioides and C. cladosporioides. In order to determine inter-and intra-species phylogenetic similarities, based on sequence analysis and the origin of isolates, we performed a Bayesian analysis (Fig 2). All sequences were clustered into the Cladosporium genus, within the C. cladosporioides complex with strong support probabilities. Cladosporium cladosporioides, was clustered with CPC-11398 (from Phragmidium griseum) in one of three well defined sub-clades at the species core (Fig 2), consistently a member of C. cladosporioides. This species is composed of one of three well defined clades clustered with Cladosporium delicatulum, so in consequence is polyphyletic. Cladosporium pseudocladosporioides was clustered with CBS-176.82 (from Pteridium aquilinum) in one well defined sub-clade ( Fig  2); all isolates were clustered together, so corresponding to a monophyletic lineage and consequently representing a true member of C. pseudocladosporioides.

Antagonism assay
The telia of P. horiana treated only with water showed no alteration and had normal appearance (Fig 3A1 and 3B1). No C. cladosporioides or C. pseudocladosporioides conidia or hyphae were observed on inoculated leaf surfaces in areas devoid of P. horiana pustules (Fig 3A2 and  3B2). In contrast, 96 h after inoculation with Cladosporium isolates, pustules showed an appearance similar to that observed on leaves from diseased plants collected in the commercial greenhouse (Figs 3A3 to 3B3 and 4). The two Cladosporium isolates showed significant (P<0.05) parasitism on P. horiana pustules (Fig 3C). Pure cultures obtained from the parasitized pustules inoculated with the two antagonists exhibited the same morphological characteristics as the original isolates (data not shown).

Interaction of Cladosporium isolates and P. horiana examined under SEM and light microscopy
The morphology of Puccinia horiana corresponded to the recently available descriptions [6,9,41], and almost all plant tissue was colonized by the CWR (Fig 5B and 5C). The healthy chrysanthemum leaves were not parasitized by the Cladosporium spp. isolates (Figs 3 and 5A). Once Cladosporium cladosporioides and C. pseudocladosporioides conidia germinated, they began to develop an intimate and active physical association with P. horiana. Teliospores from pustules colonized by C. cladosporioides (Fig 5D-5I) and C. pseudocladosporioides (Fig 5J-5O)  were collapsed but no evidence of direct penetration was observed. Their cytoplasm was disrupted showing a vacuolated appearance and the wall structures were slimmed and collapsed causing the deformation of teliospores (Fig 5F and 5G for C. cladosporioides and Fig 5K, 5N and 5O for C. pseudocladosporioides). Further invasion and damage on P. horiana sporogenous cells occurred, and conidiophores of the antagonists were observed protruding from the pustules (Fig 5D-5O).

Growth in vitro on glucanase and chitinase media
On the specific nutritional carbon-source media, Cladosporium cladosporioides grew 49.66 ± 0.47 mm on laminarin 1%, 44.66 ± 0.47 mm on glucose 1% and has no growth on colloidal chitin 1%, in case of Cladosporium pseudocladosporioides grew 32.66 ± 3.29 mm on laminarin 1%, 36.00 ± 4.32 mm on glucose 1% and has no growth on colloidal chitin 1%. Both Cladosporium isolates grew on the medium with laminarin as a sole carbon source, but not on that prepared with colloidal chitin (Fig 6).

Discussion
Morphological identification of Cladosporium spp. has been a difficult subject. Conidiophore and conidia size and shape are important characters, but usually dimensions overlap among species in the genus. However, molecular analysis has been a useful approach for the identification of Cladosporium species. Discrimination between C. cladosporioides, C. pseudocladosporioides and other taxa in the C. cladosporioides complex was made by Bensh et al., until 2010 [35], based on molecular phylogeny using ITS, ACT and EF-1α regions. These regions were used by Bensh et al. [18] to explain diversity and evolutionary trends in the Cladosporium genus. ITS alone does not give good species resolution [39] but ACT and EF-1α, in contrast, demonstrate a high degree of divergence among species [35,42]. In the present study, we used an integrated approach based on the analysis of both molecular and morphological characters to determine the mycoparasitic species isolates as C. cladosporioides and C. Velvety, brown-reddish with dark margins, reverse brown-reddish color.

On OA medium
Velvety olivaceous to grey-olivaceous color with aerial mycelia, reverse grey-olivaceous color.
Velvety grey to grey-brown, reverse greybrown color. pseudocladosporioides. Regarding intra-species relationships, our C. cladosporioides isolate showed some degree of diversification, as it clustered with another rust fungicolous isolate in a clearly supported clade inside core C. cladosporioides; likewise, there was strong clade support in C. pseudocladosporioides and C. cladosporioides lineages. Our results are consistent with Bensh et al. [35], and support the possible presence of cryptic species complexes on C. pseudocladosporioides and C. cladosporioides lineages. Both species are widely distributed and well adapted to various environments [18,35]. Cladosporium cladosporioides has already been reported parasitizing other fungi, such as Venturia inaequalis [29], Erisyphe cichoracearum [43], Botrytis fabae [44], Sclerotinia sclerotiorum [45] and rust fungi such as Puccinia graminis f.sp. tritici [30]. To our knowledge, C. pseudocladosporioides has not previously been reported parasitizing another fungus. This is the first report of both species potentially parasitizing P. horiana telia, and they occur naturally on this rust. The association of both Cladosporium spp. on the sporogenous cells, without direct penetration of spores, was previously reported in Cladosporium sp. parasitizing Exobasidium camelliae var. gracilis [46], C. phylophillum on Taphrina sp., C. exobasidii on Exobasidium vaccinii and Exobasidium warmingii, and C. epichloës on Epichloë typhina [47]. This kind of relationship, without direct penetration, was probably due to differences between the teliospores and sporogenous cells of P. horiana, such as wall structural complexity between the stroma and spores, as well as different arrangement and proportion of chitin, glucans, glycoproteins, melanin and some other structural chemical compounds between different structures, as reported on Puccinia graminis and some other Pucciniales [48,49,50,51], and it has been hypothesized that β-1,3 glucanases could be determinant for this Cladosporium spp. nutritional and spatial association [25]. Although there was evidence that C. cladosporioides and C. pseudocladosporioides were able to grow on laminarin media as the sole carbon source, we are not certain if these fungi excrete glucanases to parasitize P. horiana; however, the involvement of this enzyme in mycoparasitic Cladosporium relationships was previously reported for C. tenuissimum against Uromyces appendiculatus [25]. It is also possible that antibiotic mechanisms were affecting teliospore morphology and had repercussions probably in terms of potential loss of viability of teliospores, as previously reported in almost all mycoparasitic relationships of strains of C. tenuissimum [25,26,27], C. uredinicola [21,22,23], C. gallicola [28], C. aecidiicola [24] and Cladosporium sp. [17,52]. Various antifungal compounds have been reported and isolated from some C. cladosporioides strains [53,54] and C. pseudocladosporioides has shown some antibiotic activity [55].
Currently, control of chrysanthemum white rust in greenhouses and semi-covered growing systems is focused mainly on fungicides and some resistant cultivars. The potential resistance of P. horiana to fungicides represents a challenge for the development of new schemes to    damage by chrysanthemum white rust. Although efforts have been made to apply biological control against this disease using Verticillium lecanii, Aphanocladium album or Cladosporium spp. [4,15,16,17], so far none of them are widely used. Since C. cladosporioides and C. pseudocladosporioides isolates altered the morphology of teliospores, and possibly reduced both viability and production, they might have potential for chrysanthemum white rust management in an integrated disease management scheme. Further studies must be carried out as, among other things, it is necessary to know the disease's biology, ecology and mycoparasitism under controlled and commercial conditions, and to ascertain the antagonistic activity of the hyperparasites and their role in nature. In addition, it must be confirmed that these isolates do not harm other crops and that they have no adverse effects on humans or other animals. The present study provides a basis for such further studies.  Table 1). (TIF)

Acknowledgments
The first author (David Eduardo Torres) expresses his appreciation to CONACyT for his Mas-ter´s scholarship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We thank M.C. Mónica Pérez Pacheco from Laboratorio de Desarrollo en Plantas, Departamento de Biología Comparada, Facultad de Ciencias, UNAM and Biol. Yolanda Hornelas Cruces from Laboratorio de Microscopia Two Cladosporium species as potential new fungal antagonists of Puccinia horiana