Unveiling the Hidden Diversity of Rock-Inhabiting Fungi: Chaetothyriales from China

Rock-inhabiting fungi (RIF) are nonlichenized fungi that naturally colonize rock surfaces and subsurfaces. The extremely slow growth rate and lack of distinguishing morphological characteristics of RIF resulted in a poor understanding on their biodiversity. In this study, we surveyed RIF colonizing historical stone monuments and natural rock formations from throughout China. Among over 1000 isolates, after preliminary delimitation using the internal transcribed spacer region (ITS) sequences, representative isolates belonging to Trichomeriaceae and Herpotrichiellaceae were selected for a combined analysis of ITS and the nuclear ribosomal large subunit (nucLSU) to determine the generic placements. Eight clades representing seven known genera and one new genus herein named as Anthracina were placed in Trichomeriaceae. While, for Herpotrichiellaceae, two clades corresponded to two genera: Cladophialophora and Exophiala. Fine-scale phylogenetic analyses using combined sequences of the partial actin gene (ACT), ITS, mitochondrial small subunit ribosomal DNA (mtSSU), nucLSU, the largest subunit of RNA polymerase II (RPB1), small subunit of nuclear ribosomal RNA gene (SSU), translation elongation factor (TEF), and β-tubulin gene (TUB) revealed that these strains represented 11 and 6 new species, respectively, in Trichomeriaceae and Herpotrichiellaceae. The 17 new species were described, illustrated for their morphologies and compared with similar taxa. Our study demonstrated that the diversity of RIF is surprisingly high and still poorly understood. In addition, a rapid strategy for classifying RIF was proposed to determine the generic and familial placements through preliminary ITS and nucLSU analyses, followed by combined analyses of five loci selected from ACT, ITS, mtSSU, nucLSU, RPB1, and/or the second subunit of RNA polymerase II gene (RPB2), SSU, TEF, and TUB regions to classify RIF to the species level.


Introduction
Natural and manmade rock surfaces harbor a high diversity of lichenized and nonlichenized fungi, algae, and bacteria [1][2][3]. This niche harbors a polyphyletic assemblage of stress-tolerant

Isolation of Fungi
Fungi were isolated using previous methods [47], except that surface-disinfected samples were dried on filter papers before processing. Rock samples with black colonies were cut into pieces (2-3 mm 3 ) with an industrial stone splitter (Model CM-10, Hydrasplit, Park Industries, Inc., St. Cloud, MN, USA). Rock pieces were surface-disinfected with 95% (v/v) ethanol for 3-5 s, transferred to physiological saline containing 0.001% (v/v) tween-20, washed with sterilized distilled water, and dried on sterilized filter papers. About 1 g of each sample was pulverized in a sterilized mortar, and the powder was suspended in 2-3 mL of sterile water. Aliquot of suspensions (200 µL) were spread evenly onto DRBC in Petri plates (with 100-mg/L streptomycin added after autoclaving). Plates were incubated at 10 • C for 4 weeks. Darkly pigmented colonies developing on plates were transferred to malt extract agar (MEA; malt extract 1%, peptone 1% , glucose 1%, and agar 1.5%) in Petri plates for purification and identification. Type specimens were deposited in the Herbarium of Mycology, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China (HMAS), and ex-type living cultures were deposited in the China General Microbiological Culture Collection Center (CGMCC).

Morphology
Colony morphologies on 2% malt extract agar (MEA; BD, Sparks, MD USA) plates were observed and described after six weeks of incubation at 25 • C. Microscopic morphology was examined using slide cultures [48]. Digital images of samples were acquired using a Nikon Eclipse 80i (Tokyo, Japan) microscope and were digitally optimized with the software NIS-Elements Documentation (v. 4.00.00). The modified slide culture method was applied to observe the sporulation [49]. Pieces of 2% MEA (1 cm × 1 cm) were placed above two layers of sterilized filter paper on the bottom of a Petri dish and kept moisture with sterilized water. The edge of each agar block was inoculated with a small amount of mycelium, and a sterile cover slip was placed over the inoculated agar. After 30-60 days, the cover slide was removed from the agar block and then was placed on a slide with water mounting for morphological observation. At least 20 measurements per structure were conducted under a microscope with differential interference contrast (DIC).

DNA Extraction, PCR Amplification, and Sequencing
Genomic DNA was extracted from fungal mycelia on MEA plates, as previous described [50] and amended with cell disruption by beating for 45 s and 50 s before being incubated at 85 • C in a water bath for 45-60 min. ACT (the partial actin gene), ITS, mtSSU (mitochondrial small subunit ribosomal DNA), nucLSU (the nuclear ribosomal large subunit), RPB1 (the largest subunit of RNA polymerase II), SSU (the small subunit of nuclear ribosomal RNA gene), TEF (translation elongation factor), and TUB (the partial β-tubulin gene) fragments were PCR-amplified with corresponding primers and annealing temperatures (Table 1). PCR amplifications were made from 30-µL reaction mixtures containing 1.0-µL DNA template, 1.0-µL each forward and reverse primers, 15 µL of 2× MasterMix (TIANGEN Co. Ltd. Beijing, China), and 12 µL of H 2 O. The PCR parameters were 94 • C for 40 s, followed by 40 cycles at the temperatures appropriate for the gene amplified (Table 1) for 60 s, 72 • C for 90 s, and a final extension at 72 • C for 10 min. The PCR products were sequenced by SinoGenoMax Co. Ltd. (Beijing, China). The sequenced amplicons were compared with those in the GenBank database via BLAST searching to find the most similar taxonomic designations (Table S1). All novel sequences obtained in this study were deposited in the NCBI GenBank database (Table S1), and the final matrices were used for phylogenetic analyses in TreeBASE (www.treebase.org; accession number: S22289).

Alignment and Phylogenetic Analyses
Sequences of closely related fungi were obtained from GenBank (Table S1) and uploaded to MAFFT version 7 (http://www.ebi.ac.uk/Tools/msa/mafft/) for alignment [60]. Sequence alignments were manually adjusted with MEGA v. 5 [61]. Phylogenetic reconstructions were estimated using maximum likelihood (ML) and Bayesian inference (BI). ML analyses were run with RAXML-HPV v. 0.3 [62]. The RAxML software accommodated the GTR + GAMMA model. For the Bayesian analyses, the models of evolution were estimated by MrModeltest v. 2.3 [63]. Posterior probabilities (PP) [64] were determined by Markov Chain Monte Carlo sampling (BMCMC) in MrBayes 3.0b4 [65] under the estimated model of evolution. Four simultaneous Markov chains were run for 1,000,000 generations, and trees were sampled every 100 generations (resulting in 10,000 total trees). The first 2500 trees, which represented the burn-in phase of the analysis, were discarded, and the remaining 7500 trees were used for calculating PP values in a majority rule consensus tree.
To build a phylogenetic framework (Figure 2), the concatenated alignment of ITS (1-862) and nucLSU (863-1771) regions consisted of 1771 characters for 136 strains. All characters were assessed to be unordered and equally weighed. Missing data were treated as question marks. Based on the Akaike Information Criterion (AIC) criteria, evolution models, including a GTR + I + G with ingamma-distributed rates for ITS and nucLSU, were selected for the partitioned Bayesian inference. To resolve species boundaries, the dataset of combined sequences (3745 characters, including gaps) of five loci, including ITS (1-635), nucLSU (636-1484), mtSSU (1485-2180), RPB1 (2181-2845), and TUB (2846-3745), were aligned and used for phylogenetic reconstruction for 32 taxa, including the outgroup taxon, Exophiala bonariae CCFEE5792 (Figure 3). Ambiguous regions and introns were delimited manually and excluded from the alignments. RPB1 and TUB genes that failed to amplify from some isolates were replaced by question marks. Based on the AIC criteria, the evolution models, including a GTR + I + G with inverse ingamma-distributed rates for ITS, TUB, and RPB1; a GTR + I + G with propinv-distributed rates for nucLSU; and a GTR + G with gamma-distributed rates for mtSSU, were selected for the partitioned Bayesian inference. Trees were visualized in FigTree Figure 3. Phylogenetic tree generated by the maximum likelihood analysis using combined sequences of ITS, nucLSU, mitochondrial small subunit ribosomal DNA (mtSSU), the largest subunit of RNA polymerase II (RPB1), and the partial β-tubulin gene (TUB) loci of the family Trichomeriaceae. Bootstrap values ≥70% (left) and Bayesian posterior probability values ≥0.95 (right) are indicated at nodes (ML/BI). Thickened branches represent posterior probabilities (>0.95) from BI. Novel sequences generated in this study are indicated in bold. Exophiala bonariae strain CCFEE 5792 is used as the outgroup. Ex-type cultures are marked with " T ".
For Cladophialophora, the 5-locus sequence alignment (ITS: 1-703 characters, SSU: 704-2420 characters, TEF: 2421-2658 characters, TUB: 2659-3167, and nucLSU: 3168-4066 characters) for a total of 51 taxa was analyzed, selecting C. proteae as the outgroup. For the Bayesian analyses, HKY + I + G was selected as the best fit model for SSU and TUB, SYM+G for TEF, SYM + I + G for ITS, and GTR + I + G for nucLSU. The phylogenetic tree is shown in Figure 4. For Exophiala, the multi-locus sequence alignment (ITS: 1-750 characters, SSU: 751-2456 characters, TEF: 2457-2711 characters, TUB: 2712-3173 characters, and ACT: 3174-3716 characters) of the total 64 taxa was analyzed using E. placitae as the outgroup. For the Bayesian analyses, GTR + I + G was selected as the best model for ACT, TEF, and ITS and HKY + I + G for EF and TUB. The phylogenetic tree is shown in Figure 5. Phylogenetic tree generated by the maximum likelihood analysis using combined sequences of ITS, nucLSU, mitochondrial small subunit ribosomal DNA (mtSSU), the largest subunit of RNA polymerase II (RPB1), and the partial β-tubulin gene (TUB) loci of the family Trichomeriaceae. Bootstrap values ≥70% (left) and Bayesian posterior probability values ≥0.95 (right) are indicated at nodes (ML/BI). Thickened branches represent posterior probabilities (>0.95) from BI. Novel sequences generated in this study are indicated in bold. Exophiala bonariae strain CCFEE 5792 is used as the outgroup. Ex-type cultures are marked with " T ".
For Cladophialophora, the 5-locus sequence alignment (ITS: 1-703 characters, SSU: 704-2420 characters, TEF: 2421-2658 characters, TUB: 2659-3167, and nucLSU: 3168-4066 characters) for a total of 51 taxa was analyzed, selecting C. proteae as the outgroup. For the Bayesian analyses, HKY + I + G was selected as the best fit model for SSU and TUB, SYM+G for TEF, SYM + I + G for ITS, and GTR + I + G for nucLSU. The phylogenetic tree is shown in Figure 4. For Exophiala, the multi-locus sequence alignment (ITS: 1-750 characters, SSU: 751-2456 characters, TEF: 2457-2711 characters, TUB: 2712-3173 characters, and ACT: 3174-3716 characters) of the total 64 taxa was analyzed using E. placitae as the outgroup. For the Bayesian analyses, GTR + I + G was selected as the best model for ACT, TEF, and ITS and HKY + I + G for EF and TUB. The phylogenetic tree is shown in Figure 5.

Multi-Locus Phylogeny
The Bayesian and ML trees resulted in similar topologies-two distinct in Trichomeriaceae and Herpotrichiellaceae, as well as robustly supported monophyletic clades in the order Chaetothyriales-were evident ( Figure 2).
The phylogenic tree based on a combined sequences of the ITS and nucLSU indicated that the selected 47 isolates (in bold) were distributed among 11 independent clades, of which six are from Knufia (clade A), Bradymyces (clade C), Lithohypha (clade E), Trichomerium (clade H), Exophiala (clade I), and Cladophialophora (clade J), respectively. We propose that clade G, clearly delineated with a high bootstrap value, represents a new genus in the Trichomeriaceae of the order Chaetothyriales ( Figure 2). We propose a new genus, Anthracina, for clade G and describe the morphological characteristics of two new species in the genus ( Figure 2).
Culture characteristics: Colonies on MEA extremely slow-growing, attaining 4-mm-diam. after 20 weeks at 25 • C, compact, center-raised, deep olive-gray to olive-black, with scant, velvety aerial mycelia, flat and glossy near the periphery; black in reverse ( Figure  Notes: The cylindrical, ampulliform, and obovate hyphae of A. saxincola apparently differ from other genera, such as Knufia, Bradymyces, and Trichomerium [22,24,27,28,32], in the same family. A. ramosa and A. saxincola were clustered into a clade with strong bootstrap support (MLBP/BIPP = 100%/1.00) ( Figure 2). However, the growth rate of A. ramosa is much faster (14 mm in 20 weeks on MEA at 25 • C) than that of A. saxincola (4 mm in 20 weeks on MEA at 25 • C). A. saxincola also produced multicellular bodies in the aged cultures, and endoconidia were also occasionally observed in enlarged cells or in multicellular bodies, while those were not observed in the culture of A. ramosa.
Bradymyces  Figure 7G) and multicellular bodies ( Figure 7H-J) globose to ellipsoidal, brown, 5.2-9.2 × 9.2-15.8 μm (⎯x = 6.8 × 13.5 μm, n = 10). Endoconidia (arrow in Figure 7K) were occasionally observed in enlarged cells or in multicellular bodies. Culture characteristics: Colonies on MEA extremely slow-growing, attaining 4-mm-diam. after 20 weeks at 25 °C, compact, center-raised, deep olive-gray to olive-black, with scant, velvety aerial mycelia, flat and glossy near the periphery; black in reverse ( Figure  Notes: The cylindrical, ampulliform, and obovate hyphae of A. saxincola apparently differ from other genera, such as Knufia, Bradymyces, and Trichomerium [22,24,27,28,32], in the same family. A. ramosa and A. saxincola were clustered into a clade with strong bootstrap support (MLBP/BIPP = 100%/1.00) ( Figure 2). However, the growth rate of A. ramosa is much faster (14 mm in 20 weeks on MEA at 25 °C) than that of A. saxincola (4 mm in 20 weeks on MEA at 25 °C). A. saxincola also produced multicellular bodies in the aged cultures, and endoconidia were also occasionally observed in enlarged cells or in multicellular bodies, while those were not observed in the culture of A. ramosa.
Culture characteristics: Colonies on MEA growing slowly, attaining 3-mm-diam. after four weeks at 25 • C, blackish-brown to olivaceous black, with a velvety, aerial short hyphae, regular margin, black in reverse ( Figure 11A Notes: The difference between K. separata and K. karalitana is the growth rate on MEA, with the latter growing faster (8 mm in three weeks) than the former (3 mm in four weeks) [11]. Culture characteristics: Colonies on MEA growing slowly, attaining 3-mm-diam. after four weeks at 25 °C, blackish-brown to olivaceous black, with a velvety, aerial short hyphae, regular margin, black in reverse ( Figure 11A Notes: The difference between K. separata and K. karalitana is the growth rate on MEA, with the latter growing faster (8 mm in three weeks) than the former (3 mm in four weeks) [11].
Culture characters: Colonies on MEA growing slowly, attaining 15-mm-diam. after 20 weeks at 25 • C, dark brown, velvety, with long, grayish-brown aerial hyphae, black irregular margin, black in reverse ( Figure 14A Notes: Phylogenetic analyses and high bootstrap support values indicated that T. flexuosum is closely related to T. foliicola (MLBP/BIPP = 89%/1.00) (Figure 2). However, T. foliicola is known to produce a sexual morph, which remains unknown for T. flexuosum. The growth of T. flexuosum is slower than that of T. foliicola [27].
Notes: Phylogenetic analyses and high bootstrap support values indicated that T. flexuosum is closely related to T. foliicola (MLBP/BIPP = 89%/1.00) (Figure 2). However, T. foliicola is known to produce a sexual morph, which remains unknown for T. flexuosum. The growth of T. flexuosum is slower than that of T. foliicola [27].
Culture characters: Colony on MEA growing slowly, attaining 12-mm-diam. after four weeks at 25 • C. Colony surface cerebriform, arise centrally, velvety with grayish short aerial hyphae and lobate margin; reverse deep olivaceous gray ( Figure 20A Culture characters: Colony on MEA growing slowly, attaining 34-mm-diam. after four weeks at 25 °C. Colony surface velvety with olivaceous grey short aerial hyphae and margin irregular, reverse olivaceous black ( Figure 22A,B) Figure 5). However, they can be distinguished by conidial morphology, and E. brunnea has narrower conidia (2-3 μm vs. 2.6-5.0 μm) compared with the new species. In the meanwhile, the budding cells are absent for E. brunnea [36] but present for the new species.

Discussion
The numbers of genera and species of RIF continues to expand [11,16,71,72]. Nonetheless, rock surfaces remain an underexplored habitat for fungi. The application of phylogenetic analysis to classify strains and delimit new genera and species has led to a significant expansion of new taxa in Culture characters: Colony on MEA growing slowly, attaining 34-mm-diam. after four weeks at 25 • C. Colony surface velvety with olivaceous grey short aerial hyphae and margin irregular, reverse olivaceous black ( Figure 22A,B) Figure 5). However, they can be distinguished by conidial morphology, and E. brunnea has narrower conidia (2-3 µm vs. 2.6-5.0 µm) compared with the new species. In the meanwhile, the budding cells are absent for E. brunnea [36] but present for the new species.

Discussion
The numbers of genera and species of RIF continues to expand [11,16,71,72]. Nonetheless, rock surfaces remain an underexplored habitat for fungi. The application of phylogenetic analysis to classify strains and delimit new genera and species has led to a significant expansion of new taxa in Dothidomycetes [19,21,73,74]. However, to our knowledge, RIF in the Eurotiomycetes have been poorly investigated [11,75]. During our preliminary classification of Chinese RIF using ITS barcoding, a number of strains were assignable to novel lineages in Chaetothyriales. In this study, we examined 31 RIF isolates affiliated in Trichomeriaceae. As a result, we established a new genus and 11 new species. The richness and broad distribution of RIF in Trichomeriaceae suggests that the RIF are the dominant and ubiquitous component of the rock microbiota. The overwhelming number of new species discovery indicates higher species than that previously estimated, and more investigation in extensive niches should likely reveal more new taxa.
Exophiala and Cladophialophora are two large and frequently encountered genera in Herpotrihiellaceae, but their taxonomy are problematic because of their less morphological characteristics. Traditionally, the conidia appear to be highly taxonomically informative at the species level of Exophiala and are widely used for the taxonomy. However, along with increased discovery of the new taxa, species delimitation in the genus Exophiala became difficult, because the conidial sizes frequently overlapped among morphologically similar but phylogenically distinct species, weakening the traditional morphological species concept. For instance, conidial sizes of E. cinerea (3.6-8.8 × 2.7-5.4 µm) and E. nagquensis (4.8-10.4 × 2.6-5.0 µm) are similar, but they are phylogenetically distinct ( Figure 5). Therefore, the taxonomy of Exophiala should consider both morphological characteristics and phylogenetic relationships rather than relying on only one type of characters.
The species of Cladophialophora and Exophiala have very diverse habitats, such as plants, fruit juices, shower rooms, seawater, sports drinks, arable soil, wood pulp, oil sludge, and the decaying shell of babassu coconut [36,37,40,68,69,76], and have also been reported as opportunistic pathogens on the superficial skin or internal organs in humans and animals [35,37,38,44,46,70,77,78]. In this study, we isolated Cladophialophora and Exophiala mostly from rocks, indicating that rock is an important habitat for these species. Moreover, as a rock surface is normally dry with high solar irradiation, temperatures, and osmotic stress, the condition of the human body is in high temperature, rich in nutrients and high osmotic pressure. It is possible that species of Cladophialophora and Exophiala were isolated from and adapted to rocks.
Further extensive samplings and investigation of those fungi are necessary to generate a more complete knowledge about their biodiversity, distribution, habitats, and the adaptation mechanism to the stressing environment. Classification and species identification by DNA sequencing [79][80][81] and the multi-locus analysis [3,21] have overcome the limitations imposed by the morphological classification of organisms with cryptic vegetative features and reduced the capacity for sporulation in vitro. Rapid DNA sequencing and phylogenetic classification provide a framework for future inventories and for linking strains of species that occupy multiple habitats, e.g., natural rocks, manmade objects, plant surfaces, or animals. Although many studies have employed different loci combinations, the combination of two loci, e.g., ITS and nucLSU, have readily distinguished new species within the Trichomeriaceae and Herportichiellaceae families [21,73]. A three-gene analysis based on combined ITS, nucSSU, and mtSSU was conducted for the RIF species position within the order Capnodiales [21]. Three genes, including ITS, nucSSU, and mtSSU, and five loci of a combined nucLSU, nucSSU, mtSSU, RPB1, and RPB2 phylogenetic tree were established and compared in the statement of the phylogenetic placement of RIF within Dothideomyceta [21]. Other multi-gene phylogenetic analyses, such as four loci (nucLSU, nucSSU, mtSSU, and RPB1) and even six loci (ITS, nucLSU, nucSSU, mtSSU, TUB2, and RPB2) were also conducted in the species delimitation of Coniosporium [82] and Rupestriomyces and Spissiomyces [47]. Based on previous studies and our research experiences, a strategy for RIF identification and taxonomy can be proposed. At the outset, when new strains are accumulated, ITS sequences undergo preliminary screening as to determine if the isolates belong to major lineages of RIF or if they represent other environmental fungi that are isolated incidentally from rock surfaces. At a second step, the ITS and nucLSU combined sequences should be applied to assign the fungal isolates to order, family, and genus [11,28]. Finally, a combination of four loci sequences are selected from among the ITS, nucLSU, nucSSU, mtSSU, TUB, RPB1, and/or RPB2, etc. for species delimitation. To calibrate the phylogenetic analyses and validate the segregation of the new species proposed here, the data were analyzed by the poisson tree processes (PTP) server: a Bayesian implementation of the PTP model for species delimitation, and the results well-supported species delimitation and the taxonomic strategy of RIF (Figures S1-S4).
Although RIF can be readily isolated from well-vegetated and humid habitats, the ancestral RIF are hypothesized to originate during periods of dry climates in the late Devonian and middle-Triassic, when the paleoclimate subjected fungi to harsh environment stresses and were selected for colonizing rock surfaces in nutrient-depleted habitats [25]. Meristematic growth ensures an optimal volume/surface ratio and minimizes environmental exposure [83]. Highly melanized cell walls and the biosynthesis of mycosporines and mycosporine-like amino acids protect RIF against UV radiation [84,85]. Endospores and reproduction by vegetative fragmentation reduce resource investments in reproduction [25]. Loss of a sexual stage can simplify life cycles and enable dispersal during brief periods of optimal environmental conditions [2]. RIF produce a minimal array of metabolites, only those for essential survival-for example, extracellular polymeric substances [79], and, in general, they are poor antibiotic producers [86]. Presently, the origin, evolution, and mechanisms of stress adaptation of RIF are beginning to be understood comprehensively by a combination of genome, transcriptome, and proteome studies [26,87,88]. Although transcriptomic and metabolic analyses may explain details about their differential gene expressions and metabolic activity variations under stress, in order to reveal the survival strategies of RIF, further interesting researches are needed-for instance, how rapidly the transcriptional regulation initiated responding to fluctuating hydration conditions and how to resume metabolic activities after long periods of metabolic suspense undergoing extremes [88].
Supplementary Materials: The following are available online at http://www.mdpi.com/2309-608X/6/4/187/s1. Table S1: GenBank accession numbers of taxa used in phylogenetic analyses. Figure S1: Phylogenetic tree generated by PTP server analysis using the combined ITS and LSU sequences to calibrate the generic positions of RIF affiliated in Chaetothyriales. Figure S2: Phylogenetic tree generated by PTP server using combined sequences of ITS, nucLSU, mtSSU, RPB1, and TUB loci to validate the new species in the family Trichomeriaceae. Figure S3: Phylogenetic tree generated by PTP server using combined sequences of ITS, SSU, TEF, TUB and ACT loci to validate the new species in Exophiala. Figure S4: Phylogenetic tree generated by PTP server analysis using combined sequences of ITS, SSU, TEF, TUB and nucLSU to validate the new species in Cladophialophora.