Chaetomium and Chaetomium-like Species from European Indoor Environments Include Dichotomopilus finlandicus sp. nov.

The genus Chaetomium is a frequently occurring fungal taxon world-wide. Chaetomium and Chaetomium-like species occur in indoor environments, where they can degrade cellulose-based building materials, thereby causing structural damage. Furthermore, several species of this genus may also cause adverse effects on human health. The aims of this research were to identify Chaetomium and Chaetomium-like strains isolated from indoor environments in Hungary and Finland, two geographically distant regions of Europe with drier and wetter continental climates, respectively, and to study their morphological and physiological properties, as well as their extracellular enzyme activities, thereby comparing the Chaetomium and Chaetomium-like species isolated from these two different regions of Europe and their properties. Chaetomium and Chaetomium-like strains were isolated from flats and offices in Hungary, as well as from schools, flats, and offices in Finland. Fragments of the translation elongation factor 1α (tef1α), the second largest subunit of RNA polymerase II (rpb2) and β-tubulin (tub2) genes, as well as the internal transcribed spacer (ITS) region of the ribosomal RNA gene cluster were sequenced, and phylogenetic analysis of the sequences performed. Morphological examinations were performed by stereomicroscopy and scanning electron microscopy. Thirty-one Chaetomium sp. strains (15 from Hungary and 16 from Finland) were examined during the study. The most abundant species was Ch. globosum in both countries. In Hungary, 13 strains were identified as Ch. globosum, 1 as Ch. cochliodes, and 1 as Ch. interruptum. In Finland, 10 strains were Ch. globosum, 2 strains were Ch. cochliodes, 2 were Ch. rectangulare, and 2 isolates (SZMC 26527, SZMC 26529) proved to be representatives of a yet undescribed phylogenetic species from the closely related genus Dichotomopilus, which we formally describe here as the new species Dichotomopilus finlandicus. Growth of the isolates was examined at different temperatures (4, 15, 20, 25, 30, 37, 35, 40, and 45 °C), while their extracellular enzyme production was determined spectrophotometrically.


Introduction
Chaetomium Kunze (Ascomycota, Sordariales) is the largest genus of the family Chaetomiaceae, present in various substrates and geographical regions [1]. More than 400 Chaetomium species have been described. The type species is Ch. globosum Kunze [2].

Morphological Features of the Isolated Chaetomium Strains
Among the previously described Chaetomium species, Ch. cochliodes colonies grew rapidly on MEA, OA, and PDA ( Figure 2A) reaching 65-70 mm in diameter after 7 days at 25 • C. Hyphae were light beige on MEA, while brownish on OA and PDA, with powdery surface, undulate colony edges and without colored exudates. The strains were unable to produce ascospore-containing ascomata on MEA, while strong dark green ascospore formation was observed after 7 days on OA and PDA. Ch. interruptum ( Figure 2B) formed white mycelium on all media, brownish exudates diffusing into the media, and did not produce spores during 7 days of culturing at 25 • C. On MEA and OA, it formed regular circular colonies, while on PDA the edges of the colonies grew irregularly. Colony diameters after 7 days were 40-45, 50-60, and 30-40 mm on MEA, OA, and PDA, respectively. Ch. globosum ( Figure 2C) colonies overgrew both MEA and OA media in 7 days at 25 • C. On PDA the strains grew slowly, with colony diameters of 30-40 mm after 7 days and a lobate edge. No ascospores were produced on MEA medium, but greenish ascospores were produced on OA and PDA. Colonies ranged from beige (MEA, OA) to brown (PDA) in color, the surface texture was floccose or velvety, and a brownish exudate was produced on all media. Ch. rectangulare ( Figure 2D) completely overgrew all media in 7 days at 25 • C with white, cottony mycelium and without colored exudates. No ascospores were produced under any of the conditions tested. Ascomata of Ch. cochliodes (Figure 3(A1-A5)) were ostiolate, ovoid, greenish olivaceous, with brown wall, textura intricata. Terminal hairs were usually around the ostiolum, light brown or brown, spirally coiled, lateral hairs undulate or loosely coiled, tapering towards the tip. Mature ascospores were brown, limoniform, usually biapiculate at both Ascomata of Ch. cochliodes (Figure 3(A1-A5)) were ostiolate, ovoid, greenish olivaceous, with brown wall, textura intricata. Terminal hairs were usually around the ostiolum, light brown or brown, spirally coiled, lateral hairs undulate or loosely coiled, tapering towards the tip. Mature ascospores were brown, limoniform, usually biapiculate at both ends, bilaterally flattened. Ascomata of Ch. globosum (Figure 3(B1-B5)) were ostiolate, greenish olivaceous, with brown wall, textura intricata. Terminal hairs were light brown or brown, undulate to loosely coiled, lateral hairs brown, flexuous, tapering towards the tips. Mature ascospores were greenish or brown, subglobose or limoniform, bilaterally flattened. Ascomata of Ch. interruptum (Figure 3(C1-C5)) were ostiolate, brown, with brown wall, textura epidermoidea (tissue of closely interwoven irregularly disposed hyphae without interhyphal spaces, the walls united, usually forming a membranous or epidermislike tissue). Terminal hairs were brown undulate, lateral hairs brown, flexuous, tapering towards the tips. Mature ascospores were greenish or brown, subglobose, or limoniform, bilaterally flattened. brown wall, textura epidermoidea (tissue of closely interwoven irregularly disposed hyphae without interhyphal spaces, the walls united, usually forming a membranous or epidermis-like tissue). Terminal hairs were brown undulate, lateral hairs brown, flexuous, tapering towards the tips. Mature ascospores were greenish or brown, subglobose, or limoniform, bilaterally flattened.

Phylogeny and Taxonomy
The tef1α, ITS, rpb2, and tub2 dataset consisted of 935, 639, 525, and 571 characters, respectively. The indel-based binary dataset was 100 characters long. Isolates SZMC 26527 and SZMC 26529 resolved as members of a new species with high confidence values on the phylograms obtained from both tef1α ( Figure 1) and the other three loci (data not shown). For the final inference the four loci were concatenated and partitioned. Based on the maximum likelihood phylogenetic tree inferred from the concatenated sequences (Figure 4), isolates SZMC 26527 and SZMC 26529 formed a well-supported distinct branch inside the genus Dichotomopilus with the closest relatives being D. funicola, D. pseudofunicola, D. subfunicola, and D. variostiolatus. This new species is described below as Dichotomopilus finlandicus sp. nov.

Phylogeny and Taxonomy
The tef1α, ITS, rpb2, and tub2 dataset consisted of 935, 639, 525, and 571 characters, respectively. The indel-based binary dataset was 100 characters long. Isolates SZMC 26527 and SZMC 26529 resolved as members of a new species with high confidence values on the phylograms obtained from both tef1α ( Figure 1) and the other three loci (data not shown). For the final inference the four loci were concatenated and partitioned. Based on the maximum likelihood phylogenetic tree inferred from the concatenated sequences ( Figure 4), isolates SZMC 26527 and SZMC 26529 formed a well-supported distinct branch inside the genus Dichotomopilus with the closest relatives being D. funicola, D. pseudofunicola, D. subfunicola, and D. variostiolatus. This new species is described below as Dichotomopilus finlandicus sp. nov.

Discussion
The dominant species in this study was Ch. globosum in both countries in indoor environments, as also determined in several previous studies [5,7,62]. In both countries, the species Ch. cochliodes was found to be also common in indoor environments. Ch. interruptum was isolated only from Hungary, while Ch. rectangulare and D. finlandicus only from Finland. Due to the tendency of application of cellulose-based materials (e.g., wallpapers and drywalls) in modern buildings, cellulose-degrading fungi, such as Chaetomiaceae have an increasing relevance. Most indoor strains were isolated from house dust or surface samples, while isolates from air samples were relatively rare. Similar observations were made by Fogle et al. [63] based on the analysis of samplings performed in 794 buildings in Dallas. Although several theories have emerged to explain this phenomenon, further experiments are needed to clarify the dispersal strategy of these fungi indoors.
In a previous study, Salo et al. [5] tested 42 toxin-producing Chaetomium isolates from Finland. In addition to the most common Ch. globosum, three other species, Ch. cochliodes, Ch. rectangulare, and a Chaetomium-like species were described for the first time from Finnish buildings. In a study by Vornanen-Winquist et al. [61], unknown indoor Chaetomium-like strains were designated as Dichotomophilus sp. The molecular results presented here revealed that the Chaetomium-like isolate Ch1/tu (SZMC 26529) in Salo et al. [5] and the Dichotomopilus sp. isolate C5/LM (SZMC 26527) from Vornanen-Winquist et al. [61] belong to the same new, previously undescribed species of the genus Dichotomopilus. Strain Ch1/tu was isolated from an inlet air filter and suggested to originate from the outdoor air [5], while strain C5/LM was isolated from an exhaust air filter. This may indicate that C5/LM had a possible indoor source. On the other hand, the fact that this new species was detected in both inlet and outlet air filters may also suggest that the strains were already incorporated into the filter material during production. Contamination of gypsum wall board with Chaetomium strains during production has been described by Andersen et al. [64].
The species D. finlandicus described in the recent study could be morphologically and molecularly differentiated from related species, the results of the phylogenetic analyses of the combined dataset of ITS, tef1α, rpb2, and tub2 (Figure 4) was 100% bootstrap support. In addition, the phylogenetically closest relative species D. funicola, D. pseudofunicola, D. subfunicola, D. variostiolatus, and D. indicus are morphologically different from the strain we studied. Based on the morphological properties of these species studied by Wang et al. [7], ascomata, terminal hairs, and the asci were different while the shape and the size of ascospores were similar to D. finlandicus, which we describe here as a new species.
The enzymatic activity of the Chaetomium and Chaetomium-like strains proved to be diverse, and no correlation was found with either the isolation site or the growing substrate. These results are consistent with the findings of Abdel Azeem et al. [22], that enzyme production is isolate-dependent. The authors concluded that enzyme production has no detectable association with ecology, however, although this may be true in the case of plant host specificity, we suggest the ability to produce cellulolytic enzymes as a clear ecological advantage in the case of fungal growth on cellulose-based building materials.
In the rapid screening assays described by Salo et al. [5] and Vornanen-Winquist et al. [65], Dichotomopilus strains gave weaker responses than the Ch. globosum, Ch. cochliodes and Ch. rectangulare strains. However, strain Ch1/tu (SZMC 26529), which was designated here as the type strain of the newly described species D. finlandicus, inhibited boar sperm motility after 3 d of exposure, indicating that the strain produced a bioactive agent possibly affecting mitochondrial functions, or ion homeostasis [66]. Purification and identification of this substance and characterization of its biological activities will be the subject of further research.

Sample Collection and Isolation
Chaetomium and Chaetomium-like strains were collected and isolated from schools, flats, and offices in Finland as described previously by Salo et al. [5], as well as from houses, flats, and offices in Hungary (Table 1). To collect fungi from walls, visible colonies, or wet surfaces detected by moisture meter (Greisinger GMI 15) were sampled with sterile swabs. House dust samples were also collected with swabs. Samples were spread directly onto malt extract agar (MEA) supplemented with 2% chloramphenicol, Dichloran -Rose Bengal Agar, or Casitone Agar on site. To collect airborne fungi, air samples of 100 L were collected at 150 cm a.g.l. with 400-hole one-stage Andersen samplers [67] (MAS 100, EMD Millipore, Merck, Darmstadt, Germany; SAS IAQ, International PBI SpA, Milan, Italy; Samp l'Air MK2, AES Chemunex, Bruz, France), at a flow of 100 L/min onto MEA. Between samplings, the devices were sterilized with ethanol (abs.). Incubation of the samples was performed for 5 to 7 days at room temperature. The isolated pure cultures were deposited in the Szeged Microbiology Collection (SZMC, http://szmc.hu), Szeged, Hungary.

DNA Extraction, Identification, and Phylogenetic Analysis
Pure cultures of fungi were grown in 2% (w/v) MEA for 7 days at room temperature. Fungal genomic DNA was then extracted using the E.Z.N.A.®Fungal DNA Mini Kit (Omega Biotek, Norcross, GA, USA). The extracted genomic DNA was amplified by PCR with the primers listed in Table 2. The PCR mixture (20 µL) contained 2 µL 10× DreamTaq Buffer with 20 mM MgCl 2 , 2 µL of 2 mM dNTP mix, 4 µL of each primer (100 µM), 7 µL bidistilled water, 0.1 µL of 5 U/µL DreamTaq DNA Polymerase (Thermo Fischer Scientific, Vilnius, Lithuania) and 1 µL genomic DNA. Amplifications were performed in a Doppio Gradient 2 × 48-well thermal cycler (VWR International, Debrecen, Hungary) according to the amplification cycles shown in Table 2. PCR products were purified using NucleoSpin™ Gel and PCR Clean-up Kit (Macherey-Nagel, Düren, Germany). Sequencing was performed on the sequencing platform of Eurofins Genomics (http://www.eurofinsgenomics.com, accessed on 2 September 2021). The resulting sequences were submitted to the GenBank Nucleotide database (ncbi.nlm.nih.gov) under the accession numbers listed in Table 1. In addition to the sequences generated in this study, sequences of reference strains were obtained from the GenBank Nucleotide database ( Table 1).
Sequences of the two Dichotomopilus isolates were aligned with publicly available sequences of 12 and 11 previously described Dichotomopilus and Chaetomium species, respectively. Phylogenetic analyses were conducted using four loci (ITS, tef1α, rpb2, and tub2).
Sequences were aligned with Prank v170427 [69]. Alignments of the four loci were concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single partitions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset was divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contained relative high number of indels, therefore gaps were coded as absence/presence characters by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matrices were concatenated and added as a single partition to the dataset. Best fitting model for the phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on the Bayesian information criterion [73], with discrete gamma rate categories. Best fit models for each partition are shown in Table 3. Maximum likelihood analysis was performed using RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 1000 bootstrap replicates.  Sequences were aligned with Prank v170427 [69]. Alignments of the four loci we concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single pa tions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset w divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contain relative high number of indels, therefore gaps were coded as absence/presence charact by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matri were concatenated and added as a single partition to the dataset. Best fitting model for t phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on t Bayesian information criterion [73], with discrete gamma rate categories. Best fit mod for each partition are shown in Table 3. Maximum likelihood analysis was performed u ing RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 10 bootstrap replicates.  Sequences were aligned with Prank v170427 [69]. Alignments of the four loci we concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single par tions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset w divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contain relative high number of indels, therefore gaps were coded as absence/presence characte by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matric were concatenated and added as a single partition to the dataset. Best fitting model for t phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on t Bayesian information criterion [73], with discrete gamma rate categories. Best fit mod for each partition are shown in Table 3. Maximum likelihood analysis was performed u ing RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 10 bootstrap replicates.  Sequences were aligned with Prank v170427 [69]. Alignments of the four loci we concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single par tions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset w divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contain relative high number of indels, therefore gaps were coded as absence/presence characte by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matric were concatenated and added as a single partition to the dataset. Best fitting model for t phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on t Bayesian information criterion [73], with discrete gamma rate categories. Best fit mod for each partition are shown in Table 3. Maximum likelihood analysis was performed u ing RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 10 bootstrap replicates.  Sequences were aligned with Prank v170427 [69]. Alignments of the four loci we concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single par tions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset w divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contain relative high number of indels, therefore gaps were coded as absence/presence characte by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matric were concatenated and added as a single partition to the dataset. Best fitting model for t phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on t Bayesian information criterion [73], with discrete gamma rate categories. Best fit mod for each partition are shown in Table 3. Maximum likelihood analysis was performed u ing RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 10 bootstrap replicates.

cycles
Pathogens 2021, 10, x FOR PEER REVIEW 16 of Sequences were aligned with Prank v170427 [69]. Alignments of the four loci we concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single par tions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset w divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contain relative high number of indels, therefore gaps were coded as absence/presence characte by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matric were concatenated and added as a single partition to the dataset. Best fitting model for t phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on t Bayesian information criterion [73], with discrete gamma rate categories. Best fit mode for each partition are shown in Table 3. Maximum likelihood analysis was performed u ing RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 10 bootstrap replicates.

cycles
Pathogens 2021, 10, x FOR PEER REVIEW 16 of Sequences were aligned with Prank v170427 [69]. Alignments of the four loci we concatenated and partitioned. Tef1α and rpb2 sequences were defined as two single par tions, while the tub2 dataset was partitioned to exons and introns. The ITS dataset w divided to rDNA and ITS1-ITS2 regions. Alignments of tub2 and ITS datasets contain relative high number of indels, therefore gaps were coded as absence/presence characte by 2matrix v1.0 [70] using the simple indel coding algorithm [71]. The two indel matric were concatenated and added as a single partition to the dataset. Best fitting model for t phylogenetic inference was selected by using ModelTest-NG v0.1.4 [72], based on t Bayesian information criterion [73], with discrete gamma rate categories. Best fit mode for each partition are shown in Table 3. Maximum likelihood analysis was performed u ing RAxML-NG v0.9.0 [74]. Statistical support of the best ML tree was obtained with 10 bootstrap replicates.  Table 3. Best-fit models for each partition proposed by ModelTest-NG based on Bayesian information criterion.

Partition
Best-Fit Model

Temperature Profiling
Optimal growth temperature ranges were determined for all Chaetomium and Chaetomiumlike isolates. PDA plates were inoculated with 7 mm agar plates taken from the edge of seven-day-old colonies. The plates were incubated at 4,15,21,25,30,35,37,40, and 45 • C, with six replicates each. Colony diameters were measured after four days.