ITS rDNA Gene Analysis Versus MALDI-TOF MS For Identification of Neoscytalidium dimidiatum Isolated from Onychomycosis and Dermatomycosis Cases in Medellin (Colombia)

Within the Neoscytalidium genus, N. dimidiatum, N. oculus, N. orchidacearum, and N. novaehollandiae have been recognized. Although these species are frequently found in soil, N. dimidiatum has been identified as an etiologic agent of onychomycosis or dermatomycosis, and N. oculus has been identified as an etiologic agent of an ocular lesion. All these species can be cultured in vitro, but their morphological identification by macroscopic and microscopic traits is difficult and imprecise due to their similarity. In this study, 34 isolates of Neoscytalidium spp. from 32 onychomycosis and two dermatomycosis cases in Medellin (Colombia) were identified at the species level using sequencing of the ITS1+5.8S+ITS2 nuclear rDNA region and MALDI-TOF mass spectrometry (MS). Neoscytalidium dimidiatum strain MUM 17.21 was used to construct the reference spectrum in the in-house library to identify the clinical isolates by MALDI-TOF MS. Additionally, N. dimidiatum PPC-216 and PLAB-055 strains were used to validate the in-house constructed reference spectra. Although four groups were observed in the dendrogram obtained from the proteins of each isolate profile, MALDI-TOF MS and sequencing results are in accordance, since all isolates were identified as N. dimidiatum.


Introduction
The coelomycete Neoscytalidium (Botryosphaeriaceae) is a dematiaceous fungal genus that is commonly found in the soil of tropical and subtropical zones [1]. The taxonomy of this genus has been problematic and constantly revised. Currently, within this genus, four species are recognized: N. novaehollandiae, N. orchidacearum, N. oculus, and N. dimidiatum [2,3]. The first two species have been reported as phytopathogens, while N. oculus and N. dimidiatum have been associated to ocular or keratinized tissue (skin or nails) infections, which are indistinguishable from dermatophytosis [3][4][5].
Recently, it has been demonstrated that N. oculus can form biofilms and cause hemolysis [3]. In contrast, N. dimidiatum can also cause subcutaneous and deep infections, mainly in immunocompromised patients [6][7][8]. Furthermore, it usually shows low antifungal in vitro susceptibility with an unpredictable clinical response to treatments [9,10]. Isolates were grown in solid potato dextrose agar (PDA; Oxoid, Basingstoke, Hampshire, England) at 28 • C for 5 days. Then, approximately 0.2 cm 2 of the biomass was transferred to a 1.5-mL microtube (KIMA, Piove di Sacco, Padova, Italy) containing 1 mL of yeast malt broth (YMB; yeast extract, 3 g/L; malt extract, 3 g/L; peptone 5 g/L; glucose 10 g/L). The isolates were incubated for 10 days in constant agitation at room temperature. Subsequently, they were centrifuged for 10 min at 14,000 g, the supernatant was discarded, and 1 mL of sterile distilled water was added to remove residues from the culture medium. It was centrifuged again, and the supernatant was removed, allowing the pellet to dry and later be kept at −20 • C for DNA extraction.

DNA Extraction
DNA extraction was performed following the protocol established by Rodrigues et al. [15], with minor modifications: the biomass was transferred into Lysing matrix MP Biomedicals tubes (Santa Ana, CA, USA) containing 500 µL of lysis buffer (200 mM of Tris-HCl pH 8.5; 250 mL of NaCl; 25 mM of EDTA; 0.5% [w/v] SDS) and 300 mg of sterile glass beads of 1.25 to 1.65 mm in diameter (Sigma, St. Louis, MO, USA). Then, samples were mechanically lysed for 40 s in a FastPrep-24 TM 5G Instrument (MP Biomedicals Santa Ana, CA, USA); subsequently, they were incubated in a water bath for 30 min at 65 • C, and centrifuged for 10 min at 14,000× g. With the purpose of precipitating the polysaccharides and proteins, 500 µL of cold 3 M NaOAc pH 5.5 were added to each sample, and gently mixed by inversion before incubation at −20 • C for 10 min, and centrifugation at 14,000× g for 10 min. Nucleic acids were precipitated from a mix of 500 µL of the clear supernatant with 500 µL of isopropanol, which was incubated for 90 min, to finally be centrifuged at 14,000× g for 10 min; the precipitate was washed twice with 500 µL of cold 70% ethanol, centrifuged at 6000× g for 7 min, and air dried. DNA was resuspended in 50 µL of ultra-pure sterile water, quantified in a NanoDrop™ 1000 (Thermo Scientific™, Waltham, MA, USA), and stored at −20 • C.

PCR amplification
PCR amplification of the ITS1+5.8S+ITS2 rDNA region was performed with 50 µL of a reaction mixture containing 25 µL of NZYTaq II 2x Green Master Mix (NZTtech, Lisbon, Portugal), 1 µL of primers ITS1 (5 -TCCGTAGGTGAACCTGCGG-3 ) and ITS4 (5 -TCCTCCGCTTATTGATATGC-3 ) [16], 1 µL of DNA, and 22 µL of sterile ultra-pure water. PCR conditions were as follows: a denaturation step at 94 • C for 3 min; 35 cycles of the annealing step: 1 min at 94 • C, 1 min at 55 • C, and 1 min at 72 • C; and a final elongation step of 5 min at 72 • C. PCR amplifications were conducted in a GeneAmp PCR System 9700 thermocycler (Applied Biosystems, Foster City, CA, USA). The products were analyzed in 1% agarose gel and purified using the NZYgelpure kit from NZYtech. Sequencing of the products was carried out in the STAB VIDA Lda (Caparica, Portugal) using the Sanger/capillary method. Sequences were processed using the FinchTV program, version 1.4.0. Poor-quality end regions were removed. ITS sequences were compared with the National Center for Biotechnology Information (NCBI) database using the Basic Local Alignment Search Tool Nucleotide (BLASTN) to detect non-specific amplicons. Only sequences with 60% coverage and at least 98% identity with Neoscytalidium spp. sequences were retained for further analysis.

DNA Sequence Processing and Molecular Taxonomic Analysis
All the sequences were oriented according to the direction of the reference sequences and then globally aligned using the nested ClustalW [17] software in MEGA version 7 [18]. The best model was estimated considering the Bayesian information criterion (BIC) and the Akaike information criterion (AIC). The recovered alignments were checked manually to avoid mismatched base pairs. Reconstruction of the maximum likelihood (ML) topology with ITS1 and ITS2 sequences was performed using the Tamura-3-parameter model of nucleotide substitution, with Gamma distribution and 1000 bootstraps. Other sequences extracted from GenBank were included in the analysis, and Scytalidium lignicola was used as an outgroup (Table 1).

Protein Extraction
Protein extraction was performed following the protocol of Packeu et al. [19], with minor modifications: the fungi were cultured in Sabouraud chloramphenicol agar (Biomerieux ® , Paris, France) and incubated for 72 h at 28 • C. Mycelia were gently removed with wood sticks and suspended in a 3:1 mixture of ethanol and distilled water in a microtube of 1.5 mL (KIMA,), vortexed and centrifuged for 5 min at 13,000 rpm. The supernatants were discarded, and 50 µL of 70% formic acid was added; after being homogenized, they were left to rest for 15 min. Subsequently, 50 µL of 100% acetonitrile was added, and again, samples were allowed to rest for 15 min. Finally, they were centrifuged at 13,000 rpm for 2 min, and the supernatant was used for fungal identification.

Constructing the Reference Mass Spectra
The reference spectra were constructed from the independent reference strain of N. dimidiatum MUM 17.21, obtained from the public service fungal culture collection Micoteca da Universidade do Minho (MUM, Braga, Portugal), according to the methodology proposed by Cassagne et al. [20]. Four independent cultures of strain MUM 17.21 were incubated at 28 • C for 72 h. Proteins were extracted from each culture (as described above), and 1 µL of each supernatant was transferred to a spot onto the steel target plate (Bruker Daltonics, Bremen, Germany), allowing the samples to dry completely. Then, 1 µL of α-cyano-4-hydroxycinnamic acid matrix (Bruker Daltonics, Bremen, Germany) was added. Eight replicas for each culture were used. The spectra were acquired after at least 240 shots in linear mode using the Microflex LT table mass spectrometer (Bruker Daltonics, Bremen, Germany) in the positive ion mode, with a 337-nm nitrogen laser. A bacterial standard of Escherichia coli extract with RNase and human myoglobin was used as control and for the internal verification of equipment calibration (BTS Bruker Daltonics, Bremen, Germany). Data were acquired automatically using the AutoXecute function of FlexControl software, version 3.3, and then exported to the MALDI Biotyper software, version 3.1 (Bruker Daltonics, Bremen, Germany), to evaluate the quality of the spectrum according to the manufacturer's recommendations. Two isolates (PPC-216 and PLAB-055), previously identified by sequencing as N. dimidiatum, were used to validate the reference spectrum. Fusarium oxysporum ATCC 48112 and Aspergillus fumigatus ATCC 204305 strains were used as negative controls.

Identification of Clinical Isolates
The protein extraction of clinical isolates was performed as described above. For each isolate, four spectra were obtained and compared to the library constructed in-house using the Maldi Biotyper 3.1 software. Identification was carried out following the manufacturer's established scores: ≥2.000: species-level identification; 1.700-1.999: genus-level identification; ≤1.699: not reliably identified.

Results
Clinical isolates were presumptively identified as N. dimidiatum through classical phenotype methods and also taking into consideration that they were isolated from human patients with onychomycosis or dermatomycosis. In the beginning, all clinical isolates showed white colonies that gradually turned grey, olive green, or black. Under the microscope, mycelia that were branched, septate, hyaline, or dematiaceous with arthroconidia in chains or disarticulated, as well as occasional chlamydoconidia, were observed. In addition, molecular identification was carried out through amplification, sequencing, and phylogenetic analysis using the ITS1+5.8S+ITS2 rDNA region. The sequences of isolates included in this study were deposited in GenBank, and the accession numbers are indicated ( Table 1).
The alignment of the non-coding region ITS1+5.8SADNr+ITS2 was carried out with 40 sequences, 34 of them corresponding to clinical isolates from this study, one sequence of the ex-holotype strain of N. dimidiatum, the strain MUM 17.21, and three of the other species of the genus Neoscytalidium, and one of S. lignicola as outgroup ( Table 1). Sequences of 407 nucleotides were used for the alignment, and they were distributed as follows: 307 conserved (75.43%); 94 variables (23.10%), of which 87 were singleton (21.38% of the total nucleotides); and six were parsimony-informative (1.47%). All positions contained gaps, and missing data were eliminated from the dataset. All the clinical isolates from this study were grouped together with the type sequence of N. dimidiatum in the same branch, with a bootstrap support of 87% ( Figure 1). The presumptive phenotype identifications were totally confirmed by the ITS barcode.
For the identification of clinical isolation by MALDI-TOF MS, four reference spectra were constructed. The range of the masses detected was between 3070.643-16,682.647 Daltons (Figure 2). The 34 (100%) clinical isolates were identified through the in-house library as N. dimidiatum (score ≥2.00). The dendrogram created with the protein spectra, using the principal component analysis (PCA) method, shows the conformation of four groups (I to IV) above the critical distance (850), as shown in Figure 3. Groups I to III were formed by seven, five, and 21 isolates, respectively. Group IV was formed by the single isolate Plab-077 (MUM 19.71) with an origin in a nail infection.     Traditionally, Neoscytalidium spp. identification has been based only on macroscopic and microscopic observations [23]. Classical phenotypic methods can give inaccurate results in the identification of the species of this genus, because they do not have characteristic structures to distinguish between them [11]. Recently, using sequencing of the ITS1+5.8S+ITS2 rDNA region, or of a fragment of the 28S large subunit ribosomal (LSU), it has been possible to recognize the species N. novaehollandiae, N. orchidaceous, N. dimidiatum, and N. oculus [3,11]. In the present study, the high level of accuracy of using the ITS1+5.8S+ITS2 rDNA region to identify N. dimidiatum reveals that this barcode is fit-forpropose in the clinical field. However, this barcode alone could be unable to clearly discriminate species that exhibited a certain degree of genetic variations. In this latter case, we recommended the use of a multilocus sequence analysis (MSLA). A similar situation was obtained by Pereira et al. [24] for the dermatophyte Trichophyton rubrum, which due to its adaptation to a highly specialized ecological niche, the skin of the human host, combined with exclusively asexual reproduction, can explain a high level of genetic uniformity within the species.
Neoscytalidium dimidiatum has been recognized as an etiologic agent for nail and skin infections [25,26], possibly because it produces keratinases [27]. However, with the increase of patients with immunosuppressive conditions, it has been demonstrated that N. dimidiatum does not only affect external tissues, but also causes eye infections or invades organs such as the lungs or the brain [6][7][8]10].
Some studies in Colombia have reported the isolation of N. dimidiatum or S. dimidiatum, as it has also been named, from skin and nail samples. In general, the identification of this fungus has been based on macroscopic and microscopic observations [5,[28][29][30], with exception of the data published by Dionne et al. [7] in a case of pulmonary infection and Bueno et al. [28], who also used ITS1+5.8S+ITS2 nuclear rDNA to identify Scytalidium hyalinum (considered by [8] as an albino variant of N. dimidiatum while others [31] regard them as separate species) and F. dimidiatum as an etiologic agent of onychomycosis. However, the present study is the first one where Neoscytalidium spp. clinical isolates were identified by means of phylogenetic analysis and of MALDI-TOF MS. Phylogenetic analysis showed that all the clinical isolates were N. dimidiatum and grouped in one clade, separated
Traditionally, Neoscytalidium spp. identification has been based only on macroscopic and microscopic observations [23]. Classical phenotypic methods can give inaccurate results in the identification of the species of this genus, because they do not have characteristic structures to distinguish between them [11].
Recently, using sequencing of the ITS1+5.8S+ITS2 rDNA region, or of a fragment of the 28S large subunit ribosomal (LSU), it has been possible to recognize the species N. novaehollandiae, N. orchidaceous, N. dimidiatum, and N. oculus [3,11]. In the present study, the high level of accuracy of using the ITS1+5.8S+ITS2 rDNA region to identify N. dimidiatum reveals that this barcode is fit-for-propose in the clinical field. However, this barcode alone could be unable to clearly discriminate species that exhibited a certain degree of genetic variations. In this latter case, we recommended the use of a multilocus sequence analysis (MSLA). A similar situation was obtained by Pereira et al. [24] for the dermatophyte Trichophyton rubrum, which due to its adaptation to a highly specialized ecological niche, the skin of the human host, combined with exclusively asexual reproduction, can explain a high level of genetic uniformity within the species.
Neoscytalidium dimidiatum has been recognized as an etiologic agent for nail and skin infections [25,26], possibly because it produces keratinases [27]. However, with the increase of patients with immunosuppressive conditions, it has been demonstrated that N. dimidiatum does not only affect external tissues, but also causes eye infections or invades organs such as the lungs or the brain [6][7][8]10].
Some studies in Colombia have reported the isolation of N. dimidiatum or S. dimidiatum, as it has also been named, from skin and nail samples. In general, the identification of this fungus has been based on macroscopic and microscopic observations [5,[28][29][30], with exception of the data published by Dionne et al. [7] in a case of pulmonary infection and Bueno et al. [28], who also used ITS1+5.8S+ITS2 nuclear rDNA to identify Scytalidium hyalinum (considered by [8] as an albino variant of N. dimidiatum while others [31] regard them as separate species) and F. dimidiatum as an etiologic agent of onychomycosis. However, the present study is the first one where Neoscytalidium spp. clinical isolates were identified by means of phylogenetic analysis and of MALDI-TOF MS. Phylogenetic analysis showed that all the clinical isolates were N. dimidiatum and grouped in one clade, separated from the other species within the same genus. In addition, it was observed that N. oculus is the closest related species to N. dimidiatum.
The construction of reference spectra was necessary because the commercial library lacks spectra for the identification of Neoscytalidium spp. Although the identification of filamentous fungi using MALDI-TOF MS has limitations, and even more when they are pigmented, the present study achieved both the successful construction of reference spectra and the identification of 34 clinical isolates of N. dimidiatum. The use of young mycelium without pigmentation (72 h of growth) might explain the correct identification by MALDI-TOF MS of all N. dimidiatum clinical isolates in this study. Our results are in contrast to those obtained by Alshawa et al. [32], who only achieved identification of 77.8% of the isolates. In this latter case, the low identification rate may be explained by the incubation time (three weeks), during which pigmentation was developed more intensively in combination with possible cell wall changes, such as becoming thicker and coated by hydrophobins proteins. It is known that the presence of pigments such as melanin can interfere in the identification of some fungi through MALDI-TOF MS, since the desorption/ionization process may be inhibited [33]. Also, in the identification of Neoscytalidium spp. by MALDI-TOF MS carried out by Alshawa et al. [32], the reference spectra of N. dimidiatum had seven mass peaks with an intensity >0.25 × 10 4 u.m.a., while the ones obtained in the present study showed 34 mass peaks with an intensity >1 × 10 4 u.m.a. In this case, both the use of young mycelia without pigments and the protein extraction method are the determining factors for adequate identification. According to the critical distance level of 850 utilized by Chen et al. (2014) [34], the results observed in the dendrogram of Figure 3 suggest the presence of four phenotypes (Group I to IV) which shows the potential of MALDI-TOF MS to group the isolates below the species level [34]. This result contrasts that which has been observed in the consensus phylogenetic tree ( Figure 2) where a single homogeneous group is observed. Although ITS1+5.8S+ITS2 rDNA is considered the universal barcoding for fungal identification, it is known that it does not have enough resolution power to detect intraspecific variability as is observed with other housekeeping gene markers, such as the elongation factor, calmodulin, or β-tubulin [12,21,35]. However, for Neoscytalidium spp., sequences of those genes are not available in databases for public use, as is the case of GenBank.

Conclusions
In conclusion, the results obtained with mass analysis through dendrogram construction highlights that MALDI-TOF MS can be useful in the accurate identification of strains within N. dimidiatum species, and is a cost-effective technique with the possibility of being used in a mycological diagnosis laboratory.