Caspofungin

In vitro susceptibilities of Neoscytalidium spp. sequence types to antifungal agents and antimicrobial photodynamic treatment with phenothiazinium photosensitizers

ABSTRACT
Neoscytalidium spp. are ascomycetous fungi consisting of pigmented and hyaline varieties both able to cause skin and nail infection. Their color-based identification is inaccurate and may compromise the outcome of the studies with these fungi. The aim of this study was to genotype 32 isolates morphologically identified as N. dimidiatum or N. dimidiatum var. hyalinum by multilocus sequence typing (MLST), differentiate the two varieties by their sequence types, evaluate their susceptibility to six commercial antifungal drugs [amphotericin B (AMB), voriconazole (VOR), terbinafine (TER), 5-flucytosine (5FC), fluconazole (FLU), and caspofungin (CAS)], and also to the antimicrobial photodynamic treatment (APDT) with the phenothiazinium photosensitizers (PS) methylene blue (MB), new methylene blue (NMBN), toluidine blue O (TBO) and the pentacyclic derivative S137. The efficacy of each PS was determined, initially, based on its minimal inhibitory concentration (MIC). Additionally, the APDT effects with each PS on the survival of ungerminated and germinated arthroconidia of both varieties were evaluated. Seven loci of Neoscytalidium spp. were sequenced on MLST revealing eight polymorphic sites and six sequence types (ST). All N. dimidiatum var. hyalinum isolates were clustered in a single ST. AMB, VOR and TER were the most effective antifungal agents against both varieties. The hyaline variety isolates were much less tolerant to the azoles than the isolates of the pigmented variety. APDT with S137 showed the lowest MIC for all the isolates of both varieties. APDT with all the PS killed both ungerminated and germinated arthroconidia of both varieties reducing the survival up to 5 logs. Isolates of the hyaline variety were also less tolerant to APDT. APDT with the four PS also increased the plasma membrane permeability of arthroconidia of both varieties but only NMBN and S137 caused peroxidation of the membrane lipids.

1.Introduction
Neoscytalidium spp. (formerly Scytalidium) are ascomycetous, dematiaceous (producer of melanin or melanin-like pigments), keratinophilic and non-dermatophyte fungi (Hay 2002). The taxonomy of the genera is still very problematic and has been constantly revised (Madrid et al. 2009; Phillips et al. 2013; Valenzuela-Lopez et al. 2017). These fungal species are worldwide distributed in tropical and subtropical regions and are principally known as phytopathogens but some species can infect humans via direct contact with colonized plants (Hay 2002). Neoscytalidium dimidiatum and its albino variant N. dimidiatum var. hyalinum, which does not produce melanin, are involved in human infections (Machouart et al. 2013). Traditionally, the fungus has been characterized by producing abundant uni to tricellular dark arthroconidia (Chowdhary et al. 2014; Crous et al. 2006; Dionne et al. 2015; Miqueleiz- Zapatero et al. 2017) which are nonsexual propagules produced by hypha fission (Turian 1976). In culture, N. dimidiatum shows cottony, fast-growing colonies that become pigmented, usually with a brown to gray color (Chowdwary et al. 2014; Crous et al. 2006). There is a great variety in colony color among the so called pigmented isolates and the color-based phenotypic distinction between the two varieties is not always straightforward. Therefore, it is important to establish a more accurate methodology for the identification of the isolates and distinction between the two varieties. The accurate identification of the two varieties is important to guide clinical managements, to associate particular genotypes with medically important traits such as virulence and antifungal drugs susceptibility and to perform epidemiological studies (Alshawa et al. 2012; Madrid et al. 2009). Unlike N. dimidiatum, which is commonly isolated from plants and soil samples, the albino variety has been isolated exclusively from human infections (Machouart et al. 2004; Madrid et al. 2009; Roeijmans et al. 1997).

The two varieties have also different geographical distributions. N. dimidiatum was isolated from environmental samples and patients in Central Africa, Asia and the Indian Ocean region, whereas the hyaline variety was exclusively isolated from patients in South America, West Africa and West Indies (Machouart et al. 2013; Machouart-Dubach et al. 2002; Morris-Jones et al. 2004). The hyaline variety is responsible for chronic infections in toenail, interdigital spaces, soles and palms (Dunand & Paugam 2008; Elewski 1996; Hay 2002; Khan et al. 2009; Lacroix & Chauvin, 2008; Machouart et al. 2013; Madrid et al. 2009;) which are clinically similar to N. dimidiatum infections and indistinguishable from dermatophytosis (Dunand & Paugam 2008; Hay 2002; Lacroix & Chauvin, 2008; Madrid et al. 2009). The dematiaceus variety is responsible not only for superficial infection as N. dimidiatum var. hyalinum, but also for invasive and deeper infection such as subcutaneous infections (Dhindsa et al. 1998; Sigler et al. 1997), sinusitis (Bakhshizadeh et al. 2014; Dunn et al. 2003), endophtalmitis (Al-Rajhi et al. 1993), cerebral pheohyphomycosis (Mani et al. 2008), pneumonia (Dionne et al. 2015; Elinav et al. 2009), and disseminated infections (Ikram et al. 2009; Tan et al. 2008; Wilinger et al. 2004). As the number of mycosis caused by both varieties is increasing (Chowdhary et al. 2014; Garinet et al. 2015; Miqueleiz- Zapatero et al. 2017) there is a renewed interest in the study of these fungi.

Only a few studies reported the comparison of the antifungal susceptibility of both N. dimidiatum and N. dimidiatum var. hyalinum (Dunand & Paugam 2008; Lacroix & Chauvin, 2008; Madrid et al. 2009; Oyeka & Gugnani 1990). Amphotericin B, voriconazole and terbinafine are single-target antifungals which, respectively, interacts with fungal membrane sterols, inhibits the synthesis of ergosterol (the main fungal sterol), and inhibits the squalene epoxidase. These antifungal have shown in vitro activity against both varieties (Bueno et al. 2010; Carrillo-Munoz et al. 2007; Dunand & Paugam 2008; Lacroix & Chauvin, 2008; Madrid et al. 2009). However, unlike it was observed in infections caused by dermatophytes, the treatment of mycoses caused by Neoscytalidium spp. with conventional antifungal drugs results in poor outcome due to antifungal resistance (Elewski 1996; Hay 2002; James et al. 2017; Machouart et al. 2013). Therefore, due to the limitation of current therapeutic options, the development of new approaches to treat mycosis caused by Neoscytalidium spp. is necessary.
Antimicrobial photodynamic therapy (APDT) has been described as an interesting alternative to conventional fungicides to treat localized mycoses (Dai et al. 2008). The photodynamic process results from the combination of three elements: light of an appropriate wavelength, oxygen, and a photosensitizer (PS), which accumulates preferentially in the target microorganism (Donnelly et al. 2008). As a consequence of the photodynamic process, several reactive oxygen species (ROS) are produced and they subsequently kill the microbial cells without significant damage to the host (Donnelly et al. 2008).

More importantly, given the multiple cellular targets of ROS, the selection of tolerant microorganisms is less probable in APDT than in treatments with conventional antifungal drugs (Dai et al. 2008). Among the PS examined for antifungal use, the phenothiazinium derivatives methylene blue (MB), new methylene blue (NMBN) and toluidine blue O (TBO) are particularly attractive, mainly due to their low toxicities to mammals and other clinical uses (Dai et al. 2008; Rodrigues et al. 2013). Novel derivatives, such as the pentacyclic S137, and derivatives with basic side chains, more effective than the lead compound MB are still being synthesized (Wainwringht et al. 2011; Wainwringht et al. 2016). Previous studies have reported that APDT with phenothiazinium kills different fungal structures, such as microconidia of different species including the dermatophyte Trychophyton spp. (Rodrigues et al. 2012a; Watanabe et al. 2008) and the non dermatophyte Fusarium spp. (de Menezes et al. 2016) and conidia of Exophiala spp. (Gao et al. 2016), Colletotrichum spp. (de Menezes et al. 2014), Metarhizium anisoplia, and Aspergillus nidulans (Gonzales et al. 2010). As far as we know there is no data regarding the effect of APDT with phenothiazinium on arthroconidia of any species. The aims of this study were: (1) to determine if the multilocus sequence typing allows the distinction between the isolates of N. dimidiatum and N. dimidiatum var. hyalinum; (2) to investigate the genotype distribution of Brazilian clinical isolates of the two varieties; (3) to evaluate the in vitro susceptibility of the arthroconidia of the different sequence types to conventional antifungal drugs and to APDT with phenothiazinium photosensitizers, and (4) to compare the susceptibilities of the two varieties to antifungals and APDT.

2.Material and Methods
Análises Clínicas da Faculdade de Ciências Farmacêuticas de Ribeirão Preto (FCFRP), University of São Paulo (USP). Fungi were isolated from skin and toenail infections of patients residing in Ribeirão Preto (São Paulo, Brazil) and neighboring cities from 2004 to 2014 (Table 1). The American Type Culture Collection (ATCC) strains, ATCC®22190™ (previously identified as N. dimidiatum) and ATCC®38907™ (previously identified as N. dimidiatum var. hyalinum) were also included in the study (Table S1). This study was approved by the Research Ethics Committee of the Faculdade de Ciências Farmacêuticas de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, SP, Brazil.Fungal strains were isolated from scrapings of foot and/or toenail placed on both sabouraud dextrose agar (SDA) (Acumedia, Michigan, USA) with chloramphenicol (50 mg L-1) and SDA with chloramphenicol (50 mg L-1) and cycloheximide (400 mg L-1). Plates were incubated at 28ºC for 4 days.The arthroconidia were produced by culturing the fungi on potato dextrose agar with yeast extract (PDAY) at 28ºC for 4 days. Arthroconidia were suspended in sterile water and filtered through sterile glass wool (Sigma- Aldrich). The filtered arthroconidia were pelleted by centrifugation (3000 g) and resuspended in PBS, pH 7.4. The concentrations of arthroconidia were determined with hemocytometer.Fungal mycelia were obtained by culturing 1 × 107 arthroconidia in 50 mL of potato dextrose broth (PDB) at 28ºC for 24 h and 180 rpm (Ecotron, InforsHT, Switzerland). Subsequently, fungal mycelia were filtered through Miracloth (Calbiochem, La Jolla, CA) and frozen in liquid nitrogen.The genomic DNA (gDNA) was extracted from fungal mycelia grounded under liquid nitrogen using a mortar and pestle.

The gDNA was extracted with extraction buffer (200 mM Tris-HCl pH 8.5, 250 mM NaCl, 25 mM EDTA, 0.5% w/v SDS) and phenol-chloroform (1:1). After centrifugation, the gDNA was precipitated with isopropanol and resuspended with RNase-free water (Promega, Madison, WI).Multilocus sequence typing (MLST) was performed by PCR, DNA sequencing and nucleotide sequence analysis. All PCR reactions were performed with Phusion® High Fidelity DNA Polymerase (Thermo Fisher scientific Inc., Waltham, MA, USA) according to the manufacturer’s instructions. The following seven loci were chosen for MLST: the internal transcribed spacer regions 1 (ITS1) and 2 (ITS2) of the 28S rRNA; the chitin synthase (CHS); the histone H3 (HH3); the glyceraldehyde-3-phosphate dehydrogenase (GPD); the β-tubulin (TUB), and the elongation factor 1α (EF1). The primers for ITS1 and ITS2 amplification were designed according to White et al. 1990, and for the amplifications of the other sequences the following primers were designed based on published sequences in GenBank (Benson et al. 2013): CHS (CHS_f 5’-GTGTGCGTTGTCAGCGAC-3’, CHS_r 5’-TCCTTCAAACAGAAAAGCATC-3’), TUB (TUB_f 5’-CAGCTGAACTCTGACCTGC- 3’, TUB_r 5’- AGCAGTGAACTGGTCACC-3’), HH3 (HH3_f 5’-CGCTGCGTCGCTGTTGC-3’,HH3_r 5’-CTACAATACCGTTAGTAATGC-3’), GPD (GPD_f 5’- AGCTGACCCACGCCACCA-3’, GPD_r 5’-ACCGTCGACGGTCTTCTG-3’), EF-1 (EF1_f 5’-TCGACAAGCGTACCATCGAG-3’, EF1_r 5’- AGCCTTGAGCTTGTCAAGGA-3’).DNA sequences were determined with the ABI3730 DNA Analyzer (Applied Biosystems, Foster City, CA), analyzed with ChromasPro® software (ChromasPro 1.7.6, Technelysium Pty Ltd, Tewantin QLD, Australia) and compared with the DNA sequences database of the Basic Local Alignment Search Tool (BLAST) (Altschul et al. 1190).

The MLST analysis was performed by the multiple DNA alignment method using the ‘Molecular Evolutionary Genetics Analysis – MEGA’ 6.0 software (Tamura et al. 2013). The GenBank accession numbers of each sequence are listed in supplementary Table S1. The combination of alleles for the different loci defined the sequence type (ST) of each clinical isolate. The phylogenetic analysis was conducted by inferring the evolutionary history using the unweighted-pair group method with arithmetic averages and arithmetic mean (UPGMA) (Sneath & Sokal 1973). The robustness of the dendrogram was evaluated from 1000 bootstrap replications. The evolutionary distances were computed using the p-distance method (Bain et al. 2007; Masatoshi & Kumar 2000). The analyses were performed using MEGA 6.0 software (Tamura et al. 2013).Data regarding the antifungals susceptibility of the two varieties are scarce, particularly of the Brazilian isolates. As this information is important bothfor therapeutic and epidemiological purposes, the susceptibilities of all the clinical isolates to seven currently used antifungal drugs were determined. The susceptibility tests were conducted according to the Clinical Laboratory Standards Institute (CLSI) reference method for broth dilution antifungal susceptibility testing of filamentous fungi (CLSI document M38-A2) (Wayne 2008). The antifungal drugs amphotericin B (AMB), ketoconazole (KET), fluconazole (FLU), voriconazole (VOR), 5-flucytosine (5FC), and terbinafine (TER) were purchased from Sigma-Aldrich. Caspofungin (CAS) was purchased from Merck Sharp & Dohme Limited (Kenilworth, NJ, USA). The final concentrations were 0.03 to 32 µg mL-1 for AMB, 0.03 to 16 µg mL-1 for VOR and TER, 0.125 to 64 µg mL-1 for CAS, and 0.25 to 128 µg mL-1 for 5FC, KET and FLU.

The experiments were performed in 96-well, flat-bottomed microtitre plates (TPP, Switzerland) with RPMI 1614 culture medium (Gibco, Invitrogen Corporation, NY, USA) buffered with 0.165 M 3-(N-morpholino)propanesulfonic acid (MOPS), pH 7.0, antifungal drugs and 2 × 104 arthroconidia mL-1. After 48 h of incubation at 28ºC, minimal inhibitory concentration (MIC) was determined visually by comparison with the growth in the control well (culture medium and 2× 104 athroconidia mL-1). The MIC endpoint was defined as the minimal concentration of the drug capable of inhibiting 100% of fungal growth compared with the growth control for amphotericin B and 5-flucytosine and to inhibit 90% for azoles, terbinafine and caspofungin. Geometric means and MIC ranges were calculated for each species and antifungal drug. The standard quality control strains of the experiment were Aspergillus flavus (ATCC®204304™) and Candida krusei (ATCC®6258). The tests were repeated three times.The phenothiazinium photosensitizers methylene blue (MB), new methylene blue N zinc chloride double salt (NMBN) and toluidine blue O (TBO) were purchased from Sigma-Aldrich. The novel pentacyclic phenothiazinium photosensitizer S137 was synthesized as previously described (Wainwringht et al. 2011). The chemical structures of all the PS are show in supplementary Figure S1. Stock solutions of all PS were prepared with phosphate buffered saline (PBS), pH 7.4 and stored as previously described (Rodrigues et al. 2012a).Light was provided by an array of 96 light-emitting diodes (LED) with emission peak at 631 nm (LED96). The array was made in-house using the Cree®5-mm Round LED (model # LC503UHR1-1 5Q-Q, CREE Inc., Durham, NC, USA).

The integrated irradiance in the visible spectrum (400 to 700 nm) was 13.89 mW cm-2. Light measurement was performed by using a cosine- corrected irradiance probe (CC-3-UV, Ocean Optics, USA) screwed onto the end of an optical fiber coupled to an USB4000 spectroradiometer (Ocean Optics, Dunedin, FL, USA). Light was measured inside the well, at the sample level, to reduce interference of the plastic plate. The emission spectra of the LED array and the absorption spectra of the PS are presented in supplementary Figure S2.Minimal inhibitory concentration (MIC)-based experiments were conducted as previously described (de Menezes et al. 2014; Rodrigues et al. 2012a; Wainwringht et al. 2016) to determine the best conditions for APDT of arthroconidia of N. dimidiatum and N. dimidiatum var. hyalinum. Eleven isolates, three of the sequence type 1, one of the sequence type 2, one of the sequence type 3, two of the sequence type 4, three of the sequence type 5 and one of the sequence type 6 were evaluated. Experiments were performed in 96-well microtiter plates (TPP, Switzerland). Fifty µL of the arthroconidia suspension (4× 104 athroconidia mL-1) and 50 µL of the PS solution were added to each well. Final concentrations of MB, NMBN and TBO were 0, 1, 2.5, 5, 10, 12.5, 25, 50,75, 100, and 200 µM and the final concentrations of S137 were 0, 0.5, 1, 2.5, 5,10, 12.5, 20, 25, 30 and 40 µM. Plates were incubated in the dark for 30 min at 28°C and exposed to light fluences of 3, 6, and 11 J cm-2 or kept in the dark (dark controls).

After the exposures, 100 µL of two-fold concentrate RPMI 1614 culture medium buffered with 0.165 M MOPS pH7.0 were added to each well and plates were incubated at 28°C. Fungal growth in each well was evaluated after 7 days by visual inspection when the MICs were determined. MIC was considered the minimal concentration of the PS for each fluences, in which total growth inhibition was achieved. Three independent experiments in triplicates were performed. Based on the previous MIC experiments, the best condition were established and the effect of APDT on the survival of ungerminated arthroconidia was determined as previously described (Rodrigues et al. 2012a; Rodrigues et al. 2013). Two selected isolates of N. dimidiatum (LMC 302.01 and 301.01) and one of N. dimidiatum var. hyalinum (LMC 313.01) were evaluated. Experiments were performed in a 96 microtiter plate (TPP, Switzerland). Fifty µL of the arthroconidia suspension and 50 µL of the PS solution (MB, NMBN, TBO, or S137) were added to each well. The final concentration of the arthroconidia in the mixture was 2 × 107 artroconidia mL-1 and the final concentrations of MB, NMBN, TBO and S137 were 200, 200, 200, and 25 µM, respectively, for N. dimidiatum; and 25, 25, 10, and 10 µM, respectively, for N. dimidiatum var. hyalinum. Plates were maintained in the dark for 30 min at 28°C and exposed to light fluenc es of 3, 6, and 11 J cm-2 or kept in the dark. After exposure, arthroconidial suspensions were removed, serially diluted 10-fold in PBS, and 50 µL of each dilution was spread onto potato dextrose agar (Acumedia) supplemented with 0.12 g L-1 deoxycholic acid sodium salt (Fluka, Italy).

Plates were incubated in the dark at 28°C and colony forming units (CFU) were counted daily at 8× magnification for up to 4 days. Three replicate dishes were prepared for each treatment in each experiment and three independent experiments were performed. The effect of the different treatments was calculated as previously described (de Menezes et al. 2014).The effect of APDT was further evaluated on germinated arthroconidia. For this, the arthroconidia were previously germinated in RPMI 1614 culture medium buffered with 0.165 M MOPS pH7.0 at 28ºC for 4 h. APDT of germlings was carried out exactly as ungerminated arthroconidia.The permeability of the arthroconidial plasma membrane after APDT was evaluated by measuring the incorporation of propidium iodide (PI) by arthroconidia. PI is a positively charged fluorescent nucleic acid dye which only penetrates cells with severe membrane damage (Davey & Hexley 2011; de Menezes et al. 2016). Experiments were performed in a 24-well flat bottomed microtiter plate (TPP, Switzerland). Ungerminated arthroconidia were exposed to APDT as previously described. After APDT, arthroconidia were transferred from each well to 1.5-mL tubes, washed once with PBS and PI solution was added to a final concentration of 10 µg mL-1 immediately before flow cytometric analysis. The single-cell light scattering and fluorescence measurements were performed using a FACSCanto flow cytometer (Becton Dickinson, Sunnyvale, CA, USA).

Ten thousand events were analyzed by FACSDiva software (BD, San Jose, CA, USA). The percentage of cells stained with PI after each treatment was determined. Three independent experiments were performed. The results were compared with the ‘gold-standard’ determination of viable cells (arthroconidia survival fraction) by CFU-based experiments. Arthroconidia not treated with PS neither exposed to light were used as control of conidial viability and arthroconidia treated with 70% ethanol for 15 min were used as positive control (death control).The evaluation of lipid peroxidation in arthroconidia after APDT was determined by measuring the production of malonildialdehyde (MDA) using the Lipid Peroxidation (MDA) Assay Kit (Sigma-Aldrich, Inc. – St. Louis, MO, USA) following the manufacturer’s protocol with slight modification. MDA is a product of the oxidation of polyunsaturated lipids by reactive oxygen species (ROS). Arthroconidia were exposed to APDT as previously described. After treatments, arthroconidia were recovered from each well, transferred to a 1.5-mL microtube, centrifuged for 5 min at 5000 g and lysed with MDA lysis buffer (300 µL) and BHT 100× (3 µL). The insoluble material was removed by centrifugation for 10 min at 13000 g. Two hundred microliters of the supernatant from each sample were transferred to a new 1.5-mL microtube and incubated at 95ºC for 1 h with 600 µL of TBA solution. Samples were cooled in ice bath for 10 min and 200 µL of each sample were transferred to a 96-well flat-bottomed microtitre plate. Absorbances at 532 nm were determined using a Spectramax® Paradigm® Multi-Mode Detection Platform (Molecular Device LLC, Sunnyvale, CA). A standard curve was used to calculate MDA concentration in each sample. Three independent experiments were performed in triplicate. The results were also compared with the arthroconidia survival determined by CFU counting.Treatment comparisons were performed using one-way analysis of variance (one-way ANOVA) followed by Tukey’s post-test and P values of <0.05 were considered significant. All computations were performed using GraphPad Prism Software (v 5.0 GraphPad Software, La Jolla, CA, USA). 3.Results MLST was performed by the sequencing of ITS1 and ITS2 regions of the rDNA and the conserved genes CHS, HH3, GPD, TUB and EF1 in order to determine the genotypes of a collection of 24 N. dimidiatum and 6 N. dimidiatum var. hyalinum strains isolated from skin and toenail infections. Two ATCC strains were also genotyped (Table S1). The nucleotides sequence of ITS2 region and conserved genes CHS and GPD did not show polymorphic sites among the clinical isolates and ATCC strains. However, eight polymorphic sites were found in the sequence of ITS1 region and EF1, TUB and HH3 genes (Table 1). ITS1 has a transition from guanine to adenine at 105 bp of the sequenced region (Table 1). The nucleotide sequence of EF1 has shown four polymorphic sites throughout the intron sequence. The transitions from thymine to cytosine, cytosine to thymine, adenine to guanine and the transversion from adenine to cytosine were found at sites 103, 188, 198 and 214 bp, respectively. Two polymorphic sites were found in the TUB gene. The first site is the transition from thymine to cytosine at position 312 bp. The codon CTT (cytosine thymine thymine) encodes the amino acid leucine while the codon CCT (cytosine cytosine thymine) encodes the amino acid proline. The second polymorphic site has shown a transition from cytosine to thymine at position 319 bp which is a silent mutation since both codons encode serine. The nucleotide sequence of HH3 has shown a polymorphic site at 265 bp with a transition from adenine to guanine, which represents a silent mutation, in which both codons encode glutamate (Table 1). The combination of the eight polymorphic sites found in ITS1 region of the rDNA and in the sequences of the EF1, TUB, and HH3 genes determined six sequences types (STs) (Table 1). Most of the N. dimidiatum clinical isolates (64%) were clustered in ST1 and 5.2%, 2.6% and 12.8% of the isolates were clustered in ST2, ST3 and ST4, respectively. The N. dimidiatum var. hyalinum represents 15.4% of the clinical isolates and all of them together with the isolate ATCC®38907™ (N. dimidiatum var. hyalinum) were clustered in ST5. The only member of ST6 was the isolate ATCC®22190™ (N. dimidiatum). The evolutionary history of N. dimidiatum and N. dimidiatum var. hyalinum isolates using the 2075 bp originated from the concatenation of the seven loci dataset (ITS1, ITS2, EF1, TUB, HH3, CHS and GPD), generated the UPGMA dendrogram based on p-distance (Fig. 1). The bootstrap values were registered in the nodes of the UPGMA phylogenetic tree. The main set of the isolates could be differentiated into two subgroups (A and B), with strong bootstrap support. Significant association was found between the subgroups, regarding the presence or absence of the dark-brown pigment. All the isolates of the subgroup A produce pigments and their arthroconidia and hyphae are pigmented while none of the isolates of the subgroup B produces pigments and their arthroconidia and hyphae are hyaline. The result shows that the isolates of the two varieties can be accurately differentiated by the multilocus sequence typing. The in vitro activities of seven antifungal drugs were tested against 24 N. dimidiatum and 6 N. dimidiatum var. hyalinum clinical strains isolated from skin and toenail infections, and two ATCC strains. The antifungal drugs which demonstrated lower minimal inhibitory concentrations (MICs) against all isolates (STs 1 to 6) were AMB (MIC range 0.03 - 0.5 µg mL-1 and Geometric Mean (GM) MIC 0.126 µg mL-1), TER (0.125 - 4 µg mL-1 and 0.75 µg mL-1) and VOR (0.125 - 4 µg mL-1 and 1.68 µg mL-1). 5FC (1 - 32 µg mL-1 and 12.95 µg mL-1) and CAS (8 - 64 µg mL-1 and 27.02 µg mL-1) exhibited lower MICs in comparison with KET (0.5 - >128 µg mL-1 and 58.81 µg mL-1) and FLU (8 – >128 µg mL-1 and 95.63 µg mL-1) (Table 2). Compared with N. dimidiatum isolates (ST1 to 4), N. dimidiatum var. hyalinum isolates (ST5) were more sensitive to VOR (0.125 – 0.5 µg mL-1 and 0.27 µg mL-1), KET (0.5 – 16 µg mL-1 and 3.62 µg mL-1) and FLU (8 – 64 µg mL-1 and 39 µg mL-1 – P < 0.05). ST6 showed lower MIC than STs 1 to 4 for FLU (32 - 64 µg mL-1 and 45.25 µg mL-1) and KET (1 µg mL-1 and 1 µg mL-1 – P < 0.05) (Table 2). Exposure only to red light did not inhibit the growth of any of the isolates. In the dark, treatments with MB, NMBN and TBO in concentrations up to 200 µM did not inhibit the growth of any of the isolates, while S137 at concentrations > 50 µM inhibited the growth of all of them (Table 3). The MICs of each photosensitizer (MB, NMBN, TBO and S137) at each fluence for the 11 selected isolates of N. dimidiatum and N. dimidiatum var. hyalinum are shown in Table 5.

Based on the PS MIC, S137 was the most effective PS followed by NMBN. MB and TBO were less effective. Concentrations of S137 > 25 µM inhibited the growth of all the isolates at all the fluences, and most of the isolates were inhibited by concentrations > 10 µM. MIC-based experiments also showed that arthroconidia of the N. dimidiatum var. hyalinum isolates (ST 5) are more susceptible to APDT than the pigmented arthroconidia of the N. dimidiatum (ST 1, 2, 3 and 4) regardless the photosensitizer used (Table 5).The criteria used to select the isolates used in this experiment were the MICs for the different PS and colony color. The MICs of the isolates of the ST1, 2, 3 and 4 were similar but they differ from those of the ST5. Regarding the colony color, N. dimidiatum isolates have shown different pattern (i.e. brown and gray), while none of the isolates of N. dimidiatum var. hyalinum produced pigments (ST5) (Fig. 2). Thus, the effects of APDT on the survival of urgeminated arthroconidia was evaluated on two selected isolates of N. dimidiatum (LMC 302.01 – brown and 303.01 – gray), which belong to ST1, and one isolate of N. dimidiatum var. hyalinum (LMC 313.01 – hyaline), which belongs to ST5 (Fig. 2). The concentrations of the different PS were selected based on the results of the MIC experiments (Table 5). Thus, PS concentrations used for N. dimidiatum were 200 µM for MB, NMBN and TBO, and 25 µM for S137.

PS concentrations used for N. dimidiatum var. hyalinum were 25 µM for MB and NMBN, and 10 µM for TBO and S137. Exposures only to light andtreatments with MB, NMBN or TBO in the dark did not kill the conidia of any isolate. Treatment with S137 in the dark killed 58.3%, 39.8% and 0% of the arthroconidia of N. dimidiatum (LMC302.01), N. dimidiatum (LMC303.01) and N. dimidiatum var. hyalinum (LMC313.01), respectively. APDT with MB and TBO at 3 J cm-2 reduced the survival of N. dimidiatum (both isolates) and N. dimidiatum var. hyalinum arthroconidia by 1 log and 4 logs, respectively. At 6 J cm-2 reductions were 3 logs and 5 logs, and at 11 J cm-2 reductions were 4.5 logs and 5 logs, respectively (P < 0.05 for all treatment comparisons). APDT with NMBN at 3 J cm-2 reduced the survival of N. dimidiatum (LMC302.01 and LMC303.01) and N. dimidiatum var. hyalinum (LMC313.01) arthroconidia by 3 logs, 3 logs and 5 logs, respectively. At 6 and 11 J cm-2, reductions in the survival were 5 logs, 4 logs and 5 logs for the three isolates, respectively (P < 0.05 for all treatment comparisons) (Fig. 3). APDT with S137 at 3 J cm-2 reduced the survival of N. dimidiatum (LMC302.01 and LMC303.01) and N. dimidiatum var. hyalinum (LMC313.01) by 3.5 logs, 2.5 logs and 2 logs, respectively. At 6 J cm-2, reductions were 5 logs, 2.5 logs and 2.5 logs, and at 11 J cm-2 the reductions were 5 logs, 5 logs and 3.5 logs (P < 0.05 for all treatment comparisons) (Fig. 3). In addition, APDT with MB, NMBN, TBO and S137 at the fluence of 6 J cm-2 were carried out with 4-h germinated arthroconidia of N. dimidiatum (LMC302.01 and LMC303.01) and N. dimidiatum var. hyalinum (LMC313.01). Results were very similar to those observed with ungerminated arthroconidia (Supplementary Fig. S3). Thus, APDT with MB, NMBN, TBO and S137 are able to kill both ungerminated and germinated arthroconidia of N. dimidiatum and N. dimidiatum var. hyalinum equally well.The effects of APDT with the four PS on the arthroconidial survival and plasma membrane permeability of the three selected isolates are shown in Fig.4.Ungerminated arthroconidia were treated with MB, NMBN, TBO or S137 and exposed to light or kept in the dark. PS concentrations used for the two N. dimidiatum isolates (LMC302.01 and 303.01) were 200 µM for MB, NMBN and TBO, and 25 µM for S137, and concentrations used for the N. dimidiatum var. hyalinum isolate (LMC313.01) were 25 µM for MB and NMBN, and 10 µM for TBO and S137. After APDT, cells were treated with PI, which only penetrates cells with damage in the plasma membrane. Therefore, the higher the percentage of arthroconidia stained with PI, the greater the damage caused by APDT on the membrane. Exposure only to light did not kill arthroconidia or damage their plasma membrane. In the dark, none of the four PS killed the arthroconidia, but all of them caused some damage to the plasma membrane. The extent of the damage varied depending on the PS and isolate. Despite the lower concentration used, S137 was the PS that caused more damage to the plasma membrane of the three isolates as assessed by PI incorporation. The results also showed that PI incorporation by arthroconidia does not always mean that the stained arthroconidia are dead. For example, treatment with S137 in the dark resulted in PI incorporation by virtually all the arthroconidia, but they were still viable. APDT with all the PS killed the arthroconidia of the three isolates and also damaged their plasma membranes. PI incorporations were close to 100% after all treatments (PS and fluences) (Fig. 4).After APDT, an inverse correlation was observed between the arthroconidia survival and the percentage of conidia stained with PI (Fig. 4).Lipid peroxidation was evaluated by the cellular quantification of malondialdehyde (MDA). The effects of APDT with MB, NMBN, TBO and S137 on arthroconidial survival and lipid peroxidation are show in Fig. 5. Exposures only to light or treatments only with the PS did not kill the arthroconidia or result in lipid peroxidation. APDT with NMBN or S137 but not with MB or TBO resulted in the oxidation of the lipids in arthroconidia of the three fungal isolates (Fig. 5). 4.Discussion The accurate identification of the isolates of the two N. dimidiatum varieties is important not only to guide the therapy, since the varieties present different profiles of susceptibility to antifungal drugs, but also to allow epidemiological and biogeography studies. To date, few studies regarding the molecular distinction between the two varieties have been reported and their results varied. Chromosomal DNA of both N. dimidiatum and N. dimidiatum var. hyalinum has similar percentages of guanine and cytosine (Davison et al. 1980). Amplified ribosomal DNA restriction analysis conducted with isolates from North America, Europe and North Africa showed that N. dimidiatum is identical to N. dimidiatum var. hyalinum (Roeijmans et al. 1997). PCR-restriction fragment length polymorphism ribotyping method showed an identical sequence for the 18S ribosomal DNA gene of both N. dimidiatum and N. dimidiatum var. hyalinum (Machouart-Dubach et al. 2002). More specific molecular methods, such as the sequencing of the 18S subunit of ribosomal RNA performed with isolates from France, North America and North Africa, identified nucleotide polymorphism between N. dimidiatum var. hyalinum and N. dimidiatum isolates. Additionally, an intronic insertion was observed exclusively in N. dimidiatum isolates (Machouart et al. 2004). The genotyping of N. dimidiatum and N. dimidiatum var. hyalinum by the sequencing of four loci (the tubulin and the chitin synthase genes and the internal transcribed spacer region and D1/D2 domain of the 28S rRNA gene) revealed five sequence types (ST) (Madrid et al. 2009). In our study, the genotyping included the sequence analysis of seven loci of 24 N. dimidiatum and 6 N. dimidiatum var. hyalinum clinical isolates. Eight polymorphic sites have been detected and six ST were determined. Interestingly, all the clinical isolates of N. dimidiatum var. hyalinum evaluated in the present study and the hyaline ATCC®38907™ isolate clustered together in the same sequence type (ST5). The polymorphic sites in ITS1 and TUB allow differentiation of the ST5 isolates from the other ST (ST1 to 4 and ST6). Some of these polymorphic sites were previously described in N. dimidiatum and N. dimidiatum var. hyalinum strains isolated from different sources in Brazil, Egypt, USA, Mali and the United Kingdom (Madrid et al. 2009). Therefore, our results showed that the two varieties can be accurately differentiated based on the multilocus polymorphism analysis. Also, the genetic difference between the two varieties may be higher than it is generally assumed, involving other loci than those responsible for pigments biosynthesis. The extent of this difference will be only known when several isolates of the two varieties have their genomes sequenced. Thus, for now, differences observed between the two varieties, including their susceptibilities to antifungals and to APDT, should not be attributed only to the difference in pigmentation. The reference microdilution methodologies for in vitro antifungal susceptibility testing have not been standardized for Neoscytalidium spp. (Chowdwary et al. 2014; James et al. 2017; Valenzuela-Lopez et al. 2017). Thus, results of the in vitro susceptibility testing with these fungi should be interpreted very carefully. AMB, VOR and TER are ranked as the antifungal agents with the highest in vitro antifungal activity against N. dimidiatum and the hyaline variety (Bueno et al. 2010; Carrillo-Munoz et al. 2007; Dunand & Paugam 2008; Lacroix & Chauvin, 2008; Madrid et al. 2009). In accordance with these data, we found that the isolates of N. dimidiatum (ST 1 to 4) and N. dimidiatum var. hyalinum (ST 5) were more susceptible to AMB, VOR and TER than to the other antifungal agents tested. Additionally, our results clearly showed that the isolates of the hyaline variety are much less tolerant to the azole antifungal drugs FLU, VOR and KET than the isolates of the pigmented variety. A previous study showed that isolates of N. dimidiatum var. hyalinum from Brazil and France were less tolerant to VOR than isolates of the pigmented N. dimidiatum (Lacroix & Chauvin, 2008; Madrid et al. 2009). Antimicrobial photodynamic therapy with phenothiazinium PS has been shown to be a potential alternative to conventional antifungal agents for the treatment of mycosis. The use of MB and TBO in the photodynamic treatment of onychomycosis was recently reported. Local application of PS (MB and TBO) and illumination of infected nails produced a response in 53 out of 62 patients, without any collateral effect, and 28 patients showed complete clearance (Tardivo et al. 2015). APDT with methyl-aminolevulinate was successfully used to cure a woman with onychomycosis of the toes caused by N. dimidiatum (Aspiroz et al. 2013). In the present study, we demonstrated the in vitro efficacy of APDT with the phenothiazinium PS MB, NMBN, TBO and S137 both on urgerminated and germinated arthroconidia of the dematiaceous fungus N. dimidiatum and the hyaline variety N. dimidiatum var. hyalinum. The ability to kill ungerminated conidia is a positive aspect of APDT, since conidial tolerance to currently-used antifungal agents is frequently pointed out as one of the causes of therapeutic failure and recurrence of the infections (Coelho et al. 2008; Hashimoto & Blumenthal 1978; Martinez-Rossi et al. 2008). Arthroconidia of N. dimidiatum isolates were more tolerant to APDT with all the four phenothiazinium PS than the hyaline isolates (see Table 5). The protective effect of melanins and melanin-like pigments against conventional antifungal drugs (Baltazar et al. 2015; Nosanchuck & Casadevall 2006; van Duin et al. 2008) and also to APDT (Prates et al. 2013; Rodrigues et al. 2012b) was observed previously both in conidia and in vegetative yeast cells. Pigmented conidia of Penicillium and Metarhizium are much more resistant to APDT with different PS than the colorless mutants (Asthana & Tuvson 1992; Gonzales et al. 2010). In yeasts, Cryptococcus neoformans melanized cells are more resistant to APDT than the nonmelanized cells (Prates et al. 2013; Rodrigues et al. 2012b). Pigments, such as melanins, are present in the conidial cell wall and confer protection against APDT presumably by screening out the wavelengths that would otherwise activate the PS and by quenching reactive oxygen species (Gonzales et al. 2010). Interestingly, arthroconidia of N. dimidiatum are intrinsically much more resistant to APDT with the phenothiazinium PS than conidia of other previously studied species, such as Colletotrichum acutatum, C. gloesporioides (de Menezes et al. 2014), Aspergillus nidulans (Gonzales et al. 2010), Fusarium oxysporium, F. moniliforme and F. solani (de Menezes et al. 2016). This is not surprising, since the different types of fungal asexual propagules (i.e. conidia, microconidia and arthroconidia) are very different in relation to their ontogeny, dimensions, morphology, physiology, and physical properties and chemical composition (Turian 1976). Among the tested photosensitizers, the pentacyclic derivative S137 was the most effective. The MICs of S137 were lower than the MIC of other PS for the isolates of both the varieties N. dimidiatum and N. dimidiatum var. hyalinum (see Table 5). Higher concentrations of MB, TBO or NMBN than S137 were necessary to kill ungerminated and germinated arthroconidia of the dematiaceous isolates (see Figs. 2 and S3). Additionally, should be taken into account that, despite the higher activity of S137 compared with the other PS, the light source employed was optimal (i.e. coinciding with the longer-wavelength monomer peak) for TBO and NMBN but not for S137 or MB in terms of output wavelength (supplementary Fig. S2) (Rodrigues et al. 2013). The higher efficacy of APDT with S137 compared with the other phenothiazinium PS, including MB, was previously observed against microconidia of the dermatophytes T. rubrum and T. mentagrophytes (Rodrigues et al. 2012a); the plant-pathogenic fungi Colletotrichum acutatum, C. gloeosporioides (de Menezes et al. 2014), Fusarium oxysporum, F. moniliforme and F. solani (de Menezes et al. 2016); the saprophyte Aspegillus nidulans (Gonzales et al. 2010), and also against planktonic cells of several species of Candida (Rodrigues et al. 2013). The ability of the PS to bind to microbial cells and to produce singlet oxygen is important to the efficiency of the APDT. Although MB, NMBN, TBO and S137 do not differ much in the production of singlet oxygen, they differ markedly in their ability to bind to plasma membranes (Bacellar et al. 2014). Both MB and TBO display log Poil/water < 0 which indicate a low lipophilicity. While NMBN and S137 exhibit a higher lipophilic profile (log Poil/water > 0) (Bacellar et al. 2014; Wainwright et al. 1998; Wainwright et al. 2012). The lipophilic feature of the PS molecule determines its capacity to bind to cell membranes and is critical to define the extent of photoinduced membrane damage and consequently the efficiency of cell killing (de Menezes et al. 2016; Wainwright et al. 1998). We evaluated the effect of the APDT with the four phenothiazinium PS on arthroconidia plasma membrane by evaluating changes in membrane permeability and peroxidation of the lipids. The propidium iodide (PI) staining- based assay is widely used to evaluate the membrane integrity (Davey & Hexlley 2011) and also has been considered a good marker for cell death associated with membrane alterations in fungi (Pina-Vaz et al. 2005). PI is a membrane-impermeable fluorescent dye that only penetrates cells with damaged plasma membranes. Treatments with MB, NMBN and TBO at 200 µM in the dark increased the membrane permeability of N. dimidiatum arthroconidia, and treatment with S137 at 10 µM had an even stronger effect, causing the staining of 100% of the arthroconidia of the three isolates (see Fig. 4). Interestingly, despite the disturbance of the membrane that increased cell permeability to PI, most of the arthroconidia were still viable (see Fig. 4), which clearly indicates that the permeability to PI is not always a good marker for cell death in fungi. APDT with MB, NMBN, TBO and S137 increased the plasma membrane permeability of arthroconidia of N. dimidiatum and N. dimiditum var. hyalinum isolates, causing the PI staining of approximately 100% of the arthroconidia.

Polyunsaturated lipids are easily oxidized by reactive oxygen species resulting in a chain of reactions that has malondialdehyde as a major product (Nair & Turner 1984). APDT with S137 or NMBN but not with MB or TBO caused the peroxidation of arthroconidial lipids and consequently malondialdehyde production (see Fig. 5). Membrane damage efficiency of photodynamic treatment with phenothiazinium PS such as MB, TBO, 1,9- dimethyl methylene blue (DMMB) and S137 was previously studied using artificial liposomes (Bacellar et al. 2014). The membrane damage efficiency followed the order S137 > DMMB > MB > TBO. Only the PS that had higher membrane/solution partition (S137 and NMMB) could permeabilize the lipid bilayer and cause membrane leakage and lipid oxidation. Moreover, only S137 altered the membrane structure. To some extent, the present study confirmed these findings in living cells. In a recent study performed with microconidia of three Fusarium species we reported that APDT with the same phenothiazinium PS used in the present study damages the microconidial membrane but only with NMBN and S137 caused the peroxidation of the membrane lipids. Thus, despite the great difference between the two types of propagules (arthroconidia and microconidia), the effects on APDT on the membranes were similar and seems to be common to the different types of conidia.

5.Conclusion
The pigmented variety N. dimidiatum differs from the hyaline variety in other loci than the ones involved in the pigment biosynthesis, and the two varieties can be differentiated by multilocus sequence typing. The arthroconidia of the hyaline variety is more susceptible to antifungal drugs as well to APDT with the phenothiazinium photosensitizers. APDT Caspofungin with the four PS damages the plasma membrane of arthroconidia of both varieties increasing its permeability to PI but only NMBN and S137 caused peroxidation of the membrane lipids.