Human adaptation and diversification in the Microsporum canis complex

Abstract

The Microsporum canis complex consists of one zoophilic species, M. canis, and two anthropophilic species, M. audouinii and M. ferrugineum. These species are the most widespread zoonotic pathogens causing dermatophytosis in cats and humans worldwide. To clarify the evolutionary relationship between the three species and explore the potential host shift process, this study used phylogenetic analysis, population structure analysis, multispecies coalescent analyses, determination of MAT idiomorph distribution, sexual crosses, and macromorphology and physicochemical features to address the above questions. The complex of Microsporum canis, M. audouinii and M. ferrugineum comprises 12 genotypes. MAT1-1 was present only in M. canis, while the anthropophilic entities contained MAT1-2. The pseudocleistothecia were yielded by the mating behaviour of M. canis and M. audouinii. Growth rates and lipase, keratinolysis and urea hydrolytic capacities of zoophilic M. canis isolates were all higher than those of anthropophilic strains; DNase activity of M. ferrugineum exceeded that of M. canis. The optimum growth temperature was 28 °C, but 22 °C favoured the development of macroconidia. Molecular data, physicochemical properties and phenotypes suggest the adaptation of zoophilic M. canis to anthropophilic M. ferrugineum, with M. audouinii in an intermediate position.

INTRODUCTION

The dermatophytes comprise one of the most prevalent groups of zoonotic fungal pathogens (Chermette et al. 2008; Havlickova et al. 2008). Infections are typically superficial, involving keratinized structures such as skin, nails, and hair. Currently, seven genera of dermatophytes are accepted, of which members of Trichophyton, Epidermophyton, and Microsporum are commonly associated with human dermatophytosis (de Hoog et al. 2017). From an ecological and clinical perspective, three approximate groups are distinguished, i.e., geophilic, zoophilic, and anthropophilic species. Classically, it has been supposed that the evolution from geo- to anthropophilic lifestyles has taken thousands of years. This refers to the evolution from geophily in Arthroderma to anthropophily in Trichophyton. However, Tang et al. (2021) suggested that this may also occur at a smaller scale within single species complexes. For example, Trichophyton mentagrophytes represents an interbreeding cloud of genotypes with a preponderantly zoophilic lifestyle, while T. indotineae and T. interdigitale are anthropophilic clonal offshoots (Tang et al. 2021). Similar taxonomic structures can be observed in the prevalently zoonotic T. benhamiae species complex (Sabou et al. 2018).

The zoophilic species Microsporum canis is flanked by anthropophilic clones, M. audouinii and M. ferrugineum. The group is phylogenetically remote from all other dermatophytes. Microsporum canis is prevalent in cats and dogs, and when transmitted to humans, causes tinea capitis, tinea faciei and tinea corporis. Tinea capitis may show an often severe inflammatory host response, presenting pruritic, scaly areas of alopecia. In contrast, the anthropophilic species M. audouinii and M. ferrugineum are generally transmitted from human to human, causing tinea capitis, mainly presenting with limited inflammation and minimal symptoms. Microsporum audouinii has occasionally been detected in animals and in the environment (Brasch et al. 2015; Chah et al. 2012; Jain et al. 2011, 2012), while also some cases of severe inflammation have been reported (Fernandes et al. 2013; Smith et al. 1991; West 1982). From the early 1930s to the fifties, M. audouinii pandemics were a major cause of tinea capitis in Europe and the U.S.A. (Brito-Santos et al. 2017). With the tinea capitis eradication campaign in the 1950s, M. audouinii decreased significantly and was eventually confined to Africa (Oke et al. 2014). However, pet ownership grew in popularity from the 1990s, the zoophilic M. canis became the predominant agent of tinea capitis in Europe and Asia, and during the last decades, M. audouinii reemerged in school children (Kieliger et al. 2015; Kolivras et al. 2003; Leeming et al. 1995; Viguié-Vallanet et al. 2005). Microsporum ferrugineum is considered endemic between the Balkan, East Asia and Nigeria (Zhan et al. 2015) and has been reported in Europe and the Americas with the flow of migrants (Nenoff et al. 2020). It is worth mentioning that M. equinum, an obsolete name for M. canis strains causing horse tinea, has been reported from several locations during the last century (Aho 1987; Kane et al. 1982; Takatori et al. 1985). The identity of M. equinum remains controversial, as it differs morphologically and physicochemically from M. canis, but molecularly it is an infraspecific taxon and was synonymised with M. canis (Gräser et al. 2000). Kano et al. (2001) reported that in the chitin synthase 1 gene, M. equinum was close to Arthroderma otae. Still, when crossed with tester strains of A. otae, M. equinum failed to produce ascomata (Takatori et al. 1985).

Significant phenotypic and ecological variation exists between the three members of the M. canis complex, but evidence of sexual reproduction underlines a strong connection between the species. Fertile gymnothecia have been observed in mating experiments (Hasegawa et al. 1974), and by nomenclatural rules of that time, the sexual state was described as Arthroderma otae (= Nannizzia otae). Monoascospore cultures were demonstrated to be heterothallic (Hasegawa et al. 1974). Since then, an imbalance of the MAT idiomorph was shown in M. canis, with the predominance of MAT (-). Microsporum audouinii and M. ferrugineum were considered to be strictly clonal.

Whether the above anthropophilic and zoophilic species should be considered members of a single biological species represented by the sexual state ‘Arthroderma otae’ is still an open question. To better delineate species identities in the M. canis complex and to understand the transition process from animal to human hosts, the present study combines multiple approaches, including molecular phylogenetic and population genetic analysis, using a global set of isolates with a focus on human strains. We screened the variability of phenotypic, physiological, and genetic properties of the three taxonomic entities and determined the distribution of MAT idiomorphs.

MATERIAL AND METHODS

Strains

Microsporum canis, M. audouinii, and M. ferrugineum were isolated from patient tinea capitis, tinea corporis, and animal dermatophytosis. The 183 strains obtained from the collection conservation of the Department of Dermatology, the Third Affiliated Hospital of Sun Yat-sen University (n = 20); Clinical Microbiology, University of Liège, Belgium (n = 15); Faculty of Medicine, University of Ljubljana (n = 28); Department of Dermatology, UH Cleveland Medical Center (n = 37); Labor für Medizinische Mikrobiologie—Partnerschaft (n = 30); Microbiology, Parasitology and Pathology Graduate Program, Federal University of Paraná (n = 13); Department of Veterinary Medicine, University of Bari Aldo Moro (n = 28); and Department of Veterinary Sciences, University of Turin (n = 12). The Westerdijk Fungal Biodiversity (Utrecht, The Netherlands) provided the tester strains CBS 495.86 and CBS 496.86. Detailed information on the strains' origin is listed in Additional File 6: Table S1.

Molecular studies

A rapid DNA extraction method was used to extract DNA from 185 strains. The strains were incubated in Potato Dextrose Agar (PDA, Oxoid, U.K.) at 28 °C for 14 days. Conidia or hyphae were picked from colonies and dissolved in 300 μL Breaking buffer (containing 2.14% w/v Triton X-100, 1% w/v SDS, 0.585% w/v NaCl, 0.1575% w/v Tris–HCL, 0.0292% w/v EDTA). Samples were shaken at 1,400 rpm for 45 min at 60 °C incubation. After adding 250 μL 25:24:1 phenol–chloroform-isoamyl alcohol (Sigma, U.S.A.), the mixture was centrifuged at 13,000 rpm for 10 min at room temperature and the supernatant was collected. The rapid DNA extraction method bypassed the detection of DNA concentration and went directly to thermal cycling.

ITS1/ITS4 or ITS1F/NL4 (Aneke et al. 2021) were used to amplify the ITS (ITS1-5.8S-ITS2) rDNA region. The partial β-tubulin II (tub2) was amplified using the primers BT-2a/T2 (Choi et al. 2012). EF-DermF/EF-DermR (Rezaei-Matehkolaei et al. 2012) amplification of translation elongation factor 1 (tef-1α) was done. DNA topoisomerase I (topI) and II (topII) were amplified using the TOP1 501-F/TOP1 501-R (Stielow et al. 2015), TOPII-F1/TOPII-Mic (Shamsizadeh et al. 2020) primer sets, respectively. The 60S L10 (L1) region was amplified using 60S-908R/60S-506F (Stielow et al. 2015). The primer sequences are shown in Additional file 5.

The 25 μL PCR reaction system contained 0.25 μL high-fidelity DNA polymerase (Thermo Fisher Scientific, U.S.A.), 0.5 μL DNA, 0.5 μL dNTPs, 1 μL forward and reverse primers, 5 μL HF buffer and 17.75 μL nuclease-free water. The PCR thermal cycle profile for each primer pair is shown in Additional file 5. DNA purification was carried out according to the instructions of the QIAquick gel Extraction Kit (QIAquick, Germany).

Phylogenetic analyses

The harvested DNA forward and reverse sequences were assembled in SnapGene v6.0.2 (Evans et al. 2018), and multiple sequence alignments were performed using Clustal Omega. The maximum likelihood (ML) method was used to construct phylogenetic trees for a single locus and six loci concatenation with Lophophyton gallinae CBS 300.52 as the root in MEGA v10.2. The concatenated multi-locus sequences totaled 3299 bp in length. For tub2, ITS, topI, tef-1α, topII, 60S L10 (L1) and tandem sequences, the best substitution models were K2, HKP + G, JC, K2, K2, JC, and K2 + I, respectively. The first 25% of trees were discarded as burn-in after 1000 bootstrap replicates. The evolutionary tree was further edited and annotated in iTtol (https://itol.embl.de/) and Adobe Illustrator 2020.

The R (v4.0.2) package phylogram/dendextend was used to visualize each locus evolutionary tree relative to the others. Matching taxa were connected using auxiliary lines to minimize the number of reticulations in any biconnected network components.

Analyses of nuclear haplotypes

Nucleotide polymorphism analysis of concatenated multilocus sequences was performed in DNASP v6.12.03 (Rozas et al. 2017). Population expansion was analyzed using Tajima's D and Fu's Fs neutrality tests. Populations of 185 sequences (without outgroups) were grouped by species type, origin, and MAT type in Arlequin (v3.5.2.2). Visualization of haploid networks was achieved in PopART v1.7 (Leigh et al. 2015) using the TCS network method.

Multispecies coalescent analyses

Based on the haplotype results, the corresponding numbered strains were selected as representatives in each grouping. The Beast (v2.6.7 StarBeast3) (Douglas et al. 2022) template was used to accomplish efficient multi-species joint inference using parallel gene tree operators. The dataset was partitioned by the 6 genes. HKY for substitution Model, Strict Clock molecular clock model, and Empirical for Frequencies were applied. The Yule model was used to set the tree species prior, and the clock rate priors for all loci were set to Exponential with a mean of 1.0. The chain length was set to 5,000,000, the tree was sampled every 5,000 steps, the tracelog was set to 5,000, and the screenlog was adjusted to 10,000. After running Beast v2.6.7, the results were analyzed with Tracer v1.7 (Rambaut et al. 2018) to assess the convergence of the model parameters. The burn-in percentage was set to 10, and the posterior probability (PP) was limited to 0.75 in TreeAnnotator v1.10 and the PP support for the specified branches was displayed in the DensiTree v2.6.7.

Phenotypic and physiological studies

The 185 strains were grown on Potato dextrose agar (PDA, Oxoid) at 28 °C for 14–21 days. Colony morphology was described with reference to the Atlas of Clinical Fungi, 4th edition (de Hoog et al. 2020), using a hexadecimal color code to indicate the colour of the colony (https://coolors.co/).

For the preparation of the Tween-80 agar (1% w/v Bacto Peptone, 0.5% w/v NaCl, 0.01% w/v CaCl₂, 1.5% w/v agar and 0.5% w/v Tween-80) Petri dish is referred to Tang et al. (2021), the diameter of the halo around the colony was recorded after incubation at 28 °C for 15 days. The Deoxyribonuclease (DNase) test was performed by first inoculating the isolates on DNase Test Agar (Oxoid, U.S.A.) and cultured at 28 °C for 15 days, after which 18.25% w/v HCL was poured on the surface of the medium and the diameter of the ring-clear zone was measured after 10 min (Cafarchia et al. 2012). The protocol of the keratin azure test was based on Scott et al. (2004). Each tube contained 5.5 ml of medium, which was divided into two layers, with the lower layer of 5 ml consisting of 2.5% w/v agar, 0.05% w/v MgSO₄·7H₂O, 0.05% w/v KCl, 0.05% w/v K₂HPO₄, 0.01% w/v ZnSO₄·7H₂O, 0.01% w/v FeSO₄·7H₂O, 0.003% w/v CuSO₄, and the upper layer of 0.5 ml consisting of 1% w/v agar, 0.05% w/v MgSO₄·7H₂O, 0.05% w/v KCl, 0.05% w/v K₂HPO₄, 0.01% w/v ZnSO₄·7H₂O, 0.01% w/v FeSO₄·7H₂O, 0.003% w/v CuSO₄, 0.4% w/v keratin azure. Mycelia were transferred to the upper medium and grown at 28 °C for 4 weeks, then the degree of blue of the lower layer represented the strength of the keratin decomposition. The urea hydrolysis tests were performed using Urea Agar Base (Oxoid, Hampshire, U.K.) and incubated at 28 °C for three days, after which the colour change of the medium was observed. The results of keratinase activity and urea hydrolysis Tween opacity test were given a score ranging from 5 (strongly positive) through 4 (positive), 3 (weak), 2 (weak/negative) and 1 (negative). The effect of M. canis complex on hair was studied by culturing the strain in distilled water containing 0.06% w/v yeast extract solution (BD, Bacto, U.S.A.) and blond children's hair. After 6 weeks of culturing at 28 °C, structural alterations in the hairs were observed under the microscope (Zeiss, Germany).

The strains were inoculated in triplicate on Sabouraud Dextrose Agar (SDA; Oxoid), PDA, and Malt Extract Agar (MEA; Oxoid) and cultured at 22, 28, and 37 °C for 14 days. Colony diameters were measured to assess temperature tolerance. Three angles were chosen for each colony to measure the diameter to take the average. At least three slides per Petri dish were made to record the spore abundance, septa number, length and width of macroconidia under the microscope. Principal component analysis of morphological characteristics was constructed in GraphPad Prism 9.0.

MAT idiomorph determination

To detect the presence of MAT1-1 or MAT1-2 regions in M. canis, the primer pairs Mc alpha F/Mc alpha R and Ab HMG F/Ab HMG R were used, respectively, while the primer pair HMG for 1/HMG rev 1 (Kosanke et al. 2018) was used to amplify the MAT1-2 region in M. audouinii and M. ferrugineum. Thermal cycling procedure and conditions are shown in the above molecular studies. The mating type of 183 strains was identified using the reference strains CBS 495.86 (MAT1-2) and CBS 496.86 (MAT1-1). On agarose gel plates, the results of the bands were determined, and numerous bands for each species were chosen for sequencing to verify the correctness of the amplified sequences.

Sexual crosses

Sexual crosses were performed on Niger Seed agar (5% w/v niger seed, 0.1% w/v glucose, 0.1% w/v yeast extract, 0.05% w/v MgSO₄·7H₂O, 0.1% w/v KH₂PO₄, 2% w/v agar) and Oatmeal Agar (2% w/v Oatmeal, 0.1% w/v yeast extract, 0.1% w/v NaNO₃, 0.1% w/v MgSO₄·7H₂O, 0.05% w/v KH₂PO₄, 1% w/v agar). Based on MAT idiomorphs determination, each isolate was tested for heterothallic crosses with CBS 495.86 (MAT1-2) or CBS 496.86 (MAT1-1), respectively. Small pieces of fresh, vigorously growing cultures were cut from the colonies and placed approximately 5 mm apart in the centre of the Petri dish. All crosses were incubated in the dark at 23 − 25 °C for 4 − 6 weeks and were periodically checked for cleistothecia or pseudocleistothecia.

Statistics

Statistical significance was conducted using SPSS Statistics v26. Tween-80 opacity, keratin azure, urea hydrolysis and tests were performed using the Kruskal–Wallis test with Dunn's multiple comparisons tests. The growth rate and characteristics of macroconidia were conducted using two-way nonparametric, Scheirer-Ray-Hare test. The resultant graphs were produced using GraphPad Prism 9.0. Each dot in the figure represents a sample, and each group’s standard error of the mean (SEM) is displayed. The value of p < 0.05 was considered statistically significant. The significance layout appears as follows: *p < 0.05; ** p < 0.01; ***p < 0.001; ****p < 0.0001.

RESULTS

Phylogeny of Microsporum canis complex

Phylogenetic analysis of the M. canis complex was performed by reconstructing an evolutionary tree combining six loci. Genealogical concordance of single strains in the different loci was visualized using a tanglegram (Fig. 1), showing consistency in the topology of sequences in supported clades. Clades containing the type strains of Microsporum canis, M. audouinii or M. ferrugineum, respectively, were identified in the tef-1α and topII datasets. Each clade was supported by bootstrap (bs) > 70%, except for CBS 495.86, which remained separate from M. audouinii and M. ferrugineum only based on the tub2 region. CBS 496.86 had significant sequence identity with the major genotype known as M. canis in the topI and tef-1α regions but had multiple nucleotide substitutions at all four other loci compared to the remaining strains. In topI, CBS 495.86 and CBS 496.86 shared the same nucleotide sequence and were classified in the M. canis group (Additional file 1). Linkage lines are drawn (Fig. 1) relative to the phylogeny of the tub2 gene fragment in the tanglegram. The relationships between the three groups were consistent across the six loci of bs-supported groupings, but some isolates deviated. In tub2, 60S L10 (L1), tef-1α, topII and topI, M. audouinii and M. ferrugineum clustered in the same group, whereas in the ITS region, M. canis and M. ferrugineum belonged to the same group. Both CBS 495.86 and CBS 496.86 were assigned to the M. canis group in the topI area, while the M. audouinii strain 204 was a member of the M. ferrugineum cluster.

Fig. 1

Tanglegram of six DNA regions. The diagram shows the evolutionary tree of each locus about the others. Matching taxa are connected using auxiliary lines of the same colour, the blue line for M. canis, the green line for M. ferrugineum and the orange line for M. audouinii. The grey line connects incongruent strains

Phylogenetic analysis of the above six concatenated loci revealed three supported main groups (Fig. 2). Microsporum canis emerged as a separate clade (99% bs), and the paraphyletic clades M. audouinii and M. ferrugineum each had > 90% bs support. CBS 495.86 deviated, being intermediate between M. canis and M. audouinii (65% bs). CBS 496.86 was individualized and affiliated with the M. canis clade (> 80% bs).

Fig. 2

Multilocus phylogeny of M. canis complex. A maximum likelihood tree was constructed using tub2, ITS, topI, tef-1α, topII, 60S L10 (L1) concatenated loci to analyse the multilocus phylogeny of the M. canis complex. The tree is annotated with the morphology, colony colour and reverse colour of 185 strains

Genetic diversity and population structure of the Microsporum canis complex

Nucleic acid diversity and population structure based on DNA sequences were used to determine the genotypic and evolutionary relationships of groups within the complex. The six DNA makers were concatenated to multiple sequence alignments of 3322 bp, which included 76 variable sites, 14 singleton variable sites, 62 parsimony informative sites, and 11 indel events.

The haplotype network comprised 12 multilocus genotypes (Fig. 3, Hap1-12), of which Hap1-7 genotypes belonged to M. canis, Hap8-11 to M. audouinii, and Hap12 to M. ferrugineum. Hap1 was the predominant genotype in M. canis. Compared to Hap1, Hap2 and Hap3 isolates had single nucleotide site substitutions in the topII and ITS sections, Hap4 exhibited a single substitution in the topI region, a single substitution of Hap5 was found in the topII area, and the ITS region of Hap6 included five extra substitutions. Hap7 (CBS 496.86) was genetically distinct from the remaining genotypes. Eight base substitutions occurred between CBS 496.86 and Hap1. Hap9 was the predominant genotype of M. audouinii, with one site substitution in topII being Hap10 and three nucleotide substitutions producing Hap11. Hap8 (CBS 495.86) was situated in the middle of the network diagram and had greater variability with Hap1 and Hap9. Microsporum ferrugineum had only one (Hap12) genotype and no intragroup variation.

Fig. 3

Haplotype network analysis based on multilocus data. The haplotype network map of the different species and host sources, the relationship between geographical origin and genotype, and the distribution of MAT idiotypes and sex crosses are shown. Strains of animal origin are marked in the haplotype network; the rest are strains of human origin

Analysis of nucleotide diversity and population genetic differentiation indicated that the complex has evolved stably over a long period (Pi = 0.00513, Hd = 0.571). Hap1 was a shared haplotype and an interior clade haplotype, implying that Hap1 might be the ancestral genotype within the entire complex. The complex has yet to undergo significant population expansion or experience a bottleneck event (Tajima's D = 0.781, Fu's Fs = 0.11943). The minimum recombination event Rm was 5. There has been significant genetic divergence (Fst > 0.25) between the three clades within the complex (Additional file 4).

Coalescence analysis

The species limits of the M. canis complex were re-examined by multi-species coalescence (MSC) analysis. StarBeast3 is a template for efficient Bayesian inference under the MSC model using the Markov chain Monte Carlo algorithm. The difference with the phylogenetic analysis of multiple sequence motifs is that MSC does not perform concatenation of the loci and proceeds directly to the analysis. Based on the 12 haplotypes, we selected representative sequences from each Hap group: Hap1 included M. canis 10, 50, 65, 68 and 228, Hap2 included M. canis 138 and 244, Hap3 was M. canis 3, Hap4 included M. canis 9, 12 and 21, Hap5 was M. canis 73, Hap6 included M. canis 167, 168, 213 and 224, Hap7 was CBS 496.86, Hap8 was CBS 495.86, Hap9 was M. audouinii 178, Hap10 included M. audouinii 88, 93 and 99, Hap11 was M. audouinii 204, and Hap12 included M. ferrugineum 19 and 190. Trimming the sequences to equal length resulted in a total of 25 individuals with complex genotypes. Lophophyton gallinae was included as the root. Figure 4A shows the tree for 1000 random calculations, and Fig. 4B displays the tree with the highest posterior probability product of nodes. Posterior probability (PP) support greater than 0.7 was shown next to the branches. In the MSC tree, M. ferrugineum is a separate taxonomic entity and M. canis and M. audouinii are delimited in the same branch. Strain CBS 495.86 and remaining M. audouinii strains are considered to be paraphyletic (PP = 1). Both Hap9 and Hap10 are monophyletic (PP = 1) strain sequences that are reciprocal to one another. The Hap6 and Hap7 genotypes and the Hap1–5 genotypes constitute branches that are parallel to each other (PP = 1). Of the Hap1–5 genotypes, Hap2 differs from the other four (PP = 1) and the remaining typing is not supported by sufficient PP (Fig. 4).

Fig. 4

Multispecies coalescent tree constructed from six loci. Posterior probability support of > 75% is shown next to the node

Geography and host origin

We further analyzed the geographic and host origins of the different Hap genotypes to understand the distribution and prevalence of members of the M. canis complex (Fig. 3). Hap1 (n = 118) included isolates from Belgium, Brazil, China, Germany, Guinee, Italy, Slovenia, and the U.S.A. Of these strains, 47 were from animals (both symptomatic and asymptomatic), and the remaining strains (n = 71) were from humans. In Hap1, concerning the animal origins, 35 isolates were from cats, 6 from dogs, 1 from a cheetah kept in captivity and 1 from the environment; of the human-derived isolates, 10 strains were isolated from smooth skin, and 51 from the scalp, while for some isolates no data were available. Hap2 (n = 4) originated from Europe (Belgium, Germany, Italy, Slovenia), Hap3 (n = 1) and Hap4 (n = 1) were from Italy and Belgium, respectively, and Hap4 was only found in China. Hap6 (n = 6) was mainly from Brazil, followed by the U.S.A. and Germany. The animal isolates in Hap2 were from dogs, and those in Hap3 and Hap6 were all from cats. Thirty-six strains of M. audouinii and 14 strains of M. ferrugineum were isolated from humans. Hap9 (n = 16) was from the U.S.A., Germany, and Senegal; Hap10 (n = 19) was mainly isolated in the U.S.A., followed by African countries (Congo, Cameroun, Guinee). Hap12, M. ferrugineum, originated in China and Germany.

Physiology

For these tests, we divided the main groups defined above into six groups for comparison, viz. M. canis cat (cat-derived M. canis isolates), M. canis dog (dog-derived M. canis isolates), M. canis rabbit (rabbit-derived M. canis isolates), M. canis human (human-derived isolates), M. audouinii and M. ferrugineum groups.

Compared to isolates of human origin, all three animal groups showed higher lipolytic capacity, especially in the M. canis cat group (p < 0.001), but there were no statistically significant differences among the cat, dog and rabbit origins. The M. audouinii and M. canis human groups had similar lipase catabolic abilities, with 72.7% and 75.3% positivity, respectively. Only 14.2% of M. ferrugineum strains were positive in the Tween-80 opacity test. The keratinolytic capacity of the four M. canis groups was higher than that of both the M. audouinii and M. ferrugineum groups (p < 0.05), and no difference was observed in M. canis from human and animal origin. The M. canis rabbit group showed the highest scores in keratinase activity and urea hydrolysis tests compared to M. ferrugineum (p < 0.05). Urea catabolic capacity did not differ between the M. canis animal, M. canis human, and M. audouinii groups, but all were stronger than in M. ferrugineum. The DNase activity of M. ferrugineum was the strongest among the six groups, higher than that of M. canis (p < 0.01) (Fig. 5A).

Fig. 5

Physiological characteristics of M. canis complex. A Results of scores for the Tween-80 opacity test, keratin azure test, urea hydrolysis tests and in the M. canis animal group, M. canis human group, M. audouinii and M. ferrugineum group; B Hair changes in four groups. The mean + SEM of each group is indicated. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Hair perforations perpendicular to the hair shaft were not observed. All three species caused ectothrix, but hairs in the M. canis group showed different levels of damage at 4–6 weeks, which appeared after 8–10 weeks in M. audouinii and M. ferrugineum cultures. The hairs showed mild damage of brush-like changes in cuticle scaling, moderate damage in the form of continuous or interrupted medullae, and finally severe hair breakage (Additional File 3: Fig S2). In all M. canis animal groups, 56.5% of the hairs showed severe damage to the hair structure with breakage, 30.4% were dominated by mild hair damage, and 13% were unchanged. In the M. canis rabbit, cat, and dog groups, 75%, 42.9%, and 28.5% of the strains showed severe hair breakage, respectively. This manifestation was followed by 40.9%, 50% and 9% in the M. canis human group. In the M. audouinii and M. ferrugineum groups, 20% and 10.5% of the hairs were broken, 40% and 42.1%, respectively of the hairs remained structurally normal, and the others showed only mild damage.

The fastest growth was observed at 28 °C, followed by 22 °C (Fig. 6A). At 37 °C, only a few M. canis strains (n = 7) could grow. Microsporum audouinii and M. ferrugineum did not grow at 37 °C. M. canis showed the fastest expansion, with no difference between strains of animal and human origin, followed by M. audouinii, while M. ferrugineum grew very slowly. Growth rates were not statistically correlated with haplotypes of M. canis. Microsporum audouinii of Hap9 grew faster than M. audouinii of Hap10 (p < 0.05). Hap11 and Hap12 had the lowest growth rates under all conditions (p < 0.01).

Fig. 6

Morphological characteristics of M. canis complex at different temperatures. A Growth rate of isolates within the whole complex based on 12 genotypes, with the area shaded below the fold line indicating the magnitude of the rate; B Length, width and septa number of macroconidia in M. canis and M. audouinii under incubation at 22 and 28 °C; C Principal component analysis of morphological characteristics at 22 °C and 28 °C of incubation, including growth rate, macroconidia abundance, length, width and the septa number of macroconidia. The mean + SEM of each group is indicated. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

We further compared the state of development of macroconidia on PDA at three temperatures, including the length, width, and septa of macroconidia. These indicators showed significant differences at 22 °C of cultivation in the animal/human-derived M. canis isolates and M. audouinii isolates (Fig. 6B). At 22 °C, the length, width, and the septa number of macroconidia of M. canis were much higher than at 28 °C, especially in the animal-derived M. canis isolates (Fig. 6B). Microsporum ferrugineum 191 strain was observed to have very few macroconidia which were small and without septa, upon incubation at 22 °C, while all other strains of this species were entire without conidia. Principal component analysis (PCA) made use of a mixed model of growth rate, macroconidia abundance, length, width, and septation (Fig. 6C). The results of PCA confirmed the above differences and the principal component 1 extended X-axis distribution could separate M. canis and M. audouinii micromorphologically at 22 °C incubation, but not at 28 °C.

Macromorphology

The clusters were grouped according to haplotype diversity, and no significant morphological differences within the groups were found, but differences between groups were observed (Additional File 2: Fig S1). When strains were cultured on PDA at 28 °C for 14–21 days, M. canis colonies showed white to cream, cottony to velvety, raised, radially furrowed, and granular morphology. Fifty-two isolates of M. canis on PDA had a granular-powdered surface; 26 were of animal origin. The reverse was white to cream in 73.68% of the human-derived M. canis isolates and 48.72% of the animal-derived isolates (Fig. 2). The isolates M. canis 13, 18, 60, 67, 69, 150, 156 and 248 showed dysgonic colony. The dysgonic type colony is similar to that of M. ferrugineum, filamentous or heaped brown thallus. Macroconidia were usually absent, but microconidia were still visible.

Colonies of M. ferrugineum were white to yellow, filamentous, flat, with hyphae appressed to and submerged in the medium; two strains (2/14) had a cottony surface. The reverse colour of M. ferrugineum was predominantly yellow (78.57%) (Fig. 2).

CBS 495.86 showed a white, velvety-like colony, with cream-coloured reverse, and produced numerous macroconidia and densely associated microconidia (Fig. 2). Colonies of M. audouinii were white to brown, being predominantly downy to densely suede-like; in 17/36 strains the colonies were white in the centre, raised and downy, with the edges extending into the medium, and the hyphae were white to brown. The reverse colour was yellow to brown (96.3%), white and cream colours being uncommon.

Mating behaviour

We examined the distribution of MAT idiomorphs in the complex to examine the selection of MAT gametes on different hosts. Of the Hap1–7 genotypes of M. canis, which originated from different countries and different hosts, mating types are all MAT1-1 idiomorph. In contrast, Hap8–11 of M. audouinii and Hap12 of M. ferrugineum, from human tinea capitis, are MAT1-2 idiotypes (Fig. 3).

After 8 weeks in culture, positive responses were noted in Hap1 × Hap8, Hap6 × Hap8, Hap9 × Hap7 and Hap10 × Hap7 in a total of 11 isolates (Fig. 3), i.e., in M. canis 44, 74, 112, 156, 213, 246, M. audouinii 99, 169, 177, 186 and 211, respectively. A small amount of white powdery thallus was visible at the junction on the plates. Microscopically, sterile cleistothecia were observed, globose and light brown. The peridium of M. audouinii 177 included the peridial hyphae, and the peridium borders of the other strains were clear. However, no mature ascospores were observed in these crosses (Fig. 7).

Fig. 7

Cleistothecia production of M. canis complex. A–B Cleistothecia were yielded by M. canis 74 (Hap1) × CBS 495.86; C–D Cleistothecia were yielded by M. canis 213 (Hap6) × CBS 495.86; E Cleistothecia were yielded by M. audouinii 186 (Hap9) × CBS 496.86; F Cleistothecia were yielded by M. audouinii 177 (Hap10) × CBS 496.86 with irregular peridial hyphae visible of the peridium. Scale bars = 10 μm

DISCUSSION

The variability of phenotypic and clinical parameters in the Microsporum canis complex urges critical determination of species identity and barcoding markers and investigate the transition from animal to human hosts. Based on the molecular overview of dermatophytes by de Hoog et al. (2017), we analysed ITS, tub2 and 60S L10 (L1) supplemented with tef-1α, topI and topII molecular markers. In our results, the tef-1α and topII proved to be optimal to distinguish the three species, and the ITS and topII offered the greatest nucleotide diversity. Ciesielska et al. (2020) proposed velB as a marker to differentiate M. canis from other dermatophytes, but M. ferrugineum was not mentioned. Rezaei-Matehkolaei et al. (2012) suggested that tef-1α may be used for rapidly screening of Iranian M. ferrugineum isolates. The novel topI gene showed great promise for ascomycetes through the Pfam approach (Stielow et al. 2015) and has not been explored previously in M. canis. Shamsizadeh et al. (2020) found that the number of polymorphic sites in the topII region of dermatophytes was similar to that of ITS.

The topology of the phylogenetic tree suggested an evolutionary trajectory from zoophilic to anthropophilic species. The haplotype network combined with species origin analysis and multispecies coalescence analysis was applied to define the genetic structure of the complex further. A total of 12 haplotypes were recognized, Hap1–7 being M. canis, 8–11 being M. audouinii and 12 being M. ferrugineum. This demonstrates that the observed phenotypic differences are more than just a matter of expression. Hap1 is the prevalent M. canis genotype, and it is widely dispersed on four continents: Europe, Asia, America, and Africa. Hap2, 3, and 5 are all native to Europe, while Hap4 is found only in Asia. However, the position of Hap2–4 in the multilocus evolutionary tree lacks sufficient bs (70%) to support a separate subgroup. Hap6 has two subgroups (bs > 90%), primarily containing North and South American isolates. Within the M. canis group, there appears to be no correlation between host origin and genotype. The human isolates in Hap1 most likely concern recent direct infections from animals. Microsporum canis is known to cause small, self-limiting epidemics, e.g., in school children (Brosh-Nissimov et al. 2018; Hermoso de Mendoza et al. 2010; Subelj et al. 2014), indicating that zoophilic strains can reside on human hosts, at least temporarily. It may be surmised that host adaption in these strains remains inadequate (Sharma et al. 2007).

Subgroups Hap9 and 10 were supported by 65% bs in the phylogenetic tree. European isolates mainly have the Hap9 genotype, while African isolates predominantly Hap10. North American strains are present in both but are more prevalent in Hap10; possibly, North American strains may result from recent African immigration. During the last 20 years, an increase in the prevalence of M. audouinii infections has been reported worldwide on all continents, particularly in Belgium, France and Switzerland, and there in cities heavily populated by immigrants from African countries (Brito-Santos et al. 2017; Sacheli et al. 2021). Microsporum ferrugineum, the most derived clonal entity that lacks intraspecific diversity and has only a single genotype, Hap12, is limited geographically. A study by Zhan et al. (2015) on 60 years of change in tinea capitis in China showed that M. ferrugineum remained highly endemic in Xinjiang and Yunnan in an era of great epidemiological change in tinea capitis. Both mating tester strains, CBS 496.86 and CBS 495.86, diverge from all other isolates in the haplotype network. CBS 496.86 and CBS 495.86 derived from the F1 progeny of two feline isolates (VUT 73015 × VUT 74001). As offspring of crosses, sexual reproduction always leads to higher population diversity; they acquire different genotypes from the parents in the process of genetic recombination, thus causing positional deviations.

Microsporum canis was MAT1-1 in all our strains, while Microsporum audouinii and M. ferrugineum were all MAT1-2 idiomorph. Mating is still possible in the complex, but it is challenging to produce fertile offspring. The distribution of the MAT idiomorphs also indicated the possibility of animal-to-human host shift. Anthropophilic and zoophilic strains have opposite sex distributions, with the MAT1-2 gamete apparently being better adapted to humans. Although clonal reproduction facilitates the spread of the fungus, without the adaptability and plasticity achieved through genetic recombination, the organism would not be able to cope with long-term changes in host resistance or other natural or anthropogenic environmental changes (Drenth et al. 2019). The extinction of a MAT gamete in dermatophyte populations may be caused by the preferential transmission of strains exhibiting favourable combinations of alleles associated with higher virulence/transmission potential (Lee et al. 2010). Loss of mating ability may be triggered by epigenetic factors, but the underlying mechanism is unknown. An association of different MAT idiomorphs and reproducing sexually with virulence, ecology and pathogenicity has been demonstrated in a variety of dermatophytes and non-dermatophytes (Cerikçioğlu 2009; Nielsen et al. 2007; Persinoti et al. 2018). Only a single MAT idiomorph was identified per isolate, but thus far, sexual reproduction with fertile gymnothecia has been observed exclusively in the tester strains, which are molecular mavericks. The sexual MAT1-1 and MAT1-2 M. canis strains may be the ancestors of the entire complex, with M. ferrugineum as the most recent, still invariant human-adapted clone. Gradually, two distinct anthropophilic species evolved because of the human host adaptation of one sexual type (MAT1-2), with MAT1-1 remaining prevalent in the cat.

Phylogeny, population structure analyses and mating idiomorph distributions reveal a transition in the M. canis complex from zoophilic to anthropophilic lifestyles. The host shift is reflected in differences in physicochemical capabilities, resulting in reduced virulence in the anthropophilic species. The statistical significance in ecologically relevant parameters suggests gradual host adaptation, which is more than just phenotypic in nature. The non-sporulating phenotype of M. ferrugineum occurs occasionally in cat-derived M. canis and is then known as ‘dysgonic’; it seems that this phenotype is more suitable for infection of the human host and thus might be selected during evolution. Animal-derived M. canis isolates caused more direct hair damage than the isolates derived from humans and compared to M. canis, M. audouinii and M. ferrugineum caused structural damage to the hair of blond children twice as slowly and with lesser symptoms. Viani et al. (2001) showed that M. canis from symptomatic dogs and cats exhibited statistically higher keratinase activity than those isolated from asymptomatic dogs and cats and caused acute inflammatory responses in guinea pigs. Interestingly, Cafarchia et al. (2012) reported that M. canis isolated from healthy rabbits had higher keratinase activity than isolates from rabbits with skin lesions, whereas the latter had greater lipase activity. In our results, the lipolytic capacity of M. canis isolated from animals was higher than that of those isolated from humans. Lipase catabolism and keratinolytic capability of M. ferrugineum and M. audouinii were significantly lower than those of M. canis, matching with hair decomposition in fur by the latter species. The urease activity of dermatophytes was highly variable, as proven by the similar urea hydrolysis ability of M. audouinii and M. canis. Notably, we found that the keratinolytic and ureolytic capacities of M. canis isolated from rabbits may be stronger than those of strains isolated from cats and dogs. Microsporum ferrugineum scored the lowest of all four groups regarding lipase and keratolytic and urea hydrolytic capacities. The low virulence of M. ferrugineum may result in persistence provoking mild inflammatory response, leading to a peaceful coexistence of the anthropophilic fungus with its host, similar to T. rubrum (Zhan et al. 2018). DNase activity has been shown to facilitate the evasion of the innate immune system by Paracoccidioides (Zonta et al. 2020), Cryptococcus (Sánchez et al. 2010) and Trichosporon (Bentubo et al. 2014). It may have a similar role in M. ferrugineum.

DNA hydrolysis has been observed in other anthropophilic dermatophytes. López-Martínez et al. (1994) showed that all 47 analyzed T. rubrum strains produced DNase. The function of DNAse in M. canis is unknown; it does not vary across symptomatic and asymptomatic animal strains (López-Martínez et al. 1994). The composition of the host’s skin may impact the pathogenic effect of each dermatophyte in a given host (Cafarchia et al. 2012). The optimal temperature for in vitro growth rate of the M. canis complex was 28 °C, but at 22 °C, sporulation and macroconidial size of animal-derived M. canis were greater than those of M. canis and M. audouinii of human origin; at 28 °C, no difference was observed. It is unclear why temperature changes affect the formation of macroconidia, which may be related to changes in virulence and the shift of energy to the host during infection. A lower temperature may also indicate a distance to the host’s body increasing the need for dispersal by conidia. For example, Song et al. (pers. comm.) identified changes in gene expression that accompany the shift from aerobic residence in cat fur to invasion of hairless human skin by analyzing the genome of M. canis strains that cause deep invasion. In summary, in terms of physicochemical properties, the characteristics accompanying the animal to human shift are loss of sporulation in human tissue compared to the residence in fur, higher pigment (xanthomegnin) production, lower keratinolysis, and lower lipolysis.

There are some limitations in the present study, including that the animal-derived M. canis strains were mostly isolated from cats and dogs and were restricted to Europe and South America. Additionally, the study did not include any M. canis from horses, we cannot ignore the possibility that this is a transitional genetic event in the complex similar to cat-to-human adaptation. Consequently, the host shift from animal fur to naked human skin in the M. canis complex shows that the dysgonic phenotype is more suitable for human infection, but the question of whether this is an evolution requires a more in-depth study of wild and domesticated animals. To further explore the selective role and virulence differences between M. canis evolution in animals and humans, a more diverse panel of animal isolates is required.

CONCLUSIONS

Phylogeny, population genetics and multispecies coalescent analyses clearly distinguished the three species, not only phenotypically but also genetically. Combined with physicochemical properties and MAT idiomorph distribution analysis, we could reconstruct the host shift to the human within the M. canis complex. Microsporum canis (MAT1-1) was by far the most dominant population, still maintaining high virulence and not adapted to humans; M. audouinii (MAT1-2) and M. ferrugineum (MAT1-2) are widespread in humans, with slow growth, progressive loss of conidia, weak lipolytic and keratolytic capacity, but increased DNA hydrolysis. Mating still occurs in the complex, exemplified by MAT1-1 and MAT1-2 tester strains. Microsporum audouinii appears to be intermediate in all respects, while M. ferrugineum is the most invariable compared with M. canis.