Phylogenetic Analysis Reveals a High Prevalence of Sporothrix brasiliensis in Feline Sporotrichosis Outbreaks

1Mycotic diseases, particularly those caused by dimorphic fungi such as Sporothrix, can be considered as an emerging threat to various species of animals. Upon introduction of propagules into the mammalian host, the fungus undergoes a thermodimorphic transition to a yeast-like phase, leading to infections varying between fixed localized cutaneous lesions to severe, disseminated sporotrichosis.

2The first connection between Sporothrix and animals was made by Lutz and Splendore. Since then sporotrichosis has been reported in dogs, cats, horses, cows, camels, dolphins, goats, mules, birds, pigs, rats, and armadillos, as well as in humans. However, the cat is the animal species most affected by this mycosis. Over the last two decades, Brazil has experienced its largest epidemic of sporotrichosis due to zoonotic transmission, whereby cats were pointed out as key susceptible host. The zoonotic potential of infected cats has been demonstrated by the isolation of S. schenckii s.l. from feline skin lesions, nasal, oral cavities, and claw fragments,.

3In contrast to the classical route of infection by Sporothrix, where soil and plant material loaded with saprophytic hyphae of the pathogen were the source of contamination, transmission of Sporothrix spp. by cats to other cats and to humans via direct inoculation of yeast cells represents an alternative and a successful type of dispersal of the disease. The yeast form is more virulent than the mycelial form,. Transmission of yeast cells may enhance the appearance of more severe forms of the disease.

4Until recently, S. schenckii was considered to be the only species causing sporotrichosis. The infection has a worldwide distribution, mainly in tropical and subtropical countries –. The most common clinical manifestations in humans are the lymphocutaneous and fixed forms, but other clinical types, such as a disseminated form, may also occur,, partly depending on the immune status of the host.

5Multilocus sequencing combined with morphological and physiological data support the separation of at least four distinct Sporothrix species within the S. schenckii complex, uniting the species with high pathogenic potential to mammals. The original taxon S. schenckii (Clades IIa and IIb) and the novel species S. brasiliensis (Clade I), S. globosa (Clade III), and S. luriei (Clade VI) todays are referred to as the S. schenckii complex, while the mildly pathogenic species S. mexicana (Clade IV) takes a remote position near the environmental species S. pallida, –. The Sporothrix species differ in their pathogenic potential for mammals,, their geographical distribution,,,, and in their sensitivity to antifungal therapy. All species have been reported from Brazil.

6Endemic areas of sporotrichosis in Brazil are characterized by poor sanitation, substandard housing and little or no access to health services – a challenge to control and eradication of the disease. The oldest outbreaks of sporotrichosis among humans and cats have been reported in the states of Rio de Janeiro,, and Rio Grande do Sul,. Delayed diagnosis and treatment in cats may lead to a rapid spread of the disease through the community members. The increase in the number of cases in cats is followed by higher numbers of human cases, which constitutes a serious public health problem.

7Despite the increasing frequency and severity of cases, the eco-epidemiology of feline sporotrichosis in Brazil is still unknown. The aim of the present study was to determine the distribution and prevalence of Sporothrix species among naturally infected felines using phenotypic and molecular phylogenetic approaches.

8Thirty three (33) Sporothrix isolates from Rio de Janeiro, RJ (n = 15); Rio Grande do Sul, RS (n = 10); Paraná, PR (n = 4); São Paulo, SP (n = 3) and Minas Gerais, MG (n = 1) were obtained from lesions of cats and dogs with sporotrichosis (skin or mucosa lesions) (Fig. 1). Fungal cells were recovered directly from lesions and cultured on Mycosel agar (Difco Laboratories, Detroit, Mich.). Suspected colonies were subcultured on potato dextrose agar (Difco Laboratories, Detroit, Mich.) at room temperature. Isolates were identified phenotypically as S. schenckii s.l. As a control, human clinical isolates (n = 66) inside and outside the Brazilian feline outbreaks areas were included in the study (Table 1).

9Morphological identification of cultures was performed according to Marimon et al., including vegetative growth on PDA media at 30, 35, 37 and 40°C, colony colors on corn meal agar (Difco Laboratories, Detroit, Mich.), assimilation profiles of raffinose, ribitol and sucrose, and microscopic morphology in vitro. Growth at different temperatures was measured according to Mesa-Arango et al.: the percent growth inhibition (GI) was calculated at 37°C by the following formula [(colony diameter at 30°C – colony diameter at 37°C)/colony diameter at 30°C]×100. The GI was evaluated by analysis of variance/Tukey test using the GraphPad (GraphPad Prism v. 5.00 for Windows, San Diego California USA, www.graphpad.com), considering statistically significant when p<0.05. Observed data were used for taxonomic characterization applying the dichotomous key to species of the S. schenckii complex proposed by Marimon et al..

10For molecular analysis, genomic DNA was extracted and purified directly from mycelial colonies following the Fast DNA kit protocol (MP Biomedicals, Vista, CA, USA) with the homogenization step repeated three times with a Precellys 24 instrument (Bertin, Montigny le Bretonneux, France). DNA was quantified with NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA). The calmodulin (CAL) locus region was amplified directly from the genomic DNA by PCR, as described by O'Donnell et al., using the degenerate primers CL1 (5′-GAR TWC AAG GAG GCC TTC TC-3′) and CL2A (5′-TTT TTG CAT CAT GAG TTG GAC-3′), which generated an 800-bp amplicon corresponding to exons 3 through 5. The translation elongation factor-1 alpha (EF1α) locus region was amplified using the newly designed primers EF1-F (5′-CTG AGG CTC GTT ACC AGG AG-3′) and EF1-R (5′-CGA CTT GAT GAC ACC GAC AG-3′) which amplified an 850-bp fragment, covering the last exon of this gene, matching the same region evaluated by the consortium Assembling the Fungal Tree of Life (AFTOL).

11Thermal cycling conditions were as follows: one cycle of 5 min at 95°C, followed by 35 cycles of 1 min at 95°C, 1 min at 60°C (CAL) or 57°C (EF1α) and 1 min at 72°C, followed by one cycle of 10 min at 72°C.

12Amplified products were gel-purified with the Wizard® SV Gel and PCR Clean-Up System (Promega, USA) following the manufacturer instructions. DNA samples were completely sequenced with an ABI 3730 DNA Analyser (Applied Biosystems, Foster City, CA, USA) using BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems). The fragments were sequenced on both strands to increase the quality of sequence data and assembled into single sequences via CAP3 using bases with quality of phred ≥30. Sequences were aligned with MAFFT v. 5.667 and retrieved alignments were manually edited in order to avoid mis-paired bases.

13Calmodulin sequences deposited at GenBank belonging to the clades of clinical importance in the S. schenckii complex (Table 1) were collected and included in the present alignment as reference strains for the phylogenetic distribution. We choose the saprophytic fungus Grosmannia serpens (Ophiostomataceae), CBS 141.36 as outgroup for CAL analysis. All Sporothrix EF1α sequences used in the phylogenetic analysis were generated in this study (Table 1). The outgroup for the EF1α analysis included the saprophytic fungus Ophiostoma piliferum, CBS 158.74 (AFTOL-ID 910). This species was chosen because the genus Ophiostoma (Ophiostomataceae) is considered a close related group to Sporothrix species.

14Phylogenetic analyses were carried out using Neighbor-joining, Maximum Likelihood and Bayesian methods. Neighbor-Joining and Maximum Likelihood trees were constructed using MEGA 5 software and 1000 bootstrap replicates were used to estimate confidence values for individual clades and are shown next to the branches. The evolutionary distances were computed using the Tamura 3-parameters method and the rate variation among sites was modeled with a gamma distribution (shape parameter = 1). For Bayesian analysis by Markov Chain Monte Carlo (MCMC), two independent analyses of four chains each as default were initiated from a random tree and processed for 1.000.000 generations; sample trees were retrieved every 1000 generations. Log-likelihood scores were plotted against its generation number in order to evaluate convergence; samples collected prior to “burn-in” (25%) were ignored. The remaining samples were used to determine the distribution of posterior probability values. The posterior probabilities values of generated clades and overall topology of each replicate were compared in order to verify that each consensus tree converged on a similar phylogeny. Phylograms generated by Bayesian analysis were used to represent the phylogenetic distribution and were produced with the help of the Figtree 1.0 software (available at http://tree.bio.ed.ac.uk/software/figtree/).

15Evolutionary relationships at the intraspecific level were evaluated using haplotype networks in order to visualize differences and diversity among S. brasiliensis sequence data. The number and diversity of CAL and EF1α haplotypes were estimated using the software DNAsp v5.0. Gaps and missing data were excluded in the calculation. Median-joining networks for the CAL and EF1α dataset were obtained and visualized using the software Network 4.610 (available at www.fluxus-engineering.com).

16Evidence of recombination in S. brasiliensis population isolated from animals and humans samples was inferred by the split decomposition method, implemented by the Splitstrees4 software, version 4.8 which is used to identify branching ambiguities attributable to recombination events. The presence of recombination networks can be detected by bridges between members of the genetically isolated groups. Each isolated group will have an independent branch, showing that it does not share genetic material with the others. This analysis allowed the assessment of recombination possibilities within and between the seven phylogenetic groups considered.

17The PHI-test incorporated in the SplitsTree software was used to test signals of recombination (p<0.05, significant evidence of recombination). The test is proven to be a robust calculation and no previous knowledge about population history, recombination rate, mutation rate and rate heterogeneity across sites is necessary. Although large splits in networks do not necessarily imply recombination, split decomposition networks in conjunction with the PHI-test can easily detect which sequences in a given data set contribute the most to the recombination signal. The PHI-test is repeated after possible recombinants are deleted from the alignment until p>0.05 (no evidence of recombination). Also, DNAsp v5.1 was used to evaluate minimum number of recombination events in the history and haplotypic diversity of S. brasiliensis population. The software computes the recombination parameter R = 4Nr, where N is the population size and r is the recombination rate per sequence -or between adjacent sites.

18The animals included in this study were examined by a veterinarian with experience in small animal internal medicine. The procedures performed in these animals were approved by the Ethics in Research Committee (CEUA) of the FIOCRUZ, Rio de Janeiro, Brazil, under license number L-041/06.

19Our study included indoor and feral cats from five different geographic regions in Brazil (RJ, RS, MG, SP, and PR). Diagnosis of sporotrichosis was performed by the clinical evaluation of skin lesions and confirmed by isolation of the pathogen. The suspected colonies of Sporothrix species were grown on Mycosel agar until purification of the pathogen. The fungus was easily isolated from material from the nasal, oral mucosa and skin lesions. Lesions in the cephalic region and/or respiratory tract were observed in most of the animals (Fig. 2).

20Phenotypic characterization, i.e. growth at various temperatures, macroscopic and microscopic features, and carbohydrate assimilation, yielded data similar to those found for the reference strains of S. brasiliensis (CBS 120339) and S. schenckii (CBS 359.36) reported by Marimon et al.. Among the 33 strains of Sporothrix isolated from cats (n = 31) and dogs (n = 2) from different geographic regions of Brazil, 32 belonged to S. brasiliensis (96.9%) and 1 to S. schenckii (3%). These phenotypic results showed that S. brasiliensis is highly prevalent among cats with sporotrichosis. The two isolates recovered of canine sporotrichosis (CBS 132994 and CBS 133004 from RS and SP, respectively) were identified as S. brasiliensis.

21Using CL1 and CL2A primers we amplified 800 bp of the CAL locus. The complete alignment included 100 strains. Aligned sequences of CAL were 727 bp long, including 366 invariable characters, 214 variable parsimony-informative (29.43%), and 125 singletons. Comparison with sequences available at GenBank revealed a match of 99–100% with the type strain of S. brasiliensis (CBS 120339, AM116899) corroborating our phenotypic data. The single isolate of S. schenckii (CBS 132961) matched 99% with the S. schenckii s. str. strain (FMR 8678, AM117446) from Argentina.

22Phylogenetic analysis of isolates from cats and dogs revealed that S. brasiliensis is the prevalent species (32/33); only a single isolate clustered with S. schenckii s. str. The clade of pathogenic Sporothrix species was well supported with high bootstrap and posterior probability values. The S. brasiliensis isolates recovered from animal sources clustered in a single branch together with clinical isolates, indicating that they belonged to the same genotypes and confirming that the disease is transmitted by cats. A cryptic branch was observed in the S. brasiliensis clade composed of the isolates Ss27, Ss125, Ss128, CBS 132997, CBS 132999, CBS 133000, CBS 133001 CBS 133002 and CBS 133003, supported by bootstrap and posterior probabilities values (64/66/1) (Fig. 3A).

23Sporothrix brasiliensis presented low genetic diversity compared to its sister taxon S. schenckii when CAL was used as a marker. Elongation factor (EF1α) was used as marker to assess the genetic diversity in the species. All isolates presented similar fragments of 850 bp of the EF1α locus which were amplified and sequenced with primers EF1-F and EF1-R. Aligned sequences of EF1α were 707 bp long, including 639 invariable characters, 34 variable parsimony-informative (5.08%), and 33 singletons. The 100 OTUs were distributed into 7 main groups (Fig. 3B), which were congruent with the CAL phylogeny.

24Judging from the EF1α dataset, the S. brasiliensis isolates recovered from animal sources in RJ and RS clustered in two branches with human clinical isolates from the same states, indicating two epidemics with distinct genotypes are concerned (Fig. 3B). Sporothrix brasiliensis presented low genetic diversity in EF1α, in accordance with results obtained for the CAL locus.

25The haplotype diversity of S. brasiliensis species was assessed using the DNASp software. Only 7 haplotypes for CAL (Fig. 4A) and 3 haplotypes for EF1α (Fig. 4B) were found. The low values of haplotype (HdCAL = 0.36 and HdEF1α = 0.37) and nucleotide diversities (πCAL = 0.00152 and πEF1α = 0.00062) lead us to hypothesize that this species is clonal (Table S1). Geographical separation between the RJ and RS epidemics for the EF1α locus was clear. The median-joining network based on the EF1α haplotype showed an intraspecific separation (Fig. 4B, haplotypes H11 and H12) resulting from a nucleotide transition from A to G, between isolates from RJ and RS epidemics (Table S2). The average divergence between S. brasiliensis and its sister species S. schenckii is much higher, suggesting that the species experienced different evolutionary processes.

26Recombination analysis of S. brasiliensis was first assessed by split decomposition method and no networks linking different isolates were observed in both datasets (Fig. 5), in agreement with the concept of clonal species. Also PHI-test analysis showed no evidence of recombination (pCAL = 0.757 and pEF1α = 0.903), and no recombination events were detected by DNAsp5 software. Taken together, these analyses indicated that S. brasiliensis is a clonal species.

27Aiming to evaluate possible phenotypic characteristics that explain the success of this pathogen adaptation to the feline host we evaluated the thermal resistance of strains of clinical interest (human and animal) and environmental strains. Strains of S. brasiliensis from feline origin (n = 30) showed highest temperature tolerance, being inhibited 77.1±6.32% on average at 37°C (Fig. 6). The group differed statistically from other species evaluated herein (S. schenckii s. str., S. globosa, and S. mexicana), suggesting that this factor may confer advantage during the process of infection in the feline host.

28Supplementary information reported in this section is complementary to the results and describe the genetic diversity of the Sporothrix isolates.

29Epidemiology of fungal infections can be influenced by several factors, including: (i) biological factors such as fungal virulence and host resistance, (ii) ecological factors such as temperature, atmospheric humidity, ultraviolet radiation, geological conditions, and inter-relationships with other living beings, and (iii) socio-economic factors such as poverty, sanitation, clothing, profession, prophylactic habits and population migrations. In the Brazilian epidemic of feline sporotrichosis a combination of a highly virulent fungus and a susceptible host coupled to low sanitary conditions in the suburbs has made the state of RJ a highly endemic area of this mycosis among animals and humans. The epidemic proportions are noted only since the last two decades.

30Little is known about the eco-epidemiology of feline sporotrichosis and its impact on the epidemiology of human sporotrichosis. Cats play a significant role in outbreak areas of sporotrichosis such as RJ and RS. Classically, humans can acquire sporotrichosis by cat scratches or bites, the reason why cats are considered important source of infection in the spread of the disease. In our study we found that S. brasiliensis is the prevalent etiological agent of feline sporotrichosis in Brazil. Among cats, S. brasiliensis was identified in a total of 96.9% of the samples, by isolation of the pathogen from lesions and posterior phenotypic and molecular characterization.

31Interestingly, a correlation between cat outbreaks and prevalence of S. brasiliensis among humans was found in the same geographic area, such as in RJ (Table 1). This fact matches with our hypothesis that outbreaks among cats directly influence the prevalence of S. brasiliensis in human cases of sporotrichosis in the same geographic area. A similar situation was observed in the state of RS where S. brasiliensis was isolated with high frequency from cats as well as from humans.

32Marimon et al. analyzed 127 Sporothrix isolates using the calmodulin locus and five major clades (I–V) were recognized. The maximum likelihood, neighbor-joining and Bayesian analyses based on the calmodulin (Fig. 3A) or EF1α (Fig. 3B) loci placed our animal Sporothrix isolates in Clade I (S. brasiliensis) composed of clinical samples from the RJ State epidemic, with strong bootstrap and posterior probability support. All pathogenic Sporothrix species are known to occur in Brazil, but S. brasiliensis is relatively frequent among feline sporotrichosis outbreaks.

33The geographic origin of S. brasiliensis of the Brazilian epidemic is difficult to establish. At least two distinct genotypes occur: one in RS and another in RJ. The latter is the oldest and longest recorded in the literature,,,. Our data show that humans and animals infected in the RS epidemic harbor the same S. brasiliensis genotype, which is distinct from the one of the RJ epidemic. The RJ genotype is also present in the recent outbreaks in PR, MG and SP, which suggests spread of S. brasiliensis from RJ. Additionally, our results showed absence of recombination events in the CAL and EF1α loci, demonstrating that S. brasiliensis is a clonal species. Despite a recent indication of intraspecific variability within the species S. brasiliensis using RAPD we believe that this phenomenon is not frequent or strong enough to break the prevalent pattern of clonal population structure, i.e., recombination or scarce exchange of genetic material may occur in some point of the evolutionary course of the pathogen life without compromise or affect its population structure. This hypothesis has been discussed by Tibayrenc and Ayala through different group of pathogens including fungi.

34The existence of clonal populations has repeatedly been proven in fungal pathogens –, although most of these species are surmised to have occasional sexuality in any phase of their life cycle. Under permissive conditions, most fungi reproduce very effectively by asexual propagation. Sexual reproduction provides advantages to the pathogen under adverse conditions, generating suitable genotypes that enhance survival. Many fungal epidemics are driven by populations showing low levels of genetic diversity, as demonstrated in Penicillium marneffei,, Cryptococcus gattii, and Batrachochytrium dendrobatidis. Also feline and human sporotrichosis in Brazil caused by S. brasiliensis is driven by the spread of a clonal species. In contrast, outbreaks of other human pathogens such as Coccidioides immitis – and Paracoccidioides brasiliensis –, spread by a diversity of genotypes.

35The ecological aspects of the pathogenic species within the genus Sporothrix needs to be reevaluated, and this information can be crucial to find the source of S. brasiliensis in nature. Classically, soil, thorny plants, Sphagnum moss – and hay have been pointed as source of S. schenckii s.l. To date, just a single environmental isolate (FMR 8337) of S. brasiliensis was isolated and reported from domiciliary dust in Brazil,. Distant relatives of Sporothrix in the fungal order Ophiostomatales are mainly associates of bark beetles on woody plants,. Zhou et al. demonstrated that different ecologies are corroborated by phylogenetic separation.

36It is challenging to obtain environmental isolates of S. brasiliensis, and the low number of subjects contaminated with propagules from soil or woody plants is indeed low compared to the high occurrence in warm-blooded hosts,,. This suggests successful transmission among animals (cat-cat and cat-humans). This scenario is different from epidemics occurring in South Africa,, India,, the USA,, Australia,, China, and Japan, where patients are mainly infected through soil and decaying wood. Possibly the Brazilian epidemics of S. brasiliensis are related to the emergence of a pathogenic clone front of a highly susceptible feline host, rather than to an increase in population size of S. brasiliensis in nature. This is corroborated by the high degree of virulence observed in naturally infected cats in the outbreak area, as well as demonstrated in a murine model. Besides that, we do not discharge the hypothesis that the emergence of pathogenicity could also be attributed to a recent host-shift from an unknown host to cats as discussed in other groups of pathogens –. Feral cats present a great potential to spread the disease in a short period of time due to their mobility and digging behavior, whereas dispersal from soil or vegetable remains is ineffective.

37Classically, accumulation of mutations in fungal populations can lead to speciation processes. However, rapid emergence of a new, highly virulent pathogen which is able to explore different ecological niches may result from other processes than those observed in natural selection. In many plant-pathogenic fungi, such as Fusarium and Alternaria, pathogenicity is determined by mobile, dispensable small chromosomes,. Genetic processes such as hybridization of two distinct, sympatric species, parasexual recombination, or mechanisms of inactivation/activation of virulence genes by insertion of transposons can also drive the emergence of pathogenicity. Hybridization is one of the possible mechanisms of emergence of phytopathogenic fungi, as well as fungi pathogenic to animals. It has also been discussed in the genus Ophiostoma, which is phylogenetically related genus to Sporothrix. All these genetic processes, alone or in combination, may be the reason of the emergence of virulence in the species S. brasiliensis. The lack of variation in the populations of S. brasiliensis also may be the result of a strong selective pressure imposed by the feline host. Presence of opposite mating types and sexual reproduction leads to genetic recombination and may increase fitness and widen host ranges. So far, no evidence of sexual recombination was demonstrated experimentally for the species from the S. schenckii complex and this fact, combined with the hostile selective pressure of the cats may provide possible explanations for the lack of diversity in S. brasiliensis.

38The association of S. brasiliensis with cats may play an important role in the evolution and spread of this pathogen. The interaction between cats and S. brasiliensis is not an exclusive relationship, since S. schenckii s. str. was also found in the feline host. However, S. brasiliensis has become predominant in this host within less than a decade, indirectly indicating a recent adaptation to the conditions of the feline body. Therefore, cats represent a natural habitat for S. brasiliensis. In contrast to the situation in opportunistic fungi, Sporothrix species are able to escape from the host and be transmitted to the next host, which is one of the hallmarks of a pathogen. Transmission is either direct during fights, or indirect, the fungus returning to soil after the cat has died.

39Given the role of the mammal host in Sporothrix evolution, variance in fitness between clonal lineages of S. brasiliensis is expected to lead to populations that are better adapted to host conditions. For example, the body temperature of the feral cat Felis catus is around 38–39°C, depending on its activity. Interestingly, S. brasiliensis has the best rate of vegetative growth when incubated at 37°C, followed by S. schenckii s. str. (Fig. 6). Remaining species of Sporothrix such as S. globosa and S. mexicana appear to be more sensitive to temperature, having a maximum around 35°C. The cat's body temperature could be considered an important selective pressure event, selecting thermo-resistant strains during sporotrichosis outbreak episodes. Transmission of S. brasiliensis by cats promotes inoculation into human hosts of yeast cells of rather than of hyphae and conidia, yeast cells having been reported to be more virulent.

40The endotherm developed by mammals is a natural defense mechanism against pathogens –, and in our study this factor appears to restrict the occurrence of species of the S. schenckii complex that are sensitive to temperatures above 35–37°C,.

41Another important factor in understanding the success of the epidemic of sporotrichosis among cats in RJ, has a socio-economic character. Most cat owners are living in neglected areas and many abandon dead animals in the street, favoring contact with other feral cats, or simply bury their pets after death in their backyard or in nearby wastelands. This directly allows the return of the agent into the environment, increasing outbreak risks of the pathogen, and enhancing the spread of the clonal species. In an epidemic scenario, domestic pets such as cats and dogs are the first animals to become infected with the fungus. Subsequently human cases of sporotrichosis are likely to emerge. Thus, we believe that cats can act as sentinel animals for epidemiological services, and its notification should be compulsory by regulatory agencies as the Centers for Zoonosis Control. The predominance of a species that is highly virulent to humans and animals requires fast implementation of public health policies to contain the epidemic, lowering harmful effects to the population.