Molecular Epidemiology of Fonsecaea Species

These fungi disperse slowly, leading to changes in structure at different geographic locations.

T he genus Fonsecaea comprises etiologic fungal agents of human chromoblastomycosis (1-3), a chronic cutaneous and subcutaneous infection characterized by slowly expanding nodules that eventually lead to emerging, caulifl ower-like, mutilating and disfi guring eruptions. Infection proceeds with muriform cells in tissue provoking a granulomatous immune response. In areas where it is endemic, disease incidence is high. Yegres et al. (4) and Yëgues-Rodriguez et al. (5) noted a frequency of 16 cases/1,000 populaAtion under arid climatic conditions in rural communities of Venezuela; chromoblastomycosis in that region is caused mainly by Cladophialophora carrionii. In contrast, Fonsecaea spp. are prevalent in humid tropical climates. Esterre et al. (6) reported 1,343 cases of chromoblastomycosis from Madagascar, 61.8% of which were caused by Fonsecaea spp. Kombila et al. (7) reported 64 cases in Gabon (equatorial Africa), all caused by Fonsecaea spp., and Silva et al. (8) cited 325 cases in the Amazon region of Brazil, 98% of which had Fonsecaea spp. as the etiologic agent. In Sri Lanka, 94% of 71 chromoblastomycosis cases were caused by Fonsecaea spp. (9).
Fonsecaea contains anamorphic ascomycetes belonging to the family Herpotrichiellaceae (order Chaetothyriales), which includes black yeasts and relatives (10)(11)(12). The genus comprises 3 sibling species: F. pedrosoi, F. monophora, and F. nubica, each of which has pathogenic potential (10,13,14). Infection process and routes of dispersal are insuffi ciently clarifi ed. Humans presumably acquire the infection after being pricked by contaminated thorns or wood splinters, but some agents are substantially more clinically prevalent than their predominantly (hitherto unnamed) environmental counterparts (15), which indicates that infection is not a random process. In many published case reports, etiologic agents were referred to as Phialophora pedrosoi or identifi ed with the obsolete name F. compacta, now known to be a mutant F. pedrosoi (9,13,16). Strains are no longer accessible for molecular verifi cation. Hence, no data are available on the epidemiology of the species as defi ned by sequence data.
Phylogenetically, Fonsecaea spp. agents of chromoblastomycosis are fl anked by nonpathogenic species (10) growing on plant debris. Discovery of natural habitat and source of infection by entities emerging on the human host is essential for understanding the evolution of pathogenicity. We present an amplifi ed fragment-length polymorphism (AFLP) DNA fi ngerprinting study of a worldwide collection of clinical isolates that were identifi ed as Fonsecaea spp. by state-of-the-art sequencing methods, supplemented with environmental isolates of the same species. The AFLP technique is a powerful method for discrimination between fungal species and for providing high-resolution fi ngerprinting data within species (17)(18)(19).

Fungal Strains and Culture Conditions
We studied 81 isolates representing the 3 currently recognized Fonsecaea spp. Geographic origins and hosts of the strains are listed in Table 1; the set include reference strains from the Centraalbureau voor Schimmelcultures (CBS-KNAW Fungal Biodiversity Centre, Utrecht, the Netherlands) and fresh isolates from patients and from the environment. Stock cultures were maintained on slants of 2% malt extract agar and oatmeal agar at 24°C.

Preselective and Selective PCR
In preselective PCR, 2 μL of diluted restriction/ligation product was added to 7.5 μL of AFLP core mix (Applied Biosystems), 0.25 μL of the EcoRI core sequence (5′-GAC TGC GTA CCA ATTC-3′), and 0.25 μL of the MseI core sequence (5′-GAT GAG TCC TGA GTAA-3′). The mixture was amplifi ed in an iCycler (Bio-Rad, Hercules, CA, USA) under the following conditions: 2 min at 72°C, followed by 20 cycles of 20 s at 94°C, 30 s at 56°C, and 2 min at 72°C. Each preselective PCR was diluted 2× by adding 10 μL of dH 2 O. In selective PCR, 1.5 μL of diluted preselective PCR products was mixed with 8.5 selective PCR mix containing 0.5 μL EcoRI-AC (labeled with FAM [6-carboxy fl uorescein]), 0.5 μL MseI-A, and 7.5 μL AFLP core mix (Applied Biosystems). The selective PCR conditions were cycling for 2 min at 94°C, followed by 10 cycles of 20 s at 94°C and 30 s at 66°C (decreasing 1°C with each subsequent cycle), and a fi nal extension of 2 min at 72°C. This sequence was followed by 25 cycles of 20 s at 94°C, 30 s at 56°C, and 2 min at 72°C, and a fi nal incubation of 30 min at 60°C.

AFLP Analysis
FAM-labeled products were prepared for analysis in an ABI PRISM 377 Genetic Analyzer (Applied Biosystems) as follows: the selective PCR products were cleaned with Sephadex G-50, and selective PCR products were mixed with LIZ 500 in the new plate by several times pipetting (fi rst by preparing master mix [8.7 μL demineralized water plus 0.3 μL Liz 500], then mixing this with 1.0 μL of selective PCR product by pipetting). The total volume was adjusted to 10 μL with dH 2 O. Denaturation was done at 95°C for 5 min, and then the reaction was snap-cooled on ice water. The LIZ 500 internal size standard in each sample was used for normalization of the fi ngerprint pattern according to the instruction manual. The densitometric curves were analyzed with BioNumerics software package (version 4.61, Applied Maths, Kortrijk, Belgium), by using the cosine similarity coeffi cient and the unweighted pair group method with arithmetic means cluster analysis. Statisti cal reliability of the cluster was investigated by using a cophenetic value, which calculates the correlation between the calculated and the dendrogram-derived similarity. Subdivisions in clusters were checked visually if they were supported by the banding patterns.

Results
Profi les of 81 strains were generated with the EcoRI-AC + MseI-A PCR adaptors. Fingerprints contained ≈60-70 bands in a 50-500-bp range. Another selective PCR with EcoRI core sequence+C and MseI core sequence+A primer combination used elsewhere in related fungi (24) resulted in nonscorable fi ngerprints because of amplifi cation of too many or only faint bands. Dendrograms derived from the AFLP banding patterns of Fonsecaea spp. were generated by using the unweighted pair group method with arithmetic means cluster analysis (online Appendix Figure, www. cdc.gov/EID/17/3/464-appF.htm). At >62.50% similarity, 3 main clusters were found that matched with existing species on the basis of multilocus sequence analysis (ITS, CDC42, BT2, and ACT1), i.e., F. pedrosoi, F. monophora, and F. nubica. At an automatic cutoff value option set at <62.5% similarity, the F. monophora and F. nubica clusters were subdivided in 2 evident groups each, leading to a total of 5 clusters (1-5) interpreted as populations. Clusters 1 and 2 matched with F. nubica, clusters 3 and 4 with F. monophora, and cluster 5 with F. pedrosoi. Individual bands varied within the profi les, but further subclustering was limited, e.g., in a slightly deviating derived subclade in population 5. The groups defi ned above by AFLP analysis are interpreted as populations (1)(2)(3)(4)(5) in the text below. In population 5, some strains were nearly 100% identical, e.g., CBS 122341, 122343, 122345, and 122349, all originating from patients with chromoblastomycosis in Mexico City, Mexico (online Appendix Figure; online Appendix Table, www.cdc.gov/EID/17/3/464-appT.htm).
We determined the geographic distributions of the 5 main populations of Fonsecaea strains (Figure). Areas endemic for Fonsecaea, judging from the literature, are in tropical and subtropical climate zones. Population 1 comprised a cluster of F. nubica strains originating from humans with chromoblastomycosis in Guangdong, People's Republic of China. Population 2 of the same species comprised 4 strains, 2 of which originated from humans with chromoblastomycosis in South America, 1 from France, and 1 with unknown origin. The profi les were too different to trace to any clonal identity. Population 3 (F. monophora) comprised 15 strains, most of which were isolated from humans with chromoblastomycosis in South America; 1 originated from the United States, and 1 originated from Haikou in southern China. Two strains were isolated from decaying plants in Brazil, and the second US strain was derived from a human with a brain infection. Two other strains from human brain infections in Brazil and in Africa had unique profi les that could not be unambiguously linked to any other isolate. Another African strain, from a patient with chromoblastomycosis who lived in Spain and had acquired the infection 36 years earlier in Guinea (25) sample originated from a patient with a brain infection who lived in the United Kingdom (26); whether the patient had visited southern China could not be established. In population 5 (F. pedrosoi), most strains originated from chromoblastomycosis patients in Central and South America. Some geographic clustering was visible, i.e., the derived group of strains from South America (uppermost clade of population 5 in the online Appendix Figure) was segregated from those from Central America. Several of the strains from South American originated from soil and were isolated through mouse passage. One strain from an ear of a gazelle in Libya and 1 from a human with chromoblastomycosis in the Netherlands could not directly be linked to any other strain.

Discussion
AFLP typing is comparable to use of other DNA markers, such as random amplifi ed polymorphic DNA, restriction fragment-length polymorphism, or microsatellites, in terms of time and cost effi ciency, reproducibility, and resolution (27). The technique has emerged as a major epidemiologic tool with broad application in ecology, population genetics, pathotyping, DNA fi ngerprinting, and quantitative trait loci mapping (28). AFLP fi ngerprinting is useful for the molecular characterization of microorganisms with relatively large genomes, including various fungal species (18,19,(21)(22)(23)29,30). In a preliminary experiment that used different primer combinations, the combination EcoRI-AC + MseI-A adaptors gave excellent results, yielding readable profi les with well-separated bands.
The degree of variation in Fonsecaea appeared to differ between species. The major 5 clusters were separated at <62.5% similarity, with signifi cant differences in the presence of major fragments, several of which were unique to individual isolates or subpopulations. Populations 1 and 2, 3 and 4, and 5 corresponded with species borderlines established recently by Najafzadeh et al. (10,14) on the basis of multilocus sequencing with ITS, CDC42, BT2, and ACT1. Population 5 (F. pedrosoi) varied least at >71.7% similarity, with limited reproducible substructure being discernable. Nearly all isolates of this species originated from South and Central America (Venezuela, Brazil, Mexico, Argentina, Puerto Rico, and Uruguay). One isolate from a human with chromoblastomycosis in the Netherlands was likely to have been imported (13). One isolate from a gazelle ear in Libya, northern Africa, was the only geographic exception that could not be explained. Clusters of strains that could be grouped as being visually identical and with similarities >71.7% (online Appendix Figure; online Appendix Table) were mostly collected at close geographic distance from each other. This fi nding suggests that vectors of dispersal for Fonsecaea spp. are slow, leading to detectable regional diversifi cation. The relatively low degree of variation of F. pedrosoi and confi nement to Central and South America indicate a founder effect, the species being the most recently emerged taxon in Fonsecaea. F. monophora and F. nubica were distributed worldwide but were geographically diverse in that population 4 of F. monophora was nearly confi ned to China, with highly similar profi les (online Appendix Figure). One strain of this population 4, CBS 117238, originated from a brain infection in a human in the United Kingdom; whether this patient had emigrated from China could not be determined from the original publication (25). F. monophora population 3 was found mainly in the Western Hemisphere, particularly in Brazil. Judging from the near identity of profi les of strains isolated in 1937 (CBS 271.37) and in 1999 (CBS 102245) (online Appendix Figure), we can conclude that clones are maintained locally over decades. The 2 US strains presumably derived from immigrants from South America or Central America. Population 3 was also found in Africa and in Haikou in China, 600 km from Guangdong, where population 4 of F. monophora is prevalent. Strains of F. nubica show a similar bipartition over Asia and the Western Hemisphere, with a prevalently Chinese (population 1) and a prevalently Brazilian (population 2) population, and a presumed infected immigrant in France. Kawasaki et al. (31,32) provided similar data on the basis of restriction fragmentlength polymorphism of mitochondrial DNA, showing that Fonsecaea spp. from Japan and China differed consistently from isolates from Central and South America.
Nearly all Fonsecaea spp. isolates available in culture collections originate from mammals, mostly humans with chromoblastomycosis, and were rarely recovered from the environment of symptomatic patients despite several attempts (33). Occasionally, F. pedrosoi was isolated from mice that were euthanized for isolation of black yeasts after they had been inoculated with environmental samples (34). This information suggests that Fonsecaea spp., particularly F. pedrosoi, have a competitive advantage by using this enrichment source. Mouse passage proved to be more effi cient for environmental isolation of etiologic agents of chromoblastomycosis than general methods such as oil fl otation (35). The latter technique mostly isolates other environmental Fonsecaea spp. that are not known to be pathogenic to humans (33).
In humans with chromoblastomycosis, the male:female ratio of patients is 63:2. This male preponderance of 97% cannot be explained by different exposition rates. Distinct male preponderance is also noted in the neurotropic relative, Cladophialophora bantiana (G.S. de Hoog, unpub. data). Population 3 of F. monophora has a wider clinical spectrum than the remaining groups, comprising, in addition to chromoblastomycosis, several isolates from human brain infection. This population also comprised some isolates from soil and plant debris acquired without use of mammal baits. Coexistence of closely interrelated entities differing in pathogenicity and virulence seems likely in Fonsecaea spp., as was also suggested for black yeasts (A.H.G. Gerrits van den Ende et al., unpub. data).
Our data demonstrate that AFLP fi ngerprinting is a tool that produces highly reproducible results for molecular epidemiology. The use of AFLP showed that local Fonsecaea agents of chromoblastomycosis seem able to be maintained over 70 years, and therefore epidemiologic profi les take the structure of expanding clones. By locality, patients are infected by only a limited number of genotypes. The fungi disperse slowly, leading to appreciable geographic structuring, which ultimately may lead to allopatric speciation (diversifi cation resulting from geographic barriers). Few environmental strains have been recovered during repeated isolation experiments, whereas Fonsecaea spp. accumulates substantially in the human host. The mechanisms behind their pathology remain unexplained.