Phylogenetic analysis reveals a cryptic species Blastomyces gilchristii, sp. nov. within the human pathogenic fungus Blastomyces dermatitidis

Abstract

Background: Analysis of the population genetic structure of microbial species is of fundamental importance to many scientific disciplines because it can identify cryptic species, reveal reproductive mode, and elucidate processes that contribute to pathogen evolution. Here, we examined the population genetic structure and geographic differentiation of the sexual, dimorphic fungus Blastomyces dermatitidis, the causative agent of blastomycosis.

Methodology/principal findings: Criteria for Genealogical Concordance Phylogenetic Species Recognition (GCPSR) applied to seven nuclear loci (arf6, chs2, drk1, fads, pyrF, tub1, and its-2) from 78 clinical and environmental isolates identified two previously unrecognized phylogenetic species. Four of seven single gene phylogenies examined (chs2, drk1, pyrF, and its-2) supported the separation of Phylogenetic Species 1 (PS1) and Phylogenetic Species 2 (PS2) which were also well differentiated in the concatenated chs2-drk1-fads-pyrF-tub1-arf6-its2 genealogy with all isolates falling into one of two evolutionarily independent lineages. Phylogenetic species were genetically distinct with interspecific divergence 4-fold greater than intraspecific divergence and a high Fst value (0.772, P<0.001) indicative of restricted gene flow between PS1 and PS2. Whereas panmixia expected of a single freely recombining population was not observed, recombination was detected when PS1 and PS2 were assessed separately, suggesting reproductive isolation. Random mating among PS1 isolates, which were distributed across North America, was only detected after partitioning isolates into six geographic regions. The PS2 population, found predominantly in the hyper-endemic regions of northwestern Ontario, Wisconsin, and Minnesota, contained a substantial clonal component with random mating detected only among unique genotypes in the population.

Conclusions/significance: These analyses provide evidence for a genetically divergent clade within Blastomyces dermatitidis, which we use to describe a novel species, Blastomyces gilchristii sp. nov. In addition, we discuss the value of population genetic and phylogenetic analyses as a foundation for disease surveillance, understanding pathogen evolution, and discerning phenotypic differences between phylogenetic species.

Figure 1. Genealogies of the seven MLST gene loci sequenced for 78 isolates of B. dermatitidis.

Majority consensus maximum parsimony (MP) trees were constructed for each MLST locus region. (a) chs2, (b) drk1, (c) fads, (d) pyrF, (e) tub1, (f) arf6, and (g) its2. Isolates are coded by geographic region of isolation. For each group, the size of the patterned sector is proportional to the number of isolates from that geographic region. Values along branches represent maximum parsimony bootstrap values (MPB) and Bayesian posterior probability (BPP) values respectively. Branch values are provided where MP and Bayesian tree topology were in agreement.

Figure 2. Maximum-parsimony tree constructed from the concatenated sequences of seven nuclear genes (chs2-drk1-fads-pyrF-tub1-arf6-its2).

A majority consensus mid-point rooted maximum parsimony (MP) tree was constructed based on the concatenated sequence of seven MLST gene loci. The MP tree is displayed without logarithmic scaling so that genetic distance and geographic region of isolation could be viewed. Values along branches represent maximum parsimony bootstrap values (MPB) and Bayesian posterior probability (BPP) values respectively. Values for branches partitioning less than three isolates or those with MPB ≤70 and BPP≤0.95 are not shown. The tree is displayed with pattern coding for geographic regions as described in Figure 1. The MP tree displays a partition in the current species B. dermatitidis into two phylogenetic species: B. dermatitidis (PS1, clade 1) and B. gilchristii (PS2, clade 2).

Figure 3. Parsimony Tree Permutation Length Test (PTPLT).

A frequency distribution of tree lengths of maximum parsimony trees for 1000 artificially recombined datasets as compared to the observed tree length was calculated for: (a) All isolates in the B. dermatitidis population treated as a single population both with and without reduction to unique genotypes; (b) All isolates partitioned into two phylogenetic species (PS1) and (PS2) both with and without reduction to unique genotypes; (c) All isolates partitioned into 6 geographic regions both with and without reduction to unique genotypes; (d) Isolates of PS1 treated as a single population, partitioned into 6 geographic regions both with and without reduction to unique genotypes; (e) Isolates of PS2 treated as a single population, partitioned into 4 geographic regions, or reduced to unique genotypes (due to the low number of unique genotypes in PS2, we were unable to partition them by geographic region). Recombination was not detected between PS1 and PS2. Recombination was detected in PS1 (d) when partitioned by geographic region and reduced to unique genotypes and in PS2 (e) when reduced to unique genotypes.

Figure 4. Phylogenetic network analysis

of the B. dermatitidis. Split decomposition analysis was performed using the pairwise genetic distances of the concatenated chs2-drk1-fads-pyrF-tub1-arf6-its2 sequences. Isolated split networks within B. gilchristii (PS2) suggests recombination within the B. gilchristii (PS2) but not between both populations. Bootstrapping values less than ≤0.75 are not shown.

Figure 5. Geographic distribution of B. dermatitidis phylogenetic species across Ontario, Canada and Wisconsin, USA.

Phylogenetic species are represented by circles with light grey (B. dermatitidis, PS1) and solid (B. gilchristii, PS2) fill corresponding to each phylogenetic species. Circle size is proportional to the number of isolates from each geographic site.