Sequences of the ITS, ARF, H-anti, and OLE of Histoplasma species were downloaded from the NCBI GenBank database or abstracted from genome sequences of related isolates. In view of optimal resolution of phylogenetic relationships, clone correction was applied by retaining only one sequence per gene per isolate. After selection, a total of 879 isolates were included in this study (Suppl. material 1: table S1). Specifically, the sequences involved in the single-gene phylogenetic tree analysis were ITS (n = 627), ARF (n = 451), OLE (n = 491) and H-antigen (n = 416), respectively. In multilocus trees, 400 sequences were involved in the ARF / OLE analysis, and 274 sequences were included in four-gene multilocus analyses. The respective phylogenetic trees resulting from these analyses represent the study with the largest number of included strains to date. Strains investigated originated from 47 countries or regions and have a global distribution, with a preponderance (31.74%) of strains from Brazil. Original names and numbers were maintained as deposited in GenBank to enhance recognition of strains. A search of original literature revealed living type materials for all names formally introduced in Histoplasma. For several, living strains proved to be extant but not always available for analysis of all genes covered in this study. Strains were then attributed to their clades on the basis of data available in GenBank, while in some cases the missing genes were sequenced in-house or abstracted from genomes according to the original papers.

Ten clades maintained at phylogenetic species level in seven earlier studies were selected as guidelines in this study (Kasuga et al. 1999; Kasuga et al. 2003; Teixeira et al. 2016; Sepúlveda et al. 2017; Rodrigues et al. 2020; Almeida-Silva et al. 2021; Jofre et al. 2022). The entities have been indicated in these papers as NAm-1, NAm-2, Africa, LAm-A (LAm-A1 and LAm-A2), LAm-B (LAm-B1 and LAm-B2), LAm-C, LAm-D, LAm-E, RJ and India, and an additional Indonesia clade identified in this study.

In-house sequenced strains (CBS 215.53) were grown on malt extract agar (MEA) for 14 days. Approximately 1 cm² of material was added to a screw-capped tube containing 490 µL CTAB-buffer (2% cetyltrimethylammonium bromide, 100 mM Tris-HCl, 20 mM EDTA, 1.4 M NaCl) and 6–10 acid-washed glass beads. The above procedures were performed inside a class II biological safety cabinet under conditions of biosafety level 3 (BSL3) containment. Ten units of proteinase K were added to the mixture and vortexed with a MOBIO vortex for a few min. Tubes were incubated at 60 °C for 60 min. After incubation, the tubes were again vortexed and 500 µL of chloroform: isoamyl alcohol (24 : 1) were added followed by shaking for 2 min. Tubes were spun at 14,000 r.p.m. in a microfuge for 10 min and the upper layer was collected in new sterile tubes with 0.55 volume ice-cold iso-propanol and spun again. The pellets were washed with 70% ethanol, air-dried and re-suspended in 50 µL TE buffer. DNA amplification was performed for ARF and H-antigen. Primers used for amplification and sequencing of ARF were ARF1 and ARF2, for H-anti these were H-anti3 and H-anti4 (Kasuga et al. 2003). The amplifications were carried out in 50 μL reaction mixtures [10 μL PCR buffer, 1 μL dNTPs, 1 μM of each primer (10 pmol), 0.5 μL Taq polymerase, 0.5 μL DNA and 36 μL water].

Sequences of ITS, ARF, OLE, H-anti of related isolates were edited using BIOEDIT v7.2 (Hall 1999). Alignments were made by MAFFT v7 (http://mafft.cbrc.jp/) and optimized manually using MEGA v7.2 (Kumar et al. 2012) and BIOEDIT v7.2. Missing data for partial or complete sequences in some taxa were coded as ‘missing’ (Wiens 2006). To address the phylogenetic relationships among taxa, Maximum Likelihood (ML) and Bayesian inference (BI) algorithms were used (http://www.phylo.org/). For better comparison, we also used the IQ-TREE (http://www.iqtree.org/doc/Concordance-Factor) web server and applied the Maximum Likelihood algorithm to construct the phylogenetic trees. One strain with a long branch in all trees was selected as outgroup. Trees were edited using TREEVIEW v1.6.6 and completed with Adobe ILLUSTRATOR CS v5. The bootstrap of 11 clades in different phylogenetic trees are shown in Table 1.

The haplotype diversities were estimated based on two genes, ARF and OLE. Haplotype networks for the genes analyzed were plotted using the geneHapR library v1.1.9 (Zhang et al. 2023). Haplotype networks were built using the same sequences as those used for constructing the ARF and OLE phylogenetic trees. Due to the low diversity of ITS and the short sequences of H-anti, these two genes were excluded from the haplotype network analysis. We used seven different colored spheres to represent seven different categories: six categories represent six clades H. capsulatum, H. (var.) duboisii, H. mississippiense, H. ohiense, India and Indonesia. The seventh represents strains from Panama, as these include the type strain of Histoplasma capsulatum. The size of each sphere is proportional to the number of strains belonging to that haplotype. The geographic of each clades was calculated and shown in the maps.

To assess the concordance of lineages among different genes of studied strains, we designed two sets of analyses. The first set involved a selection of strains all having sequences for four genes, and constructing individual single-gene phylogenetic trees. These trees were pairwise compared using a custom-designed encoding program (ITS vs. ARF, ITS vs. OLE, ARF vs. OLE). The combined results were then used to determine potential phylogenetic species. In the second set of analyses, only strains previously defined as LAm-A, LAm-B, LAm-C, LAm-D, LAm-E, RJ, NAm-1, NAm-2, Africa, Indonesia and India were retained. Individual phylogenetic trees for the ARF and OLE genes were constructed, and their lineage relationships were analyzed using the custom-designed encoding program.

To enhance the reliability of the phylogenetic tree in this study, three tree models were applied: a maximum likelihood (ML) tree based on the CIPRES platform, ML trees based on the IQ-TREE platform, and Bayesian trees based on the CIPRES platform. The results indicate that the topologies of these three trees are highly consistent, with the main differences in branch support values. The ML tree generated with IQ-TREE in general yielded highest branch support, with some branches that did not receive significant support in the other trees showing support rates above 90% in the IQ-TREE ML tree. In the following parts, only the ML tree based on the CIPRES platform is presented, with branch support values of these three trees listed in the order of ML (CIPRES), ML (IQ-TREE), and Bayes (CIPRES).

Through a search in GenBank and extraction from genomes, we collected a total of 627 ITS sequences deposited under the generic name Histoplasma or Ajellomyces, after removing some duplicate sequences, 499 sequences were retained for analysis. This extensive data set comprehensively represents all currently known taxa in Histoplasma. Aligned ITS sequences were 470 bp long with the following base frequencies: pi(A) = 0.186443, pi(T) = 0.204952, pi(C) = 0.309767, pi(G) = 0.298837. ML tree generated by CIPRES was shown (Suppl. material 1: fig. S1, table S1). Sequences of strains CADAM, AC02, and RS36 stood out in a long branch which could not be explained by their low number (n = 7) of SNPs. The results indicate that most ITS sequences of Histoplasma share the same sequence and few differences between strains. Four supported clades contained reference strains of described phylogenetic species (NAm-1, NAm-2, Africa, India). In the ITS tree, nine strains form a distinct clade (NAm-2, BS 98%, BS 99%, PI 100%) with 11 additional strains from North America and two strains from South America. CBS 136.72 was described as ex-type strain of Histoplasma capsulatum. G217B was described as type strain of Histoplasma ohiense. Both strains clustered together in this clade. Similarly, nine strains cluster together in a clade referred to as NAm-1 with support rate (BS 77%, BS 87%, PI 100%); this clade includes 11 further U.S.A. strains, plus 5 strains from Iraq and 1 strain from Austria. A third supported clade (BS 98%, BS 99%, PI 100%) comprises strains exclusively from Africa. Eleven strains from India clustered together into a subclade with support rate (BS 87%, BS 100%, PI 100%). The strains previously grouped as the phylogenetic species LAm did not aggregate in supported clades. Strains formerly indicated as RJ, LAm-A, LAm-A1, and LAm-A2 were found scattered throughout the tree. Similarly, strains referred to as LAm-B, LAm-B1, and LAm-B2 were dispersed in several clades of the tree. The three clades reported by Rodrigues et al. (2020), i.e. LAm-C, LAm-D, and LAm-E were intermixed with strains from the LAm-A and RJ group. Twelve isolates from India and 9 isolates from Indonesia clustered together in a clade containing strains all originating from Asia, but with low bootstrap support.

Groups with identical sequences, even when lacking statistical support, mostly originated from the same geographic region. The upper half of ITS tree comprising 70 identical sequences contains only five strains from outside South America, i.e. IFM 41330 from Japan, and four strains from North America. The backbone of the lower part of the tree also contains an overabundance of identical sequences from South America(n = 83), but also from other continents. Clusters of strains that deviated by a few SNPs from the main groups, despite absence of statistical support, nearly always originated from a single country.

Aligned ARF sequences were 418 bp, with the following base frequencies: pi(A) = 0.298862, pi(T)= 0.257376, pi(C) = 0.239235, pi(G) = 0.204527. In the ML tree (Fig. 1A; Suppl. material 1: fig. S2, table S1), numerous subclusters that deviate in a single SNP are nearly always composed of strains from a single or from adjacent countries. One cluster with 66% BS contains twelve strains from Europe and one strain H90 from Egypt, which was described as type strain of Histoplasma capsulatum var. farciminosum. Strains from Central America (Honduras, Mexico) are relatively variable, belonging to several small clusters. The reference strains of previously distinguished taxa are found in an African clade (BS 95%, BS 99%, PI 100%), and strains referred to as NAm-1 (BS 97%, BS 99%, PI 100%) and NAm-2 (BS 98%, BS 99%, PI 100%). A clade with support rate (BS 84%, BS 99%, PI 100%) with non-South American strains contains three subclusters: one subclade contains strains exclusively from India (BS 95%, BS 99%, PI 100%), one subclade contains strains exclusively from Indonesia (BS 76%, BS 99%, PI 99%).

Figure 1. 

Collapsed phylogenetic tree of Histoplasma based on ARF, OLE, two genes (ARF and OLE) or multilocus sequences (ITS, ARF, OLE and H-anti), obtained by maximum likelihood. All the bootstrap are shown. Different colors represent different clades. Single clade with long branch were selected as outgroup. A ARF tree and OLE tree; B two-gene tree and Combined-Gene.

Comparing the current ARF tree with previously recognized groups, a group referred to as LAm-A has been recognized as an independent clade (Kasuga et al. 2003; Rodrigues et al. 2020). Teixeira et al. (2016) divided LAm-A into two subclades, LAm-A1 and LAm-A2. In the current ARF tree, the strains of these subclades do not show significant differences; instead, they were scattered in one big clade with other LAm strains. LAm-B was accepted as a phylogenetic species in the studies of Kasuga et al. (2003) and Rodrigues et al. (2020). Teixeira et al. (2016) distinguished two subclades, LAm-B1 and LAm-B2. In the current ARF data, the bipartition of LAm-B was recognized: a subclade (BS 89%, BS 98%, PI 100%) contained three RJ strains, 16 LAm-B or LAm-B1 strains, in addition to some other isolates, and a subclade (BS 79%, BS 98%, PI 92%) with 12 strains of LAm-B or LAm-B2 groups, 13 strains of RJ group, 2 strains of LAm-D, and some other strains. Exceptions were strains H66 and H69, which were considered to belong to LAm-B but were found to be forming independent clades rather than clustering into any of the LAm-B clades. All the strains in these clades originated from South America.

Teixeira et al. (2016) recognized a clade ‘RJ’ as a possible cryptic species, of which isolates of the LAm-A group were collected in Rio de Janeiro and São Paulo states in Brazil. This clade was confirmed by Rodrigues et al. (2020). However, in the current ARF tree, strains identified as RJ were mixed with other Latin American strains, scattered across different clades rather than forming a distinct clade. The reason might be that ‘RJ’ indicates a collection, with not all strains originating from the Rio de Janeiro area.

Rodrigues et al. (2020) recognized several further cryptic entities, named LAm-C, -D and -E. All strains referred to as LAm-C clustered together in a major clade. Despite the support rates not being high (BS 60%, BS 98%, PI 84%), this clade contained strains originating exclusively from Brazil. The strains of LAm-D were divided into two parts. Two strains clustered with LAm-B (described above), while three strains were close to LAm-A strains. LAm-E comprised only two strains, which were close to each other but did not form an independent clade. Consequently, the ARF data did not support these Latin American groups.

The OLE gene alignment contained 491 sequences. Aligned OLE sequences were 407 bp, with the following base frequencies: pi(A) = 0.226902, pi(T) = 0.269648, pi(C) = 0.281335, pi(G) = 0.222114. An African and two North American clades were recognized, but within the sequences from Latin America no supported clades were found, strains belonging to subclades with other genes being intermixed, without forming distinct clades(Fig. 1A; Suppl. material 1: fig. S3, table S1). Twenty strains from Africa all clustered in a clade with support rate (BS 97%, BS 99%, PI 100%), containing not a single strain from another continent. Fourteen strains from the U.S.A, and one from Austria formed a clade previously referred to as NAm-1 (Kasuga et al. 1999; Kasuga et al. 2003; Teixeira et al. 2016; Rodrigues et al. 2020), with support rate (BS 99%, BS 100%, PI 100%). Twenty one from the U.S.A. and three from Colombia formed the NAm-2 branch (BS 96%, BS 100%, PI 100%). One of these Colombia strains (1986) was described as Histoplasma suramericanum by Sepúlveda et al. (2017). Additionally, 12 strains from Indonesia clustered with one strain from Australia in a clade with support rate (BS 87%, BS 98%, PI 100%); 16 strains from India, one strain from The Netherlands (probably concerning an immigrant) and one unknown strain formed a sister clade next to the Indonesian strains (BS 80%, BS 98%, PI 100%). A highly supported clade (BS 99%, BS 100%, PI 100%) comprised 27 strains, including those isolates of groups previously referred to as LAm-A, LAm-B, LAm-D, and RJ, the ex-type strain of H. suramericanum MZ5 also cluster in this clade (Teixeira et al. 2016; Sepúlveda et al. 2017; Rodrigues et al. 2020). The origins of these strains are diverse, spanning five countries from Brazil to Honduras.

The H-antigen gene tree contained 416 sequences (Suppl. material 1: fig. S4), with an alignment length of 223 bp. However, due to the short length of the sequences and the limited number of informative sites, this gene provided insufficient resolution for the purpose of the present paper.

Concatenated two-gene trees were made (n = 400, Fig. 1B; Suppl. material 1: fig. S5, table S1), and a multilocus tree using all four genes under study (n = 274, Fig. 1B; Suppl. material 1: fig. S6, table S1). The alignment of ARF + OLE was 826 bp long, with the following base frequencies: pi (A) = 0.264853, pi (T) = 0.262417, pi (C) = 0.259700, pi (G) = 0.213030. The alignment of the multilocus tree was 1467 bp long, with the following base frequencies: pi (A) = 0.237208, pi (T) = 0.245648, pi (C) = 0.263872, pi (G) = 0.253273.

In the ARF-OLE phylogenetic tree, LAm-C (BS 48%, BS 96%, PI 96%), India (BS 96%, BS 98%, PI 100%), Indonesia (BS 99%, BS 99%, PI 100%), Africa (BS 100%, BS 100%, PI 100%), NAm-1 (BS 100%, BS 100%, PI 100%), and NAm-2 (BS 100%, BS 100%, PI 100%) formed distinct clades. All LAm-A strains, two strains of LAm-E, five strains of LAm-D, and some RJ strains were found scattered throughout the tree, without forming supported clades. Most strains of LAm-B were distributed among two subclades, with low support values. One supported clade (BS 99%, BS 99%, PI 100%) contained 12 strains from Indonesia and one from Australia. 15 strains from India form a single clade (BS 96%, BS 98%, PI 100%).

Similar to the ARF-OLE tree, the multilocus tree revealed supported clades for LAm-C (BS 55%, BS 97%, PI 64%), India (BS 81%, BS 97%, PI 100%), Indonesia (BS 100%, BS 100%, PI 100%), Africa (BS 99%, BS 98%, PI 100%), NAm-1 (BS 100%, BS 100%, PI 100%) and NAm-2 (BS 100%, BS 100%, PI 100%). The strains from South and Central America did not form distinct clades; instead, they were dispersed in unsupported subgroups or intermixed with each other. Consequently, the multilocus tree does not support the distinction of LAm-A, LAm-B, LAm-C, LAm-D, LAm-E and RJ as phylogenetic species.

Based on the analyses of the above single-gene and multilocus phylogenetic trees, we identified two genes, ARF and OLE, which most effectively resolved the phylogenetic relationships within the genus Histoplasma. We then analyzed the concordance between these two genes by comparing the compositions of supported clades. The trees resulting from this analysis were labeled as AFR2 and OLE2 to differentiate them from the previously mentioned single-gene tree. To clarify the results, only 190 strains defined as 10 clades (Africa, NAm-1, NAm-2, LAm A-E, RJ, India) by previous studies, and an additional Indonesia clade identified in this study, were retained for concordance analysis of these two genes (ARF, Fig. 2A; OLE, Fig. 2B; Table 2). Twenty-eight strains of LAm-C cluster in separate clades with bootstrap values of 73% and 48%, respectively. Five strains of LAm-D cluster together in OLE2 (Suppl. material 1: fig. S7), but only three strains cluster together in ARF2 (Suppl. material 1: fig. S8), indicating that the lineage relationships are consistent for only three strains of LAm-D. The strains of groups LAm-A, LAm-B, LAm-E and RJ did not show phylogenetic consistency in ARF 2 and OLE 2. Some of the clusters were interspersed with each other, suggesting gene flow. For the Africa clade, six strains show consistent lineage relationships, while the remaining two African strains, H88 and H143, do not form a clade with other African strains but are located adjacent to the African clade in ARF 2 although included in the Africa clade in the OLE 2. Other clades, NAm-1, NAm-2, India and Indonesia all were concordant in these two genes, suggesting reproductive isolation. We also performed pairwise comparisons of the four single-gene phylogenetic trees. The results were similar to those of the two-gene comparisons, and therefore are not displayed.

Figure 2. 

Genealogical concordance of analysed sequences of ARF and OLE. Tanglegram comparisons were applied to analysis of gene exchange in these species. A Clades did not show phylogenetic consistency in ARF 2 and OLE 2; B clades show phylogenetic consistency in ARF 2 and OLE 2.

Based on the results above, we classified all the Histoplasma strains in this study into six major groups: H. mississippiense (NAm 1), H. ohiense (NAm 2), H. duboisii (Africa), India, Indonesia, H. capsulatum / suramericanum (LAm A-E, RJ, and other strains). Strains in the group Panama do not form an independent clade in any of the phylogenetic trees. Since Panama contains the type strain of H. capsulatum which typifies the genus, understanding its position in the haplotype network is very important. The haplotypes are color coded according to these seven groups. In the haplotype networks of ARF (Fig. 3A) and OLE (Fig. 4A), the strains of H. capsulatum / suramericanum exhibit the most extensive haplotype diversity, likely due to this group having the largest sample size. Additionally, some strains of this group share haplotypes with certain strains of H. ohiense in the ARF haplotype network. In both networks, the H. ohiense strains show relatively complex haplotypes. Besides four unique haplotypes clustering together in ARF and OLE, some strains also share haplotypes with H. capsulatum and Indonesia only in ARF. In contrast, the H. mississippiense group has only two haplotypes. Panama strains have only one unique haplotype in both networks. Similarly, H. duboisii strains share the same haplotypes in both networks, of which eight unique haplotypes cluster together. Indian strains have only one haplotype too. In the ARF network, Indonesia strains share haplotypes with H. ohiense. In the OLE network, they have two unique haplotypes.

Figure 3. 

A Haplotype networks and distribution patterns of H. capsulatum ARF sequences used in this study. The size of the circumference is proportional to the haplotype frequency. Seven colors were coded according to the genetic group representing Histoplasma duboisii, H. mississippiense, H. ohiense, H. capsulatum, India, Indonesia and Panama, respectively. B Percentages of individuals per locality assigned to the most probable populations defined by the ARF sequences analysis.

Meanwhile, we mapped the geographic origins of the strains in the seven groups for the ARF (Fig. 3B) and OLE (Fig. 4B) genes separately. The geographic distributions of the strains involved in both genes are almost identical. The strains of H. capsulatum are distributed almost globally. The strains H. mississippiense have two origins, the United States and Colombia. H. ohiense strains also have two origins, the United States and one strain from Austria. Most of the H. duboisii strains originate from Africa, although one strain (H88) is recorded from Belgium, which is likely from an immigrant. One strain (H176) diagnosed in the Netherlands forms a sister clade with Indian strains in the ARF tree and clusters within the Indian branch in the OLE tree. Similarly, one strain (H157) from Australia forms a sister clade with Indonesian strains in the ARF tree and clusters within the Indonesian clade in the OLE tree.

Figure 4. 

A Haplotype networks and distribution patterns of H. capsulatum OLE sequences used in this study. The size of the circumference is proportional to the haplotype frequency. Seven colors were coded according to the genetic group representing Histoplasma duboisii, H. mississippiense, H. ohiense, H. capsulatum, India, Indonesia and Panama, respectively. B Percentages of individuals per locality assigned to the most probable populations defined by the OLE sequences analysis.

This study includes a total of 879 Histoplasma strains from 47 countries or regions (Fig. 5), covering all continents except Antarctica. Among them, the strains from South America are the most abundant, with 454 strains, accounting for 51.65% of the total. North America contributes 266 strains, accounting for 30.26% of the total. Africa has 36 strains, representing 4.10% of the total. There are 73 strains from Asia, comprising 8.30% of the total, and 29 strains from Europe, accounting for 3.30% of the total. Only two strains originate from Australia. As of the current study, there have been no reports of Histoplasma from Antarctica.

Figure 5. 

The global distribution of all strains in this study.

Most strains from Asia lacked protein-coding sequence depositions, so that only ITS rDNA could be included in the analysis. The ITS tree comprised a cluster only containing strains from Asian countries, although lacking significant bootstrap support. The clade referred to as NAm-1 contains, besides strains from the U.S.A., five strains from Iraq. In all trees, the majority of African strains cluster together to form an independent branch (98% BS in ITS, 95% BS in ARF, 97% BS in OLE). Eleven strains from Indonesia formed clades with relatively high support in the ARF (BS 76%, BS 99%, PI 99%), OLE (BS 87%, BS 98%, PI 100%), two genes (BS 99%, BS 99%, PI 100%) and multilocus (BS 100%, BS 100%, PI 100%) trees. No significant group support was achieved with H-antigen and ITS. Sixteen strains from India clustered together almost in all trees with high bootstrap values, including ITS (BS 87%, BS 100%, PI 100%), ARF (BS 95%, BS 99%, PI 100%), OLE (BS 80%, BS 98%, PI 100%), two genes (BS 96%, BS 98%, PI 100%) and multilocus (BS 81%, BS 97%, PI 100%). In all trees, strains from Brazil were scattered over several clades, but the NAm-1, NAm-2, and the African, Indonesian and India clades did not include Brazilian strains.

Given the pronounced clustering with geographical preferences, the North American clades consistently being separate from the Latin American clades, strains from the Central American country Mexico, which borders the U.S.A., consistently clustered with Central and South American strains, defined as the LAm clade. In contrast, three strains from Colombia, 1986, COL_H_001 and COL_H_002, consistently appear in the NAm-2 clade, while remaining Colombian strains are scattered throughout the tree along with strains from South America. The strains of the Panama cluster was located inside the distribution of the L-Am clade.

The majority of African strains clustered together in most of the trees. The ITS tree comprises a total of 30 strains from Africa. Fifteen strains clustered in a single clade with bootstrap support of 98%, along with two strains from the U.S.A. Additionally, 15 strains were scattered in the lower half of the tree and did not form a single clade. In the ARF tree, six strains from Africa clustered in a clade, while four African strains did not form a separate clade. In the OLE tree, twenty one strains from Africa formed an independent clade (BS 97%, BS 99%, PI 100%) without contributions from other origins. This gene is the most effective in distinguishing African strains among all single-gene trees. A similar scenario occurs in the ARF-OLE tree, which contains a total of ten strains from Africa in a clade with high support rate (BS 100%, BS 100%, PI 100%). Six African strains clustered in one clade in multilocus trees with high support rate (BS 99%, BS 98%, PI 100%). The H-antigen sequences failed to effectively resolve the systematic classification of African strains. Although the strains are clustered, the groups lack significant support.

As a sample study, we reviewed 50 case reports published from 2023 to 2024 (Suppl. material 1: table S2), involving 50 patients aged between 2 and 83 years, predominantly middle-aged men. Among these patients, 25 had normal immune function, and 25 were immunocompromised. These patients came from various countries, including U.S.A., India, Colombia, Brazil, Venezuela, Bangladesh, Ecuador, Argentina, Nepal, Panama, Cameroon, Nigeria, China, Mexico, Palestine and Australia. Thirty-seven cases were diagnosed as disseminated histoplasmosis, while the remaining cases had infections localized to single organs such as the lungs, trachea, vertebrae, oral cavity, adrenal glands, bones, heart, thyroid, eyes, skin, and tongue. The majority was described with presence of small yeast cells inside macrophages. This indicates that histoplasmosis can have a severe impact on individuals with normal immune function and can affect multiple organs. Based on these 50 case reports, no clinical differences attributable to regional variations in histoplasmosis could be identified. Also for the diseases caused by the North American, Indian (Agrawal et al. 2020) and Indonesian (Baker et al. 2019) genotypes as yet were not reported to differ significantly from classic histoplasmosis. However, strains from tropical Africa are known to cause African histoplasmosis which is characterized by subcutaneous nodules and histopathology comprising large, broad-based budding cells outside macrophages (Develoux et al. 2021). The variety farciminosum is known as the cause of histopathological disease in donkeys (Powell et al. 2006), but no sequence was available to draw a solid taxonomic conclusion.

Kasuga et al. (1999) utilized a multilocus sequence typing (MLST, ARF, OLE, H-anti, TUB1) technique to assess the phylogenetic relationships among 46 Histoplasma strains from different areas, prior to study representing three varieties: vars capsulatum, duboisii, and farciminosum. The results identified six clades: (i) North American H. capsulatum (NAm-1), (ii) North American H. capsulatum (NAm-2), (iii) Central American H. capsulatum (Panama), (iv) South American H. capsulatum group A (SAm-A), (v) South American H. capsulatum group B (SAm-B), and (vi) African H. capsulatum var. duboisii. The concordance among the four gene genealogies suggests that the above six groups have been reproductively isolated from each other for a long time. A few years later, the same authors (Kasuga et al. 2003) introduced additional strains while retaining the original ones, bringing the total number of strains to 137. After the same MLST analysis, the authors redefined eight clades. Three previously recognized clades (NAm-1, NAm-2, Africa) were retained, and clades SAm-1 and SAm-2 were renamed as LAm-A and LAm-B. The previous clade Panama was deleted. Three new clades (Netherlands, Australia, Eurasia) were added. Seven clades should be considered as phylogenetic species, except for the Eurasian clade, as it was considered to have emerged from the largest clade, South American LAm-A. The agent of Egyptian horse disease, classically referred to as H. capsulatum var. farciminosum, was not recognizable as a monophyletic group, as strains clustered in different clades. Histoplasma capsulatum var. duboisii , together with strains previously identified as var. capsulatum, composed a separate African clade. The authors suggested that the three varieties in their classical circumscription are phylogenetically meaningless. Of the seven clades finally distinguished by Kasuga et al. (2003), current data confirmed NAm-1, NAm-2, and Africa-duboisii, all showing concordance in three genes. LAm-A and LAm-B had no concordance in all genes and were not separated, while the Netherlands and Australia (possibly concerning immigrants) were represented by single isolates.

Teixeira et al. (2016) further increased the number of analyzed Histoplasma sequences to 246 and employed various methods of phylogenetic analysis and population genetics. Their study retained the six clades described by Kasuga et al. (2003), while they suggested that the two Latin American clades comprised at least 6 phylogenetic entities. Among these clades, particularly LAm-A exhibited a complex population genetic structure, supporting at least 4 monophyletic clades named LAm-A1, LAm-A2, RJ, and BAC-1. The LAm-B clade could be divided into two highly supported clades, which were geographically restricted to either Colombia/Argentina or Brazil, respectively. Despite observed structured diversity with a strong geographic component in our current data, clade concordance indicated significant gene flow between South American groups. Multiple mechanisms of genetic exchange can be involved, which is a subject for later studies.

Valero et al. (2018) studied African Histoplasma using ITS, RBP1, and OLE genes. In that study, nine strains from Africa clustered in a single clade with a BS support rate of 96%. However, in our ITS tree, these nine strains did not cluster together; only two clustered in the duboisii clade (98% BS), while the remaining seven were scattered in the LAm-A clade.

Rodrigues et al. (2020) utilized data from protein-coding loci (ARF, H-antigen, OLE), rDNA barcoding (ITS), AFLP markers, and mating type analysis to assess the genetic diversity, population structure, and identified distinct phylogenetic species among 436 strains of Histoplasma from around the world but focused on Brazil. The study distinguished three separate groups in South America, namely LAm-C, LAm-D, and LAm-E, all originating from Brazil. In our data, these groups were difficult to recognize.

In addition to MLST, genomic sequence analysis has been introduced as a next level taxonomic approach in Histoplasma. Sepúlveda et al. (2017) performed genome-wide population genetics and phylogenetic analyses with 30 Histoplasma isolates representing four endemic areas but with a strong focus on South America. Four phylogenetic species were formally introduced (Sepúlveda et al. 2024), i.e. Histoplasma capsulatum s. str. (Panama), H. mississippiense (NAm-1), H. ohiense (NAm-2), H. suramericanum (LAm-A), and one African clade without attribution of a name. In our current data, Lam-A sequences were distributed over several clades. Almeida-Silva et al. (2021) sequenced 18 Histoplasma genomes additional to the 30 from earlier research (Sepúlveda et al. 2017) and constructed a genome-level phylogenetic tree. Apart from the four previously recognized clades, this study reinstated LAm-B as an independent clade. In addition, Histoplasma suramericanum was composed of at least two populations, one in the northern part of South America, and another in the southern portion of the continent. Jofre et al. (2022) sequenced genomes of 16 Histoplasma isolated from India and combined those with the 30 genomes of Sepúlveda et al. (2017) to construct a genomic level phylogenetic tree. A new phylogenetic species was recognized in India. This conclusion matches with our barcode and genealogical concordance study, where a clearly separate Asian (India / Indonesia) species was distinguishable.

In summary, populations of Histoplasma show strong regional structuring, which has led authors to distinguish several entities with decreased gene flow. In main traits, all authors agree that Histoplasma comprises several species, while in contrast H. farciminosum is indistinguishable from H. capsulatum / suramericanum. The expansion of the number of studied strains reveals that the sample has less common genetic events, indicating that not all genetically different groups deserve recognition as species. Histoplasma capsulatum/suramericanum is a species that occurs globally, while the regional species H. ohiense, H. mississippiense (Tenório et al. 2024) and H. duboisii (Ameni et al. 2022) are consistently deviating. A separate Asian species (Jofre et al. 2022) might deserve formal recognition.

Histoplasma is an environmental pathogen (Carpouron et al. 2022) that grows on guano of bats, pigeons, and chickens in sheltered habitats. It is designated to infect the local animals, develop an invasive yeast phase inside macrophages, and return to the environment via feces and carcasses (Taylor et al. 2022). Human infections are fulminant in patients with impaired acquired immunity (Damasceno et al. 2019), but disseminated histoplasmosis may also occur in immunocompetent individuals. It has been suggested that the severity of infection partly depends on the size of the inoculum inhaled (Ling et al. 2018). Humans are not the preferred host of the fungus. The sheltered habitats of the fungus explain the fragmented, regionally differing population structure. For many of these populations, clinical differences are being investigated (Damasceno et al. 2019; Jones et al. 2020), statistically significant deviations between groups as yet being unconvincing, with the exception of the variety duboisii. Histoplasma capsulatum is a pulmonary, intracellular disease finally leading to cutaneous pustules, while infections caused by var. duboisii are distinguished by a high propensity for involvement of skin, lymph nodes, and bone. Osseous lesions are typically osteolytic processes that can simulate cancer (Pakasa et al. 2018; Amona et al. 2021; Develoux et al. 2021). Phylogenetically, it is a distinct lineage (Ameni et al. 2022) at short distance from others (Almeida-Silva et al. 2021). The disseminated form of the disease can manifest as skin lesions, which have been confirmed through direct examination in HIV-infected patients (Wahyuningsih et al. 2021). The histological features of infection caused by Histoplasma duboisii, which are also distinctive from those of H. capsulatum, are characterized by large ovoid to globose yeast-like cells measuring 8–15 μm in diameter with broad-based budding. By comparison, there are currently no data supporting a clinical or histological distinction of disease caused by the H. ohiense, or H. mississippiense, which thus could be viewed as cryptic species. Whether or not the Asian genotype is clinically similar to H. capsulatum (Sayal et al. 2003; Krıshnamurthy et al. 2021) needs to be established. The current database (Suppl. material 1: table S2) for distinguishing the clinical, immunological, and ecological characteristics among patients infected with different Histoplasma species may lack sufficient statistical power to distinguish among the three species. A multicenter, prospective study which identifies key epidemiological and clinical variable with isolates collected for whole genomic sequencing, proteomics, mating studies, and host-fungus in vitro properties may be able to further distinguish clinical, physiological, epidemiological, and immunological properties among these different species.

This study primarily focuses on reevaluating previously defined phylogenetic species based on gene sequences. For whole genome phylogeny, three studies have as yet been conducted in Histoplasma, involving a total of 62 genomes. Sepúlveda et al. (2017) distinguished NAm-1, NAm-2, LAm and the African clade as the main entities, similar to the main conclusions in our gene studies. This suggests a broad correspondence of genome analysis and barcoding studies in Histoplasma.

The primary and most significant limitation in the phylogenetic understanding of Histoplasma diversity lies in sample collection. Although the present comparison encompasses a total of 879 strains, only 274 possess sequences for all four genes, with 400 strains having sequences for both ARF and OLE. The absence of sequences for the majority of strains is primarily due to their unavailability in public databases. Currently, only 39 living strains are stored in the CBS fungal collection. Preserving strains in publicly accessible collections will facilitate future retrospective studies. Another important constraint in collecting Histoplasma strains is their classification as Biosafety Level 3 organisms, requiring isolation and cultivation in Biosafety Level 3 cabinets. This poses a challenge for many developing countries and contributes to the significant underestimation of Histoplasma prevalence in Asia and Africa.

Through analysis of multiple gene trees, we found that the Africa, NAm-1, NAm-2, India and Indonesian clades are supported in most gene trees in addition to the main species, i.e. the global species containing the ex-type of H. capsulatum. Most clusters composing the different variants of LAm showed profuse genetic recombination and are therefore judged to represent a single species. For example, sequences in the clusters of LAm-A and RJ are distributed over several clusters throughout the entire tree with the four genes applied in this study. Clusters LAm-D and LAm-E do not form supported clades in all trees. The strains of LAm-B cluster into two well-supported branches in both the ARF and two gene trees, supporting the suggestion of Teixeira et al. (2016) to divide the LAm-B clade into two subclades. Additionally, one of the genome studies (Almeida-Silva et al. 2021) also suggests retaining the phylogenetic species status of LAm B. However, the genealogical concordance analysis indicates that only a subset of LAm-B strains exhibit consistent relationships in ARF and OLE phylogenies. Therefore, we postpone a final decision on LAm-B until after analyzing a larger number of strains with better global representation, and the influence of the bias of South American material is evaluated. As for LAm-C, the strains cluster together in all phylogenetic trees, but without obtaining effective support. Hence, future studies should introduce genome data for LAm C to further determine its classification status.

≡ Monilia capsulata (Darling) Lindner & Knuth – Z. Infektionskrankh. 17: 299, 1916 [n.v.].

= Cryptococcus farciminosus Rivolta & Micelloni – Giorn. Anat. Fisol. Patol. Anim. Dom. 15: 162, 1883; as “criptococcus farciminosus”. Type: Rivolta, Dei Paras. Veg.: fig. 153b, 1873 (lectotype designated here, MBT 10022527); Egypt: isolated from horse with epizootic lymphaginitis, 1983, comm. S.A. Selin, CDC B-3786 [dried culture]. Epitype designated here, CDC B-3786, MBT 10022528). Ex-epitype cultures ATCC 58332, CBS 536.84.

≡ Cryptococcus rivoltae Farmi & Arnch – Centralbl. Bakteriol. Parasitenk. 1 Abt. 17: 597, 1895 (name change).

≡ Saccharomyces farciminosus (Rivolta & Micelloni) Tokishige – Zentralbl. Bakteriol. Parasitenk., Abt. 1, 19: 112, 1895.

≡ Leishmania farciminosa (Rivolta & Micelloni) Galli-Valerio – Centralbl. Bakteriol. Parasitenk., Abt. 1, 44: 577–582, 1909.

≡ Endomyces farciminosus (Rivolta & Micelloni) Nègre & Bouquet – Bull. Soc. Pathol. Exot. 10: 274, 1917.

≡ Parendomyces farciminosus (Rivolta & Micelloni) Mello & L.G. Fern. – Arq. Hig. Pat. Exot. 6: 29, 1918.

≡ Grubyella farciminosa (Rivolta & Micelloni) M. Ota & Langeron – Ann. Parasitol. Humaine Comp. 3: 78, 1925.

≡ Coccidioides farciminosus (Rivolta & Micelloni) Vuill. – Champ. Paras. Myc. Homme Anim. p. 140, 1931.

≡ Torulopsis farciminosus (Rivolta & Micelloni) F.P. Almeida – Ann. Fac. Med., São Paulo 9: 76, 1933.

≡ Histoplasma farciminosum (Rivolta & Micelloni) Redaelli & Cif. – Boll. Sierot. Milan. 10: 851, 1934; as “farcinimosus”.

≡ Zymonema farciminosum (Rivolta & Micelloni) C.W. Dodge – Med. Mycol.: 169, 1935.

≡ Histoplasma capsulatum var. farciminosum (Rivolta & Micelloni) Weeks et al. – Mycologia 77: 969, 1985; nom. inval. (Art. 41.5).

= Histoplasma suramericanum Sepúlveda et al. – mBio 8(6): e01339-17, 13, 2017; nom. inval. (Art. 40.7).

≡ Histoplasma suramericanum Sepúlveda et al. – mSphere 9(6): e00009-24, 11, 2024. Type: CBS 145499, strain 3/11, Guatemala.

Type. Darling, J. Amer. Med. Assoc. 46: 1284, fig. 1, 1906, lectotype designated here, MBT 10022526; Panama: isolated from human with histoplasmosis CBS 145499, metabolically inactive culture preserved in liquid nitrogen, epitype designated here, MBT 10021411. Ex-epitype culture: CBS 145499.

≡ Histoplasma mississippiense Sepúlveda et al. – mBio 8(6): e01339-17: 12–13, 2017 nom. inval. (Art. 40.7). Type: CBS 145498, strain CI#19, Missouri, USA.

≡ Histoplasma ohiense Sepúlveda et al. – mBio 8(6): e01339-17: 13, 2017, nom. inval. (Art. 40.7). Type: CBS 145496, strain Cl#17, Missouri, USA.

= Emmonsiella capsulata Kwon-Chung – Science, N.Y. 177: 368, 1972. Type: USA: BPI 71811, Arkansas, Miller County, isolated from soil samples under bird roosts, K.J. Kwon-Chung (ATCC 22635, ATCC 22636, CBS 136.72, CBS 137.72 – ex-type cultures of opposite mating types).

≡ Ajellomyces capsulatus (Kwon-Chung) McGinnis & Katz – Mycotaxon 8: 158, 1979.

≡ Histoplasma capsulatum var. duboisii (Vanbreus.) Ciferri – J. Amer. Med. Assoc. 2: 342, 1960.

Type. CBS 215.53, isolated from guinea pig previously injected with a strain from human with African histoplasmosis, Congo, R. Vanbreuseghem (RV 4754).

Saccharomyces equi Marcone – Atti Reale Ist. Incoragg. Napoli 8–6: 1–19,1895.

Cryptococcus tokishigei Vuillemin ex Guéguen – Champ. Paras. l’Homme Anim. Domest.: 108, 1907 ≡ Parendomyces tokishigei (Vuillemin ex Guéguen) Mello –Arq. Hig. Pat. Exot. 6: 295, 1918.

Posadasia pyriformis M. Moore – Ann. Missouri Bot. Gard. 21: 347, 1934 ≡ Histoplasma pyriformis (M. Moore) C.W. Dodge – Med. Mycol.: 155, 1935.

The authors have declared that no competing interests exist.

No ethical statement was reported.

All the fungal strains used in this study have been legally obtained, respecting the Convention on Biological Diversity (Rio Convention).

Funding China Postdoctoral Science Foundation (2023M740784).

Conceptualization: YQ. Data curation: YQ, MMT, NL, DM, XZ, JBS, RBL. Formal analysis: YQ. Funding acquisition: DW, YQ. Investigation: YQ. Methodology: TJW, RBL, YQ, AC. Project administration: YQ. Resources: RW, YQ. Software: XZ, YQ. Supervision: DLLH, TJW, SH, DW, SZ, AC. Validation: DLLH. Visualization: DW, SH. Writing - original draft: SH, YQ. Writing - review and editing: TJW, SH, DLLH, DW, RW.

Yu Quan https://orcid.org/0000-0002-3103-3118

Xin Zhou https://orcid.org/0000-0003-2094-7288

Ricardo Belmonte-Lopes https://orcid.org/0000-0002-3122-0271

Na Li https://orcid.org/0000-0002-4713-0705

Retno Wahyuningsih https://orcid.org/0000-0002-3294-5792

Anuradha Chowdhary https://orcid.org/0000-0002-2028-7462

David L. Hawksworth https://orcid.org/0000-0002-9909-0776

Sean Zhang https://orcid.org/0000-0003-1166-2563

Marcus de Melo Teixeira https://orcid.org/0000-0003-1763-3464

Daniel Matute https://orcid.org/0000-0002-7597-602X

Sybren de Hoog https://orcid.org/0000-0002-5344-257X

Dong Wu https://orcid.org/0000-0002-2842-1144

All of the data that support the findings of this study are available in the main text or Supplementary Information.