Evolution of the sex-Related Locus and Genomic Features Shared in Microsporidia and Fungi

Background Microsporidia are obligate intracellular, eukaryotic pathogens that infect a wide range of animals from nematodes to humans, and in some cases, protists. The preponderance of evidence as to the origin of the microsporidia reveals a close relationship with the fungi, either within the kingdom or as a sister group to it. Recent phylogenetic studies and gene order analysis suggest that microsporidia share a particularly close evolutionary relationship with the zygomycetes. Methodology/Principal Findings Here we expanded this analysis and also examined a putative sex-locus for variability between microsporidian populations. Whole genome inspection reveals a unique syntenic gene pair (RPS9-RPL21) present in the vast majority of fungi and the microsporidians but not in other eukaryotic lineages. Two other unique gene fusions (glutamyl-prolyl tRNA synthetase and ubiquitin-ribosomal subunit S30) that are present in metazoans, choanoflagellates, and filasterean opisthokonts are unfused in the fungi and microsporidians. One locus previously found to be conserved in many microsporidian genomes is similar to the sex locus of zygomycetes in gene order and architecture. Both sex-related and sex loci harbor TPT, HMG, and RNA helicase genes forming a syntenic gene cluster. We sequenced and analyzed the sex-related locus in 11 different Encephalitozoon cuniculi isolates and the sibling species E. intestinalis (3 isolates) and E. hellem (1 isolate). There was no evidence for an idiomorphic sex-related locus in this Encephalitozoon species sample. According to sequence-based phylogenetic analyses, the TPT and RNA helicase genes flanking the HMG genes are paralogous rather than orthologous between zygomycetes and microsporidians. Conclusion/Significance The unique genomic hallmarks between microsporidia and fungi are independent of sequence based phylogenetic comparisons and further contribute to define the borders of the fungal kingdom and support the classification of microsporidia as unusual derived fungi. And the sex/sex-related loci appear to have been subject to frequent gene conversion and translocations in microsporidia and zygomycetes.


Introduction
Microsporidia are obligate intracellular pathogens that mainly infect animals including fish and insects, and also some protists and crustaceans [1,2]. In microsporidia, ,150 genera and approximately 1,300 species are known [3]. Thirteen microsporidian species infect humans causing chronic diarrhea in immunocompromised individuals, mainly AIDS patients; in some cases, infection can also occur in otherwise healthy hosts [4]. Microsporidia have a uniquely specialized infection device called the polar tube [5]. The polar spore tube is coiled within dormant spores and when stimulation occurs by encountering and recognizing the host, the polar tube is everted and penetrates the host cell membrane. The polar tube then serves as a conduit for delivery of the infectious material, the sporoplasm. Micro-sporidial cells lacking a cell wall replicate inside the host and produce meronts, which eventually form mature spores that are released from the host.
The phylogenetic placement of microsporidia has long been debated [6,7]. Originally, microsporidia were placed within an artificial group, the schizomycetes (reviewed in [7]). They were then long considered to be related to other spore forming parasites, but due to the perceived lack of mitochondria, microsporidia were eventually considered to be an ancient eukaryotic lineage [8,9]. Phylogenetic analysis based on protein coding sequences then aligned the microsporidia within the fungal kingdom [10,11,12,13,14,15], and the lack of mitochondria was soon undermined by the discovery of a reduced ''mitosome'' in microsporidia [8,16]. In addition, recent findings that microsporidian genomes harbor two syntenic ribosomal genes (RPL21 and RPS9) that are also syntenic throughout the fungi (with a few exceptions, such as Schizosaccharomyces pombe), supports a fungal origin. The RPL21-RPS9 synteny is only found in microsporidia and fungi [17,18] and not in other opisthokonts or other eukaryotic groups, where these genes are unlinked.
Within the fungi, molecular phylogenetic analyses have suggested a relationship between microsporidia and either ascomycetes and basidiomycetes [19,20] based on some genes, or with zygomycetes based on other genes [21,22]. A phylogenetic relationship with the zygomycetes is also supported by the findings that microsporidian genome architecture is similar to that of zygomycetes, and strongly differs from any other extant fungal phyla [17]. One of the shared loci is of particular potential interest because it may represent a sex locus in microsporidia. This microsporidian sex-related locus is highly similar to the sex locus of the zygomycetes: both contain genes for a triose phosphate transporter (TPT), a high-mobility group (HMG) protein, and an RNA helicase. In zygomycetes, the sex locus orchestrates and regulates sexual reproduction [17,23,24,25,26,27]. This synteny is unique to microsporidia and zygomycetes and absent in other fungi whose genomes are available, including chytridiomycetes, ascomycetes, and basidiomycetes. Therefore, the microsporidian sex-related locus could be involved in an extant sexual cycle, if it is indeed homologous. Furthermore, several recent studies identified the existence of the meiosis specific genes for 'core meiotic recombination machinery' including Spo11, Rad50/Mre11, Dmc1, Rad51, and Mlh1 in the microsporidian genomes [17,28], which could be an indication of the possible existence of sex in microsporidia [29,30]. However, bona fide sexual development of microsporidia has not been reported although there are some observations and inferences about the possibility based on morphological approaches [31,32].
Here, we have examined the genomic architecture of microsporidia to identify further conserved characteristics to help determine the relationship between microsporidia and other opisthokonts. We have identified unique genomic characteristics shared only in microsporidia and fungi rather than other lineages in the opisthokonts and more divergent eukaryotic lineages that contribute to define the boundaries of the fungal kingdom by features independent of sequence based phylogenetic approaches. We have also analyzed the sex-related locus from multiple isolates of Encephalitozoon species to test for the presence of idiomorphic HMG genes as observed in the zygomycete sex locus, in which divergent sexP and sexM genes are encoded by the sex locus of (+) and (2) mating type strains, respectively. Further, we analyzed the TPT, HMG, and RNA helicase genes to determine the phylogenetic relationship with the homologous genes of zygomycetes. We discuss the evolutionary trajectory of the sex locus within the basal fungal lineages for the zygomycetes and microsporidia.

Results and Discussion
Genome structure data support the Microsporidia as fungi Virtually all available evidence now supports a phylogenetic relationship between microsporidia and the fungi, but whether they are within or sisters to the fungi remains contentious, as does exactly what fungal lineage they might be most closely related to. Individual gene phylogenies place the microsporidia close to the ascomycetes, basidiomycetes, or zygomycetes [19,20,22,33], whereas combined tubulin phylogenies place them with the zygomycetes [21] and a four gene phylogeny suggests that microsporidia are related to Rozella [15]. We have found that microsporidian genomes share an overall higher degree of gene order conservation with zygomycetes than with any other fungal lineage that has been well sampled at the genomic level. Interestingly, one conserved gene pair found in microsporidia (RPS9-RPL21) was also identified in all known fungal genomes except that of Schizosaccharomyces pombe (Figure 1 and Figure S1) [17,18,34]. To test whether this gene pair is a stable marker for fungi, we searched more broadly within the fungi and other opisthokonts. Three other Schizosaccharomyces fission yeast species, S. japonicus, S. octosporus, and S. cryophobus were also found to lack this gene pair, and interestingly the chytrid Spizellomyces punctatus was also found to lack the RPS9-RPL21 pair (Figure 1), and in these species homologs for both genes are present but unlinked. Thus, independent chromosome translocations might have occurred in the two fungal lineages that unlinked the two ribosomal genes in these exceptional species in the fungal kingdom.
Searching for similar genomic markers revealed two other highly conserved characteristics that help distinguish microsporidia from non-fungal opisthokonts. First, a gene encoding a fusion protein between ubiquitin and the ribosomal S30 protein is found in choanozoans as well as metazoans [35], whereas in all fungal lineages, including microsporidia, these two domains are present as separate genes that are unlinked ( Figure 1 and Figure S2). Similarly, a glutamyl-prolyl tRNA synthetase fusion protein is also found in metazoans, choanoflagellates and the filasterean Capsaspora [36]. Here, two tRNA synthetase domains are linked by an RNA binding domain that regulates gene expression through controlling translation [37,38]. We found no such fusion gene in the fungal genomes analyzed, including microsporidia, zygomycetes, chytridiomycetes, ascomycetes, and basidiomycetes ( Figure 1 and Figure S3). Thus, these two fusion proteins appear to have arisen within the opisthokonts, after the divergence of the ancestors of fungi and metazoans. While these fusions do not suggest a specific fungal lineage that is particularly close to microsporidia, they do support the specific relationship between microsporidia and fungi as a whole to the exclusion of other major opisthokont lineages.

Conservation of sex-related loci in microsporidia
Among the many gene pairs found to be conserved between microsporidia and zygomycetes in our previous study, one of particular interest was a putative sex-related locus in microsporidia [39]. This locus, which comprises an HMG domain-containing gene flanked by RNA helicase and triose phosphate transporter (TPT) genes, is similar in architecture to the sex loci of the zygomycetes Mucor circinelloides, Phycomyces blakesleeanus, and Rhizopus oryzae. In microsporidia, an additional hypothetical protein is found downstream of the TPT gene, and the E. cuniculi sex-related locus also contains a second novel gene with limited identity to HMG genes (weak HMG protein gene) ( Figure 2). Interestingly, the sex-related locus of the recently sequenced Nosema ceranae genome is similar to that of Antonospora locustae in that the RNA helicase gene is not linked to the HMG gene in either genome ( Figure 2). Molecular phylogenetic analysis consistently shows Encephalitozoon is more closely related to Nosema than either are to Enterocytozoon, so this pattern suggests the locus was present in the ancestor of Nosema. Moreover, in N. ceranae a hypothetical protein that is found elsewhere in the genomes of all other microsporidia investigated is found in place of the hypothetical protein linked to the HMG gene. Overall, the greater diversity of this locus now apparent suggests that gene conversion and translocation may have occurred frequently around/within the sex-related locus in microsporidia. This variation is also found in the sex locus of three zygomycetes, where the orientation of the TPT and SexP/M genes differs and there is a repetitive element or additional ORF found in the P. blakesleeanus and R. oryzae sex locus, respectively [17,23,40].

Variation in the putative sex-related locus of microsporidia
If this region is a sex locus in microsporidia, we might expect to find variation corresponding to distinct sex alleles. To test for the presence of such variation, we compared the sequences of sexrelated locus genes for the presence of multiple paralogs, and also characterized the putative sex-related locus from three distinct species and multiple strains in the genus Encephalitozoon.
Characterization of the sex-related loci of E. intestinalis and E. hellem revealed an overall architecture identical to that of E. cuniculi: TPT, HP, HMG, weak HMG, and RNA helicase gene ( Figure 2). The overall sequence conservation between homologous genes in these three species was also very high (for example see Figure 3). All Encephalitozoon isolates were found to be highly conserved (i.e., not an opposite mating type. See below) and the HMG proteins in all Encephalitozoon species as well as A. locustae have two HMG domains in the HMG protein ( Figure 3) [17], whereas in the HMG proteins in N. ceranae and E. bieneusi only one HMG domain was found (data not shown) [17]. The HMG proteins have ,90% identity between the three Encephalitozoon species, whereas the weak HMG domain proteins have relatively lower alignment scores (,45% identity between E. cuniculi vs. E. intestinalis, ,46% identity between E. intestinalis vs. E. hellem, and ,51% identity between E. cuniculi vs. E. hellem). Comparison of the hypothetical proteins showed .78% identity between each species (data not shown). Thus, only the weak HMG domain protein was shown to be more diverged than the other proteins encoded by the sexrelated locus between the three species. However the similarity is much higher than between SexP and SexM in zygomycetes. Given that the SexP and SexM proteins share only 14% and 20% similarity in M. circinelloides and P. blakesleeanus, respectively, the level of variation in microsporidian HMGs does not support the presence of idiomorphic HMG/weak HMG genes in the Encephalitozoon species complex.
To examine the level of variation within species for possible allelic heterogeneity, the sex-related locus was characterized from eleven genetically distinct E. cuniculi isolates and three strains of E. intestinalis from different host environments. The sex-related locus of the three E. intestinalis isolates was also very similar (.99% identity), as were all eleven isolates of E. cuniculi.
Given the level of sampling at least for E. cuniculi, it seems unlikely that multiple distinct sex type alleles exist with equal frequency at this locus, so if E. cuniculi sexual development occurs, is the sex-related locus involved and if so how? One possibility is that an opposite mating type allele/idiomorph is infrequent or rare, as is in case in the human pathogenic basidiomycete Cryptococcus neoformans [41]. A second possibility is that E. cuniculi is homothallic (self-fertile) [42] and sex might involve unisexual reproduction, similar to C. neoformans [43,44,45] or Candida albicans [46,47]. The E. cuniculi HMG protein contains two HMG domains whereas zygomycete SexP and SexM have only one HMG domain ( Figure 3) [17]. The two HMG domains may function separately and play the equivalent roles of SexP and SexM. This does not, however, explain the single HMG-domain containing proteins of N. ceranae and Enterocytozoon bieneusi. Alternatively, the weak HMG protein in the Encephalitozoon sex-related locus [17] (Figure 2) could play functions equivalent to SexP and SexM, but once again this protein was not observed in the other microsporidia.
Linking the sex-related locus to sexual development in microsporidia As of yet, there is no direct evidence linking the sex-related locus to sex determination or sexual reproduction in the microsporidia. The types of evidence that would be necessary to assign such a functional role could include a demonstration of opposite alleles or idiomorphs present in different isolates. We have sequenced the sex-related locus in eleven different E. cuniculi isolates, however, and found them to be highly conserved with the exception of a few point mutations, which tellingly are restricted to the flanking genes and are not found in the HMG or weak HMG genes (Figures 3  and 4).
Of course the zygomycete sex locus is just one strategy of many used in fungi. Given the independent data favoring a zygomycete origin of microsporidia, both from genome structure and some molecular phylogeny, it is the most reasonable system to search for initially, but others should also be considered if the sex-related locus proves not to be related to sex after all. For example, basidiomycetes utilize homeodomain (HD) transcription factors to orchestrate and regulate sexual development, such as the Ustilago maydis b alleles that encode two divergently transcribed HD1 and HD2 class proteins (reviewed in [48]), or the Coprinopsis cinerea A locus that also encodes two divergently transcribed homeodomain proteins [49]. In Cryptococcus, the a and a mating type locus encodes different classes of HD proteins, in which the MATa allele encodes only an HD2 factor (Sxi2a) while the MATa allele encodes only an HD1 factor (Sxi1a), in contrast to other basidiomycete MAT loci [50]. It would be worthwhile to investigate whether microsporidia have a pair of HD genes that are idiomorphic between isolates, as in U. maydis, or two functionally different HD in the same locus, as in different isolates as in Cryptococcus. We investigated three homeodomain gene clusters described previously (six HD genes: ECU03_0600, ECU03_0610, ECU04_0970, ECU04_1030, ECU10_1470, and ECU10_1480) [17,51] in three E. cuniculi isolates that were previously proposed to represent candidate mating type loci [52]. Once again, however, no evidence for distinct alleles or idiomorphs was apparent from sequence analysis across the three homeodomain gene clusters (GenBank accession at HM049491 to HM049502) ( Figure S4).
Although there is some morphological data that has been interpreted to suggest that some microsporidian species may undergo sexual reproduction [31,32], this is not known for any of the six species in the current analysis. The type of evidence that would be necessary to show this definitively includes a demonstration of marker exchange (recombination) following co-infection of distinct isolates, or the finding that ploidy changes occur in the population (such as the finding of diploid or dikaryon isolates, or isolates heterozygous for genomic markers). Finally, to provide evidence linking the sex-related locus to sexual reproduction will require, for example, documentation that the genes therein are expressed at an appropriate time in the life cycle, or to show a candidate protein binds physically to the promoters for meiotic gene homologs by chromatin immunoprecipitation studies. These and other studies are ongoing to test whether sexual reproduction occurs in E. cuniculi, and whether the sex-related locus participates in this process.
Oblique ortholog assignment between the sex-related/ sex locus genes of microsporidia and zygomycetes The sex-related locus in the E. cuniculi genome was originally identified using Blast searches with the sexP gene from the functionally-defined sex locus from Phycomyces blakesleeanus [23]. This revealed the homologous E. cuniculi HMG gene that is unique in the genome, and the region around this gene was then subjected to manual annotation and inspection, revealing the presence of a flanking triose phosphate transporter (TPT) gene and an RNA helicase gene homologs, strikingly similar to the organization of the sex locus in P. blakesleeanus, and also in Mucor circinelloides and Rhizopus oryzae. Here we address whether the genes flanking and within the sex and sex-related loci are orthologs or paralogs.
With respect to the HMG gene contained in the sex-related locus, while Blast searches with P. blakesleeanus SexP identify the E. cuniculi gene as homologous, reciprocal Blast searches with the E. cuniculi gene against other fungi return a variety of related HMG proteins in the R. oryzae genome. Construction of phylogenetic trees based on either the isolated HMG domains or the full-length proteins suggests that the high rate of divergence makes identifying orthologs problematic. Indeed, no microsporidian genes that are clear candidates to be orthologous with either SexP/M were found, although one domain of the E. cuniculi sex-related locus gene did branch with the M. circinelloides SexM with no bootstrap support. Further, there was low bootstrap support for virtually all nodes in the tree ( Figure 4A). In fact, the zygomycete sex genes do not form a clade even though there is no question that the sexM and sexP genes are paralogous given the level of sequence identity and functional analysis ( Figure 5A). Similarly, the microsporidian HMG domains (also including both domains from the dual-HMG genes) do not form a clade either, even though they are found in nearly identical genomic contexts and the orthology of the microsporidian TPT and RNA helicase genes is not in question (see below). In reciprocal Blast searches, there is a second HMG domain gene in the P. blakesleeanus genome that does share a modestly higher level of identity with the microsporidian HMG domain gene in the sex-related locus (jgi|Phybl1|79113|estExt_ fgeneshPB_pg.C_220136). The HMG domain gene family is very divergent and subject to accelerated rates of change in the case of the sex locus. In addition the evolutionary trajectory for an HMG domain determinant in the zygomycetes, which are heterothallic and have two opposite mating type genes (sexM and sexP), may be very different in a hypothesized homothallic species that may have a single sex determinant. It may simply be the case that this is an example of syntenic orthologs that do not share the highest level of identity with the syntenic partner, a problem for which there are precedents in other fungi.
We also performed phylogenetic analysis of the microsporidian TPT, which identified another gene in the zygomycete genomes . Alignments for HMG and weak HMG proteins in Encephalitozoon species. Alignment for the HMG proteins shows the HMG gene is highly conserved in eleven E. cuniculi, three E. intestinalis, and one E. hellem isolates suggesting divergent HMG genes are absent in this sample collection in contrast to the zygomycete sex locus which has sexP and sexM alleles. Tandem HMG domains are observed across the three species (highlighted boxes). Alignment for the weak HMG proteins also displays a relatively high level of conservation across the three species. doi:10.1371/journal.pone.0010539.g003 that is more closely related compared to the sex locus linked TPT gene ( Figure 5B). In maximum likelihood analysis, the zygomycotan sex-linked TPT genes were also closely related, so we find it hard to conclude with much certainty which of these genes are orthologous.
There are multiple paralogs of the RNA helicase present in microsporidian genomes, and again phylogenetic analysis reveals that the RNA helicase genes linked to the sex locus and the sexrelated locus are also either paralogs or highly derived orthologs ( Figure 5C). The microsporidian RNA helicases of the sex-related locus group in a clade that is distinct from the zygomycotan RNA helicases of the sex locus. However, within that clade, the support for the position of the proposed orthologs is extremely low. In fact, the entire clade is unsupported, so there is no evidence that the microsporidian helicases of the sex-related locus are more closely related to the ones proposed in Figure 5C than to any zygomycete helicase in this family. Overall, therefore, both the TPT and RNA helicase genes flanking the sex or sex-related locus are either highly derived orthologs or paralogs.
It is interesting to ask, if the RNA helicase and TPT genes are not orthologous to those in the zygomycete sex locus, why are the microsporidian, especially E. cuniculi/intestinalis/hellem and E. bieneusi, HMG genes also surrounded by a TPT and RNA helicase gene? The functional and genomic architecture of the microsporidian sex-related locus is highly similar to the zygomycotan sex locus, even though the sequences of the genes are highly derived. Moreover, this architecture (TPT/HMG/ RNA helicase or TPT/HMG gene cluster) is conserved only between these two groups and is not shared with other fungal phyla or representative outgroups, including the choanoflagellate Monosiga brevicollis and the filasterea Capsaspora owczarzaki (see also reference [17]).
The question, therefore, is whether there is a functional relationship between these three genes in the microsporidian sexrelated locus and those in the zygomycete sex locus? There are several possible interpretations of these findings. First, the three genes may represent positional orthologs that are rapidly diverging as a result of gene duplication and conversion ( Figure 6) and therefore appear less closely related. Gene duplications tend to occur locally, leading to linked paralogs. Gene loss can then result in paralog-ortholog syntenic gene pairs compared to orthologortholog gene pairs. There are clear examples of this throughout the Ascomycota in which positional information has been necessary to correctly assign orthologous relationships [53]. Given the nature of the mating type locus and genes resident therein as rapidly evolving [54,55,56,57], it is likely that similar criteria will be necessary here as well. There are also clear cases in hemiascomycetous fungi in which true orthologs can be defined by synteny, though they are not always the gene pair that shares the highest level of identity in pairwise comparisons of the two genomes. One striking example involves the six kinesin genes in Saccharomyces cerevisiae compared to Ashbya gossypii. There is another important example for this in the microsporidia. A cellular hallmark of all microsporidia is the presence of a polar tube. The proteins that compose the tube (PTP1, 2 and 3) are extremely divergent between members of the group, to the point that synteny is actually the only available way (outside antibody precipitation) to identify and annotate these proteins in newly sequenced genomes [58]. In N. ceranae, two of the PTP proteins only share 16.7% and 19.6% percent identity, respectively, when aligned with their putative E. cuniculi orthologs. Thus, it is simply not possible to assign the correct orthologous relationships by Blast searches, but syntenic position information can reveal which are orthologs based on conservation in the flanking genes [53]. Another striking example of difficulty in assigning orthologs based on sequence similarity is the case of the BAR1 gene in Candida albicans and S. cerevisiae. The Bar1 protein encoded by BAR1 is an aspartyl protease that degrades the alpha factor pheromone in S. cerevisiae [59]. In studies by Schaefer et al., Blast searches of the C. albicans genome with S. cerevisiae Bar1 revealed 14 homologous proteins, and the 12 th protein, based on the sequence comparison % similarity, was found to be the bona fide Bar1 ortholog based on functional studies. The 11 paralogs that share greater similarity than the bona fide ortholog are members of the SAP protease family [46,60].
Second, the HMG genes may be orthologs and one or both of the flanking genes may share a paralogous relationship (see Figure 7). The promoter of the TPT gene is part of the M. circinelloides sex locus, indicating that these genes lie at the junction spanning both the common and diverged regions of the genome at the border of the sex locus. From detailed analyses of the mating type locus of Cryptococcus, it is clear that genes can be evicted from this locus, and genes which are quite divergent and clearly part of MAT are fixed as one of the two paralogs (IKS1, NMC1, BSP3) in other closely related lineages [61,62]. Moreover, gene duplication and conversion events can occur within MAT which change the phylogenetic relationship of the resident genes ( Figure 6). Thus, one possible scenario is that the TPT, HMG, and helicase genes were all part of an ancestral locus, and that one or the other paralog for the TPT and helicase gene was fixed and linked to the HMG gene in two descendent lineages. Sequence analysis of other basal fungi will allow this model to be tested in further detail. This notion has some interesting implications developed in more detail below (see Figure 7).
Third, the three genes of the sex-related locus identified in the microsporidia may each be paralogous with the genes in the sex locus of zygomycetes. This model is the most straightforward in terms of the available sequence data, but it invokes the convergent Figure 4. sex-related locus comparison between E. cuniculi isolates. The sex-related locus of EC1 and the reference isolate share 100% sequence identity, whereas the EC2 and EC3 sex-related locus has four synonomous (black asterisks) and three nonsynonomous (red asterisks) base substitutions (I38V in the hypothetical protein and C109S and I197M in the RNA helicase). These limited base changes do not support assignment as idiomorphs. doi:10.1371/journal.pone.0010539.g004 evolution of two similar gene clusters including a TPT transporter, an HMG domain protein, and an RNA helicase. Remembering that genome structure and some phylogenetic analysis both support a close relationship between microsporidia and zygomy-cetes [12], it seems unlikely that a gene cluster with such functional significance in zygomycetes would also assemble by chance independently in their sisters, the microsporidia (see supplementary discussion File S1).  . Evolutionary trajectory of the sex locus within zygomycetes and microsporidia. An ancestral sex locus might have spanned the TPT and RNA helicase genes. Thus, two divergent TPT, HMG, and RNA helicase genes were present in the two sex-alleles. In the microsporidian lineage, only one allele was retained. In the zygomycetes, local recombination might have fixed the TPT and RNA helicase genes resulting in one allele for these genes, whereas two alleles of the HMG genes remain in the two opposite mating types. doi:10.1371/journal.pone.0010539.g007 A model for the evolution of mating type loci by gene capture and eviction A P. blakesleeanus TPT gene unlinked to the sex locus is present in the genome that may be more closely related to the TPT gene linked to the sex-related locus of microsporidia. Again this may simply be a case in which the gene is rapidly evolving and sequence similarity is not sufficient to correctly assign the orthologous relationship. However if we assume that the TPT genes linked to the sex and the sex-related locus are paralogous, it leads to an interesting model for the evolution of the mating type locus (Figure 7). The assignment of the sex locus in P. blakesleeanus and M. circinelloides has been definitively established based on functional approaches that show the sex genes control sexual identity and orchestrate sexual reproduction [23] (Lee, Gryganskyi, Li, Vilgalys, and Heitman, unpublished data). One interesting feature involves the border of the sex locus and the TPT gene [17]. While the TPT gene and its promoter flank the sex locus in P. blakesleeanus, the ORF of the TPT gene lies in the region flanking the sex locus in M. circinelloides, and the promoter lies within the sex locus, and is thus sexually dimorphic. Thus, the TPT gene straddles the border of the sex locus, similar to the amelogenin gene at the border between the pseudoautosomal and sex-specific regions of the mammalian sex chromosomes [63].
This raises the possibility that in a last common ancestor, and possibly also in some extant species, the TPT gene promoter including the ORF was a bona fide component of the sex locus ( Figure 7). There are clear examples throughout the fungal kingdom of genes that have been captured into the mating type locus, lost from the mating type locus, or evicted from the mating type locus [64,65,66,67,68,69]. A striking example of this last process is found in the detailed comparisons of the mating type locus in the Cryptococcus neoformans and Cryptococcus gattii lineages (serotypes A, D, and B that have diverged over ,40 million years of evolution) [61,62]. In serotypes A and B, three genes (IKS1, NCM1, and BSP3) are clearly part of the mating type locus, present in both the a and a alleles, and sex-specific. In contrast, in the serotype D C. neoformans var. neoformans lineage, these three genes have been evicted from MAT by a gene conversion event, and fixed as the a allele in the 39 flanking region of the MAT locus. Thus, these three genes were within MAT in the last common ancestor, and remain within MAT in two extant lineages, but are no longer within MAT in a related diverged lineage.
Based on this analogy, it is possible that the TPT gene was part of an ancestral sex locus, but then evicted by similar gene conversion events (Figure 7). If one TPT allele was fixed in the zygomycete lineage as part of the sex locus and the other allele was fixed in the microsporidian lineage but evicted from MAT, then the observed situation would arise. To test this model additional genome or sex and sex-related loci must be characterized from zygomycete and microsporidian lineages, and these studies are in progress.
Based on this model, at one point there may have been that an ancestral sex locus that contained two HMG domain genes, similar to heterothallic to homothallic transitions that occur commonly in other fungi. One or the other was lost or transposed elsewhere in the genome in a transition to heterothallism. As of yet it is unknown how the two HMG domains in the microsporidia arose, but each of the two domains appears to be more closely related to SexP than to SexM of the zygomycetes. Thus, this might have resulted from a SexP-SexP gene fusion that subsequently underwent accelerated evolution, and its orthologous relationship to the ancestral gene is now less clear based solely on percent identity comparisons for genes in this large, divergent family of transcriptional regulators.
Currently these comparisons are restricted to speculation by the paucity of genomic information from basal fungal lineages, currently limited to three Mucorales and two chytrid species, and from microsporidia, currently limited to three genomes (E. cuniculi/intestinalis/hellem, the last two of which are in progress to be released) and genome survey/draft genomes (A. locustae, N. ceranae, and Octosporea bayeri) [39,70,71,72]. As additional whole genome analyses are reported these hypotheses can be tested more directly, and it is also possible that species with currently unsampled genomes may emerge as even more closely related to the microsporidia than are the Mucorales.

E. cuniculi cultures and genomic DNA extraction
Three E. cuniculi strains, EC1, EC2, and EC3 (Table 1), were kindly provided by Dr. Louis Weiss. These strains were isolated from the kidney of an infected rabbit, mouse, and dog, respectively and represent three genotypes of E. cuniculi [73,74]. The strains were maintained in RK13 (rabbit kidney) cells. RK13 cells were grown in MEM media containing 7% FBS supplemented with penicillin-streptomycin (Invitrogen, Co.) as described previously [75]. Monolayers of RK13 cells were subject to E. cuniculi infection. The MEM media was changed twice a week and spent media was collected to accumulate spores. Genomic DNA from EC1, EC2, and EC3 was extracted as described [75]. Encephalitozoon intestinalis and Encephalitozoon hellem (Table 1) previously harvested from humans were maintained in RK13 cells [74,76,77]. Spores were isolated by sequential washing with dH 2 O, TBS-Tween 20 (0.3%), and TBS, followed by Percoll TM centrifugation, a final wash with TBS-SDS (0.1%), and genomic DNA was extracted using established protocols [78].

Identification and sequencing of the sex-related locus
The sex-related locus from the EC1, EC2, and EC3 isolates was amplified with high fidelity Taq polymerase (Roche) using primers, JOHE20578-CCGGTGTTCATCCTTCTGTT and JOHE-20579-GCACGTCTCACAGTTGACCA. The amplicons were sequenced with primers JOHE20578, JOHE20579, JOHE20720-CAGTAAAAAGGCGCAAGGAC, and JOHE20721-CTAGA-ATGCGCCCCATACAT. Three independent PCR reactions were performed and analyzed. The sex-related locus of E. intestinalis and E. hellem was identified through in-depth genome surveys. This locus was samples in all isolates analyzed here using the following ''forward'' primer sets: TPTF1-TGTGCAGAGTTTACGCTCGTT; TPT-F2-TGAACTATGTTGGGCTGACCATAA; TPTF3-CTGACC-ATAAGCATTGCAGGAAT in combination with the following reverse primers: HelicaseR1-GGCTTGACATTCCGTATCTC-GAC and HelicaseR2-TACGACCTATGTGACGAGAGAACAT The sex-related locus of N. ceranae and RPL21-RPS9 synteny were identified by Blast with the E. cuniculi TPT, HMG, and RNA helicase genes using Formatdb against a local copy of the NCBI nr database.

Annotation and phylogenetic analyses
Sequences were annotated by using FGENESH or ORFfinder (NCBI). For phylogenic analyses for the sex-related locus components, deduced amino sequences were aligned with ClustalW and the alignment results were manually inspected and corrected if needed. Maximum likelihood trees (bootstrap number = 100) were generated by using the PhyML 3.0 software [79].

Supporting Information
File S1 Calculation of the probability of convergence to similar gene clusters.