An Expanded Inventory of Conserved Meiotic Genes Provides Evidence for Sex in Trichomonas vaginalis Authors: Shehre-Banoo Malik, Arthur W. Pightling, Lauren M. Stefaniak, Andrew M. Schurko, John M. Logsdon Published: 2008, PLOS ONE DOI: 10.1371/journal.pone.0002879 ABSTRACT Meiosis is a defining feature of eukaryotes but its phylogenetic distribution has not been broadly determined, especially among eukaryotic microorganisms ( i.e. protists)—which represent the majority of eukaryotic ‘supergroups’. We surveyed genomes of animals, fungi, plants and protists for meiotic genes, focusing on the evolutionarily divergent parasitic protist Trichomonas vaginalis . We identified homologs of 29 components of the meiotic recombination machinery, as well as the synaptonemal and meiotic sister chromatid cohesion complexes. T. vaginalis has orthologs of 27 of 29 meiotic genes, including eight of nine genes that encode meiosis-specific proteins in model organisms. Although meiosis has not been observed in T. vaginalis , our findings suggest it is either currently sexual or a recent asexual, consistent with observed, albeit unusual, sexual cycles in their distant parabasalid relatives, the hypermastigotes. T. vaginalis may use meiotic gene homologs to mediate homologous recombination and genetic exchange. Overall, this expanded inventory of meiotic genes forms a useful “meiosis detection toolkit”. Our analyses indicate that these meiotic genes arose, or were already present, early in eukaryotic evolution; thus, the eukaryotic cenancestor contained most or all components of this set and was likely capable of performing meiotic recombination using near-universal meiotic machinery. INTRODUCTION Meiosis is a necessary part of sexual reproduction and a hallmark of eukaryotes that distinguishes them from prokaryotes, yet we are only beginning to understand its origin and evolution. Recent work has revealed that many meiotic genes are conserved not only among animals, fungi and plants (AFP) and some eukaryotic microorganisms (protists), but also in the putatively early-diverged protist Giardia intestinalis [1] which is not known to be sexual per se but was recently shown to have genetic recombination [2] , [3] and to use orthologs of meiosis-specific genes in putatively parasexual recombination processes [4] . The breadth of eukaryotic diversity lies among the protists [5] , [6] , yet much remains to be elucidated about their meiotic machinery [1] , [7] , [8] . Thus, we have continued and expanded our search for conserved meiotic genes in public databases and particularly in the recently completed genome of Trichomonas vaginalis [9] , a member of the Parabasalia. Parabasalids are a highly diverged eukaryotic lineage in which the molecular mechanisms of meiosis are unexamined; they are related (albeit distantly) to diplomonads ( e.g. Giardia ) [10] – [17] . Morphological and molecular phylogenetic data, while controversial in details, divide parabasalids into two groups, the hypermastigotes (symbionts of roaches and termites, e.g. Trichonymphida) and the trichomonads (parasitic and free-living flagellates, e.g. Trichomonadidae) [18] – [23] . T. vaginalis is sexually transmitted between people's urogenital tracts, and acute infections are associated with increased risk of pelvic inflammatory disease, HIV-1 infection, infertility and problems with pregnancy [24] , [25] . T. vaginalis is estimated to cause 174 million new infections annually worldwide and is the most common non-viral sexually transmitted human pathogen [25] , [26] . Metronidazole is commonly used to treat T. vaginalis infections, but resistance to the drug is increasing [27] . It is not known whether genetic exchange occurs in populations of T. vaginalis ; however, genetic exchange could mediate the proliferation of drug-resistant mutations or increased virulence in populations of the parasite. While neither meiosis nor sex has been observed in Trichomonas or other trichomonads, various observations suggest the presence of sexual processes in Parabasalids. Using light microscopy, Cleveland described insect-hormone-induced divisions in hypermastigotes as one-step meiosis, and suggested that this was a more primitive form of meiosis than typical two-step meiosis in AFP [28] – [30] . For trichomonads, the finding of six genetically identical strains of T. vaginalis with clonal population structure was taken as evidence against meiotic recombination [31] . However, recent phylogenetic analyses of 731 polymorphic molecular markers show genetic variation among 20 strains of T. vaginalis that may be sufficient to indicate meiotic recombination [32] . Similar analyses reveal that closely-related T. vaginalis strains shared the phenotype of resistance to metronidazole, but this pattern had no correlation with geographical origin [33] , suggesting the genetic spread of resistance by recombination and strong selection. The recent identification of mariner , Maverick and other DNA transposons in high copy number [34] – [36] in the highly repetitive >160 Mb genome sequence of T. vaginalis strain G3 [9] , [37] could be the result of a predominantly asexual mode of reproduction or the recent loss of sexual reproduction [38] . In contrast, the presence of intact retroposons and reverse transcriptase homologs in the T. vaginalis strain G3 genome sequence [9] is consistent with an expectation that such elements, that are predominantly vertically transmitted, are only maintained in sexual lineages [39] . Although meiosis was not observed in extensive cytological studies of cell division in T. vaginalis [40] , [41] , it has been noted that its six chromosomes may be synapsed in 0.1% of cells, suggesting that meiosis may occur transiently in lab populations [42] . Quadrinucleated cells transiently observed in lab populations were noted but dismissed as not indicating a developmental stage in T. vaginalis [43] . Thus, with little direct evidence for meiosis in T. vaginalis , an inventory of meiotic genes will be an informative tool with which to assess its ability to undergo sexual processes. Meiosis remains to be described at the molecular level in parabasalids and in most other protist lineages. Since parabasalids could represent one of the earliest diverging lineages on the tree of eukaryotes [11] – [14] and they may employ non-canonical meiosis, their meiotic processes could represent an ancestral state. Thus, an understanding of the molecular mechanisms underlying meiosis in these protists is important. Comparative studies of meiotic machinery (i) may indicate the presence of sexual reproduction in recent ancestors of organisms that are sexual or truly asexual ( vs. just facultatively sexual), (ii) could indicate the absence of sex, (iii) will be useful for studies of the evolutionary advantages of sex, and (iv) can provide data that are valuable for ecological and epidemiological studies [44] . Surveys of meiotic genes have not been performed in most protists, making a comparative analysis of meiosis incomplete due to the limited available gene sequence data from diverse protists [45] . Thus, the universality of meiotic machinery in eukaryotes remains an open question. We surveyed the genomes of Trichomonas vaginalis (strains G3 and NIH-C1) and other diverse eukaryotes for a previously-described set of 17 conserved meiotic genes [1] and 12 additional meiotic genes also conserved among eukaryotes ( Table 1 and Figure 1 ). Our search included 9 genes that are “meiosis-specific” since they are only known to function in meiosis in AFP and thus hypothesized to only be present in organisms with sexual ancestry ( Spo11 , Hop1 , Hop2 , Mnd1 , Dmc1 , Msh4 , Msh5 , Mer3 , Rec8 ). We also surveyed 20 additional genes whose products are required for meiosis in AFP but also have general functions in DNA repair or mitosis ( Mre11 , Rad50 , Rad1 , Rad52 , Rad51 , Msh2 , Msh6 , Mlh1–Mlh3 , Pms1 , Smc1–Smc6 , Rad21 , Scc3 , Pds5 ). We searched public databases to find homologs of meiotic proteins in diverse eukaryotes with complete (at least 7× coverage) genome sequence. Completed and near-complete genome sequences include AFP and representative apicomplexan, ciliate, chromist, amoebozoan, trypanosomatid and diplomonad protists, in addition to T. vaginalis . Meiotic gene sequences from T. vaginalis strain G3 found by this method were amplified by the polymerase chain reaction (PCR) from strain NIH-C1 and sequenced. Homology of the meiotic proteins was validated by phylogenetic analysis, by which we determined if the homologous genes were orthologs (related by speciation events) or paralogs (related by gene duplication). Our results indicate that homologs of a diverse set of meiotic genes are widespread among eukaryotes. The names of meiosis-specific proteins are highlighted in green. Exact stoichiometry is not implied. In meiosis I, cohesins bind to sister chromatids (A), after which double-strand DNA breaks are made by Spo11 (accessory proteins not shown) and the axial elements (Hop1) of the synaptonemal complex are formed (B). Double strand break repair is initiated (coupled with (B) in S. cerevisiae ) and Hop1 forms lateral elements of the synaptonemal complex (C). Strand exchange proteins are attracted to the double-strand break (accessory proteins not shown) (D). The resulting heteroduplex (E) may be resolved by crossovers, which utilize meiosis-specific proteins (F), or by gene conversion, which does not (G, proteins not shown). This model is based primarily upon details from S. cerevisiae , but includes details from mammals for Msh4 and Msh5, and speculates on the role of Drosophila Mei-9 (Rad1) in (F) as reviewed by [54] , [97] – [100] . Table 1 gives additional details and references. Genes encoding meiosis-specific proteins are highlighted in grey. TABLE: Core meiotic genes and some key functions of their encoded proteins in meiosis. Protein Function Spo11 Transesterase, creates DNA double-strand breaks (DSB) in homologous chromosomes [103] – [105] . Mre11 3′–5′ dsDNA exonuclease and ssDNA endonuclease, trims back broken DNA ends and hairpins [106] – [108] . Rad50 Dimer, holds broken DNA ends together while Mre11 trims. ATPase, has DNA-binding activity [105] , [108] . Rad1 (Mei9) Forms a heterodimer with Rad10 (Ercc1) [109] . 5′>3′ endonuclease, essential for nucleotide excision repair. Required for meiotic crossing over, normal meiotic chromosome disjunction, to repair mismatches in heteroduplex DNA and to resolve reciprocally exchanged recombination intermediates in Drosophila [110] . Hop1 Protein that binds DSBs and oligomerizes early during meiotic prophase, and forms axial and lateral elements of the synaptonemal complex [111] . Hop2 With Mnd1 , ensures accurate and efficient homology searching, downstream of Rad51 and Dmc1 , during pachytene stage of meiotic prophase [112] . Mnd1 With Hop2 , functions after meiotic DSB formation, and required for stable heteroduplex DNA formation [113] . Rad52 Binds to the ssDNA ends of DSBs and initiates DSB repair by homologous recombination [114] . Stimulates Rad51-mediated strand invasion by interaction with Rad51 and RPA, and promotes single strand annealing (SSA) [115] . Dmc1 Meiosis-specific homolog of Rad51, has similar function but promotes interhomolog recombination [116] – [118] . Rad51 Forms helical filaments on single-stranded and double-stranded DNA and catalyzes homologous DNA pairing and strand exchange. (Intrahomologous recombination) [116] , [118] . Msh4 Forms a heterodimer with Msh5 , interacts with Mlh1/Mlh3 heterodimer. Directs Holliday junction resolution towards crossover with interference [119] . Msh5 Forms a heterodimer with Msh4 , interacts with Mlh1/Mlh3 heterodimer. Directs Holliday junction resolution towards crossover with interference [119] . Msh2 Forms a heterodimer with Msh3 or Msh6 [119] . Msh6 Forms a heterodimer with Msh2, binds base-base mismatches [119] . Mlh1 Mismatch repair of dinucleotide and trinucleotide sequences, interacts with Msh2, forms heterodimers with Mlh2, Mlh3 and Pms1 [119] . Mlh2 Forms a heterodimer with Mlh1. Interacts with Msh2/3 or Msh2/6 for removal of cisplatin adducts [119] . Mlh3 Forms a heterodimer with Mlh1. Interacts with Msh2/3 or Msh2/6 for frameshift repair in mitosis or meiosis, or with Msh4/5 to promote meiotic crossovers [119] . Pms1 Mismatch repair. Interacts with Msh2/3 or Msh2/6 as a heterodimer with Mlh1 [119] . Mer3 Meiosis-specific DEAD-box helicase that promotes Holliday junction resolution with crossover interference together with ZMM proteins, including Msh4 and Msh5 [57] , [58] , [120] – [122] . Smc1 Forms a heterodimer with Smc3 to form core sister chromatid cohesin subunits, with ring shape around sister chromatids [123] , [124] . Smc2 Forms a heterodimer with Smc4 to form core condensin subunits, ring shape, essential for chromosome assembly and segregation. [123] Smc3 Forms a heterodimer with Smc1 to form core sister chromatid cohesin subunits, with ring shape around sister chromatids [123] , [124] . Smc4 Forms a heterodimer with Smc2 to form core condensin subunits, ring shape, essential for chromosome assembly and segregation [123] . Smc5 Forms a heterodimer with Smc6 (Rad18) and is involved in DNA repair and checkpoint responses [123] . Smc6 (Rad18) Binds ssDNA, has important role in postreplication DNA repair [125] . Forms a heterodimer with Smc5 and is involved in DNA repair & checkpoint responses [123] . Rad21 (Scc1) Holds Smc1 and Smc3 heads together by binding N-terminal domain to Smc3 and C-terminal domain to Smc1, thus holding sister chromatids together during mitosis and meiosis [124] . Rec8 Meiotic homolog of Rad21. Holds Smc1 and Smc3 heads together by binding N-terminal domain to Smc3 and C-terminal domain to Smc1, thus holding sister chromatids together during meiosis [126] . Scc3 Necessary for sister chromatid cohesion, and required for DSB repair [127] . Interacts with Smc1, Smc3 and Rec8 /Rad21 in holding cohesin ring together. Pds5 Important for maintenance of sister chromatid cohesion in late prophase [127] . RESULTS AND DISCUSSION We present our inventory of meiotic genes found in T. vaginalis in the context of an expanded set of 29 meiotic genes conserved among over 30 AFP and protist genomes ( Table 2 ). Of the 29 meiotic genes surveyed, 27 have homologs in the T. vaginalis genome; homologs of the meiosis-specific sister chromatid cohesin Rec8 and the DNA repair protein Rad52 were not found. The 29 genes surveyed from the five of six major eukaryotic lineages [11] for which complete genome sequence data are available (Opisthokonta, ‘Amoebozoa’, ‘Archaeplastida’, ‘Chromalveolata’ and ‘Excavata’) include 17 genes previously reported as “core meiotic machinery” [1] . The large number of meiotic genes shared by T. vaginalis , mammals ( e.g. , Homo ) and fungi ( e.g. , Saccharomyces ) suggest that putative meiotic processes in T. vaginalis could resemble those in mammals and fungi. In contrast, Giardia intestinalis , the other putatively asexual early-branching protist in our study, lacks eight of 29 meiotic genes, and Drosophila melanogaster , a sexual organism, is missing ten. G. intestinalis was recently shown to utilize three of its meiosis-specific protein homologs (Spo11, Hop1 and Dmc1) to mediate homologous recombination (but not meiosis) in the nuclei of cysts in a process named diplomyxis [4] . Together, our data and these observations suggest that T. vaginalis may be equipped to perform meiotic recombination or similar parasexual process by using its meiotic gene homologs. Bold columns highlight meiosis-specific proteins, and T. vaginalis is highlighted at the beginning. The names of genera with the most completely annotated genome sequences are highlighted in bold. The presence of orthologs is designated on the basis of phylogenetically verified data (P) obtained by BLASTp searches of public protein sequence databases and by cloning/sequencing (S) of selected genes (from T. vaginalis ) in this study or our previous study [1] . The presence (B) of putative homologs identified by BLASTp from recently released sequence data and verified only by bi-directional BLAST searches ( Table S1.3 in Supporting Information File S1 ) are also indicated. If more than one gene is present, copy number follows this designation parenthetically. The absence of meiotic genes from completed genome sequencing projects is indicated by (–), while empty cells indicate putatively missing data (from unfinished or incompletely annotated genomes). Protein homology was inferred by multiple sequence alignment and Bayesian phylogenetic analyses (see Figure 2 and Figures S1.1–S1.33 in Supporting Information File S1 for details), except for cells designated by “B”. Some data excluded from Figures S1.1–S1.33 in Supporting Information File S1 were phylogenetically verified in equally or similarly rigorous prior studies [1] , [50] , [101] , [102] . We found non-identical copies of eleven meiotic genes in T. vaginalis that are usually found as single copies in other eukaryotes ( Tables 2 , 3 and Table S1.1 in Supporting Information File S1 ). We cannot discern whether all of the copies are functional given the limited gene expression information available. However, our phylogenetic analyses show that these genes have evolved by recent duplications (within parabasalids) and, in most cases, one copy is more conserved than the others. Consistent with that observation, duplications within families of Trichomonas vaginalis protein-coding genes likely occurred after its divergence from sister taxon Trichomonas tenax [9] . The frequency of recently duplicated meiotic genes observed in T. vaginalis approaches that seen in some plants and fish ( Table 2 ), which are both thought to have polyploid origins. Six of the eleven genes present in multiple copies in T. vaginalis ( Table 3 ) are uniquely duplicated in T. vaginalis and not in any other organism included in our study. Duplications of chromosomes, segments of chromosomes, and possibly whole genomes as in some plants and fish might explain the presence of extra copies of meiotic genes in T. vaginalis . If so, these genes are putative homeologs or ohnologs – duplicated genes arising from polyploidization events [46] . Since the T. vaginalis genome sequence is highly repetitive and consists of 17,290 unordered scaffolds [9] , additional data and analyses to better assemble the genome sequence into a smaller number of longer and ordered scaffolds are required to understand how chromosome- and genome-scale duplication contributed to its genome architecture. Ten of eleven recently duplicated meiotic genes were located on separate scaffolds. However, the two Rad50 gene copies were found in an inverted tandem repeat on the same scaffold of assembled genome sequence, an arrangement which is inconsistent with polyploidization and that might result from ectopic meiotic sister chromatid recombination [47] . Our sequences from T. vaginalis strain NIH-C1 revealed that Rad51b , Scc3b and Pds5 differed in the number of short tandemly repeated sequences within their coding regions when compared with the genome sequence of T. vaginalis strain G3. These duplicate genes ( Rad51b and Scc3b ) may also be derived from allelic divergence resulting from the accumulation of mutations during extensive asexual (mitotic) reproduction [48] . If this is the case, T. vaginalis may be facultatively sexual or asexual. Summarized from Table S1.2 in Supporting Information File S1 . All duplicates trace to within parabasalids since their phylogenies show no intervening lineages, with three exceptions ( Smc1 , Smc3 , Mlh2 ) that can be attributed to rapid rates of evolution. To assess if the T. vaginalis meiotic genes are functional, we queried two expressed sequence tag (EST) databases and found evidence of the transcription of some genes ( Table S1.1 in Supporting Information File S1 ). ESTs derived from normal asynchronized cultures were found for Rad50a , Rad1 , Dmc1 , Msh2 , Smc1b , Smc5b , Smc6 , Rad21b and Pds5 . In low-iron conditions, Msh5 , Smc1a , Smc5a , Rad21b and Scc3a are transcribed, while ESTs encoding Rad51a were found from G2/M trophozoites. Cells exhibiting vaginal epithelial cell mediated cytoadherence expressed Smc2 , and ESTs encoding Smc1b and Smc4b were found from cold-induced pseudocysts (“compact non-motile forms without a cyst wall” [40] ). These data are consistent with the expression of many (18 of 37) of the genes in our survey. However, the available ESTs may not represent conditions that promote meiosis, and their small numbers are consistent with the possibility that a given gene is expressed at low levels, or even post-transcriptionally down regulated. Of the 18 meiotic genes found in ESTs, Dmc1 and Msh5 are the only meiosis-specific gene orthologs found to be expressed in the available small sample of ESTs. The expression of Dmc1 and Msh5 orthologs suggests that meiotic recombination may occur in T. vaginalis in asynchronous cells and during low iron conditions (stress). In sum, the EST data are consistent with homologous recombination (and possibly meiosis) occuring in asynchronous cells, G2/M trophozoites and pseudocysts, as well as during low iron conditions and cytoadherence to vaginal epithelial cells. Our revised inventory of meiotic proteins that spans an additional breadth of organisms with completely sequenced genomes allows us to elucidate the sexual status of basal eukaryotes. We have identified homologs of meiotic genes among organisms that may span most of the deepest divergences among eukaryotes by using phylogenetic inference to assess the evolutionary history of each of meiotic protein homolog. This approach allows evaluation of the origin and evolution of meiosis in the context of the common ancestor of eukaryotes ( Figure 2 and Table 2 ). All trees shown are the consensus tree topologies determined from ≥700 best trees ( i.e. those with the highest posterior probabilities) inferred by Bayesian analysis using alignments of inferred proteins. Animals are indicated in red text, fungi brown, ‘Amoebozoa’ teal, ‘Archaeplastida’ in green, Alveolates plum, ‘Chromista’ purple, ‘Excavata’ blue and prokaryotes shown in black. Branches with the best support – i.e. , those with 0.95 to 1.00 Bayesian posterior probabilities – have thicker lines. Numbers at the nodes indicate Bayesian posterior probability followed by percent bootstrap support from 100 replicates of PROML. An asterisk (*) denotes topological constraints placed upon the nodes uniting Fungi and Opisthokonts for Bayesian analysis. Scale bars represent 0.1 amino acid substitutions per site. Details for each tree and the accession numbers for all sequences are provided in Figures S1.1–S1.4 in Supporting Information File S1 . (A) Hop2 homologs, unrooted. 167 aligned amino acid sites were analyzed, this consensus topology derived from 900 trees, α = 3.86 (2.71<α<5.37), pI = 0.014 (0.0004