Evolution of the mating type gene pair and multiple sexes in Tetrahymena.

The multiple mating type system of the Ciliate Tetrahymena thermophila is a self/non-self recognition system, whose specificity resides in a head-to-head, functionally distinct pair of genes, MTA and MTB. We have now sequenced and analyzed these mating type genes in nine additional Tetrahymena species. We conclude that MTA and MTB are derived from a common ancestral gene and have co-evolved for at least ∼150 Myr. We show that T. shanghaiensis, a perpetual selfer (unisexual) species, has a single mating type gene pair, whose MTA and MTB genes likely have different mating type specificity. We document the recent replacement of a complete different set of mating type specificities for another, illustrating how quickly this can happen. We discuss how varying conditions of reproductive stress could result in evolutionary co-adaptations of MTA and MTB genes and changes in mating type determination mechanisms.

Introduction

Sex is an evolutionary conserved process among organisms, including the Ciliated Protozoa (= Ciliates), a unicellular eukaryotic phylum (Bachtrog et al., 2014). Mating type systems generally ensure sexual self-incompatibility and promote outbreeding. Most eukaryotic species have two mating types or sexes and thus a binary mating system. However, systems with more than two mating types exist in some groups such as ciliates and mushrooms (Phadke and Zufall, 2010; Kües, 2015). Ciliate mating systems are very diverse: they vary in such features as number of mating types, mechanism of mating type determination (MTD), and molecular nature of the mating type proteins. This suggests that many fundamental changes in mating type biology have independently evolved in the major Ciliate clades (Phadke and Zufall, 2010). The Tetrahymena genus of ciliates is thus an excellent model system for studying multiple mating type systems and their evolution.

As in other ciliates, cells of most Tetrahymena species possess two kinds of nuclei: a diploid, silent germline nucleus (the micronucleus or MIC) and a polyploid, highly expressed somatic nucleus (the macronucleus or MAC). The Tetrahymena life cycle consists of two stages: asexual reproduction by binary fission when food is abundant and conjugation triggered by starvation (reviewed in (Orias et al., 2011; Orias et al., 2017)). Key life cycle features of genetic significance, illustrated in Figure 1, are:

Figure 1

Schematic diagram of the Tetrahymena life cycle

The life cycle of sexual Tetrahymena species consists of two stages: sexual and asexual. To start the sexual stage, two starved cells conjugate (1). The micronucleus (MIC) of each conjugant first undergoes meiosis (2). A randomly chosen meiotic product in each conjugant divides mitotically and generates two gamete pronuclei (3). Next, the two conjugants exchange one of their two gamete pronuclei, after which the pronuclei fuse to form a genetically identical diploid fertilization nucleus in each conjugant (4); thus, each conjugant gets a haploid genome from each parent. The fertilization nucleus undergoes two rounds of mitotic division, generating four nuclei in each conjugant (5). Two of these nuclei differentiate into new macronuclei (MACs) and the other two remain MICs (6). The old MAC is degraded, and the two “exconjugants” separate from one another. After feeding, the exconjugants undergo their first postzygotic fission; each of the new MACs is distributed into a different daughter cell, the “karyonide” cells (7). At this stage, the one MIC and one MAC condition has been restored. Subsequently, during the asexual part of the life cycle, the MIC divides mitotically and the MAC divides amitotically at every succeeding cell cycle. Amitotic division of the polyploid MAC results in the random segregation of parental chromosomes among daughter cells, leading to segregation of genetic diversity among individuals (allelic assortment). Ultimately, by asexual reproduction and allelic assortment, an individual MAC tends to become homozygous for its entire genome.

Only one of the four MIC meiotic products is retained in each conjugant.

Reciprocal fertilization generates genetically identical, diploid zygote nuclei in each conjugant.

The zygote nucleus in each conjugant divides twice mitotically, two products are the new MICs, while the other two differentiated into the new MACs.

During MAC differentiation the five MIC chromosomes undergo programmed site-specific fragmentation, resulting in 180 MAC chromosomes. These acentromeric chromosomes are then amplified (∼45 G1 copies).

The two exconjugant cells from a pair divide, resulting in four cells with genetically identical MICs but independently differentiated MACs, called “karyonides”.

Phenotypic assortment: when asexual multiplication resumes after conjugation, amitotic division of the polyploid MAC results in the random distribution of daughter chromosome copies at every fission. This allows segregation at all heterozygous loci present in a newly differentiated MAC and ultimately generates whole-genome homozygous MACs.

Sexual progeny are initially sexually immature; in T. thermophila they must undergo 40–70 fissions before they reach sexual maturity and can mate again.

The diversity of Tetrahymena mating type systems was described in detail in a series of papers by Nanney, Elliott, and their collaborators, beginning in the early 1950s (reviewed in (Orias, 1981; Orias et al., 2017)). Among sexual species in this genus, the number of known mating types per species ranges from three to nine. Certain rare species (e.g., T. shanghaiensis) are unisexual (selfers), meaning that sexually mature cells in a clonal population mate with one another upon starvation (Chen et al., 1982; Simon et al., 2009) and all their progeny are selfers. Cells of other species (e.g., T. pyriformis and T. vorax) have lost their MIC (i.e. the germline nucleus) and thus can only reproduce asexually (Gruchy, 1955; Doerder, 2014).

The best characterized species belong to either the “Australis” or the “Borealis” clades, which diverged ∼150 Myr ago (Xiong et al., 2019). Earlier studies suggested that Tetrahymena MTD patterns can also be classified into two categories that co-branch with the phylogenetic tree. Investigated species in the “Australis” clade exhibit “synclonal MTD”; where the four genetically identical karyonides of a mating pair (the synclone) express the same mating type with a Mendelian inheritance pattern when they reach sexual maturity (Figure S1A). In T. pigmentosa, for example, mating types are controlled by three alleles of a single mat locus that show “peck-order” dominance (Simon, 1980). In contrast, investigated species in the “Borealis” clade show “karyonidal MTD”, where mating type is randomly and independently determined in each new MAC. This results in four genetically identical karyonides which often express different mating types (Figure S1B).

The molecular basis of mating type specificity has only been investigated in T. thermophila (“Borealis” clade). Mating type is determined by a mating type gene pair (mtGP), a head-to-head arrangement of two mating type genes (MTA and MTB) (Cervantes et al., 2013) (Figure 2A). In this study, we investigated the molecular evolution of Tetrahymena mtGPs in a phylogenetically wide range of Tetrahymena species, including two asexual species (pyriformis and vorax), a unisexual obligatory selfer species (shanghaiensis), and the species furthest removed from the “Australis” and “Borealis” clades (paravorax) (Figure 2B). We provide evidence for the evolution of all Tetrahymena mating type proteins from an ancient member of the “Furin-like repeat” protein family, for the coevolution of MTA and MTB genes, for the evolution of a heterotypic MTA-MTB gene pair leading to perpetual selfing, and for the recent replacement of one multiple mating type system with another within a subgroup of the genus Tetrahymena.

Figure 2

Tetrahymena mating type genes, species phylogeny and mating type gene expression

(A) Schematic diagram of a T. thermophila mtGP and the domain composition of the two encoded proteins. Top line: MTA gene (green), MTB gene (blue), and intergenic region (black). Blue vs. green colors are used to indicate that, other than the furin-like repeats, there is no sequence conservation between MTA and MTB proteins. Darker color thick lines: conserved 3′ terminal exons; lighter color thick lines: mating type-specific exons. Bottom line: MTA and MTB proteins, showing detail of terminal exon domains conserved in all Tetrahymena mating type genes. Vertical red line: predicted transmembrane helices; squares: cysteine-rich “Furin-like repeat” domain. The MTA 3′ terminal exon sequences are >99% conserved among six published mating types, and likewise for the MTB genes.

(B) Tetrahymena species phylogeny and MTA, MTB gene homologs present in each species. The phylogenetic tree is based on 18S rRNA sequences. Green check mark, homologous gene identified; blue strikeout check mark, homologous gene with low sequence similarity identified; red ×, no homologous gene identified.

(C) Gene expression measurements, using RNA sequencing data, for mating type genes of all ten species. Numbers represent FPKM values. All mating types were measured and averages are shown when data was collected for more than one mating type in the species.

Results

Mating type gene pair homologs exist in all but the most distantly related Tetrahymena species examined

In T. thermophila, the MTA and MTB genes shares some similar features but have completely different sequences. The sequence of the terminal exons is highly conserved between MTA genes encoding different mating types (Figure 2A) (nucleotide identity >0.99); the same is true among the MTB alleles. In contrast, the remainder of the MTA and MTB genes and the intergenic region are mating type-specific (nucleotide identity <0.6). The 3′-terminal exons of the MTA and MTB genes both encode five predicted transmembrane helices and a cysteine-rich “Furin-like repeat” domain. These features are diagrammed in Figure 2A; see also (Cervantes et al., 2013; Orias et al., 2017).

We looked for homologs of the T. thermophila MAC mating type genes in nine additional species spanning the ∼300 Myr old Tetrahymena genus (Xiong et al., 2019) (Table 1). Six species (thermophila, malaccensis, pyriformis, vorax, borealis and canadensis) are in the “Borealis” clade, three (shanghaiensis, americanis, and pigmentosa) are in the “Australis” clade, and one (paravorax) diverged from both clades at the base of the Tetrahymena genus (Figure 2B). T. pyriformis and vorax cells never mate, they lack an MIC and only reproduce asexually. And to our knowledge, sexual reproduction has not been observed in paravorax. Altogether, we investigated mating type genes in 19 Tetrahymena strains (Table 2). We found mating type gene homologs in all sexual species, as well as in three asexual species (pyriformis, vorax and paravorax) and in the unisexual strain, shanghaiensis (a “selfer” species, in which starvation of sexual mature cells triggers intraclonal mating). Within each sexual species, we verified that mating only occurs between starved cells of different mating types; no mating was observed between starved cells from different species (for experimental details see Transparent methods).

Table 1

Mating type systems of Tetrahymena species investigated in this article

Cladea	Subcladea	Species	# Mating types	MTD patternb
Borealis	“The-Mal”	Thermophila	7	Karyonidal
Borealis	“The-Mal”	Malaccensis	6	Karyonidal
Borealis	“Pyr-Vor”	Pyriformis	Asexual	NA
Borealis	“Pyr-Vor”	Vorax	Asexual	NA
Borealis	“Bor-Can”	Borealis	7	Synclonal
Borealis	“Bor-Can”	Canadensis	5	Synclonal
Australis	“Pig-Ame”	shanghaiensis	Perpetual selfer	NA,
Australis	“Pig-Ame”	pigmentosa	3	Synclonal
Australis	“Pig-Ame”	americanis	9	Synclonal
Paravorax	NA	paravorax	ND	ND

N/D: not determined.

The phylogeny is illustrated in Figure 2B.

Mating type determination pattern observed in sexual progeny (see text and Figure S1 for explanation). N/A: not applicable; only strain and only species characterized in this clade.

Table 2

Relevant strain and mating type information on the Tetrahymena strains used in this work

Species	Strain IDa	Mating typeb
thermophila	SD01580	II
thermophila	SD01653	III
thermophila	SD01582	IV
thermophila	SD01656	V
thermophila	SD01584	VI
thermophila	SD01585	VII
Malaccensis	SD01608	X
Pyriformis	SD00707	NA
Vorax	SD30421	NA
borealis	SD01609	X
borealis	SD19502	Y
borealis	SD19803	Z
canadensis	SD30770	X
shanghaiensis	SD205039	Selfer
pigmentosa	SD19481	III
pigmentosa	SD20427	I
americanis	SD21194	X
americanis	SD21244	Y
paravorax	SD205177	N/D

N/A: not applicable; these strains lack a micronucleus and are asexual. N/D: not determined.

Tetrahymena Stock Center ID numbers.

Where the relationship to previously published mating types is undetermined, we have used capital letters X, Y, and Z to avoid confusing the literature.

Our searches revealed a single, mating type-specific mtGP with MTA and MTB homologs in head-to-head orientation for each mating type of every species, with the sole exception of the T. paravorax strain, which has a truncated MTA gene (named MTAL for MTA-like) and lacks an MTB homolog (Figures 2B and S2). Additionally, we determined that all mtGPs of species with sequenced genomes (Xiong et al., 2019), with the single exception of the asexual T. vorax mtGP, have syntenic chromosomal locations (Figure S3).

It has long been known that starvation conditions are required for conjugation in Tetrahymena. Consistent with this, the T. thermophila MTA and MTB genes are highly expressed during starvation, but are essentially silent during vegetative growth (Cervantes et al., 2013). To investigate whether mtGPs of other Tetrahymena species might function similarly in mating, their expression levels were measured during growth and starvation (Figure 2C). In the seven sexual species (including the “selfer” species, T. shanghaiensis), the expression pattern of the mating type genes is identical to that of T. thermophila, consistent with conservation of mtGP function in mating. In the asexual species (pyriformis, vorax, and paravorax), the mating type genes are not induced by starvation, suggesting that they are no longer functional in mating.

Tetrahymena MTA and MTB genes are likely derived from a common ancestral gene

All 37 MTA and MTB genes reported in the previous section belong to the same gene superfamily which has a conserved cysteine-rich “Furin-like repeat” domain (Figures 3A and S2). Alignment of the “Furin-like repeat” domains in all MTA and MTB homologs shows that all 14 cysteine residues are highly conserved with one another, although four cysteines are missing in MTAL (Figure 3A). Since cysteine residues play many important roles, such as forming covalent disulfide bonds with each other (Sela and Lifson, 1959; Thornton, 1981), the conservation of these residues may be essential to the secondary structure of the mating type proteins and their function.

Figure 3

Highly conserved features of Tetrahymena MTA and MTB genes: FLR repeats and introns

(A) Sequence alignment of the “Furin-like repeat” domains. Cysteine sites are conserved in MTA and MTB proteins; note that four conserved cysteines (darker background) are missing in T. paravorax MTAL. Pink letters: partially conserved amino acids. P.tet: Paramecium tetraurelia mtA gene.

(B) Conservation of intron location and phase among MTA and MTB genes. Thick green lines, MTA exons; thick blue lines, MTB exons; thick dark green or dark blue lines, terminal exons. Colored dots, introns: blue, phase 0, inserted between two codons; green, phase 1, inserted between the first and second codon nucleotide; red, phase 2, inserted between the second and third codon nucleotide. Mating type allele shown, for those species with multiple sequenced mating types: T. thermophila: mt II; T. borealis: mt X; T. pigmentosa: mt III; T. americanis: mt X. Green triangle, phase 1 intron that exists in T. borealis MTAY (but not in MTAX and MTAZ). Note that mating type genes in the asexual species, T. pyriformis, T. vorax and T. paravorax, are essentially silent during growth and starvation (see Figure 2C) and therefore lack a reliable RNA-Seq-based intron/exon annotation for most exons.

The full length Tetrahymena MTA and MTB genes investigated here contain 5-8 introns. In the sexual species, four introns (#3, #5, #6, and #7) show total conservation of phase and approximate location in the 32 sequenced MTA and MTB genes (Figure 3B and Table S1, details in Figures S2 and S4). Intron #7 precedes the 3’-terminal exon, which encodes the “Furin-like repeat” domain and transmembrane helices of the MTA, MTAL, and MTB genes. A fifth intron (#2) is conserved among the MTA and MTB genes of all mtGPs, with the exception of T. canadensis, which has just one sequenced mtGP (MTAX).

The conservation of intron location, in both MTA and MTB, is particularly striking within subgroup alignments (Figure S4) that include at least two species for each subgroup. Astonishingly, 27 (of 38) introns occur at identical codon locations within their gene. In another seven cases, the introns are located at an adjacent codon. In the four remaining cases, the intron is located a few codons further away, but there are clearly insertions/deletions of one or more codons in the immediate neighborhood. The only outlier is intron #2 in the “Bor-Can” MTAs, which has changed phase in T. borealis MTAZ and is missing in T. canadensis MTAX.

The absolute conservation of the “Furin-like repeat” domain, together with the high degree of conservation of intron location and phase in such a large (∼1,500 aa) protein, along with chromosomal synteny, are strong evidence that all MTA and MTB genes were derived from a common ancestral gene which existed prior to the divergence of the “Australis” and “Borealis” clades in Tetrahymena, ∼150 Myr ago (Xiong et al., 2019).

To confirm the evolutionary relationship between MTA and MTB proteins, we generated phylogenetic trees of these proteins from the sexual species. For each protein, we made separate trees for the entire protein (Figure S5), for the C-terminal, transmembrane exon (distal ∼1/3 of the protein), and for the rest of the protein (proximal ∼2/3 of the protein (Figures 4A and 4B, respectively)). The results show that all three trees have two main branches, such that all MTA proteins fall cleanly into one branch, while all MTB proteins fall cleanly into the other branch. This is consistent with the nearly complete lack of overall sequence similarity observed between the MTA and MTB proteins when they are co-aligned (Figure S2). Thus the trees support the conclusion that the MTA and MTB genes diverged structurally and functionally from a common ancestral gene early in Tetrahymena evolution, prior to the divergence of the “Australis” and “Borealis” clades.

Figure 4

Phylogenetic tree of mating type proteins

(A) Protein phylogenetic tree based on the C-terminal exon (distal-third).

(B) Phylogenetic tree based on the rest of sequence (proximal-two-thirds). The best-fit models were calculated by ProtTest (version 3.4.2) (Darriba et al., 2011). (A) is under JTT + I + G + F model; (B) is under VT + I + G + F model. Numbers at each node, bootstrap values (1000 replicas). Branch length, number of base substitutions per site. Horizontal dashed lines: boundaries between the species clades/subgroups. Red boxes: deviations from the species tree. The MTA and MTB branches are shown opposite to one another to facilitate comparisons between them.

Interestingly, conserved intron #4 is absent from the MTA genes of all sequenced mating types in the “The-Mal” subgroup. These genes all have another intron (intron #6.5), located between conserved introns #6 and #7. Unexpectedly, the T. borealis MTAY gene contains an intron at exactly this location (#6.5) and phase (Figure 3B, green triangle and Table S1) but also has intron #4. Conceivably, the MTA genes of the “The-Mal” subgroup and the T. borealis MTAY gene may share a recent common ancestor.

Paramecium, like Tetrahymena, belongs to the Ciliate class Oligohymenophorea. The two genera are estimated to have diverged from one another nearly a billion years ago (Xiong et al., 2019). The mating protein (mtA) in Paramecium tetraurelia is also a member of the superfamily of genes having a terminal exon containing “Furin-like repeat” domains, which conserves all the cysteines found in Tetrahymena and the transmembrane helices (Singh et al., 2014). However, the Paramecium gene has only three introns, all at different locations than in the Tetrahymena genes. Thus, the Paramecium and Tetrahymena mating type genes likely had a common ancestor but have undergone extensive independent evolution.

The Tetrahymena mating type proteins exhibit a special type of incomplete lineage sorting

The topology of the individual MTA and MTB branches of the phylogenetic tree of the two entire proteins (Figure S5) does not exactly match the topology of the species tree (Figure 2B). This finding represents an example of “incomplete lineage sorting”. The discrepancy is limited to the MTA and MTB proteins of T. malaccensis, borealis, canadensis, and shanghaiensis.

Interestingly, the topologies of the branches of both the MTA and MTB proteins in the phylogenetic tree for the distal third (C-terminal exon) (Figure 4A) are almost identical to the topology of the species tree (Figure 2B). Indeed, essentially all of the incomplete lineage sorting seen for the whole proteins is accounted for by that in the proximal (N-terminal) roughly two thirds of each protein (Figure 4B). In this context, it is important to note the very different functions of the two segments of the mating type proteins. The distal third (encoded by the C-terminal exon) includes the only predicted intracellular segment of the protein and can thus be inferred to be involved in mating type-non-specific interactions with cell machinery required, for example, for the structural remodeling of the cell in preparation for mating: re-shaping and de-ciliation of the anterior ventral surface where two cells will form a temporary junction lasting many hours (“tip transformation” (Wolfe and Grimes, 1979)). This is the protein segment that has evolved at a rate commensurate with that of other conserved, mating type-unrelated cell proteins. On the other hand, the proximal two thirds of the mating type proteins are predicted to be extracellular and to be the main site of the mating type-specific (positive and negative) interactions that allow the self vs. non-self recognition required to initiate or inhibit mating between two cells. Given that most Tetrahymena species possess multiple mating type systems, and that speciation has been accompanied by conservation of some mt protein specificities and evolutionary radiation of others (as described in subsequent sections), the incomplete lineage sorting observed for this protein segment becomes readily understandable. This clear distinction between nearly complete lineage sorting in the distal third and significantly incomplete lineage sorting in the proximal two thirds, seen for both of the two proteins, represents an example of what could be called “composite lineage sorting”.

The individual examples of incomplete lineage sorting detected in this proximal two thirds of the MTA and MTB proteins are addressed in more detail below:

T. malaccensis MTAX and MTBX proteins co-branch with T. thermophila MTA4 and MTB4, respectively. The two species are very closely related (Figure 2B). This was already reported (Cervantes et al., 2013) and interpreted to mean that these two mtGPs recently evolved from the same mtGP in a common ancestor of the “The-Mal” subgroup.

T. borealis MTAZ and MTBZ proteins co-branch with T. canadensis MTAX and MTBX, respectively, which has a proposed analogous explanation to the previous case. These two species are among the most closely related Tetrahymena species pairs known. Their SSUrRNA genes are identical, and their COX1 barcodes show only 4.4% polymorphisms (data not shown). Four percent is the COX1 threshold that best corresponds to the ultimate criterion of the Tetrahymena species difference, the failure to mate (Doerder, 2019).

The two other discrepancies are more intriguing.

The co-branching of T. borealis mating type Y with all the T. thermophila and T. malaccensis mating types is unexpected because these species are in two different phylogenetic subgroups (“Bor-Can” and “The-Mal”, respectively). This finding suggests that this particular mtGP has been retained and has changed relatively little since the divergence of the two subgroups. Further supporting this hypothesis, and as described in an earlier section, the T. borealis MTAY gene contains an intron at exactly the same location and phase as the additional intron (#6.5) found in all the sequenced MTA genes in the “The-Mal” subgroup and nowhere else (Figure 3B).

The last anomaly is the co-branching of the MTA and MTB proteins of the perpetual selfer T. shanghaiensis with T. pigmentosa MTA3 and MTB3, respectively; this finding is addressed in the T. shanghaiensis section, further below.

A better understanding of the molecular basis of these cases of incomplete lineage sorting will require sequencing additional mating type genes of these and other species in the “Bor-Can” subgroup, as well as additional knowledge of the sequence and organization of the genes encoding these proteins in the germline (micronuclear) mat locus. The latter information is currently available only for T. thermophila (Cervantes et al., 2013).

MTA and MTB genes have coevolved within the different Tetrahymena phylogenetic subgroups

T. thermophila MTA and MTB gene products have non-redundant functions required for mating (Cervantes et al., 2013). Intriguingly, when we compared the two mating type genes in species belonging to different Tetrahymena phylogenetic groups, we noticed several cases where recent evolutionary changes in the MTA gene have been mirrored by corresponding changes in the MTB gene. For example, in the “The-Mal” subgroup, the GC content of 3’-terminal exons of both genes is significantly higher than that of other regions of the mtGP (Figures 5A and 5B). This difference is not observed in the other subgroups.

Figure 5

Coevolution of the MTA and MTB genes as demonstrated by GC content and length of conserved regions

MTA (green) and MTB (blue) genes of the mtGP are shown parallel to each other to facilitate comparison.

(A) mtGPs structure and GC content. Dark green and dark blue exons, terminal exons of MTA and MTB genes, respectively.

(B) Statistical comparison (ANOVA) of the data shown in panel (A) Only exon sequence was included to avoid intron sequence influence. This analysis confirms that the GC content of terminal exons relative to that of other exons is statistically significantly higher in T. thermophila and T. malaccensis, and only in those to species.

(C) Sequence conservation among different mating type genes within species with more than one sequenced mtGP. Dots within the gene lines: introns (blue, phase 0; green, phase 1). Vertical dotted line: boundary between conserved and specific regions. Horizontal dashed lines: mean allele sequence similarity within each region. Specific regions and statistical comparisons (t-test) are shown on the right in each chart. ∗: p < 0.05; ∗∗: p < 0.01; ∗∗∗: p < 0.001. Note the co-variation in the length of highly conserved sequence in the MTA and MTB genes of the same species.

More striking evidence of coevolution is the length of mating type specific region. Previous work in T. thermophila (Cervantes et al., 2013) had revealed that the 3’-terminal exons in MTA genes, comprising about 1/3 of each gene, are highly conserved among alleles for the different mating types, while the rest of each gene is mating type-specific; the same is true for MTB alleles (Figure 5C, T. the). We examined different mtGPs in species of the “Bor-Can” subgroup (T. borealis) and the “Australis” clade (T. americanis and T. pigmentosa) to see if they also shared a distinct sequence conservation boundary. Sequence conservation plots (Figure 5C) of the MTA and MTB genes of these species show that none of their mating type genes have an abrupt conservation boundary at conserved intron #7. Instead, in the two species of the “Australis” clade, the conserved regions of the MTA and MTB genes are about twice as long as in T. thermophila (Figure 5C, T. ame and T. pig), so that only the 5’-terminal ∼1/3 of each gene is unique for each mating type. In further contrast, for the three sequenced mating types of T. borealis, essentially the entire length of the MTA and MTB genes is mating type-specific (Figure 5C, T. bor).

In an attempt to shed more light on the question of MTA and MTB protein coevolution, we also did an amino acid usage analysis of all the MTA and MTB proteins in the sexual species (Data S1, Figure S6). The results showed some regularities but did not provide clear conclusions. A rigorous answer to this question will likely have to wait for additional experimental investigations and knowledge of the 3D structure of these proteins.

T. shanghaiensis cells have one mtGP allele and mate with one another

In contrast to the multiple mating type systems of other Tetrahymena species, T. shanghaiensis has been reported to be a selfer species by Chen et al. (1982) and Feng et al., 1988, in which sexually mature cells within every T. shanghaiensis clone mate with one another upon starvation. By DNA sequencing, we identified a single mtGP in the T. shanghaiensis MAC genome, whose MTA and MTB genes are homologous with the respective mating type genes of the other Tetrahymena species investigated here (Figures S2 and S4). Furthermore, the T. shanghaiensis MTA and MTB homologs have identical expression profiles to those of all the other sexual Tetrahymena species during growth and starvation Figure 2C.

To verify the previously reported observation that the non-assorting selfing trait is transmitted to sexual progeny, we did RNA-Seq experiments and de novo sequence assembly on two additional starved T. shanghaiensis populations, obtained as sexual progeny of independent selfing populations (i.e. biological replicates) (Figure S7, frames 5 and 7, highlighted with red stars). Cells in both populations contained transcripts from the MTA and MTB genes of the only previously detected single, genomic mtGP.

To explain the selfing of T. shanghaiensis, we proposed four a priori hypotheses, the first three of which differ with respect to what is encoded in the MAC genome, as illustrated in Figure S8.

Two normal mtGPs with different mating type specificities are present in the MAC of every cell; these mtGPs cannot be purified by assortment because the MAC mat locus is homozygous for both genes. MTA and MTB proteins with different mating type specificities are expressed upon starvation and trigger selfing.

Two normal mtGPs with different mating type specificities are present in the homozygous MAC genome but only one can be expressed. Frequent gene conversion causes a random mtGP to be expressed in every cell. Thus, even a clonal population will have a mixture of cells expressing different mating type specificities, leading to selfing upon starvation.

Only one type of mtGP is present in the homozygous MAC genome, containing MTA and MTB genes of different mating type specificity, i.e. a heterotypic mtGP. The proteins encoded by this mtGP are sufficient to trigger mating upon starvation.

Our finding of a single mtGP in every population is consistent with only the third hypothesis, a heterotypic mtGP.

It could be argued that mating in T. shanghaiensis is controlled by genes other than the mtGP found in the genome. This alternative is unlikely for the following reasons. As already mentioned, the T. shanghaiensis MTA and MTB genes show identical expression pattern during the life cycle as MTA and MTB genes of other Tetrahymena species (Figure 2C). Furthermore, the locations of both genes in the mating type protein phylogenetic tree (Figure 4) and in the amino acid frequency-based clustering (Figure S6) all correspond closely with the location of T. shanghaiensis in the species phylogeny tree (Figure 2B). Therefore, it seems highly probable that the mtGP of T. shanghaiensis functions in mating. Thus the results are entirely compatible with T. shanghaiensis having a heterotypic mtGP which is responsible for the perpetual selfing, as shown in Figure S8C.

The above conclusion is supported by the findings of Lin and Yao (2020) in Tetrahymena thermophila, published while this article was under review. These authors report that starved cells expressing one complete MTA gene of one mt specificity and one complete MTB gene of different mt specificity behave as non-assorting selfers, exactly as T. shanghaiensis, regardless of which pair of different mt specificities are involved. This finding strengthens our conclusion that the basis for the non-assorting selfing of T. shanghaiensis is the possession of MTA and MTB genes of different specificity. The non-assorting T. shanghaiensis selfer and the selfers investigated by Lin & Yao illustrate a new type of molecular basis for selfing in Tetrahymena (see Data S2 for more details about classification of Tetrahymena selfers). Rigorous proof of our hypothesis must await the development of molecular genetic tools to experimentally modify mating type genes in T. shanghaiensis.

Discussion

Proposed steps in the evolution of the Tetrahymena mtGP

Two mating type proteins (MTA and MTB) embody mating type specificity in the multiple mating type system of T. thermophila. These proteins contain ∼1,500 amino acids each, and are encoded by adjacent head-to-head genes in the mtGP (Cervantes et al., 2013). The two genes have very similar organization and both belong to the superfamily of FLR domain proteins, but they only share about 6% sequence similarity. In this report, we identified and compared the MAC mating type protein-coding loci in nine additional Tetrahymena species, encompassing the phylogenetic diversity of the genus. Eight of the additional species were found to have homologs to T. thermophila MTA and MTB, while an asexual strain of the most distantly related species, T. paravorax, has an MTA homolog but lacks any trace of an MTB homolog. The MTA and MTB genes of the sexual species share in common at least 5 introns that have conserved location and phase.

The findings we have reported here suggest the major steps in the evolution of the Tetrahymena mtGP over the last 150 Myr, illustrated in Figure 6:

Figure 6

Postulated steps in the evolution of the Tetrahymena mtGP

Key to symbols: Thick gray arrows: major evolutionary steps. Color-matched thin arrows: paralogs encoding different mating type specificities (not shown) co-evolve from MTA and MTB genes. Symbols within each gene: vertical lines: introns (blue, green and red: phase 0, 1 and 2, respectively); hatched rectangle, “Furin-like repeat” domain. Step 1. Ancient gene, containing a cysteine-rich “Furin-like repeat” domain and five transmembrane helixes, acquires mating-related function, and evolves into distinct mating type genes in Tetrahymena and Paramecium. Step 2. In Tetrahymena, a DNA rearrangement generates a copy of the mating type gene (MTA′). Step 3. The MTA′ gene acquires a new mating-related function, becoming the MTB gene. A set of paralogs encoding different mating type specificities (colored arrows) co-evolve from each gene. Intron location and phase are conserved during this differentiation and following events. Step 4. A second translocation causes MTA and MTB to become adjacent genes in opposite orientation, thus generating the head-to-head mtGP. A new wave of co-evolutionary mutational changes generates additional mating type specificities. Step 5. An MTA rearrangement deletes intron #4 (black vertical dotted line) and creates intron #6.5 (marked with triangle) within an MTA gene in a “The-Mal” subgroup ancestor, followed by a new wave of co-evolved changes that generate additional mating type specificities. Note: if the MTA and MTB alleles brought into contiguity at step 4 had encoded different mating type specificities, this evolutionary intermediate initially would have been a perpetual selfer, like present day T. shanghaiensis.

A cysteine-rich “Furin-like repeat” domain protein acquired mating-related activity in an Oligohymenophorean ancestor of Tetrahymena and Paramecium. A terminal exon, encoding a “Furin-like repeat” domain and five transmembrane helixes, are shared by mating type genes of P. tetraurelia and all ten Tetrahymena species examined. The asexual T. paravorax strain, the earliest Tetrahymena to diverge, with a clearly differentiated MTA-like gene (MTAL) but lacking an adjacent MTB gene, could represent a relic of this evolutionary stage.

A duplication of the “Furin-like repeat” family protein gene occurred in a common ancestor of the “Australis” and “Borealis” Tetrahymena clades, and generated identical MTA and MTB genes sharing the location and phase of multiple introns still observed today.

Subsequent co-evolution of the duplicate copies generated functionally differentiated MTA and MTB genes.

A subsequent DNA rearrangement brought into contiguity two cognate MTA and MTB genes, in head-to-head orientation, generating the Tetrahymena mtGP that we see today.

A rearrangent in MTA, marked by the loss of intron #4 and the appearance of intron #6.5, occurred in an ancestor of the “The-Mal” subgroup.

It seems very likely that every step was quickly followed by waves of paralogous diversification of MTA and MTB to generate different mating type specificities.

It is possible that steps 2 and 4 occurred at once, i.e., step 2 was a “palindromic duplication”, and thus the head-to-head MTA-MTB contiguity preceded their functional differentiation. However, this seems unlikely because intra-strand gene conversion within the palindrome would have precluded the functional differentiation of the two genes, due to mutual DNA sequence self-correction among the two copies, such as described in metazoan Y chromosome palindromic duplications (Trombetta and Cruciani, 2017).

These findings raise the question, what could have been gained by having MTA and MTB genes immediately adjacent to one another? The tight linkage of mating related genes provides important advantages, such as to “facilitate the coordinated expression” and “cosegregation of the interacting genes” (Uyenoyama, 2005). One additional consequence of MTA-MTB contiguity in Tetrahymena is that it minimizes the frequency of selfing among sexual progeny that would otherwise occur as a consequence of independent allelic assortment of the two genes in the MAC. If the MTA and MTB genes were located on different MAC chromosomes, the two genes would assort independently in double heterozygotes. MACs that are pure for non-cognate (heterotypic) MTA and MTB genes would then be frequently generated, ultimately resulting in non-assorting selfers (see Data S3 and Figure S9 for a detailed explanation). Thus, reducing the length and the sequence similarity of the intergenic segment between MTA and MTB genes has the effect of minimizing selfer-generating germline or somatic recombination events. The evolution of the mtGP, with its tight contiguity of the cognate (homotypic) MTA-MTB genes, likely was a major step in the evolution of the cross-breeding genetic economies generally observed in Tetrahymena species today.

Finally, the mtGP organization has also proven its versatility by allowing the evolution of additional genetic and molecular mechanisms of MTD capable of adjusting selfing frequency in Tetrahymena. These mechanisms can promote outbreeding or inbreeding under conditions of low or high reproductive stress, respectively (see below for details). A better understanding of these mechanisms will come when studies of the MIC organization of the mating type loci of the various species become available.

True-breeding selfing in T. shanghaiensis

Our work has confirmed that T. shanghaiensis is a true-breeding selfer and has shown that it contains a single mtGP in its MAC that behaves structurally and functionally like the mtGPs of the other species investigated here (Figures 2, 3, 4, S2, S4, and S6). The simplest explanation for its obligatory selfing behavior is that T. shanghaiensis mtGP is heterotypic, i.e. its MTA and MTB genes have different mating type specificity. This conclusion is supported by the finding that T. thermophila mutants, which fail to complete MTD and are left in the MAC with intermediates containing a complete MTA and a complete MTB gene but with different mt specificity, are non-assorting selfers (Lin and Yao, 2020) just like wild-type T. shanghaiensis. Evolutionarily, the T. shanghaiensis chimeric mtGP could have been generated by a simple DNA rearrangement, such as a non-homologous meiotic recombination event between two normal mtGPs of different mating types occurring at the MTA-MTB intergenic region in a heterozygote, resulting in the replacement of either gene with a homolog of different mating type specificity. An alternative way in which the heterotypic mtGP could have arisen is by successive mutations in the two genes of an initially homotypic mtGP that promoted increasingly strong interactions between their encoded proteins ultimately leading to efficient selfing, favored under conditions of high reproductive stress.

As a perpetual selfer, T. shanghaiensis can be considered to be unisexual, in the sense used to describe mating in the absence of any intra-species diversity at the mating type locus, as occurs in some fungi (reviewed by Heitman (2015)). Indeed, that author argues that the last eukaryotic common ancestor, was unisexual, i.e., a perpetual selfer using the Tetrahymena terminology. The species tree (Figure 2B) shows that T. shanghaiensis is surrounded by species with homotypic mtGPs. Thus, it seems most probable that the unisexuality (perpetual selfing) of T. shanghaiensis is derived, rather than ancestral in the genus Tetrahymena. That does not exclude, however, the possibility of a unisexual common ancestor of the “Borealis” and “Australis” Tetrahymena clades.

The intron #4 and #6.5 rearrangement provide a glimpse into the evolution of a different mating type system

While sexual reorganization events (meiosis, fertilization, MAC differentiation) are highly conserved among Ciliates, a striking variety of mating type systems have evolved within this group (Phadke and Zufall, 2010). For example, in the multiple mating type system of the hypotrich Ciliate Euplotes, the two proteins that embody ligand and receptor function for each mating type specificity are the products of intron-splicing variants of the same gene; both proteins are very small, in the order of 40 amino acids, reminiscent of cytokines of multicellular eukaryotes (Luporini et al., 1986; Miceli et al., 1992). In the binary mating type system of the heterotrich Ciliate Blepharisma, the mating type ligand is not a protein but a small tryptophan-related molecule (Miyake, 1996; Sugiura et al., 2005). This diversity implies that the molecules that embody mating type specificity in Ciliates have independently undergone major successive replacements. However, the lack of evolutionary intermediates makes it extremely challenging to trace how this diversity evolved among major groups.

The results reported here have allowed us to infer a succession of replacement waves, occurring within the genus Tetrahymena, which generated a diversity of mating type proteins from an ancestral “Furin-like repeat” protein (Figure 6). Serendipitously, this work also uncovered a more recent replacement wave in which an MTA allele, generated along the way by the loss of conserved intron #4 and the gain of intron #6.5, is inferred to have de novo replaced the MTA alleles of mtGPs of every mating type specificity among the sequenced MTA genes in the “The-Mal” subgroup. Each of the rearrangements that resulted in the two intron changes likely happened only once, with the final rearrangement presumably resulting in one MTA gene of one particular mating type specificity. This variant MTA gene then had to spread and diversify, to ultimately be able encode every known MTA mating type specificity found today in the “The-Mal” subgroup. It seems reasonable to expect that MTB genes also had to co-evolve, in order to allow all the appropriate positive and negative mating type protein interactions required to promote mating between different mating types and prevent selfing in the multiple mating type system. The generation of “raw material” for the re-evolution of multiple mtGPs of different mating type—all containing the variant introns in the MTA gene—was likely facilitated by two special features: the tandem array organization of multiple mtGPs in the T. thermophila germline (micronuclear) genome (Cervantes et al., 2013), in combination with unequal meiotic crossing over, a capacity which has been well documented in the case of Tetrahymena leucine-rich repeat genes (Xiong et al., 2019).

The mtGP: a durable and effective vehicle for Tetrahymena unicell adaptation to reproductive stress fluctuations

T. thermophila cells, maintained by asexual reproduction in the laboratory for long periods in the absence of mating, eventually become sterile (Simon and Nanney, 1979). This failure was inferred to be due to the random accumulation of deleterious mutations in the MIC. The time-sensitive deterioration of their germline results in the susceptibility of Tetrahymena cells to reproductive stress.

Most Tetrahymena species tend to be primarily outbreeders. As previously discussed (Orias et al., 2017), this represents a balance between mechanisms that promote outbreeding and inbreeding. Features of Tetrahymena biology that promote outbreeding include a long sexual immaturity period, intranuclear coordination during MTD, allelic assortment and, in the “Australis” clade, synclonal MTD. Features capable of promoting inbreeding include multiple mating type systems, selfing, and karyonidal MTD in the “Borealis” clade. This investigation of the mating type genes of a broader set of Tetrahymena species has contributed an additional finding relevant to the balance between inbreeding and outbreeding, namely the rare occurrence of a putative heterotypic mtGP, which ensures obligatory, perpetual selfing in T. shanghaiensis.

Some asexual Tetrahymenas, such as T. pyriformis and T. vorax also studied here, have lost their MIC (so called amicronucleates, amics) and can no longer conjugate. Such an extreme feature avoids the germline deterioration that would affect sexual Tetrahymena cells under conditions of severe sexual reproductive stress. Tetrahymena amics are presumably capable of long-term adaptation to changing environments by virtue of allelic assortment in the polyploid MAC. The number of copies of favorable mutations can increase by random assortment and come to phenotypic expression, while unfavorable mutations can be eliminated.

A puzzling feature of the asexual strains investigated here is that their mtGP retain open reading frames (at least for the sequenced exons), even though the proteins are no longer needed for mating, and their expression is not induced by starvation. One trait that would delay the emergence of internal in-frame stop codons is the variant Ciliate genetic code, which has a single stop codon, UGA. It is also possible that the mtGP has other useful function(s), unrelated to mating and expression is induced by some condition other than starvation, which keeps it under selection.

The species investigated here were chosen to provide a sample of the diversity of the Tetrahymena genus, which now contains nearly 100 identified species, with no end in sight. As the breeding systems of additional Tetrahymena species are molecularly characterized, our current picture of how they have evolved in reaction to fluctuating levels of reproductive stress will no doubt be enlarged and enriched.

Limitations of the study

In this work, we mainly focused on the MAC mtGP. Even though these results provide some clues of the evolution of the MIC mating type locus, there are still many unknowns. Subsequent studies of MIC mtGPs should provide a more elaborate picture of the evolutionary process. In addition, the most distantly related species, T. paravorax, seems to be asexual, so it will be more informative if we can find and investigate a sexual outgroup species in the future.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Wei Miao (miaowei@ihb.ac.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

The accession number for the data reported in this paper is GEO: PRJNA510545.

Methods

All methods can be found in the accompanying Transparent methods supplemental file.