Evolutionary divergence of the sex-determining gene MID uncoupled from the transition to anisogamy in volvocine algae

Gamete size dimorphism (anisogamy or oogamy) is a nearly ubiquitous trait in multicellular eukaryotes, and is thought to have originated from an ancestrally isogamous state that is still found in most unicellular eukaryotes (Bell, 1978; Lehtonen et al., 2016; Togashi and Cox, 2011). Although gametic differentiation plays a crucial role in the evolution of sex, the molecular evolutionary bases for the transitions from isogamy to anisogamy (unequally sized gametes) and oogamy (small motile sperm, large immotile eggs) have been difficult to study in most extant lineages such as plants and animals owing to the ancient origins of this innovation.

Volvocine algae form a monophyletic clade that encompasses the unicellular genus Chlamydomonas and multicellular genera with different gradations of size and complexity, including Gonium, Pleodorina and Volvox, the latter of which contains a few thousand cells and exhibits germ-soma differentiation and other developmental innovations that result in a functionally integrated multicellular organism (Matt and Umen, 2016). Members of the smaller, less complex genera such as Chlamydomonas and Gonium have isogamous mating systems with two mating types, while anisogamy characterizes most intermediate genera such as Eudorina and Pleodorina, and oogamy is found in the most complex genus Volvox (Nozaki, 1996, 2003; Nozaki et al., 2000) (Fig. 1).

In the isogamous species Chlamydomonas reinhardtii, cells have either a minus or plus mating type. minus and plus gametes are morphologically similar, yet express mating type-specific genes that allow fusion with a partner of the opposite mating type (Goodenough et al., 2007). The differentiation of minus and plus gametes in C. reinhardtii is governed by a mating locus (MT) whose two haplotypes, MT+ and MT−, are large, rearranged multigenic regions that are suppressed for recombination and segregate as single Mendelian alleles (De Hoff et al., 2013; Ferris and Goodenough, 1994). The C. reinhardtii MID gene (CrMID) encodes a putative RWP-RK family transcription factor that is found in the MT− haplotype and is necessary and sufficient to specify the minus mating type (Ferris and Goodenough, 1997). Proteins from the RWP-RK family have also recently been shown to play a key role in the life cycles of plants, where they control gametophyte identity or gametophyte-sporophyte transitions (Koi et al., 2016; Kőszegi et al., 2011; Rövekamp et al., 2016; Waki et al., 2011), and RWP-RK or Mid-like proteins were reported in possible prasinophyte algal sex-determining regions (Blanc-Mathieu et al., 2017; Worden et al., 2009) and in the mating locus of the ulvophyte green alga Ulva partita (Yamazaki et al., 2017). Yet, little is known about how RWP-RK proteins have undergone functional diversification within and between lineages in the Viridiplantae, where they are ubiquitous (Chardin et al., 2014).

Vegetatively (asexually) reproducing Volvox carteri spheroids of either sex are morphologically identical, but upon exposure to the glycoprotein hormone sex-inducer (Kochert and Yates, 1974; Starr and Jaenicke, 1974; Tschochner et al., 1987), both sexes undergo modified developmental programs that result in differentiation as egg-bearing females or sperm-bearing males (Kochert, 1968; Starr, 1969). Like the case for C. reinhardtii mating types, sexual differentiation in V. carteri is under the control of a dimorphic mating locus with two haplotypes, MTF (female) and MTM (male), where the V. carteri MID gene (VcMID) is found only in MTM (Ferris et al., 2010; Umen, 2011). We have previously found that VcMID is sufficient to induce spermatogenesis when expressed in V. carteri females, and that in its absence germ cell precursors differentiate as eggs (Geng et al., 2014). Therefore, Mid protein has maintained a homologous function in volvocine algae as a dominant determinant of minus/male sexual differentiation.

MID genes have been found in the minus or male mating haplotypes of other volvocine algae, including several isogamous Gonium species (Hamaji et al., 2008, 2013; Setohigashi et al., 2011) and in anisogamous Pleodorina starrii (Nozaki et al., 2006), suggesting that the genetic basis of sex or mating type determination is conserved throughout the volvocine lineage (Fig. S1A,B). Interestingly, a MID gene from the homothallic species V. africanus (VaMID) showed expression correlating with the degree of male differentiation in monoecious versus male sexual spheroids, suggesting that MID is associated with the male-female differentiation switch even when sexes are not determined by a dimorphic mating locus (Yamamoto et al., 2017). Other than the MID gene, no sex-related genes are universally conserved among the MT loci of V. carteri, Gonium pectorale and C. reinhardtii (Hamaji et al., 2016). Even though MID is a rapidly evolving gene, the finding that MID from C. incerta [now reclassified as C. globosa (Nakada et al., 2010)] can substitute for CrMID indicates that functional conservation of Mid proteins can be retained after speciation (Ferris et al., 1997). However, CrMID was not able to substitute for its ortholog in V. carteri, suggesting that significant changes in Mid sequence or in its regulatory network were required for the transition to anisogamy and oogamy in the volvocine lineage (Geng et al., 2014).

Here we set out to more clearly define the point at which Mid acquired its male-determining function in volvocine algae by ascertaining whether MID genes from the isogamous species G. pectorale (GpMID) and/or from the anisogamous species P. starrii (PsMID) were able to functionally substitute for VcMID in inducing spermatogenesis in V. carteri females. We found that PsMID was able to induce spermatogenesis when expressed in V. carteri females and, unexpectedly, that GpMID also had this capability, although it was not as effective as PsMID or VcMID. By contrast, neither VcMID nor GpMID could complement a C. reinhardtii mid mutation, a result most likely due to instability of the non-native Mid proteins in C. reinhardtii. Our data suggest that early functional divergence of Mid proteins between unicellular and multicellular volvocine clades underlies the lack of interspecific compatibility between Mid orthologs, and that changes in the sex-determination network unrelated to Mid protein were mainly responsible for the transition from isogamy to anisogamy and oogamy.

In order to test whether PsMid is capable of inducing spermatogenesis in V. carteri, we transformed a female V. carteri strain (Eve #15) with the PsMID gene expressed under the control of its own promoter/terminator and fused to a blue fluorescent protein (BFP) and a hemagglutinin (HA) epitope tag at its C-terminus (pPsMID-BH), similar to pVcMID-BH that was used previously for generating the V. carteri MID transgenic strains (Geng et al., 2014) (Fig. 2A, Fig. S2A). Briefly, MID-containing expression plasmids were co-transformed with a nitA plasmid (encoding nitrate reductase) into a nitA⁻ strain, and nit⁺ transformants were selected and further tested.

When wild-type vegetative stage females are exposed to sex-inducer their reproductive cells (gonidia) undergo modified embryogenesis to produce sexual progeny containing 32-48 eggs and ∼2000 somatic cells (Kochert, 1968; Starr, 1969; Umen, 2011) (Fig. 2B). When wild-type vegetative males are exposed to sex-inducer, their gonidia also undergo modified development to produce sexual progeny containing 128 somatic cells and 128 sperm packets, with each sperm packet containing 64 or 128 sperm cells (Fig. 2C, Fig. S6A). In control experiments with Eve::VcMID-BH transformants, we observed 100% conversion of presumptive eggs into sperm packets after sexual induction (Geng et al., 2014).

We identified four Eve::PsMID-BH-containing transformants (#1-#4, Materials and Methods), two of which (#2, #3) were examined in more detail (Table 1 and see below). All four transformants showed normal vegetative development (Fig. S3A), and upon sexual induction produced sperm packets and eggs in different proportions, which ranged from 95% sperm packets to equal ratios of sperm packets and eggs (Fig. 2D,E, Table 1). All four transformants also exhibited self-fertility, which is a phenotype we previously observed in V. carteri male MID partial knockdown strains that had a homothallic monoecious (hermaphroditic) phenotype (Geng et al., 2014) (Fig. 2E, Table 1). We chose transformant line #2, which made mostly sperm packets (95% sperm packets, 5% eggs) (Fig. 2D), and line #3, which produced about an equal ratio of sperm packets and eggs (Fig. 2E), to assess the expression of PsMID-BH by semi-quantitative RT-PCR and to assess PsMid-BH protein levels by immunoblotting. Although expression of PsMID mRNA in P. starrii males is normally induced by nitrogen starvation (Nozaki et al., 2006), we found that expression levels of PsMID-BH in V. carteri were similar to those of VcMID-BH (Fig. 2F) and were not influenced by sexual induction or nitrogen availability (Fig. S4A). Full-length PsMid protein was detected in both strains and its levels were higher for line #2 than line #3, correlating with its higher level of spermatogenesis induction (Fig. 2G, Table 1). However, neither of the PsMID-BH-expressing strains produced as much protein as VcMID-BH controls (Fig. 2G). It is not clear why there are similar mRNA levels (detected using BFP primers) but different protein levels for lines #2 and #3, but it is possible that the upstream portions of the mRNAs in each transgenic strain are slightly different due to position effects at their respective genomic insertion sites (i.e. transcription start site or 5′ UTR differences) and impact translation efficiency. It is also possible that the RT-PCR assay loses some accuracy at the higher cycle numbers required to detect MID transgenes compared with the internal control transcript from RPS18. Our main interest was in assessing relative protein levels of PsMid-BH, so we did not pursue this observation further.

Similar to previously described phenotypes for VcMID-BH transgenic females (Geng et al., 2014), both of the characterized PsMID-BH-expressing strains had spermatogenesis-related developmental defects, including sperm cell morphological abnormalities and delayed sexual development (Figs S5 and S6). They also showed delayed or incomplete hatching of sperm packets from their vesicles that was more severe than the delayed vesicle hatching in VcMID-BH transformants (Table 1, Fig. 2E, Fig. S3B,C, Fig. S6). However, unlike sperm from VcMID-BH transformants, which were eventually released from their mother spheroid (Fig. S3B, Fig. S6B), most of the PsMID-BH transformant sperm from both lines #2 and #3 were unable to leave the mother spheroid and remained trapped within a partially dissolved sperm packet vesicle (Fig. 2E, Fig. S3C, Table 1).

Although the BFP-HA tag fused to VcMid had no detectable impact on the function of endogenous VcMid-BH protein expressed in V. carteri females (Geng et al., 2014) (Table 1), the weaker penetrance and more severe sperm hatching defects in Eve::PsMID-BH versus Eve::VcMID-BH transformants could have been partly due to the impact of the epitope tag on PsMid function. Therefore, we generated untagged PsMID constructs (Fig. S2B) and tested four independent Eve::PsMID transformants identified by PCR genotyping (Geng et al., 2014). Unlike the case for tagged PsMID-BH constructs, where conversion to sperm packets was incomplete, in all four untagged Eve::PsMID transformants we observed 100% conversion of eggs to sperm packets, indicating that the BFP-HA tag may weaken Mid activity and/or partially destabilize the protein. Interestingly, however, the phenotypes of incomplete vesicle hatching and inability to release from mother spheroids were not alleviated in Eve::PsMID strains (Fig. S3C, Fig. S6B, Table 1). We note that these hatching-related phenotypes are unlikely to be just the result of reduced PsMID expression as they were not observed in V. carteri male strains in which endogenous VcMID expression was partially reduced by RNAi (Geng et al., 2014). We conclude that PsMid protein is likely to have somewhat weaker or altered activity compared with VcMid protein, and that these differences with native VcMid function led to more severe defects in sperm packet hatching from their vesicles and in the release of sperm from the mother spheroid in PsMID-expressing females. Although the BFP-HA tag caused some functional impairment of Mid proteins, use of the tag also served as a diagnostic tool to help evaluate relative Mid activity, so we continued to use it in combination with untagged constructs for experiments with MID from the isogamous species Gonium pectorale, as described below.

As we did for PsMID, we generated tagged and untagged GpMID transgenic constructs driven by its native promoter/terminator (Fig. 3A,B, Fig. S2C,D) and assessed their phenotypic effects when expressed in female V. carteri transformants. We also generated a tagged construct that expressed the GpMID cDNA (GpMIDcDNA-BH) (Fig. 3C, Fig. S2E) to control for possible defects in RNA processing that we observed in transformants expressing the native GpMID gene (see below). Five independent untagged Eve::GpMID transformants (#1-#5) all had self-fertile hermaphrodite/homothallic phenotypes with 90-95% conversion of eggs to sperm packets. However, among all five independent Eve::GpMID lines we very rarely saw complete conversion to sperm packets (<1% of transformants), in contrast with what we observed for untagged PsMID or VcMID transgenics (Fig. 3D, Table 1). Supporting the idea that the BFP-HA epitope tag somewhat weakened the expression or activity of Mid proteins, we found that tagged Eve::GpMID-BH and Eve::GpMIDcDNA-BH strains both had only ∼40% of their eggs converted to sperm packets (Fig. 3E, Fig. S7A, Table 1). Nonetheless, it was remarkable that GpMid from an isogamous species could largely substitute for VcMid in driving ectopic spermatogenesis in V. carteri females.

Other than the degree to which eggs were converted to sperm packets, strains expressing each of the three different GpMID constructs we tested (untagged genomic, tagged genomic, tagged cDNA; Fig. 3A-C) had nearly identical developmental phenotypes as described below. Besides the formation of ectopic sperm packets and self-fertility (Fig. S7A-C), the additional phenotypes included sperm cell developmental abnormalities and incomplete or delayed hatching from the sperm vesicle that were even more severe than those observed for PsMID transformants: in GpMID transformants, most sperm packets never hatched from their vesicle, but instead dissolved within it; and we never observed sperm or sperm packets released from parental spheroids (Fig. S6B, Fig. S7A, Table 1). Finally, unlike the case for natural V. carteri males or for female VcMID and PsMID transformants, we never observed spontaneous production of sexual spheroids (i.e. self-induction) with any of the GpMID transformants, all of which had to be treated with exogenously applied sex-inducer to initiate sexual development (Table 1).

Besides the sperm development and hatching abnormalities observed in GpMID-expressing transformants, all of the different GpMID constructs we tested caused additional sexual stage and vegetative stage phenotypes that were never observed in PsMID or VcMID transgenic strains. These included smaller vegetative and sexual spheroids with fewer vegetative gonidia or sexual germ cells than in control strains (Fig. S8, Table S1). About 30-50% of the vegetative and sexual spheroids that expressed GpMID had unevenly spaced and disorganized somatic cells (Fig. 3F,G, Fig. S8A), and ∼30-40% of the vegetative-phase spheroids from the same strains were misshapen, which might have been caused by incomplete or aberrant inversion – the post-cleavage reversal of embryo curvature that turns the spheroid right-side out (Fig. S8B) (Kirk and Nishii, 2001; Sessoms and Huskey, 1973). We classified these abnormal phenotypes as pleiotropic because they did not relate to any processes associated with normal sexual development or germ cell formation.

The pleiotropic defects in GpMID transformants prompted us to examine whether the expected mRNA and proteins were being produced. mRNA for GpMID-BH, as detected using BFP primers, was expressed at similar levels to VcMID-BH and to PsMID, and was not affected by nitrogen availability or presence of sex-inducer (Fig. S4B,C), even though in its native context GpMID is induced by nitrogen starvation (Hamaji et al., 2008). However, most of the GpMID mRNA produced in V. carteri female transformants expressing GpMID and GpMID-BH was partially processed, with a variety of isoforms, the most abundant of which skipped exon 3 and used an internal 3′ splice acceptor within exon 4 to produce a shorter mRNA containing a predicted frameshift (Fig. 3H, Fig. S4D, Fig. S9B). On the other hand, the only message detected for GpMIDcDNA-BH transformants was full length (Fig. 3I) and it was expressed at similar levels to VcMID-BH (Fig. S4E), indicating that altered mRNA processing and the resulting frameshifted polypeptides are unlikely to underlie the phenotypic effects associated with GpMID expression in V. carteri females.

Immunoblots from strains expressing tagged GpMIDcDNA-BH or GpMID-BH detected proteins that migrated as predicted for full length or near full length, but also smaller products that are likely to be stable degradation fragments containing the C-terminal epitope tag (Fig. 3J, Fig. S7E). The total amount of GpMid-BH protein signal detected in transgenic GpMIDcDNA-BH strains was less than for VcMid-BH (Fig. 3J) and this reduction was most likely attributable to reduced translation or protein stability, as the detectable mRNA for this construct was all full length (Fig. 3I) and expressed comparably to VcMID-BH mRNA (Fig. S4E). Together, these findings suggest either inefficient translation or protein instability as the cause for reduced steady-state levels of GpMid-BH in transgenic strains. Importantly, the pleiotropic developmental defects caused by GpMID expression in V. carteri females were not influenced by the presence/absence of the epitope tag or by the substitution of a cDNA for the genomic version (Table S1). In summary, the pleiotropic developmental phenotypes observed when GpMID was expressed in V. carteri were likely to be due to neomorphic (i.e. off-target) interactions between native GpMid protein (or truncated versions) and other developmental regulators in V. carteri.

We previously found that CrMID and chimeric constructs expressing combinations of N-terminal and C-terminal domains of VcMID and CrMID could be expressed in V. carteri females but could not induce spermatogenesis (Geng et al., 2014). Experiments to test heterologous MID genes in C. reinhardtii have only been reported for one closely related sister species, C. incerta [now called C. globosa (Nakada et al., 2010)], where cross-species complementation was observed (Ferris et al., 1997). In control experiments, we found that complementation of a C. reinhardtii mid deletion mutant strain CC3712 by the endogenous CrMID gene or a C-terminally FLAG epitope-tagged version of CrMID-6XFLAG occurred at low frequency and was often just partial (evidenced by self-agglutination), as was previously reported (Ferris and Goodenough, 1997; Lin and Goodenough, 2007) (data not shown). To overcome the problem of low efficiency expression and complementation with native CrMID, we made a version of the FLAG epitope-tagged MID construct in which the native MID promoter/terminator were replaced with those of ribosomal protein RPL23 (pL23:CrMID-6XFLAG) (Fig. S10A), which we have previously used to generate high-level transgene expression in C. reinhardtii (Li et al., 2016; López-Paz et al., 2017). We co-transformed CC3712 with the pL23:CrMID-6XFLAG construct and a hygromycin resistance (HygR) marker gene (Berthold et al., 2002) and screened HygR colonies for presence of the transgene (Materials and Methods). Around 24% of the HygR colonies tested positive for having the MID transgene (22/91), and most of these positive colonies (19/22) showed full complementation of the mid deletion phenotype (100% minus mating phenotype; no self-agglutination) and produced CrMid protein that was detectable by immunoblotting (Fig. 4A, Table 2).

The high-frequency complementation of the mid deletion mutation obtained with this control construct facilitated testing of heterologous MID genes for complementation in C. reinhardtii. We transformed RPL23 promoter-driven FLAG-tagged cDNAs for GpMID and VcMID (Fig. S10B,C) into CC3712 and assessed co-transformation rates, mating type complementation, mRNA expression, and protein levels. Co-transformation rates with these constructs were similar to those we observed for pL23:CrMID-6XFLAG (24/112 and 13/96, respectively, by colony PCR), but in no case did we see even partial complementation of the mid deletion phenotype; nor were we able to detect tagged protein for any of the GpMID and VcMID transformants despite detecting the expression of full-length GpMID and VcMID mRNAs in several selected strains (Fig. 4B-D). Although in some circumstances codon optimization can improve transgene expression (Barahimipour et al., 2015), the endogenous CrMID gene shows almost no codon bias relative to optimal codon usage in C. reinhardtii (Ferris et al., 1997), yet the endogenous transgenic CrMid protein was easily detectable by immunoblotting (Fig. 4A). Moreover, the GpMID and VcMID genes have C. reinhardtii codon adaptation indices (CAIs) in the same range as the endogenous CrMID gene (CrMID, 0.370; GpMID, 0.429; VcMID, 0.324). By contrast, well-expressed genes such as TUA1 (alpha tubulin) and RPL23 (cytoplasmic ribosomal protein) have CAIs of 0.813 and 0.751, respectively. We conclude that heterologous Mid proteins from G. pectorale and V. carteri are poorly expressed in C. reinhardtii, most likely because of protein instability and possibly inefficient translation, and that poor expression may have precluded complementation.

Our prior work on MID in V. carteri and previous studies of MID in C. reinhardtii showed that the presence/absence of native MID gene expression in sexually induced individuals is necessary and largely sufficient to determine mating type or germ cell differentiation programs (Ferris and Goodenough, 1997; Geng et al., 2014; Lin and Goodenough, 2007). Further work showing the inability of ectopic CrMID or chimeras between CrMID and VcMID to induce spermatogenesis in V. carteri females led to a simple hypothesis that the evolution of sexual dimorphism and oogamy in the volvocine lineage was due to molecular changes in the Mid protein (Geng et al., 2014). The experiments described here rule out the simplest version of this hypothesis because GpMid from the isogamous species G. pectorale could induce spermatogenesis when ectopically expressed in V. carteri females, albeit at reduced efficiency (Fig. 3, Table 1). It would still be possible that anisogamy/oogamy coevolved with Mid if Gonium or related clades had been ancestrally anisogamous/oogamous but secondarily lost this trait. If so, then GpMid might still have residual functionality in a dimorphic sex-determining regulatory circuit. However, secondary loss of anisogamy/oogamy is unlikely since all species of the genus Gonium and other genera at or near the base of the multicellular volvocine radiation (Astrephomene and Tetrabaena/Basichlamys) are isogamous (Nozaki, 1996, 2003; Nozaki et al., 2000, 1996).

Our data on cross-species complementation with Mid instead support a phyletic model in which Mid functional divergence was not coupled to the evolution of gamete dimorphism, but occurred in a more gradual manner that was proportional to divergence times between lineages. Under this model, the Mid protein from the Chlamydomonas lineage has accumulated too many changes for it to function in the multicellular volvocine taxa (and vice versa), but the core interactions between Mid, its target DNA sequences, and other transcriptional regulators have been sufficiently conserved within the multicellular taxa (Volvox, Pleodorina, Gonium) for it to retain basic function across these genera (Fig. 1). We note that functional divergence and loss of interspecific compatibility between transcription factor (TF) orthologs is not an inevitable outcome over the time scale of 200-300 million years (MY) during which volvocine algae diversified (Herron et al., 2009). For example, the well-known metazoan Eyeless/PAX6 TFs retained function in eye development between fruit flies and mice over 800 MY despite the evolution of anatomically completely different eye structures between arthropods and vertebrates (Kumar et al., 2017; Shubin et al., 2009). Similarly, an ortholog of the circadian TF CONSTANS (CO) from Chlamydomonas was able to complement an Arabidopsis CO mutant (Serrano et al., 2009), although the two CO orthologs share significantly less sequence similarity (32% identity/38% similarity) than the most diverged Mid orthologs in volvocine algae (41% identity/52% similarity) (Fig. S1), and exhibit a much deeper phylogenetic divergence of ∼1000 MY (Kumar et al., 2017). Ideas about how Mid proteins and the Mid sex-determination network might have diverged in function within the volvocine algae are discussed below.

The ability of GpMid from an isogamous species to partially function in V. carteri male differentiation means that the transition to anisogamy/oogamy was not driven directly by changes in Mid, but mainly by changes in its direct and indirect target genes that evolved to control the developmental programs leading to gamete dimorphism. Changes in TF target genes can result from alterations in the DNA binding specificity of a TF (and/or its co-regulators) or from loss/gain of cis-regulatory elements (Wray et al., 2003). Sexual development and mating behaviors in V. carteri are significantly more complex than mating type specification and gamete interactions in C. reinhardtii or Gonium (Nozaki, 1996; Umen, 2011), implying that during the evolution of anisogamy/oogamy Mid must have gained control over new genes and/or regulons responsible for this additional complexity. In the isogamous, unicellular C. reinhardtii a few hundred genes exhibit mating type-specific differential regulation (Joo et al., 2017; Lopez et al., 2015; Ning et al., 2013) but very little is known about mating type and sex-regulated genes in multicellular volvocine algae (Ferris et al., 2010; Umen, 2011). We predict that many new direct or indirect targets of Mid arose in the transition from isogamy to anisogamy/oogamy, and future work aimed at identifying Mid targets in isogamous and sexually dimorphic volvocine species will help illuminate how the Mid network changed and/or expanded in this lineage.

Our data suggest that the largest changes in the Mid network were unrelated to sexual dimorphism and occurred after the split from the last common ancestor shared by C. reinhardtii and the multicellular volvocine clade (V. carteri, P. starrii and G. pectorale) (Fig. 1). We hypothesize that, after this split, Mid proteins from the two lineages gradually became incompatible due to altered network interactions, ultimately leading to the instability of heterologous Mid proteins expressed in C. reinhardtii (Fig. 4) and the inability of CrMid to function in V. carteri (Geng et al., 2014). These incompatibilities could have arisen from neutral drift in rapidly evolving MID genes (Ferris et al., 1997; Yamamoto et al., 2017) that resulted in altered protein-protein interactions, perhaps similar to what has been observed in TFs that control fungal mating type regulatory circuits (Tuch et al., 2008a,b). TFs are often targeted for rapid degradation and may be especially sensitive to changes in structure or conformation that are coupled to regulated turnover (Kodadek et al., 2006; Yao and Ndoja, 2012). We speculate that our inability to detect expression of heterologous Mid proteins in C. reinhardtii (Fig. 4) and the prevalence of non-full-length Mid protein fragments for GpMid and CrMid proteins expressed in V. carteri (Geng et al., 2014) (Fig. 3J, Fig. S7E) reflect the lability of Mid protein and sensitivity to changes in its interaction network. For example, if Mid is protected from degradation by binding to a second stabilizing protein with which it has coevolved, then a weakened interaction between these two partners in the case of non-endogenous Mid proteins could lead to decreased Mid stability. The pleiotropic developmental phenotypes caused by expression of GpMid in V. carteri might further reflect promiscuous interactions of GpMid with off-target partners or cis-regulatory motifs that are also thought to coevolve with TFs (Baker et al., 2011; Nadimpalli et al., 2015; Sayou et al., 2014).

Although the core functions of spermatogenesis were fulfilled by PsMid and GpMid proteins when expressed in V. carteri, we highlight below new or enhanced phenotypes caused by heterologous MID gene expression that shed light on different aspects of V. carteri sexual development and the Mid pathway: self-induction, sperm packet release from vesicles, and sperm packet release from parental spheroids.

An interesting property of the V. carteri sex induction pathway is its ultra-sensitivity to the sex-inducer hormone, a glycoprotein related to other hydroxyproline-rich glycoproteins that constitute most of the V. carteri extracellular matrix (ECM) (Gilles et al., 1984; Hallmann, 2003; Mages et al., 1988; Starr and Jaenicke, 1974). Males and females can produce sex-inducer in response to stress (Kirk and Kirk, 1986; Nedelcu and Michod, 2003), but sex-inducer is also produced spontaneously by males (Starr, 1970). In a poorly understood process, an initial sex-inducer signal can be amplified so that a single spontaneously produced sexual male can induce the remaining spheroids in an entire culture (or, presumably, an entire natural pond environment) to undergo sexual development, a process termed self-induction. Wild-type V. carteri males frequently self-induce in culture (Kirk, 1998; Starr, 1970) and we previously observed the same phenomenon in V. carteri females expressing VcMID (Geng et al., 2014). We also observed self-induction in V. carteri PsMID transformants, but never for GpMID transformants (Table 1). Although the lack of self-induction in GpMID transformants could be attributed to low levels of GpMid protein in transgenic strains, we note that the best sperm packet conversion rates of GpMID transgenics (∼95%) were higher than those of the weakest PsMID transformants (∼50%), yet the PsMID transformants all showed normal self-induction (Table 1). This suggests that some aspect(s) of the sex-inducer production/sensing/amplification process are under the control of Mid, and that the activity of GpMid protein is unable to support the process of self-induction and/or signal amplification.

During normal male development in V. carteri, sperm must breach three ECM-based barriers in order to be released and fertilize a female. The first barrier is the vesicle in which each sperm packet is formed, a structure analogous to the vesicle that encapsulates a vegetative V. carteri embryo, or to a mother cell wall in a postmitotic C. reinhardtii division cluster. The second barrier is the ECM within the parental spheroid and the sheath material surrounding the parental spheroid (Jaenicke and Waffenschmidt, 1981). The third barrier is the female mating partner's ECM, which must be entered through a fertilization pore that forms after a sperm packet contacts a sexual female (Kochert, 1968; Starr, 1969).

In PsMID-expressing and GpMID-expressing V. carteri female transgenic strains, defects in vesicle hatching and parental spheroid hatching were both evident to different degrees, suggesting that production of the relevant sexual stage hatching enzymes was inadequate in these transgenic strains due to weakened or altered Mid activity (Table 1). By contrast, dissolution of sperm packets, possibly mediated by V. carteri sperm lysin (Waffenschmidt et al., 1990), still occurred in all transgenic MID strains (Fig. 2, Figs S3 and S8). Release of sperm packets from their vesicle is likely to be mediated by proteolytic hatching enzymes such as the VHE/lysin/sporangin subfamily of proteases, which in C. reinhardtii are secreted through the daughter cell flagella to enable hatching (Kubo et al., 2009). Once released from their vesicle, sperm packets are still constrained within their parental spheroids by ECM and the outer wall or boundary zone (sheath), which constitutes the second barrier to sperm release (Jaenicke and Waffenschmidt, 1981). An analogous process to sperm packet release from spheroids is daughter colony release during V. carteri vegetative reproduction, a process that is mediated by specific hatching enzymes VheA and LSG2 (Fukada et al., 2006; Nishimura et al., 2016), the former of which is homologous to C. reinhardtii vegetative hatching enzyme (Kubo et al., 2009). During their sexual cycle, C. reinhardtii cells produce a second type of hatching enzyme called gamete lytic enzyme (GLE or G-lysin), which belongs to a family of matrix metalloproteases (MMPs) distinct from vegetative hatching enzyme (Buchanan et al., 1989; Kinoshita et al., 1992; Matsuda et al., 1985). We predict that one or more sexual stage-specific GLE-like or Vhe-like activities might be defective in PsMID and GpMID transformants, leading to defects in sperm packet release from vesicles and parental spheroids, although this remains to be determined. In V. carteri, many ECM genes and ECM MMP-related gene families are expanded relative to those in C. reinhardtii (Prochnik et al., 2010), and a subset of these might be dedicated to sperm packet release/hatching and be produced under the control of VcMid.

Female V. carteri strain Eve (V.c. f. nagariensis UTEX 1885), male V. carteri strain Aichi M (V.c. f. nagariensis NIES 398), P. starrii (NIES 1363) and G. pectorale (NIES 1710) were obtained from UTEX (https://utex.org) or NIES (http://mcc.nies.go.jp/) stock centers. The female nitA⁻ V. carteri strain Eve #15 used for transformations was described previously (Geng et al., 2014). All V. carteri strains were cultured in Standard Volvox Medium (SVM) at 32°C (unless otherwise specified) and synchronized on a 48 h developmental cycle with a 16 h:8 h light:dark regime (Kirk and Kirk, 1983) with a combination of 100 µE blue (465 nm) and 100 µE red (465 nm) LED lights with bubbling aeration. Eve::VcMID-BH lines were grown at 28°C for vegetative cultures and 32°C to obtain sexual cultures. Sexual development was induced by adding pre-titered crude sex-inducer to juveniles 24 h prior to embryonic cleavage (Starr and Jaenicke, 1974). For the nitrogen (N)-free SVM growth experiments, V. carteri cultures were first grown in 300 ml SVM in flasks with 3000 spheroids/flask, then collected and washed with 500 ml N-free SVM using a magnetic funnel. The washed cultures were returned to 300 ml N-free SVM flasks and grown for 24 h prior to RNA isolation.

The C. reinhardtii strains used include the MT− mid deletion mutant CC3712 obtained from the C. reinhardtii Resource Center (www.chlamy.org) and the mating testers CC620 (referred to as R3, MT+) and CC621 derivative CJU10 MT− (Umen and Goodenough, 2001). CC3712 differentiates as a pseudo-plus mating type and agglutinates with minus gametes (but cannot fuse) (Ferris and Goodenough, 1997). Strains were maintained on TAP agar plates and grown in liquid TAP (Harris, 2009).

Multiple sequence alignments of Mid orthologs from volvocine algae were generated in MEGA7 (Kumar et al., 2016) using MUSCLE (Edgar, 2004) and formatted using BoxShade (https://www.ch.embnet.org/software/BOX_form.html). Identity and similarity were calculated using SIAS with default settings (http://imed.med.ucm.es/Tools/sias.html).

Oligonucleotide primers used for plasmid construction are listed in Table S2.

Plasmid pPsMID was created as follows. The PsMID promoter and 5′ UTR (−444 bp to −1 bp) were amplified from P. starrii (N1363) genomic DNA using primers PlestMidpromoter f1 and PlestMidpromoter r1, which introduced flanking SacI and NcoI restriction sites, with the ATG start codon of PsMID within the NcoI site (ccATGg). The PsMID gene region (coding exons and intervening introns, 1354 bp) was amplified using primers PlestMid GOI f1 and PlestMid GOI r1, which introduce flanking NcoI and NheI sites, respectively. The PsMID 3′ flanking region (378 bp following the Stop codon) was amplified with primers PlestMid3′UTRf1 and PlestMid3′UTRr1, which introduce flanking BamHI and KpnI restriction sites. The above three fragments were digested at the appropriate flanking restriction enzyme sites, ligated together, and the full-length product amplified with primers PlestMidpromoter f1 and PlestMid3′UTRr1, and ligated into pGEM-T Easy vector (Promega) to obtain pPsMID (Fig. S2C). pPsMID-BH was created by digesting plasmids pPsMID and pVcMID-BH with NheI and BamHI, and replacing the VcMID gene from pVcMID-BH with the PsMID gene (Fig. S2B).

Plasmid pGpMID was created as follows. The GpMID promoter and 5′ UTR (−185 bp to −1 bp) were amplified from G. pectorale genomic DNA using primers GpMid promoter f1 and GpMid promoter r1, which introduced flanking SacI and NcoI restriction sites, with the ATG start codon of GpMID within the NcoI site. The GpMID gene region (coding exons and intervening introns, 760 bp) was amplified from G. pectorale genomic DNA using primers GpMidGOI.f1 and GpMidGOI.r1, which introduce flanking NcoI and NheI sites, respectively. The GpMID 3′ region (760 bp following Stop codon) was amplified with primers GpMid3UTR.f1 and GpMid3UTR.r1, which introduce flanking BamHI and KpnI restriction sites. The above three fragments were digested at the appropriate flanking restriction enzyme sites, ligated together, and the full-length product amplified with primers GpMid promoter f1 and GpMid3UTR.r1, and ligated into pGEM-T Easy to obtain pGpMID (Fig. S2C). pGpMID-BH plasmid was created by digesting pGpMID and pVcMID-BH with NheI and BamHI, and replacing the VcMID gene from pVcMID-BH with the GpMID gene (Fig. S2D). The GpMID cDNA (495 bp) was amplified from G. pectorale cDNA using primers GpMidGOI.f1 and GpMidGOI.r1, which introduce flanking NcoI and NheI sites, respectively. The cDNA fragment and plasmid pGpMID-BH were digested with NheI and BamHI, and the cDNA fragment was ligated to the backbone of pGpMID-BH to form pGpMIDcDNA-BH (Fig. S2E).

Constructs pL23:CrMID-6XFLAG, pL23:GpMIDcDNA-6XFLAG and pL23:VcMIDcDNA-6XFLAG were assembled and cloned into the EcoRI site of pGEM-T Easy using Gibson assembly reactions (New England Biolabs, E2611) via ∼20 bp overlapping sequences generated by PCR. PCRs were performed with Phusion polymerase (Thermo Scientific) according to the manufacturer's protocol. A spacer (GG) and a 6XFLAG sequence (5′-CGCGACTACAAGGACCACGATGGGGACTACAAGGACCATGACATCGACTACAAGGACGACGACGACAAGGGCGGCCGCGACTACAAGGACCACGATGGGGACTACAAGGACCATGACATCGACTACAAGGACGACGACGACAAGGGCGGCCGC-3′) were inserted immediately before the Stop codon for all constructs. pL23:CrMID-6XFLAG was generated as follows. The stretch of C. reinhardtii MID native promoter-5′ UTR, coding region and 6XFLAG tag sequence were amplified from CrMID-6XFLAG expression plasmid [FLAG Stop #11, a gift of Dr Huawen Lin (Lin and Goodenough, 2007)] using primers L_CrProEcoRIF and 6FLAG_AvrIIR1. Another set of primers, 6FLAG_AvrIIF1 and pGEM-T Easy EcoRI_p3R2, was used to amplify a 725 bp stretch of the terminator-3′ UTR of the C. reinhardtii RPL23 gene (Cre04.g211800) (López-Paz et al., 2017). The plasmid backbone was prepared by digesting pGEM-T Easy vector with EcoRI. Finally, all three fragments, including the EcoRI-cut vector backbone, were assembled using Gibson assembly to yield pCr:CrMID-6XFLAG:L23. Next, the promoter-5′bUTR region of pCr:CrMID-6XFLAG:L23 was replaced with a 1005 bp stretch of RPL23 promoter-5′bUTR sequence to yield pL23:CrMID-6XFLAG:L23 (López-Paz et al., 2017). The 1005 bp L23 sequence was amplified with primers pGEM-T Easy EcoRI_p3F2 and p3promoter_R. The stretch of C. reinhardtii MID coding sequence through the L23 terminator-3′ UTR was amplified from pCr:CrMID-6XFLAG:L23 using primers p3promoter_cN_F and pGEM-T Easy EcoRI_p3R2. Finally, these two PCR products and EcoRI-cut pGEM-T Easy vector backbone were assembled using Gibson assembly.

pL23:GpMIDcDNA-6XFLAG and pL23:VcMIDcDNA-6XFLAG were generated as follows. pGEM-T Easy vector backbone containing RPL23 promoter-5′ UTR and 6XFLAG-L23 terminator-3′ UTR sequence was generated by PCR from pL23:CrMID-6XFLAG:L23 using primers p3promoter_R and 6FLAG_F2. Next, GpMID and VcMID coding sequences were amplified from cDNA-containing plasmids with primers p3promoter_gN_F and GpMID-6FLAGR, and with p3promoter_vN_F and VcMID-6FLAGR, respectively. Finally, each coding sequence was assembled with the PCR-generated vector backbone using Gibson assembly.

PCR genotyping, PCR amplification conditions, RNA and cDNA preparation and RT-PCR on V. carteri strains were as described previously (Geng et al., 2014) using primers listed in Table S2. Selected RT-PCR products were cloned into pGEM-T Easy vector for sequencing. The V. carteri DGAT1 gene was used as a control for nitrogen starvation (www.phytozome.org, Volvox carteri V2.1 gene ID Vocar.0001s0624.1). Primers for amplifying DGAT1 cDNA (DGAT1-F1 and DGAT-R1), BFP cDNA (BFP-F and BFP-R), GpMID-BH (BFP-F and BFP-R; GpMidGOI.f1 and GpMidGOI.r1), PsMID-BH (PlestMid GOI f1 and PlestMid GOI r1) and ribosomal protein S18 cDNA (VcS18-1 and VcS18-2) are listed in Table S2.

Preparation of whole-cell lysates and immunoblotting of V. carteri cultures were performed as described previously (Geng et al., 2014).

The diameters of 20 randomly selected juvenile spheroids were measured 24 h after the start of embryogenesis (i.e. first embryonic cleavage). Numbers of germ cells and germ cell differentiation patterns (eggs or sperm packets) were counted 40 h after the start of embryogenesis. Measurements and photomicrographs were made using a Leica DMI6000 microscope with 10×, 20× or 40× objectives using DIC optics, a Leica DFC 450 camera, and Leica LAS v4.0 software.

The mid deletion strain CC3712 was co-transformed using the glass-bead method (Kindle, 1990) with 2 µg MID expression construct DNA (see above), and 200 ng hygromycin resistance plasmid pHyg3 (Berthold et al., 2002). After recovery overnight under dim light and gentle shaking, the cells were collected by centrifugation (5 min, 2000 g) and spread onto TAP plates supplemented with 35 µg/ml hygromycin B (Gold Biotechnology). Plates were kept under continuous light for 5-6 days, when visible colonies were picked and tested.

Individual colonies that grew on the hygromycin B plates were transferred to individual wells of 96-well plates containing 200 µl TAP medium. After 2 days under continuous light, the transformants were spotted onto TAP plates and grown 5-7 days as small spots ∼5 mm in diameter. For genotypic screening, a tiny amount of cells from each spot was collected using a 10 µl pipette tip and suspended in 100 µl TE (10 mM Tris pH 8.0, 1 mM EDTA). The suspensions were boiled at 95°C for 10 min, centrifuged at 3000 g for 10 min, and 1 µl supernatant was used as template for PCR to detect the presence of the MID transgene using primers CRMID_ATG_F and crMID_HA_crR, or p3promoter_gN_F and GPMID-6FLAGR, or p3promoter_vN_F and VCMID-6FLAGR for CrMID, GpMID and VcMID, respectively. Genotyping PCR reaction mixtures contained 1 μM each primer, 0.2 mM each dNTP, 1× Ex-Taq Buffer (with 2.5 mM MgCl₂) (Takara), 2% DMSO, and 1-2 U Taq polymerase (Invitrogen). Products were amplified using 40 cycles of 94°C for 15 s, 60°C for 30 s, and 72°C for 30 s.

Transformants that tested positive for an MID transgene were randomly selected and tested for mating as follows. About one-quarter of each small spot from a TAP agar plate (see above) was suspended in 200 µl of nitrogen-free HSM (NF-HSM) (Harris, 2009) in a well of a 96-well plate and kept under continuous light for 2 h. Then, 100 µl each suspension was mixed with gametes of MT+ mating tester R3 to observe mating. The other 100 µl was kept to observe potential self-agglutination. Selected non-complemented lines were mixed with MT− mating tester strain CJU10 and observed for agglutination to confirm that gametogenesis had been induced. Mating and agglutination were scored using a Zeiss Axiostar compound microscope with 40× objective and phase optics. Transformants that either mated with MT+ strain R3 or self-agglutinated were scored as positive transgenic lines (Table 2).

Gamete RNA extraction was as follows. About 5×10⁷ cells were harvested and flash-frozen in liquid nitrogen. The pellets were quickly thawed in 3-4 ml Trizol (Invitrogen) by pipetting at room temperature. After 5 min incubation at room temperature, lysates were centrifuged at top speed (10 min, 13,200 g) to remove debris. The supernatants were used for RNA extraction according to the manufacturer's protocol. cDNA was prepared from 2-3 µg total RNA according to the manufacturer's protocol for Superscript III (Thermo Fisher Scientific, 18080–044) using a 10:1 mixture of oligo(dT) and random hexamer for priming. The reaction conditions were: 25°C 10 min, 42°C 10 min, 50°C 20 min, 55°C 20 min, 60°C 20 min, 85°C 5 min. The reactions were diluted 1:7 with TE and used as templates for semi-quantitative RT-PCR with the primers indicated in Table S2. The C. reinhardtii 18S ribosomal RNA gene (GenBank KX781336) was used as an internal control. Primers for amplifying VcMID cDNA (VcMID_ATG_F and VcMid-C-R3), GpMID cDNA (GpMID-BamATG and GpMID-XhoTTA), CrMID (CrMID_F and CrMID_R) and the 18S ribosomal RNA gene (Cr18SrRNA-1 and Cr18SrRNA-2) are listed in Table S2.

Lawns of selected C. reinhardtii strains were spread on TAP plates and grown under continuous light for five or six days. A pea-sized loopful of cells was scraped into NF-HSM and kept in the light for 2 h. Aliquots were tested briefly for mating, then gametes were spun down (5 min, 2000 g) and washed in 1 ml lysis buffer [1× PBS supplemented with 1× ProteaseArrest (G-Biosciences), 5 mM benzamidine, 1 mM PMSF] and finally resuspended in 150 µl lysis buffer before flash-freezing in liquid nitrogen. The cells were thawed at room temperature and spun down at full speed (10 min, 13,200 g) to pellet debris. The supernatant was collected and, prior to mixing with sample buffer, the protein concentration was determined using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) with bovine serum albumin as a standard. Equal amounts of protein (∼30 µg per lane) were loaded for each sample and resolved on 12% SDS-PAGE gels. The gels were blotted to Immobilon-P PVDF membranes (Millipore) for immunodetection. The blotting was performed in an XCell II Blot Module (Invitrogen) with transfer for 48 min at 45 V in buffer comprising 25 mM Tris pH 8.6, 192 mM glycine, 0.1% SDS, 20% methanol. Blocking was performed overnight at 4°C in 9% nonfat dried milk in TBST (TBS with 0.3% Tween 20), followed by incubation with primary antibody in blocking buffer overnight at 4°C. Anti-FLAG rabbit polyclonal (Rockland 600-401-383) was used at 1:10,000 to detect the FLAG epitope tag. The blots were washed three times with TBST at room temperature for 15 min followed by incubation for 1 h with HRP-conjugated goat anti-rabbit secondary antibody (Thermo Scientific Pierce PI31460) at 1:20,000 in TBST. Blots were washed three times as described above and signal was detected by chemiluminescence (Luminata Forte Western HRP substrate, Millipore) using a ChemiDoc CCD camera system (Bio-Rad).

We thank the staff of the Tissue Culture and Transformation core facility at Donald Danforth Plant Science Center for technical assistance with biolistic transformation; Takashi Hamaji for helpful discussion; Takashi Hamaji, Gavriel Matt and Yi-Hsiang Chou for feedback on the manuscript; and Richard Davenport and Jie Li for excellent technical support.