A cis-acting bidirectional transcription switch controls sexual dimorphism in the liverwort

Conserved MYB transcription factors are specifically expressed in the female gametophytes of Marchantia polymorpha and Arabidopsis thaliana

To identify evolutionarily conserved regulators of female sexual differentiation in land plants, we compared transcriptome datasets of archegonia and thalli of M. polymorpha (see Data availability). Genes preferentially expressed in archegonia were screened and further selected for enrichment of related genes in the published transcriptome datasets of A. thaliana female gametophytes (embryo sacs; Yu et al, 2005; Steffen et al, 2007; Wuest et al, 2010). Among the 23 genes thus identified (Table EV1), we focused on Mapoly0001s0061 encoding a MYB-type transcription factor. In a phylogenetic tree constructed from the amino acid sequences of MYB domains, Mapoly0001s0061 was found to be closely related to three Arabidopsis genes, AtMYB64, AtMYB119, and AtMYB98, as well as two homologs in the moss Physcomitrella patens (Fig 2A and B). Because our genetic analyses indicated a role of Mapoly0001s0061 in female gametophyte development in M. polymorpha (see below) as do the three Arabidopsis homologs in the embryo sac, the female gametophytes of flowering plants (Kasahara et al, 2005; Punwani et al, 2007; Rabiger & Drews, 2013), we named Mapoly0001s0061 FEMALE GAMETOPHYTE MYB (MpFGMYB), following the Marchantia nomenclatural guidelines (Bowman et al, 2016), and hereafter refer to the clade including these genes as the FGMYB subfamily (Fig 2A).

PCR analyses detected MpFGMYB in both male and female genomic DNAs, indicating the autosomal localization of MpFGMYB (Fig 2C). As expected from the transcriptome data (Bowman et al, 2017), MpFGMYB was predominantly expressed in the archegoniophores of female plants and the sporophytes (Fig 2D). Plants harboring a transcriptional MpFGMYB reporter construct (MpFGMYBpro:Citrine-NLS:MpFGMYB3′) exhibited reporter fluorescence in the archegonia (Fig 2E). In a functionally complemented translational reporter line (gMpFGMYBresist-Citrine, see below), Citrine fluorescence was localized in the nuclei of rescued archegonia (Fig 2F). As reported previously (Rabiger & Drews, 2013; Waki et al, 2013), our transcriptional and translational reporters for AtMYB64 were specifically expressed in embryo sacs of A. thaliana (Fig 2G and H; Rabiger & Drews, 2013; Waki et al, 2013).

Loss-of-function MpFGMYB alleles confer a male morphology to female liverworts

Previous studies demonstrated that AtMYB64 and AtMYB119 have critical roles in the development of the embryo sac (Rabiger & Drews, 2013). Similarly, another gene, AtMYB98, is known to be required for the differentiation and function of the synergids, two of the seven cells constituting the embryo sac (Kasahara et al, 2005; Punwani et al, 2007). Thus, the preferential expression of MpFGMYB in the liverwort archegonia suggests that the FGMYB genes have evolutionarily conserved roles in the development of female gametophytes in land plants. To explore this possibility, we generated knockout mutants of MpFGMYB using clustered regularly interspaced short palindromic repeats/CRISPR-associated endonuclease 9 (CRISPR/Cas9) technology (Sugano et al, 2018). We obtained four independent loss-of-function Mpfgmyb lines with insertions or deletions that created premature stop codons in the MYB domain-coding region (Fig 3A and Appendix Fig S1A), of which three (Mpfgmyb-1^ge, Mpfgmyb-2^ge, Mpfgmyb-6^ge) were genetically female, while the other (Mpfgmyb-4^ge) was genetically male (diagnosed by sex chromosome-linked markers; Fig 3B and Appendix Fig S1C).

The Mpfgmyb mutants were morphologically indistinguishable from the wild-type plants during the vegetative growth period (Appendix Fig S2). Genetically male Mpfgmyb mutants (hereafter designated Mpfgmyb [Y]) were also indistinguishable from wild-type males during reproductive growth (Fig 3F, G, K and L). By contrast, genetically female Mpfgmyb mutants (hereafter designated Mpfgmyb [X]) exhibited a striking sex conversion phenotype; antheridiophores developed in place of archegoniophores (Figs 3C and D, and EV2A). Furthermore, the antheridiophores of Mpfgmyb [X] contained antheridia (Fig 3I, compare with 3H, and Fig EV2B). These phenotypes were rescued by introducing a MpFGMYB genomic fragment containing synonymous mutations to resist the remaining CRISPR/Cas9 activity (gMpFGMYBresist) in the mutants (Fig 3E and J, and Appendix Fig S1B), confirming a causal relationship between the female-to-male sex conversion phenotype of Mpfgmyb [X] and the loss of MpFGMYB function.

The sex conversion phenotype of Mpfgmyb [X] could also be rescued by expressing MpFGMYB-Citrine fusion proteins under the same regulatory sequences as those used in the transcriptional reporter lines (gMpFGMYBresist-Citrine; Fig 2F). This line exhibited no Citrine florescence in the apical notch region of vegetative thalli (Fig EV1A and B). After induction of reproductive growth by far-red (FR) light treatment (Ishizaki et al, 2016), MpFGMYB-Citrine proteins accumulate in ventral apical notch regions where archegoniophores will develop (Fig EV1A and C), and later in developing archegoniophores (Fig EV1A and D), consistent with the requirement of MpFGMYB in female sexual differentiation of M. polymorpha.

Click here to expand this figure.

Mpfgmyb mutant females produce sperm with nearly normal morphology but lacking motility

The results presented so far indicate a key role for MpFGMYB in the female sexual differentiation of M. polymorpha. In its absence, male sexual differentiation proceeds as the default program. This female-dominant mode of sex differentiation is consistent with the classical observation that rare diploid gametophytes of M. polymorpha carrying both X and Y chromosomes exhibit a female morphology (Haupt, 1932); however, it was still unclear whether the loss of MpFGMYB function alone was sufficient to generate functional sperm in the absence of the Y chromosome.

To address this question, we performed a histological analysis and found that spermiogenesis proceeds in Mpfgmyb [X] essentially as it does in the wild-type and Mpfgmyb [Y] antheridia (Fig EV2C–J). Moreover, sperm collected from Mpfgmyb [X] antheridia exhibited nuclear condensation and flagella formation (Fig 4A–C). Consistently, two autosomal genes implicated in sperm morphogenesis and known to be specifically expressed in the antheridiophores, PROTAMINE-LIKE (MpRPM) and DYNINE LIGHT CHAIN7 (MpLC7; Higo et al, 2016), were expressed in the antheridiophores of Mpfgmyb [X] (Fig 4D). Furthermore, while the expression of the archegoniophore-specific autosomal genes was suppressed in Mpfgmyb [X], X chromosome-linked genes expressed in vegetative thalli (Bowman et al, 2017) were still expressed in Mpfgmyb [X] antheridia (Fig EV3A). These data indicate that the feminization capacity of MpFGMYB is primarily associated with the sex-specific expression of autosomal genes involved in sexual differentiation, and not with the expression of sex chromosome-linked genes.

Click here to expand this figure.

Click here to expand this figure.

The sperm of Mpfgmyb [X] plants exhibited abnormal morphologies, such as incompletely condensed nuclei and short flagella (Fig 4B, compare with 4A and C). Transmission electron microscopy revealed that most Mpfgmyb [X] flagella had irregular axonemes, lacking the “9 + 2” arrangement of microtubules (Carothers & Kreitner, 1968) typically seen in wild-type sperm (Fig 4E and F). Consistently, sperm produced in Mpfgmyb [X] antheridia were immotile and did not enter wild-type archegonia (Fig EV3B–D). Thus, while the loss of MpFGMYB function resulted in an almost complete female-to-male sex conversion, the formation of functional sperm requires additional factors that are likely encoded by the Y chromosome (Bowman et al, 2017).

Expression of MpFGMYB is suppressed by its cis-acting antisense gene SUF in males

The striking sex conversion phenotype caused by the autosomal Mpfgmyb mutations raised the question as to how MpFGMYB expression is tightly suppressed in males. A close inspection of our RNA sequencing data revealed the male-specific accumulation of antisense lncRNAs derived from the MpFGMYB locus (Fig 5A and B), which we named SUPPRESSOR OF FEMINIZATION (SUF). Real-time RT–PCR analyses revealed that SUF transcripts accumulated in male gametophytes and in sporophytes, while negligible accumulation was detected in female gametophytes. In male gametophytes, antheridiophores accumulated a significantly higher amount of SUF transcripts than thalli (Fig EV4A). Importantly, 5′ and 3′ RACE PCR revealed an invariable 5′ end and a polyadenylation site of SUF transcripts, indicating that SUF constitutes a strictly defined transcription unit of RNA polymerase II (Appendix Fig S3). Strand-specific RT–PCR analyses confirmed SUF transcript accumulation in wild-type males both before and after the induction of sexual reproductive growth by far-red irradiation (Fig EV4B).

Click here to expand this figure.

The mutually exclusive expression patterns of MpFGMYB and SUF in gametophytes suggested a possible regulatory mechanism in which SUF suppresses MpFGMYB expression in the males, as has been reported for other antisense lncRNAs (Xue et al, 2014; Kopp & Mendell, 2018). To test this possibility, we deleted a 1-kb region at the predicted transcription start site (TSS) of SUF in the wild-type males using genome editing, without affecting the MpFGMYB-coding sequence (Fig 5A and Appendix Fig S4). As expected, the resulting male mutants (hereafter designated suf^ge [Y]) lost SUF expression and instead gained MpFGMYB expression after the induction of reproductive growth by far-red irradiation (Fig 5B). Furthermore, suf^ge [Y] plants formed archegoniophores and archegonia that expressed autosomal genes whose expression is female-specific in the wild type (Bowman et al, 2017), the opposite phenotype to that of Mpfgmyb [X] (Figs 5C and D, and EV5). However, archegonia produced in suf^ge [Y] did not produce egg cells, suggesting that this process may require the function of genes present on the X chromosome (Fig 5D). This male-to-female sex conversion phenotype was suppressed by additional loss-of-function mutations in the MpFGMYB-coding region (suf^ge Mpfgmyb^ge [Y]; Figs 5A, E, F, and EV5, and Appendix Fig S4). Furthermore, male plants lacking the entire MpFGMYB/SUF locus (designated complete deletion [Y]) developed normal antheridiophores and antheridia expressing the male-specific genes (Figs 5A, G, H, and EV5, and Appendix Fig S4). These data indicate that the male-to-female sex conversion phenotype of suf [Y] is caused by the misexpression of MpFGMYB. Taken together, our data support the notion that SUF suppresses MpFGMYB expression in males and thereby allows a default male differentiation program to proceed.

Click here to expand this figure.

To determine whether SUF suppresses MpFGMYB expression in cis or in trans, we made a SUF overexpression construct using the strong constitutive MpEF1α promoter (Althoff et al, 2014) and introduced it to suf^ge [Y] in a non-targeted manner. Interestingly, SUF overexpression did not rescue the sex conversion phenotype of suf^ge [Y] (Fig 6A), despite strong accumulation of SUF transcripts (Fig 6B). Furthermore, genetically male plants harboring an MpFGMYB transgene lacking the putative SUF promoter and the TSS (gMpFGMYB-S [Y]; Fig 6C) accumulated MpFGMYB transcripts to the level comparable to or higher than that of wild-type females (Fig 6D) and morphologically converted to females (Fig 6E and F). In contrast, genetically male plants harboring the entire MpFGMYB/SUF locus as a transgene, i.e., containing the putative SUF promoter and the TSS (gMpFGMYB-L; Fig 6C), did not accumulate MpFGMYB transcripts to the level comparable to wild-type females (Fig 6D) and retained male morphologies (Fig 6G and H). These observations indicate a key role of SUF transcription in suppressing MpFGMYB at the same locus in males, and inability of the endogenous SUF locus to suppress unlinked transgenic copies of MpFGMYB. Taken together, these data indicate that SUF-mediated MpFGMYB suppression is locus-specific and hence acts in cis.

Phylogenetic analysis

Amino acid sequences of the R2R3-type MYB proteins in clades 11, 12, and 14–16 (Bowman et al, 2017) were used in the phylogenetic analysis. These sequences were retrieved using the MarpolBase (http://marchantia.info/), Phytozome (https://phytozome.jgi.doe.gov/), and TAIR (https://www.arabidopsis.org/) databases (Rensing et al, 2008; Dubos et al, 2010; Bowman et al, 2017). Regions encompassing the ~110-residue MYB domain were aligned using the MUSCLE program (Edgar, 2004) in AliView v1.18.1 (Larsson, 2014). After manually removing the gaps, a phylogenetic tree was constructed using the maximum-likelihood algorithm based on the Le_Gascuel_2008 model (Felsenstein, 1981; Le & Gascuel, 2008) and evaluated using a bootstrap method (Felsenstein, 1985) with 1000 replicates in MEGA7 (Kumar et al, 2016).

DNA construction

Plasmids used in this study were constructed using gateway cloning system (Ishizaki et al, 2015) or SLiCE method (Motohashi, 2015). Primers used for DNA construction are listed in Table EV2. See Appendix Supplementary Methods for details.

Generation of transgenic Marchantia polymorpha

The genome editing constructs pMpGE010_MpFGMYBge01, pMpGE010_MpFGMYBge02, pMpGE018_SUFge, and pMpGE018_complete-deletion (see Appendix Supplementary Methods for details) were introduced into M. polymorpha as described previously (Ishizaki et al, 2008). Other constructs were introduced essentially as described previously (Kubota et al, 2013). For mutant isolation and sex diagnosis, gRNA-targeted regions of MpFGMYB were amplified from the genomic DNAs prepared from the thalli of T1 plants using the primer pairs ge01-Fw/ge01-Rv and ge02Fw/ge02-Rv for MpFGMYBge01 and MpFGMYBge02, respectively. PCR products were directly sequenced using Bigdye Terminator v3.1 (Thermo Fisher Scientific) and the primers ge01-Fw or ge02-Fw. The genetic sex of the Mpfgmyb^ge lines was diagnosed as described previously (Fujisawa et al, 2001) using the primer sets listed in Table EV2.

Histology and microscopy

Excised tissues were fixed with a formaldehyde/acetic acid (FAA) solution overnight at 4°C, and then dehydrated in an ethanol series and embedded in Technovit 7100 resin (Heraeus Kulzer, Hanau, Germany). Sections of 2 μm thickness were made using a Leica RM2155 microtome (Leica Microsystems, Wetzlar, Germany) and stained with toluidine blue. Image contrast was enhanced equally over the entire area using the Fiji program (Schindelin et al, 2012).

Confocal laser scanning microscopy was carried out using a Nikon C2 confocal laser scanning microscope (Nikon Instech, Tokyo, Japan). Archegonia were stained with 0.4% (v/v) SCRI Renaissance 2200 (Renaissance Chemicals, Selby, UK; Musielak et al, 2015) in 4% (w/v) paraformaldehyde and 1× PBS. Sperm released from antheridiophores were collected by centrifugation and stained with 1 μg/ml DAPI in 4% (w/v) paraformaldehyde and 1× PBS. Stained sperm were dropped on a slide glass and dried up by incubating for a few minutes at room temperature, then observed using a confocal microscope. The sperm attraction assay was performed as described previously (Koi et al, 2016).

For electron microscopy, the excised tissues were transferred to vials containing a fixative solution composed of 4% (v/v) glutaraldehyde and 0.05 M phosphate buffer (pH 7.0), and then vacuum-infiltrated until the specimens sank to the bottom. After fixation for 6 h at room temperature, samples were rinsed with 0.05 M phosphate buffer and post-fixed in 1% (w/v) osmium tetroxide for 3 h at 4°C. The samples were dehydrated in a graded ethanol series and embedded in Spurr's plastic resin using a graded series of propylene oxide and the resin. Ultra-thin sections (80 nm) were prepared with a diamond knife on an ultramicrotome (Ultracut R; Leica Microsystems). Sections were stained sequentially with uranyl acetate and lead citrate, and then observed under a transmission electron microscope (JEM 1010; JEOL, Tokyo, Japan).