Research Article

Transcription factor AP2 controls cnidarian germ cell induction

See allHide authors and affiliations

Science  14 Feb 2020:
Vol. 367, Issue 6479, pp. 757-762
DOI: 10.1126/science.aay6782

Conserved gene specifies germ cell

Germ cells are the exclusive progenitors of gametes. In most studied animals, including humans, germ cells are produced only once during embryogenesis and are not replenished in adult life. DuBuc et al. studied germ cell induction in the clonal cnidarian Hydractinia symbiolongicarpus, an animal that forms germ cells continuously in adult life from stem cells that also generate somatic cells. A single transcription factor is capable of converting the animal's adult stem cells to germ cells. A similar gene also controls germ cell induction in mammalian embryos, but its action there is limited to a single event in early embryogenesis.

Science, this issue p. 757


Clonal animals do not sequester a germ line during embryogenesis. Instead, they have adult stem cells that contribute to somatic tissues or gametes. How germ fate is induced in these animals, and whether this process is related to bilaterian embryonic germline induction, is unknown. We show that transcription factor AP2 (Tfap2), a regulator of mammalian germ lines, acts to commit adult stem cells, known as i-cells, to the germ cell fate in the clonal cnidarian Hydractinia symbiolongicarpus. Tfap2 mutants lacked germ cells and gonads. Transplanted wild-type cells rescued gonad development but not germ cell induction in Tfap2 mutants. Forced expression of Tfap2 in i-cells converted them to germ cells. Therefore, Tfap2 is a regulator of germ cell commitment across germ line–sequestering and germ line–nonsequestering animals.

Segregation of germ cells from somatic fate is an irreversible, once-in-a-lifetime event that is induced during embryonic development by maternal or zygotic factors in many bilaterians (1). The introduced barrier between soma and germ line (known as the Weismann barrier) prohibits somatic cells from contributing to gamete production, and vice versa, thereby preventing the transmission of somatic mutations to future generations. By contrast, clonal animals, such as sponges and some cnidarians, do not sequester a germ line (24). Instead, these animals maintain a population of adult stem cells throughout life that retain the ability to differentiate into somatic cells and into gametes (Fig. 1A). Other animals—such as sea urchins, snails, and annelids—specify their germ cells after embryogenesis, but it is unknown whether this process occurs only once or multiple times, as in clonal animals (5).

Fig. 1 Sexual development in Hydractinia.

(A) Timing of germ cell induction in germ line–sequestering and germ line–nonsequestering animals. (B) Tissue architecture and location of i-cells (pink) and germ cells (green) in Hydractinia feeding polyp and a hypothetical sexual polyp with both sexes. (C) Expression of Piwi1 in feeding and sexual polyps. Solid blue line indicates the body’s epidermal outline. Dashed green line indicates the basement membrane (mesoglea) separating the epidermis and gastrodermis. Piwi1+ cells in the epidermis (i-cells) are encircled in purple. Piwi1+ cells in the gastrodermis are germ cells. Asterisks denote the oral pole. The distribution of i-cells can vary between polyps and extends more orally in sexual polyps compared with feeding polyps.

The molecular mechanisms that induce germ cell commitment are understood in a few germ line–sequestering animals (69), but the genes that induce germ cell fate in clonal species remain unknown. This raises the question of whether the differences in the timing of animal germ cell specification are temporally distinct manifestations of a shared molecular program or have independent evolutionary origins.

We find that a single gene, transcription factor AP2 (Tfap2), is sufficient to induce germ fate when expressed in adult stem cells in the clonal cnidarian Hydractinia. Tfap2 is also required non–cell autonomously for proper gonad development. A homologous gene, Tfap2C, is a major regulator of mammalian germ cell induction, which is consistent with this gene being an ancient regulator of animal germ cells.

Hydractinia as a model for germ cell induction in clonal animals

Hydractinia symbiolongicarpus is a clonal, colonial hydrozoan cnidarian [see (3) for a definition of coloniality]. Adult stem cells in hydrozoans, known as i-cells (10), generate progenitors to somatic lineages and to gametes (11). Commitment to germ cell fate in Hydractinia occurs continuously after reaching sexual maturation in an anatomically defined location (12, 13) (Fig. 1B), making the animal an accessible and attractive model system to study this alternative, continuous mode of germ cell specification. Hydractinia colonies are composed of genetically identical (clonal) modular units called polyps that arise by asexual budding from a single sexually produced individual (fig. S1A). All polyps in a colony are connected by stolonal tissue, allowing i-cell migration throughout the colony. A newly formed colony consists exclusively of nonreproductive feeding polyps. Sexual polyps, which are morphologically distinct (Fig. 1B and fig. S1, B and C), appear approximately 2 months after metamorphosis. The body columns of both polyp types are composed of outer epidermal and inner gastrodermal tissues (Fig. 1B). The animal’s stem cells (the i-cells) are located exclusively in the interstitial spaces between epithelial cells in the epidermis and are marked by germline multipotency program (GMP) gene expression (14); this includes, e.g., Piwi1 (Fig. 1C and figs. S1 and S2), Vasa, and Pl10 (15). In sexual polyps, i-cells can acquire germ cell fate and become gamete progenitors (Fig. 1C and fig. S1C). Early germ cells concentrate in a narrow tissue stripe at the neck of the sexual polyp, referred to as the germinal zone (12, 13), from which they migrate into the sporosacs and mature. Germ cells express GMP genes, similar to i-cells from which they were derived, making them the only GMP+ gastrodermal cells in Hydractinia colonies and, therefore, easy to recognize (Fig. 1, B and C). Hydractinia is gonochoristic, and the sexual polyp is the exclusive site of gametogenesis, making it functionally equivalent to gonads in bilaterians. Early stages of male and female sexual polyp development are morphologically indistinguishable (fig. S1C).

Tfap2 is expressed in male and female germ cells

To identify candidate regulators of germ cell commitment in Hydractinia, we compared gene expression between feeding and sexual polyps. A previous study (16) compared the transcriptomes of different Hydractinia polyp types using pooled male and female samples. Analyzing these data, we found that some genes reported to be up-regulated in sexual polyps are primarily female specific (fig. S3) and are probably involved in oogenesis rather than in the earlier-occurring germ cell induction, which is likely shared by males and females (17, 18). Therefore, we repeated this experiment but generated separate male and female RNA sequencing (RNA-seq) libraries from the heads and bodies of sexual and feeding polyps. This enabled us to identify genes that are commonly up-regulated in both sexes during sexual development and allowed us to test whether they are differentially expressed between the polyps’ oral and aboral regions (Fig. 2, A and B, and table S1).

Fig. 2 Sexually up-regulated genes in Hydractinia.

(A) Schematic of tissue sampling strategy. (B) Differentially expressed genes in different tissue compartments. (C) Live image of a Tfap2 transgenic female reporter animal expressing GFP in the germinal zone (green arrowhead). (D) mRNA fluorescence double in situ hybridization of Piwi1 and Tfap2 in a male sexual polyp. Arrowhead points to double-positive cell in the epidermis. Asterisks denote the oral pole. In the schematic, pink cells are Piwi1 positive, yellow cells are Piwi1 and Tfap2 positive, and green cells are Tfap2 positive.

Tfap2 emerged as a potential candidate gene for germ cell induction, given the known role of one of its homologs (Tfap2C) in mammalian germ cell specification (8, 1923) and given that it is not sex-specific in Hydractinia (Fig. 2, B to D, and table S1). Tfap2 genes are found across the Metazoa, including the four nonbilaterian phyla: Ctenophora, Porifera, Placozoa, and Cnidaria (fig. S4).

The Hydractinia genome encodes two Tfap2-like genes (Tfap2a and Tfap2b), with phylogenetic analyses suggesting that they are paralogs (fig. S4). Tfap2b mRNA could not be detected by either reverse transcription polymerase chain reaction (RT-PCR) or in situ hybridization (fig. S4), making it a likely pseudogene. On the other hand, Tfap2a (henceforth Tfap2) was exclusively expressed in the germinal zone of female and male sexual polyps (Fig. 2, C and D, and fig. S4). Given that Piwi1 marks i-cells (in the epidermis) and germ cells (in the gastrodermis; Fig. 1C), we performed double-fluorescence in situ hybridization in sexual polyps to detect Piwi1- and Tfap2-expressing cells. We found double-positive cells in both the epidermis and the gastrodermis, with the latter containing the majority (Fig. 2D). Some Tfap2+ cells in the gastrodermis were Piwi1. The function of these cells is currently unknown.

We developed Tfap2 transgenic reporter animals that enabled us to study the localization of Tfap2 in vivo (Fig. 2C, fig. S5, and movie S1). The expression pattern observed in the transgenic reporter animals and the in situ localization of Tfap2 mRNA are consistent with Tfap2 being expressed not only in recently induced germ cells but also in early gametogonia and in Piwi1 cells whose function remains unknown. Notably, Tfap2 is down-regulated in late gametogonia and gametocytes, as well as in gametes.

To gain insight into the genes acting downstream of induction to activate the germ cell transcriptional program, we compared the transcriptomes of isolated germ cells with those of their i-cell progenitors and with somatic cells. For this, we dissociated sexual polyps from male and female Tfap2 reporter animals and feeding polyps from a Piwi1 reporter animal (15). We established a fluorescence-activated cell sorting (FACS) protocol to sort GFP+ cells at high purity (figs. S6 and S7). We also collected GFPlow and GFP cells from the Piwi1 reporter, representing all somatic lineages (fig. S6C). Transcriptomic analysis of these cells revealed up-regulation of conserved germ cell genes in Tfap2+ cells as compared with i-cells (fig. S6, D and E). However, many germ cell genes, such as Piwi1/2, Nanos1/2, and Pl10, were expressed in both cellular fractions (fig. S6D and dataset S1). This reflects the dual competence of i-cells to contribute to both somatic cells and germ cells. The long half-life of GFP resulted in the inclusion of not only recently induced germ cells but also of gametocytes that no longer expressed Tfap2. This was made evident by the up-regulation of late female and male gametogenesis genes and meiosis genes (fig. S6E and dataset S1). These results are in line with those of previous studies showing that the metazoan germ cell transcriptional program downstream of induction is partly conserved across clades (24).

Tfap2 is essential for germ cell commitment and gonad development

Next, we performed CRISPR-Cas9–mediated mutagenesis experiments to study the role of Tfap2 in sexual development (2528). Two single guide RNAs (sgRNAs) were designed to target the 5′ and 3′ ends of the predicted DNA binding domain of the Tfap2 gene, respectively (fig. S8); these were then injected into zygotes, together with recombinant Cas9. Injected embryos were allowed to develop into larvae, metamorphose, and grow to ages at which sexual maturity is normally reached. They were then genotyped by PCR and sequencing to check for Tfap2 mutations. Several types of mutations were identified in these injected animals, including small or large deletions and insertions, frameshifts, and base substitutions (Fig. 3A and fig. S8). The mutants displayed various defects in sexual development that included too few (but otherwise normal) sexual polyps, deformed sexual polyps, and rudimentary sexual polyps that never matured and contained no germ cells (Fig. 3, C to G, and fig. S8). G0 mutants were mosaics with multiple alleles, including wild-type alleles in several cases. Crossing a G0 mosaic mutant with a Tfap2 wild-type animal, we found that G1 heterozygote mutant i-cells (i.e., Tfap2+/−) could still commit to germ fate, though less effectively (Fig. 3B), generating fertile, mutant gametes. This allowed us to breed these animals to homozygosity (i.e., Tfap2−/−; Fig. 3, C, F, and G). Tfap2−/− animals were sterile, with no detectable germ cells and only rudimentary sexual polyps (Fig. 3, C, F, and G), which implies that Tfap2 is required for both germ cell induction and gonad development. The mutants had normal distributions of i-cells, and their growth and regenerative ability was not compromised (fig. S9).

Fig. 3 Breeding strategy for generating Tfap2−/− knockout animals.

(A) Genomic structure of wild-type and mutant alleles of Tfap2. (B) G1 generation that includes homozygote Tfap2 wild-type and heterozygote mutant animals; the latter produced fewer gametes. All animals shown also carry a Piwi1 reporter transgene inherited from their Tfap2 wild-type father. (C to G) The G2 generation resulting from breeding G1 siblings. (C) Overview of Tfap2−/− homozygote mutant. Only rudimentary sexual polyps are present (arrowheads); the colony appears otherwise normal. (D and E) Tfap2 wild-type rudimentary male and female sexual polyps. Early gastrodermal germ cells express GFP, driven by the Piwi1 reporter transgene. (F) Rudimentary sexual polyp of a Tfap2−/− mutant. This animal also carries a Piwi1 reporter transgene but has no germ cells. (G) Close-up of the same polyp in (F), showing GFP+ epidermal i-cells.

Tfap2 acts non–cell autonomously to induce sexual polyp development

To identify a possible non–cell autonomous role for Tfap2 in sexual development, we transplanted cells from an animal with a wild-type Tfap2 gene into a sterile Tfap2 mutant (Fig. 4A). To facilitate the tracking of Tfap2 wild-type cells in the Tfap2 mutant’s tissues, we generated a transgenic fluorescent reporter animal as a cell donor that carried two reporter transgenes: a Piwi1 reporter that expressed GFP in i-cells and germ cells (15) and a β-tubulin reporter that expressed mScarlet in all other cell types, except for i-cells (fig. S10). Hence, all cells in the donor animal were fluorescent (Fig. 4B and fig. S10) and, therefore, visible after transplantation in the background of the nonfluorescent and sterile Tfap2 mutant that lacked germ cells and mature gonads (Fig. 4 and movie S2). Donor animals were fertile (Fig. 4E), having a wild-type Tfap2 genotype, and were genetically histocompatible (29) with the mutant.

Fig. 4 Transplantation of wild-type, allogeneic cells into a Tfap2 mutant.

(A) Experimental setup. Dashed red line denotes the interface between the two grafted animal colonies. (B) Female Tfap2 wild-type, fluorescent donor feeding polyp. (C) Recipient mutant colony, lacking sexual polyps. (D) The interface between the grafted donor and recipient colonies (dashed red line), viewed from above. Donor-derived cells are visible in the recipient’s tissues. They are more numerous closer to the interface. (E) Fluorescent, Tfap2 wild-type sexual polyp of the donor animal. Oocytes are encircled by a dashed white line. (F) Immature chimeric sexual polyp composed of donor (fluorescent, Tfap2 wild type) and recipient (nonfluorescent, Tfap2 mutant) cells. (G) Mature chimeric sexual polyp. Oocytes encircled by dashed line are exclusively donor-derived. Animals were pictured live and the red and the green channels representing the β-tubulin::mScarlet and Piwi1::GFP reporter transgenes were merged and false-colored green in (B) and (D) to (G) for simplicity. The blue color in (E) to (G) represents DNA. All 10 grafts that were performed resulted in induction of sexual polyps in the mutant.

The grafting procedure (Fig. 4A) allowed i-cells and progeny to migrate between the partners, generating chimeras whose cellular origin could be directly observed in vivo by fluorescence microscopy (30) (Fig. 4, D to G, and movie S2). We found that cells from the wild-type animal, which had migrated into the mutant, induced development of sexual polyps that consisted somatically of mutant and wild-type cells (Fig. 4F). However, the gametes produced by the chimeric sexual polyps were exclusively fluorescent and, thus, donor derived; no mutant-derived gametes were obtained (Fig. 4G). Given that nonchimeric mutant animals only produced rudimentary sexual polyps, we conclude that Tfap2 expressed in donor-derived cells acted non–cell autonomously to promote sexual polyp development in the mutant but could not induce mutant i-cells to germ fate. In bilaterians, germ cells are necessary for proper gonad development in some species (31, 32), and our results are consistent with a previously described phenomenon in animals.

Tfap2 acts cell autonomously to induce germ fate in i-cells

To investigate a cell-autonomous role of Tfap2, we used a random-integration transgenesis approach to ectopically express Tfap2 in three different cellular contexts using three transgenic constructs (fig. S10). This generated mosaic transgenic animals expressing Tfap2-GFP in different cell types. First, we used the Wnt3 promoter to drive Tfap2-GFP expression in the oral region, where Wnt3 is normally expressed (3335) (fig. S10A). As i-cells are normally not present in the oral pole (15), we expected to observe the consequences of Tfap2 expression in differentiated head cells. However, Wnt3 promoter-induced Tfap2 expression resulted in phenotype-free animals (Fig. 5A).

Fig. 5 Ectopic expression of Tfap2 in i-cells induces germ fate.

(A) Wnt3 promoter-driven Tfap2-GFP. Transgene expression is restricted to the oral end and causes no visible phenotype. (B) β-tubulin promoter-driven Tfap2-GFP. Transgene is expressed in somatic cells and causes no visible phenotype. (C to K) Piwi1 promoter-driven Tfap2-GFP. Transgene is expressed only in i-cells, transforming them to germ cells. (C) Ectopic early oocyte in the gastrodermis of a 48-hour-old larva, identified by morphology. (D) Sperm progenitors in the gastrodermis of a 48-hour-old larva, identified by H2B3/4-expression. (E and F) Ectopic oocytes in the gastrodermis of a mosaic transgenic feeding polyp expressing Tfap2-GFP under the Piwi1 promoter, where the transgene had been suppressed for 2 weeks by a shRNA. (G to I) Double mRNA fluorescence in situ hybridization showing colocalization of GFP and Cel in a mosaic transgenic feeding polyp treated as in (E). (G) GFP mRNA. (H) Cel mRNA. (I) Merge. (J) Schematic illustrating the localization of (G) to (I) in the polyp. (K) Tfap2 expression in i-cells converts them to germ cells.

Next, we drove Tfap2-GFP expression by the β-tubulin promoter that is active in all differentiated cells but not in i-cells (fig. S10, B and C). This approach also resulted in no visible phenotype (Fig. 5B), suggesting that Tfap2 can induce neither germ cells nor gonads when expressed in somatic cells.

Finally, we expressed Tfap2-GFP under the Piwi1 promoter to restrict transgene expression to i-cells (Fig. 5, C to J, and fig. S10, B and C). This experiment resulted in large GFP+ cells that morphologically resembled early stage oocytes in mosaic transgenic embryos that were probably females (Fig. 5C). Other embryos (probably males) displayed cells that expressed H2B3/4, a spermatogenesis marker (36) (Fig. 5D and fig. S11). Normally, germ cells appear 2 to 3 months after metamorphosis. However, ectopic germ cells in embryos never matured and vanished after metamorphosis. This suggests that, whereas Tfap2 was effective in inducing germ fate in embryonic Piwi1+ cells, the larval tissue microenvironment could not support gametogenesis downstream of germ cell induction.

We hypothesized that ectopic oocytes would develop to a later stage if Tfap2 expression commenced only after metamorphosis. In the absence of a conditional expression system in Hydractinia, we focused on inhibiting transgene expression until after metamorphosis. For this, we coinjected zygotes with the Piwi1::Tfap2-P2A-GFP ectopic expression construct alongside a short hairpin RNA (shRNA) (37), targeting the GFP sequence in the transgene’s mRNA. Injected embryos and larvae remained GFP-free, consistent with effective transgene repression by the shRNA (fig. S12). The suppressive effect of the shRNA dissipated ~1 week after metamorphosis, and transgenic feeding polyps developed ectopic GFP+ oocytes in the gastrodermis that appeared morphologically more mature than ectopic oocytes in larvae (Fig. 5, E and F). Furthermore, they expressed Cnidarian egg lectin (Cel)—an exclusive early oogenesis marker (38)—in GFP+ cells (Fig. 5, G to J), which was undetectable in larval ectopic oocytes. Taken together, our results show that Tfap2 acts cell autonomously and is essential and sufficient to induce germ cell fate in Piwi1+ i-cells but not in differentiated cells (Fig. 5K). Tfap2 acts non–cell autonomously downstream of germ cell induction to drive sexual polyp maturation.

The evolution of bilaterian sequestered germ lines

Tfap2 is a critical regulator of germ cell induction in Hydractinia, an animal that does not sequester a germ line, as well as in germ line–sequestering animals, such as mice (8) and humans (39). Drosophila and C. elegans germ lines are specified maternally, a mechanism that is thought to be evolutionarily derived (1). In mammals, AP2γ (encoded by Tfap2C) acts in concert with other transcription factors, such as Blimp1, Prdm14, PAX5, and SOX17, to induce germ cell fate in epiblast cells in a species-specific combination (7, 8, 40). Partners of Tfap2 in Hydractinia are yet unknown, but, unlike in mammals, its expression in i-cells—which are similar to epiblast cells in being somatic and germ cell–competent—is sufficient to induce germ cell fate. It has been suggested that a nonsequestered germ line is an ancestral trait in metazoans (4). Under this hypothesis, a key event in the evolution of bilaterian sequestered germ lines would have been the redeployment of the hypothesized ancestral adult germ cell induction program, which is still present in extant Hydractinia, during embryogenesis.

A sequestered germ line prevents transmission of somatic mutations to future generations and was also proposed to help select for mitochondrial quality in complex bilaterians with high mutation rates (4). Why, then, would clonal animals not sequester a germ line? We suggest that a sequestered germ line in clonal animals would be detrimental, exposing them to the risk of stochastically generating new clonal individuals that lack germ cells. Therefore, maintaining uncommitted cells after embryogenesis, which can provide progenitors to any lineage of somatic cells or germ cells, grants full developmental potential to new clonemates. This may represent the selective pressure that prevented the evolution of a sequestered germ line in clonal animals.


Materials and Methods

Figs. S1 to S12

Table S1

References (4163)

Movies S1 and S2

Datasets S1 and S2

References and Notes

Acknowledgments: We thank our laboratory members for lively discussions, our colleagues C. Morrison and G. Schlosser for comments on the manuscript, and the NIH Intramural Sequencing Center (NISC) for generating the sequence data. All flow cytometry and imaging cytometry analyses were performed in the Flow Cytometry Core Facility at NUI Galway. Funding: U.F. is a Wellcome Trust Investigator in Science (grant no. 210722/Z/18/Z, co-funded by the SFI-HRB-Wellcome Biomedical Research Partnership). This work was also funded by a Science Foundation Ireland Investigator Award to U.F. (grant no. 11/PI/1020); by CURAM, SFI Centre for Research in Medical Devices (to U.F.); and by the Intramural Research Program of the National Human Genome Research Institute, National Institutes of Health to A.D.B. (ZIA HG000140). T.Q.D. was an EMBO Long-Term Fellow (grant no. ALTF 68-2016). S.G.G. was a Marie Curie Incoming International Fellow (project 623748) and was also supported by a Science Foundation Ireland SIRG award (grant no. 13/SIRG/2125). F. is a Hardiman Scholar and is also supported by Thomas Crawford Hayes Research Grant. Funding in support of imaging cytometry was received from Science Foundation Ireland under research infrastructure grant no. 16/RI/3760 and from the European Regional Development Fund. Author contributions: T.Q.D. and U.F. conceptualized this project. T.Q.D. collected all RNA samples, generated stable transgenic animals, created CRISPR-Cas9 mutants, conducted short hairpin experiments, and performed all microinjections, IF, and in situ hybridization experiments. T.B. and T.Q.D. performed mutant screens. E.C., S.H., and T.Q.D. performed the FACS experiments. C.E.S., S.N.B., P.G., S.G.G., and A.D.B. analyzed all RNA-seq data and performed the computational analysis. E.T.M. generated the Piwi1 antibody. F. designed and tested the shRNAs. J.M.G. cloned the β-tubulin regulatory regions. T.Q.D. and U.F. wrote the paper. Competing interests: The authors declare no competing interests. Data and materials availability: The raw reads utilized to generate the tissue- and cell-specific differential expression analyses (table S1 and dataset S1) are available through the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) at Tissue-specific reads from Hydractinia echinata are available as accession numbers SRR9332370 to SRR9332387. Bulk cell reads from H. symbiolongicarpus are available under the accession numbers SRR9331388 to SRR9331403. Detailed descriptions of each dataset can be found in the SRA Data tab of table S1 and dataset S1. Transcriptomes generated for this manuscript and draft genomes for both species are available for download at the Hydractinia Genome Portal (

Stay Connected to Science

Navigate This Article