Transcriptional control of human gametogenesis.

The maintenance of cell identity in FGCs requires the repression of somatic lineages in concert with the activation of germ cell programs ( Figs 2 and 4 ). Of note, human FGCs are defined by a unique transcriptional network that comprises germ cell-specific genes together with somatic lineage specifiers and pluripotency genes. How these transcription factors, which are master regulators for various cell types, function differently from their canonical roles in driving germ cell fate is an intriguing and fundamental question in the field of human germ cell developmental genetics. One hypothesis is that transcription factors work in different protein complexes to perform cell-type-specific roles. In support of this, Fang et al. (2018) observed that OCT4 switches partners from SOX2 in hPSCs to PAX5 and PRDM1 proteins in human FGCs ( Fang et al. , 2018 ). While continuing to repress differentiation toward ectoderm, importantly in human FGCs, OCT4 shifts its binding from pluripotency-related genes to germline-specific genes to activate germ cell fate, coincident with the switch in activity of the respective OCT4 complexes. Proposed model for intrinsic transcription forces that drive human germline and somatic differentiation. Solid ovals represent gene expression programs that are activated, and dotted ovals represent gene expression programs that are silenced. An alternate but not necessarily mutually exclusive hypothesis is that the delicate balance of different transcription factors defines germ cells. Lineage specifiers that belong to the three germ layers and trophectoderm (SOX17, EOMES, PAX5 and GATAs), as well as pluripotency proteins (OCT4, NANOG), are all expressed in human FGCs and function as drivers of the germ cell lineage, as demonstrated by diverse functional studies. Thus, these master regulators of differentiation and pluripotency must be regulated to modulate their canonical functions ( Fig. 4 ). For example, while the SOX17 gene encodes a classical endoderm specifier and plays a critical role in normal hPGC specification, over-expression of SOX17 beyond a PGC-competent window favors the expression of endoderm genes rather than direction toward a germ cell fate ( Kobayashi et al. , 2017 ). In response to hPGC specification, the dosage and actions of these transcription factors must be balanced so that cell identity extends beyond pluripotency but is not co-opted toward any specific somatic lineages. It is likely that other germ cell determinants are activated to inhibit further somatic lineage differentiation and reinforce commitment of cells to the germline. For example, as a transcriptional repressor, PRDM1 activity in hPGCs represses expression of somatic lineage genes ( Irie et al. , 2015 ; Sasaki et al. , 2015 ). In support of this, a recent study demonstrated that high-dosage overexpression of SOX17 in hPSCs leads to aberrant expression of endoderm markers, which could be rescued by simultaneously providing a comparable dose of PRDM1 protein ( Kobayashi et al. , 2017 ). Given the importance of PRDM1 in hPGCs, its activation may be safeguarded by multiple transcription factors to assure its appropriate expression to maintain germ cell identity.

A number of reviews have contrasted, analyzed and discussed gametogenesis across species including humans. Two excellent recent examples are the reviews of Li et al. (2020) and that of Saitou and Hayashi (2021) . In these reviews, in vivo and in vitro development are compared and contrasted, and differences between the processes across species are also highlighted. Here, we briefly provide an overview of human gametogenesis that distills details in specification, migration, sex determination and male- and female-specific development and then we focus on transcription factors and their functions and associated pathologies. Recent reviews and this work largely concur on major aspects while providing different content; this is indicative of the field of gametogenesis in vivo and in vitro maturing toward a common set of foundational developmental and genetic principles. In vivo , hPGCs are first identified in the posterior region of the yolk sac, and begin to migrate to the genital ridge about 4 weeks post-conception ( McKay et al. , 1953 ; Motta et al. , 1997 ; Culty, 2009 ; Leitch et al. , 2013 ). Data from studies in the mouse indicate that signaling via bone morphogenetic proteins (BMPs) released from the extraembryonic ectoderm and proximal endoderm, including BMP4, BMP8b and BMP2, is essential for PGC specification ( Lawson et al. , 1999 ; Ohinata et al. , 2009 ). Analysis of human fetal ovary also demonstrates that the expression of BMP2 and BMP4 may regulate the survival and migration of hPGCs ( Childs et al. , 2010 ). In addition to BMPs, WNT (Wingless-related integration site) signaling, which is an evolutionarily conserved pathway in embryonic development, is required to activate the expression of many transcription factors that are indispensable in the specification of PGCs ( Aramaki et al. , 2013 ). Finally, it is notable that a recent study of non-human primates demonstrates that cynomolgus monkey PGCs (cyPGCs) originate from the dorsal amnion instead of the posterior epiblast as seen in murine development ( Sasaki et al. , 2016 ), suggesting the potential for distinct environmental cues for primate PGC specification versus other mammals. Following specification, hPGCs gradually proliferate as they also gain motility and initiate migration at 4–5 weeks ( Pereda et al. , 2006 ; Mamsen et al. , 2012 ; Gomes Fernandes et al. , 2018 ). Despite significant differences in terms of migration rates and distances traveled, PGC migration in all species has conserved elements ( Pereda et al. , 1998 ; Richardson and Lehmann, 2010 ; Grimaldi and Raz, 2020 ) including: first, the acquisition of motility/initiation of migration; second, directed migration; and third, termination of migration at the developing gonad. In terms of acquisition of motility and initiation of migration, once PGCs are specified, specific molecular pathways direct the detachment from neighboring cells and the extracellular matrix as a prerequisite to motility. For example, studies in different organisms indicate that downregulation of the cell–cell adhesion protein, E-cadherin, initiates the migration process of PGCs. Directed migration is regulated by attractive and repulsive cues. Following initiation of migration, PGCs require cues for directionality. PGCs from different organisms migrate along different paths while interacting with diverse cell types and the extracellular matrix. Immunohistochemistry and electron microscopy studies suggest that hPGCs preferentially migrate along autonomic nerve fibers and Schwann cells from the dorsal hind gut mesentery to the developing gonad ( Mollgard et al. , 2010 ; Mamsen et al. , 2012 ). The migration is accompanied by a wave of chemical cues expressed by the surrounding somatic cells. Appropriate migration and survival of PGCs are instructed by both an intrinsic transcriptional program and external guidance cues. Stem cell factor, lipids and c-KIT (receptor tyrosine kinase) as well as G protein-coupled receptor signaling are implicated as attractive guidance cues for PGC migration to the genital ridge ( Molyneaux et al. , 2003 ; Hoyer et al. , 2005 ). Similar to mouse PGCs ( Hayashi et al. , 2007 ; Saitou and Yamaji, 2012 ), migratory hPGCs maintain a gene expression program characteristic of pluripotency, with sustained expression of pluripotency factors such as OCT4 and NANOG . hPGCs also maintain a broad developmental potential, retaining the capacity for both germ cell and somatic cell differentiation. Concerning termination of migration at the developing gonad, although there is no evidence for sex-specific differences during PGC migration, once PGCs arrive at the target gonad, motility is lost as the PGCs acquire sex-specific properties to contribute to gonad formation with somatic cells. Studies in mouse PGC development indicate that a change in cell adhesion may play a role in reduced mobility ( Bendel-Stenzel et al. , 2000 ; Di Carlo and De Felici, 2000 ). However, the set of proteins responsible and their precise modes of action have yet to be identified and characterized in full. PGCs that fail to exit the nerve branches at the gonadal site may continue to migrate to other organs, such as the abdomen, adrenal glands, heart, lungs, and central nervous system. If they are not eliminated by apoptosis, these stray germ cells may give rise to germ cell tumors ( Mamsen et al. , 2012 ). Upon arriving at the genital ridge, PGCs interact with somatic cells and form the bipotential gonads. Sex determination of the gonad is a process by which the bipotential gonads differentiate into either testes or ovaries at gestational weeks 6–7 onward ( Baker, 1963 ; Jorgensen et al. , 2012 ). Interestingly, sex determination of germ cells is dependent on external signals from the somatic environment rather than solely on the sex chromosome composition (XX or XY). Studies in mouse models confirmed this mechanism by demonstrating that XY germ cells can develop into oocytes in female chimeric embryos and XX germ cells can develop into prospermatogonia in male chimeric embryos ( Ford et al. , 1975 ; Burgoyne et al. , 1988 ; Palmer and Burgoyne, 1991 ; Patek et al. , 1991 ). In the XX testis, the XX germ cells enter spermatogenesis and become prospermatogonia; however, they are eliminated before differentiation into spermatogonia. In the XY ovary, the XY germ cells enter meiosis and continue to differentiate as the primary oocytes; however, their fertility depends on species, genetic background and causes of sex reversal ( Taketo-Hosotani et al. , 1989 ; Heard and Turner, 2011 ). The developmental fate of the bipotential gonad is dependent on a delicate balance of pro-testis and pro-ovary pathways in the supporting somatic cell lineage. To initiate male differentiation to the testis, the pro-testis pathway, characterized by the SRY (Sex-determining Region on the Y chromosome)-SOX9 (SRY-related HMG bOX-containing gene 9)-FGF9 (Fibroblast Growth Factor 9) gene network, needs to be activated to induce differentiation of the somatic cells into the male-specific Sertoli cells, and simultaneous repression of the ovarian pathway. In females, continuous activation of pro-ovary pathways, characterized by the RSPO1 (R-Spondin1)-WNT4-β-catenin signaling pathway, promotes differentiation of somatic cells to granulosa cells, leading to ovarian development. Once the somatic sex of the gonad is determined, sexual development of the rest of the embryo can progress. In males, the testes produce testosterone and anti-Müllerian hormone (AMH) to induce the formation of other organs in the male reproductive system and promote degeneration of the Müllerian duct. In females, the ovaries produce estrogen, which triggers development of the uterus, oviducts and cervix from the Müllerian duct. In response to somatic sex-determining cues, germ cells in female embryos initiate oogenesis and enter meiosis before birth. In contrast, male germ cells enter a mitotic arrest and do not enter meiosis until after birth. Most of what we know of female germ cell development in vivo derives from studies in mice and rats with similarities observed in human fetal development, as well. After arriving at the genital ridge, female germ cells continue to proliferate through mitotic divisions with incomplete cytokinesis, to form oogonia cysts. In response to retinoic acid signals, oogonia cells then start meiosis and differentiate into primary oocytes ( Bowles et al. , 2006 ; Koubova et al. , 2006 ). Meiosis initiates with prophase I stage, which is classically divided into five distinctive sub-stages based on the conformation of chromosomes: leptotene (prophase begins, chromosome start to condense), zygotene (synapsis begins), pachytene (crossing over), diplotene (synapsis ends) and diakinesis (prophase ends, nuclear membrane disintegrates). Primary oocytes arrest at the dictyate stage and become quiescent until sexual maturation. Around this time, the germ cell cyst breaks down, and the majority of oocytes that are not surrounded by somatic cells succumb to apoptosis and/or autophagy ( Goldsmith, 1990 ; Pepling and Spradling, 2001 ; Escobar et al. , 2010 ). Surviving oocytes are assembled into primordial follicles with pre-granulosa cells; the primordial follicles are the reservoir of germ cells for the entire female reproductive life. At birth in humans, there are approximately 400 000 primordial follicles, and this number gradually declines with age ( Block, 1953 ; Forabosco et al. , 1991 ; Gougeon, 1996 ). During a woman’s reproductive life, approximately 400 follicles will undergo ovulation. With the onset of puberty, oocyte meiotic maturation is initiated by hormone stimulation, particularly by LH signaling molecules ( Mehlmann, 2005 ). LH releases oocytes from meiotic prophase arrest and induces them to complete the first meiotic division and produce the first polar body. The second meiotic division begins immediately but pauses at metaphase, where the oocyte remains arrested until fertilization. The second meiotic division is triggered by the penetration of the sperm, and the second polar body will be formed at the same time. Upon arriving at the genital ridge of a male embryo, male fetal germ cells (FGCs) will not enter meiosis en masse . Instead, at this stage in normal testis development, somatic cells and FGCs begin to differentiate into seminiferous tubules with germs cells in the center and Sertoli cells at the periphery ( Wilhelm et al. , 2007 ). Somatic cells will provide the niche for developing FGCs. Spermatogenesis starts in early puberty, and it is a continuous cellular differentiation process that can be classified into four distinctive stages: Mitotic proliferation and maturation to generate spermatogonia (SPG). Spermatogonia are composed of three subtypes of cells: Type A (dark) cells (spermatogonia stem cells (SSCs) that do not undergo active mitosis), Type A (pale) cells (SSC that undergo active mitosis and divide to produce Type B cells), and Type B cells, which undergo growth and become spermatocytes. Two rounds of meiotic division to form haploid spermatocytes (SPC). Morphological transformation of spherical SPCs to elongated spermatids (SPT), a process also referred to as spermiogenesis. Final maturation of SPT to spermatozoa and release into the lumen of the seminiferous tubules, with the sperm passing through the epididymis to undergo final maturation ( Clermont, 1972 ). These four processes are interdependent and regulated by the somatic niche of the seminiferous tubules that is composed of three major cell types: Sertoli, peritubular and Leydig cells. It is estimated that the entire process of human spermatogenesis takes about 74 days ( Heller and Clermont, 1964 ; Amann, 2008 ).

Despite advances in our knowledge of transcription factor function in PGC specification, in vitro -derived hPGCLCs do not progress further down the germ cell lineage efficiently and do not readily enter or complete meiosis to produce functional germ cells. Accordingly, gene expression analysis of in vivo hPGCs from developing human embryos has revealed clear differences with hPGCLCs ( Table III ) (Gkountela et al. , 2013 , 2015 ; Tang et al. , 2015 ; Chen et al. , 2018 ; Sybirna et al. , 2020 ). Most notably, late-stage hPGC markers, such as DAZL (deleted in azoospermia), VASA / DDX4 (DEAD-box helicase 4), and PIWIL1 (Piwi like RNA-mediated gene silencing 1), are not activated in hPGCLC models ( Irie et al. , 2015 ) suggesting a lack of activation of necessary transcription factors to induce the transcriptional program of later germ cell stages in vitro . Indeed, a time-course analysis of early- versus late-gestation cyPGCs found that the gene expression signature of hPGCLCs is more similar to early-stage PGCs than later-gestation PGCs, which have been most commonly profiled from human samples ( Sasaki et al. , 2016 ). Several transcription factors enriched in expression in ‘late’ cyPGCs, such as RNF17 (ring finger protein 17) and KRBOX1 (KRAB Box domain containing 1), lack expression in hPGCLCs, and are known to function in late-stage germ cell development. Transcription factors reported to be expressed in bona fide hPGCs and in vitro derived hPGCLCs. PiRNA, Piwi-interacting RNA. The wealth of expression data now available from hPGCs in vivo , as well as different models of hPGCLC generation, might enable the unbiased identification of new transcription factors that may drive hPGC specification beyond the current state achievable in vitro ( Gkountela et al. , 2015 ; Chen et al. , 2018 ). We have identified transcription factors whose expression is induced in hPGCs or hPGCLCs and found several notable transcription factor classes present in vivo that are not activated in vitro ( Table III ). For example, DMRT-family transcription factors are known to play a role in sex determination in many organisms, and both DMRT1 and DMRTC2 are found in hPGCs ( Guo et al. , 2015 ), but are not expressed in hPGCLCs. Consistent with their well-known role in embryonic development, many homeobox-containing genes, including both classic Hox cluster transcription factors as well as other homeodomain-containing transcription factors, such as ALX4 (ALX homeobox 4), EMX1/2 (empty spiracles homeobox 1/2), ESX1 (ESX homeobox 1), SIX1/2 (Sineoculis homeobox homolog 1/2) and LHX2/8 (LIM-homeodomain protein 2/8), are present in hPGCs but not hPGCLCs. This enrichment of homeodomain transcription factors is somewhat surprising given that repression of homeobox genes is a well-known feature of developing PGCs, so this may reflect contamination from somatic cells. Finally, some members of other transcription factor classes, such as PAX and SOX-domain transcription factors, are active in in viv o but not in vitro models of PGCs. Different gene expression profiles could be linked to different developmental timing (or different development trajectories between in vitro differentiation or in vivo development) or aberrant gene expression. It is likely that future efforts to derive germ cell-like cells in vitro will require the expression or activation of transcriptional programs of one or more of these transcription factors, either by exogenous gene introduction or identification of upstream signaling pathways, which can be activated by the addition of extracellular factors. In addition to identification of stage-specific gene expression profiles, it is equally important to compare the epigenetic status of in vitro derived germ cells with that of bona fide germ cells. Global erasure of DNA methylation, with the exception of some repetitive elements, is the hallmark of in vivo hPGC development ( Leitch et al. , 2013 ; Irie et al. , 2015 ). Several transcription factors are involved in the epigenetic modeling of hPGCs. PRDM1 and SOX17 function in maintaining the epigenetic program of 5-methylcytosine (5mC) erasure from week 4 to week 9. Partial erasure of 5mC and enrichment of 5hmC is also observed in the in vitro differentiation system. In hPGCLCs, loss of PRDM1 inhibits the initiation of DNA demethylation while the expression of SOX17 activates PRDM1, which then sustains the epigenetic program toward global 5mC erasure ( Tang et al. , 2015 ). Comparisons of the epigenome of in vitro derived hPGCLCs to that in vivo provides instruction on the progress of generation of germ cells.

Although both mutations linked to infertility and gene expression in various stages of human germ cell development contribute to identification of genes that act at specific stages of development, functional analysis is necessary to validate their developmental roles and pinpoint underlying mechanisms. Recent developments in stem cell biology and gene editing provide an opportunity to recapitulate human germ cell development in vitro and to functionally dissect genetic requirements. A key step in developing in vitro gametogenesis is identifying and characterizing genetic determinants in a robust model for germ cell specification. While it is clear that in vitro -derived germ cells lack important characteristics of authentic FGCs (notably, the ability to efficiently develop into gametes in vitro ), in vitro gametogenesis provides a viable system to explore the core transcriptional machinery for germ cell specification. Numerous studies have contributed to protocols for directing germ cell differentiation from hPSCs, starting from both hESCs and human induced pluripotent stem cells (hiPSCs). Currently, human primordial germ cell-like cells (hPGCLCs) can be induced using rationally designed cocktails of growth factors and small molecules that have emerged over the years (Kee et al. , 2006 , 2009 ; Easley et al. , 2012 ; Irie et al. , 2015 ; Sasaki et al. , 2015 ; Sugawa et al. , 2015 ; Jung et al. , 2017 ; Yamashiro et al. , 2018 ; Murase et al. , 2020 ), or induced by ectopic expression of genes associated with germ cell development, especially transcription factors ( Kee et al. , 2009 ; Qiu et al. , 2013 ; Yu et al. , 2014 ; Irie et al. , 2015 ; Medrano et al. , 2016 ; Panula et al. , 2016 ; Jung et al. , 2017 ; Fang et al. , 2018 ; Kojima et al. , 2021 ). To induce maturation and more advanced differentiation in vitro , hPGCLCs are often co-cultured with somatic cells (Park et al. , 2009 ; Lin et al. , 2014 ; Yamashiro et al. , 2018 ). In addition, xenotransplantation assays have been extensively applied to promote in vivo maturation of in vitro -derived germ cells. Two recent review articles have summarized progress and strategies of in vitro gametogenesis ( Li e t al. , 2020 ; Saitou and Hayashi, 2021 ). Delineation of the conditions and factors required to promote in vitro PGC differentiation have set the stage for genetic studies that can probe the function and hierarchies of transcription factors during hPGC specification ( Fig. 2 ). Several transcription factors which are usually involved in lineage specification during embryogenesis have been reported to be repurposed in PGCs to form a specific transcriptional network that may act to safeguard human germ cell fate by maintaining pluripotent status while repressing differentiation. In response to WNT ( Kojima et al. , 2017 ) and ACTIVIN signals, the mesoderm specifier EOMES activates SOX17 , an endoderm specifier, which in turn upregulates PRDM1 . Deletion of EOMES in hPSCs significantly impacts their competence toward hPGCLC differentiation ( Chen et al. , 2017 ; Kojima et al. , 2017 ). SOX17 can also be induced directly by BMP signaling to activate germ cell programs: conversely, SOX17 -null hESCs cannot undergo hPGCLC specification ( Irie et al. , 2015 ; Tang e t al. , 2015 ). The trophoblast marker GATA3 is an immediate effector of the BMP pathway and regulates SOX17 and TFAP2C . Accordingly, GATA3 null mutations significantly decreased hPGCLC induction efficiency in response to BMP signals ( Kojima et al. , 2021 ). PRDM1 is a transcriptional repressor that acts as one of the key signature genes for germ cell fate ( Ohinata et al. , 2005 ; Irie et al. , 2015 ; Kobayashi et al. , 2017 ). PRDM1 function is tightly controlled by multiple transcription factors to repress somatic differentiation during the process of hPGCLC specification; in PRDM1 -deficient or PRDM1 -knockdown cells, germline differentiation potential is significantly impaired, and somatic lineage genes are de-repressed ( Lin et al. , 2014 ; Sasaki et al. , 2015 ). Part of the role of PRDM1 protein is to suppress SOX2 expression and consequently inhibit neuronal differentiation directly. A study that examined hPGCLC specification in TFAP2C −/− cells found that TFAP2C acts upstream of PRDM1 and plays a dominant role in repressing somatic programs in hPGCLCs ( Kojima e t al. , 2017 ). Another study used single-cell sequencing in TFAP2C −/− cells during hPGCLC specification and confirmed that TFAP2C functions upstream of both PRDM1 and SOX17 , acting to prevent cells from adopting somatic fates and thus safeguard germ cell fate ( Chen et al. , 2019 ). The core transcriptional network in human pluripotent stem cells and primordial germ cells. Arrows with pointed tips represent activation, and arrows with vertical line tips represent inhibition. In pluripotent stem cells, FGF2, WNT and ACTIVIN signaling pathways are essential to activate the gene expression program for pluripotency. In response to the signals, OCT4, SOX2 and NANOG are activated and form a core transcriptional network that suppresses the somatic and germline gene expression program. Once human pluripotent stem cells (hPSCs) start to differentiate toward germline, WNT3, ACTIVIN and BMP4 signals activate EOMES and GATA3, which then activate the expression of a few transcription factors essential for germ cell development, including SOX17, TFPAP2C and PRDM1. Moderate expression of pluripotency transcription factor OCT4 is also critical for human germ cell development. Upon differentiation, the expression of OCT4 is gradually reduced, and the expression of its functional partner in hPSCs, SOX2, is diminished. Instead, OCT4 partners with PAX5 in human primordial germ cells (hPGCs) to activate the expression of PRDM1. In addition to transcription factors that act in specifying lineages in development, pluripotency transcription factors are important in germ cell development. One of the unique features of hPGCs compared with other cell types of the body during development is that they share with hPSCs the expression of several pluripotency genes, including a pluripotency master regulator OCT4 / POU5F1 . In both mouse and human embryo development, OCT4 is initially expressed in all blastomeres of the embryo; subsequently, expression is restricted to the pluripotent stem cells of the inner cell mass. During gastrulation OCT4 level is maintained in epiblast cells and after gastrulation OCT4 expression is confined exclusively to germ cells ( Scholer et al. , 1990 ; Scholer, 1991 ; Yeom et al. , 1996 ; Nichols et al. , 1998 ; Pesce and Scholer, 2001 ). Mouse embryos depleted of OCT4 fail to form an inner cell mass and the cells are committed to the trophoblast lineage ( Nichols et al. , 1998 ); However, conditional knock out of OCT4 in mouse PGCs leads to apoptosis of PGCs rather than cell fate change to the trophectodermal lineage ( Kehler et al. , 2004 ), suggesting that OCT4 is playing distinct roles in these two distinct cell types. To dissect the roles of OCT4 in hPSCs and hPGCs, Fang et al. (2018) compared genome-wide binding of OCT4 in hPSCs and hPGCs (the latter from human fetal testis samples). They discovered that OCT4 repressed neuronal differentiation in both hPSCs and hPGCs, while it regulated a unique set of genes during germ cell differentiation by switching partners from SOX2 to PAX5 (paired box 5). In hPSCs, OCT4 and SOX2 interact and form a protein complex to cooperatively bind and regulate target genes in order to activate or maintain pluripotency ( Herr and Cleary, 1995 ; Nichols et al. , 1998 ; Wegner, 1999 ; Niwa et al. , 2000 ; Avilion et al. , 2003 ). As hPSCs begin to differentiate toward a germline fate, the expression of SOX2 is diminished and OCT4 switches functional partners to PAX5 as germ cells are specified. The PAX5-OCT4 complex functions in activation of PRDM1 expression and other genes implicated in PGC specification. PAX5 null mutations have significantly reduced PRDM1 expression and impaired germ cell potential in hPSC xenotransplants in vivo . Hence, the PAX5-OCT4-PRDM1 proteins function as a genetic switch in the transition from a pluripotent state to germline ( Fang et al. , 2018 ). PRDM14 , another human pluripotency gene, is also critical for the acquisition and maintenance of the hPGCLC-competent state ( Sybirna et al. , 2020 ); it functions to activate OCT4 expression and to upregulate PRDM1 ( Chia et al. , 2010 ). Loss of PRDM14 function results in significantly reduced efficiency of in vitro differentiation and an aberrant transcriptome of the resultant hPGCLCs ( Sybirna et al. , 2020 ). Overexpression of SOX17 leads to the generation of hPGCLCs without BMP induction, suggesting that SOX17 is at the top of the transcriptional hierarchy for hPGCLC specification and is sufficient for human germ cell fate acquisition ( Irie et al. , 2015 ). Although a major role of SOX17 is to activate TFAP2C and PRDM1 , forced expression of TFAP2C could not generate hPGCLCs, even in conjunction with PRDM1 overexpression ( Kobayashi et al. , 2017 ). These results suggest that while TFAP2C is indispensable for hPGCLC specification, it is insufficient on its own for germline induction. Recent work reported that the GATA family of transcription factors ( GATA2 / 3 ), combined with SOX17 and TFAP2C , act as a minimum requirement to replace BMP signaling and confer germ cell fate on incipient mesoderm-like cells (iMeLCs) ( Kojima et al. , 2021 ). In total, these genetic studies of hPGCs in vitro begin to allow the construction of a network of transcription factors that are involved in hPGC specification and maturation ( Fig. 2 ).

Human embryo development, like that of other organisms, is characterized by a series of cell-fate transitions from one cell type to another, starting from pluripotent stem cells (PSCs) and progressively specifying different lineages including extra-embryonic tissues, germ cell and somatic cell lineages. The primordial germ cells (PGCs) arise early in development as a small group of embryonic cells that will ultimately give rise to sperm and oocytes, and pass on genetic information to subsequent generations ( Waters and Trainer, 1996 ; Donovan, 1998 ; Tang et al. , 2016 ; Kobayashi and Surani, 2018 ). The correct functioning of lineage specification is obviously critical; dysfunction during gametogenesis may lead to defects in germ cell development and/or function underlying diverse genetic fertility syndromes ( Krausz and Riera-Escamilla, 2018 ; Xavier et al. , 2021 ). In this review, we use the term ‘specification of cell fate or identity’ in reference to when a cell is committed to differentiate down a specific pathway if left in its normal environment. Germ cell development is dependent on the regulators of gene expression that function at multiple levels, including transcription factors that orchestrate expression at the transcriptional level by binding to enhancer or promoter regions of target genes. Following embryonic genome activation, a series of transcription factors sequentially regulates the activity of a host of genes involved in cell fate decisions, including PGC specification and migration, sex determination, meiosis and germ cell maturation. Concurrently, developmentally regulated protein expression is also proceeding with coordination by RNA-binding proteins, beginning at fertilization with the translation of maternally inherited mRNA and continuing throughout germ cell development, as evidenced by the number of RNA-binding proteins defined as markers of late stages of germ cell lineages ( Clark and Reijo Pera, 2006 ; Makar and Sasaki, 2020 ). PGCs exhibit many properties of classic pluripotent cells, including the property of pluripotency itself, and yet they are committed to the germ cell lineage ( Kuijk et al. , 2011 ). The prime example or archetype of a pluripotent cell type, namely embryonic stem cells (ESCs), maintain their undifferentiated state via the activity of a defined set of transcription factors, coordinately regulating those genes necessary for reinforcing the pluripotent state, and suppressing lineage-specific genes that would otherwise drive differentiation ( Kim et al. , 2008 ; Niwa, 2009 ; Ng and Surani, 2011 ). PGCs appear to employ a subset of members of this set of genes while also adopting a distinct subset or circuitry of transcription factors to define their identity and complete three crucial developmental events: repress somatic programs; reacquire pluripotency; and reprogram genome-wide epigenetics. For example, although human PGCs (hPGCs) are committed to the germ cell lineage, they share expression of a subset of pluripotency genes with human ESCs (hESCs), notably OCT4 (also known as POU5F1 , POU class 5 homeobox 1) and NANOG ( Kehler et al. , 2004 ; Hoei-Hansen et al. , 2005 ); however, other key pluripotency genes, such as SOX2 (SRY-related HMG box-containing gene 2), are not expressed in hPGCs ( Perrett et al. , 2008 ). Co-expression of pluripotency transcription factors, as well as lineage specifiers, distinguishes hPGCs from all other human embryonic cell types as well as mouse PGCs (mPGCs) ( Tang et al. , 2015 ). To maintain cell identity, hPGCs likely require a precise regulation/balance of pluripotency-related and lineage-specific transcription factors to repress somatic differentiation and concurrently activate germ cell programs. A continuum of in vivo and in vitro models, based on human, mouse and non-human primate cells, has been explored and leveraged to study germ cell development, including the formation of PGCs, and their specification from PSCs or ESCs ( Li et al. , 2020 ; Saitou and Hayashi, 2021 ). While mouse models are extraordinarily useful given their genetic malleability and ability to probe in vivo development of engineered cells, the genetics of germ cell development has both similarities and differences between species ( Sasaki et al. , 2016 ; Kojima et al. , 2017 ; Stirparo et al. , 2018 ). For example, efforts to define a core set of transcription factors sufficient for PGC specification have succeeded in driving or even actively directing mouse cells further down the germ cell lineage than what has been achieved in human cell models ( Niwa, 2009 ; Magnúsdóttir et al. , 2013 ). Indeed, it is likely that the microenvironment of the mouse gonad provides as yet undefined signals to induce germ cell differentiation of PGCs; moreover, xenotransplantation and co-culture with somatic cells have provided a superior microenvironment for further development of in vitro -derived PGC-like cells ( Dominguez et al. , 2014 ; Durruthy Durruthy et al. , 2014 ; Ramathal et al. , 2014 ). Finally, recent analyses of bona fide germ cells in developing human embryos have provided insight into transcription factor expression as well as their interactions and functions during development ( Otte et al. , 2017 ; Wen and Tang, 2019 ; Estermann and Smith, 2020 ; Li et al. , 2020 ; La et al. , 2021 ). Further analyses of these experiments are likely to add to our library of transcription factors potentially required for later stages of PGC function and germ cell development.

Prior to recent advances in stem cell biology and sequencing technologies, our understanding of germ cell development relied almost solely on animal models. However, germline commitment occurs within a limited window of embryo development, when the morphology of embryos and the timing of germ line specification diverges significantly between different species, including mice and humans ( Sybirna et al. , 2019 ) ( Fig. 3 ). Notably, mouse embryos develop as an egg cylinder, and mouse PGCs (mPGCs) are clustered in the proximal epiblast around the time of primitive streak formation ( Tam and Behringer, 1997 ; Anderson e t al. , 2000 ; McLaren, 2003 ). Human embryos present as a bilaminar disc and, based on the studies in non-human primates, hPGCs probably arise prior to primitive streak formation from the dorsal amnion, which is physically separate from the posterior epiblast ( Behringer et al. , 2000 ; Rossant, 2015 ; Sasaki et al. , 2016 ; Kobayashi et al. , 2017 ). Given such major developmental differences in terms of timing, shape and cell origin, it is not surprising that the intrinsic transcriptional network required for PGC emergence also has divergent components and functions. In mPGCs, the Prdm1, Tfap2c and Prdm14 proteins constitute a core transcriptional network that is essential for PGC specification in vivo ( Ohinata e t al. , 2005 ; Yamaji et al. , 2008 ; Weber et al. , 2010 ) and sufficient to induce germ cell fate in vitro ( Magnúsdóttir et al. , 2013 ; Nakaki et al. , 2013 ). In contrast, the expression of PRDM14 in humans is strongly downregulated from hPSCs to hPGCs ( Sugawa et al. , 2015 ), and SOX17 has instead emerged as the critical determinant for hPGC specification ( Irie et al. , 2015 ). Pluripotency genes OCT4 and NANOG are re-expressed in both mPGCs ( Murakami et al. , 2016 ) and hPGCs ( Guo et al. , 2015 ; Tang et al. , 2015 ), and SOX2 is absent in hPGCs ( Lin et al. , 2014 ) although it is required for mouse PGC survival and proliferation ( Campolo et al. , 2013 ). In addition, the mesoderm specifier gene T/Brachyury is essential for robust activation of Prdm1 and Prdm14 in mPGCs ( Aramaki et al. , 2013 ) but this role is replaced by another mesoderm gene, EOMES , in hPGCs ( Chen et al. , 2017 ; Kojima e t al. , 2017 ). The co-expression of pluripotency genes and lineage specifier genes persists to the sex determination stage when PGCs differentiate to spermatogonia and oogonia in both mice and humans. Differences in transcription factor expression between murine and human germ cells . E, embryonic day; ICM, inner cell mass; WK, week. Green marks the genes specifically expressed in mouse germ cell development, and red marks the genes specifically expressed in human germ cell development. The expression of pluripotency genes also differs during germ cell development in mice and humans. In male mice, expression of OCT4 persists as cell fate transits from PGCs to the undifferentiated SPG stage, and expression is downregulated once cells enter meiosis ( Pesce et al. , 1998 ; Tadokoro et al. , 2002 ). However, OCT4 expression, specifically isoform OCT4A (translated from transcript variant 1), is more restricted during male development in humans, being confined to hPSCs and hPGCs. In female mice, the expression of OCT4 is downregulated by the onset of meiotic prophase and then re-activated after birth in oocytes within primary follicles and at the onset of folliculogenesis ( Pesce et al. , 1998 ; Anderson et al. , 2007 ). In humans, the number of OCT4 positive cells peaks by gestational week 8 and diminishes after week 9, as oogonia enter meiosis ( Kerr et al. , 2008 ). NANOG has a similar expression pattern to OCT4 ( Hoei-Hansen et al. , 2005 ). Once FGCs arrive at the gonads and progress toward meiosis, pluripotency-related transcription factors undergo significant downregulation and are diminished in all adult reproductive tissues. The FOXO subclass of the forkhead box transcription factors are key regulators of mouse reproduction ( Brosens et al. , 2009 ). FOXO3 is required to suppress primordial follicle activation in females as FOXO3 -null female mice display age-dependent infertility ( Castrillon et al. , 2003 ; Hosaka et al. , 2004 ), and FOXO1 is essential for SSCs maintenance and the initiation of spermatogenesis in males ( Goertz et al. , 2011 ). While FOXO1 expression in spermatogonia and granulosa cells is conserved between humans and mice ( Richards e t al. , 2002 ; Liu et al. , 2013 ), FOXO3 is not expressed in primordial oocytes in humans ( Tarnawa et al. , 2013 ), suggesting that other members of the FOXO transcription factor family may replace its function. FOXO6 , which has been identified as an upregulated gene in human oocytes by scRNA-seq, may be a potential candidate for this substitution of FOXO3 function.

The National Institute of Health (1R01HD096026); the National Natural Science Foundation of China (32070830); and the Fundamental Research Funds from University of Science and Technology of China (WK9110000141, YD9100002007).

PubMed database was used to search articles and reviews with the following main keywords: human gametogenesis; transcription factors; infertility; germ cell; germ cell tumors; infertility; gene mutations; single-cell RNA-sequencing; pluripotent stem cells; in vitro differentiation; and other key terms related to these subjects. The search period included all publications until now (November 2021).

Previous research in multiple different stem-cell based experimental systems has demonstrated that cell and tissue transplantation is the gold standard for testing cell identity and function ( Hanna et al. , 2007 ; Nelson et al. , 2009 ; Weissman, 2012 ; Takahashi and Yamanaka, 2013 ). As noted in a recent review, transplantation may provide the first ability to demonstrate the feasibility of generating PGCLC-derived, fully mature gametes in primates, including humans ( Saitou and Hayashi, 2021 ). Three types of studies have been reported with different transplantation strategies and are summarized below. In the first studies, whole slices of testicular tissue were transplanted from non-human primates ( Macaca fascicularis ) to a subcutaneous region of the back or side of male nude mice ( Liu et al. , 2016 ). The results indicated that xenotransplanted testis tissue from young mammals, including mice and monkeys, is capable of undergoing full and complete spermatogenesis in xenografts. Moreover, monkey xenografts to nude mice were capable of generating sperm capable of producing monkey offspring when sperm derived from the xenografts testis tissues from juvenile wild-type (WT) and transgenic cynomolgus monkeys ( M. fascicularis ) were used for assisted reproduction. These results may inform future strategies for in vitro gametogenesis. Second, in an elegant set of proof-of-concept experiments, Rhesus macaque testicular tissue that was cryopreserved as an experimental validation of fertility preservation for prepubertal human patients was used in autologous transplantation ( Fayomi et al ., 2019 ). Prepubertal testicular tissue that had been cryopreserved was autologously grafted under the back skin and scrotal skin of castrated pubertal R. macaque monkeys. Results indicated that mature sperm were produced that were capable of fertilizing R. macaque oocytes and culminating in live birth. This suggests that in non-human primates there is the capability for full gametogenesis, from spermatogonia to mature sperm, through autologous grafting. Third, in studies reported in several manuscripts, the ability of germ cells to engraft was used to assess validity and quantitative and qualitative aspects of in vitro gametogenesis. In these studies, the ability of hPGCLCs to engraft in mouse seminiferous tubules following differentiation in vitro was tested ( Dominguez et al. , 2014 ; Durruthy-Durruthy et al. , 2014 ; Ramathal et al. , 2014 ; Medrano et al. , 2016 ; Fang et al. , 2018 ). Results demonstrated that the ability to differentiate and produce engraftable germ cells was dependent on the genetic composition of the parental iPSC lines used to produce PGCLCs. Thus, PGCLCs differentiated from iPSC lines with Y chromosome deletions that are linked to poor or no spermatogenesis were shown to have limited engraftment relative to PGCLCs without genetic abnormalities derived from fertile men; similar results were also observed with PGCLCs produced from lines with numerical sex chromosome abnormalities ( Dominguez et al. , 2014 ; Durruthy-Durruthy et al. , 2014 ; Ramathal et al. , 2014 ). In contrast, over-expression of key germ cell-specific genes was associated with increased engraftment ( Medrano et al. , 2016 ; Panula et al. , 2016 ; Fang et al. , 2018 ). Furthermore, these studies documented the migration of PGCLCs to the basement membrane of spermatogenic tubules and expression of spermatogonial markers. As expected, however, complete spermatogenesis was not observed owing to the evolutionary distance between mice and humans. Instead, as reported previously, in transplantation with human SSC the germ cells migrate to the seminiferous tubule basement membrane and proliferate to form chains and patches of spermatogonia that persist in the long term but do not appear to initiate or complete meiosis ( Nagano et al. , 2002 ; Hermann et al. , 2010 ; Dovey et al. , 2013 ).

The authors declare that there are no conflicts of interest.

Considering the complexity of the development pathways of germ cells in humans and the relation to the processes outlined above, it is clear that given the rarity of germ cells in developing human embryos and the poor resolution of germ cell isolation methods, it is not possible to profile development and gene expression patterns at all stages. In addition, bulk RNA-seq or microarray analysis cannot resolve the heterogeneity within germ cells, which is essential for understanding the precise trajectory in which development occurs ( Raser and O'Shea, 2005 ; Plass et al. , 2018 ). Transcriptome profiling at the single-cell level (i.e. single-cell RNA-sequencing: scRNA-seq) has been used to overcome this limitation by comprehensively measuring mRNA levels within all individual germ cells at a given developmental stage, and has been applied to diverse biological systems to begin to explore the potential molecular mechanisms for development ( Junker and van Oudenaarden, 2014 ; Raj et al. , 2018 ; Genga et al. , 2019 ; Han et al. , 2020 ). Since 2013, a handful of reports have characterized the transcriptional dynamics during human germ cell development by analyzing human fetal and adult tissues using scRNA-seq. These studies were recently reviewed ( Li et al. , 2020 ). Here, we focus on transcription factors that potentially act as master regulators to activate the unique gene expression program for each specific stage of germ cells ( Fig. 1 ; Table I ). We note that expression of a gene does not imply function; moreover, it is highly likely that genes are expressed at stages other than those that have been assayed and/or only briefly during development. Transcription factors expressed/upregulated in developing human germ cells identified by scRNA-seq . EPI, epiblast; FGC, fetal germ cells; PGC, primordial germ cells; scRNA-seq, single-cell RNA-sequencing. Transcription factors expressed in developing human germ cells, determined by single-cell RNA-sequencing. FOXP1, SOX2, NANOG and KLF4 are involved in the creation of pluripotency in EPI cells. ESSRB: maintain ESCs with OCT4 and SOX2. PRDM14: essential for pluripotency and germ cell formation. NR5A2: important for embryonic development. SMAD: cell proliferation, apoptosis, and differentiation. KLF17: involved to spermatid differentiation and oocyte development. ARGFX: related to pre-implantation embryo. TFCP2L1: important for establishment and maintenance of pluripotency in ESCs. ZFP42: involved in the reprogramming of X-chromosome inactivation during the acquisition of pluripotency and ESCs self-renewal. DPPA2: associated with developmental pluripotency SALL1: transcriptional regulation of PSCs. HAND1: acts as a transcriptional repressor of SOX15. UTF1: involved in differentiation of embryonic carcinoma and ESCs. ELF5: involved in tumorigenesis. Early PGC:   NANOG, POU5F1, SALL4, KLF4, ZFP42, TFAP2C, T, SOX15, SOX17 Mitotic PGC:   NANOG, POU5F1, SALL4, SOX17, SOX15, SOX13, PRDM1, PRDM14, TFAP2C, TFCP2L1, KLF4 Male PGC mitotic arrest:   EBF3, SATB2, ZNF267, FEZF1, SOX12, DMRTB1, ZSCAN5A, SOX30, GATA3, FOXD1 Pro-Meiotic FGC:   ZNF208, YBX1, ZNF791, PRDM9, LHX8, NR4A2, ZNF382, MGA, ZIC4, HSF2, DMRTA2, L3MBTL1, SOX30, GATA3, FOXD1 EBF3: inhibits cell survival through the regulation of genes involved in cell cycle arrest and apoptosis. ZNF267 and ZSCAN5A: involved in transcriptional regulation. FEZF1:(GO) annotations related to this gene include RNA polymerase II proximal promoter sequence-specific DNA binding. DMRTB1: gene Ontology (GO) annotations related to this gene include DNA-binding transcription factor activity and sequence-specific DNA binding. ZIC3: this nuclear protein probably functions as a transcription factor in early stages of left-right body axis formation. KLF6: transcriptional activator. LEF1: participates in the WNT signaling pathway. PGC:   NANOG, POU5F1, SALL4, TCF3, KLF4, STAT3, MBD3, DMRT1 SSC:   DMRT1/6, BCL6, ID4, SALL4, ETV5, TCF3, KLF4, KLF2, STAT3, MBD3 SPG:   NR6A1, SOHLH2, TCF3, KLF4, KLF2, STAT3, DMRT1 DMRT1/6: plays a central role in spermatogonia by inhibiting meiosis in undifferentiated spermatogonia and promoting mitosis, leading to spermatogonial development and allowing abundant and continuous production of sperm. BCL6: downregulated during maturation of dendritic cells by selective stimuli such as bacterial lipopolysaccharide. ID4: implicated in regulating a variety of cellular processes, including cellular growth, senescence, differentiation, and apoptosis. SALL4: plays a key role in the maintenance and self-renewal of embryonic and hematopoietic stem cells. NR6A1: may be involved in the regulation of gene expression in germ cell development during gametogenesis. SPG:   ETV5, ID4, SOHLH1, RHOXF1, ZBTB16, BCL6B, FOXO1, DMRT1, SOHLH2, NEUROG3, MYBL1, CHD5, GATA4, NR5A1 SSC:   ETV5, ID4, SOHLH1, RHOXF1, ZBTB16, BCL6B, FOXO1, DMRT1 SPC:   ETV5, SOHLH1, SOHLH2, NEUROG3 SPT:   ETV5, ZBTB16, BCL6B, FOXO1, NEUROG3, HSFY2, WT1, GATA4, NR5A1, DMRTB1 SPG:   MYBL2, E2F4, ZNF709, HINFP, DMRTB1, CTCFL SSC:   JUND, HMGA1, ZBTB16, POU3F1, UTF1, SALL4, FGF2, SMAD1, SMAD5, SMAD9, ID1, ID2, ID4, TCF7L2, SSRP1, SOX4, CLOCK, RHOXF1, TFDP2, PARP1, PMS1, E2F4, HINFP, DMRT1 SPC:   OVOL1, OVOL2, BRCA1, CREB1, PMS1, MTF1, CEBPZ, MYT1L, PARP1 SPT:   HDAC4, NFKBIB, CHD5 SPG:   DMRT1, DMRTB1, SOHLH1, SOHLH2, BRCA1, TBX3, HOXA3 SSC: UTF1, SOHLH1, EGR4, ID4, TCF3, ETV5, DMRT1, MSL3 SPC:   DMRT1, DMRTB1, DMRTC2, DMRT3, SOX4, SOX5, SOX30, SOHLH1, SOHLH2, HOXB4, HOXC6, PRDM9 SPT:   DMRTB1, CREM, SOX30 CREM: plays a role also in human spermatogenesis and that the absence of the CREM switch can be associated to spermatogenic arrest. TBX3: acts as a negative regulator of PML function in cellular senescence. SSC:   EGR4 SPG:   SOHLH1/2, DMRT1 EGR4: the functional loss of EGR4 blocked spermatogenesis, leading to a significant reduction in spermatozoa production. SOHLH1: plays a pivotal role in the transition of germ cells from primordial to primary follicles and in the differentiation of spermatogonia. SOHLH2: plays a pivotal role in the transition of germ cells from primordial to primary follicles and in the differentiation of spermatogonia. SPG:   UTF1 SPC:   PRDM9 PGC:   POU5F1, NANOG, TFAP2C SOX17 SSC:   EGR4, MSL3 RA-Responsive female FGC:   ID3, RARB, ZGLP1, HES6, ZNF362, ZBTB11, HOXA5, HOXB6, HMGB3, PBX1, SOX30, GATA3, FOXD1 Oogenesis:   NOBOX, FIGLA, STAT1, TBX3, AFF1, JARID2, NFKB2, NR3C2, SOX30, GATA3, FOXD1 ID3: luteinization, oogenesis, oocyte maturation. RARB: receptor for retinoic acid. ZGLP1: germ cell development HES6: members of this gene family regulate cell differentiation in numerous cell types. ZNF362, ZBTB11: may be involved in transcriptional regulation. HOXA5, HOXB6: provides cells with specific positional identities on the anterior-posterior axis. HMGB3: plays a fundamental role in DNA replication, nucleosome assembly and transcription. PBX1: may have a role in steroidogenesis, sexual development and differentiation. PRDM9: the zinc finger array recognizes a short sequence motif, leading to local H3K4me3, and meiotic recombination hotspot activity. LHX8: plays a role in tooth morphogenesis, oogenesis and in neuronal differentiation. NR4A2: encodes a member of the steroid-thyroid hormone-retinoid receptor superfamily. ZNF382: may play roles in differentiation, proliferation and apoptosis. MGA: functions as a dual-specificity transcription factor, regulating the expression of both MAX-network and T-box family target genes. ZIC4: members of this family are important during development. HSF2: activates heat-shock response genes. DMRTA2: may be involved in sexual development. L3MBTL1: probably plays a role in cell proliferation. NOBOX: oogenesis. FIGLA: functions in postnatal oocyte-specific gene expression. STAT1: induces a cellular antiviral state. TBX3: regulates developmental processes. AFF1: implicated in human childhood lymphoblastic leukemia. NFKB2: the endpoint of a series of signal transduction events. JARID2: stem cell differentiation and normal embryonic development. NR3C2: functions as a ligand-dependent transcription factor that binds to mineralocorticoid response elements. Primordial follicle:   SOX30 Primary follicles:   GTF2I, CSDE1, SOHLH2, SMARCE1, TUB, HBP1, SOX30, HIF1A Secondary follicles:   KLF2, YBX2, FOXO6, SOX13, ETV5, TEAD2, OTX2 Antral follicles:   PINX1, PBX1, MTF1, SOX15, UBTF, SOX13 MII oocytes of preovulatory follicles:   ATF2, EOMES In the oocytes, the expressions of GTF2I, CSDE1, SOHLH2, SMARCE1, TUB, HBP1, SOX30 and HIF1A were upregulated in primary follicles, indicating that these TFs may play a critical role in the transition from the primordial to the primary stage. KLF2, YBX2, FOXO6, SOX13, ETV5, TEAD2 and OTX2 were overexpressed in the oocytes from secondary follicles compared to those from primary follicles, implying that they are likely the regulators of the primary-to-secondary stage transition. PINX1, PBX1, MTF1, SOX15, UBTF, SOX13 and POU2F1 had higher expression levels in the oocytes of antral follicles compared to those of secondary follicles, indicating possible regulatory roles in the cytoplasmic and nuclear maturation of oocyte at the antral stage. ATF2 and EOMES were abundantly expressed in the MII oocytes of preovulatory follicles, indicating their potential roles in the unique transcription networks. EPI, epiblast; FGC, fetal germ cells; hESCs, human embryonic stem cells; PGC, primordial germ cells; PSC, pluripotent stem cells; SPG, spermatogonia; SSC, spermatogonia stem cells; WNT, Wingless-related integration site. The gene expression patterns of migrating and mitotic PGCs are similar in male and female germ cells. There is continued expression of transcription factors associated with pluripotency and ESCs, such as POU5F1/OCT4, NANOG and PRDM14 (PR/SET domain 14) ( Guo et al. , 2015 ), although at different levels relative to pre-implantation epiblasts (EPI) ( Yan et al. , 2013 ; Blakeley et al. , 2015 ; Guo et al. , 2015 ; Petropoulos et al. , 2016 ; Li et al. , 2017 ; Stirparo et al. , 2018 ; Zhou et al. , 2019 ). Concurrently, however, transcription factors that are diagnostic of germline cells, such as PRDM1(PR/SET domain 1) and TFAP2C (Transcription Factor AP-2 gamma) , and somatic lineages, such as BRACHYURY (T) and EOMES (eomesodermin), are also expressed in the same cells ( Guo et al. , 2015 ; Tang et al. , 2015 ). A recent finding has also demonstrated that during hPGC specification, the classic endodermal transcription factor marker protein, SOX17 (SRY-related HMG box-containing gene 17), is required for hPGC commitment in an in vitro model of hPGC differentiation ( Irie et al. , 2015 ). Moreover, scRNA-seq data of human gonadal PGCs in vivo confirmed the presence of SOX17 in early migrating and mitotic PGCs, consistent with its essential role in hPGC function ( Guo et al. , 2015 ). To shed light on the critical transcriptional regulation in sex determination, scRNA-seq analyses were performed on both germ cells and their gonadal niche cells in multiple studies ( Guo et al. , 2015 ; Li et al. , 2017 ; ChitiashviLi et al. , 2020 ; Guo et al. , 2021 ; Zhao et al. , 2021 ). Male sex determination initiates with activation of the Y chromosome-specific transcription factor, SRY, a dominant determinant for testis differentiation ( Berta et al. , 1990 ; Gubbay et al. , 1990 ; Koopman et al. , 1990 ; Sinclair et al. , 1990 ; Kashimada and Koopman, 2010 ). Transcription factors that regulate SRY function and have been shown likely to be required for male sex determination include WT1(Wilms’ tumor gene), NR5A1(nuclear receptor subfamily 5 group A member 1), GATA4(GATA binding protein 4), FOG2(FOG family member 2) and CBX2 (chromobox2) ( Sekido and Lovell-Badge, 2008 ). Once SRY is activated, it acts by upregulating the expression of SOX9, which then activates a cascade including AMH, prostaglandins and steroidogenic genes, to promote complete organogenesis of the testis in humans and suppress the pro-ovary pathways ( Koopman, 2001 ; Kozhukhar, 2012 ). SRY is also a direct target of the WT1. WT1 is a zinc finger containing DNA-binding protein that activates the expression of SRY in the initial sex determination process in humans ( Shimamura et al. , 1997 ; Hossain and Saunders, 2001 ; Matsuzawa-Watanabe et al. , 2003 ). Other transcription factors that are essential for early testis differentiation include NR5A1, a highly conserved nuclear receptor transcription factor that interacts with SRY to regulate SOX9 expression during the differentiation of Sertoli cells ( Sekido and Lovell-Badge, 2008 ; Rotgers et al. , 2018 ; Stevant and Nef, 2019 ). GATA4, a zinc finger transcription factor, also cooperatively interacts with NR5A1 to regulate downstream genes critical for testis differentiation ( Viger et al. , 2008 ). Similarly, FOG2, a zinc finger cofactor, is suggested to be involved in testis determination through interaction with Gata4, potentially by modulating the activity of GATA4, and regulating the expression of SRY and SOX9 ( Zaytouni et al. , 2011 ). CBX2, a component of the polycomb group (PcG) complex of regulatory proteins, has been reported to act in testis determination by activating the expression of NR5A1 and SRY and repressing genes involved in fetal ovarian development ( Biason-Lauber et al. , 2009 ). Transcription factors belonging to the doublesex and mab-3 related transcription factor (DMRT) family, including DMRT1, 2 and 3, are found to be evolutionarily conserved sex-determining transcription factors. DMRT1 is a male-specific transcription factor gene which functions at multiple stages during male germ cell and Sertoli cell development to support spermatogonial development by antagonizing FOXL2 (forkhead box L2) activity and repressing the oogenesis program ( Matson et al. , 2011 ). Mouse models found that DMRT1- mutant mice fail to develop functional testes, and continued expression of DMRT1 is necessary to prevent female reprogramming in the postnatal testis ( Matson et al. , 2011 ). Mutation of DMRT transcription factors causes abnormal testicular formation and feminization ( Ottolenghi and McElreavey, 2000 ). Female sex determination is regulated by transcription factors associated with RSPO1-WNT4-β-catenin signaling pathways. FOXL2 is considered a gatekeeper transcription factor for ovarian identity ( Uhlenhaut et al. , 2009 ; Pannetier and Pailhoux, 2010 ) and promotes ovary development by blocking testis development through transcriptional repression of SOX9 ( Crisponi et al. , 2001 ; De Baere et al. , 2002 ; Udar et al. , 2003 ; Nallathambi et al. , 2007 ; Hersmus et al. , 2008 ; Shah et al. , 2009 ; Auguste et al. , 2011 ). Consistent with its critical role in ovarian cell function, somatic mutations in FOXL2 are found in nearly all cases of adult granulosa cell tumors of the ovary ( Jamieson and Fuller, 2012 ). Other genes, such as NR5A1 , may regulate anti-testis gene expression in the ovary; in 46, XX individuals, NR5A1 synergizes with β-catenin to upregulate the expression of anti-testis genes (e.g. DAX1/NR0B1(Dosage-sensitive sex reversal-Adrenal hypoplasia congenita critical region on the X chromosome, gene 1)) and possibly pro-ovarian genes ( Gummow et al. , 2003 ; Hossain and Saunders, 2003 ; Jordan et al. , 2003 ; Mizusaki et al. , 2003 ). Upon arriving at the genital ridge of a male embryo, germ cells arrest mitotically and transcription factors involved in cell cycle arrest, such as EBF3 (EBF transcription factor 3), are specifically upregulated ( Guo et al. , 2015 ). Several groups have profiled the transcriptional trajectory across the entire spectrum of human adult spermatogenesis. Transcription factors mainly involved in repressing gene expression (e.g. E2F4 (E2F transcription factor 4), HMGA1 (high mobility group AT-hook 1)) are enriched in SSCs, consistent with their slow proliferation rate. After progressing to the differentiating SPG, cell cycle activation-associated genes, such as KIT and KI67, are significantly upregulated to ensure active proliferation and differentiation. Later, transcription factors involved in meiotic sex chromosome inactivation, homolog synapsis and meiotic recombination, such as OVOL1 (Ovo like transcription repressor 1), SOHLH1 (spermatogenesis and oogenesis specific basic helix-loop-helix 1) and DMRT1, are upregulated to initiate the meiotic gene expression program ( Guo et al. , 2017 , 2018 ; Hermann et al. , 2018 ; Wang et al. , 2018 ; Sohni et al. , 2019 ). As SPC complete their differentiation into SPT, nearly all these transcription factors are downregulated as the overall level of transcription gradually declines ( Wang et al , 2018 ), with the exception of CHD5 (chromodomain helicase DNA binding protein 5), which is highly enriched in early SPT ( Wang et al. , 2018 ). This is probably because of its involvement in the process of condensation of spermatid chromatin by regulating histone hyperacetylation and the replacement of histones by transition proteins in chromatin ( Li et al. , 2014 ). After arriving at the genital ridge, female germ cells rapidly lose expression of pluripotency transcription factors, for example, POU5F1/OCT4 ( Rajpert-De Meyts et al. , 2004 ; Stoop et al. , 2005 ). Oogonia cells then undergo three sequential stages instructed by stage-specific transcription factors to generate fertilization-competent oocytes: the retinoic acid-responsive stage, the meiotic prophase stage and the folliculogenesis stage. Li et al. provided a thorough study to identify master transcription factors for germ cell development in the fetal stage ( Li et al. , 2017 ) using the ARACNe (algorithm for the reconstruction of accurate cellular networks) algorithm. ARACNe identifies master regulators of development by correlation of expression of transcription factors and their target genes across various cell types. Their analyses indicate that ZNF208 (ZiNc Finger protein 208) , YBX1 (Y-BoX-binding protein 1) and ZNF791 might be critical for the female mitotic phase, whereas HES6 ( HES family BHLH transcription factor 6 ), MAEL (Maelstrom spermatogenic transposon silencer 6), ZGLP1 (Zinc finger GATA-Like protein 1), ZNF362, ZBTB11 (ZiNc Finger and BTB domain containing 11), HOXA5 (HomeobOX A5), HOXB6, HMGB3 (High Mobility Group box3) and PBX1 (PBX homeobox 1 ) are the potential transcriptional regulators in the retinoic acid-responsive phase. Meiotic recombination transcription factor proteins LHX8 (LIM homeobox 8), together with NR4A2, ZNF382, MGA (MAX dimerization protein), RLF (RLF zinc finger), ZIC4 (Zic Family Member 4), PAXBP1 (PAX3 and PAX7 binding protein), HSF2 (heat shock transcription factor 2), DMRTA2 (DMRT like family A2) and L3MBTL1 (L3MBTL histone methyl-lysine binding protein 1), are implicated in shaping the gene expression program for meiosis in the meiotic prophase ( Guo et al. , 2015 ). Then cells start to express master transcriptional regulators, such as NOBOX (NOBOX oogenesis homeobox) and FIGLA (factor in germline alpha, also known as FIGLα or FIGα), to initiate the unique transcription network for folliculogenesis ( Li et al. , 2017 ; Wagner et al. , 2020 ; Ye et al. , 2020 ). Human folliculogenesis is a complex process comprising five key stages (primordial, primary, secondary, antral and preovulatory follicles). The development of follicles is considered to be associated with highly dynamic transcriptional regulation ( Aquila and De Amicis, 2014 ). Zhang et al. , explored the dynamic transcriptomes of the human oocyte, together with the neighboring granulosa cells across the entire process of follicular development, and identified potential master transcription factors for each stage using the ARACNe algorithm ( Zhang et al. , 2018 ). Interestingly, once cells begin follicular development, the DNA methyltransferases DNMT1, DNMT3A and DNMT3B are highly expressed at all stages of oocyte development, suggesting that maintaining a high level of DNA methylation is essential for oocyte maturation.

All authors contributed to the identification and critical evaluation of the relevant literature and reviewed all sections and critical discussion of the manuscript. F.F., P.J.I. and R.A.R.P. proposed the outlines, made the figures and drafted the main text. N.X., L.L. and L.D. performed the literature search and prepared the data for all the tables.

The identification and functional characterization of transcription factors implicated in human germ cell development may not only increase our understanding of the fundamental genetic basis of pluripotency and heredity, but also may help to dissect the causative mechanisms of germline diseases and provide potential strategies for treatment ( Fig. 5 ). With recent advances in stem cell biology and genetic and epigenetic profiling, there is the ability to compare and contrast data across different species, including human-specific aspects of germ cell development. Transcription factors that are specifically expressed or upregulated at critical developmental stages have the potential to act as master regulators of germ cell development. In addition to methods such as scRNA-seq that provide cell type-specific transcriptome information, rapid development of other single-cell ‘multi-omics’ technologies will integrate various components of genomic and epigenomic information, including DNA methylation, histone modification, chromatin accessibility, RNA expression and protein abundance within the same cell, enabling an in-depth understanding of gene regulatory mechanisms. Furthermore, advances in single-molecule imaging allow us to track the dynamic processes of development within a single cell at high spatial and temporal resolution. Together with these new technologies, we can develop a more comprehensive set of hypotheses to test our understanding of cell fate decisions, identity and the function of cells in normal germ cell development, physiology and disease. Knowledge of key transcription factors required for directed stem cell differentiation promises to provide additional tools to directly probe the function of putative master regulatory transcription factors in germ cell development through gain and loss-of-function studies. Improving human reproduction research through transcription factor-induced in vitro differentiation models . ESC, embryonic stem cell Successes in modeling mouse germ cell differentiation and gamete formation have thus far translated to human systems only in part, indicating gaps in knowledge of human-specific factors involved in mature germ cell development. Current methods of human germ cell differentiation in vitro still remain inefficient, asynchronous, and prone to variability, with a majority of the initial pluripotent cells differentiating toward various somatic lineages spontaneously. The efficiency of hPGCLC progression to more mature germ cells is also still very low ( Tang et al. , 2016 ). Developing multiple, testable strategies to differentiate hPSCs toward functional oogonia/spermatogonia efficiently is essential to model the entire germline development and diseases. In recent years, the biomedical community has witnessed the development of iPSC technology, which spawned a new era of controlling cell fate by modulating master transcription factors ( Takahashi and Yamanaka, 2006 ; Takahashi et al. , 2007 ). The successes in mouse germ cell differentiation enhance the prospect of transcription factor-induced differentiation of hESCs to hPGCs, and beyond. However, multiple transcriptome analyses have revealed that human germ cells acquire a unique transcriptional network, which is distinct from that of mouse germ cells and from that of any other human cell type. The germ cell transcriptome is composed of a surprising combination of core germ cell-specific genes, somatic lineage specifiers, and pluripotency genes. We and others have hypothesized that these lineage specifiers cooperate with the pluripotency networks to act as the earliest molecular switch in the developmental transition from PSCs to the germ cell lineage ( Fig. 4 ). Counteracting the activity of lineage specifiers, and continued transcription by pluripotency proteins, synergistically represses somatic lineages and activates germ cell programs to maintain germ cell identity. The identification and overexpression of these master transcription factors of human germ cells in hPSCs may enable faithful and efficient in vitro production of bona fide human germ cells and contribute to diagnosis and applications in human germ cell pathologies. Numerous protocols over the years have attempted to mimic the spatial component of embryonic development by stimulation of differentiating cells in 3D aggregates. However, these methods can expose cells at different positions in the aggregates to varying levels of extracellular signals, leading to heterogeneity in differentiation. The ability to maintain cells in a uniform monolayer and induce cell fate change by overexpression of transcription factors could enhance differentiation efficiency while improving final purity. More importantly, this strategy would reduce the complexity and variability of current protocols and enhance the widespread availability of germ cell research, as well as reproducibility, between laboratories. Simplifying and increasing the efficiency of differentiation of germ cells will also allow previously challenging studies, such as high-throughput drug discovery and genetic screening assays, thereby accelerating our understanding of human germ cell biology, and expediting infertility research. Finally, the ability to culture hPSCs and hPGCs indefinitely in vitro , in addition to providing an excellent model for germ cell research, could permit the repair of genetic defects in germ cell development by gene editing tools, for example, clustered regularly interspaced short palindromic repeats (CRISPR) technology. Note that in vitro hPSC differentiation-based methods or disease-modeling studies cannot diminish the importance of animal disease models for studying disease mechanisms or the downstream characterization and validation of putative drugs. However, the use of in vitro hPSC differentiation-based studies is providing additional tools for the identification of human-specific genes and pathways of development, and thereby accelerate the overall drug development process to treat human reproductive disease.

Human germ cell tumors (GCTs) are neoplasms presenting in the gonads, primarily in the testes. The transcriptome of GCTs is highly similar to authentic FGCs; thus, GCT cell lines are frequently used as a model to study the function of FGCs ( Irie et al. , 2015 ). GCTs can be broadly categorized into seminoma and non-seminomatous GCTs ( Oosterhuis and Looijenga, 2005 ; Vasdev et al. , 2013 ). Seminoma GCTs grow and spread more slowly and are sensitive to chemotherapy and/or radiation therapy. Non-seminomatous GCTs are divided into four subtypes: embryonal carcinoma, yolk sac carcinoma, choriocarcinoma and teratoma. Compared with seminoma, non-seminomatous GCTs are very variable in phenotype and prognosis. Non-seminomatous GCTs tend to grow faster, have an earlier mean age at the time of diagnosis, and have a lower 5-year survival rate ( Litchfield et al. , 2016 ; Costa et al. , 2017 ; Shen et al. , 2018 ). Identification of molecular signatures to differentiate subtypes of GCTs is therefore crucial for determining prognostication and subtype-based selection of treatment. Thus, a number of studies have been conducted to identify signature genes for each subtype, and transcription factors are promising to be useful as distinct biomarkers for different categories of GCTs ( Alagaratnam et al. , 2011 ; Litchfield et al. , 2017 ). GCTs are thought to originate from FGCs since pluripotency transcription factors are highly expressed in the precursor lesion of GCTs. Master transcription factors for pluripotency, namely OCT4, NANOG, SOX2 and LIN28 (Lin-28 homolog A), are key markers of certain types of GCTs, implicating their roles in maintenance of these malignant cells in the growth of this tumor ( Skakkebaek, 1972 , 2002 ; Looijenga et al. , 2003 ; Cheng et al. , 2004 ; Hart et al. , 2005 ; Hoei-Hansen et al. , 2005 ; Cheng et al. , 2007 ; West et al. , 2009 ; Gillis et al. , 2011 ). Clinically, these pluripotency factors are emerging as diagnostic markers for both testicular and ovarian GCTs ( Gillis e t al. , 2011 ). Immunohistochemistry studies in primary samples have suggested OCT4 and NANOG as sensitive and specific markers for identifying GCTs ( Jones et al. , 2004 ; de Jong et al. , 2005 ; Richie, 2005 ; de Jong and Looijenga, 2006 ; Jung et al. , 2006 ). However, these two transcription factors alone do not provide the specificity necessary to distinguish between seminomatous and non-seminomatous tumors ( Ulbright and Young, 2005 ). Recent gene expression profiling and immunohistochemistry analyses have suggested that the combination of expression patterns of multiple transcription factors may serve as a feature to differentiate seminomatous and subtypes of non-seminomatous GCTs ( Santagata et al. , 2007 ). For example, seminomas are found to be positive for OCT4 and NANOG and negative for SOX2, whereas embryonal carcinomas are positive for all three pluripotency markers. Besides pluripotency transcription factors, other crucial transcriptional regulators of FGC development are also indicated as diagnostic markers for GCTs. For example, the expression pattern of SOX17, a critical regulator of hPGC specification, can also distinguish seminoma from embryonal carcinoma when combined with SOX2 ( Nonaka, 2009 ). Immunohistochemistry of TFAP2C, another essential transcription factor for germ cell development, has also been evaluated for the diagnosis of multiple subtypes of GCTs ( Pauls et al. , 2005 ).

The data underlying this article will be shared with all reasonable requests to the corresponding author.