| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gamete Biology; |
Centre for Human Development, Stem Cells and Regeneration,3 Human Genetics Division,4 and Developmental Origins of Health and Disease Division,5 University of Southampton, Southampton SO16 6YD, United Kingdom
Department of Growth and Reproduction,6 Copenhagen University Hospital, Rigshospitalet, DK-2100 Copenhagen, Denmark
ABSTRACT
NANOG, POU5F1, and SOX2 are required by the inner cell mass of the blastocyst and act cooperatively to maintain pluripotency in both mouse and human embryonic stem cells. Inadequacy of any one of them causes loss of the undifferentiated state. Mouse primordial germ cells (PGCs), from which pluripotent embryonic germ cells (EGCs) are derived, also express POU5F1, NANOG, and SOX2. Thus, a similar expression profile has been predicted for human PGCs. Here we show by RT-PCR, immunoblotting, and immunohistochemistry that human PGCs express POU5F1 and NANOG but not SOX2, with no evidence of redundancy within the group B family of human SOX genes. Although lacking SOX2, proliferative human germ cells can still be identified in situ during early development and are capable of culture in vitro. Surprisingly, with the exception of FGF4, many stem cell-restricted SOX2 target genes remained detected within the human SOX2-negative germ cell lineage. These studies demonstrate an unexpected difference in gene expression between human and mouse. The human PGC is the first primary cell type described to express POU5F1 and NANOG but not SOX2. The data also provide a new reference point for studies attempting to turn human stem cells into gametes by normal developmental pathways for the treatment of infertility.
embryonic, gamete biology, gene regulation, human, human development, human stem cell biology, primordial germ cells, SOX2
From original experiments in mice, embryonic stem cells (ESCs) and embryonic germ cells (EGCs) are, unequivocally, two types of pluripotent stem cell [1]. The former are derived from the inner cell mass (ICM) taken into laboratory culture; the latter are presumed to represent an analogous in vitro conversion of the primordial germ cell (PGC), the diploid precursor of the male and female gametes. A third pluripotent stem cell, the embryonal carcinoma cell (ECC), arises from in vivo malignant transformation of the PGC. Comparative studies of these different cell types and their respective parent cells offer a collective view on the derivation and maintenance of pluripotent stem cells. In addition, improved knowledge of the germ cell lineage can benefit research into the different types of germ cell tumor (GCT), particularly those affecting the testis, which are of increasing frequency in the Western world [2]. Information will also provide a reference point for studies aiming to replicate normal development for the in vitro generation of gametes from human stem cell populations for future fertility treatment.
Studying the development of human germ cells is hampered by restricted access to appropriately aged material and by their intractability in culture [3]. Human embryonic material is inaccessible from 14 days (UK legal limit for in vitro culture of the preimplantation embryo) until potential access to material with appropriate consent at social/voluntary termination of pregnancy (rarely earlier than 4 wk postconception) [4]. Thus, researchers have extrapolated heavily from other species, most notably the mouse, with implicit and, where analysis has permitted, mostly correct assumption of high interspecies conservation [5, 6]. Mouse PGCs become apparent during gastrulation and over the following days migrate via the gut mesentery to the gonadal ridge [7]. This same path seems to be followed in human embryos, as first-trimester material, stained for alkaline phosphatase (AP) activity, identifies germ cells in the gut mesentery and gonad [1]. A few groups, including our own, have taken these cells into culture to generate cells that have been termed human EGCs [3, 8, 9]. In our experience, derivation is defined by acquiring highly proliferative cells positive for AP activity and cell surface antigens, such as SSEA4, TRA1–60, and TRA1–81, commonly used as markers of pluripotent cell types [3]. However, contrary to the equivalent mouse cell type, common experience has indicated a difficulty maintaining this cell state over numerous passages in prolonged culture, suggesting value in further characterizing the human germ cell lineage.
Three of the main transcription factors identified in the ICM and ESC are POU5F1 (previously known as OCT4), NANOG, and SOX2 [10]. They have been suggested to act collaboratively in the promoter region of genes encoding two sets of transcription factors: activating expression of those that maintain pluripotency, while repressing transcription from those that enact differentiation [11]. In this study, we have analyzed the expression of POU5F1, NANOG, and SOX2 during early human germ cell development and compared the data to those acquired from human ESCs and various ECCs. We have also studied the expression in the fetal gonad of many genes described as stem cell restricted [11] and that have been reported to require POU5F1 and SOX2 [12–14] or POU5F1, NANOG, and SOX2 [11] for their expression.
Collection of Human Embryonic and Fetal Material and Germ Cell Tumor Samples
Ethical approval, collection, and staging of human embryonic and fetal material was carried out as described previously, using the Carnegie classification and fetal foot length to provide a direct assessment of gestational age as days or weeks postconception (dpc or wpc) [15–17]. Human preimplantation embryos were obtained with ethical permission and informed consent under a licence from the UK Human Fertilisation and Embryology Authority (RO142). Table 1 details the numbers and ages of human fetal material used in this study. The Regional Committee for Medical Research Ethics in Denmark approved the use of human germ cell tumor material. The tissue samples from adults with testicular neoplasms were obtained directly after orchidectomy and macroscopic pathological evaluation.
|
Immunohistochemistry/Immunocytochemistry and Alkaline Phosphatase Activity Staining
Tissue processing of human embryonic and fetal samples, immunohistochemistry (IHC), immunocytochemistry (ICC), and alkaline phosphatase (AP) activity staining were performed as described previously [3, 17, 18]; primary antibodies used are detailed in Supplemental Table 1 available online at www.biolreprod.org. Adult testicular samples were fixed overnight at 4°C in 4% paraformaldehyde or formalin and subsequently embedded in paraffin. A series of 12 testicular tumors were analyzed by IHC, including classical seminomas and various nonseminomatous tumor components. Either biotin- or fluorescently labeled secondary antibodies were used according to the manufacturer's instructions. Anti-rabbit (1:800), anti-goat (1:300), and anti-mouse (1:100) biotinylated antibodies were from Vector Laboratories. Fluorescently labeled secondary antibodies were fluorescein isothiocyanate (FITC) anti-mouse (1:64) or anti-goat (1:64) (both from Sigma-Aldrich). For biotinylated secondary antibodies, either streptavidin horseradish peroxidase (SA-HRP; 1:200; Vector Laboratories), SA-FITC (1:150; Sigma-Aldrich), or SA-Texas Red (1:200; Vector Laboratories) conjugates were used according to the manufacturer's instructions. Controls omitted primary or secondary antibody. For bright-field immunohistochemistry, the color reaction was developed using diaminobenzidine (Merck) containing 0.1% hydrogen peroxidase (Sigma-Aldrich) with toluidine blue counterstaining or, for the GCT samples, using aminoethyl carbazole substrate (Zymed) counterstained with Mayer hematoxylin.
Protein Preparation, SDS-PAGE Electrophoresis, and Western Blotting
Tissues and cells were rinsed with PBS and treated with ice-cold lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM sodium chloride, 0.5% Triton X-100 [Sigma-Aldrich]) containing a set of protease inhibitors (Complete; Roche Diagnostics) for 30 min with gentle trituration. Lysates were stored at –80°C. Single-dimension SDS-PAGE was carried out vertically in buffer (25 mM Tris-HCl [pH 8.3], 250 mM glycine, 0.1% SDS). Cell lysates containing 15 µg of protein were combined with an equal volume of 2x SDS gel-loading buffer (100 mM Tris-HCl [pH 6.8], 200 mM DTT, 4% SDS, 20% glycerol, and 0.2% bromophenol blue) and heated at 95°C for 5 min before gel loading. Proteins were electrotransferred onto a nitrocellulose membrane (Hybond-C Extra, Amersham Biosciences) in transfer buffer (48 mM Tris-HCl [pH 8.3], 39 mM glycine, 20% methanol). Blocking of nonspecific binding sites was carried out by immersion in PBS containing 5% nonfat powdered milk and 0.1% Tween-20 (Sigma-Aldrich) for 1 h at room temperature. Primary antibodies were diluted in blocking buffer and incubated with the membrane overnight at 4°C. Membranes were washed three times in PBS containing 0.1% Tween-20 and then incubated with peroxidase labeled anti-rabbit antibody for SOX2 (Amersham Biosciences; 1:50 000) or anti-goat antibody for POU5F1 (Sigma-Aldrich; 1:200 000) in blocking buffer for 1 h at room temperature. β-actin was detected by 1-h incubation with mouse anti-β-actin peroxidase conjugated antibody (Sigma-Aldrich; 1:50 000) at room temperature. Membranes were washed three times in PBS/0.1% Tween-20 and once in PBS, and developed using advanced enhanced chemiluminescence reagents (Amersham Biosciences) according to the manufacturer's instructions.
Total RNA was isolated from tissues using Tri-Reagent (Sigma-Aldrich) and cDNA synthesized from 1 µg per sample with Superscript III (Invitrogen). Wherever possible, intron-spanning primer pairs were designed (Supplemental Table 2 available online at www.biolreprod.org). Negative (water and –RT) and positive (genomic) control reactions were performed concurrently.
Human Germ Cell, ESC, and ECC Culture
Human gonad-derived cells were processed, plated, and cultured as described previously [3]; in total, 25 cultures were initiated, as detailed in Table 1. The hECC lines NTERA-2 (clone D1 [ECCAC], TERA-1, and 2102Ep [both gifts from Peter Andrews, University of Sheffield] were cultured in Dulbecco modified Eagle medium (DMEM; PAA Laboratories) containing 10% fetal calf serum (PAA Laboratories) and 1% penicillin/streptomycin (Invitrogen) in 5% CO2 at 37°C. Human ECCs were passaged using glass beads (VWR International) and maintained at high density. The hESC lines HUES1 and HUES7 were obtained under MTA from Harvard University [19] and maintained in KO-DMEM containing 1% penicillin/streptomycin, 1% Gluta-MAX, 1% nonssential amino acids, 10% KO-Serum Replacement (all from Invitrogen), 10 ng/ml FGF2 (Peprotech), and 0.1% 2-mercaptoethanol (Sigma-Aldrich) in 5% CO2 at 37°C. Human ESCs were cultured on mitotically inactivated mouse embryonic fibroblasts and passaged using 0.05% Trypsin/EDTA (PAA Laboratories).
Germ Cell Expression of SOX2 Differs Between Human and Mouse
The human fetal gonad is apparent as a distinct structure from 32 dpc with sex determination marked by the expression of SRY and SOX9 at 
41–44 dpc in the male [18]. Thereafter, sex cords comprised of Sertoli cells and germ cells become increasingly apparent within the testis from 48 dpc during late embryonic (up to 56 dpc) and early fetal (thereafter) development [18]. POU5F1, NANOG, and SOX2 are known as three critical transcription factors in human ESCs [11] (Fig. 1A). Nuclear POU5F1 and NANOG proteins were clearly detected within germ cells in the gonad of the embryonic ovary and testis at Carnegie stages 19–22 (48–54 dpc; Fig. 1, B and C). In sharp contrast, SOX2 was not detected within the human female or male gonad either prior to sex determination or later during the first trimester (ranging from 48 to 73 dpc for both sexes; Table 1, Fig 1, B and C; data not shown). The expression of some genes, such as Ddx4, is altered on the arrival of the PGC to the gonadal ridge in mice [20]. Therefore, we also studied PGCs within the gut mesentery at 48 dpc, a location consistent with their migration from yolk sac wall to gonadal ridge. Nuclear POU5F1 and NANOG were again clearly detected; however, SOX2 was not (Fig. 1D). The validity of the SOX2 antibody within the same human embryonic sections was confirmed by clear staining of the neuroprogenitors in the spinal cord and stomach epithelial cells (Fig. 1, E and F). Sox2 transcripts have been demonstrated in the gonadal ridge of mouse embryos [6]. Consistent with this finding, nuclear SOX2 protein was present in primordial germ cells within the mouse embryonic testis and ovary (E13.5 and E12.5, respectively; Fig. 1, G and H). Nuclear SOX2 was also clearly detected within the ICM of mouse blastocysts (Fig. 1I) [21]. In human blastocysts, SOX2 detection was more diffuse but included nuclear localization within the ICM.
|
The SOX2-negative human germ cells included ones that were proliferative in situ as marked by dual immunoreactivity for POU5F1 and the proliferative marker MK167 (recognized by the Ki67 antibody; Fig. 2). There were more of these double-stained cells in the embryonic ovary at 54–56 dpc than in the embryonic testis of the same age, consistent with male PGCs entering mitotic arrest following testicular cord formation.
|
These data, demonstrating an interspecies difference in PGC SOX2 expression, were further validated by immunoblotting of protein isolated from fetal gonads (Table 1, Fig. 3). SOX2 was present as a 35-kDa band in protein isolates from mouse gonads at E14 but not from the corresponding human organs. POU5F1 detection is shown as a positive control for the presence of PGCs.
|
Redundancy Is Unlikely to Compensate for Lack of SOX2 in Human Germ Cells
One potential explanation for this species difference is redundancy for SOX2 within group B of the SOX gene family, which comprises SOX1, SOX2, SOX3, SOX14, and SOX21. By RT-PCR for 35 cycles, SOX2, SOX3, SOX14, and SOX21 transcripts were detected in both human ESCs and NTERA-2 D1 ECCs (Fig. 4). The weak identification of SOX1 in ESCs may represent low-level expression per se or, feasibly, minor spontaneous ectodermal differentiation as is commonly encountered in human ESC culture. However, transcripts for all SOX group B genes were undetected in four samples of human testes collected during the late embryonic and early fetal periods (51–61 dpc; Fig. 4, Table 1). In five ovarian samples ranging from 51 to 61 dpc, a very faint band was just discernible for SOX1 and SOX21 (Fig. 4, Table 1). Only SOX14, encoding a putative repressor of SOX2 function [22], was clearly detected within the ovary after amplification for 35 cycles (Fig. 4). The failure to detect SOX2 transcripts convincingly within these testicular and ovarian samples following extensive PCR cycles corroborates the protein data from IHC and immunoblotting (Figs. 1 and 3).
|
SOX2 in Germ Cell Tumors and Embryonal Carcinoma Cell Lines
The ECC represents malignant transformation of the PGC. Our findings led us to interrogate SOX2 expression in different GCTs in vivo and in vitro. Previously, SOX2 transcripts have been reported in a "germ cell carcinoma" sample [6]. In agreement with data from others [23–25], we did not detect SOX2 in samples of seminoma, a nullipotent human GCT; however, it was clearly visualized as nuclear protein in pluripotent nonseminomatous embryonal carcinoma. Both seminomas and pluripotent GCTs expressed nuclear POU5F1 and NANOG (Fig. 5, A and B). In contrast to NTERA-2 D1 cells, once taken into culture some embryonal carcinomas have yielded nullipotent cell lines that are no longer capable of differentiation to derivatives of all three germ layers (e.g., TERA-1 and 2102Ep cells). SOX2 was clearly expressed equivalently in both pluripotent and nullipotent hECCs localizing indistinguishably to the nucleus in all lines tested (Fig. 5, C–F).
|
Human Germ Cells Persisting in Culture In Vitro Remain SOX2 Negative
We have previously assigned germ cell cultures as "poorly proliferative" (PP) or "vigorously proliferative" (VP), according to their growth characteristics, with the latter taken as indicative of conversion to the EGC state [3]. Akin to ESCs, human EGCs demonstrate expression of the nuclear transcription factor POU5F1 and AP activity [1]. However, SOX2, required for the maintenance of pluripotent ESCs [21], was not expressed in the human germ line in situ. These findings led us to question whether SOX2 expression was induced in culture. Numerous human PGC cultures were established, and samples of both PP and VP/EGC cultures were analyzed. POU5F1 protein and POU5F1 transcripts and AP activity served as positive controls for the presence of the germ cell lineage (Fig. 6). Neither SOX2 protein nor SOX2 transcripts were detected by immunoblotting or RT-PCR, respectively, in either early, PP, or VP/EGC cultures.
|
SOX2 Appears Dispensable for the Expression of Genes Previously Identified to Require OCT-SOX Interaction or Cooperative POU5F1-SOX2-NANOG Function
Given the lack of SOX2 in human germ cells yet the presence of both POU5F1 and NANOG, we investigated a selection of genes recognized as targets of these critical transcription factors. POU5F1 and SOX2 are known to act cooperatively in up-regulating FGF4 [12], UTF1 [13], and FBX15 [14]. However, despite the absence of SOX2, UTF1, and FBX15 transcripts were detected in the human fetal ovary and testis. In contrast, FGF4 was not detected (Fig. 7A, Table 1). A further set of 13 genes were analyzed that are expressed in ESCs when the respective promoter regions are bound by POU5F1, NANOG, and SOX2 [11]. All 13 stem-cell-restricted transcripts were detected in fetal gonads after RT-PCR for 35 cycles albeit for DKK1, SET, ZIC, and STAT3 more weakly in the human fetal ovary (Fig. 7B, Table 1).
|
Previous studies have revealed highly concordant gene expression profiles across different pluripotent stem cells, the cells of the ICM and PGCs [5, 24, 26]. Here, our analysis of the human germ cell lineage was borne of recognized difficulties in the long-term maintenance of human germ cell cultures [1]. We hypothesized that the starting human PGCs lack a critical factor(s) or fail to induce one in in vitro culture that is required for maintained self-renewal and pluripotency. Contrary to predictions from other studies [6] or previously presumed conservation [5], we have identified that SOX2/SOX2 expression, identified as fundamental to stem cell pluripotency [10], was not detected in human PGCs as either transcripts or protein. In contrast, SOX2 was present in the human ICM, ESCs, and various ECCs. Amounts of SOX2 and POU5F1 are known to alter in cells derived from different stages of mouse development; mouse germ-line stem cells, which are cultured spermatogonial stem cells of highly restricted developmental potential, express abundant Sox2 and Pou5f1 transcripts but limited corresponding protein [27]. However, mouse PGCs clearly express SOX2 [6 and herein]. At equivalent stages of human development, human PGCs are distinguished by expressing POU5F1 and NANOG but lacking SOX2.
Genome sequencing projects have revealed remarkable conservation across seemingly diverse species. Therefore, a likely explanation underlying cross-species differences in phenotype is altered regulation of gene expression. Thus, the regulation of SOX2 in human PGCs must differ from that in ESCs (and presumably the epiblast) where an upstream enhancer is bound by POU5F1 and a further downstream composite element by both POU5F1 and SOX2 [28, 29]. The high interspecies sequence conservation of these elements and the wider SOX2 locus suggest that alternative factors, such as epigenetic mechanisms, may be responsible for divergent SOX2 expression [30, 31]. Although the earliest germ cell lineage was inaccessible in our human specimens (i.e. prior to those cells in the gut mesentery at 6–7 wk postconception), our data imply that SOX2 expression must cease either during germ cell specification from cells of the SOX2-positive epiblast or soon afterward. As reports unfold of human ESCs differentiated toward functional germ cells [32], it will be interesting to observe whether SOX2 disappears in mimicry of normal human development. Redundancy among group B SOX family members would limit functional consequences of undetectable SOX2 for the human PGC. This appears unlikely from our data. Only SOX14 transcripts were detected clearly in fetal ovary samples and only after relatively high numbers of PCR cycles. Furthermore, to date, SOX14 has been recognized to counteract rather than substitute for the role of SOX2 [22].
The collective data infer that, alongside POU5F1 and NANOG, SOX2 is necessary but not sufficient for pluripotency. The epiblast of SOX2 null mice fails to progress in vivo, and SOX2 null blastocyst outgrowths fail to generate ESCs in vitro [21]; however, whereas mouse PGCs give rise to robust pluripotent EGC lines, the SOX2-positive PGCs themselves are nullipotent [33]. Similarly, although previous data on nullipotent human ECC lines were limited largely to analysis of transcripts [24], our data reveal relatively equivalent quantities of correctly localized nuclear SOX2 protein regardless of a capacity for broad differentiation. SOX2 expression, as demonstrated here and by others [23–25], identifies patients with a pluripotent GCT phenotype rather than a nullipotent seminoma. Discovery of ways to induce SOX2 expression in cultured human germ cells might allow improved models of human GCTs. For instance, with the initiation of neoplastic transformation of human germ cells proposed to occur in utero [34], the lack of SOX2 in proliferative PGCs may explain the predilection for SOX2-negative testicular seminomas in humans compared to mice.
Identifying the consequences of missing SOX2 offers comparative insight into its role when present in other cell types. Where coexpressed, SOX2 is considered to act cooperatively with POU5F1 and NANOG in regulating critical cohorts of target genes. From bioinformatic studies, their collective association with gene promoters is proposed to induce the expression of genes that determine self-renewal, while repressing those associated with differentiation [11]. Our data suggest that SOX2 is dispensable in the former function, as, without its detection, all 13 of our arbitrarily selected "ESC-restricted" target genes from Boyer and colleagues, as well as UTF1 and FBX15, were expressed in the fetal ovary and testis. Indeed, most recently, evidence has been published that the major role for SOX2 in ES cells is in maintaining POU5F1 expression [35]. This implies that the regulation of both POU5F1 and SOX2 is different in human germ cells—SOX2 because of its apparent absence and POU5F1 because of its independence from SOX2 protein. In contrast, our data do support a pivotal role for SOX2 in promoting FGF4 expression [12, 36]; FGF4 transcripts were undetected in the SOX2-negative germ cell lineage. The difference between FBX15 (detected) and FGF4 (not detected) is interesting, as both have been shown hypermethylated and correspondingly absent in mouse germ-line stem cells [27].
Finally, it has been discussed whether, on the basis of gene expression profiles, the closest equivalent of the human ESC is an early germ cell and, indeed, whether ESC derivation arises from an early germ cell phenotype [37]. On the basis that the earliest human germ cells detected are SOX2 negative, this hypothesis appears unlikely. In conclusion, we have discovered a fundamental difference in gene expression between human and mouse, demonstrating the limitations of interspecies extrapolation. The data highlight SOX2 as an important transcription factor for further investigation in attempts to understand the relationship between human PGCs, GCTs, and the derivation, self-renewal, and pluripotency of human EGCs. The information is also instructive in attempts to generate human gametes from stem cell sources for ambitious fertility treatments.
ACKNOWLEDGMENTS
The authors are grateful for the gift of human cell lines from Professor Peter Andrews and the collection of human fetal material by Anne Chad and colleagues at the Quays Clinic and the Princess Anne Hospital.
FOOTNOTES
1Supported by a UK Department of Health Clinician Scientist award (to N.A.H), Wellcome Trust project grant GR074320MA (to N.A.H.), International Research Mobility Award from the Worldwide Universities Network (to R.M.P.), and the Danish Cancer Society (to E.R.M.). R.M.P. is the recipient of an MRC Ph.D. studentship in stem cell research. L.T. is an MRC/Juvenile Diabetes Research Foundation (JDRF) Career Development Fellow in stem cell research. The authors declare no conflicts of interest. 
Correspondence: 2Neil A. Hanley, Human Genetics Division, Duthie Building, Mailpoint 808, Southampton General Hospital, Tremona Rd., Southampton SO16 6YD, U.K. FAX: 44 0 23 8079 4264; e-mail: N.A.Hanley{at}soton.ac.uk
Received: 25 October 2007.
First decision: 20 November 2007.
Accepted: 9 January 2008.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  A. J. Childs, P. T.K. Saunders, and R. A. Anderson Modelling germ cell development in vitro Mol. Hum. Reprod., September 1, 2008; 14(9): 501 - 511. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
