Signaling for germ cells

Establishment of the germ cell lineage

What is the evidence? The earliest that primordial germ cells have been identified in the mouse is midway through gastrulation, ∼7.25 days after fertilization (Ginsburg et al. 1970). Recognizable by their high alkaline phosphatase activity, they are seen as a cluster of cells located in the extraembryonic region posterior to the primitive streak, and can subsequently be tracked from that location, along their migratory route and into the future gonads. All attempts to identify germ cells or germ cell determinants (e.g., germplasm, ‘nuage’) earlier in development have failed. Experiments in which a genetically marked cell was introduced into the embryo either at the 4-cell (Kelly 1977) or at the mid- or late-blastocyst stage (Gardner 1977) established that cells with primordial germ cells among their descendants also gave rise to somatic cells, thus no exclusive germ cell lineage existed in the preimplantation embryo.

More recently, this approach has been extended to the postimplantation period by Lawson and Hage (1994). Immediately before gastrulation and during the early stages (6.0 and 6.5 days after fertilization), the mouse embryo can be visualized as a thick-walled cup of tissue (the epiblast or embryonic ectoderm), which will give rise to the entire fetus as well as contributing to some of the placental membranes. On top of this is inverted a second thick-walled cup of tissue (the extraembryonic ectoderm, derived from the trophectoderm) which will give rise to the main part of the placenta. Both cups are enclosed in a thin bag of primitive endoderm. Lawson and Hage removed the embryos from the uterus, then injected a single epiblast cell in each embryo with a fluorescent lineage marker, so that after culturing for some 40 hours they could identify the descendants of the injected cell and ascertain their fate. They found that the only cells giving rise to primordial germ cells (recognizable again by their high alkaline phosphatase activity) were those located around the rim of the epiblast cup (i.e., proximal), immediately adjacent to the inverted extraembryonic ectoderm cup. At 6.0 days, the germ cell progenitors were distributed all round the rim; by 6.5 days they were found nearer to the side of the cup where the primitive streak had formed. This entire proximal population of cells moves, perhaps under pressure from cell proliferation, through the primitive streak and upwards, into the territory of the extraembryonic ectoderm. Some of the cells settle down, to form the initial cluster of primordial germ cells; others nearby will give rise to the allantois; the majority continue to migrate, forming extraembryonic mesoderm and separating the cavities of the two cups.

Crucially, Lawson and Hage found that none of their injected cells gave rise only to germ cells, proving that even at this late stage no exclusive germ cell lineage had been established. By counting the clonal descendants and the marked and unmarked germ cells, they were able to carry out a clonal analysis. This indicated that germ cell lineage restriction was taking place in a group of some 45 cells at ∼7.2 hours after fertilization, by which time the migrating cells would have reached the extraembryonic location in which the initial cluster of primordial germ cells had earlier been seen.

Ancestors or neighbors?

So far so good. The germ cell lineage was founded in the place and at the time when primordial germ cells first could be identified. But were those cells programmed to gather in that location and become germ cells, as a result of some asymmetric segregation of determinants at each of the many cell divisions since fertilization? Given the extensive cell mingling that occurs in the epiblast before gastrulation (Gardner and Cockcroft 1998), the likelihood of such a happening seemed small. It was reduced to zero by a recent experiment, simple in design but very demanding in execution (Tam and Zhou 1996).

Tam and Zhou knew from the work of Lawson and Hage that the ancestors of the germ cells were normally located in the proximal region of the epiblast, around the rim of the epiblast cup. In contrast, cells at the distal tip of the epiblast, at the bottom of the cup, normally give rise only to neurectoderm and surface ectoderm. Using 6.0 and 6.5 day transgenically marked donor embryos and unmarked host embryos, they carried out an in vitro transplantation experiment, grafting small groups of marked cells both homotypically (proximal to proximal, distal to distal) and heterotypically (proximal to distal, distal to proximal). They cultured the grafted embryos for two days, then processed them for β-galactosidase to identify the donor cells, and alkaline phosphatase activity to identify primordial germ cells. The results were clear-cut. Cells placed in the proximal rim of the epiblast cup were capable of giving rise to germ cells, whether they had been taken from the proximal or from the distal part of the donor epiblast; conversely, germ cells never appeared in the bottom of the cup, even when donor cells from the proximal region had been transplanted there. The cell types that differentiated depended on where the grafts were placed, not where they had been taken from.

Signals—but from where?

So germ cell status is not ‘achieved’ in the mouse embryo by virtue of some C. elegans-like segregation of cytoplasmic determinants, but rather is ‘thrust upon’ certain cells by means of local signals. But when and where do these signals occur, and what do they consist of? Suspicion must fall first on the extraembryonic cluster location, where the movement of the germ cell progenitors is halted and germ cell specification takes place. This would time the signal for ∼7 days after fertilization. The requirement for progenitors to have come from the proximal region of the epiblast cup could then arise merely because these are the cells that move through the primitive streak and into the extraembryonic region during gastrulation.

The first indication that an earlier signal might be involved came from the observation (Y. Masui and T. Yoshimizu, pers. comm.) that proximal epiblast tissue taken at 6.5 days, disaggregated, and then cultured for several days, could give rise to cells showing high alkaline phosphatase activity, resembling primordial germ cells. These cells never could have been exposed to any cluster-location signals. Cultures derived from other parts of the epiblast never yielded any alkaline phosphatase-positive cells. When proximal epiblast was taken from earlier stages, the putative germ cells only appeared if some extraembryonic ectoderm was included in the cultures.

So would an earlier signal suffice? Or are there two signals? Or more than two? Our understanding of the situation has been greatly enhanced by the important paper in this issue by Lawson et al. (1999).

Bone Morphogenetic Protein 4 (BMP4) is a member of the TGFβ superfamily of intercellular signaling proteins. Its role in development extends far beyond the morphogenesis of bone. Most mouse embryos homozygous for a null mutation in the Bmp4 gene die in the early stages of gastrulation; but Lawson et al. (1999) found that some survive long enough to show that, although the proximal epiblast cells move through the primitive streak and into the extraembryonic region, no primordial germ cells are seen in the cluster location or anywhere else, nor does an allantois develop. This is the first report of a total inhibition of primordial germ cell formation.

Bmp4 is expressed before gastrulation in the inverted cup of extraembryonic ectoderm that abuts onto the proximal rim of the epiblast cup. In chimeras made between ES cells (which contribute to the epiblast but not to the extraembryonic ectoderm) and homozygousBmp4-null embryos (which therefore form the entire extraembryonic ectoderm), neither germ cells nor allantois developed. The authors conclude from these chimera studies that some signal arising from the expression of Bmp4 in the extraembryonic ectoderm is an essential requirement for the subsequent establishment of the germ cell lineage and for the formation of the allantois. Whether or not the secreted molecule BMP4 itself constitutes the signal remains to be investigated; but the absolute requirement forBmp4 to be expressed is unquestionable.

Mice heterozygous for the Bmp4-null mutation are fertile, butLawson et al. (1999) report that the number of primordial germ cells in heterozygous embryos lags behind that in their wild-type littermates: the germ cell population in the heterozygotes increases at the same rate as in the controls, but for any given somite number it is reduced ∼50%. A similar though smaller reduction in germ cell number was reported for W^e heterozygotes by Buehr et al. (1993). Lawson et al. (1999) interpret their findings as indicating a smaller founding population of germ cells in the heterozygotes, rather than a delay in germ cell allocation or proliferation, or an early loss of germ cells. A smaller founding population could indicate a weaker signal in the heterozygotes, suggesting that germ cell allocation is dosage dependent. The allantois, however, is of normal size in the heterozygotes.

One signal or two?

Given that a Bmp4-dependent signal from the extraembryonic ectoderm to the proximal epiblast cells is necessary for germ cell development, is it sufficient? Lawson et al. (1999) discuss both one and two signal models.

On a one-signal model, one would have to assume that only a minority of the proximal epiblast cells receiving the Bmp4-dependent signal, perhaps those above a certain threshold level, became programmed to give rise to allantois and germ cells, whereas the majority ignored the signal and continued to migrate as extraembryonic mesoderm. Of the responding subset, ∼45, perhaps those receiving the highest levels of signal, would settle down in the cluster region as the germ cell founder population, whereas an unknown number would settle nearby and subsequently give rise to the allantois. The cells would halt in the cluster region not because of the properties of the region, but because it represented the limit of the time or distance or number of cell divisions for which they (or their pioneers) had been programmed to travel. Given that only a small fraction of the descendants of any germ cell progenitor in the epiblast at 6.0 days actually contributes to the germ cell population (Lawson and Hage 1994), it would be necessary to postulate that the signal continued to act on the migrating cells until shortly before germ cell allocation.

More attractive seems a two-signal model (Fig. ), with a Bmp4-dependent signal from the extraembryonic ectoderm sensitizing the proximal epiblast cells, and a further signal or signals of unknown origin capturing the allantois progenitors and the germ cell founder population from the sensitized cells moving past towards an extraembryonic mesoderm fate. Lawson et al. (1999) postulate that a single population of epiblast cells is allocated to both the allantois and the germ cell lineage. To explain the normal-sized allantois in Bmp4-null heterozygous embryos, they suggest that when the founder population is reduced in number, the allantoic rudiment would retain its normal complement of cells since an allantois is required for further development, while the less essential germ cell population could be correspondingly decreased. It may be, however, that the rapidly proliferating allantois is capable of a considerable degree of compensatory growth (Snow et al. 1981), unlike the more slowly dividing germ cell population (Lawson and Hage 1994). In that case one could consider separate signals for allantois and germ cells.

From what tissues could such a signal or signals emanate? We know little of the siting and make up of the allantoic primordium: more detailed investigation of this essential organ is badly needed. What of the germ cell cluster? One side is bounded by primitive extraembryonic visceral endoderm; the other side is bounded by extraembryonic mesoderm at the time when the primordial germ cells can first be identified (Ginsburg et al. 1990), but the first epiblast cells to move into the extraembryonic region would be in contact with extraembryonic ectoderm. There are thus three possibilities. Any effective signal emanating from extraembryonic ectoderm would need to be strictly regulated in time and space to target the first cells coming through the posterior primitive streak, leaving later immigrants to become extraembryonic mesoderm. It would therefore be quite distinct from the earlier broadBmp4-dependent signal. Epiblast-derived extraembryonic mesoderm has been shown by the chimera experiments of Lawson et al. (1999) to be incapable on its own of inducing germ cell formation; the postulated second signal, however, would be acting on cells already sensitized by exposure to the first signal. Again it would need to be highly localized, in contrast to Bmp4, which Lawson et al. (1999) report to be widely expressed in the extraembryonic mesoderm. The third possibility is the primitive visceral endoderm, the thin bag of tissue that surrounds both the epiblast and the extraembryonic ectoderm. The extraembryonic germ cell cluster lies closely apposed to this endodermal layer. Anterior patterning in the mouse has been shown to depend on signals coming from the embryonic visceral endoderm (Thomas and Beddington 1996) but nothing is known of gene expression in the extraembryonic visceral endoderm posterior to the primitive streak.

Are mammals so strange?

The establishment of a germ cell lineage must be about the most fundamental issue ever to have faced the Metazoa throughout their long evolutionary history. Once evolution has devised a system based on unequal distribution of germ cell determinants within the fertilized egg and early embryo, a system that appears to work well for C. elegans and Drosophila, and which has extended into vertebrate evolution, at least in zebrafish and frogs, is it not strange that mammals should do things so differently?

Mammals may not be alone. Although frogs and toads (Anura) have visible ‘germ plasm’ present from the oocyte stage onwards, their cousins the newts (Urodela) lack any obvious germ plasm in the early embryo. Germ cells are first seen in the ventrolateral region during gastrulation, and were reported by Nieuwkoop and Satasurya (1979) to be induced in the mesoderm. EvidentlyUrodeles warrant further investigation, using modern methods. In birds, the evidence remains inconclusive (Karagenc et al. 1996).

But if mammals do turn out to be unique, is that so surprising? The first and most crucial challenge in mammalian development is to achieve complete and stable implantation in the uterus of the mother. To that end, the first cell lineages to differentiate in the mammalian embryo are trophectoderm and primitive endoderm, both of which contribute exclusively to extraembryonic structures (placenta and other extraembryonic membranes). Until the basic life-support systems to nourish and protect the future fetus have been laid down, no further differentiation takes place. In a sense the dramatic increase in proliferation rate of the epiblast that heralds the completion of implantation and the onset of gastrulation is analogous to early embryonic development in other animals.

In Drosophila, the unequal distribution of determinants within the oocyte is induced by signals emanating from the surrounding somatic tissues, which make no material contribution to later development. In mammals, the two extraembryonic lineages (trophectoderm and primitive endoderm) make no material contribution to fetal development, but they may turn out to be the source for many of the signals that regulate that development. The distinction could be regarded as essentially one of timing.