Closing the circle of germline and stem cells: the Primordial Stem Cell hypothesis

Germline determination is believed to occur by either preformation or epigenesis. Animals that undergo germ cell specification by preformation have a continuous germline. However, animals with germline determination by epigenesis have a discontinuous germline, with somatic cells intercalated. This vision is contrary to August Weismann’s Germ Plasm Theory and has led to several controversies. Recent data from metazoans as diverse as planarians, annelids and sea urchins reveal the presence of pluripotent stem cell populations that express germ plasm components, despite being considered to be somatic. These data also show that germ plasm is continuous in some of these animals, despite their discontinuous germline.

Here, based on recent molecular data on germ plasm components, I revise the germline concept. I introduce the concept of primordial stem cells, which are evolutionarily conserved stem cells that carry germ plasm components from the zygote to the germ cells. These cells, delineated by the classic concept of the Weismann barrier, can contribute to different extents to somatic tissues or be present in a rudimentary state. The primordial stem cells are a part of the germline that can drive asexual reproduction.

Molecular information on the expression of germ plasm components is needed during early development of non-classic model organisms, with special attention to those capable of undergoing asexual reproduction and regeneration. The cell lineage of germ plasm component-containing cells will also shed light on their position with respect to the Weismann barrier. This information will help in understanding the germline and its associated stem cells across metazoan phylogeny.

This revision of the germline concept explains the extensive similarities observed among stem cells and germline cells in a wide variety of animals, and predicts the expression of germ plasm components in many others. The life history of these animals can be simply explained by changes in the extent of self-renewal, proliferation and developmental potential of the primordial stem cells. The inclusion of the primordial stem cells as a part of the germline, therefore, solves many controversies and provides a continuous germline, just as originally envisaged by August Weismann.

Keywords: Weismann barrier, Germ plasm, Nuage, Epigenesis, Preformation, Germline, Stem cells, Germ cells, Regeneration, Neoblasts

The concept that the germline is distinct from the soma was introduced in the late 19th century by August Weismann [1]. In his Germ Plasm Theory he emphasized the continuity of the germline, which is both totipotent and immortal. The germ plasm, a collection of genetic determinants found in egg cells (Table 1), is inherited by the cells that will later give rise to the gametes of the new organism. Thus, in Weismann’s theory, the germ plasm is present in a continuous line of cells that expands from the zygote of one generation to the zygote of the next generation. This line of cells is known as the germline (Table 1). The germline is capable of giving rise to all somatic cells, while it cannot be affected by them. The concept of the so-called Weismann barrier derives from this vision, stating that hereditary information can transit from the germline to the soma, but not vice versa and, therefore, the germline is not affected by any acquired mutation in the somatic cells (Table 1).

Later studies on the embryology of different model organisms revealed the existence of metazoans to which the concept of germline continuity can be hardly applied. Ultimately, this led to the proposal that two different paths of germline specification exist: preformation and epigenesis (Table 1) [2]. The organisms that specify their germline by preformation (Figure 1A) present germ plasm components in their zygotes which are inherited by the primordial germ cells (PGCs, Table 1) first and then by the germ cells (GCs, Table 1). Therefore, both germline continuity and germ plasm continuity are observed in this model, as proposed by Weismann. The Weismann barrier (Figure 1A) can be easily imagined in these organisms, since any mutation in their somatic tissue will eventually disappear with the death of the somatic cell or ultimately with the death of the organism. The germline will, therefore, stay clear of somatic mutations. However, the majority of metazoans are believed to specify their germline cells by epigenesis (Figure 1B), a process whereby epigenetic signals specify a group of somatic cells, sometimes called the presumptive primordial germ cells (pPGCs), to become the germline [2], of which the so-called PGCs are the first cells. In this model, the continuity of the germline is broken in early embryonic development, as somatic cells that exist after the zygote stage are induced to form the germline of the organism (see [3]). Therefore, in this mode of development somatic mutations can trespass into the germline, eliminating the concept of the Weismann barrier in these organisms.

It has become clear that the germ plasm is composed of several proteins and mRNAs, including for instance Vasa, Nanos and Piwi, which are typically organized in electron-dense granules. These granules, often known by different names in different species but collectively known as nuage (Table 1) and their molecular components are found in the egg cells and the zygotes of many species, but also in the germ cells. Only recently, the study of the expression patterns of several germline components in non-classical model organisms (collectively reviewed in [4]) has added some extra controversy to this old question [3]. In particular, several germ plasm components have been found to be expressed in tissues and cells classically considered to be somatic, such as planarian neoblasts [5-8], the mesodermal posterior growth zone (MPGZ) in polychaetes [9,10], the interstitial stem cells (I-cells) of hydrozoan cnidarians [11-13], and the small micromere lineage in echinoderm embryos [14]. Together, these findings have led to the proposal of a conserved germline multipotency program (GMP, Table 1) [4] that operates in both germline cells and somatic multipotent cells [15]. In most of these animals the germ plasm seems to be continuous, flowing from the zygote to these multipotent cells and forth to the GCs (Figure 1C). The classical model of preformation and epigenesis is, therefore, challenged, with a growing number of animals in which a germ plasm continuity is observed at the molecular level [8-10,13,14,16-23], despite their classification as epigenesis (Figure 1C). The specification of these multipotent cells and the germline often occurs in parallel, as has been shown in Platynereis dumerilii[10,19], generating controversy on the mechanism of this specification.

The root of the controversy, however, can be traced back to the original definition of the germline. This definition never considered the wide variety of metazoans capable of undergoing asexual reproduction and with high regenerative capacities. Here, I revisit the original definition of germline and the Weismann barrier and revise accordingly the concept of germline in order to include asexually reproducing metazoans. Out of this revision, and after careful inspection of the available data, a model that accurately describes the germline specification of most described organisms emerges. The existence of a highly conserved population of zygote-derived multipotent cells is deduced from this model. I call these cells the primordial stem cells (PriSCs), which have their representatives in planarian neoblasts, the 4d lineage of annelids and molluscs, the small micromere lineage of sea urchins, the inner cell mass (ICM) of the mammalian embryos, and the precursors of the PGCs of model organisms whose germline specification has been classically described to proceed by preformation. Only a few characteristics of these PriSCs change in each of the classes here described, hence providing an evolutionarily dynamic framework that explains the multiple transitions observed, even within animals belonging to the same phylum, across evolution.

Freshwater planarians are free living platyhelminthes that often undergo both sexual and asexual reproduction by fission and present remarkable regenerative capacities. This is largely based on a population of undifferentiated stem cells present in their adult stage, the so-called neoblasts [24-26]. Neoblast cells present striking similarities to GCs, both at the morphological and molecular level. The presence within these cells of a kind of perinuclear RNA granules, called chromatoid bodies, constitutes a remarkable similarity with the germline granules, collectively known as nuage, that are known to be present in the germline of almost any studied metazoan [27,28]. The molecular components of the germline granules, which constitute what has been called the GMP [4] and include, for instance, Piwi, Vasa, and Tudor proteins, are also expressed in planarian neoblasts [5-8]. Despite these similarities, neoblasts are capable of giving rise to all somatic tissues [29].

It is currently unknown if or how planarians specify their germline during embryonic development, basically due to the inaccessibility of the highly derived triclad embryonic development [16,30] (reviewed in [31]). Briefly, the zygote of Schmidtea polychroa cleaves into blastomeres that are dissociated and embedded in a yolk syncytium. Embryonic cells (blastomere derivatives) proliferate in this syncytium and also give rise to the somatic structures of what is believed to be a cryptic larval stage [16,30]. Later, a wave of differentiation of the embryonic cells gives rise to the adult tissues. The first cells resembling neoblasts arise after this stage [8]. The embryo hatches as a juvenile worm with all adult structures except the gonads, which develop later. PGCs, however, have been detected in late planarian embryos [32], although this point remains controversial [33].

However, it has been shown that, regardless of their origin, neoblast cells always retain the potential of regenerating GCs when lost [32-34] and, hence, their specification has been considered to proceed by epigenesis [2]. The few existing data about the embryonic expression patterns of GMP genes point, however, to a very early expression of these genes in cleaving blastomeres, as has been described for the Tudor homologue Spoltud-1 [8], which is also localized in nuage-like granules present in these cells [16]. Embryonic germline specification, therefore, occurs most likely from a population of GMP-positive neoblast-like pluripotent stem cells. Therefore, the germ plasm, understood as both the presence of nuage material and GMP components, seems to be continuous in planarians.

The specification of the germline in planarians, then, likely follows the model proposed in Figure 2. A population of GMP-positive nuage-containing undifferentiated and pluripotent stem cells is derived from the early blastomere divisions, and the GCs are singled out from it either during embryonic development or later. This embryonic population is maintained in the adult organism, the neoblasts, and is capable of repeatedly regenerating the GCs (Figure 3A). The GCs of the adult are responsible for sexual reproduction but asexual reproduction (and regeneration) is driven by the population of neoblast cells.

The classical germ-to-soma boundary in planarians is indicated by the dotted blue line in Figure 2. Neoblast cells have been classically considered to be somatic stem cells, since they give rise to somatic tissues. Therefore, the germline only contains the GCs and their derivatives, the gametes and the zygote. However, while the germ plasm, evidenced by GMP components and nuage, seems to be continuous in planarians, this line breaks the continuity of the germline and the germline cycle here described, and I believe that this breakage is the root of the controversies explained above. Therefore, I propose a revision of the germline concept, in order to include the neoblasts, as indicated by the dotted red line in Figure 2. This proposal is made on the following basis. First, neoblasts resemble GCs both morphologically and at the molecular level and, therefore, might function in a similar way. Second, neoblasts have the ability to drive reproductive processes; both by asexual reproduction and by generating GCs. Third, the ability to give rise to somatic tissue cannot be used to exclude neoblast cells from the germline, since they share this property with the zygote itself. Lastly, and importantly, this definition would comply with the concept of the Weismann barrier (Figure 2, black line). The hereditary information can flow from both GCs and neoblast cells to somatic tissues, but it cannot flow in the opposite direction. For instance, mutations occurring in neoblast cells can be incorporated in both the soma (by means of cell differentiation, asexual reproduction and germline regeneration) and in the GCs (since these can be regenerated from neoblast cells). On the contrary, mutations occurring in differentiated somatic cells will be lost without possibly being incorporated into either neoblasts or GCs.

The old concept of a Weismann barrier is, therefore, useful in order to delineate the germline.

The idea of neoblasts constituting a kind of asexual germline that operates during asexual division and by regenerating the sexual germline is new. At first sight, it seems counter-intuitive, due to the fact that asexually reproducing animals have been traditionally understudied, and the germline concept has only been associated with sexual reproduction. However, there is little doubt in the current literature that neoblasts are the cell lineage that contributes the genetic information to forthcoming generations of asexually dividing planarians by driving the regeneration and cell turnover of animals after fission. This is at least true for the subset of neoblasts (cNeoblasts) that have been experimentally demonstrated to be pluripotent, clonogenic and capable of restoring regeneration in lethally irradiated hosts [29,35]. In these experiments, lethally irradiated sexual strain worms were rescued by the implantation of a single neoblast and completely transformed to the genotype of the asexual donor strain, showing that neoblasts contribute their genetic information to the soma [29]. The opposite experiment, with a sexual donor strain, would demonstrate whether cNeoblasts can restore all tissues including the germline. Further work is needed in order to determine the level of heterogeneity of the neoblast population (see [25]) and, particularly, what is the proportion of cNeoblasts in this population [29,35,36].

Another line of argumentation for such a revision of the germline conception lies in the function of nuage particles and the GMP. Little is known about this matter, although it is clear that piwi genes and related piwi-interacting RNAs (piRNAs), for instance, function in the control of selfish transposable elements [37-39]. The inclusion of neoblasts as a full member of the germline cycle in freshwater planarians would explain the need for such cellular machinery, which is not essential in somatic cells due to the Weismann barrier limitation, but is needed in neoblasts since they propagate the genome during asexual reproduction and permanently retain the ability to give rise to GCs.

Molecular data from other platyhelminthes indicate that the expression of GMP genes in neoblast cells and GCs is widespread among the members of the phylum. For instance, piwi and Vasa homologues are expressed in both GCs and neoblasts in Macrostomum lignano[40,41]. However, the questions of whether these cells both derive from embryonic GMP-positive populations and the presence of germ plasm components in the zygote itself remain to be addressed. Future lines of research should try to clarify these questions by studying the localization of GMP components coupled with lineage tracing studies during the embryology of basal platyhelminthes [42], such as Macrostomum lignano or the polyclad Hoploplana inquilina, since their spiral cleavage is more amenable to lineage tracing studies [43,44].

Molecular studies on the expression of GMP components in cnidarians have revealed as well the expression of these genes in undifferentiated stem cells and their propagation through a continuous germ plasm. The vasa transcript is missing in the zygote of the hydroid Hydractinia echinata, but Vasa protein is present in a specialized region of the zygote, implying it is a kind of germ plasm [13]. The adult stage of H. echinata and other cnidarians also contains a population of I-cells which are capable of giving rise to both somatic tissues and gametes [45]. Furthermore, it has been shown in the hydrozoan Clytia hemisphaerica that GMP components are also localized as a germ plasm in the oocyte and are inherited by I-cells [18]. Recently, RNA-seq molecular profiling of gene expression has been achieved in planarian neoblasts [46-49] and cnidarian I-cells [50], revealing extensive similarities in their gene expression patterns. Therefore, cnidarian I-cells resemble neoblast cells in their dual somatic and germ potential, their expression of GMP markers and their sustained perdurance over the adult phase. Hence, cnidarians also likely follow the model proposed in Figure 2.

It has been reported recently that both archeocyte and choanocyte cells express piwi homologues in the freshwater sponge Ephydatia fluviatilis[51]. Archeocytes are undifferentiated stem cells which have the ability to differentiate to at least several of the differentiated cell types that constitute these organisms (reviewed in [52]), and they progressively lose the expression of piwi as the differentiation process takes place. However, it has been reported that archeocytes can undergo differentiation to choanocytes as well, and the expression of piwi is sustained during this process [51]. Choanocytes, remarkably, have the ability to give rise to gametes, mostly sperm. Interestingly, choanocytes are very similar at the morphological and functional level to the unicellular choanoflagellates, believed to be the sister group of all metazoans and, therefore, their closest unicellular relative.

Spiralians are a group of lophotrochozoans that share the archetypal spiral cleavage [53], making evolutionary comparisons easy and illustrative. However, their respective embryonic development programs differ after spiral cleavage. For instance, a Müller’s larva is formed in some polyclads [42]. Trochozoans, however, form a trochophore larva, which is shared by groups as different as annelids and molluscs [54], the most widespread and well-studied trochozoan groups, and others [55]. Annelidian trochophore larvae grow posteriorly through the MPGZ, giving rise to the segments that will shape the adult animal. Molluscan trochophore larvae are coiled instead to shape another kind of larva, the veliger larva, from which the adult animal develops.

A broad conservation of cellular fates has been observed in different spiralian embryos. For example, the 4d blastomere of spiralian embryos (the so-called ‘mesentoblast’) is known to give rise to endomesoderm in most examples [53,56,57]. The germline is also known to arise consistently from this particular blastomere in many spiralians [9,10,17,19,53,58-64] (Figure 3B-C). Cell lineage tracing studies, however, are not possible in every organism, and this question is, therefore, difficult to address rigorously.

One beautiful and illustrative example of a combination of cell lineage tracing and molecular expression profiling of germ plasm components is that carried out by Rebscher and co-workers in the annelid species Platynereis dumerilii[10,19] (for an account of P. dumerilii embryonic development see [65]). In P. dumerilii Vasa protein is localized to a specific region of the oocyte and, upon the beginning of development, it becomes progressively restricted to the D quadrant and highly enriched in the 4d blastomere (Figure 3B). This blastomere further divides to give rise to four cells that will later migrate anteriorly and give rise to the GCs, the PGCs of P. dumerilii, and to a highly proliferative area, the MPGZ, which will give rise to somatic tissue exclusively. Both the PGCs and the MPGZ express Vasa and other GMP components [10,19]. These characteristics establish a striking parallelism between the 4d lineage of P. dumerilii, understood as including both the germ plasm-containing precursors and descendants of the 4d blastomere, and planarian neoblasts in their dual somatic and germ potential and the presence of germ plasm components (Figure 3D). Essentially, annelidian 4d lineage cells and planarian neoblasts behave very similarly and their only difference is their final fate and developmental potential (Figure 3E). In freshwater planarians, neoblasts are present throughout the adult stage; however, in annelid worms, it is not yet clear if multipotent GMP-positive cells are present in the adult. Nevertheless, the continuity of the germ plasm is still present in P. dumerilii, flowing from the zygote to the 4d blastomere and forth to GCs and MPGZ cells.

Studies in the annelid species Capitella teleta indicate that the expression of GMP components in this distant polychaete species [66] is similar to that of P. dumerilii[17] and that posterior regeneration observed in some annelid species is driven by GMP-positive cells [9]. However, whether these regenerative cells are derivatives of the MPGZ or express GMP components de novo is still a question.

The embryonic development of gastropod and bivalve molluscs also proceeds by spiral cleavage and the formation of a planktonic trochophore larva. The expression pattern of several GMP components has already been described during the embryonic development of several molluscs. The early embryonic expression of a vasa homologue in the oyster Crassostrea gigas was found to be similar to that of P. dumerilii[58]. The mRNA from the oyster vasa gene is found in the vegetal pole of oocytes and becomes progressively restricted to the 4d blastomere (Figure 3C). Then, the mRNA is detected in two cells, presumably two daughter cells of the 4d blastomere (Figure 3C). These two cells persist without further proliferation during the subsequent larval stages and are therefore believed to be the PGCs of C. gigas.

Similar patterns have been described in other mollusc species with some modifications. In the snail Ilyanassa obsoleta the mRNA of a vasa gene is broadly expressed in the early cleavages and progressively restricted to the 4d blastomere and its descendants. Later, however, the expression of vasa disappears and it is only seen later in the adult stage [64]. Similarly, a nanos transcript is also restricted to the 4d blastomere and its descendants in I. obsoleta, and its knock down has been shown to affect somatic cells [67]. In the mollusc Haliotis asinina, a more basal gastropod [68], both vasa and nanos mRNAs are expressed broadly in the zygote and become restricted later to the 4d blastomere. Their expression can be followed until the mesodermal bands of the trochophore larva, which likely give rise to somatic tissues and to the germline [69].

Therefore, some molluscs seem to follow a model similar to that of the annelid P. dumerilii (Figure 3E), but in contrast, the case of the mollusc C. gigas shows a more restricted fate for the embryonic GMP-positive cells (Figure 3F). The GMP-positive descendants of the 4d blastomere in C. gigas do not exhibit any self-renewing properties, nor do they seem to contribute to any somatic tissues, but only to the germline, and they can, therefore, be called PGCs. However, the parallelism with different species of closely related molluscs and annelids unveils that while the 4d blastomere only generates one germ plasm-containing cell type, believed to be the PGCs, in C. gigas, it generates populations of self-renewing, germ plasm-containing cells with somatic potential in other spiralians, such as the MPGZ in P. dumerilii. The model depicted for the mollusc C. gigas resembles what has been described for most organisms whose germline is specified by preformation (Figure 1A), including Drosophila melanogaster, Caenorhabditis elegans and Xenopus laevis, among others: a group of cells that are specified by preformation are set aside during embryonic development. Only a few divisions separate the zygote from the PGCs. The cells undergoing these divisions express GMP components and contribute to the soma. Through the comparison of the distinct embryonic developments of spiralian organisms, I propose that these cells can sometimes give rise to populations of stem cells with both somatic and germ potential and, in some animals, often with regenerative potential and asexual reproduction, these populations can even be present in the adult stage.

I thus call these cells the primordial stem cells (PriSCs) and define them as a group of multipotent or pluripotent cells that can have mixed germ/soma potential and that are always intercalated between the zygote and the PGCs during embryonic development. I propose that the PriSCs are part of the germline cycle, and this fact explains the presence of germ plasm components in them. The inclusion of the PriSCs in the germline cycle solves the long-standing question of germ plasm continuity in diverse groups such as annelids and platyhelminthes. Unlike the pPGCs introduced earlier in the literature, the PriSCs can have unlimited self-renewal and be present in the adult stage. Also, the use of the Weismann barrier concept allows delineating the PriSCs as part of the germline, since mutations occurring in any of these cells are competent to trespass the GCs and to be incorporated in forthcoming generations by either sexual or sometimes asexual reproduction. For instance, mutations occurring in the 4d lineage of either annelids or molluscs, or in the neoblast cells of triclad flatworms can all be transmitted to the GCs of these animals and, therefore, to forthcoming generations, and this fact might explain the presence of germ plasm components in them.

I describe three models of the germline cycle, depending on the behavior and potential of PriSCs. In the first one (unlimited, Figure 3D), the PriSCs can continuously self-renew, even throughout the adult stages of the animal, and PriSCs always retain a mixed germ/soma potential. This model is likely followed by triclad planarians, but also by acoel worms, cnidarians, sponges, and other animals with high regenerative capabilities and the potential for asexual reproduction. In the second model (restricted, Figure 3E), the PriSCs retain a mixed germ/soma potential but their self-renewing capacities are limited. This model can be observed in P. dumerilii and other annelids and molluscs, for instance. In the third model (rudimentary, Figure 3F), the PriSCs derived from the zygote only exhibit somatic potential and self-renewing properties for a very limited number of divisions during early embryonic development. The mollusc C. gigas likely follows this model, as well as the organisms considered to specify their germline by preformation, such as C. elegans, D. melanogaster and X. laevis.

Although the idea of multipotent or pluripotent progenitors being the precursors of the germline has been proposed previously by others [2-4,15,25], the addition of an asexual germline role for some of these cells helps in clarifying the variability observed in their different fates and potentials. Previous attempts relied on the somatic nature of these cells, a vision that has led to controversies. Their inclusion in a germline cycle observed in many phylogenetically distant animals strongly argues for the inclusion of these cells, the PriSCs, in a continuous germline concept according to the Weismann Barrier.