Applying “Gold Standards” to In vitro-Derived Germ Cells

Abstract

Germ cells are the ultimate stem cells and reports of their in vitro derivation generate excitement due to potential applications in reproductive medicine. To date there is no firm evidence that meiosis, the hallmark of gametogenesis, can be faithfully replicated outside the gonad. We propose benchmarks for evaluating in vitro derivation of germ cells, facilitating realization of their potential.

The last decade marked great excitement and substantial progress in the methods for in vitro development of germline cells. These efforts are vital for several reasons. First, these are the ultimate stem cells. Second, of all the cells in the body, germ cells go through the most dramatic epigenetic reprogramming, including specification of imprinted genes. Third, infertility rates are relatively high historically (~15% of couples in the U.S.), and societal changes in modern countries have exacerbated reproductive problems because both women and men have been delaying having children. Fourth, there is pressing need to preserve the fertility of patients undergoing cancer therapy. Fifth, the lack of robust in vitro systems for study of meiosis relegates the field to more difficult and expensive in vivo studies. Finally, the ability to recapitulate gametogenesis in vitro would enable functional evaluation and even correction of mutant infertility alleles using new genome editing technologies.

Recently, there have been several intriguing studies claiming the existence of female germline stem cells in adults, differentiation of various stem cells into germ cells in vivo and in vitro, and recapitulation of meiosis in culture systems. The aforementioned promise of these studies for translational medicine applications has generated excitement amongst clinical practitioners and hope for patients desperate for fertility restoration. However, because the stakes are high when it comes to artificial reproductive technologies involving unprecedented germ cell manipulation, it is imperative that the scientific community applies the highest standards when conducting and evaluating research concerning in vitro gamete generation. Indeed, some findings in this area remain controversial, and even for publications that are accurate in what they report, there are serious concerns about interpretation of the findings. Here, we lay out rigorous criteria by which to evaluate claims of recapitulating meiosis and proper gametogenesis from putative stem cells of several kinds, discuss past and future results in the context of the distinction between meiotic development and gamete development, and predict the potential for reproductive medicine and research if gametogenesis can indeed be recapitulated accurately.

Gold Standards for Proof of Meiosis

Deriving useful numbers of consistently competent germ cells from stem cells depends absolutely on the fidelity of meiosis, which is a defining event of gametogenesis. Meiosis reduces nuclear DNA content (“C”) in sperm to 1C, in ovulated oocytes to 2C, and the nuclear chromosome number (“N”) of both sperm and oocytes to 1N (see below and Figure 1). Meiosis also reshuffles gene alleles by a form of genetic recombination called crossing over, which is programmed to occur at least once between each pair of homologous chromosomes or chromosome arms. Crossovers ensure bi-polar orientation and normal segregation in the first (reductive) meiotic division. In many cases, the prolonged meiotic prophase, characterized by open chromatin, also serves to provide templates for the transcription of many genes required for subsequent phases of gametogenesis and/or early post-fertilization development. It is these hallmarks of meiosis that must be rigorously established in order for a claim to be made of successful derivation of gametes from stem cells. How should these benchmarks be established and proven? Below, we outline what we consider to be key benchmarks – a checklist –to substantiate successful in vitro spermatogenesis and oogenesis (Figure 1).

Figure 1. Checklist of key events during mammalian meiocyte development.

Diagrammed are the processes of spermatocyte and oocyte development, and key events in these processes. Of particular importance for assessing accurate meiocyte development are features denoted by the red checkmarks. Synaptonemal complexes can be visualized by immunolabelling proteins such as SYCP1 or SYCP3. Markers of recombination include RAD51 (which is bound at hundreds of foci in early Meiosis I prophase) and MLH1 (which binds 25 or so foci at sites of chiasmata in mice). Metaphase figures can be visualized in various ways cytologically (for example, by DAPI staining of DNA and alpha tubulin immunolabeling of the spindle). Mature oocytes should have extruded a single polar body, and fertilized oocytes (zygotes) should have extruded another (occasionally the first polar body will undergo a division too). A final test is that any zygotes formed from in vitro derived gametes should be able to form viable progeny following transfer to pseudopregnant female hosts.

DNA Content

Flow cytometry (particularly for spermatogenesis) and/or quantitative cytology should be employed to verify DNA content at four critical stages: 1) Pre-meiotic stage (2C/2N), 2) primary gametocyte stage (after meiotic S-phase, when germ cells are 4C/2N, note that this is not a “tetraploid” chromosome content as is often erroneously stated, but is simply the normal post-S phase DNA content of diploid cells, found also in post-S phase mitotically proliferating somatic cells.), 3) secondary gametocyte stage (after the first, reductional, meiotic division when homologs have separated) when the cell nuclei are 2C/1N and are formally haploid, and 4) after the second meiotic division when individual spermatids are 1C/1N (but unfertilized oocytes never reach this point unless parthenogenetically activated).

The meiotic division process itself is not suggested as a practical gold standard, because it is rapid, transient and difficult to monitor. Nonetheless, it should be kept in mind that there is substantial evidence for regulation of both timing (Duncan et al., 2009) and preferential segregation (Wu et al., 2005) during the oocyte meiotic divisions, either of which could influence developmental outcomes. Moreover, oocyte-like cells that have not entered meiosis can produce a polar body with mitotic segregation of chromatids (Dokshin et al., 2013).

In sum, although reductions in DNA content occur during meiosis, they do not constitute proof that all events essential to meiosis have occurred. It is crucial to emphasize that mammalian oocytes are normally never 1C since the maternal genome is reduced to 1C only after fertilization and the introduction of a 1C male genome. Thus, claims of in vitro-derived “haploid oocytes” based on flow cytometric detection of a 1C population portray an abnormal biological state not found in real life and are unacceptable as evidence for in-vitro derivation of normal oocytes.

Chromosome Content and Organization

Either chromosome counts or fluorescence in situ hybridization (FISH) analysis (and preferably both) should be used to establish chromosome counts in prophase gametocytes and MII metaphase nuclei (1N chromosome count, 2N chromatid count), as well as in the first mitotic metaphase of embryos produced by in vitro-derived gametes (Figure 1). In the case of in vitro-derived sperm, FISH analyses for several chromosomes, including the X & Y, can verify 1N chromosome count. For both male and female putative gametes, it must be demonstrated that paired homologs, each with two chromatids, are correctly oriented in opposition at the metaphase I spindle equator.

Recombination

Recombination is essential for the pairing and proper segregation of homologous chromosomes during mammalian meiosis, and it is stimulated by the programmed induction of double stranded breaks (DSBs). Although it is difficult to obtain definitive evidence for molecular resolution of DSBs, recombination can be assessed in spread or whole-mount chromatin prepared from primary gametocytes by using immunolabeling with antibodies against proteins involved in homologous synapsis and recombination to demonstrate appropriate nuclear and chromosomal localization (Bolcun-Filas and Schimenti, 2012; Handel and Schimenti, 2010). However, cytological or molecular detection of these proteins (by RT-PCR or Western analysis, for example) does not alone constitute evidence that putative germ cells are executing meiosis (see below).

Viable Euploid Offspring

The ultimate gold standard for faithful meiosis is the production of chromosomally-normal, healthy offspring. The robustness of a method for in vitro gamete derivation should be assessed by determining the frequency of offspring produced in relation to the number of gametes participating in fertilization. Simply attaining fertilization or a two-cell embryo is insufficient, because gametes from a number of meiotic mutants can undergo fertilization but not transition into viable embryos, and because even “oocyte-like cells” that fail to undergo meiosis can be fertilized (Dokshin et al., 2013). Furthermore, embryos with grossly abnormal chromosome content can develop significantly beyond the two-cell stage before arrest.

These “gold standards” constitute the baseline of what must be demonstrated to convincingly show that meiosis has occurred in vitro in a manner that generates functional gametes. Moreover, in addition to meiosis, a number of other parameters are important for fully accurate gametogenesis and subsequent health of offspring, including epigenetic reprogramming, a normal transcriptome, and maintenance of genome stability. These events are critical for success of assisted reproductive technologies, and, indeed, remain a challenge for both cloned animals and iPS cells (Okae et al., 2013) (Lister et al., 2011). Although we do not view recapitulation of these parameters as among the “gold standards” for demonstrating successful meiosis in vitro, they are essential for successful gametogenesis (see comments below).

Reports on the Derivation of Gametes in vitro

The generation of gametes in vivo can be divided roughly into 4 stages: 1) primordial germ cell (PGC) specification, which normally occurs in utero; 2) gonial maturation; 3) meiosis; and 4) postmeiotic development, which is limited to spermiogenesis in males. Among the most promising reports of “gametogenesis in vitro” are the development of conditions for differentiation of mouse embryonic stem (ES) or induced pluripotent stem (iPS) cells into primordial germ cell-like cells (PGCLCs) that, upon transplantation into an environment of appropriate somatic cells in vivo, appear to undergo meiosis and produce functional sperm and oocytes that contribute to normal progeny following in vitro fertilization (IVF) (Toyooka et al., 2003) (Hayashi et al., 2012; Hayashi et al., 2011). These studies have established that early steps of germ cell specification (steps 1 and 2) can be successfully recapitulated in culture, and have shown that transcription factor reprogramming of iPS or ES cells can be used for facile derivation of PGCLCs (Nakaki et al., 2013). Nevertheless, although Hayashi et al. documented zygotene-like “oocytes” derived in culture from PGCLCs, it has not yet been fully demonstrated that the definitive and temporal aspects of meiosis occur normally in vitro (Hayashi et al., 2012). Overall, this work established that ES or iPS cells can be converted into PGC-like cells that appear to be competent for meiosis in vivo but behave less well than PGCs following aggregation with embryonic gonad somatic cells.

Meiosis in vitro remains a significant challenge and, indeed, virtually no studies have fully met the “gold standards” for faithful meiosis (step 3). For example, Eguizabal et al. reported in vitro meiosis from differentiated ES and iPS cells; the criteria used were DNA content, epigenetic landscape and marker expression, but major hallmarks of meiosis were not shown, and fertilization of putative gametes was not attempted (Eguizabal et al., 2011). Geijsen et al. produced embryoid bodies from ES cells, and from these, PGCLCs were isolated and differentiated into 1C cells that, following intra-cytoplasmic sperm injection (ICSI), led to blastocysts. However, hallmarks of meiosis were not shown and development beyond the blastocyst stage was not achieved (Geijsen et al., 2004), leading to concerns about chromosome content of the putative embryos. Nayernia et al. came very close to achieving the “gold standards.” They derived spermatogonial cells lines that putatively underwent meiosis (however, the crucial hallmarks of meiosis were not demonstrated) and differentiated into cells with 1C DNA content that produced progeny following ICSI (Nayernia et al., 2006). In this study however, the liveborns were abnormal, and faithful transmission of parental genotypes was not clear. Collectively, these studies are promising by showing that PGCLCs can be derived successfully in vitro from iPS or ES cells; however, it is still the case that definitive and successful meiosis requires the gonadal environment in vivo. Thus, these systems are quite premature with respect to clinical applications of in vitro gametogenesis, and also fail to develop in vitro approaches for basic research into normal meiosis.

Although findings to date do not provide an in vitro system meeting the basic research needs of those interested in meiosis or spermatogenesis, the successful in vitro development of PGCLCs from iPS cells does open the door for gene therapy in men with mutations causing germ cell-deficiency. Although gonadal fate of iPS cells depends on donor genetic background (Ramathal et al., 2014), theoretically, patient iPS cells could be genetically corrected, differentiated into PGCLCs or spermatogonia, then transplanted into the patient’s seminiferous tubules. For females, the requirement for implantation of PGCLC:embryonic gonad somatic cell aggregates, followed by explantation for IVF, provides neither a resource for meiosis research nor a facile solution to infertility. Indeed, for fertility medicine, the requirement for embryonic gonad somatic cells seems to pose a major practical problem, currently insurmountable. Moreover, each culture is a “one-shot” process, because all the PGCLCs that develop into oocytes in the implanted aggregates mature synchronously (with no formation of a resting pool of primordial follicles). Nevertheless, this promising technology lends itself to the iPS gene therapy paradigm.

Distinctions between Meiotic and Developmental Programs

Successful gametogenesis requires both meiosis and acquisition of developmental competence; as mentioned above, meiosis alone is not sufficient for development of gametes. Recent studies have demonstrated that “meiotic” and “developmental” programs of gametogenesis are separate entities. Meiotic entry in C. elegans is genetically separable from the developmental decision to differentiate into sperm or oocytes (Morgan et al., 2013). Cells that have already entered meiosis and are genetically programmed for spermatogenesis can switch to oogenesis by inhibiting the Ras/ERK (RAS/MAPK3/1) signaling pathway. The meiotic and oocyte development programs in mice are also formally separable. Mice deficient for the gene Stra8 (stimulated by retinoic acid 8), which is required for oocytes to enter meiosis, are infertile and eventually undergo complete loss of oocyte-like cells. However, a recent study found that a small number of follicles and oocyte-like cells survived to birth in mutant females (Dokshin et al., 2013). Even though these cells failed to undergo any chromosomal hallmarks of meiosis (premeiotic DNA replication, recombination, synapsis, etc.), they were morphologically indistinguishable from oocytes, developed a zona pellucida, behaved like oocytes by organizing follicular development, were ovulated, formed a polar body during ovulation and in vitro maturation, became fertilized and developed into two-cell embryos. Yet, these cells cannot be identified as “oocytes” because of failure to enter meiosis.

How are these findings important with respect to reports of recapitulation of oogenesis in vitro or observation of de novo creation of oocytes in vivo from stem cells in adults (White et al., 2012)? Obviously, it is a cautionary tale that what looks like an oocyte may not be a true oocyte, and this underscores the necessity of a checklist applying “gold standards” for meiosis outlined above. In particular, there are many cases in the literature where “meiosis-specific” markers of recombination and chromosome synapsis (such as SYCP3 and DMC1) have been used as evidence that an oocyte-like cell is a real oocyte. However, these markers are not always restricted to meiotic cells and, and, when expressed ectopically, are not actually localized to structures in which they are critical for meiosis (such as the synaptonemal complex or recombination repair sites).

Claims for postnatal germline stem cells in the ovary (Woods et al., 2013) remain highly controversial (Lei and Spradling, 2013; Yuan et al., 2013; Zhang et al., 2012) (Oatley and Hunt, 2012). The distinctions between the meiotic and developmental programs of oogenesis are relevant here. Reports (White et al., 2012) that cells from human ovaries can be isolated and cultured with properties (germ cell markers; rare mitotic division) of “oogonial stem cells (OSCs)” and capable of producing fertilizable “haploid” oocytes when xenotransplanted must be evaluated not only by “gold standard” criteria, but also from the perspective of normal oocyte biology (where haploid oocytes are not found), and with respect to separable meiotic and developmental pathways. Although putative OSCs may be induced to undergo the oocyte-like development pathway, they may not execute the meiotic pathway. Thus, both challenges and opportunities arise to understand what it takes to coordinate the two processes. The generation of oocyte-like cells that are not really oocytes presents a fascinating opportunity to experimentally unravel the coordinated regulatory processes required to produce a functional oocyte.

Research and Clinical Applications for Germline Stem Cells

In spite of the caveats (and in some cases because of the caveats), developments in germline stem cell research have enormous potential on several fronts. First, the benefits for mechanistic understanding of human gametogenesis are great. Most of our understanding of the genetic control of gametogenesis, and meiosis in particular, comes from analysis of normal and mutant mice. Even though samples can be obtained, experiments are largely restricted to the in vivo situation. This is still tedious, especially for studies of the early stages of PGC growth and differentiation. With the ability to recapitulate aspects of gametogenesis in culture, it becomes feasible to do experiments that involve cell synchronization, facile addition or inhibition of genes and combinations of genes. However, it is important to consider the fact that there has never been reliable evidence of mammalian meiosis in vitro without somatic cells supporting gametogenic processes. Indeed, there is tight coordination between germ cells and the surrounding somatic cells (the follicle cells that surround differentiating oocytes and the Sertoli cells that surround spermatogenic cells) in vivo, and it is likely that meiosis and all aspects of gametogenesis depend on reciprocal signaling between these two cellular compartments (Sugiura et al., 2005) (Su et al., 2008). It is possible that co-culture methodologies will be required for accurate gamete generation in vitro.

The advent of efficient genome editing technologies is likely to have manifold benefits for both basic and clinical germline stem cell research. It would immediately become possible to create null alleles of any gene in stem cells, and then test the consequence of such mutations upon gametogenesis. Using homology-directed repair stimulated by site-specific induction of double-strand breaks (DSBs) by CRISPR/Cas9 or TALENs, it becomes feasible to recapitulate putatively deleterious single nucleotide polymorphisms (SNPs) in gametogenesis genes, and test the actual consequence of these SNPs or de novo mutations in patients who have undergone genome sequencing to identify potential causes for infertility.

Identification of germline stem cells, or their derivation from iPS cells, could provide impetus to developing new reproductive technologies for species conservation or improvement of domestic animals of economic importance. Even more “blue-sky” would be the ability to use patient-specific iPS or germline stem cells to perform these genome editing procedures. In the case of men, if spermatogonia are unaffected, it should be possible to utilize established culture methods, coupled with genome editing, to reconstitute a functional germline. In females, the iPS route (iPS>PGCLCs>oogonia) could be taken. However, the extended manipulations might have unacceptable and insurmountable side effects, as suggested by the low success rate even without manipulation. If putative OSCs could be proven to exist and if they could be harvested and conduct proper meiosis, manipulation in culture might be less than in generating iPS cells, but the rarity of OSCs would preclude screening them for proper editing. Certainly the iPS route has the greater potential for clinical application than OCSs because of more ready availability, but the challenge remains to demonstrate the “gold standards” for meiosis.

The good news is that today there are multiple routes by which plausible “reprogenetic” therapies can be envisioned. However, at least in the near term, the greatest benefit of the recent advances in derivation in vitro of germ cells may be in facilitating our understanding of the basic biology and mechanistic control of germ cell development and meiosis.