Last May, to much fanfare, an international group of researchers published two papers describing a new in vitro system that had maintained human embryos in culture for 13 days.1,2 The experiments could have continued beyond two weeks, if not for the “14-day rule”—a widely recognized limit to how long scientists are permitted to maintain human embryos for research purposes. Bioethicists first proposed the rule, which was subsequently enshrined in the laws of several countries and as a guideline in the U.S., in 1979. Three and a half decades elapsed before the technology existed to keep embryos alive outside of a womb past the implantation stage, which typically occurs about a week after egg and sperm cells fuse. Now, the rule was finally coming into play.
“The decision to stop this beautiful amazing structure that [was] moving forward with self-organization . . . was the toughest I’d ever done in my professional career,” says Rockefeller University embryologist Ali Brivanlou, a senior author on one of the papers. “I did it because of respect for guidelines.”
The researchers stopped the experiment by flash-freezing the human embryos in liquid nitrogen, suspending them in time. “I have no idea if we will be able to thaw them again and have them come back. But my hope is that one day—hopefully within my lifetime; if not, the next generation of my students and postdocs and others—we’ll have the opportunity to go back to the liquid nitrogen and thaw these embryos and ask a very simple question as to how far this self-organization can sustain itself [in culture]. Because it’s impossible to imagine that this can go on much farther than 14 days.”
The research has reinvigorated the ethical discussion concerning the culturing of human embryos for scientific study, while providing the means to study embryos postimplantation—a period of development that has remained largely mysterious until now. “What happens [during the second week and] later has been the black box of development, because we could not successfully culture embryos beyond implantation,” says Magdalena Zernicka-Goetz, a developmental and stem cell biologist at the University of Cambridge in the U.K. whose lab developed the new system. Meanwhile, other technological advances are yielding major insights into the very first week of embryonic development—a period that involves the reprogramming of two highly differentiated cells, a sperm and an egg, into a totipotent cell from which an entire organism will form.
“It seems to be a very hot area of research, I think in part because we’re trying to understand what creates this very interesting tabula rasa state of the genome where it’s totipotent—it can turn into anything,” says MIT biophysicist Leonid Mirny.
With the advent of single-cell technologies, scientists are, for the first time, able to take a peek inside the individual cells of two-, four-, and eight-cell embryos, as well as inside the individual pronuclei—one from mom and one from dad—of the initial one-cell embryo, called a zygote, formed upon fertilization. Just in the past few years, experimental results have begun to reveal how the zygotic genome is reorganized and reprogrammed to transfer control of development from maternal factors harbored by the egg to the embryo’s own genes. “Given how few of these cells there are, it’s really amazing we can now look into these early stages of development,” says Mirny. “This progress is totally driven by the single-cell techniques.”
“It’s only now that we start having a glimpse of what takes place [in the first hours and days],” agrees Didier Trono of École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland. “These last couple of years—and in the few years to come—we’re making tremendous progress in understanding what happens during this period.”
In the hours and days that follow fertilization, the genomes of the newly united egg and sperm cells begin to express genes important in early development. Prior to this activation, maternal factors packaged in the oocyte are in charge. But changes to the overall chromatin structure of the paternal and maternal genomes, which are housed in separate pronuclei within the zygote, permit access by transcription factors shortly after fertilization—at about 13 hours in mouse embryos. The exact nature of these dynamics, however, has remained shrouded in mystery for decades.
This March, Kikuë Tachibana-Konwalski, a cell biologist at the Institute of Molecular Biotechnology (IMBA) of the Austrian Academy of Sciences, and her colleagues published the first in-depth look at how chromatin structure changes from the oocyte to the single-cell embryo in mice. Her group teamed up with Mirny’s lab at MIT to refine a method known as Hi-C (high-resolution chromosome conformation capture) so it could be applied to individual nuclei. During Hi-C, pieces of DNA that are close in space—regardless of their genomic distance—are glued together at contact points, or contacts, before enzymes digest the DNA. The glued pieces are then chemically ligated into single DNA fragments. These hybrid DNA molecules are sequenced, and researchers use computational techniques to map the sequences to determine the higher-order, 3-D structure of the intact genome.
See “Nuclear Cartography”
The problem was that traditional Hi-C approaches require thousands or millions of cells. This is because researchers would filter out those reads believed to be hybrids, and they needed to retain enough material to generate a complete map of chromatin conformation. Mirny’s team found that they could skip this filtering step, isolating as much DNA as possible from a single cell, and then sort out computationally those reads that are productive. “So you spend more money on sequencing but you’re trying to minimize DNA loss,” Mirny explains. “As a result, we got 10 times more contacts per cell” than the only other published single-cell Hi-C technique. In total, the method yielded “about a million contacts per individual cell,” he says, which “gives you enough information to reveal major features of chromatin organization.”
Applying this approach to paternal and maternal pronuclei of mouse zygotes, Tachibana-Konwalski’s team analyzed the chromatin structure of the two genomes. According to their results, both the paternal and maternal genomes appeared to have already reestablished local features known as loops and topologically associated domains (TADs)3—a finding in conflict with two other studies published this summer, which did not detect these structures until the embryo reached the eight-cell stage or became a blastocyst, a hollow ball of cells that implants in the uterine wall.4,5 Tachibana-Konwalski says she and her colleagues “are confident that TADs and loops form within hours after fertilization in zygotes,” having found evidence of TADs in an as-yet unpublished reanalysis of the other groups’ data “with greater statistical power and appropriate controls.”
Tachibana-Konwalski’s team also found a surprising difference between the two pronuclei of the zygote. While the paternal genome also contained higher-order formations called compartments, the maternal genome contained only the local structures, but no compartments—global features of chromatin in which transcriptionally active DNA associates more closely with other transcriptionally active regions, while silent stretches associate more closely with one another. That the paternal pronucleus contained these features while the maternal pronucleus did not “was really unexpected,” says Tachibana-Konwalski. The paternal genome “seems to be winning the [reprogramming] race.”
One area of the genome where restructuring appears important for early development is the heterochromatin—highly compacted regions of DNA that are normally silent but that suddenly become active in the zygote. For example, retrotransposons, one of the main components of heterochromatin, are highly transcribed at this time. “The activation of these retrotransposons is very peculiar for the developmental process,” notes Maria Elena Torres-Padilla, an epigeneticist at Helmholtz Zentrum München in Germany. “It only happens otherwise in disease and cancer and very specific situations; in most of our cells these transposons are silent.”
Most researchers had considered retrotransposon activation to be a side effect of the overall reprogramming process, says Torres-Padilla—as the chromatin restructured, transposons were freed from their normal repression, the thinking went. But that explanation didn’t sit well with her. So she and her colleagues used transcription activator-like effectors (TALEs), a gene-editing technology, to selectively manipulate the transcription of LINE-1 transposable elements in mouse embryos during the first few days following fertilization. When the researchers prevented LINE-1 activation, they observed decreased rates of development. However, adding LINE-1 mRNAs to make up for the lack of transcription did not rescue the phenotype.6 “That was the most surprising finding,” says Torres-Padilla—“that it’s not the messenger RNA itself, but it was really what we were doing on the DNA loci at the chromatin level.”
Just what’s going on remains to be seen, but she suspects that retrotransposon activation somehow initiates zygotic gene expression. “You have thousands of genes that are going to be activated from the genome of the embryo for the very first time,” she says. “I think what the LINEs are doing is to help open up the chromatin, so that perhaps other elements that direct transcription in [other] genes can function more efficiently.”
Still, whether changes in chromatin structure are driving early embryonic transcription eludes researchers. And there’s still another piece of the puzzle that scientists are working to fit in: at the same time that the chromatin of embryonic genomes is restructuring, the vast majority of cytosine methylation on the DNA is lost. But the exact timing and causative relationship of these changes is unclear. “I think the most exciting aspect of zygote biology is to combine these approaches to precisely understand how individual modifications will change overall chromatin structure,” Tachibana-Konwalski says. “To me, the next natural step is to merge these two levels of organization.”
Methylation overhaul and transcription initiation
While genome-wide DNA methylation analyses have documented the global removal of cytosine methylation from the maternal and paternal genomes in the zygote, as well as the reestablishment of these marks over the first few days of embryonic development, the pathways that control this epigenetic revamp have been hard to pin down. In recent years, analyses focused on individual cells within the embryo, along with the application of gene-editing technologies to selectively block or activate enzymes thought to play a role, have begun to elucidate these enigmatic processes. “At present, our knowledge of epigenetic reprogramming is accumulating at a dizzying pace,” one group of researchers wrote in a 2014 review of the field.7
In the maternal genome, passive dilution of the methylation marks occurs over a few days, while the paternal genome undergoes active and rapid demethylation—often accompanied by replacement with alternative modifications, including hydroxymethylation and carboxylation—shortly after fertilization. One proposed mechanism of this active demethylation process, first posited by Azim Surani of the Gordon Institute and colleagues in 2010,8 is the breaking and repairing of DNA, and several studies over the years have lent support to this hypothesis. “Of course, [inducing DNA breaks] would be very dangerous at this stage when it’s a single-cell embryo,” Tachibana-Konwalski notes. “It’s not exactly what one would expect evolution to do.”
See “The Role of DNA Base Modifications”
Luckily, as she and her colleagues discovered last year, the cell has a surveillance mechanism to ensure that development does not continue if the breaks go unrepaired. By knocking out key components of the DNA repair pathway, Tachibana-Konwalski and a colleague found that when lesions remained, the zygote did not undergo its first cell division.9 “This was the first evidence that epigenetic reprogramming is monitored in the context of the cell,” she says. “So if reprogramming is delayed, then the zygotes will not enter first mitosis.”
Although many questions remain, continued study of the reprogramming process—both at the level of overall chromatin structure and of DNA methylation—will be important for understanding exactly what controls the initiation of embryonic transcription. While transcriptomic surveys over the past several years have begun to document which genes are expressed very early in development, what triggers those transcriptional changes remains a key question in the field. This year, taking a closer look at one of the first genes turned on, EPFL’s Trono and colleagues identified what they think might be an important clue.
It all started with the discovery in the 1990s that patients suffering from facioscapulohumeral muscular dystrophy harbor mutations in a gene called DUX4 that cause the gene to be overexpressed. Then, in 2012, Stephen Tapscott of the Fred Hutchinson Cancer Research Center and colleagues forced the production of DUX4 protein—which is normally epigenetically repressed—in cultured human myoblasts and observed the upregulation of a suite of genes known to be active during early embryonic development.10 This caught the attention of Trono, who decided to probe deeper into DUX4’s potential role in embryonic genome activation.
Existing data on gene expression in human and mouse embryos confirmed DUX4 is expressed just before full embryonic genome activation. When Trono and his colleagues overexpressed the gene (known simply as DUX in mice) in mouse embryonic stem cells, they also saw an induction of the expression of other genes active in early development. The team further demonstrated that DUX bound to the promoters of some of these genes. Finally, deleting DUX in mouse embryos just before the two-cell stage—a tricky methodological feat achieved using the CRISPR-Cas9 gene-editing system—the researchers blocked embryonic genome activation altogether.11 “That was the nail in the coffin, I would say,” Trono says. “What this strongly suggests is that DUX is the gene product that kicks it off.”
“With the identification of the DUX transcription factors, this has opened up an avenue to understand the first wave of transcription factors,” agrees Tachibana-Konwalski. But the question remains—what initiates DUX expression? “Even with DUX, it appears that there must be some upstream factors, and this we are still totally ignorant on,” she says. “The jury is still very much out on what the master totipotency factor is in mammals.”
While many groups continue to hash out the molecular factors governing embryonic totipotency (which differs from pluripotency; see box on opposite page), others are looking forward to the next important milestone in embryonic development—determining what dictates which cells will form the baby itself and which cells will form the placenta. “When one follows later lineages, there will be differences that one would like to trace back, and ultimately one will trace them back to the zygote and its initial cell-fate separation,” says Rickard Sandberg, a computational geneticist at the Karolinska Institutet in Sweden. Once again, single-cell technologies are allowing researchers to do just that.
Over the past several years, the labs of Zernicka-Goetz at the University of Cambridge and Nicolas Plachta at the A*STAR Institute of Molecular and Cell Biology in Singapore have independently shown that, in mammals, this decision isn’t black-and-white. Although cells of mammalian embryos differ from one another early on, they retain flexibility in cell-type specification, Zernicka-Goetz explains. “Those fate decisions happen gradually, starting at the four-cell stage and possibly even earlier.”
The big question, then, was how cells became biased toward forming one lineage over the other. Last year, Plachta and colleagues found that transcription factors such as Sox2 bind to mouse DNA for different periods of time at the four-cell stage, and that this correlates with cell fate.12 In the same issue of Cell, Zernicka-Goetz’s group published a study that further explained why: those murine cells with longer Sox2 binding start to express genes, including Sox21, that repress the expression of transcription factors associated with differentiation.13 As a result, these cells preferentially form the interior population of cells that give rise to the fetus. “I think that this is one of the important discoveries over the last few years,” Zernicka-Goetz says.
Of course, this all ties back to the epigenetic reprogramming that the zygote undergoes during its very first hours and days: the length of SOX2 binding is regulated by CARM1, an enzyme that methylates arginine 26 on histone H3 (H3R26). “So as far as we know for now, everything starts with this particular epigenetic modification—methylation of histones—and this drives cell-fate specification,” Zernicka-Goetz says. But what initiates CARM1’s methylation of H3R26? “The situation is complex,” she says. “Our group and many others are still trying to discover what it is that breaks the symmetry for the very first time.”
Still, the progress that has been made in the past few years toward understanding the first hours and days of embryonic development is promising, Mirny says. “Single-cell techniques are still in their infancy across the board, so these are challenging techniques in general, but I think the picture is coming together.”
The next frontier
Developmental biologists appear poised to answer many of the remaining questions about the transition from maternal to embryonic control of development that happens in the first few days after fertilization. The next challenge lies in the weeks that follow, says Zernicka-Goetz, a period into which researchers are just now getting their first glimpses. And so far, her group and others have demonstrated that embryos are more self-sufficient than previously appreciated.
Initially published in 2012,14 with refinements made a couple of years later,15 the new culture system designed by Zernicka-Goetz’s team has successfully been used to sustain both mouse and human embryos until the point of gastrulation, when the three distinct embryonic cell layers—the ectoderm, the mesoderm, and the endoderm—form following implantation.1,2 This work has demonstrated that embryos self-organize without input from their maternal host—at least, up to 13 days postfertilization. “Our work and Ali [Brivanlou]’s work show the same thing: that the embryo can organize itself outside the body of the mother,” says Zernicka-Goetz. “It doesn’t need the maternal information at that stage of its life, which I think is incredible and unexpected.”
In addition, these experiments have revealed how the different types of cells in the early embryo interact with one another. This year, Zernicka-Goetz and her group used that knowledge to replicate those interactions using mouse embryonic stem cells and extra-embryonic trophoblast stem cells. Placed in a dish with a 3-D scaffold that resembled the extracellular matrix, the cells assembled to create the first-ever synthetic mouse embryos.16 While these entities will likely also be the subject of regulations that limit their development in culture, they provide yet another window into the “black box of development” that is the period following implantation, says Zernicka-Goetz.
In combination with advances being made in the study of the first week of development, the study of embryogenesis continues at an unprecedented pace. The next few years should see the publication of new insights into the miracle of life, says Tachibana-Konwalski. “It’s an amazing and dynamic field.”