A one-cell zygote before its first cell division Credit: Kevin Eggan © Harvard University

Research with unfertilized eggs usually means racing the clock. Under the egg's influence, mouse nuclei specialized to run the affairs of, say, a skin cell or an antibody-making blood cell can be reset to create entire animals or new lines of embryonic stem cells. But eggs' power over nuclei declines rapidly after they are collected, and even the freshest eggs do not reprogram specialized nuclei perfectly. Especially for humans, the scarcity of unfertilized eggs for research limits attempts to study development or to create embryonic stem cells (ES cells) genetically tailored for particular diseases or patients.

Fertilized human eggs are much easier to come by, but transferring nuclei from adult cells into fertilized mammalian eggs routinely fails to create embryos, so for the past 20 years research has focused on better ways to collect, and even create, unfertilized eggs. Now, Kevin Eggan and his colleagues at Harvard University have taken a different tack. They have shown that, if chromosomes rather than whole nuclei are transferred, fertilized eggs, or one-cell zygotes, can be used to generate cloned mouse embryos from which ES cells can be cultured1. They've even been able to get embryos cloned by this method all the way through development to adult mice. The team has already begun experiments to see if human ES cells can also be generated from fertilized human eggs in this way.

Click here to read an interview with Davor Solter, whose work helped convince scientists that oocytes, not zygotes, should be used for cloning.

Fertilized eggs offer several advantages over unfertilized eggs. The potential of unfertilized eggs to develop into embryos starts to decline soon after ovulation, while fertilized eggs are more stable and robust. Moreover, fertility therapies routinely produce fertilized human eggs with chromosomal abnormalities, and these zygotes are more available for research than are fresh human oocytes. Currently they are often discarded.

So why has the Harvard team succeeded when previous attempts with fertilized eggs failed? A crucial difference, Eggan believes, is leaving some factors in the zygote nucleus behind. The researchers used drugs to arrest the cell cycle right before a zygote divides for the first time to form a two-cell embryo. At that point, the nuclear membrane has broken down and the chromosomes are coiled up and attached to a structure known as a spindle. Instead of removing the nucleus, the researchers removed the chromosomes on their spindle. Then, the Harvard team replaced the original chromosomes with chromosomes at a similar stage from skin cells from donor mice.

Eggan thinks that removing only the chromosomes, rather than the whole nucleus, from the egg leaves essential proteins in place that can then reprogramme the donor chromosomes for embryonic development. That's a strong possibility, but other factors should be considered, says Tony Perry of the RIKEN Center for Developmental Biology in Kobe, Japan, who was part of the team that made the first cloned mouse, Cumulina, in 19982. Factors generally enriched at metaphase (the point of the cell cycle when chromosomes align on the spindle) or even disruption of the spindle microtubules during chromosome transfer might also spur the zygote to reprogramme the donor chromosomes. If true, this might have practical applications. Perry says the Harvard team's work raises the possibility that if treated similarly, cell types other than eggs, such as the stem cells present in adults, might be able to reprogramme donor chromosomes as well. That might augment or supplement other reprogramming technologies.

Resetting the programme

Despite their unconventional technique, Eggan and his colleagues have still come up against the same high failure rates and other drawbacks as conventional nuclear transfer—mouse embryos that fail to implant, miscarriages, stillbirths, and live pups with fatal respiratory problems.

What the oocyte uniquely does is convert the egg and sperm chromosomes into an embryonic genome. When you ask it to do the same thing on another genome, you're giving it a completely different substrate. , Keith Latham, Temple University

Introduced nuclei don't completely substitute for the zygote genome, explains Keith Latham of Temple University in Pennsylvania. That probably explains why cloning generally has low success rates between 1 and 7%3. Cloned embryos have different culture requirements from normal embryos, and those requirements correlate with the source of the donor nucleus4. “What the oocyte uniquely does is convert the egg and sperm chromosomes into an embryonic genome. When you ask it to do the same thing on another genome, you're giving it a completely different substrate.” For example, DNA in somatic cells is more exposed and more active than DNA in a gamete. That's not necessarily a limitation, says Latham, “One could make the assumption that the best clone would look like a normal embryo, but that's an assumption.”

Still, a better understanding of how an egg reprogrammes the gamete nuclei —the sperm and egg nuclei — when they come together to form the embryonic genome would hint at what might work best for chromosomes coming from other cells. “Once you understand what reprogramming is, you'll be able to do it much more efficiently in a dish,” says Renee Reijo Pera, head of Stanford's Center for Human Embryonic Stem Cell Research, who hopes to clone human embryonic stem cells using human eggs that failed to fertilise in fertility clinics. But that understanding is far from complete. “It's the first step in our own development, and we don't know the genes involved,” she says.

Programming of the egg chromosomes to direct embryonic development starts as far back as the pre-ovulation egg, or oocyte5. As it matures, an oocyte unpacks and repacks its chromosomes, exposing new strings of DNA to the cellular machinery that controls gene expression through a process called 'chromatin remodelling'. Chemical tags, such as methyl groups, on DNA get reshuffled, which also affects the potential for gene expression. By the time fertilization has taken place, the chromosomes of the egg and the sperm have become programmed to set the newly diploid zygote on its way to develop into an embryo and beyond.

One possible drawback of Eggan's technique, therefore, is that the introduced chromosomes will miss out on reprogramming that occurs before the zygote stage. Outi Hovatta of the Karolinska Institute in Stockholm, Sweden, is looking for the molecular factors that make up the human egg's reprogramming machinery. She and her colleagues have been able to profile gene expression in human oocytes at different stages of development and have found genes expressed that are known also to be active in mouse and cow oocytes. Not surprisingly, several of the genes active in oocytes control the cell cycle — the cycle of DNA replication, chromosome segregation and cell division — which is partly on hold in oocytes. Others are part of known signaling pathways in embryonic stem cells. But of the more than 10,000 genes expressed in human oocytes, nearly half were of unknown function6. And homing in on the differences that matter is difficult. “In mice, you can use an inbred strain, and they are all similar. In humans, there are large amounts of variability,” she says.

And gene expression won't tell the full tale. A mature oocyte is already full of RNA transcripts laid down earlier, and their translation into proteins is tightly regulated. Work by Kaiquin Lao, a biophysicist at Applied Biosystems in Foster City, California, and Azim Surani at the University of Cambridge, UK, indicates that the small RNAs called microRNAs (miRNAs) are crucial for regulating the translation of messenger RNAs in mammals in embryonic development. MicroRNAs prevent the production of particular proteins by binding specifically to their messenger RNAs. The team compared miRNA levels in mature mouse oocytes and early embryos and found that miRNAs fall precipitously in the two-cell embryo, but rise again at the four-cell stage, presumably as a result of new transcription from the embryonic genome7. Lao and colleagues have analysed miRNAs in individual mouse oocytes and early embryos, including ones lacking the enzyme Dicer, which is essential for miRNA function. Oocytes without Dicer were dramatically abnormal and failed to develop any further. Many of the miRNAs studied so far seem to be involved in regulating the formation of the microtubule spindle, but Lao doesn't discount a role for miRNAs in nuclear reprogramming, pointing out that spindle defects are simply easier to spot.

Vital controls

The two processes might even be connected. The regulation of cell division need not be distinct from nuclear reprogramming. “The next big stage is to bring the cell cycle and the [chromatin] remodelling activity together,” says Mary Herbert, who studies cell-cycle regulation in mammalian oocytes at the University of Newcastle upon Tyne, UK. She is part of a team that identified many defects in cell-cycle components when attempting nuclear transfer in oocytes from fertility clinics that failed to fertilize8. As an oocyte develops, it sheds chromosomes in the two divisions of meiosis (the nuclear division that reduces chromosome number from diploid to haploid during the formation of gametes). The first meiotic division creates an egg ready for fertilization; the second happens after the egg is fertilized. A human oocyte takes much longer to complete the first meiotic division than a mouse oocyte — 20 hours compared with only 6 to 12 hours for mice. “It seems to bide its time,” says Herbert, describing how the meiotic spindle with its attached chromosomes moves across the cell. “The idea that there might be large-scale reprogramming seems like a reasonable speculation.”

More reprogramming occurs at fertilization. A sperm's entry into the egg initiates intracellular pulses of calcium ions. The rise in calcium causes the degradation of egg proteins that prevent cell division, and also activates histones, proteins that package DNA and so control the structure of chromatin. Other mechanisms for controlling chromatin structure and gene expression are also implicated. “The calcium oscillations induced by the sperm can be important for demethylation [of DNA],” says Rafael Fissore, a developmental biologist at the University of Massachusetts in Amherst. “That is something that has not been tested but needs to tested.”

In conventional somatic-cell nuclear transfer, calcium waves are artificially induced after an unfertilized egg's nucleus has been removed and replaced with the donor nucleus, potentially exposing the foreign nucleus to reprogramming caused by this artificial activation. Under Eggan's techniques, donor chromosomes are not introduced until just before the one-cell zygote prepares to divide for the first time. Nonetheless, while reprogramming is certainly imperfect, it seems no worse than in unfertilized eggs, indicating that the calcium waves are not essential.

Eggan speculates that chromosome replacement in cells from a two- or even a four-cell embryo might also be possible. “As time goes on in development, you might lose reprogramming activity,” he says, “but I'm not convinced.” Still, at this stage of technology, obtaining any live clones or well-characterized ES cells wins accolades. If techniques improve, reprogramming activity before and after fertilization will receive further scrutiny.

But studying reprogramming is hard to do, particularly for human eggs and embryos. Categorizing gene expression and RNA transcripts provides only a hint of what proteins are being produced and what they are doing. If the mechanism of reprogramming was understood, technologies could be invented to improve it, says John Gurdon, the first person to clone an animal, a frog, in the 1960s using nuclear transfer from a differentiated somatic cell. But most papers in the field deal with what can be made to happen rather than how it happens, frets Gurdon. To really solve the mystery, he says, researchers must identify the proteins involved.

The frog egg instructs introduced nuclei to turn on stem-cell genes, says Gurdon. The key is to find precisely what molecules effect that switch. To this end, Gurdon and his team at Cambridge University, UK, are placing mouse nuclei into frog eggs, whose size and abundance make otherwise unthinkable experiments possible. They recently screened frog eggs for proteins that promote the expression of Oct4, a gene that is normally active in ES cells, and identified Tpt1, a protein previously associated with cancer9. Gurdon's group is currently modifying the screen to explore additional control regions of Oct4, as well as to find proteins that interact with nanog, another stem-cell marker gene.

The problem is that getting closer to mechanism usually means studying species further from humans, and even descriptions of proteins don't tell you what's important. In fact, pluripotent cells (cells with almost the same developmental potential as an egg) can be created, albeit imperfectly, through techniques that have nothing to do with the egg, like fusing somatic cells with ES cells or even by forcing the expression of a handful of genes, points out Rudolf Jaenisch at the Whitehead Institute in Cambridge, MA10. “Does the egg do it the same way?” he asks, “Probably not.” The next steps require moving from description to function. Right now, he says, “we're cataloguing a lot. What does what: that's a big issue.”