CHAPTER 8

image

NOW CREATE A CLONE

I have, up until this point, been very clear that mammoths will not be resurrected by cloning. What I say next, therefore, is likely to be confusing. The next step in bringing a mammoth back to life is to create a clone.

In my defense, the cells that we would clone at this stage would be very different from the cells that the Japanese and South Korean teams hope to find and use in their cloning experiments. By the time we arrive at this stage of de-extinction, we might have spent years (even decades) in the lab, painstakingly engineering changes to the elephant genome within our cells. We would not be beginning our cloning experiments with miraculously well-preserved mammoth cells. Nonetheless, the next step in de-extinction would be to “clone” our cells, and thereby turn them into an entire elephant (with some mammoth genes).

Some de-extinction projects will, of course, be able to skip the genome-engineering steps and proceed directly to cloning. These projects may move ahead much more quickly than those that require genome engineering. Of course, that simply means that they will be first to arrive at the next hurdle. Consider the example of the bucardo.

NOT QUITE THE FIRST DE-EXTINCTION

In the summer of 2003, a young female bucardo, which is a subspecies of Spanish ibex (a type of wild goat), was born. Bucardos had been endemic to the Pyrenees, the mountain range that forms the border between Spain and France. When this baby bucardo was born, however, bucardos had been extinct for just over three and a half years.

The baby bucardo was a genetic clone of the last living bucardo, an elderly female named Celia. Unfortunately, the baby suffocated within minutes of birth. An autopsy revealed that she had been born with a malformed lung and had no chance of survival. Nonetheless, the birth of this baby bucardo is often held up as the first successful de-extinction. I disagree. To me, if she had no chance of survival, this is not de-extinction.

As de-extinction projects go, the bucardo project has a lot of promise. Bucardo cells were harvested from Celia ten months prior to her death and immediately frozen, and the DNA within those cells is in very good condition. Several closely related subspecies of Spanish ibex are still thriving, so finding appropriate egg donors or surrogate mothers should be straightforward. The bucardo has also not been extinct for very long, and its extinction was likely due to overhunting and not to the disappearance of its habitat. As long as we can control our guns, resurrected bucardos could be returned to their native habitat without the need for extensive environmental impact studies or political maneuvering.

When the team of Spanish and French scientists began the bucardo project in 1989, bucardos were not yet extinct. Cross-species cloning had also not yet been achieved in large mammals, and the challenges facing the project were immense.

In 2001, and in a separate effort to perform cross-species cloning, the biotechnology company Advanced Cell Technologies successfully cloned a gaur, which is an engendered species of cattle native to South and Southeast Asia, using a cow as a surrogate host. The cloned gaur lived only forty-eight hours before succumbing to dysentery, but its birth demonstrated that cross-species cloning was possible. Two years later, the same company successfully cloned a banteng, another endangered cattle species from Southeast Asia, again using a cow as a surrogate host. This banteng lived in the San Diego Zoo for seven years—less than half its lifespan in the wild—before dying of what appeared to be natural causes.

The bucardo project was similar to the gaur and banteng projects in that there was no need for genome sequencing or genome engineering for cloning to be possible, and in that surrogate hosts were available. There were, however, two important differences that distinguished the bucardo project from the others. First, assisted reproductive technologies were already established for cattle, but had not yet been developed for Spanish ibex. Second, by the time the team of scientists developed this technology, the bucardo had gone extinct.

Unfortunately, the bucardo cloning experiment did not succeed, and it is not entirely clear why. It is possible that the experiment failed because the scientists simply did not make enough embryos. Cloning by nuclear transfer is, after all, notoriously inefficient. The team transferred copies of Celia’s somatic cells into 782 eggs, but only 407 eggs developed into embryos. Of these, 208 embryos were transferred into potential surrogate hosts, but only seven pregnancies were established. Only one pregnancy lasted to term, and the baby bucardo that was born lived for less than ten minutes. If one were to count this baby bucardo as a successfully established clone, which I will do here only for the purpose of illustration, the success rate of bucardo cloning would be 0.1 percent.

Alberto Fernández-Arias, who is the director of the Aragón Hunting, Fishing, and Wildlife Service and who was brought in to the bucardo project in 1989 to develop assisted reproductive technologies in Spanish ibex, believes that I am being unfair in my characterization of this as a “failed” de-extinction. He points out that if his team had known that the bucardo would be born with a lung deformity, they could have been prepared to remove the malformed part of the lung immediately after birth. Such surgery has been performed successfully on human babies with similar birth defects and, indeed, may have been able to save the baby bucardo’s life. Of course, it is not possible to know either what caused the lung deformity or what might have happened—how the bucardo would have developed, how it might have fared in adulthood—if the bucardo had survived. The project continues, however, and we may soon have another opportunity to find out whether bucardos will once again roam the Pyrenees.

De-extinction by Nuclear Transfer

Once we have a cell that contains the genome of the animal we want to create, whether that cell has been grown from frozen tissues that were harvested prior to extinction or subjected to genome editing, the next step is to create an embryo from that cell. This involves using a living host as a surrogate. In many candidate species for de-extinction (I will describe some exceptions later in this chapter), this involves cloning by nuclear transfer. And, as one might anticipate, some candidate species for de-extinction will be considerably easier to clone than others will be. Cloning the bucardo, for example, should be much simpler than cloning edited elephant cells would be. For that reason, I will begin the exploration of this phase of de-extinction using the bucardo as an example. Once the basics have been covered, I will move on to the bigger challenges to be faced when cloning genetically engineered elephant cells. And, finally, I will introduce an obstacle to de-extinction that took me completely by surprise: it is not possible to clone birds.

The Making of a Bucardo

Nuclear transfer is a complicated process with potential disaster lurking in each step. Even what should be the simplest steps can harbor significant obstacles. With dogs, for example, it is nearly impossible to harvest mature eggs—that is, the eggs into which the somatic cell would be transferred—from female dogs. Unlike the eggs of other mammals, which mature in the ovary, dog eggs mature as they move from the ovary into the uterus. Because domestic dogs also tend to have unpredictable ovulation cycles, knowing when to harvest mature eggs requires both careful monitoring of the dog’s hormones and a bit of luck.

The most challenging step in nuclear transfer is, however, reprogramming. During reprogramming, the cell forgets how to be a somatic cell and becomes, essentially, an embryonic stem cell. Only cells that have reset completely can later differentiate into all of the various cell types that make up an organism. This step is, however, particularly inefficient. Incomplete reprogramming is thought to explain why so few embryos develop after nuclear transfer and the high frequency with which developmental defects are observed among those embryos that do develop.

Reprogramming is not the only step that can fail. Even if cells are reprogrammed correctly and develop into viable embryos without developmental defects, the egg may fail to implant in the uterus of the surrogate mom, or the pregnancy may fail after implantation. This could be due to a poor understanding of the reproductive cycle or to some kind of incompatibility between the developing embryo and the surrogate mother. Such incompatibilities are likely to be more common among cross-species clones (including de-extinction experiments, where all or part of the genome is from a different species than the surrogate mother) compared with same-species clones. Also, there is little doubt that experimental manipulation is stressful for the surrogates and that this stress may contribute to the elevated rate of failed pregnancies in cloning experiments.

Anxious Ibexes and a Hybrid Solution

Stress was certainly a limiting factor in the bucardo cloning experiments.

In preparation for working with bucardo cells, the scientific team leading this work first attempted a cross-species cloning experiment with a different and relatively common subspecies of Spanish ibex. Once these technologies had been developed and fully tested, the team would proceed with the bucardo.

The plan required Spanish ibex embryos. To create these embryos, the scientists first had to capture Spanish ibex from the mountains. Then, they needed to rear the captured ibex in captivity to observe their reproductive behavior and develop a way to make the females ovulate. After mating had been observed, the scientists would harvest fertilized eggs, implant the developing embryos into domestic goats, and hope for the best.

Harvesting Spanish ibex eggs turned out to be much more difficult than the team anticipated. Accustomed to climbing the steep faces of rocky slopes, the Spanish ibex escaped manipulation by taking refuge along high ledges in the walls of the animal facility (plate 15). When the team finally harvested their eggs, the eggs were all unfertilized. The animals, it seemed, were too stressed by the captive breeding environment to mate successfully.

The team was able to develop less stressful ways of manipulating the ibex and, eventually, they recovered fertilized eggs from captive Spanish ibex. Any excitement derived from this success was short-lived, however, as another serious setback to the experiment was revealed: none of the embryos continued to develop after implantation in domestic goats. It seemed that the domestic goat uterus was incompatible with the Spanish ibex embryo. This was bad news for bucardo cloning.

Believing that genetics might be the solution, the team decided that a different surrogate mother—one that was genetically more similar to the developing embryo—might be just what they needed. The most genetically similar surrogate would be a subspecies of Spanish ibex. They knew, however, that Spanish ibex were difficult to manipulate and did not fare well in captivity. Not wishing to spend every day coaxing ibex down from the walls, they decided on a compromise: they would create hybrids. Female domestic goats crossed with male Spanish ibex would produce kids with 50 percent Spanish ibex DNA and, most crucially, that would probably keep their feet on the ground. When these hybrid females reached adulthood, they would become the surrogate mothers for the Spanish ibex embryos.

Around a year later, the team transferred Spanish ibex embryos into hybrid goat-ibex females and, again, hoped for the best. Excitingly, half of the embryos established successful pregnancies and developed into healthy Spanish ibex.

I should point out that this success rate—50 percent survival of implanted embryos—is high because this particular experiment did not involve nuclear transfer. This experiment began with healthy embryos taken from living ibex, not with somatic cells that needed to be reprogrammed. As I noted before, this reprogramming step—which is the first step in bucardo de-extinction—has an extremely low success rate.

Unanticipated Barriers to De-extinction

In developing assisted reproduction technology for Spanish ibex, the bucardo-cloning team learned that bucardo embryos, should they get that far in the bucardo-cloning experiment, might develop within surrogate mothers that were hybrids of domestic goats and Spanish ibex, but they were unlikely to develop within purebred domestic goats. The team had discovered a barrier of some sort to cross-species cloning that had arisen during the evolutionary divergence between these two lineages.

Importantly for de-extinction, the probability that barriers such as this may arise increases with evolutionary distance. Extinct species with no close evolutionary relatives might not have any suitable living potential maternal host. The ibex experiment revealed, however, that barriers can also exist between closely related species. Genome editing could even cause barriers—for example, by disrupting important interactions between the embryo and the maternal host. In this way, even de-extinction projects that involve only minimally edited genomes may be frustrated by unanticipated incompatibilities between the developing embryo and the surrogate host.

Some incompatibilities may manifest even before the implantation phase. If, for example, the egg cell into which the nucleus of the somatic cell is injected is incompatible with the somatic cell, then none of these eggs will develop into embryos even if the somatic cell is completely and correctly reprogrammed. Such a problem may arise, for example, if the nuclear genome from the somatic cell is incompatible with the mitochondrial genome in the egg cell.

Mitochondria are organelles that live within the cytoplasm of cells and are not part of the nuclear genome. All of the mitochondria that an organism will have in all of its cells are descended from the mitochondria in the egg cell. Mitochondria have their own genome, and this genome codes for some of the proteins that are necessary for cellular respiration—that is, the process by which cells use oxygen and sugars to make energy. Other proteins necessary for cellular respiration are made by genes in the nuclear genome. Incompatibility between the mitochondrial and nuclear genomes can lead to incompatibilities between these genes. If these genes can’t work together to make the cell respire, this can lead to metabolic disease, neurologic disease, and even death. Thus far, all cross-species cloning has involved the transfer of only nuclear DNA—not mitochondrial DNA.

Researchers in David Rand’s lab at Brown University demonstrated how nuclear-mitochondrial mismatch can produce unusual phenotypes in otherwise genetically normal cross-species hybrids. Rand’s lab created fruit flies with nuclear DNA from Drosophila melanogaster and mitochondrial DNA from Drosophila simulans, two fly species that diverged from each other around 5.4 million years ago. The resulting mismatched-genome flies had whiskery bristles on their backs, were half the length of normal flies, were developmentally delayed, reproduced poorly, and, as might be expected if energy production is off, got tired more quickly than did flies with matched genomes.

Mismatches between the mitochondrial and nuclear genomes might be a problem for de-extinction, but not one without an obvious solution. If the mitochondria don’t match, why not replace them with mitochondria that do match the nuclear genome? Or, why not edit the mitochondrial genome to replace the problematic sites? This could presumably be achieved using the same genome-editing approaches as would be used to alter the sequence of the nuclear genome. Neither of these approaches is simple, and neither is feasible today. Both are, however, theoretically possible.

MAMMOTH PROBLEMS

Now that I’ve introduced some of the challenges to be faced during the cloning and prenatal development stage of de-extinction, let’s return to the mammoth as a specific example. As I discussed in the preceding chapter, we have the technology today to edit the elephant genome so that it contains the mammoth version of least some genes. Assuming that genome engineering takes place in a cell that either is a stem cell or can be reprogrammed to become a stem cell, we are ready to move on to the next step: creating a living animal that contains the edited genome and, hopefully, expresses the traits that we aim to resurrect.

To complete this step, the cell needs to develop into an embryo, and because we cannot grow an elephant in the lab, that embryo needs to be transferred into a surrogate host. Once inside the surrogate, the embryo needs to implant in the uterine wall and establish a pregnancy. The pregnancy then needs to proceed without problems and culminate in the birth of a healthy baby animal whose genome contains several carefully selected and painstakingly engineered mammoth genes.

The simplest way to transform the edited cell into an embryo is to use an egg. We know that eggs contain proteins that activate cells—that is, they reset cells that have already differentiated and turn them into embryonic stem cells. The most appropriate egg to activate our edited elephant cell is, unsurprisingly, an elephant egg. Elephant eggs are not particularly easy to come by. When an Asian elephant ovulates, she releases only one egg at a time. Once released, the egg travels through her reproductive system to the uterus, which is, predictably, elephant sized. An elephant that is not pregnant will ovulate once every two to three months. Given the poor efficiency of nuclear transfer, it’s reasonable to assume that collecting a single egg every two months—assuming we can find that egg within the elephant’s reproductive tract—will not provide enough eggs. We’ll need hundreds, even thousands of elephant eggs for this to work. Frankly, that seems unfair. Elephants are struggling to make enough elephants to sustain healthy populations; the last thing they need is for us to be snooping around their ovaries stealing their precious few mature eggs. In fact, if harvesting mature eggs from adult elephants were the only way to get elephant eggs, my opinion would be that mammoth de-extinction research should stop immediately.

Fortunately, there may be another way. In 1998, researchers at Purdue University and the Advanced Fertility Institute at Methodist Hospital of Indianapolis created mice that could grow elephant eggs. Dr. John Crister, who led the study, wanted to develop a way to increase the reproductive rate of endangered species, and he hoped that coaxing laboratory mice to grow their eggs would be a good start. He and his team transplanted ovarian tissue—the tissue in which immature eggs are found—that Crister collected from three South African elephant carcasses into several laboratory mice. A few of these mice developed egg-producing follicles and, ten weeks later, one of these follicles produced a slightly misshapen elephant egg. Crister and his team did not attempt to fertilize the egg with elephant sperm, so it’s not possible to say whether it would have developed into a viable embryo. It is, however, an optimistic start.

Hopefully, scientists will invent an efficient means to collect a large number of elephant eggs without jeopardizing any elephants. We could then collect a ton (perhaps literally) of elephant eggs, remove their nuclei, and insert nuclei that contain our edited genomes. Then, we would sit back and allow the egg to perform its reprogramming magic. If this goes smoothly and we end up with viable, developing elephant embryos (with slightly modified genomes), we can then transfer these embryos into the uteri of adult female elephants, where they can develop into baby elephants (with slightly modified genomes).

The entrance to an elephant’s uterus is blocked by a membrane called a hymen. In elephants, the hymen stays in place throughout pregnancy, ruptures during birth, and then grows back in preparation for the next pregnancy. To establish a healthy pregnancy in a surrogate elephant mother, the embryo and whatever tool is used to deliver it into the uterus must pass through the only opening in the hymen—a four-millimeter hole designed to allow only sperm to penetrate—without destroying the membrane and thereby compromising the pregnancy.

Let us assume this is possible. Let’s also assume that the pregnancy takes hold, and the embryo begins to develop. The next step is to wait patiently while the pregnancy proceeds. A typical gestation period for an Asian elephant is somewhere in the realm of eighteen to twenty-two months. Hopefully, no compatibility issues will develop between the embryo and the surrogate mother during this time. Hopefully, the surrogate mother’s genetic makeup won’t influence the expression of the genes we changed. Hopefully, her diet, hormones, and stress level won’t alter the developmental environment in a way that influences the expression of the genes that we changed. Hopefully, the birth goes well for both the surrogate mother and the neonate.

Size Matters

In designing cross-species cloning experiments for the purpose of de-extinction, it is important to consider physical differences between the two species involved. Mammoths that lived during the Late Pleistocene varied considerably in size. The largest of these mammoths were about the same size as big African elephants and the smallest were similar in size or even smaller than average-sized Asian elephants. It is not known whether this size variation was genetically determined or simply reflected differences in the amount and quality of available food. Regardless, this variation might be important in choosing surrogate hosts. Interestingly, the two baby mammoth mummies that have been found were both around ninety centimeters tall, which is approximately the same size as a newborn Asian elephant, suggesting that the most closely related elephant species to a mammoth might make a reasonable surrogate.

Physical differences in size can cause problems in gestation and birth, however. Imagine, for example, if sperm from a Great Dane was used to impregnate a Chihuahua. The embryos would begin to develop and fill whatever space was available, but development would stall as they ran out of room to grow. Ultimately, the embryos may die, the mother may die, or both may die. If a natural birth were attempted, the mother would almost certainly suffer terribly. Returning to de-extinction, what might happen if a very large auroch were to develop within a much smaller, domestic cow? Or if a dugong tried to gestate a Steller’s sea cow? Size differences between species, even between closely related species, should definitely be considered when proposing surrogate hosts.

A possible solution might be to make miniature versions of some extinct species. We could identify which genes or suite of genes are most critical to determining body size and tweak these using genome editing. A useful clue about which genes to target could come from genetic analysis of the population of mammoths that lived on California’s Channel Islands. These so-called pygmy mammoths grew to only around two meters tall at the shoulder, compared with four meters or more for mainland mammoths, and probably weighed just under 800 kilograms, compared with 9,000 kilograms or more for mammoths on the mainland. There is one problem with this idea. Tiny mammoths may be easier to gestate, but they might not be sufficiently large to replicate the ecological interactions between normal-sized mammoths and the ecosystems in which these normal-sized mammoths lived. Resurrecting pygmy mammoths therefore might not achieve the environmental goals of mammoth de-extinction.

Another potential solution is to give up entirely on surrogates and instead use artificial wombs. I’m imagining something similar to the artificial wombs that Aldous Huxley envisaged to grow children in his book Brave New World. Or, even better, the giant nutrient-filled flasks in which human clones were grown on the planet Kamino to fight for the good side in the movie Star Wars: Episode II. In the artificial womb scenario, embryos would develop to term in a completely artificial environment—an idea known as ectogenesis. Modern medicine is a long way from functional artificial wombs and successful ectogenesis, but there is little doubt that innovations in this realm would have considerable impact on neonatal and perinatal care. Plus, by using artificial wombs, any animal suffering caused by surrogacy would be avoided entirely. The use of artificial wombs assumes, however, that developing within a real uterus is not critical to normal mammalian development. This is something that science does not yet know.

CLONING IS FOR THE BIRDS (NOT)

Although my focus until now has been on mammoth de-extinction, the present discussion provides an ideal opportunity to shift to the other de-extinction project with which I am involved—resurrecting the passenger pigeon. I hinted earlier that some species would not be cloned using nuclear transfer. The passenger pigeon is one of those species.

Because birds develop on the outside, rather than within the bodies of their surrogate moms, birds would seem to be a good choice for cloning by nuclear transfer. Yet, none of the species listed as having been cloned using this approach were birds. Why is that?

The simple answer is that birds cannot be cloned in this way.

A bird begins its long journey to becoming a bird as a yolk. The yolk is a single unfertilized cell—the oocyte—that lives inside the bird’s ovary. The first step in bird development is to release the yolk into the oviduct. As it begins its journey down this very long, very spirally tube, it meets sperm and is fertilized. Then, for the next twenty-four hours or so, the fertilized egg travels slowly through the oviduct, tumbling around dramatic twists and through spiraled coils. As it bobs along its path, layers of albumin and structural fibers gradually cover the egg. This is the stuff we know as egg white. As it is moving, the fertilized cell starts to divide. The egg tumbles through the oviduct, twisting the structural fibers around the yolk, anchoring it firmly within the egg white. Toward the end of the oviduct, and just before the egg is laid, the hard shell is deposited as the final layer around the developing embryo. By the time it completes the journey from inside its mother’s ovary to the outside world, the embryo comprises around 20,000 cells. These will have begun to differentiate into different cell types.

At what point in this process would it be possible to perform nuclear transfer? In a mammal, the egg whose nucleus is removed and then replaced is collected from the female reproductive tract after it has matured but before it is fertilized. At precisely this stage, the egg is primed to reprogram the nucleus of the somatic cell. It turns out to be extremely challenging to collect bird eggs that are at this stage of development. The reproductive tract in birds is long and sinuous, and the yolk is tricky to recover prior to fertilization. If we wait until the egg has been laid, the cells in the embryo will have already started to differentiate, and the embryo—which is held in position within the egg by many layers of twisted fibers—will be too large to remove. Even if the embryo could be removed and replaced without destroying the egg, the replacement embryo would have to be at the same developmental stage as the egg’s natural embryo. Growing embryos to such a late stage in the lab is also proving to be extremely challenging. For the moment, therefore, it seems that cloning birds may never be possible.

Fortunately, there is another way. When the bird egg is laid, the embryo is still in an early developmental stage. The primordial germ cells—those cells that will later develop into either the sperm cells or the egg cells of the developing embryo—have formed but have not yet found their way to the sex organs, as the sex organs do not yet exist. Around twenty-four hours after the egg is laid, the primordial germ cells migrate through the developing embryo’s bloodstream to the sex organs (which are now starting to develop), where they settle in until the time comes when they mature into sperm or eggs.

Primordial germ cells are the key to genetically manipulating birds. Primordial germ cells can be grown in a dish in the lab, which makes their genomes accessible for editing. Primordial germ cells are also tiny, which means they can be injected into the egg during that second twenty-four-hour window of development during which the egg is on the outside and the primordial germ cells are making their way to the bird’s developing sex organs. In this way, the injected edited primordial germ cells will travel with the embryo’s primordial germ cells to the sex organs. When these cells mature, the edited cells will participate in making the next generation of birds.

When the chick hatches from the egg into which the genetically modified primordial germ cells were injected, that chick itself will not be genetically altered. Instead, the genetically altered cells will be hiding out in its sex organs. The first time the genetically altered genes will be expressed will be when that chick grows up and has its own baby chicks.

Let’s walk through how this process would work for passenger pigeon de-extinction. Band-tailed pigeons are the closest living relative of passenger pigeons. The intention of the passenger pigeon de-extinction project, although these experiments have not yet begun, is to create band-tailed pigeons that look and act like passenger pigeons. To achieve this, we will isolate primordial germ cells from band-tailed pigeons and grow these in the lab. We will then edit the genomes within the primordial germ cells using the genome-engineering technologies described several chapters ago, replacing band-tailed pigeon genes with the passenger pigeon version of these genes as appropriate. We will then inject these edited band-tailed pigeon primordial germ cells into developing band-tailed pigeon eggs at precisely the right time during development. The chicks that are born when these eggs hatch will be genetically pure band-tailed pigeons, except that some of their germ cells (sperm or eggs) will contain passenger pigeon DNA. The offspring created by these edited germ cells will contain passenger pigeon DNA throughout their bodies.

Cloning by Germ Cell Transfer

Cloning by transferring germ cells into a developing embryo has one important advantage when compared with cloning by nuclear transfer. Edited primordial germ cells do not need to be reprogrammed. This is huge. So why has all the focus been on cloning a mammoth, when cloning passenger pigeons or dodos would apparently be so much simpler?

It is not entirely clear why cloning birds has received far less attention than cloning mammals has. Primordial germ cell transfer works remarkably well as a means to genetically modify birds. The technology has been developed mainly with the chicken industry in mind, but it has been used both for conservation purposes and in pure scientific research. There is no obvious reason to suspect it would not work well for the purposes of de-extinction.

Some of the applications of primordial germ cell transfer are, admittedly, unusual. The Roslin Institute, the facility that was responsible for cloning Dolly, has used the technology to create chickens that glow a bright green color under ultraviolet light. To make their chickens glow, they insert a protein into their genomes called green fluorescent protein, or GFP, which is found naturally in the North American jellyfish Aequorea victoria. The scientific community uses GFP to track biological changes within an organism. For example, if tissues whose cells express GFP are grafted onto an organism whose cells do not express GFP, scientists can track what happens to the grafted cells by watching them under fluorescent light. Scientists interested in using glowing chickens for their research can go to the Roslin Institute’s Web site and order them. For now, there is no charge.

In addition to making chickens glow, the technique of injecting primordial germ cells into developing embryos has been used to boost the population sizes of rare or endangered chicken breeds. Primordial germ cells can be harvested from the blood of developing embryos without killing the embryo. These cells can then be kept alive in the lab and introduced into the developing embryos of common breeds. When these birds reach sexual maturity, they can then be fertilized with sperm (which are much easier to collect than eggs) from the rare breed. When these sperm fertilize eggs that develop from the injected primordial germ cells, the result is a purebred rare-breed chicken that hatches from an egg laid by a common chicken.

The most exciting application of primordial germ cell transfer from the perspective of bird de-extinction has been the successful transfer of primordial germ cells between species. Scientists at the Central Veterinary Research Laboratory in Dubai injected primordial germ cells from a chicken into duck eggs. When the ducks hatched from these eggs, they looked like perfect ducks. Remember, only the germ cells are different in the first generation. The scientists later harvested sperm from these ducks, and used those sperm to fertilize a hen. When the eggs laid by this hen hatched, perfect chickens were born. With a duck for a dad.

Fascinatingly, ducks and chickens are not the only animals that have been coaxed to give birth to a different species using this approach. Recently, Professor Goto Yoshizaki of the Tokyo University of Marine Science and Technology transferred rainbow trout eggs and sperm into the reproductive tracts of adult Masu salmon. When these adults mated, some of their eggs hatched into rainbow trout. Rainbow trout and salmon are closely related species, which may explain the experiment’s success. However, there is hope that the technique can be extended to other fish species. Yoshizaki also produced tiger puffer fish using grass puffer fish and intends to use mackerel to produce bluefin tuna, which, if successful, would provide an inexpensive way to increase tuna production without removing juveniles from the wild.

Germ cell transfer is certainly an exciting technology, and one that may have a variety of uses in conservation biology. There are some drawbacks, however, to using germ cells for the purpose of de-extinction.

First, primordial germ cells are haploid; they either become sperm or eggs. When a sperm with an edited genome fertilizes an egg that does not have an edited genome (or vice versa), the offspring’s diploid genome will have only one copy of the edited gene. The edits, therefore, may not be visible in the offspring’s phenotype. To produce offspring with two copies of the edited gene, genomes from both the sperm and the egg have to be edited.

Second, the injected primordial germ cells are not the only primordial germ cells that make it to the sex organs. In the duck example above, the duck was the dad of the chicken, but his sperm were chimeric—some of his sperm were duck sperm and other sperm were chicken sperm. When his duck sperm fertilized a chicken egg, nothing happened. No hybrid “duckens” were born. But, when his chicken sperm—those of his sperm that were derived from the chicken primordial germ cells that were injected into his egg while he was a developing embryo—fertilized a chicken egg, a chicken was born. That chicken had a genome that was 100 percent chicken-like. But its father was, nonetheless, a duck.

Third, in the experiments that have been done thus far, scientists observed that the efficiency with which the injected primordial germ cells go on to become the next generation is poor. Only a small fraction of the eggs and sperm that are eventually made by the embryos develop from the injected primordial germ cells.

Mike McGrew of the Roslin Institute has a plan to overcome these obstacles. He is genetically engineering chickens that cannot make primordial germ cells. The only way these chickens would make eggs or sperm would be if primordial germ cells were injected during the appropriate developmental stage. In this way, he can produce hens in which 100 percent of eggs contain the edited genome, and cockerels in which 100 percent of sperm contained the edited genome. Mating these together would result in offspring that are 100 percent genetically engineered.

While there has been some success in transferring primordial germ cells between distantly related bird species, I imagine that there are still limits to how far this can be taken. Chickens, for example, may struggle to (and probably should not be caused to) lay eggs that contain developing moa or elephant bird embryos, for example. And there is little doubt that the hormonal and genetic environment within the mother—even for just the first twenty-four hours of development—plays some role in early embryonic development. This technology is exciting, however, and will certainly find use in the preservation of avian biodiversity, at the very least among chicken breeds.

And perhaps someday a chicken will be persuaded to lay an egg that contains a baby dodo. If that were to happen, the next question might be, just what is that chicken going to do with a baby dodo?