This piece was co-authored with @Metacelsus
“For a successful technology, reality must take precedence over public relations, for nature cannot be fooled.”— Richard Feynman
The black-faced honeycreeper — or po’o-uli, as the Hawaiians called it — was native to Maui. Two birds were spotted in 2000 and then never seen again. The species was declared extinct in 2019; it lives on only as cells in a cryogenic freezer.
Or, consider the Pyrenean Ibex, a large goat native to Andorra’s mountains. The Ibex is the only animal to go extinct twice: Once in January 2000 and again in July 2003, after scientists cloned an animal and watched it die, shortly after birth, from a lung defect.
Hundreds of other plants and animals are listed as “possibly extinct” by the International Union for Conservation of Nature, which means they haven’t been seen in the wild for years. The list includes 156 amphibians, 22 birds, 29 mammals, and 93 insects. Captain Cook’s Bean Snail, the Wyoming Toad, and the She Cabbage Tree are all extinct in the wild. Today, they live only in zoos.
Numbers alone can’t convey the Sixth Extinction, because numbers alone aren’t inherently visual. Each digit is an entire species, with thousands or millions of years of history, that is at risk of disappearing from the Earth. We mask extinction by visiting zoos or fawning over new technologies. “Humans made this mess,” we think, “and humans will fix it.”
De-extinction, unfortunately, will not be our savior.
Many academic scientists and companies aim to ‘resurrect’ long-extinct animals using gene-editing and advanced reproductive technologies. One of them, Dallas-based Colossal Biosciences, recently raised $150 million to bring back the woolly mammoth, dodo, and thylacine, a fox-like marsupial that once roamed Tasmania. The dodo bird, native to Mauritius, was hunted to extinction by Dutch sailors in 1662. Colossal launched less than a year-and-a-half ago, and yet is already valued at over $1 billion.
The company, then, is a unicorn building technologies to make unicorns.
Press releases, though, won’t make de-extinction come true. No papers have been published by the company, and resurrecting the Pyrenean Ibex, which ended in failure, was massively simpler in comparison; living cells were already available. True de-extinction is nowhere near possible.
“It’s impossible to bring something back that’s an identical copy to something that used to be alive,” said Beth Shapiro, a Colossal advisor, in an interview for Fast Company. Instead, “we are going to be able to bring back traits and behaviors and characteristics of extinct species that I think we can use to revitalize and reinvigorate existing ecosystems.”
But even that — making hybrid animals, like elephants with mammoth genes — is plagued by technical issues. It is still difficult to make ultra-precise genome edits, for instance, and there are vast gaps in our understanding of genetics and development, especially for elephants and birds.
This essay breaks down de-extinction technologies, step-by-step, to separate facts from fiction.
George Church, a Harvard geneticist, first pondered de-extinction in the mid-2000s. Nicholas Wade, a journalist at The New York Times, was an unlikely source of inspiration. In 2008, as DNA sequencing costs were falling, Church’s lab in Boston was devising a method to make many edits across a cell’s genome. In that same year, scientists uncovered “a large fraction of the mammoth genome” by sequencing DNA isolated “from clumps of mammoth hair.” These three advances, together, made de-extinction — which once seemed impossible — seem somewhat plausible. The ever-charismatic Church even put a price tag on bringing back the mammoth: $10 million.
“This is something that could work,” he told Wade, the Times reporter, “though it will be tedious and expensive.” The estimated budget was far too small, of course, and the timelines were way off. But here, finally, was a spark toward a blue-sky, scientific aim.
Journalists have written about mammoth de-extinction every single year since 2008. And, although Church is the face of the mammoth project, de-extinction was, until recently, the least funded of all his group’s efforts. Scientists in Melbourne have been working to resurrect the thylacine for many years, too. They’ve already sequenced the genome from a museum specimen that was preserved in alcohol for 110 years. A nonprofit, called Revive & Restore, also aims to de-extinct the passenger pigeon, which once numbered five billion but crashed to zero within a 40-year span. Martha, the last of her species, died in the Cincinnati Zoo in September 1914.
Each of these projects — the mammoth, dodo, pigeon, and thylacine — has the same problem: True de-extinction is not going to happen. To understand why, let’s break down each step in the woolly mammoth project and look at the technologies required.
Claim: The first step in de-extinction is to collect biological tissue, or cells, from the extinct organism, and then sequence their DNA. Well-preserved mammoth tissues have been retrieved from melted Siberian permafrost. After sequencing the extinct species, one must also sequence the genome of their closest living relative. For the mammoth, that’s the Asian Elephant. For the thylacine, it’s the dunnart, a mouse-y marsupial in Australia that weighs about 1,000-times less than its extinct relative.
Response: Sequencing is the easiest step. While too much time has passed to recover dinosaur DNA from fossilized amber, the woolly mammoth lived much more recently and so it was possible to isolate DNA from specimens stored in museums (or permafrost). Although the mammoth DNA has degraded somewhat, modern sequencing technologies have enabled the assembly of genome sequences for the mammoth, dodo, and thylacine.
A high-quality mammoth genome was published in the journal Current Biology in 2015, based on DNA harvested from two woolly mammoths that died 4,300 and 44,800 years ago. Beth Shapiro, the Colossal advisor and an evolutionary molecular biologist at the University of California - Santa Cruz, also sequenced the dodo genome last year, though we’re not aware of a published manuscript. The thylacine genome was sequenced by Andrew Pask and colleagues in Melbourne. Surprisingly, Asian elephants were first sequenced several years after mammoths.
Disclosure: Metacelsus’ research on ovarian organoids was partially funded by Colossal Biosciences. Colossal was not involved in writing or editing this article.
Claim: After sequencing the genomes, the next step is to compare them and decide which genes should be modified to turn one organism into another. For the mammoth project, scientists have identified traits that could be added to Asian Elephants to enhance cold resistance, such as “smaller ears, shaggy fur, hemoglobin adapted to cold, and excess fat tissue.” It’s expected that somewhere between 50-100 precise genome edits will be required to make an “Arctic Elephant” — basically an Asian Elephant with some added mammoth DNA.
After selecting traits to modify, the next step is to edit the genome. This is not the most difficult part of de-extinction, but it’s still incredibly challenging. The general plan is to use CRISPR-Cas9 gene editing to “cut and paste” regions of the genome, slowly turning an Asian elephant cell into an elephant-mammoth hybrid. Church says that they will edit both protein-coding and regulatory genes.
Response: Reading DNA is simple; precise editing is not. It is likely impossible to create a woolly mammoth genome in any reasonable timeframe (e.g. 10 years or less) by stitching together big chunks of synthesized DNA. The yeast genome is just 12 million bases in length, for example, and efforts to synthesize it have been ongoing for more than eight years. The (haploid) Asian elephant genome is 3.94 billion bases, or 328 times larger. Tools to assemble chunks of DNA also don’t work well in animal cells, and so entirely new methods would be required to actually build a mammoth genome.
This is why de-extinction scientists are not synthesizing entire, ancient genomes de novo, but rather editing the genomes of living relatives, such as the Asian elephant. The woolly mammoth genome differs from the elephant genome at about 1.4 million sites, and the two animals have roughly 2,000 distinct, protein-coding genes. Colossal Biosciences has obtained Asian elephant cell lines, and their scientists are currently editing them to be more like the mammoth.
Current gene editing technology, of course, cannot edit all these sites at once! State-of-the-art editing technologies can, at most, target a few dozen sites in the genome, with efficiencies at each site ranging from 5% – 80%. “A lot of the edits we have to make are precise,” says Church, “and precision editing is not a healthy field yet.”
Making 50–100 edits to the Asian elephant genome, then, will require multiple rounds of experiments. Researchers also cannot recombine animal chromosomes in cell culture, so each round of editing must be done sequentially, and not in parallel. If each round of editing takes about two weeks, and (optimistically) assuming that 5 edits are made each round, it would take about 8 years to modify just the protein-coding genes, and about 5,000 years to do the rest of the genome.
If all goes to plan, the best possible outcome would be a hairy, cold-tolerant “Arctic elephant;” not a true woolly mammoth. There is also the question of whether the 50 genome changes will actually have the intended outcome. Some traits, like cold resistance, could possibly be tested in cell culture. But it’s impossible to know what the hybrid animal will actually look like until it’s born and grows up. Elephants have a gestation period of at least 18 months, and that’s a long time to wait for uncertain outcomes.
“Some tests will require that elephants be born and walk around in the snow in minus-40 degrees plus,” says Church, while others traits, like hair and cold resistance, could perhaps be tested using “teratomas or possibly in vitro differentiation.”
Claim: After editing the genome of an Asian Elephant cell, the next step is somatic cell nuclear transfer, or SCNT. Colossal’s website explains: “The nucleus from a donated Asian elephant egg is removed, and the hybrid nucleus, which is the Asian elephant nucleus edited with the woolly mammoth DNA, is inserted in its place. Electrical pulses are applied to the egg to stimulate fertilization. The egg then begins to divide and grow into an embryo.”
SCNT has been used to make cloned cows, pigs, horses, cats, dogs, monkeys, and sheep. The technique has never been tested on elephant cells.
Response: In vitro fertilization, or IVF, is the simplest reproductive technology. It works by combining sperm and eggs in a test tube to make a fertilized embryo. And yet, despite its simplicity, IVF has never been done with elephants.
Why? Because it’s difficult to harvest eggs from massive, endangered animals! The ultrasound devices that are normally used to pinpoint eggs in, say, a human, do not work in an elephant — their ovaries are too deep within their bodies. Fertile, female Asian elephants are also a bit of a rarity; the species numbers just 50,000 in the wild.
Assuming that you had some kind of ‘egg-retrieving elephant farm,’ there’s also no guarantee that one could collect enough eggs for SCNT. You’d first have to perform superovulation, or use hormones to stimulate the ovaries to release more than one mature egg at a time. This works well in humans and mice, but we have no idea whether this would work in elephants, says stem cell expert Sergiy Velychko, which have a hormonal cycle distinct from other mammals.
In the early days of SCNT, the success rate in well-studied animals, like mice, was one live clone from 277 embryo transfers, or about 0.36 percent. The current success rate for SCNT is between 5 and 10% for those same animals. ViaGen, a pet cloning company, charges $50,000 for their services because it is really expensive to create and implant the dozens of SCNT embryos required to achieve a healthy birth.
To make a viable Arctic Elephant, then, one would need to collect dozens (hundreds?) of eggs from an endangered species using technology that does not yet exist. “I think promising to create a woolly mammoth right now is like promising interstellar travel when we can’t even travel to Mars yet,” says Velychko.
SCNT would be even more challenging for the other species, such as the thylacine and dodo, because it has never been done in marsupials or birds. Simply growing marsupial embryos in vitro is highly challenging, according to a 2019 review, because the embryos are extremely fragile and require a shell coat for most of development.
For the dodo de-extinction project, Colossal plans to edit primordial germ cells (or PGCs, which are early precursors of eggs and sperm) and transplant them into a recipient embryo. There is some precedent for producing sperm from gene-edited PGCs in chickens and quails, so it’s not a crazy idea. Still, developing PGC culture conditions for a new species is not trivial. PGCs are unable to form sperm if their growth conditions are not exactly right. Embryo culture and PGC transplantation are also challenging and, since sperm would be produced instead of an embryo, multiple rounds of breeding would be required to generate an animal with the desired genome.
Claim: After SCNT, grow an embryo with the elephant-mammoth hybrid genome, and then implant it into a surrogate African (not Asian) elephant. The gestation period, for these animals, is between 18 and 22 months. A newborn would “be a hybrid with genetic traits from the extinct Woolly Mammoth and the Asian Elephant, its living relative.”
Response: We’ve already described the absurd difficulties in getting Asian elephant eggs. But that’s not the biggest problem, according to Church: If the success rate of SCNT for pigs is about 10 percent, you would need “8 shots on goal” to make one pregnancy, he says. But pigs have a gestation period of just 4 months, and elephants carry just one embryo at a time and have a gestation period that is five-times longer!
Assuming an initial SCNT success rate of 1 percent, one would have to implant dozens of fertile Asian elephants to get just one viable birth. “This is not something that you can just throw money at,” says Church, “because there’s a limited number of fertile Asian elephants in the world. They’re an endangered species.”
Perhaps that’s why de-extinction scientists plan to implant embryos in African, rather than Asian, elephants. There are roughly eight times more African than Asian elephants in the wild, and so it will be easier to find surrogates. But African and Asian elephants are more distant relatives than humans and chimps, and so there’s no guarantee that an edited Asian Elephant embryo, implanted into an African Elephant, would even survive.
Church and other scientists are working on an alternative solution: An artificial womb that could nourish and grow the embryo during its development period, thus taking the onus away from endangered elephants. “Artificial wombs are even higher risk in a certain sense,” says Church, “but at least you can throw money at it.” There’s just one problem: “Nobody has ever gotten any ex vivo tissue to survive for more than a week outside the body, and we need to do it for 22 months.”
An artificial womb doesn’t yet exist for mice, and we’ve been studying them — with hundreds of billions of dollars in funding — for more than 100 years. The most advanced technology, to date, can grow non-implanted mouse embryos for the first 9.5 days of development in little flasks. Other work has focused on the end of development. Modern artificial wombs can help save human babies that are born up to 20 weeks prematurely.
The problem is that these devices are not “designed to dovetail,” says Church. Nobody has made an artificial womb that can do both early and late development. “You need an umbilical cord for a few weeks of gestation,” he says. “Without a cord, you’re pumping precious stem cells into the media…There are all kinds of cells that circulate in fetal blood, which is fine if you have a tight circuit through the placenta, but not fine if working with a giant vat.”
Perhaps it will be simpler to make an artificial womb for the thylacine, because these extinct marsupials spend just 12 days in the uterus during development, and then receive their nutrients through milk. But Church says the thylacine project “is distracting” and uses “probably less than 5 percent of our resources.”
De-extinction groups often talk about how they are BRINGING BACK THE MAMMOTH! OR THE DODO! OR THE THYLACINE! When, in reality, they are trying to make elephant, pigeon, and dunnart hybrids.
Of all these de-extinction projects, the mammoth is the most likely to succeed. Asian elephants and woolly mammoths are closely related. Projects to de-extinct the thylacine and dodo are less likely because of poorly-developed basic science and challenges in collecting unfertilized eggs.
If scientists make a mammoth-elephant hybrid, they will be released into game reserves in Africa and America. They would have hundreds of square kilometers of space, says Church, and would be strategically placed on carbon-rich, endangered soil “where methane could be released any day now.” The animals would eat plants to keep the soil “cold and grassy,” thus protecting the methane reserves.
Even if de-extinction never happens, the money hasn’t necessarily gone to waste. With $150 million on hand, Colossal is developing more precise gene-editing tools, crafting artificial wombs that could one day increase the survival rate of premature infants, and studying endangered species that have long been neglected. Elephants have strong cancer resistance, can live far longer than most mammals, and don’t experience neurodegeneration — we’ll learn a lot by studying them!
For the other technologies — like artificial wombs and gene-editing — the commercialization strategy seems to be to get approval to use them, first, in animals. Veterinary products are sometimes approved five-times faster than human products, says Church, and so reproductive technologies will first be used on mice, and then elephants, and then people. Mammoth de-extinction, in a sense, is a flagship project that unites many technologies into a single vision. “It’s a very charismatic project, it has bipartisan support, and it has indigenous people support,” says Church. “There’s something magical about it.”
Although we’re critical of the technology and timelines behind de-extinction (it’s a bit like Musk’s promises of full self-driving Teslas), we’re not skeptical of the conservation aims. Critics say that Colossal’s money would be better spent in protecting existing species, rather than de-extinct animals that have been dead for thousands of years. Maybe that’s true, but it’s extremely unlikely that this $150M would have gone to conservation otherwise! If rich people want to spend their money on de-extinction, and bolster reproductive technologies in the meantime, then we say “let ‘em.”
Ben Lamm, Colossal’s CEO, has also said that the company will make all patents freely available to conservation groups. Church says the company will publish all findings in scientific journals. (Although Church hasn’t published anything specific to elephants, his laboratory has pioneered tools to make multiple edits across the genome, reprogram tissues and gametes, and has spun out a company, called eGenesis, that made pigs with 42 genome edits.)
Regardless of whether you’d put your money on de-extinction or not, these projects will not fail due to a lack of money or talent. The cause, at least, is worthy. And if there’s anyone on Earth who can prove us wrong, it’s George Church.
Ok let's collapse this claim to the part that is relevant.
Because animals lose information with each generation (due to no written or symbolic language with error check symbols), copying their gene pool exactly would de-extinct the species. We do not have full gene pools for extinct species but usually a few individual samples. So perfect de extinction is impossible.
Synthetic DNA printing is not up to the task at present of printing full scale genomes. So for now, no de extinction projects can be carried out.
Some scammers claim they are doing it, but due to paragraph 3 we can ignore.