In 1932, Heinz Heck declared his back-breeding experiment a success. A bull was born that he felt looked similar enough to what he believed an auroch should look like that his new bull could be called an auroch. According to Heinz’s records (which he stopped keeping after the birth of this bull), the bull was 75 percent Corsican cattle, 17.5 percent Gray cattle, and the remaining 17.5 percent was a mix between Scottish Highland, Podolic Gray, Angeln, and Black Pied Lowland cattle. Selective breeding continued after the birth of this bull, eventually giving rise to what is today known as Heck cattle. Around two thousand Heck cattle are alive today, living in zoos and roaming pastures, mostly in Europe.
Are Heck cattle aurochs? Heck cattle certainly look primitive, particularly to someone who (like the Heck brothers) might not have access to accurate reconstructions of real wild aurochs. Heck cattle have dark coats and long, curved horns, which are two characteristics that were definitely found in wild aurochs. Heck cattle are also more cold tolerant than many other domestic breeds and can survive under relatively poor forage conditions, much as their wild ancestors must have done during Pleistocene glacial cycles. That, however, is about where the similarity ends. Heck cattle are large for domestic cattle, but not as large as the average auroch bull would have been. A Heck bull stands around 1.4 meters high at the shoulder and weigh up to 600 kilograms. An auroch bull, with a shoulder height around 2 meters, would have been taller than the average European man. Also, while the coat color of Heck bulls is similar to what we believe was characteristic of auroch bulls, Heck cows are lighter and more variable in color than auroch cows were. The overall body shape of Heck cattle is also different from that of aurochs, mainly in that it is smaller and, like all domestic taurine cattle, lacks the prominent neck musculature of the wild ancestors. Finally, while the horns of Heck cattle are long, relative to those of domestic cattle breeds, their shape and curvature are somewhat different from an auroch’s: they curve slightly too close to the head and point a little bit too far outward.
It is safe to conclude that the Heck brothers did not quite hit the mark. This failure, however, does not spell doom for the present back-breeding project. Today, we know much more than the Heck brothers knew in the early twentieth century about what traits defined aurochs. We have better descriptions of the various breed phenotypes and a better understanding of the temperaments of these breeds. We have abundant genetic data that help to determine which breeds are the most primitive. We even have ancient DNA data from actual aurochs. Using all of these data, there is little doubt that we will make different and more scientifically justified choices about what animals to use in the back-breeding project, which will eventually lead to the birth of animals that better resemble wild aurochs.
Of course, these animals will not actually be aurochs. Not exactly, anyway. Selective breeding is a process by which individuals that display the desired phenotype are bred together to try to replicate that phenotype in the next generation. The phenotype, however, is a consequence of the interaction between genotype and environment. Genetically, the gradual concentration of genes that code for auroch-like traits has to happen by chance. When the gametes—the sperm or egg cells that will go on to become the next generation—are formed, each one contains a shuffled version of that organism’s parents’ genomes. This shuffling of genetic material, called recombination, is an important source of genetic variation within populations. Recombination shuffles genes or parts of genes from mom’s chromosome onto dad’s chromosome and vice versa. When the sperm or eggs are formed, they will contain some DNA from mom and some DNA from dad. If a phenotype that we want to select is coded for by a gene that came from mom, but the egg that is fertilized contains dad’s version of that gene, then the offspring, despite our best intentions, will not display that phenotype.
We can guide the process of concentrating specific traits into a single lineage by selective breeding, but we cannot selectively choose which gametes go on to become that next generation. Some offspring will get the right genes and display the desired phenotype, and others will not. This does not mean that the process will never work. It will, however, be slow. Selecting for multiple traits simultaneously will be particularly challenging, as the genes for each trait need to wind up, by chance, in the same fertilized egg. Despite this, selective breeding is and has been a powerful tool in our species’ history, as attested by the variety of forms of domestic plants and animals that we encounter every day. There is no reason why, with sufficient time, resources, and patience, we cannot recover at least some traits of the wild auroch using the selective breeding approach.
As the auroch back-breeding experiment proceeds, I anticipate that the animals will gradually become more and more auroch-like in their physique and behavior. Some auroch traits may, however, never be recoverable from living cattle breeds. The DNA sequence that coded for a particular trait may have been lost, for example, or the trait may have been the product of a genetic interaction with an environment that no longer exists. Some people (myself included) would argue that this does not matter—that by filling the niche of the auroch, even partially, the experiment is a success. De-extinction purists will, however, never be satisfied with a back-breeding product, because the result will always be something new, not something old. Auroch 2.0 will not be an auroch. Not precisely, anyway.
IS SIMPLER NECESSARILY BETTER?
One advantage to back-breeding as a means of de-extinction is that it relies so little on molecular biology technologies. Genomes don’t have to be sequenced, genes don’t have to be identified, and genetic variants don’t have to be linked to specific traits. The gradual transition from one form to another happens without embryonic stem cells and long hours spent in a lab. And the results of the experiments are validated qualitatively: does it or does it not look more like an auroch?
The simplicity of back-breeding, however, may also be its downfall. While traits such as dark coat color, long forward-facing horns, or strongly expressed neck and shoulder musculature may emerge in the population after some generations of selective breeding, the genes that code for these traits once the traits reemerge may be different genes from those that coded for the same traits in the extinct species.
Does it actually matter? If we want long forward-facing horns, and the bull has long forward-facing horns, does it really matter what specific genes are making it happen? It might matter. Genes don’t always, or even often, have just one function. A gene that makes curved horns might have other consequences on the resulting cattle phenotype that we don’t want. It might make their skull slightly differently shaped, for example, or somehow influence the shape or texture of their hooves. In addition, genes don’t act in isolation but instead act in concert with other genes that are also expressed in the cell.
An example of an interaction between genes that is used in introductory biology classes is the way that coat color in horses is determined. Horses have a single gene that determines whether their coat will be red or black. The dominant allele produces a black coat and the recessive allele produces a red coat. If this gene acted alone, individuals that carried either two copies of the dominant allele or one dominant and one recessive allele would have black coats and individuals that carried two copies of the recessive would have red coats. However, there are many different types of red or reddish horses. This comes about because of yet another gene—the cream dilution gene—that modifies the expression of the red alleles. A horse that has two copies of the recessive red allele can be chestnut colored, palomino, or even white or cream colored, depending on how many copies of the cream dilution allele it carries.
While not all interactions among genes are known, and very few are well understood, this does not mean that selective breeding for specific traits is impossible. Through multiple generations of back-breeding, using different crosses involving different individuals or different breeds, the right combination of genes, or at least combinations of genes that provide the right phenotypes, may eventually be disco
vered. How long it will take depends on several factors, including how many traits are being selected, how easy the animals are to breed, and how long it takes to go from one generation to the next.
TOO SLOW FOR SUCCESS?
The generation time of cattle is short compared to many species. Female cattle can breed for the first time when they are between one and two years old, and gestation takes around nine months. A selectively bred individual can be born, develop into an adult, get pregnant, and give birth to the next generation all within two to three years. While not a breakneck pace, one can imagine how a selective breeding program could progress reasonably quickly in cattle.
Progress would be much, much slower for some of the other candidate species for de-extinction. For example, male elephants begin making sperm between ten and fifteen years old, and female elephants in the wild will become pregnant for the first time around age twelve. Gestation time in elephants is between twenty and twenty-two months. That means there would be a fourteen-year wait between when the first selectively bred offspring is born and when that offspring can produce the next generation. At that pace, only five generations could be produced in a human lifetime. There must be a better way.
Of course there is. An easy way to minimize the time it takes to selectively breed a trait into a lineage is to make sure that every individual in the next generation contains the target trait. This is not possible with back-breeding, where the offspring of two parents may or may not inherit the target trait or traits. However, new technologies—specifically, the genome engineering technologies that are behind the second presently feasible (and the more magical) pathway to de-extinction—make it possible to edit the genome directly. By manipulating the DNA sequence in a cell and then using that cell to create living individuals, we can be certain that the target trait is present in the next generation. We can make the entire process of resurrecting extinct traits in living species move along much more quickly and efficiently.
For example, we know that mammoth hemoglobin—the protein in red blood cells that takes up oxygen in the lungs and then distributes it via the circulatory system to the rest of the body—differs from elephant hemoglobin by exactly four mutations. These four differences modify the performance of the hemoglobin by making the mammoth version more efficient than the elephant version at delivering oxygen when the temperature in the body is very low (think mammoth feet standing in the snow).
We will not find a living elephant that has the mammoth version of these hemoglobin genes. The common ancestor of mammoths and living elephants lived in the tropics, and adaptations to life in the cold would have evolved in mammoths only after the mammoth lineage diverged from the Asian elephant lineage. Since all mammoths are extinct, there are precisely no individuals alive who have these particular genes. In order to create an elephant that makes mammoth hemoglobin, we will have to make the mammoth version of those genes from scratch and then somehow insert that version of the gene into an elephant cell. We can do that.
CHAPTER 6
RECONSTRUCT THE GENOME
In 2010, J. Craig Venter made life from scratch. He and his team synthesized the complete genome of a tiny, free-living bacterium, which they named Mycoplasma mycoides JCVI-syn1.0, and transplanted it into a recipient cell whose own genome had been removed. In addition to stringing together all of the genetic bits and pieces required to make the genome function (which it did) and the cell replicate (which it did) they added watermark sequences—the names of the researchers involved in the project translated into a genetic code—to distinguish the synthetic genome from the real genome on which the copy was based.
Venter and his team began the life-creating process by learning the complete genome sequence of a living Mycoplasma mycoides bacterium. This digitized genome, which was nothing more than lines of text stored in a file on a computer’s hard drive, became their blueprint for constructing life. They chose this particular bacterial genome because it was short—a little over a million base-pairs long—and because the bacteria grew quickly, which meant that it would not take a long time to complete the experiment.
One million base-pairs is a very small genome, even for a bacterium. It is not, however, sufficiently small to synthesize all at once. When strands of DNA are produced in a lab, machines do this by stringing together single nucleotide bases—the As, Cs, Gs, and Ts that make up entire genomes—in order. The longer the fragment is, the more mistakes will be made during synthesis. If this bacterium was going to be able to survive and reproduce, they would have to make a synthetic genome that was as identical to the blueprint genome as was possible.
To get around the problem of synthesizing long fragments, Venter’s team designed a four-step process to construct the complete genome. First, they synthesized, one base-pair at a time, 1,078 fragments of DNA that were each 1,080 base-pairs long. These fragments were sufficiently short to construct reliably in the lab, but also long enough to each contain unique identifying information that would be used to orient them correctly in the final genome. Then, taking ten fragments at a time that they knew were adjacent to each other in the blueprint genome, they inserted these smaller fragments into yeast cells and allowed the yeast’s cellular machinery to stitch these fragments together. That process provided 100 fragments of bacterial DNA that were each around 10,000 base-pairs long. They then took ten of these at a time and stitched them together, making eleven fragments of around 100,000 base-pairs long. Finally, they stitched these eleven fragments together to create a single million-base-pair-long bacterial genome. They removed this genome from the yeast cell and inserted it into a bacterial cell, where it began to make all the proteins necessary for life. The entire process took fifteen years and cost more than US$40 million.
Creating the first synthetic life was an awesome accomplishment. It does not, however, get us any closer to being able to create mammoths or passenger pigeons. First, bacteria are prokaryotes, which means they lack a nucleus. Because of this, Venter and his team were able to skip an important and as yet unsolved step in the life-creation process: they did not have to assemble a genome comprising multiple distinct chromosomes within a nuclear membrane, as would have been necessary to create a eukaryote. Until somebody figures this out—and I’m watching this space with a close eye on the J. Craig Venter Institute—there will be no mammoths or passenger pigeons running around with completely synthesized genomes. Second, bacterial genomes are small. The mammoth’s genome is more than four billion base-pairs long. Birds tend to have much smaller genomes than mammals, but even their genomes are mostly more than one billion base-pairs long. Not all of these base-pairs code for genes that make proteins, but we still don’t really know how much of the other stuff in a genome is essential for life. More importantly, we do not and probably cannot know the complete sequence of any extinct species’ genome. Even if scientists were to discover a means to synthesize an entire eukaryotic genome within the nucleus of a cell, we may never have the template for that synthetic genome.
Let’s take a closer look at the mammoth. Over the years, ancient DNA scientists have sequenced billions of base-pairs of mammoth DNA from dozens of mammoth bones and other remains. The fragments of DNA that are recovered from these tend to be short—somewhere in the range of thirty to ninety base-pairs long—and they are damaged, as is expected for very old DNA. Returning to the puzzle analogy from chapter 2, our mammoth DNA fragments are the puzzle pieces and the picture on the box top is of the African elephant; while we know from comparing their mitochondrial DNA sequences that the Asian elephant is more closely related to the mammoth than the African elephant is (figure 9), thus far only the African elephant nuclear genome has been reconstructed, so only the African elephant genome is available to act as a guide. Also, only about 80 percent of the African elephant genome is known, so there is something not quite right about the photo on the box top. In essence, we have billions of microscopic, slightly misshapen puzzle pieces and a slightly blurry photographic key that solves a differe
nt puzzle.
The easiest pieces of the puzzle to put together will be those that come from the most conserved regions of the genome. These are the parts of the genome where mammoths and both species of living elephant, and indeed all mammals, are identical or nearly identical. We should also be able to assemble pieces that come from genomic regions in which the mammoth and the African elephant are similar but not perfectly matched. The hardest pieces to assemble will be those that come from regions of the genome in which the mammoth and the African elephant are very different. These differences might be due to reshuffling of genes or even duplication or deletion of genes.
Figure 9. The evolutionary relationships between mammoths, mastodons, and Asian and African elephants, based on the fossil record and their mitochondrial genome sequences.
In the book Jurassic Park, scientists filled in the bits of the dinosaur genome that they were unable to sequence with frog DNA. Similarly, we might solve this problem by simply filling in the holes of the mammoth genome with elephant DNA. This does not seem to me to be a very good idea, however. The common ancestor of mammoths, Asian elephants, and African elephants lived around four million years ago. This means the mammoth is separated from both the Asian elephant genome and the African elephant genome by more than eight million years of evolutionary change—a long time in which to collect evolutionary differences. Some of the hardest parts of the genome to assemble will be those regions that have changed in mammoths since their divergence from the other elephants. Arguably, these will be among the most important genomic regions to change in order to create an elephant that looks and acts like a mammoth instead of like an elephant. For the purposes of de-extinction, these might be the most critical regions of the genome to get right.
How to Clone a Mammoth Page 12