by Pbo, Svante
Naïve as always, I thought our paper pointing out the chemical impossibility of DNA survival over millions of years would halt the search for such super-old DNA. But rather than being the end of things, the Idaho plant fossils were the beginning of a whole new area of research. The next super-old DNAs to pop up were found in amber. Amber is resin that was exuded from trees millions of years ago and solidified into translucent golden clumps; it is found in large quantities in quarries in the Dominican Republic and on the shores of the Baltic Sea, among other places. Not uncommonly, insects, leaves, and even small animals such as tree frogs can be found entombed in resin. Such inclusions often preserve exquisite details of organisms that lived millions of years ago, and many investigators hoped that the same would be true for their DNA. One of the first such papers came in 1992, when a group at the American Museum of Natural History published a paper in Science presenting DNA sequences from a 30-million-year-old termite encased in a piece of Dominican amber.{24} This was followed in 1993 by a whole series of papers from a lab headed by Raul Cano at California Polytechnic State University, in San Luis Obispo, including one on DNA from a weevil between 120 million and 135 million years old found in Lebanese amber{25} and another on DNA from a 35- to 40-million-year-old leaf from the Dominican tree that produced the amber.{26} Cano went on to found a company that claims to have isolated more than twelve hundred organisms from amber, including nine ancient strains of live yeast. Leaving such outlandish claims aside, it seemed to me that one could not rule out the possibility that DNA might be preserved for an extraordinarily long time in amber, since it was probably protected from water and oxygen, the two factors most destructive to the chemistry of DNA. That supposition, however, didn’t necessarily mean protection from the effects of background radiation over millions of years, nor did it explain why we had struggled so mightily to amplify traces of DNA a thousand times younger.
The opportunity to find out came in 1994, when Hendrik Poinar joined our lab. Hendrik was a jovial Californian and the son of George Poinar, then a professor at Berkeley and a well-respected expert on amber and the creatures found in it. Hendrik had published some of the amber DNA sequences with Raul Cano, and his father had access to the best amber in the world. Hendrik came to Munich and went to work in our new clean room. But he could not repeat what had been done in San Luis Obispo. In fact, as long as his blank extracts were clean, he got no DNA sequences at all out of the amber—regardless of whether he tried insects or plants. I grew more and more skeptical, and I was in good company. In 1993, Tomas Lindahl, who had been interested in ancient DNA ever since my 1985 visit to his lab, published a highly influential review on DNA stability and decay in Nature, in which he devoted a section to ancient DNA.{27} He pointed out—as I had with Allan earlier—that survival beyond a few hundred thousand years was unlikely. He left open the possibility that DNA from specimens encased in amber was an exception; in the meantime, however, I had given up even on the amber.
Tomas had also found the perfect term for super-old DNA: antediluvian DNA. We loved it, applied it, and it stuck. But this ridicule could not, of course, deter the enthusiasts. The inevitable happened in 1994, when Scott Woodward of Brigham Young University in Utah published DNA sequences that he and his colleagues had extracted from 80-million-year-old bone fragments—bone that “likely” came from a dinosaur or dinosaurs.{28} Not unexpectedly, this paper appeared in one of the two journals that compete for headline-worthy work and enjoy an often undeserved scientific prestige. This time it was Science. The authors had determined many different mtDNA sequences from the bone fragments, and some of them seemed to the authors to be as distant from birds and reptiles as from mammals. They suggested that these might be dinosaur DNA sequences. This seemed ludicrous to me. Thoroughly frustrated by the way the field had developed, Hans Zischler, a meticulous, even slightly pedantic, postdoc in my lab, decided to go after this particular piece of work. When we did a more rigorous analysis of the DNA sequences that the Utah group had published, they seemed closer to mammalian—and indeed human—mtDNA than to birds or reptiles.
Still, they didn’t quite seem to be human mtDNA. Explaining what they were takes a bit more explanation of the nature of mtDNA. Recall that mitochondrial genomes are circular DNA molecules of 16,500 nucleotides that reside in mitochondria, organelles located outside the cell nucleus in almost all animal cells. These organelles, as well as their genomes, derive originally from bacteria that almost 2 billion years ago entered primordial animal cells and were hijacked by those cells to produce energy. Over time, the hijacked bacteria transferred most of their DNA to the cell nucleus, where the DNA became integrated into the major part of the genome, situated on chromosomes. Even today in the human germ line, when eggs and sperm cells are formed, a mitochondrion will occasionally break, and fragments of its DNA will end up in the cell nucleus. There, efficient repair mechanisms recognize the ends of broken DNA strands and join them to other DNA ends that may exist if the nuclear genome also happens to carry a break. Thus, now and again, pieces of mtDNA become integrated in our nuclear genome, where, without having any function, they are passed on to new generations. Each of us carries hundreds if not thousands of such misplaced mitochondrial DNA fragments in our cell nuclei that have integrated into our genome at various times in the past. These fragments have different degrees of similarity to our real mitochondrial mtDNA; although they resemble ancestral mtDNA sequences, they have accumulated mutations, unconstrained as they are by any functional requirements in their new life as genetic garbage embedded in nuclear DNA. Hans Zischler had worked in our lab on detecting such integrations of mtDNA into the nuclear genome, and as we considered the putative dinosaur DNA, we wondered whether such mtDNA fragments might be what the Utah group had found. Indeed, given our experience with contaminating human DNA, it seemed probable to us that they had found nuclear versions of human mtDNA with unusual mutations. We decided to look in the human nuclear genome for the sequences they had published. The problem with our plan was that any normal preparation of DNA from human cells ended up containing a mix of both nuclear and mtDNA, and the hundreds and thousands of copies of real mtDNA in the mitochondria of most cells would get in the way of our attempts to detect any mtDNA segments that had left the mitochondrion and settled among the nuclear DNA. Here we were helped by biology. As noted in Chapter 1, we inherit our mtDNA exclusively from our mothers, via the egg, and get no mtDNA from our fathers. This is because the heads of the sperm, which penetrate the egg, contain no mitochondria. So to get nuclear DNA without accompanying mtDNA, our lab simply needed to isolate sperm heads.
I talked to my male graduate students, and there was enough enthusiasm for our work that we all went our separate ways one morning and generated sperm, from which Hans carefully isolated the heads by centrifugation. He then purified the DNA from the sperm heads and used the same primers for the PCR as had been used by the Utah group. As expected, he obtained many sequences from nuclear mtDNA fragments, which we then sifted through for any similarity to the “dinosaur” sequences from Utah. Indeed, we found two that were almost identical to the published sequences. This meant that instead of dinosaur DNA, the Utah group had sequenced bits of translocated human mtDNA from the human nuclear genome. Because these segments had left the human mtDNA genome in the distant past, they had picked up enough mutations to appear somewhat distant from humans, yet still similar to mtDNAs from mammals, birds, and reptiles. I could not prevent myself from being slightly facetious when writing the “Technical Comment” for Science{29} and suggesting that there were three possibilities to explain how we could obtain DNA sequences very similar to the ones from Utah using our own DNA in our lab. The first was that we had contamination in our laboratory from dinosaur DNA, which I suggested was unlikely. The second was that dinosaurs hybridized with early mammals before becoming extinct some 65 million years ago. This alternative, too, was dismissed as unlikely. The third (and most plausible) scenario was contamination by human DNA i
n the dinosaur experiments. Science published our comment along with comments from two other groups, both of which pointed out deficiencies in the DNA-sequence comparisons that had led the Utah group to claim that the mtDNA sequences looked ancestral to birds.
The comment was fun to write but also somewhat bitter, given that studies such as the Utah one had become a constant feature of the ancient DNA field. The problem of high-profile but dubious results still plagues research on ancient DNA today. As my students and postdocs have often remarked to me, it is easy to generate outlandish results with the PCR but difficult to show that they are correct; nevertheless, once results are published, it is even more difficult to show that they are wrong and explain where the contamination came from. In this instance, we were successful, but our efforts involved a lot of work and did not take our research forward. To this day, it is unclear just where the amber sequences that were published in Nature and Science came from. With enough work, I was sure the sources could be found, yet I decided we had had enough. As one student put it, “Let’s stop playing the PCR police.” We determined from then on to ignore those reports we thought were wrong and concentrate on our own work. Our best service to the field, we felt, was to establish methods to retrieve DNA from sources that were some tens of thousands of years old and to show that the results were genuine and correct. With ancient remains of humans this was hard if not impossible, as modern human DNA was lurking almost everywhere. So even though it pained me, I needed to forget about human history for the time being and divert my work to ancient animals. After all, I was a professor in a zoology department. I decided that we would focus on questions concerning the relationships of extinct animals and their present-day relatives.
Chapter 5
Human Frustrations
____________________________
During his collecting expeditions in the 1830s in South America, Charles Darwin was fascinated but puzzled by fossil remains of various large, plant-eating mammals. These creatures seemed much bigger than any animals currently living in the area. Along with examples of every living animal and bird he could capture, Darwin collected a number of fossils to send back to England, including a large lower jawbone that was eroding out of a coastal cliff in Argentina. The anatomist Richard Owen analyzed the jaw and attributed it to a giant ground sloth the size of a hippopotamus, which he dubbed Mylodon darwinii (see Figure 5.1). Even more interesting than the idea of such a bizarrely large herbivore was the idea that it might even still exist, alive, somewhere in the wilds of Patagonia. In 1900, the sensational discovery of apparently fresh dung and skin remnants of what appeared to be giant ground sloths motivated an expedition by a Mr. Hesketh Prichard in search of this marvel. After journeying some two thousand miles through Patagonia, Prichard briskly concluded that he had found “no single scrap of evidence of any kind which would support the idea of the survival of the Mylodon.”{30} This was with good reason: we now know that it became extinct during the last Ice Age, some 10,000 years ago.
Two- and three-toed sloths exist today in South America, but, weighing in at a mere ten to twenty pounds, they are of modest size compared to Mylodon. And unlike Mylodon, both two- and three-toed sloths live in trees. But they seem to have adapted to life in the trees only rather recently in evolutionary terms, since they are rather large for tree-dwelling mammals, not particularly agile aloft, and prefer to descend to the ground for such mundane routines as defecation. A big question was whether the ancestors of the tree sloths had become adapted to the arboreal lifestyle just once, and not particularly gracefully, or whether the two forms of tree sloths were examples of parallel adaptations, whereby ground-dwelling sloths in the past had at least twice independently taken to the trees. If similar adaptations happened independently more than once—if history repeated itself, so to speak—it suggests that there are a limited number of ways in which animals can adapt to an ecological challenge. Each such case of convergence, when two or more unrelated organisms independently evolve similar behaviors or body shapes, is evidence that evolution follows rules—and is helpful in deducing how these rules work. An example of this was the marsupial wolf that we had studied in Zurich and Berkeley. In the case of the tree sloths, just as in the case of the marsupial wolf, we could determine whether convergence had occurred if we could clarify how Darwin’s extinct giant ground sloth was related to the two-toed and three-toed tree sloths.
I visited the Natural History Museum in London and spent some time there with the amiable curator of Quaternary mammals, Andrew Currant, an expert on mammal paleontology with a build not unlike that of a large Pleistocene mammal. He showed me some of the fossilized bones that Darwin had brought back, and he allowed me to cut a small piece from two of the Patagonian Mylodon bones in their collection. I also visited the American Museum of Natural History, in New York, and got samples for our study there. But it was in Andrew’s museum that I experienced a vivid demonstration of how readily the ancient animal specimens we studied might become contaminated. As I was examining sloth bones with Andrew, I asked him if they had perhaps been treated with varnish. To my amazement, he picked up a bone and licked it. “No,” he said, “these have not been treated,” explaining that if a bone had been treated with varnish, it would not absorb saliva. In contrast, an untreated bone would do this so efficiently that one’s tongue tended to adhere to the bone. I was horrified and wondered how many times this “test” had been done during the hundred years or more that some of the bones we worked with had been in museums.
Figure 5.1. Reconstruction of ground sloth skeleton. Source: http://commons.wikimedia.org/wiki/.
Once the samples were back in Munich, Matthias Höss applied his skills to them. As always, I insisted that we first pay attention to the technical side of things. My interest in sloths was after all driven mainly by an interest in how to retrieve ancient DNA. Matthias used a rough assay to estimate the total amount of DNA in his Mylodon extract and another crude assay to measure how much of that was similar to modern sloth DNA. It turned out that about 0.1 percent of the DNA in our best Mylodon bone extract was from the animal itself, the rest having come from other organisms that had lived in the bones after the giant sloth died. This has turned out to be typical of many ancient remains we have since studied.
Focusing on mitochondrial DNA fragments, Matthias managed to use the PCR to reconstruct a stretch of Mylodon mtDNA more than a thousand nucleotides long by amplifying short overlapping pieces. By determining and comparing the same sequences from samples from living sloths, he could show that the giant ground sloth, which stood ten feet tall on its hind legs, was more closely related to the present-day two-toed tree sloth than to the three-toed tree sloth. This was important, since if the two- and three-toed sloths had been most closely related to each other and more distantly related to Mylodon (which was the opinion of most scientists at that time), it would have suggested that they had a common ancestor who became tree-dwelling. Our result suggested that sloths had at least twice evolved into forms that were small and spent most of their lives in trees (see Figure 5.2).
Figure 5.2. A tree showing that the Mylodon is more closely related to the two-toed than to the three-toed sloth, suggesting that sloths started to live in the trees twice during their history. From Matthias Höss et al., “Molecular phylogeny of the extinct ground sloth Mylodon darwinii,” Proceedings of the National Academy of Sciences USA 93, 181–185 (1996).
That both the marsupial wolf and the tree sloths turned out to be examples of convergent evolution was to me a strong message that morphology is often an unreliable indicator of relatedness among organisms. It seemed that almost any body shape or behavior could evolve independently if a change of environment created a pressure for a change in lifestyle. To me, DNA sequences seemed to offer a much better chance to correctly gauge how species were related to each other. DNA sequences can accumulate hundreds and thousands of mutations over time, each of which occurs independently of one another, and most of which have no influen
ce on how an organism looks or behaves. In contrast, measurement of morphological features necessarily is done on traits that might very well affect the survival of the organism, and the sizes of different features, such as various bones, might be linked to one another. Because of the greater number of independent, randomly varying data points that can be accumulated, DNA sequences allow reconstruction of relationships with greater rigor than morphological features. In fact, in contrast to morphological features, even the timing of divergences from a common ancestor can be derived from the number of differences that have accumulated in DNA sequences, since these differences occur roughly as a function of time, at least within a group of related animals.
Matthias used such a “molecular clock” approach and calculated the numbers of nucleotide differences and underlying mutations that had accumulated in the mtDNAs of members of the group of animals to which the sloths belong, which includes armadillos and anteaters. He found that this group of animals is surprisingly old. They began diversifying before the dinosaurs became extinct, some 65 million years ago. This timetable holds for some other groups of mammals, as well as for birds, such that many groups of present-day animals trace back to ancestors that originated at a time when dinosaurs dominated the earth. Once, many different forms of ground-dwelling sloths existed, but today there are only tree sloths. Until our discovery that today’s tree sloths do not have a common ancestor, it had been reasonable to think that the arboreal forms might share some unknown but important physiological adaptation that allowed them to survive, perhaps in the face of climatic change during the last glaciation. But if they did not share a common ancestor, this seemed less likely. It was more plausible that the crucial factor in their survival was the most obvious one: that they live in trees. We ended our paper with the speculation that living in trees might have helped them survive the arrival of humans, who seem to have hunted ground-dwelling, slow-moving sloths to extinction.{31} Although the debate continues about whether ecological factors or overhunting by humans caused the disappearance of American megafauna, such as ground-dwelling sloths, by approximately 10,000 years ago, we were happy that ancient DNA could add a piece to the puzzle. We had shown that reliable DNA sequences could be retrieved from animals that lived thousands of years ago, and that this could yield enough information to provide a new perspective on their evolution.
