The Tangled Tree

by David Quammen

  But there’s more in this vein preserved in the memories of Woese’s friends and colleagues. One of them, a chemist at the Institute for Genomic Biology, tells a story (though she wouldn’t repeat it to me for quotation) reflecting Woese’s late attitude toward evolution and his own place in its history. She went to see an important science bureaucrat in Washington—maybe a program officer at the National Science Foundation, someone like that—and noticed that high on his office wall hung portraits of two men: Charles Darwin and Carl Woese. Returning to Urbana, she mentioned that to Woese, presuming he might be flattered. Woese said: Why Darwin?

  77

  Molecular biologists now understand, Jan Sapp wrote in his draft introduction to the book that never happened, that “bacteria evolve by leaps and bounds through the inheritance of acquired genes.” It sounded more Lamarckian than Darwinian, he well knew. HGT was what he had in mind.

  And this wasn’t true just of bacteria, evolving by leaps and bounds. Animals do it too, sometimes. And not just insects and bdelloid rotifers. Mammals do it too, sometimes. “The cells that make up our bodies have also not arisen gradually in the typical Darwinian manner of gene mutation and natural selection.” Some of the changes occurred by quantum leaps. Our mitochondria came aboard suddenly, deep in the past of our eukaryote or pre-eukaryote lineage, as captured bacteria. Plants acquired their chloroplasts the same way. Our genomes are mosaics. We are all symbiotic complexes, even us humans.

  “Consider too,” Sapp wrote, “that a great percentage of our own DNA is of viral origin.” The figure most commonly cited is 8 percent: roughly 8 percent of the human genome consists of the remnants of retroviruses that have invaded our lineage—invaded the DNA, not just the bodies, of our ancestors—and stayed. We are at least one-twelfth viral, at the deepest core of our identities. Consider that, Sapp urged.

  So I did. And when I learned that few scientists if any have studied our viral genomic content more carefully than Thierry Heidmann, of the Gustave Roussy Institute on the south outskirts of Paris, it seemed worth a trip to see him.

  78

  Thierry Heidmann is another biologist who trained originally in physics and math. He grew up in Paris and got his education there at some of the best institutions in town: the École Normale Supérieure, followed by the University of Paris and then the Pasteur Institute. His father was an astrophysicist, and he considered astrophysics himself until he tipped toward neurobiology and the science of complex neuronal networks, roughly the same sort of complexity (with emergent properties) that intrigued Woese in his late collaboration with Nigel Goldenfeld. Heidmann, after finishing his doctorate in that area, hankered to do something more germane to human health, so he began to think about tumors and their relation to transposons and retroviruses.

  A retrovirus is a virus that works backward relative to the usual DNA transcription process. Instead of DNA-makes-RNA-makes-protein, the normal path from genetic information to living application, a retrovirus uses its RNA genome to make DNA, the double-stranded molecule. That trick, plus a few others, allows it not only to invade a cell but also to enter the cell nucleus and patch a DNA version of itself into the cell’s DNA, becoming a permanent part of the cell’s genome. Whenever the cell or its descendants replicate, that alien stretch is copied too. If the retrovirus happens to infect germline cells—eggs or sperm or the cells in ovaries and testes that produce them—then the inserted viral sequence will be inherited, passed along as a permanent part of the genome. At this point, it’s no longer alien to the organism. It’s now endogenous, meaning native, inherent. Such viruses are known as endogenous retroviruses (ERVs) because they become endogenous to the lineage of creatures they have infected.

  If a retrovirus inserts itself into the human genome, then that’s a human endogenous retrovirus, or HERV. Those are the viruses, HERVs, that constitute 8 percent of the human genome. Trust me, this is leading to a point that will boggle you at the implications of Thierry Heidmann’s work.

  Some retroviruses cause cancer. Mouse leukemia virus, for instance. Heidmann studied that one. The most notorious of retroviruses, of course, is HIV-1, which causes AIDS. It would have been logical at the time, the late 1980s, when Heidmann established his own lab, for a scientist such as he with an interest in retroviruses to launch a research program on HIV-1. The money would have flowed, the significance was vast and urgent. Instead, he did something less obvious.

  Tumor biology had its own urgency, and the connection of tumors with retroviruses continued to fascinate him. He knew of an old observation, made by more than one electron microscopist in the years before genome sequencing, that virus-like particles show up abundantly in placental tissue and in some tumors. Placental tissue? That seemed curious, a potentially fruitful digression, so Heidmann began looking for evidence of retroviruses in human placenta, freshly recovered from hospitals. He and his team found a new family of viruses, embedded in the placental DNA, that they named HERV-L. Further investigation showed that similar sequences, other ERV-L variants, exist in the mouse genome and the genomes of other mammals. “We made a sort of viral archeology,” Heidmann said in a 2009 interview.

  They found that the presence of ERV-L in animal genomes, including ours, dates back about a hundred million years, to before the big divergence of mammals. In primate genomes, this thing has copied itself, like a transposon, about two hundred times. Did it have a function? Did it serve as a gene? they wondered. Maybe and maybe not. It might just be selfish DNA, perpetuating itself by bouncing and replicating throughout various genomes. Heidmann’s group didn’t claim to know. But that was the beginning of Heidmann’s decades-long exploration of the viral component of human genomes, human identity. What he found, in his long search, is that some HERVs, if not that one, have indeed acquired roles as human genes.

  Thierry Heidmann is a generous man as well as a distinguished researcher. When he heard where I’d be staying in Paris, he said: Ah, very near my neighborhood. Forget the Metro, I’ll pick you up. So I stood outside my little hotel in the 17th Arrondissement and, promptly eight in the morning, a small white Volkswagen pulled to the curb. Out stepped a man with a graying beard, robust eyebrows, and a blue blazer worn over his sweater, welcoming me to hop in. Instead of swinging north to the nearby Boulevard Périphérique, the great ring road, as I had expected, he drove southeast on a picturesque avenue toward the heart of the city. Traffic wasn’t bad, and along the way, he pointed out some sights: there’s the church de la Madeleine, this is the Place de la Concorde, the Louvre on our left as we cross the Seine, Notre-Dame, of course, just upriver, now here’s Boulevard St.-Germain, and the Sorbonne, and over there the École Normale Supérieure, where I went to school, he said. This man, every day, makes the world’s most elegant commute. We reached the Gustave Roussy Institute in less than an hour, talking the whole way about his background and his work, and then talked for another six hours in his lab office, interrupted barely by lunch.

  The office, at the end of a lab corridor, is a small room with windows overlooking the treetops of Villejuif, a suburban village within which the Gustave Roussy Institute sits. The room is lined with shelves, the shelves filled with journal papers in piles, folders, and boxes, many of them printed out on colored paper—blue, green, orange, pink, as well as white—lending a nice sort of Mondrian décor to a very serious workspace. We sat at a table beside his MacBook Pro, on which he showed me slides of figures and graphs as he spoke, illustrating what he had been doing for twenty years. It began with the HERV-L work and proceeded into the story of another human endogenous retrovirus his group discovered not long afterward. What had changed between the two discoveries was that the human genome had been sequenced, and that sequence was publicly available. Heidmann’s methods changed accordingly. His team screened the entire genome, looking for evidence of unknown ERVs. What they sought in particular were signs of one kind of gene: a familiar, recognizable gene that makes a viral envelope, which is a sort of sticky wrapping around the h
ard capsule of a viral particle. They saw twenty.

  “There are all these envelope genes,” Heidmann told me. “Among them, two are very important.”

  One of the two had already been discovered by other researchers and named syncytin. This gene expressed itself, turning out its protein, most notably in placental tissue. What was it doing there? Nobody knew, not at first, but its name derived from a capacity to cause cells (cyt) to fuse together (syn). This effect had been proven with laboratory cell cultures. Fusion of cells into aggregate cell masses with multiple nuclei instead of individual walls is a crucial step in building one layer of the human placenta. That layer, a sort of permeable protoplasmic cushion, is the part of the placenta that mediates between the maternal blood and the fetal blood. (Brace yourself for a fancy bit of lingo: it’s called the syncytiotrophoblast. Okay, relax and forget.) So the hypothesis arose that syncytin might help assemble that layer. Heidmann’s group found another envelope gene, left behind from an entirely different retrovirus, with a similar capacity to fuse human cells. They named it syncytin-2. (The first became syncytin-1.) Their lab testing revealed the same capacity for causing cell-to-cell fusion, adding support to the hypothesis that these genes help build placentas.

  Soon after that, Heidmann’s team recognized two other such genes in the common laboratory mouse. These were different enough from the human syncytins to get their own names: syncytin-A and syncytin-B. Further screening of genomes revealed close equivalents also in rats, gerbils, voles, and hamsters. So the two mouse genes were old, having entered the rodent lineage at least twenty million years ago, before those branches split.

  “At this stage,” Heidmann told me, “there was a question we asked. How is it possible in fact that a gene, which has been captured by chance, could be so important?”

  Building placentas is obviously important, but the team needed more evidence linking syncytins to that function. They got it by experiments with genetically modified mice—“knockout” mice, in which the syncytin-A gene had been decommissioned by molecular manipulation. Breeding those mice, they watched the embryos all die in utero after no more than thirteen days of gestation. The usual gestation period in mice is nineteen to twenty-one days. These dead mice weren’t coming close. Dissection revealed structural defects in the boundary between placenta and fetus, which constrained the fetal blood vessels, inhibited fetal growth, and killed the unborn mice. It was persuasive. But, of course, you can’t do that experiment on humans.

  Heidmann’s curiosity spiraled outward. He and his team found a syncytin gene in the European rabbit. They found a carnivore syncytin in the genomes of dogs and cats. They found one in cows and sheep. They found one in ground squirrels. “We worked with many, many university and labs and zoos.” They collaborated particularly with a zoo in France, “with a lot of animals, because we get placenta when possible.” They even found one of these genes in marsupials.

  Marsupials, they have a placenta?

  “Very transient placenta. Because, in the marsupials there is opossum, or the kangaroo, wallaby, okay. So they have a very short-lived placenta.” Some people assumed otherwise, he said—that a marsupial has no placenta “because the embryo is going to a pouch outside of the mother.” So a marsupial female with her transient placenta and her gestation in an external pouch, yes, even she has a viral gene targeted to helping with the placenta. Heidmann’s group named that gene syncytin-Opo1, for the marsupial in which they first found it: the gray short-tailed opossum.

  All of these genes have four things in common. Each one derives from the envelope gene of a retrovirus that inserted itself in the mammal genome. Each one expresses itself as a protein suffusing the placenta. Each one causes cell-to-cell fusion (at least in lab cultures), suggesting that it can create that special fused-cell protoplasmic layer, which helps mediate between placenta and fetus, letting nutrients and gases seep in from the mother, letting wastes seep out. And each is an ancient gene, preserved for many millions of years in functional form (against the disarray of random mutations) by natural selection. Such preservation proves the genes have been useful. They aren’t junk DNA. They are tools, not scrap. They have helped the fittest mammals survive.

  Those four points represent the canonical standards, articulated by Heidmann’s group, for what constitutes a syncytin. But equally notable, his team realized, is what these genes do not have in common: They all came from different sources. They represent independent captures, independent domestications, of viral genes from entirely different retroviruses. That independence, Heidmann suspects, accounts for the high diversity of types of placenta, which among mammal species is an extremely variable structure. He gave me a whole disquisition on placental structure and taxonomy, wondrous and arcane, which I’ll spare you.

  “They capture different syncytins,” he said, “they” meaning the different mammal lineages throughout evolutionary history. “They capture different envelope genes, because they capture it from different viruses. And the difference in the capture is responsible for the variability of the structures. Okay.” The captures had occurred, furthermore, at vastly different times. The primate syncytin-2 dates back at least forty million years. The rodent version, as I mentioned, has sat in that lineage for twenty million years. The cow-and-sheep syncytin seems at least thirty million years old, and the one in marsupials may have entered that lineage more than eighty million years ago. What this reflected, Heidmann said, was a continuous bombardment of animals and their genomes by retroviruses. Most of those infections didn’t result in the viruses inserting themselves into the genomes, but a tiny fraction did, and those few yielded the endogenous retroviruses. A still smaller fraction of those ERVs gave their envelope genes to be transmogrified into syncytins.

  But wait. This whole pattern of different syncytins appearing in different mammal lineages at different times raises another question, a logical dilemma, which had struck me as I read through Heidmann’s papers before ever boarding the plane for Paris. If some of these essential syncytins are twenty million years old, and some are thirty million years old, and some are forty million, how did the mammalian lineage arise at all—how did the first placenta evolve—before those gene captures occurred? They have been acquired intermittently and by chance, through the course of mammal evolution, but they have always been necessary. You can’t be a placental mammal without a placenta. Which came first, chance or necessity?

  “Yeah. Exactly,” said Heidmann. “This is the paradox.”

  79

  Heidmann and his young colleagues answered the paradox with a hypothesis. Their hypothesis involves another nifty capacity of syncytin genes, probably also derived from the viral envelope genes from which they have been anciently modified. That capacity is immunosuppression.

  Retroviral envelopes are complex and versatile structures, and the genes that code for them, equally so. Besides their ability to cause cell-to-cell fusion, they can also suppress the antiviral immune response of a host. This has obvious value to an invading virus. It has other value, less obvious, to a mammal for use in its placenta. Both the fetus and the placenta of an individual mammal carry a different genome from the mother’s. Half their DNA comes from the father. If the mother’s immune system were entirely on alert, her white blood cells might attack the fetus and reject it. Part of the role of the placenta, a uniquely adaptive organ among placental mammals, is to keep peace between the mother’s immune system and the fetus by dampening that response. This allows for internal pregnancy and birth, an innovation back at the time when early mammals diverged from the reptile lineage, which evidently offered certain advantages over egg laying. Birds still exist, Heidmann reminded me, meaning that the advantage wasn’t absolute. Birds don’t have placentas. They eject their embryos early within hard-shelled oval packets, sending them out self-contained, with an endowment of nutritious yolk and a mere promise to keep them warm. That is, they lay eggs. Crocodiles, likewise. But the placenta gave an edge to some vertebrate lineages in
some situations. One lineage of mammals seized that advantage while another, the ancestors of what we now call monotremes (platypuses and spiny anteaters, egg-laying mammals) did not. What’s so advantageous about pregnancy and live birth? Well, it allows a mother to walk around, for one thing, carrying the fetus to safety inside her body instead of squatting on it like a sitting duck.

  What this all suggests, according to Heidmann’s hypothesis, is that the earliest syncytial gene that was captured in preplacental mammals may have served such a purpose of immunosuppression toward the fetus, and then gradually acquired its additional role also as an intermediating layer in the evolving placenta. Later syncytins may then have been substituted into mammalian lineages as improvements upon the first one.

  Heidmann and I went to lunch nearby, then came back for a final surge of talk. What does all this tell us, I asked him, about how evolution works? About the tree of life?

  He sighed at the breadth and flat-footedness of my question. “That our genes are not only our genes,” he said. Then he laughed, and I laughed too, but uneasily, because I was unsure I’d heard right. I asked him to repeat.