The Gene

by Siddhartha Mukherjee

  Two decades later, the transformation of the landscape of genetics was remarkable: human genes had been mapped, isolated, sequenced, synthesized, cloned, recombined, introduced into bacterial cells, shuttled into viral genomes, and used to create medicines. As the physicist and historian Evelyn Fox Keller described it: once “molecular biologists [had discovered] techniques by which they themselves could manipulate [DNA],” there “emerged a technological know-how that decisively altered our historical sense of the immutability of ‘nature.’

  “Where the traditional view had been that ‘nature’ spelt destiny, and ‘nurture’ freedom, now the roles appeared to be reversed. . . . We could more readily control the former [i.e., genes], than the latter [i.e., the environment]—not simply as a long-term goal but as an immediate prospect.”

  In 1969, on the eve of the revelatory decade, Robert Sinsheimer, the geneticist, wrote an essay about the future. The capacity to synthesize, sequence, and manipulate genes would unveil “a new horizon in the history of man.”

  “Some may smile and may feel that this is but a new version of the old dream of the perfection of man. It is that, but it is something more. The old dreams of the cultural perfections of man were always sharply constrained by his inherent, inherited imperfections and limitations. . . . We now glimpse another route—the chance to ease and consciously perfect far beyond our present vision this remarkable product of two billion years of evolution.”

  Other scientists, anticipating this biological revolution, had been less sanguine about it. As the geneticist J. B. S. Haldane had described it in 1923, once the power to control genes had been harnessed, “no beliefs, no values, no institutions are safe.”

  * * *

  I. In 1978, two other researchers, Y. Wai Kan and Andree Dozy, had found a polymorphism of DNA near the sickle-cell gene—and used it to follow the inheritance of the sickle-cell gene in patients.

  II. The high prevalence of the mutant cystic fibrosis gene in European populations has puzzled human geneticists for decades. If CF is such a lethal disease, then why was the gene not culled out by evolutionary selection? Recent studies posit a provocative theory: the mutant cystic fibrosis gene may provide a selective advantage during cholera infection. Cholera in humans causes a severe, intractable diarrhea that is accompanied by the acute loss of salt and water; this loss can lead to dehydration, metabolic disarray, and death. Humans with one copy of the mutant CF gene have a slightly diminished capacity to lose salt and water through their membranes and are thus relatively protected from the most devastating complications of cholera (this can be demonstrated using genetically engineered mice). Here too a mutation in a gene can have a dual and circumstantial effect—potentially beneficial in one copy, and lethal in two copies. Humans with one copy of the mutant CF gene may thus have survived cholera epidemics in Europe. When two such people reproduced, they had a one-in-four chance of creating a child with two mutant genes—i.e., a child with CF—but the selective advantage was strong enough to maintain the mutant CF gene in the population.

  “To Get the Genome”

  A-hunting we will go, a-hunting we will go!

  We’ll catch a fox and put him in a box,

  And then we’ll let him go.

  —Children’s rhyme from the eighteenth century

  Our ability to read out this sequence of our own genome has the makings of a philosophical paradox. Can an intelligent being comprehend the instructions to make itself?

  —John Sulston

  Scholars of Renaissance shipbuilding have often debated the nature of the technology that spurred the explosive growth of transoceanic navigation in the late 1400s and 1500s, ultimately leading to the discovery of the New World. Was it the capacity to build larger ships—galleons, carracks, and fluyts—as one camp insists? Or was it the invention of new navigation technologies—a superior astrolabe, the navigator’s compass, and the early sextant?

  In the history of science and technology too, breakthroughs seem to come in two fundamental forms. There are scale shifts—where the crucial advance emerges as a result of an alteration of size or scale alone (the moon rocket, as one engineer famously pointed out, was just a massive jet plane pointed vertically at the moon). And there are conceptual shifts—in which the advance arises because of the emergence of a radical new concept or idea. In truth, the two modes are not mutually exclusive, but reinforcing. Scale shifts enable conceptual shifts, and new concepts, in turn, demand new scales. The microscope opened a door to a subvisual world. Cells and intracellular organelles were revealed, raising questions about the inner anatomy and physiology of a cell, and demanding yet more powerful microscopes to understand the structures and functions of these subcellular compartments.

  Between the mid-1970s and the mid-1980s, genetics had witnessed many conceptual shifts—gene cloning, gene mapping, split genes, genetic engineering, and new modes of gene regulation—but no radical shifts in scale. Over the decade, hundreds of individual genes had been isolated, sequenced, and cloned by virtue of functional characteristics—but no comprehensive catalog of all genes of a cellular organism existed. In principle, the technology to sequence an entire organismal genome had been invented, but the sheer size of the effort had made scientists balk. In 1977, when Fred Sanger had sequenced the genome of the phiX virus, with 5,386 bases of DNA, that number represented the outer limit of gene-sequencing capability. The human genome contains 3,095,677,412 base pairs—representing a scale shift of 574,000-fold.

  The potential benefit of a comprehensive sequencing effort was particularly highlighted by the isolation of disease-linked genes in humans. Even as the mapping and identification of crucial human genes was being celebrated in the popular press in the early 1990s, geneticists—and patients—were privately voicing concerns about the inefficiency and laboriousness of the process. For Huntington’s disease, it had taken no less than twenty-five years to move from one patient (Nancy Wexler’s mother) to the gene (one hundred and twenty-one years, if you count Huntington’s original case history of the disease). Hereditary forms of breast cancer had been known since antiquity, yet the most common breast-cancer-associated gene, BRCA1, was only identified in 1994. Even with new technologies, such as chromosome jumping, that had been used to isolate the cystic fibrosis gene, finding and mapping genes was frustratingly slow. “There was no shortage of exceptionally clever people trying to find genes in the human,” John Sulston, the worm biologist, noted, “but they were wasting their time theorizing about the bits of the sequence that might be necessary.” The gene-by-gene approach, Sulston feared, would eventually come to a standstill.

  James Watson echoed the frustration with the pace of “single-gene” genetics. “But even with the immense power of recombinant DNA methodologies,” he argued, “the eventual isolation of most disease genes still seemed in the mid 1980s beyond human capability.” What Watson sought was the sequence of the entire human genome—all 3 billion base pairs of it, starting with the first nucleotide and ending with the last. Every known human gene, including all of its genetic code, all the regulatory sequences, every intron and exon, and all the long stretches of DNA between genes and all protein-coding segments, would be found in that sequence. The sequence would act as a template for the annotation of genes discovered in the future: if a geneticist found a novel gene that increases the risk for breast cancer, for instance, she should be able to decipher its precise location and sequence by mapping it to the master sequence of the human genome. And the sequence would also be the “normal” template against which abnormal genes—i.e., mutations—could be annotated: by comparing that breast-cancer-associated gene between affected and unaffected women, the geneticist would be able to map the mutation responsible for causing the disease.

  Impetus for sequencing the entire human genome came from two other sources. The one-gene-at-a-time approach worked perfectly for “monogenetic” diseases, such as cystic fibrosis and Huntington’s disease. But most common human diseases do not arise f
rom single-gene mutations. These are not genetic illnesses as much as genomic illnesses: multiple genes, spread diffusely throughout the human genome, determine the risk for the illness. These diseases cannot be understood through the action of a single gene. They can only be understood, diagnosed, or predicted by understanding the interrelationships between several independent genes.

  The archetypal genomic disease is cancer. That cancer is a disease of genes had been known for more than a century: in 1872, Hilário de Gouvêa, a Brazilian ophthalmologist, had described a family in which a rare form of eye cancer, called retinoblastoma, coursed tragically through multiple generations. Families certainly share much more than genes: bad habits, bad recipes, neuroses, obsessions, environments, and behaviors—but the familial pattern of the illness suggested a genetic cause. De Gouvêa proposed an “inherited factor” as the cause of these rare eye tumors. Halfway across the globe and seven years prior, an unknown botanist-monk named Mendel had published a paper on inherited factors in peas—but de Gouvêa had never encountered Mendel’s paper or the word gene.

  By the late 1970s, a full century after de Gouvêa, scientists began to converge on the uncomfortable realization that cancers arose from normal cells that had acquired mutations in growth-controlling genes.I In normal cells, these genes act as powerful regulators of growth: hence a wound in the skin, having healed itself, typically stops healing and does not morph into a tumor (or in the language of genetics: genes tell the cells in a wound when to start growing, and when to stop). In cancer cells, geneticists realized, these pathways were somehow disrupted. Start genes were jammed on, and stop genes were flicked off; genes that altered metabolism and identity of a cell were corrupted, resulting in a cell that did not know how to stop growing.

  That cancer was the result of alterations of such endogenous genetic pathways—a “distorted version of our normal selves,” as Harold Varmus, the cancer biologist, put it—was ferociously disquieting: for decades, scientists had hoped that some pathogen, such as a virus or bacterium, would be implicated as the universal cause of cancer, and might potentially be eliminated via a vaccine or antimicrobial therapy. The intimacy of the relationship between cancer genes and normal genes threw open a central challenge of cancer biology: How might the mutant genes be restored to their off or on states, while allowing normal growth to proceed unperturbed? This was—and still remains—the defining goal, the perennial fantasy, and the deepest conundrum, of cancer therapy.

  Normal cells could acquire these cancer-causing mutations through four mechanisms. The mutations could be caused by environmental insults, such as tobacco smoke, ultraviolet light, or X-rays—agents that attack DNA and change its chemical structure. Mutations could arise from spontaneous errors during cell division (every time DNA is replicated in a cell, there’s a minor chance that the copying process generates an error—an A switched to a T, G, or C, say). Mutant cancer genes could be inherited from parents, thereby causing hereditary cancer syndromes such as retinoblastoma and breast cancer that coursed through families. Or the genes could be carried into the cells via viruses, the professional gene carriers and gene swappers of the microbial world. In all four cases, the result converged on the same pathological process: the inappropriate activation or inactivation of genetic pathways that controlled growth, causing the malignant, dysregulated cellular division that was characteristic of cancer.

  That one of the most elemental diseases in human history happens to arise from the corruption of the two most elemental processes in biology is not a co-incidence: cancer co-opts the logic of both evolution and heredity; it is a pathological convergence of Mendel and Darwin. Cancer cells arise via mutation, survival, natural selection, and growth. And they transmit the instructions for malignant growth to their daughter cells via their genes. As biologists realized in the early 1980s, cancer, then, was a “new” kind of genetic disease—the result of heredity, evolution, environment, and chance all mixed together.

  But how many such genes were involved in causing a typical human cancer? One gene per cancer? A dozen? A hundred? In the late 1990s, at Johns Hopkins University, a cancer geneticist named Bert Vogelstein decided to create a comprehensive catalog of nearly all the genes implicated in human cancers. Vogelstein had already discovered that cancers arise from a step-by-step process involving the accumulation of dozens of mutations in a cell. Gene by gene, a cell slouches toward cancer—acquiring one, two, four, and then dozens of mutations that tip its physiology from controlled growth to dysregulated growth.

  To cancer geneticists, these data clearly suggested that the one-gene-at-a-time approach would be insufficient to understand, diagnose, or treat cancer. A fundamental feature of cancer was its enormous genetic diversity: two specimens of breast cancer, removed from two breasts of the same woman at the same time, might have vastly different spectra of mutations—and thereby behave differently, progress at different rates, and respond to different chemotherapies. To understand cancer, biologists would need to assess the entire genome of a cancer cell.

  If the sequencing of cancer genomes—not just individual cancer genes—was necessary to understand the physiology and diversity of cancers, then it was all the more evident that the sequence of the normal genome had to be completed first. The human genome forms the normal counterpart to the cancer genome. A genetic mutation can be described only in the context of a normal or “wild-type” counterpart. Without that template of normalcy, one had little hope that the fundamental biology of cancer could be solved.

  Like cancer, heritable mental illnesses were also turning out to involve dozens of genes. Schizophrenia, in particular, sparked a furor of national attention in 1984, when James Huberty, a man known to have paranoid hallucinations, strolled casually into a McDonald’s in San Diego on a July afternoon and shot and killed twenty-one people. The day before the massacre, Huberty had left a desperate message with a receptionist at a mental health clinic, pleading for help, then waited for hours by his phone. The return phone call never came; the receptionist had mistakenly spelled his name Shouberty and neglected to copy his number. The next morning, still afloat in a paranoid fugue, he had left home with a loaded semiautomatic wrapped in a checkered blanket, having told his daughter that he was “going hunting humans.”

  The Huberty catastrophe occurred seven months after an enormous National Academy of Sciences (NAS) study published data definitively linking schizophrenia to genetic causes. Using the twin method pioneered by Galton in the 1890s, and by Nazi geneticists in the 1940s, the NAS study found that identical twins possessed a striking 30 to 40 percent concordance rate for schizophrenia. An earlier study, published by the geneticist Irving Gottesman in 1982, had found an even more provocative correlation of between 40 and 60 percent in identical twins. If one twin was diagnosed with schizophrenia, then the chance of the other twin developing the illness was fifty times higher than the risk of schizophrenia in the general population. For identical twins with the severest form of schizophrenia, Gottesman had found the concordance rate was 75 to 90 percent: nearly every identical twin with one of the severest variants of schizophrenia had been found to have a twin with the same illness. This high degree of concordance between identical twins suggested a powerful genetic influence on schizophrenia. But notably, both the NAS and the Gottesman study found that the concordance rate fell sharply between nonidentical twins (to about 10 percent).

  To a geneticist, such a pattern of inheritance offers important clues about the underlying genetic influences on an illness. Suppose schizophrenia is caused by a single, dominant, highly penetrant mutation in one gene. If one identical twin inherits that mutant gene, then the other will invariably inherit that gene. Both will manifest the disease, and the concordance between the twins should approach 100 percent. Fraternal twins and siblings should, on average, inherit that gene about half the time, and the concordance between them should fall to 50 percent.

  In contrast, suppose schizophrenia is not one disease but a family of diseases
. Imagine that the cognitive apparatus of the brain is a complex mechanical engine, composed of a central axle, a main gearbox, and dozens of smaller pistons and gaskets to regulate and fine-tune its activity. If the main axle breaks, and the gearbox snaps, then the entire “cognition engine” collapses. This is analogous to the severe variant of schizophrenia: a combination of a few highly penetrant mutations in genes that control neural communication and development might cause the axle and the gears to collapse, resulting in severe deficits of cognition. Since identical twins inherit identical genomes, both will invariably inherit mutations in the axle and the gearbox genes. And since the mutations are highly penetrant, the concordance between identical twins will still approach 100 percent.

  But now imagine that the cognition engine can also malfunction if several of the smaller gaskets, spark plugs, and pistons do not work. In this case, the engine does not fully collapse; it sputters and gasps, and its dysfunction is more situational: it worsens in the winter. This, by analogy, is the milder variant of schizophrenia. The malfunction is caused by a combination of mutations, each with low penetrance: these are gasket-and-piston and spark-plug genes, exerting more subtle control on the overall mechanism of cognition.

  Here too identical twins, possessing identical genomes, will inherit, say, all five variants of the genes together—but since the penetrance is incomplete, and the triggers more situational, the concordance between identical twins might fall to only 30 or 50 percent. Fraternal twins and siblings, in contrast, will share only a few of these gene variants. Mendel’s laws guarantee that all five variants will rarely be inherited in toto by two siblings. The concordance between fraternal twins and siblings will fall even more sharply—to 5 or 10 percent.