12. A triangle of considerations—extraordinary suffering, highly penetrant genotypes, and justifiable interventions—has, thus far, constrained our attempts to intervene on humans. As we loosen the boundaries of this triangle (by changing the standards for “extraordinary suffering” or “justifiable interventions”), we need new biological, cultural, and social precepts to determine which genetic interventions may be permitted or constrained, and the circumstances in which these interventions become safe or permissible.
13. History repeats itself, in part because the genome repeats itself. And the genome repeats itself, in part because history does. The impulses, ambitions, fantasies, and desires that drive human history are, at least in part, encoded in the human genome. And human history has, in turn, selected genomes that carry these impulses, ambitions, fantasies, and desires. This self-fulfilling circle of logic is responsible for some of the most magnificent and evocative qualities in our species, but also some of the most reprehensible. It is far too much to ask ourselves to escape the orbit of this logic, but recognizing its inherent circularity, and being skeptical of its overreach, might protect the weak from the will of the strong, and the “mutant” from being annihilated by the “normal.”
Perhaps even that skepticism exists somewhere in our twenty-one thousand genes. Perhaps the compassion that such skepticism enables is also encoded indelibly in the human genome.
Perhaps it is part of what makes us human.
* * *
I. Another system to deliver “programmable” cuts in specific genes using a DNA-cutting enzyme is also being developed. Termed a “TALEN,” this enzyme can also be used for genome editing.
II. One important technical detail is that since individual ES cells can be cloned and expanded, cells with unintended mutations can be identified and discarded. Only prescreened ES cells, carrying the intended mutation, are transformed into sperm or egg.
Epilogue: Bheda, Abheda
Sura-na Bheda Pramaana Sunaavo;
Bheda, Abheda, Pratham kara Jaano.
Show me that you can divide the notes of a song;
But first, show me that you can discern
Between what can be divided
And what cannot.
—An anonymous musical composition inspired by a classical Sanskrit poem
Abhed, my father had called genes—“indivisible.” Bhed, the opposite, is its own kaleidoscope of a word: “to discriminate” (in its verb form), “to excise, to determine, to discern, to divide, to cure.” It shares linguistic roots with vidya, “knowledge,” and with ved, “medicine.” The Hindu scriptures, the Vedas, acquired their name from the same root. It arises from the ancient Indo-European word uied, “to know” or “to discern meaning.”
Scientists divide. We discriminate. It is the inevitable occupational hazard of our profession that we must break the world into its constituent parts—genes, atoms, bytes—before making it whole again. We know of no other mechanism to understand the world: to create the sum of the parts, we must begin by dividing it into the parts of the sum.
But there is a hazard implicit in this method. Once we perceive organisms—humans—as assemblages built from genes, environments, and gene-environment interactions, our view of humans is fundamentally changed. “No sane biologist believes that we are entirely the product of their genes,” Berg told me, “but once you bring genes into the picture, then our perception of ourselves can no longer be the same.” A whole assembled from the sum of the parts is different from the whole before it was broken into the parts.
As the poem in Sanskrit goes:
Show me that you can divide the notes of a song;
But first, show me that you can discern
Between what can be divided
And what cannot.
Three enormous projects lie ahead for human genetics. All three concern discrimination, division, and eventual reconstruction. The first is to discern the exact nature of information coded in the human genome. The Human Genome Project provided the starting point for this inquiry, but it raised a series of intriguing questions about what, precisely, is “encoded” by the 3 billion nucleotides of human DNA. What are the functional elements in the genome? There are protein-coding genes, of course—about twenty-one to twenty-four thousand in all—but also regulatory sequences of genes, and stretches of DNA (introns) that split genes into modules. There is information to build tens of thousands of RNA molecules that do not get translated into proteins but still seem to perform diverse roles in cellular physiology. There are long highways of “junk” DNA that are unlikely to be junk after all and may encode hundreds of yet-unknown functions. There are kinks and folds that allow one part of the chromosome to associate with another part in three-dimensional space.
To understand the role of each of these elements, a vast international project, launched in 2013, hopes to create a compendium of every functional element in the human genome—i.e., any part of any sequence in any chromosome that has a coding or instructional function. Ingeniously termed the Encyclopedia of DNA Elements (ENC-O-DE), this project will cross-annotate the sequence of the human genome against all the information contained within it.
Once these functional “elements” have been identified, biologists can move to the second challenge: understanding how the elements can be combined in time and space to enable human embryology and physiology, the specification of anatomical parts, and the development of an organism’s unique features and characteristics.I One humbling fact about our understanding of the human genome is how little we know of the human genome: much of our knowledge of our genes and their functions is inferred from similar-looking genes in yeast, worms, flies, and mice. As David Botstein writes: “Very few human genes have been studied directly.” Part of the task of the new genomics is to close the gap between mice and men—to determine how human genes function in the context of the human organism.
For medical genetics, this project promises several particularly important payoffs. The functional annotation of the human genome will enable biologists to discover novel mechanisms of illness. New genomic elements will be linked to complex medical diseases, and these links will allow us to determine the ultimate causes of diseases. We still do not know, for instance, how the intersection between genetic information, behavioral exposures, and random chance causes hypertension, schizophrenia, depression, obesity, cancer, or heart disease. Finding the correct functional elements in the genome that are linked to these diseases is the first step to solving the mechanisms by which they arise.
Understanding these links will also reveal the predictive power of the human genome. In an influential review published in 2011, the psychologist Eric Turkheimer wrote, “A century of familial studies of twins, siblings, parents and children, adoptees, and whole pedigrees has established, beyond a shadow of a doubt, that genes play a crucial role in the explanation of all human differences, from the medical to the normal, the biological to the behavioral.” Yet, despite the strengths of these links, the “genetic world,” as Turkheimer calls it, has proved much more difficult to map and deconvolute than expected. Until recently, the only genetic changes that were powerfully predictive of future illnesses were those with high penetrance that caused the severest of phenotypes. Combinations of gene variants were particularly difficult to decipher. It was impossible to determine how a particular permutation of genes (i.e., a genotype) would determine a particular outcome in the future (i.e., phenotype), especially if that outcome was governed by a multitude of genes.
But this barrier might soon collapse. Imagine a thought experiment that might seem far-fetched at first glance. Suppose we could comprehensively sequence the genomes of one hundred thousand children prospectively—i.e., before anything is known about the future of any child—and create a database of all the variations and combinations of the functional elements of each child’s genome (one hundred thousand is an arbitrary number; the experiment can be extended to any number of children). Imagine, now, creating a
“fate map” of this cohort of children: every illness or physiological aberrancy identified and recorded on a parallel database. We might describe this map as a human “phenome”—the full set of all phenotypes (attributes, features, behaviors) of an individual. And now imagine a computational engine that mines the data from these gene map/fate map pairs to determine how one might predict the other. Despite remnant uncertainties—even deep ones—the prospective mapping of one hundred thousand human genomes to one hundred thousand human phenomes would provide an extraordinary data set. It would begin to describe the nature of fate that is encoded by a genome.
The extraordinary feature of this fate map is that it need not be restricted to illness; it can be as wide and deep and detailed as we would like it to be. It could include the low birth weight of a child, a learning disability in preschool, the transient tumult of a particular adolescence, a teenage infatuation, an impulsive marriage, coming out, infertility, a midlife crisis, a propensity for addiction, a cataract in the left eye, premature baldness, depression, a heart attack, an early death from ovarian cancer or breast cancer. Such an experiment would have been inconceivable in the past. But the combined power of computing technology, data storage, and gene sequencing has made it conceivable in the future. It is a gargantuan twin study—except without twins: millions of virtual genetic “twins” are created computationally by matching genomes across space and time and these permutations are then annotated against life events.
It’s important to recognize the inherent limitations of such projects or, more generally, of trying to predict diseases and destinies from genomes. “Perhaps,” as one observer complained, “the fate of genetic explanations will [culminate in] de-contextualising etiological processes, underrepresenting the role of environments, producing some stunning medical interventions, [but] revealing little about the fate of populations.” But the power of such studies is precisely that they “decontextualize” illness; genes provide the context to understand development and fate. Situations that are context dependent or environment dependent get diluted and filtered out—and only those powerfully affected by genes remain. With enough subjects, and enough computational power, nearly all of the predictive capacity of the genome can, in principle, be determined and computed.
The final project is perhaps the most far-reaching. As much as the ability to predict human phenomes from human genomes was limited by the lack of computational technologies, the ability to intentionally change human genomes was bounded by the paucity of biological technologies. Gene delivery methods such as viruses were inefficient and unreliable at best, and lethal at worst—and intentional gene delivery into the human embryo was virtually impossible.
These barriers too have started to collapse. Novel “gene editing” technologies now allow geneticists to make remarkably precise alterations in the human genome with equally remarkable specificity. In principle, a single letter of DNA can be mutated to another letter in a directional manner, leaving the 3 billion other bases of the genome largely untouched (this technology might be likened to a copyediting device that scans sixty-six volumes of the Encyclopaedia Britannica and finds, erases, and changes one word, leaving all other words untouched). Between 2010 and 2014, a postdoctoral researcher in my laboratory tried to introduce a defined genetic change into a cell line using the standard gene-delivery viruses, but with little success. In 2015, having switched to the new CRISPR-based technology, she engineered fourteen alterations of genes in fourteen human genomes, including the genomes of human embryonic stem cells, in six months—a feat unimaginable in the past. Geneticists and gene therapists around the world are currently exploring the possibility of changing the human genome with renewed verve and urgency—in part, because the current technologies have brought us to a precipice. A combination of stem cell technologies, nuclear transfer and epigenetic modulation, and gene-editing methods has made it conceivable that the human genome can be broadly manipulated, and that transgenic humans can be created.
We have no knowledge of the fidelity or efficiency of these techniques in practice. Does making an intentional change in a gene run the risk of creating an unintended change in another part of the genome? Are some genes more easily “edited” than others—and what governs the pliability of a gene? Nor do we know whether making a directed change in one gene might cause the entire genome to become dysregulated. If some genes are indeed “recipes,” as in Dawkins’s formulation, then altering one gene may cause far-reaching consequences for gene regulation—potentially unleashing a volley of downstream consequences, akin to the proverbial butterfly effect. If such butterfly-effect genes are common in the genome, then they will represent fundamental limitations for gene-editing technologies. The discontinuity of genes—the discreteness and autonomy of each individual unit of heredity—will turn out to be an illusion: genes may yet be more interconnected than we think.
But first, show me that you can discern
between what can be divided
and what cannot.
Imagine, then, a world in which these technologies can be routinely deployed. When a child is conceived, every parent is given the choice of testing the fetus using comprehensive genome sequencing in utero. Mutations that cause the most severe impairments are identified, and parents are given the option of aborting such fetuses at the earliest stages of pregnancy, or selectively implanting only the “normal” fetuses after comprehensive genetic screening (we might call this comprehensive preimplantation genetic diagnosis, or c-PGD).II
More complex combinations of genes that might cause tendencies toward disease are also identified by genome sequencing. When children with such predicted tendencies are born, they are offered selective interventions throughout childhood. A child with a predilection toward a genetic form of obesity, for instance, might be monitored for changes in body mass, treated with an alternative diet, or metabolically “reprogrammed” using hormones, drugs, or genetic therapies in childhood. A child with a tendency for an attention-deficit or hyperactivity syndrome might undergo behavioral therapy or be placed in an enriched classroom.
If and when the illnesses do emerge or advance, gene-based therapies are deployed to treat or cure them. Corrected genes are delivered directly into the affected tissues: the functioning cystic fibrosis gene, for instance, is aerosolized and injected into the lungs of patients, where it partially restores the normal function of the lung. A girl born with ADA deficiency is transplanted with bone marrow stem cells carrying the correct gene. For more complex genetic diseases, genetic diagnostics are combined with genetic therapies, with drugs and with “environment therapies.” Cancers are comprehensively analyzed by documenting the mutations responsible for driving malignant growth of one particular cancer. These mutations are used to identify culprit pathways that fuel the growth of cells and to devise exquisitely targeted therapies to kill malignant cells and spare normal cells.
“Imagine you are a soldier returning from war with PTSD,” the psychiatrist Richard Friedman wrote in the New York Times in 2015. “With a simple blood test looking at gene variants, we could discover whether you were biologically adept at fear extinction. . . . If you had a mutation that reduced your ability to extinguish fear, your therapist would know you might just need more exposure—more treatment sessions—to recover. Or, perhaps a different therapy altogether that doesn’t rely on exposure, like interpersonal therapy, or medication.” Perhaps drugs that can erase epigenetic marks are prescribed in combination with talk therapy. Perhaps the erasure of cellular memories can ease the erasure of historical memories.
Genetic diagnoses and genetic interventions are also used to screen and correct mutations in human embryos. When “intervenable” mutations in certain genes are identified in the germ line, parents are given the choice of genetic surgery to alter their sperm and eggs before conception, or prenatal screening of embryos to avoid implanting mutant embryos in the first place. Genes that cause the most detrimental variants of illness are thus excised from the huma
n genome up front by positive or negative selection, or by genome modification.
If you read that scenario carefully, it inspires both wonder and a certain moral queasiness. The individual interventions may not push the boundaries of transgression—indeed some, such as the targeted treatment of cancer, schizophrenia, and cystic fibrosis, represent landmark goals for medicine—but aspects of this world seem distinctly and even repulsively alien. It is a world inhabited by “previvors” and “post-humans”: men and women who have been screened for genetic vulnerabilities or created with altered genetic propensities. Illness might progressively vanish, but so might identity. Grief might be diminished, but so might tenderness. Traumas might be erased but so might history. Mutants would be eliminated but so would human variation. Infirmities might disappear, but so might vulnerability. Chance would become mitigated, but so, inevitably, would choice.III
In 1990, writing about the Human Genome Project, the worm geneticist John Sulston wondered about the philosophical quandary raised by an intelligent organism that has “learned to read its own instructions.” But an infinitely deeper quandary is raised when an intelligent organism learns to write its own instructions. If genes determine the nature and fate of an organism, and if organisms now begin to determine the nature and fate of their genes, then a circle of logic closes on itself. Once we start thinking of genes as destiny, manifest, then it is inevitable to begin imagining the human genome as manifest destiny.