The CRISPR/Cas9 (CRISPR) technique has been used to modify genes in animals, plants and fungi, organisms different from and more complex than the bacteria in which the molecular components originally evolved. It has undergone several refinements since its introduction, each iteration proving more accurate, with fewer off-target effects. The Stanford University bioethicist Hank Greely contemplates using CRISPR to touch up human embryos that have been produced by in vitro fertilization and prescreened for overall suitability by gene sequencing. George Church, a Harvard University genetic technologist and entrepreneur, advocates a more aggressive program of CRISPR-mediated genetic improvements to future generations.
Claims of the near-infallibility of CRISPR may be overstated, but even if it could be made to operate perfectly, would using CRISPR to improve humans by altering embryos ever be justified? Since CRISPR acts on genes, not traits (which are presumably the target of any prospective modification), the answer to this question depends the relationship between them.
In fact, there is a growing realization that DNA is far from the “code of life” it has long been claimed to be. Geneticists commonly use terms like “epigenetics” (functional effects from chemical modifications of genes), “epistasis” (consequences of interaction between the products of different genes), and “incomplete penetrance” (failure of a gene to have its default effect), to signal their expectation that, apart from such exceptions, a gene, or ensemble of genes will influence a trait in a reliable fashion. But it appears that it is the well-behaved gene that may be the exception. A study in the journal Genome Research reported that the genes of monozygotic (“identical”) twins exhibit different patterns of activity-affecting modification from early stages of development. A review article in the journal Human Genetics discussed the implications of the “many known examples of ‘disease-causing mutations’ that fail to cause disease in at least a proportion of the individuals who carry them.” The authors noted that in some cases the ability of a “bad” gene to cause disease “appears to require the presence of one or more genetic variants at other loci.” An unstated implication of this is that when a typically pathogenic genetic variant is compensated by a second one, replacing it by its “wild type” or common counterpart would likely cause problems.
Surveying a rash of new data casting doubt on soundness of the received corpus of human genetics, a recent editorial in the journal Nature asserted in that “many [human] genetic mutations have been misclassified as harmful.” This accompanied a news feature that began, “Lurking in the genes of the average person are 54 mutations that look as if they should sicken or even kill their bearer. But they don’t.”
Part of the reason for the disarray in the field is the notion, long rejected by geneticists but difficult to completely dispel, that individual genes map one-to-one to specific traits or diseases. But recent research suggests that the problems of genotype-phenotype mapping go much deeper, to the concept of the gene itself. One problem is the fact that genomes have unique evolutionary histories. The genes that helped establish the basic body plans and organ structures of animals around 600 million years ago still operate in present-day species, but they have diverged in their precise functions, partnering with different accessory genes in different kinds of animals, even when making the same structure (an eye, a heart, a limb). Consequently, members of the same species (including individual humans) can use variable genetic means to accomplish the same or similar ends. This “rewiring” effect is known to evolutionary biologists as “developmental system drift.”
Another even more serious difficulty in assigning definite functions to genes is that their protein products do not have fixed identities. For more than half a century molecular biology was dominated by what came to be called “Anfinsen’s dogma,” the doctrine that the polypeptide chains specified by a gene fold in unique fashions, and that the resulting proteins therefore perform similarly in all contexts. It is now recognized, however, that many proteins have one or more “intrinsically disordered” domains, and the context-dependent interactions among them constitute a protein-based system of inheritance that does not depend on changes in DNA. Intrinsic disorder is particularly prevalent among gene products that control the expression of other genes in complex, multicellular organisms, undermining standard ideas of how gene regulatory networks regulate embryonic development and organ physiology.
Thus, even the most precise alteration of a known gene with CRISPR is fraught with uncertainties. This may be worth the risk in an existing person with a disabling or mortal condition for which there is no other effective treatment. But it would never be so in an embryo, where the intention would be to improve a prospective individual’s biological characteristics. Certainly a trait could be altered by gene editing, but not without the possibility of deranging other traits that may well have turned out normally in the unmodified embryo. Stated differently, “engineering” an organism, in analogy to engineering a mechanism or machine, is an inapplicable notion.
Commentators writing about reproductive biotechnologies with an ethical orientation often express valid concerns about the prospects of inequitable distribution or eugenic hazards of the anticipated benefits of gene manipulation – improvements to health, intelligence, physical beauty – but rarely question the ability of the purveyors to deliver on their promises. It can be seen from the foregoing that, as with anyone else trying to sell something, it makes sense to look behind the smoke and mirrors.