Shadows of Forgotten Ancestors

You can see that the substitution of one nucleotide for another might have only minor consequences—you could, for example, substitute one structural amino acid for another (in the “handle” of the machine tool) and in no way change what the resulting protein does. But it could also have a catastrophic effect: A single nucleotide substitution might convert the instructions for making a particular amino acid into the signal to stop the transcription; then, only a fragment of the molecular machine in question will be manufactured, and the cell might be in trouble. Organisms with such altered instructions will probably leave fewer offspring.

The subtlety and nuance of the genetic language is stunning. Sometimes there seem to be overlapping messages using the same letters in the same sequence, but with different functional import depending on how it’s read: two texts for the price of one. Nothing this clever occurs in any human language. It’s as if a long passage in English had two completely different meanings,⁹ something like

ROMAN CEMENT TOGETHER NOWHERE …

and

ROMANCEMENT TO GET HER NOW HERE …

but much better—on and on for pages, perfectly lucid and grammatical in both modes, and, we think, beyond the skill of any human writer. The reader is invited to try.

In “higher” organisms, many long sequences seem to be nonfunctional genetic nonsense. They lie after a “STOP” and before the next “START” and generally remain ignored, forlorn, untranscribed. Maybe some of these sequences are garbled remnants of instructions that, long ago, in our distant ancestors, were important or even keys to survival, but that today are obsolete and useless.* Being useless, these sequences evolve quickly: Mutations in them do no harm and are not selected against. Maybe a few of them are still useful, but elicited only under extraordinary circumstances. In humans some 97% of the ACGT sequence is apparently good for nothing. It’s the remaining 3% that, as far as genetics goes, makes us who we are.

Startling similarities among the functional sequences of As, Cs, Gs, and Ts are seen throughout the biological world, similarities that could not have come about unless—beneath the apparent diversity of life on Earth—there was an underlying and fundamental unity. That unity exists, it seems clear, because every living thing on Earth is descended from the same ancestor 4 billion years ago; because we are all kin.

But how could machines of such elegance, subtlety, and complexity ever arise? The key to the answer is that these molecules are able to evolve. When one strand is making a copy of the other, sometimes a mistake occurs and the wrong nucleotide—an A, say, instead of a G—will be inserted into the newly assembled sequence. Some of them are honest replication errors—good as it is, the machinery isn’t perfect. Some are induced by a cosmic ray or another kind of radiation, or by chemicals in the environment. A rise in temperature might slightly increase the rate at which molecules fall to pieces, and this could lead to mistakes. It even happens that the nucleic acid generates a substance that alters itself—perhaps thousands or millions of nucleotides away.

Uncorrected mistakes in the message are propagated down to future generations. They “breed true.” These changes in the sequence of As, Cs, Gs, and Ts, including alterations of a single nucleotide, are called mutations. They introduce a fundamental and irreducible randomness into the history and nature of life. Some mutations may neither help nor hinder, occurring, for example, in long, repetitive sequences—containing redundant information—or in what we’ve called the handles of the molecular machine tools, or in untranscribed sequences between STOP and START. Many other mutations are deleterious. If you’re crafting superb machine tools and, while you’re not looking, someone introduces a few random changes into the computer instructions for manufacture, there isn’t much chance that the resulting machines, built according to the new, garbled instructions, will work better than the earlier model. Enough random changes in a complex set of instructions will cause serious harm.

But a few of the random changes, by luck, prove advantageous. For example, the sickle-cell trait we mentioned in the last chapter is caused by the mutation of a single nucleotide in the DNA, generating a difference of a single amino acid in the hemoglobin molecules that nucleotide helps code for; this in turn changes the shape of the red blood cell and interferes with its ability to carry oxygen, but at the same time it eventually kills the plasmodium parasites those cells contain. A lone mutation, one particular T turning into an A, is all it takes.