Dna: The Secret of Life

by Watson, James

  Indeed, recombinant technology owes its very existence to the extraordinary ability of bacteria to incorporate new pieces of DNA (usually plasmids). Not surprisingly, then, microbial evolution too bears the footprint of dramatic gene-importing events of the past. E. coli, normally a benign inhabitant of our intestines (and of petri dishes), has morphed through gene importation into a killer variant. The toxins produced by one strain that occasionally causes outbreaks of food poisoning (killing twenty-one people in Scotland in 1996-97) and headlines about "Killer Burgers" are attributable to massive genetic "borrowing" from other species.

  Genetic material normally moves vertically down a lineage – from ancestor to descendant – so this importation of DNA from outside is known as "horizontal transfer." Comparison of the genome sequence of normal E. coli to that of the pathogenic strain has revealed a shared genetic "backbone," identifying both strains as members of a common species, but there are many "islands" of divergent DNA unique to the pathogen. Overall, the pathogen lacks 528 of the normal strain's genes and has instead a staggering 1,387 genes not present in the normal strain. In that 528-for-1,387 exchange lies the key to the transformation of one of nature's most innocuous products into a killer.

  Other bacterial nasties also show similar evidence of wholesale horizontal transfer. Vibrio cholerae, the agent of cholera, is unusual for a bacterium in that it has two separate chromosomes. The larger one (about 3 million base pairs in length) appears to be the microbe's original equipment, containing most of the genes essential to the functioning of the cell. The smaller one (about 1 million base pairs in length) seems to be a mosaic, made up of bits and pieces of DNA imported from other species.

  Complex organisms, especially large ones like humans, are by design fairly inviolable gatekeepers of their own internal biochemistry: in most cases, if we don't ingest or inhale a substance, it cannot alter us profoundly. And so the biochemical processes of all vertebrates have tended over time to remain very similar. Bacteria, on the other hand, are much more exposed to the chemical vagaries of the environment; a colony may find itself suddenly awash in a noxious chemical – say, a disinfectant like household bleach. Little wonder these highly vulnerable organisms have evolved a stunning variety of chemistries. Indeed bacterial evolution has been driven by chemical innovation, the invention of enzymes (or the retrofitting of old ones) to do new chemical tricks. One of the most fascinating and instructive instances of this evolutionary pattern occurs among bacteria whose secrets we have only recently begun to learn about, a group known collectively as the "extremophiles" because of its members' predilection for the most inhospitable environments.

  Bacteria have been found in Yellowstone hot springs (Pyrococcus furiosus thrives in boiling water and freezes to death at temperatures below 70°C [158°F]) and in the superheated water of deep-sea vents (where the high pressure at depth prevents the water from boiling). They have been found living in environments as acidic as concentrated sulfuric acid and in acutely alkaline environments as well. Thermophila acidophilum is an all-around extremophile, withstanding, as its name suggests, both high temperatures and low pH. Some species have been discovered in rocks associated with oil deposits, converting oil and other organic material into sources of cellular energy, rather like so many tiny sophisticated automobiles. One of these species inhabits rocks a mile or more down and dies in the presence of oxygen; appropriately, it is named Bacillus infernus.

  Perhaps the most remarkable microbes discovered in recent years are the ones that subvert what was once considered a key dogma of biological science – that all energy for living processes comes ultimately from the sun. Whereas even Bacillus infernus and oil-consuming bacteria found in sedimentary rocks are connected to the organic past – the sun shone eons ago on the plants and animals whose remains are today's fossil fuels – so-called lithoautotrophs are capable of extracting the nutrients they need from rocks created de novo by volcanoes. These rocks – granite is an example – bear no traces of organic material; they contain no vestige of the energy of sunny prehistoric days. Lithoautotrophs have to construct their own organic molecules out of these inorganic materials. They live, literally, on a diet of rock.

  There has been no more persuasive indicator of our general ignorance of the microbial universe than our belated discovery of the bacterial genus Prochlorococcus, whose planktonic cells photo-synthesize as they float in the open ocean. As many as 200,000 may inhabit a single milliliter of seawater, making this arguably the most abundantly represented species on the planet. It is certainly responsible for a huge proportion of the ocean's contribution to the global food chain. And yet Prochlorococcus was unknown to us until 1988.

  The extraordinary microbial universe around us reflects the phenomenal power of eons of natural selection. Indeed the history of life on our planet can be told mostly as a tale of bacteria; more complicated organisms, ourselves included, are embarrassingly late arrivals – a virtual afterthought. Life appears to have originated as bacteria some 3.5 billion years ago. The first eukaryotes – cells whose genes are enclosed within nuclei – arose around 800 million years later, but they remained single-celled for about a billion years after that. Only about half a billion years ago did the breakthroughs occur that would ultimately give rise to the likes of the earthworm, the fruit fly, and Homo sapiens. The predominance of bacteria is reflected in the DNA-based reconstruction of the tree of life first carried out by Carl Woese at the University of Illinois: the tree of life is a bacterial tree, with a few multicellular beings forming a late-growth twig. Now generally accepted, Woese's ideas were at first strenuously opposed within the biological establishment. Still some of the implications of the DNA-based approach to the tree of life have been difficult to take: they have shown, for instance, that animals are not, as was once supposed, closely related to plants; rather, the closest relatives of animals are fungi. Humans and mushrooms stem from the same evolutionary root.

  The Human Genome Project has proved Darwin more right than Darwin himself would ever have dared dream. Molecular similarities stem ultimately from the way in which all organisms are related through common descent. A successful evolutionary "invention" (a mutation or set of mutations that is favored by natural selection) is passed down from one generation to the next. As the tree of life diversifies – existing lineages splitting to produce new ones (reptiles persist as such, but also bud off into both bird and mammal lineages) – that invention may eventually appear in a huge range of descendant species. Some 46 percent of the proteins we see in yeast, for example, also appear in humans. The yeast (fungal) lineage and the one that ultimately gave rise to humans probably split about 1 billion years ago. Since each has subsequently developed independently, free to follow its own evolutionary trajectory, there have been in effect 1 billion years of evolutionary activity since that yeast/human common ancestor; and yet, through all that time, that set of proteins that existed in the common ancestor has changed only minimally. Once evolution solves a particular problem – for example, designing an enzyme to catalyze a particular biochemical reaction – it tends to stick with that solution. We have seen how this kind of evolutionary inertia is responsible for the centrality of RNA in cellular processes: life started in an "RNA world," and the legacy remains with us to this day. And the inertia extends to the biochemical details: 43 percent of worm proteins, 61 percent of fruit fly proteins, and 75 percent of fugu proteins have marked sequence similarities to human proteins.

  Comparing genomes has also revealed how proteins evolve. Protein molecules can typically be envisioned as collections of distinct domains – stretches of amino acid chains that have a particular function, or form a particular three-dimensional structure – and evolution seems to operate by shuffling domains, creating new permutations. Presumably most new permutations are as useless as they are random, doomed to be eliminated by natural selection; but in the rare instance that a new permutation proves beneficial, a new protein is born. Some 90 percent of the domains that have
been identified in human proteins are also present in fruit fly and worm proteins. In effect, therefore, even a protein unique to humans is likely nothing more than a reshuffled version of one found in Drosophila.

  There is no better demonstration of this fundamental biochemical similarity among organisms than so-called rescue experiments, the aim of which is to eliminate a particular protein in one species and then use the corresponding protein from another species to "rescue" the missing function. We have already seen this strategy implemented in the case of insulin. Because human and cow insulins are so similar, diabetics who fail to produce their own can be given the cow version as a substitute.

  In an example evocative of B-movie science fiction, researchers have been able to induce fruit flies to grow eyes on their legs by manipulating a particular gene that specifies where an eye should go. That gene then induces the many genes involved in producing a complete eye to go to work in that designated location. The mouse's corresponding gene is so similar to the fruit fly's that it will perform the same function when situated – by the genetic engineer's sleight of hand – in a fruit fly whose gene has been eliminated. That this can be done is nothing less than remarkable. Fruit flies and mice have been separated by evolution for at least half a billion years, so – following the logic applied above to humans and yeast evolving simultaneously along independent lines – the gene has in fact been conserved over a billion years of evolution. This is all the more astonishing when we consider that fruit fly and mouse eyes have fundamentally different structures and optics. Presumably each lineage perfected an eye appropriate for its respective purposes, but the basic machinery for determining the location of that eye, needing no improvement, stayed the same.

  The most humbling aspect of the Human Genome Project so far has been the realization that we know remarkably little about what the vast majority of human genes do. To use the hard-won information properly requires us to devise methods for studying the function of genes on a genomewide scale.

  In the wake of the HGP, two new postgenomic fields have duly emerged, both of them burdened with unimaginative names incorporating the "-omic" of their ancestor: proteomics and transcriptomics. Proteomics is the study of the proteins encoded by genes. Transcriptomics is devoted to determining where and when genes are expressed – that is, which genes are transcriptionally active in a given cell. If the genome is ultimately to be understood in its more dynamic reality, not as a mere set of instructions for life's assembly but as the screenplay for life's movie – all the drama described in the precise order it is meant to occur – then proteomics and transcriptomics provide the keys to glimpsing the live action. The more we learn, the more we see of Life, the Movie.

  We have long appreciated that a protein is a great deal more in biological terms than the linear string of amino acids that compose it. How the string folds up to produce a distinctive three-dimensional configuration is really the key to its function – what proteomics seeks to know. Structural analysis is still done using X-ray diffraction: the molecule is bombarded with X rays that bounce off its atoms and scatter in a pattern from which the three-dimensional shape may be inferred. In 1962, my one-time colleagues at the Cavendish Lab at Cambridge University, John Kendrew and Max Perutz, received the Nobel Prize in Chemistry for their elucidation of the structures of, respectively, myoglobin (which stores oxygen in muscle) and hemoglobin (which transports oxygen in the bloodstream). Theirs was a monumental effort. The complexity of the X-ray diffraction images they had to interpret made me appreciate the relative simplicity of DNA!

  Knowledge of a protein's three-dimensional structure greatly assists the work of medical chemists in their hunt for new drugs that work, as many do, by inhibiting protein functioning. In the ever more specialized and automated world of pharmaceutical research, several companies now offer to determine the structure of proteins as if they were production-line commodities. And the work is now immeasurably easier than it was in the day of Perutz and Kendrew: with more powerful X-ray sources, automated data recording, and faster computers driven by increasingly clever software, the time needed for solving a structure can be reduced from many years to a matter of weeks.

  All too often, however, the three-dimensional structure itself provides no particular indication of that protein's function. Important clues may come instead from studying how the mystery protein interacts with other known ones. A simple way to identify such interactions involves spotting out samples of a set of known proteins on a microscope slide and then dousing them with the mystery protein, which has been previously treated so it will fluoresce under UV light. Where our test protein "sticks" to a particular spot on the slide's protein grid, it has become bound to the protein in that spot, causing it too to become fluorescent. Presumably, then, these two proteins are engineered to interact within the cell.

  Ideally, to know life's screenplay, to "see" life's movie, we need to discover all the precise changes in protein composition that occur over the individual's development, from the moment of fertilization all the way through to adulthood. Though many proteins will be found to be active throughout the process, some will prove specific to a particular developmental stage, so in each growth phase we should expect to see different sets of proteins. Adult and fetal hemoglobins, for example, are subtly different. Similarly, each variety of tissue produces its own profile of proteins.

  The most reliable way to sort out the various proteins from a given tissue sample is still the long-established method that uses two-dimensional gels to separate protein molecules on the basis of differences in their electrical charge and molecular weight. The several thousand protein spots thus differentiated can then be analyzed with a mass spectrometer, an instrument that can determine each one's amino acid sequence. Unfortunately, to apply proteomics like this to the vast number of proteins coded by an entire genome requires more funding than academic scientists typically have. For the most part, such expensive enterprises are left to the better-endowed researchers of large pharmaceutical companies. But because of the method's limitations, even their labs can't routinely find proteins that are present in very small amounts (see Plate 47).

  This type of high-throughput proteomics, with all its expensive hardware and industrial-scale automation of complicated procedures, is therefore not the way most scientists nowadays study gene function at the genome level. Instead, methods of transcriptomics have been adopted, because they are cheaper and easier to apply: the functioning of all genes in a genome can be tracked by measuring the relative amounts of their respective messenger RNA (mRNA) products. If you are interested in the genes being expressed in, say, a human liver cell, you isolate a sample of mRNAs from liver tissue. This represents a snapshot of the mRNA population in the liver cell: very active genes, those most heavily transcribed and that produce many mRNA molecules, will be more abundantly represented, whereas genes that are rarely transcribed will contribute only a few copies to the mRNA sample.

  The key to transcriptomics is a surprisingly simple invention known as a DNA microarray. Imagine a microscope slide with a grid of 25,000 tiny dot-shaped wells etched onto it. Using precise micropipetting techniques, DNA sequences from just one gene are deposited in each well so that the grid contains every gene in the human genome. Critically, the location on the microscope slide of each gene's DNA is known. Affymetrix, a company near Stanford, has managed to miniaturize these arrays even further by etching them onto a sliver of silicon the size of a small computer chip, yielding a "DNA chip."

  Using standard biochemical techniques, you can tag your liver mRNAs with a chemical marker so, like the proteins mentioned above, they will fluoresce obligingly under UV light. Then comes the step where the power and simplicity of the technique becomes wonderfully apparent: you simply dump your sample of mRNAs onto the microarray with its minuscule chessboard of 25,000 gene-filled wells. The very same base-pairing bonds that hold together the two strands of the double helix will compel each mRNA molecule to pair off with the gene from w
hich it was derived. The complementarity is precise and foolproof: the mRNA from gene X will bond only to the very spot occupied by gene X on the microarray. The next step is merely to observe which spots have picked up the fluorescent mRNAs. One spot on the microarray may show no fluorescence, implying that there was no complementary mRNA in the sample – and thus, we may infer, no active transcription of that gene in the liver cell. On the other hand, a number of spots do fluoresce, some with particular intensity; this indicates that many mRNA molecules have bound to it. Conclusion: a very active gene. Thus, with a single simple experimental assay, you have identified every one of the genes active in the liver. And such molecular panoramas have been made possible thanks to the success of the Human Genome Project and the new mind-set it has ushered into biology: we no longer need be content to study bits and pieces – we can now see the whole picture in all its spectacular glory.

  It is hardly surprising that Stanford's Pat Brown, one of the method's leading practitioners, sees DNA microarrays as "a new kind of microscope." Marveling at the technology's potential to reveal a whole new genetic universe, he has declared: "We're toddlers now just starting to discover our world."