13—
f X
The genes were the goal and viruses were to be the key. Viruses bridge the boundary between the living and the nonliving. As do all life forms, they reproduce and undergo inheritable variation (mutation), but, lacking any metabolism, they can only do so within the confines of living cells.
Viruses were discovered toward the end of the nineteenth century as mysterious agents of infection. Invisible under an ordinary microscope, their features could not be discerned until the invention of the electron microscope. Many varied forms are known that can infect animal or plant or bacterial cells. Max Delbrück first recognized that viruses—in particular, bacteriophage, which infect the simpler bacterial cells—afford the potential for detailed analysis of the processes of reproduction and mutation.
Genes, known since the time of Mendel to be the determinants of hereditary traits, were for decades biochemically utterly obscure. Early on, they were thought, by a process of elimination, to most likely be protein. No other structures seemed sufficiently complex. As the true dimensions of the nucleic acids became more apparent, several lines of indirect evidence moved them, in some minds, into contention for the genetic role.
Whether proteins or nucleic acids, the properties of genes seemed nearly inexplicable. How could a complex substance, of whatever chemical form, be reproduced with high precision through thousands of cell generations? Even more strange was the fact that, when a gene was
altered, mutated, by some unknown process, the mutant form was then reproduced just as faithfully as the original. And the mutant form could in turn be mutated further—or restored to its original state. Some, including Delbrück, proposed that a new principle of nature, a new physical law, might be needed to account for these phenomena. This speculation proved to be wrong—the ordinary principles of chemistry proved to be sufficient—but the proposal stimulated much discussion and fruitful research.
The choice of the bacteriophage-bacterium system proved to be most felicitous, providing a continual stream of new insights over a period of some forty years. In this period, the tools of molecular biology were developed and the study of bacterial virus infection provided a straight-forward and rewarding field for their application and refinement. The simpler viruses proved to be only a coated set of specialized genes. By simultaneous infection of a population of bacteria, one could initiate within these cells a completely novel sequential pattern of genetic activity that could be dissected into discrete steps.
My research on the bacterial virus f X174 over a period of twenty-three years was a significant component of these advances, sometimes leading the stream, sometimes being borne by it. It was the centerpiece of my scientific career and is illustrative of the spectacular advances in our knowledge during this time.
Our aim was bold. We were to use this virus to pry open the processes of heredity and infection. We sought to find its genes, to count them, to deduce their mode of action, to understand their reproduction, to learn the basis for their stability and mutability, and to identify the agencies of their control. To know the genes of even one virus this well would be of profound importance in itself, as all life is related and the principles we found should be general. Knowing the genes would lead to understanding the nature of viral infection, to learning how the synthetic machinery of a cell is subverted to the production of more virus. And this might, in time, provide a basis for therapy of viral disease.
Initially, we had only the rudimentary knowledge that f X was probably small and could grow in certain strains of E. coli . At the end of our research, we knew the complete sequence of its DNA and the details of its genetic structure and had an extensive if incomplete picture of its architecture and the processes of its replication. This was a classic period for virology, especially bacterial, and research on f X was one of the central features. But the path was not straight, the way was often uncertain, and there were many surprises en route.
During my six months in Max Delbrück's laboratory in 1953, I had decided that while bacterial viruses provided in principle the simplest system in which to study DNA structure and expression, significant advance (using the techniques of that day) would require the use of the simplest of the bacterial viruses available. The T-even bacteriophages, then under the most intensive study, seemed much too complex to me. Proceeding on the (partially erroneous) assumption that the reproduction of smaller viruses would be simpler to analyze, I scoured the literature in search of small bacteriophages. Two—f X174, discovered in Paris, and S13, discovered in England—seemed suitable. The evidence as to their size was only qualitative and insecure, for each line of evidence could have alternative explanations. But these viruses seemed a good place to start. Remarkably, cultures of each of these were available in laboratories in France and England, and an available strain of E. coli proved to be a suitable host cell.
After my return to Ames, I began work on these viruses in 1953. Work with any new virus requires that it be "domesticated"—that one learn the better media in which to cultivate the host and the virus, the conditions suitable for its storage, how to purify it, and so on. In early experiments, f X proved hardier than S13, which lost viability rapidly on storage. From then on, we concentrated on f X though, as we later learned, these two viruses are in fact closely related.
After learning how to produce cultures of high infectivity, we increased the scale of culture to provide quantities (in milligrams) sufficient for purification. The most useful techniques involved various forms of centrifugation. After some stages of purification, we were able to associate infectivity with a particular component that, because it moved relatively slowly under centrifugal force, supported our hope that the virus would be small. I then examined, in the electron microscope, virus from preparations that appeared nearly pure in the ultracentrifuge. Small, approximately spherical particles about twenty-five to thirty nanometers in diameter appeared, providing evidence that the total mass could not exceed eight to ten million daltons.
All viruses, of the small number that had been analyzed, were known to contain a nucleic acid, either DNA or RNA. We assumed f X would too, but which would it be? No RNA bacteriophage were then known, but animal and plant viruses containing RNA were common. After trying several methods, we were able to separate the viral protein from its nucleic acid and demonstrate that the latter was DNA. But an oddity. was immediately evident; its rate of movement under centrifugal force
was such that were it a conventional double-stranded DNA, its molecular weight would exceed that of the entire virus. How could this anomaly be resolved?
The research to this point had required three years. I reported it at the ASBC meeting in April 1957.
Research was interrupted for a time by my move to Caltech in the summer of 1957, but I resumed the phage studies as soon as possible.
I set up a light-scattering apparatus at Caltech that permitted me to ascertain first the absolute molecular weight of the f X virus particle (6.2 million daltons) and then that of its DNA, which proved to be 1.7 million daltons. This molecular weight, together with its centrifugal and other properties, suggested strongly that the DNA could not be the usual stiff, double-helical form. Could it be a single-stranded DNA, then unknown in nature? Or more likely, was it a denatured, crumpled form of a double-stranded DNA in which the stiffening bonds between the strands had been disrupted somehow by the extraction procedure?
Several lines of evidence suggested that the state of DNA inside the virus particle was no different than that we observed after extraction. But the convincing fact proved to be the quantitative determination of the nucleotide composition of the DNA by the methods we had developed in our earlier work. This DNA did not have the Chargaff equalities of A = T and C = G. The DNA of f X had to be a single-stranded, completely novel form.
These results were published to some astonishment in the spring of 1959. In June of that year, in a lecture at the Brookhaven Symposium, I compared this discovery to that of "finding a unicorn in the ruminant section of the zoo."
In the six years after its formulation, the double helix structure of DNA had become dogma. It so neatly accommodated the known facts concerning DNA structure and replication that the proposed presence of a single-stranded DNA—in of all things a virus, a form specialized for reproduction—aroused amazement and doubt and stirred the imagination. Soon, many of the best graduate students at Caltech wanted to work in my laboratory, and I received numerous applications from postdoctoral fellows from around the world. Success breeds success, and the talent thus attracted surely facilitated further progress.
How could the single-stranded DNA of the f X virus reproduce? Was there another mechanism in addition to the mode of complementary replication implied by the Watson-Crick model of double-stranded DNA and demonstrated subsequently by the famous Meselson-Stahl
experiment? To study this question, we undertook to follow the fate of the viral DNA after it was introduced into the bacterial cell. This was a novel, even daring, proposition that called on several of the newly developed techniques of molecular biology. Many of these techniques—the use of nonradioactive and radioactive isotopes, light scattering, the various forms of ultracentrifugation—were based on specific applications of physical principles. My background in biophysics served me well. The small size of the f X DNA, which permitted us to manipulate it with minimal degradation, was also critical.
We developed three distinct means to identify viral DNA within the infected cell and to determine its integrity. The most rigorous of these required measurement of the infectivity of naked viral DNA. According to the most common but not yet universally accepted theory, the protein coat of the virus merely served to protect the DNA and assure its entry into the infected cell. Once inside, the DNA, the genetic material alone, was thought probably sufficient to carry on the infection and give rise to progeny virus particles. If this were so, the free DNA would be infective if we introduced it into a bacterial cell.
We were able to accomplish this by preparing, under appropriate conditions, bacterial protoplasts—cells that lacked a layer of their cell wall and therefore were somewhat permeable to the free DNA alone. George Guthrie, then a graduate student, developed this technique and was able to achieve an infective efficiency of one infected protoplast per one hundred to one thousand viral DNA molecules. This figure was low compared to that of intact virus, but it was sufficient for our purpose. Theory aside, this was the first complete demonstration that a viral DNA was sufficient for infection. Combining all of the techniques, we could infect a cell and subsequently identify, track, and assay the infective potential of the intracellular DNA molecules of the parental virus. We could isolate these at various times after infection.
These combined experiments demonstrated conclusively that the parental viral f X DNA remained intact and infective on infection of the host cell. But within one to two minutes it was converted to a different form of DNA, which we called the "replicative form." Physical and chemical studies of the replicative-form DNA demonstrated that in fact it was the double-stranded version of the single-stranded viral DNA, paired with its newly made complementary strand. This replicative form was also infective to protoplasts. Further studies revealed that, after its initial formation, the replicative form reproduced as such, in the usual semiconservative manner of double-stranded DNA to produce some ten
to twenty copies. This set of replicative-form DNA molecules then served as the template from which the distinctive single-stranded DNA of the progeny virus particles was made during the latter half of the infection. So replication of this single-stranded DNA in fact proceeded by the usual, complementary strand mechanism, albeit with some atypical aspects.
However, another major surprise was now in store for us. We had available from f X, for the first time, a homogeneous DNA of defined size (about fifty-four hundred nucleotides long) and nucleotide composition, with a specific biological activity (infection). It seemed a reasonable question to inquire whether there was anything special about the ends of this DNA. Was there a specific nucleotide or a specific sequence at either end of the chain? (Nucleic acid chains have a polarity and the two ends are distinguishable.) We had available two enzymes that we knew would degrade a polydeoxynucleotide chain from respectively one or the other end taking off one nucleotide at a time. Walter Fiers, a postdoctoral fellow, undertook to study the effects of these enzymes on the purified viral DNA and to characterize the products of the enzymatic action.
His results, repeated many times, were surprising. When he used highly purified enzymes, only a small fraction of the viral DNA appeared as mononucleotides, and these included all four possible in similar proportions. Further, there was no perceptible decline in the infectivity of the DNA after the enzyme digestion. Clearly, the bulk of the viral DNA, all of the infective DNA, was resistant to the action of these enzymes. Why? Were both ends of the DNA "blocked" by some special chemical group?
One day while walking back to the laboratory after lunch at the Athenaeum and pondering this question, I began to consider seriously a possibility we had occasionally tossed off in jest that the DNA might really be a ring—not a linear molecule, but a ring. A ring would of course have no ends to degrade, but if it were a ring it would behave slightly differently—it would move slightly faster under centrifugal force—than would the corresponding linear DNA of similar weight. Now in fact, in all of our DNA preparations, we had actually observed two components in the ultracentrifuge with closely similar rates of movement, present in varying amounts in different preparations. Preparations with the greater proportion of the faster-moving component had appeared to have higher specific infectivity. Our more recent preparations, made with better technique, had only small amounts of the
slower component We had had no explanation for this observation.
If the viral DNA were a ring, which would be the faster component, then the slower component could be rings that had accidentally been cleaved, linearized, at some random point and were therefore noninfective. Further, the ends of such randomly cleaved rings could account for the small yield of terminal nucleotides of all four kinds that we had observed on enzyme treatment.
An experiment to test this hypothesis became clear to me. I would start with a preparation of (almost) pure rings, then cleave them deliberately and slowly with an enzyme that would break the chains randomly. The proportion of rings cleaved could be followed by the loss of infectivity. As the rings were cleaved, DNA molecules would be transferred, first from the faster component to the slower component and then, as a second cleavage was made, out of the slower component into smaller fragments of random size.
The following afternoon I worked out the equations relating residual infectivity to the proportions of DNA in the first component, second component, and smaller fragments. Walter Fiers performed the experiment; it precisely confirmed the theoretical expectation. f X DNA had to be a ring.
I still consider this experiment, combining measurements of biological infectivity, ultracentrifugal analysis, and enzymatic activity, as probably the most elegant with which I have been involved.
The result, which I thought unarguable, met initially with considerable skepticism, particularly from nucleic acid biochemists. They were wedded to the concept of a polymer that, starting at one end, was extended unit by unit to the other. A ring molecule provided no purchase for such extension. If the single-stranded viral DNA was a ring, was the double-stranded replicative form also a ring? Alice Burton, then a postdoctoral fellow, carried out the analogous experiment, degrading the replicative form with an enzyme known to cleave both strands at the same site. She thus demonstrated its conversion to a linear, slower moving form in equal proportion with its loss of infectivity.
Then came an opportunity to satisfy the skeptics—seeing is believing. Dr. Albrecht Kleinfelter, then at UC Berkeley, had developed a method to visualize DNA in the electron microscope. It could only, however, be applied to double-stranded DNA. I took some replicative-form DNA to his laboratory, and we prepared the electron microscope grids and took pictures. When we examined the small pictures with a hand lens, the very first DNA we saw was a ring. Indeed, the field was filled with
double-stranded DNA rings of the circumference to be expected for the known weight of DNA. The skeptics were convinced.
Later, when it became possible to visualize single-stranded DNA in the electron microscope, the rings of the viral DNA were quite obvious.
Why had this result been so hard to deduce? In retrospect, we had been mesmerized by the predominantly linear character of DNA. To have a ring, one had to have an intact DNA. Prior to f X, everyone had studied fragments of DNA. Since then, many other ring DNAs have been found in viruses, plasmids, mitochondna, chloroplasts, and so on. In fact, it has become clear that for those DNAs that are not rings (as in eucaryotic chromosomes) special means must be provided to replicate their termini.
Now we had a complete DNA molecule, the complete genome of this virus. It was all there in our test tube. But how did it multiply? How did this relatively small molecule subvert the machinery of the infected cell to achieve its own reproduction?
Biochemically, the next logical step in our research would have been to determine the nucleotide sequence of the viral DNA, which could plausibly be expected to lead us to the genes through the genetic code then being deciphered. However, the technology to do this was not then available. We were able to separate out all the pyrimidine nucleotide tracts from the intervening purine nucleotide tracts and then to fractionate these according to size (thus learning that the longest consecutive run of pyrimidine nucleotides was eleven), but we could neither further fractionate nor sequence these tracts. So the information gain was limited.
We turned, therefore, to genetics. Using the recently developed methodologies for conditional lethal mutations (i.e., mutants able to grow under one set of conditions or in certain cells but unable to grow under other conditions or in other hosts), we were able to obtain a large number of mutants of f X. These could be classified into "complementation groups," or genes, by determining whether a successful infection would ensue in a cell infected with two different mutants. If so, then each mutant must supply the function defective in the other so that they were in different complementation groups. If such a joint infection were not successful, it was plausible that the mutants were from the same complementation group, that is, defective in the same gene. Nine complementation groups were discovered and named A–H and J. The mutations could be mapped—arranged in a linear order along a (necessarily circular) path—by the usual means of recombinant formation in genetic
crosses. Only the mutants in genes D and E could not be consistently ordered for reasons that much later became apparent.
Studies of the abortive infective process in cells infected with these mutants provided clues as to the functions of the several genes. Gene E, for instance, was required for disruption of the cell membrane at the end of the infection, thereby releasing the progeny virus. In gene E mutants, the membrane failed to disrupt and viral synthesis continued extensively. As we could then artificially disrupt these cells, these mutants were useful to provide high yields of virus. Some genes proved necessary for early stages of DNA synthesis. Others were crucial for intermediate stages of viral particle assembly, while others coded for the four proteins of the viral coat.
Parallel with these studies, a long-term collaboration with Arthur Kornberg of Stanford proceeded intermittently. Professor Kornberg, a distinguished enzymologist, had devoted years to the biochemistry of DNA synthesis and had isolated and purified an enzyme (DNA polymerase) from the E. coli bacterium that could synthesize DNA using as a template whatever DNA it was given to copy. While chemical analysis indicated that the enzyme was making faithful copies, the ultimate test would be to demonstrate that the copy DNA had a true biological activity. With its defined size and a functional assay (infectivity), f X DNA was an obvious choice to test the faithfulness of Kornberg's enzyme. Could it make infective DNA copies in vitro?
Using our DNA, Kornberg's enzyme, and our infectivity assay, we tried in 1960. While considerable amounts of DNA were made in excess of that initially added, infectivity fell rapidly. Attributing this result to impurities (degradative enzymes) in the DNA polymerase preparation, we tried the experiment again in 1963 with a more highly purified enzyme. The same result.
With the discovery that f X DNA was a ring, it became evident that effective synthesis would require an agent to close the newly made DNA strands into a continuous circle. A recently discovered enzyme, DNA ligase, had the ability to do this.
In 1967, with the help of Mehran Goulian, a postdoctoral fellow in Kornberg's laboratory, we set out again to make infective DNA using DNA polymerase together with DNA ligase and f X DNA. To insure that the synthesized (and, we hoped, infective) DNA was distinct from the input viral DNA, we incorporated a density label so that the product could be physically separated from the template. The experiment was eminently successful. Together, we had made for the first time an in-
fective agent in the test tube capable of thereafter reproducing itself indefinitely in host cells.
In truth, we had only copied faithfully an existent design. But our copies were, in fact, potentially immortal.
We reported this work in the December 1967 issue of the Proceedings of the National Academy of Sciences . By some circumstance, this report landed on the desk of an aide to President Johnson on the day the president was to speak at the National Institutes of Health. It was incorporated into the president's speech as "the nearest approach to the synthesis of life in a test tube," an instance of the remarkable research supported by NIH. As a consequence, this result was reported in newspapers and news magazines everywhere. A result of such coverage was the receipt of many letters, some from long lost friends or relations, some quite touching. The newspaper articles, referring to the small size of f X, sometimes called it a "dwarf virus," and I received several letters from people suffering from dwarfism who asked if my research could alleviate their affliction Sadly, not.
In the early 1970s, our ability to analyze DNA was greatly augmented by two developments. The restriction enzymes were discovered, which cleaved DNA only at specific nucleotide sequences. Using these, a DNA such as that of f X could be broken reproducibly into a small number of specific, isolatable regions. By the use of partial digests, Amy Lee in my laboratory was able to order these regions into a continuous map, the first complete restriction fragment map of a DNA. By hybridizing one or more of these fragments with the viral DNA, Lee Compton was able to visualize specific regions of phage DNA in the electron microscope.
Even more significant, Fred Sanger at Cambridge University was developing means to sequence the DNA of restriction fragments nucleotide by nucleotide. As a defined DNA of specific size with an ordered restriction fragment map, f X was the natural choice for Sanger's first application of this technique. After obtaining his Ph.D. in my group, Clyde Hutchison went to Sanger's laboratory to help with this project. In 1977, Sanger announced the complete nucleotide sequence of f X, 5,386 nucleotides. This was the first complete sequence of any DNA, a major accomplishment. As such, it marked the end of an era of f X research and the beginning of a vast new period of DNA research.
Sanger's sequence indicated clearly the location of the genes we had laboriously sited on the genetic map. It also contained one major surprise. We had been unable to map genes D and E in a consistent man-
ner; the reason now became clear. The two genes used in large part the same region of DNA, reading it in different, overlapping trinucleotide frames!
This possibility had been discussed before but was usually dismissed with the argument that such an overlapping pattern would result in such great constraints of the choice of amino acids that it seemed very unlikely that two functional proteins could result. Yet here it was!
In small viruses such as f X, the physical constraint on the size of the DNA evidently drives the virus in an evolutionary sense to make the very maximum use of its fifty-four hundred nucleotides. Other small overlaps, at the ending of one gene and the start of the next, attest to the same pressure, with the use of one region of DNA to code for portions of two proteins.
At this same time, f X was also serving an important role in the attempt in Arthur Kornberg's laboratory to reconstruct from E. coli extracts an entire system able to initiate, extend, and complete DNA synthesis. The use of f X DNA as a small, defined template enabled this laboratory to extract and successfully refine the many factors needed, in addition to the DNA polymerase and ligase, to reproduce DNA in a truly biological fashion. Out of this work has come the recognition that while the smaller viruses, with less genetic material, contribute fewer components to the infective process, the host cell must therefore contribute more. In brief, the smaller viruses are the more parasitic; their infective process, resulting in viral reproduction, may not be appreciably simpler than for larger viruses.
While certain aspects of f X replication, in particular the manner of particle assembly, remained obscure, by 1977 f X had fulfilled its primary role in the advance of molecular biology. My original conception in undertaking f X studies—the need for a simple, defined, biologically assayable DNA as an experimental subject—had been validated. The research with f X had played a significant role in a classic period of advance in our understanding of genes. The technology was now available to attack more complex problems.
This tale—the unraveling and elucidation of the structure and process of replication of one virus strain—can serve as a prototype for much of biological research. Starting with but a few empirical observations, each finding led to deeper questions of detail or process. To answer these questions required the development, application, and refinement of new techniques and new approaches. Some came from our laboratory; others were borrowed from Kornberg's laboratory, or Sanger's,
or were culled from the literature on restriction enzymes, conditional lethal mutations, or ultracentrifugation. Out of this study of one virus came often surprising discoveries of great generality—that there are viruses containing single-stranded DNA; that many DNAs are rings; that single-stranded DNAs reproduce via the conventional complementary mechanisms; that, in small viruses, the same DNA region can be used to determine more than one protein—as well as the important demonstrations that an infective DNA can be synthesized in vitro and that a complete genome can be sequenced.
These are the rewards of a sustained, collective effort: meaningful contributions to human knowledge. They were not easy won—one does not write of the failed experiments, the misconceived, the accidents, the equipment failures, the blunders. One cherishes the nuggets of knowledge, and the students who mastering new skills on these problems have gone on to make advances in many other fields.