replicated within bacteria, and could be transmitted between bacterial strains. After a long morning of presentations, Boyer fled to the beach for a respite, and spent the afternoon nursing a glass of rum and coconut juice.
Late that evening, Boyer ran into Stanley Cohen, a professor at Stanford. Boyer knew Cohen from his scientific papers, but they had never met in person. With a neatly trimmed, graying beard, owlish spectacles, and a cautious, deliberate manner of speaking, Cohen had the “physical persona of a Talmudic scholar,” one scientist recalled—and a Talmudic knowledge of microbial genetics. Cohen worked on plasmids. He was also an expert on Frederick Griffith’s “transformation” reaction—the technique needed to deliver DNA into bacterial cells.
Dinner had ended, but Cohen and Boyer were still hungry. With Stan Falkow, a fellow microbiologist, they strolled out of the hotel toward a quiet, dark street in a commercial strip near Waikiki beach. A New York–style deli, with bright flashing signs and neon-lit fixtures, loomed providentially out of the shadows of the volcanoes, and they found an open booth inside it. The waiter couldn’t tell a kishke from a knish, but the menu offered corned beef and chopped liver. Over pastrami sandwiches, Boyer, Cohen, and Falkow talked about plasmids, gene chimeras, and bacterial genetics.
Both Boyer and Cohen knew about Berg’s attempts to create gene-hybrids in the lab. Cohen also knew that Mertz, Berg’s graduate student, was making the rounds among the microbiologists at Stanford, seeking to learn techniques to transfer her novel gene hybrids into E. coli.
The discussion moved casually to Cohen’s work. Cohen had isolated several plasmids from E. coli, including one that could be reliably purified out of the bacteria, and easily transmitted from one E. coli strain into another. Some of these plasmids carried genes to confer resistance to antibiotics—to tetracycline or penicillin, say.
But what if Cohen cut out an antibiotic-resistance gene from one plasmid and shuttled it to another plasmid? Wouldn’t a bacterium previously killed by the antibiotic now survive, thrive, and grow selectively, while the bacteria carrying the non-hybrid plasmids would die?
The idea flashed out of shadows, like a neon sign on a darkening island. In Berg’s and Jackson’s initial experiments, there had been no simple method to identify the bacteria or viruses that had acquired the “foreign” gene (the hybrid plasmid had to be purified out of the biochemical gumbo using its size alone: A + B was larger than A or B). Cohen’s plasmids, carrying antibiotic-resistance genes, in contrast, provided a powerful means to identify genetic recombinants. Evolution would be conscripted to help their experiment. Natural selection, deployed in a petri dish, would naturally select their hybrid plasmids. The transference of antibiotic resistance from one bacterium to another bacterium would confirm that the gene hybrid, or recombinant DNA, had been created.
But what of Berg and Jackson’s technical hurdles? If the genetic chimeras were produced at a one-in-a-million frequency, then no selection method, however deft or powerful, would work: there would be no hybrids to select. On a whim, Boyer began to describe the DNA-cutting enzymes and Mertz’s improved process to generate gene hybrids with greater efficiency. There was silence, as Cohen and Boyer tossed the idea around in their minds. The convergence was inevitable. Boyer had purified enzymes to create gene hybrids with vastly improved efficiency; Cohen had isolated plasmids that could be selected and propagated easily in bacteria. “The thought,” Falkow recalls, was “too obvious to slip by unnoticed.”
Cohen spoke in a slow, clear voice: “That means—”
Boyer cut him off mid-thought: “That’s right . . . it should be possible. . . .”
“Sometimes in science, as in the rest of life,” Falkow later wrote, “it is not necessary to finish the sentence or thought.” The experiment was straightforward enough—so magnificently simple that it could be performed over the course of a single afternoon with standard reagents: “mix EcoR1-cut plasmid DNA molecules and rejoin them and there should be a proportion of recombinant plasmid molecules. Use antibiotic resistance to select the bacteria that had acquired the foreign gene, and you would select the hybrid DNA. Grow one such bacterial cell into its million descendants, and you would amplify the hybrid DNA a millionfold. You would clone recombinant DNA.”
The experiment was not just innovative and efficient; it was also potentially safer. Unlike Berg and Mertz’s experiment—involving virus-bacteria hybrids—Cohen and Boyer’s chimeras were composed entirely of bacterial genes, which they considered far less hazardous. They could find no reason to halt the creation of these plasmids. Bacteria, after all,