by Paul Rabinow

Early in January 1984 Mullis shifted his target once again. Perhaps, he reasoned, custom designing the target would facilitate amplification. [1] He shifted from the lambda phage system to a 100 base-pair synthetic oligonucleotide he made in his lab. During the course of the next several months, he began working on distinguishing beta-globin alleles.

The results were disappointing.

In the spring - approximately one year after Mullis conceived of PCR - he returned to the 58-base-pair region of the human beta-globin gene containing the sickle-cell mutation that Erlich's group had been working on. By June 1984 Mullis was satisfied that he had succeeded in achieving some amplification. In a report he submitted at the beginning of June, covering his work from June '83 to June '84, Mullis wrote:

These claims would eventually prove to be true.

Arnheim had little contact with Mullis during the year 1983-84. During the course of his stay there, Arnheim came to adopt the view of Mullis as someone with many "blue-sky" ideas and few data to back them up. At the beginning of the summer of 1984, Arnheim returned to Stony Brook to ponder a job offer to head Cetus's Human Genetics department, which he ultimately accepted. At the end of the summer, the day he came back to Cetus, White outlined for him Mullis's PCR idea. This was the first time Arnheim had heard a cogent explanation of the principle of PCR. It would be prudent, White explained, to explore the approach further, even though, in his opinion, Mullis had not been able to get it to work. Specifically, White was impressed by Mullis's strategy of multiplying the target as a way to increase sensitivity. It might well be possible, White told Arnheim, to increase the target ten- or even fiftyfold, a qualitatively significant increase in sensitivity for diagnostic purposes.

During the summer of 1984, Arnheim had heard vague references to the Mullis method and remembers, "No one but Kary believed the data he showed ... demonstrated what he said it did." White adds, "Any independent scientist looking at the data today would come to the same conclusion." [3] White thought that Arnheim and Erlich were the people at Cetus best able to help Mullis to either make the concept work experimentally or find any fundamental flaw that had been overlooked. Consequently, White formally asked Arnheim, in his new role as head of Human Genetics, to monitor what Mullis was doing, even though Mullis was officially still in the Chemistry department. Arnheim talked to Mullis (and Faloona) but was not convinced by the data and suggested they do some Southern blots in order to determine whether any of the multiple bands they were producing were what they claimed them to be. Arnheim remembers Mullis reacting with hostility to what he viewed as intrusive and unwarranted oversight of his work. In essence, he refused to cooperate.

For White, what was needed to determine whether PCR would work was independent analytical methods to establish whether the bands on the gels were actually the intended bands, to ascertain exactly what was being amplified. He didn't care if that required Southern blotting or sequencing or OR, or some other independent and scientifically demonstrable method. He wanted some reliable control. Because PCR's initial potential commercial value for Cetus lay in diagnostic tests, White felt strongly that the work had to be done on total genomic human DNA. Arnheim proposed going for the "gold standard" - a single-copy gene in a microgram of total human DNA. It was natural to choose the most developed experimental system available at Cetus, the beta- globin system.

At White's suggestion, Erlich and Arnheim formed "the PCR group," which was staffed by scientists and technicians in Erlich's lab. The work of the group was codirected on a daily basis by Arnheim and Erlich. In addition, the group met regularly on Friday afternoons. Arnheim and Erlich (and others) fondly remember these Friday meetings - at least for the first year. Excellent interchanges over the data took place, and the group members were open to constructive criticism, and goal-directed in seeking and proposing experimental strategies. For Arnheim, this was the first time in his scientific career that he had worked in a large group setting on an exciting, well-defined project; he found the brainstorming and feedback exhilarating. In his view, the experimental results generated from ideas that arose during the Friday afternoon meetings are testimony to the fact that PCR was made to work by means of a team effort. Mullis and Faloona had a standing invitation to these meetings, and usually one or both attended. Although Mullis was adamantly opposed to setting up the group, tension between Mullis and the others waxed and waned; during the first year (summer 1984 to late spring 1985), working relations were often collegial and productive.

By the end of October 1984, White agreed to assign another technician to the project. The choice was Stephen Scharf, who had unquestioned technical ability with proven experience in performing Southern blotting. An added plus was that Scharf was a friend of Mullis's. There were three "technical" people centrally involved during the crucial development period of PCR - Fred Faloona, Randy Saiki, and Stephen Scharf. Their careers and personal trajectories present a contrastive range. None of the three had a Ph.D. Faloona, unambiguously at the bottom of the scientific hierarchy, maintained good relations throughout this period with the other scientists and technicians performing PCR-related experiments. His low status and affable personality enabled him to move from lab to lab with materials, (often) inconclusive results, and ideas without provoking any controversy. It was as if people didn't feel obliged to take a stand on Faloona; he was simply the bearer of reagents, autoradiographs, and news from Mullis as well as a conduit for returning the same. The praise and blame went to Mullis. Randall Saiki had moved through the ranks at Cetus and had been given a great deal of autonomy in refining his work. He was respected throughout the company. Stephen Scharf occupied a niche between Faloona and Saiki. Everyone involved expressed a high opinion of his technical abilities, and there was (and continues to be) a general confidence in his lab performance.

By the summer of 1984 quite a lot was known about beta globin. Making the OR system work, however, continued to present problems: the signal-to-noise ratio remained poor. In the fall, Scharf was working full time on the beta globin-PCR project. In these experiments, cell lines of two types were used. They contained either the normal gene or the sickle-cell gene but not both. Scharf's experiments had specificity problems: PCR seemed to amplify both the normal allele and the mutation. There were several possible reasons for this confusing result. Perhaps the double signal had been caused by simple sloppiness in loading the gel. A precaution as simple as separating the samples by the width of the gel yielded better, although not perfect, results. Finding out why this and other inconsistencies remained demanded systematic (and time-consuming) exploration of experimental conditions. After a string of consistent results, Scharf remembers one experiment in which all the probes gave strong signals, including the empty control lane: "That was awful." [4] Though product was clearly being amplified and signals were being produced, they were simply too bountiful and were not always in the right places.

On 15 November 1984 Scharf was convinced that he really had "knock-out" experimental data. In his lab notebook he wrote, "IT WORKS!" The experiment clearly demonstrated that there had been amplification of the right-sized product. There was now no question that PCR was working and that specificity of amplification could be demonstrated. Seen from the lab bench - twenty months after Mullis's Eureka! experience - this experiment was the moment of triumph.

Erlich had no doubts that something was being amplified (although he doubted the pellet was all DNA), but the question was, precisely what was it? How specific was this operation?

These results, however encouraging, were still far from constituting convincing proof that the PCR group had worked out all the parameters required for a stable system. Nonetheless, demonstrating that the system could be made to work as it was supposed to, at least once, was a lift. Mullis, Faloona, and Scharf were now convinced that it was plausible and potentially powerful. Erlich and Arnheim didn't doubt that PCR was a powerful amplification technique, but they required more consistent data demonstrating specificity to be fully convinced of its viability. During the course of the following months, experimentation continued. Some experiments worked well and others didn't; some problems were solved and others remained. Although this lack of consistency was frustrating, there was nothing especially unusual about it - even in established systems, unexplainable problems crop up. However, as PCR-OR was not yet an established system, and as the goal was to produce a method reliable enough for commercial diagnostic tests, these remaining problems needed to be accounted for. The techs spent the winter sorting through the variables, seeking reliable standard procedures. For example, systematic work on varying the number of cycles of heating and cooling functioned as predicted (more cycles, more product). While there seemed to be some cross-reactivity between the amplified products, at least such cross-reactivity took place at a constant ratio. For months, the group maintained a rhythm of a completed experiment every three or four days (a day to do the amplification, a day to do the OR and run the gel, a day to develop it).

During the winter and spring of 1985, there was a consensus in the group that the results were encouraging, even exciting, but not yet quite definitive. At this juncture, Arnheim and Erlich decided to assign Randy Saiki - a consummate experimental practitioner - full time to the project. The investment paid off handsomely because Saiki eventually executed almost every experiment published in the first PCR publication. By this point the OR assay had been successfully worked out. A great deal of truly laborious and precise lab work was involved; nothing was automated and many of the frequent steps required constant monitoring over long periods of time. It was a tremendously labor-intensive undertaking, since it involved manually transferring tubes from a near-boiling water bath to one at 37° C, adding more polymerase, and repeating these steps over and over. Steve Scharf performed more of these repetitive tasks than anyone will ever have to do again, and has no regrets at seeing them taken over by a machine.

What had been achieved by spring 1985? The team members had demonstrated with reliable and quantifiable data that they could amplify genomic DNA hundreds of thousands of times, thus amplifying the target they were aiming at. The precision of the result was ascertained by quantitating the amount of a specific sequence radioactively and using the oligomer restriction assay. The data showed that not only had there been exponential amplification but that it was the beta-globin gene that had been amplified. Given the objectives of the experimental system, and given the fact that OR could be used to detect a specific product, the fact that other DNA was also being amplified did not pose an insurmountable problem. The system's specificity was good enough for diagnostic purposes.

Making PCR - developed by Paul Rabinow and colleagues, this new website will "chart the versatility of the techniques, concepts and experimental systems that make PCR a continually emergent invention." Site areas include What Is PCR?; Foundational Papers; and PCR Forum, a dialogue/debate on the past, present, and future of PCR. The site was developed as part of the Science and Technology in the Making (STIM) project, whose purpose is to provide interactive fora for contemporary science and technology; sponsored by the Alfred P. Sloan Foundation.

Eureka It Ain't - review of Making PCR in Fortune magazine, June 10, 1996

A Decade of PCR - an HMS Beagle Featured Essay - audio clips from A Decade of PCR, a 7-videotape set from Cold Spring Harbor Laboratory Press - In 1994, Cold Spring Harbor Lab sponsored a meeting to celebrate ten years of PCR. The HMS Beagle Essay features clips from talks given at the conference by James Watson, Kary Mullis, Andrew Ellington, and Stephen Fodor.

Heat Wave - The Thermal Cyclers of 1997 - an overview from The Scientist of many new or specialized thermal cyclers. Includes sections on Cycling Strategies; Handling High Throughput; In situ Cycling; and Block Swapping. Also provides PDF charts of features and specifications of cyclers, and information on PCR licenses.

PCR II - A Practical Approach Mutational Analysis: New Mutations - protocols for using PCR in searching for mutations and sequence polymorphisms. Protocols include PCR amplification of genomic DNA; Preparation of a polyacrylamide gel for SSCP; Standard SSCP; Consecutive SSCP; more.

PCR Applications Manual - from Boehringer Mannheim. The manual's table of contents is provided, along with PDF sample sections from chapters. Chapters include General Guidelines for PCR; Preparation of PCR Templates; Basic PCR: Protocols and Optimization Strategies; Use of Nonradioactive Labels in PCR; Reverse Transcription-PCR (RT-PCR); more.

Selected Events in the Chronology of PCR - a list of important dates, from the publication of the first paper on PCR, to the patenting of the technique, to its use in human forensics.