NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Committee on Technological Innovation in Medicine; Rosenberg N, Gelijns AC, Dawkins H, editors. Sources of Medical Technology: Universities and Industry. Washington (DC): National Academies Press (US); 1995.

Cover of Sources of Medical Technology: Universities and Industry

Sources of Medical Technology: Universities and Industry.

Show details

7Incentives and Focus in University and Industrial Research: The Case of Synthetic Insulin


Within biotechnological research, university departments and firms are often organized as complementary inputs into research activities. The degree of complementarity is somewhat surprising (Arora and Gambardella, 1989), given the competition that exists between university departments and between firms. University molecular biology labs (or biologists) view each other as competitors for scarce funding and academic prestige, while similarly sized firms compete through markets and the development of appropriable products or processes. Why, then, do biotechnology firms and universities regard each other as complements, and is this relationship the result of the particular historical sequence that marked the early days of biotechnology?

This question will be attacked through a case study of insulin research in the late 1970s, an intense period which utilized and expanded upon the emerging techniques of genetic engineering. Observed interdisciplinary and interinstitutional research links will be examined to shed light upon the heterogeneous set of goals and trade-offs facing different researchers, the relative success of researchers in meeting these goals, and the effect of the research outcomes on future patterns of university-industry interaction. To provide focus, the research that is examined is centered around the expression of human insulin in E. coli bacteria.1

Research teams at Harvard University, the University of California at San Francisco (UCSF), and the City of Hope National Medical Center all played key roles in the development of the techniques, substances, and concepts utilized. However, Genentech, a small start-up biotechnology firm with no income and limited resources, was the organization first able to synthesize and patent human insulin, resulting in a royalty agreement with Eli Lilly. This paper contends that Genentech's commercial success can be understood through contrasting and comparing its goals with those of university departments, such as Harvard or UCSF.

While many authors have characterized the insulin research as a "race"—where researchers are solely focused on appropriating the commercial benefits of gene expression2—a great deal of insight is gained by analyzing the heterogeneity of these groups, the different strategies they pursued, and the divergence of their goals. Further, the insulin research, which resulted in the first significant commercial application of recombinant deoxyribonucleic acid (rDNA), conditioned the future organization of university-industry interaction. Genentech's success, often cited as a catalyst to investment in biotechnology, had the additional consequence of providing a framework for complementary university-industry interaction.

The insulin research was effective in demonstrating the relative strengths and weaknesses of firms and universities. In a nutshell, firms face strong incentives to produce products or processes in a minimum-cost manner. To achieve this goal, small start-up firms attempt to focus their efforts on a small number of projects. If these projects are unsuccessful, the firm will go bankrupt. Conversely, successful projects result in financial benefits to the scientists and investors in the corporation. In contrast, the mission of a university department or researcher is not so neatly characterized. At the very least, senior university researchers attempt to achieve three distinct goals: the training of graduate students and postdoctoral fellows, the resolution and explication of discipline-specific research inquiries, and the production of information that will be appropriable by the lab and the university. These three goals compete for priority. In other words, university researchers face trade-offs between devoting resources towards a project with potential commercial applications and devoting those resources towards the training of students and the development of experiments that will answer general scientific questions. Commercial firms do not face these trade-offs. This straightforward difference sheds light on the particular pattern of research relationships observed within this case study and, more consequentially, helps explain the success or failure of these research relationships.

The paper is organized as follows. The first three sections summarize the research under study by examining (1) the context out of which recombinant DNA techniques emerged, (2) the cast of characters who made up the primary research teams working on the insulin project, and (3) the timetable of the actual research. The next two sections detail the interdisciplinary and interinstitutional links observed within the frame of the study. These sections provide the background for the analysis and conclusions of the study, which examine the incentives of institutions and individuals in establishing and maintaining particular relationships, the cluster of characteristics that are present in the most successful of these relationships, and the effect of the insulin research on future university-industry interaction.

The Emergence Of Genetic Engineering

The insulin research is best understood in the context of the development of molecular biology and related disciplines. In the 1970s, critical advances in the techniques, instrumentation, and theory of molecular biology expanded the power and scope of DNA research, revolutionizing the practice of molecular genetics. Molecular biology, previously a prestigious basic science, was fundamentally transformed along with a host of allied disciplines. New avenues of research were opened, some with potential commercial application.

The proposal of a double helix structure for DNA by Watson and Crick in 1953,3 the most public achievement of molecular biology, ensured the place of molecular biology as an elite science. Subsequent to Watson and Crick's work, a diverse group of researchers resolved empirical and theoretical questions concerning the function and structure of genetic material. Activation and repression of protein production, the intergenerational transmission of genetic instructions, and the relationship between the sequence of genetic information and the production of amino acids presented fundamental but manageable puzzles to researchers.

Before 1970, most of the important advances within molecular genetics were the result of studying prokaryotic (lower) organisms. Conversely, important questions within prokaryotic biology could be resolved through the use of molecular genetics. Particular research avenues within bacteriology, the study of phage,4 for example, utilized molecular genetics to advance understanding of bacteriology as well as more general genetic phenomena. While molecular biology stood at the heart of these advances, it can be argued that the rapid pace and surprising direction of discovery led to fundamental ambiguities concerning the scope of molecular biology in relation to other disciplines. Bacteriologists, organic chemists, geneticists, and classical biochemists5 all contributed to the growing body of tools, techniques, and theory. Paul Berg, one of the most innovative researchers within the field and a biochemist by training, argues that molecular biologists were united by similar beliefs about the proper level at which to understand cell behavior: "Someone who would have called themselves a molecular biologist … would want to understand the phenomena at the molecular level" (telephone interview with Paul Berg, Professor of Biochemistry, Stanford University, February 17, 1993). This approach necessitated asking questions of a more fundamental nature than those asked within much of the mainstream of classical biochemistry. While perhaps providing the advantage of analytical depth and rigor, pre-1970s molecular biology suffered from tools that were inadequate for seriously studying eukaryotic (higher) organisms. Thus, the whole of biochemistry, much of it focused on "the characterization of metabolic pathways … of the more numerous and immediately useful proteins" (Kenney, 1986, p. 12), was affected but not transformed by molecular genetics. This distinction held as long as understanding of the physiological events of eukaryotes could not be greatly enhanced through the use of molecular genetic approaches.

After 1970, critical advances in technique, instrumentation, and theory overcame many of the barriers that had slowed the adoption of molecular genetics. The most public and startling of these advances was the gene-splicing technique pioneered by Stanley Cohen and Herbert Boyer in 1973. Along with work by Jackson, Symons, and Berg, the potential to manipulate—to change—the genetic code and subsequent protein production of an organism became feasible through use of restriction enzymes6 developed by Boyer (Johnson, 1983). This technique "has allowed for the first time the analysis of individual eukaryotic genes as well as the study of the organization of genetic information in higher organisms," as well as "enabled modification of the genetic makeup of bacteria and unicellular eukaryotic organisms so as to render them capable of producing gene products encoded by the DNA of higher eukaryotes" (Cohen, 1982, p. 21). The Cohen-Boyer technique had a broad effect on biology and biochemistry. Not only did it aid in the resolution of long-standing questions, it opened up the possibility of asking fundamentally new questions. Not surprisingly, the fundamental novelty of the technique led to serious questions of ethics and safety. Perhaps a bit more surprising, heated debate among molecular biologists resulted in a self-imposed two-year moratorium on recombinant DNA research.7

While recombinant DNA experimentation was halted, complementary techniques for DNA sequencing, gene synthesis, and gene detection advanced considerably. DNA sequencing yields the order of the nucleotides (A, C, T, or G)8 along a strand of DNA. This information is sufficient to predict the amino acids and proteins that can be produced by an organism. DNA sequencing methods were advanced through refinement of gel electrophoresis.9 The most important advances were simultaneously pioneered by Walter Gilbert10 and Allan Maxam at Harvard University and A. R. Coulson and Frederick Sanger at Cambridge University. Their primary achievement was "the separation of DNA molecules that differ in length by only one nucleotide" (Kolata, 1976, p. 645). The sequencing techniques developed at Harvard and Cambridge were reported to "herald a new era in molecular biology—an era in which … long-standing problems of DNA structure, sequence, organization and function will become clear" (Kolata, 1976, p. 647). The Gilbert-Maxam technique effectively reduced sequencing time by an order of magnitude. Moreover, sequencing was (and is) one of the most time-consuming activities of the molecular biologist. It is a useful measure of the technique's utility to note that while Herbert Boyer and Stanley Cohen have not received the Nobel Prize for their gene-splicing technique, Gilbert and Sanger shared the 1980 Nobel Prize in Chemistry for the sequencing advances.

Artificial gene synthesis was also pioneered during the moratorium on rDNA research. Har Gobind Khorana, of the Massachusetts Institute of Technology, fabricated an artificial gene by stringing together nucleotides in the appropriate order.11 The synthetic technique, requiring "9 years of work by 24 postdoctoral fellows," was "hailed … as a major accomplishment in genetics" (Maugh, 1976). Khorana's work was considered an excellent piece of basic research, whereby "the goal is to learn more about the gene itself and how it is regulated" (Maugh, 1976). However, the commercial application of the technique was also noted: "One suggestion is to incorporate a gene for insulin in a bacteria such as E. coli so that the valuable protein could be harvested from bacterial fermentation instead of from animals" (Maugh, 1976). As will be seen, Genentech pursued exactly this strategy.

Innovations in gene splicing, DNA sequencing, and gene synthesis required advances in both technique and instrumentation.12 Further, molecular genetic theory was informed, indeed transformed, by the expanded range of manipulation and observation of genetic material. With the issuance of National Institutes of Health (NIH) regulations in June 1976, rDNA experimentation resumed. Research proceeded at an extraordinary pace. However, it is the direction of some of this research, rather than its pace, which is of primary concern here. An influential group of researchers perceived the opportunity to apply the new techniques. Labs exploring gene expression with rDNA technology perceived the possibility of expressing commercially useful proteins in inexpensive bacteria cultures as part of their broader research agendas. Thus, for certain researchers, the lifting of the moratorium marked the beginning of a quest to produce higher organism proteins, such as insulin, in lower organisms, most notably the bacteria E. coli.13

Research Coordinators

This case study focuses upon research whose goal was the expression of human insulin in bacteria. Research of this sort was organized by a relatively small number of molecular biology and biochemistry labs. In Stephen Hall's popular account of the "race" to synthesize human insulin,14 he divides the research activities into three groups. The groups' coordinators were Walter Gilbert, the team of William Rutter and Howard Goodman, and Herbert Boyer, respectively.15 Organizing analysis of the insulin research by lab chief provides sharp focus for identifying interdisciplinary and interinstitutional links. However, this focus comes at a cost. There is underemphasis on the influence and mobility of postdoctoral researchers and the degree of cooperation between the groups. However, providing the context and backgrounds of each of the principal labs will allow for the identification of crossover and cooperation later on.

Walter Gilbert's group worked primarily out of the Harvard Biology Labs. Gilbert, a physicist by training, was perhaps the most celebrated scientist to be involved in the insulin research. In the late 1960s, Gilbert completed seminal work on the lac repressor, providing important insights into the regulation of gene expression. Further, Gilbert was instrumental in the revolution in DNA sequencing in the mid-1970s. Gilbert maintained a large, active, and diverse lab at Harvard. Only a few of the research projects being pursued had direct bearing on the insulin research. Gilbert recalls that his lab was pursuing research on sequencing, genetic control mechanisms, and general methods for the expression of proteins in bacteria. He characterizes the insulin research as merely a specific application of these more general research themes (telephone interview on February 18, 1993, of Walter Gilbert, Professor, Department of Molecular Biology and Biochemistry, Harvard University, February 18, 1993).

The respective labs of Howard Goodman and William Rutter, both of UCSF, decided to collaborate on the insulin research. The collaboration was initiated because members of both labs were independently attempting to express rat insulin in E. coli. Rutter, chairman of biochemistry at UCSF during the 1970s, pursued the project as part of his lab's larger goal of exploring ''differentiation of the pancreas" (telephone interview on June 14, 1993, of William Rutter, Professor of Biochemistry, University of California, San Francisco, and President, Chiron Corporation, June 14, 1993). Much of the research within Rutter's lab at the time "related in one way or another" to the insulin research. Indeed, expression of insulin was a particular and useful application of the projects being pursued by the lab at the time.16 Howard Goodman, on the other hand, was a younger researcher with a smaller lab. Collaborative work with Herbert Boyer earlier in the decade had given Goodman's lab more experience with recombinant technology than Rutter's lab. Both labs were attempting to capitalize upon the new methods and technologies: "When it became evident that you could consider cloning … then it was obvious we should shift our program prominently, or even dominantly—and, as it turned out, almost solely—to the issues of isolating the genes as a prelude to finding out how they were expressed" (William Rutter, quoted in Hall, 1988, pp. 91-92).

Lastly, Herbert Boyer had organized a medium-size lab at USCF. Boyer's lab was one of the most advanced recombinant technology labs in the world. In particular, Boyer's lab specialized in the construction of plasmids, also known as vectors. Vectors are the circular DNA material that can be easily transferred in and out of cells. Restriction enzymes developed by Boyer's lab resulted in a method for "cutting and pasting" DNA strands onto plasmid DNA. This method and the resulting vectors that could be easily manipulated, both fortes of Boyer's lab, were crucial components of the new gene-splicing technology.

Shortly before the NIH regulations were announced, though, Boyer's potential research base expanded. Along with Bob Swanson, a venture capitalist, Boyer formed Genentech,17 the first biotechnology firm focused on recombinant DNA technologies.18 Genentech, after an initial $1,000 investment split evenly between Boyer and Swanson, was initially capitalized by Kleiner & Perkins, a San Francisco-based venture capital firm for which Swanson had previously worked. The initial investment by Kleiner & Perkins was $100,000. Boyer's corporate affiliation affected the research activities of Boyer's lab—Genentech's first research contract was directly with Boyer's lab. Boyer and Swanson soon expanded the set of labs that Genentech invested in. Within nine months, Genentech contracted with researchers at City of Hope National Medical Center outside of Los Angeles to explore synthesis of a human gene.

These sketches are intended to provide a backdrop for understanding the research strategies and outcomes of each group. With the addition of a descriptive history of the research activities of each group, the interdisciplinary and interinstitutional links that were formed during this period can be examined and interpreted.

A Short History Of The rDNA Insulin Research Projects

The expression of human insulin in bacteria was identified as a feasible research goal by the time that rDNA experimentation resumed in the summer of 1976. The most influential labs, equipped with a novel set of procedures, tools, and questions, attempted to carve out important but "doable" research projects. Projects were chosen that would allow for experimentation, refinement, and expansion of the developing technology as well as resolution of long-standing questions in molecular biology and biochemistry. Insulin, a relatively small gene with interesting properties, was a prime candidate for a major lab's research agenda.

Expression of insulin in bacteria could be achieved by a number of distinct methods, namely, complementary DNA (cDNA) cloning, shotgunning, or synthesis.19 The precise procedures, as well as the instrumentation, for each of these methods were still rudimentary. More importantly, each method revealed different types of scientific information. Complementary DNA cloning, by starting with messenger RNA (mRNA) as its source material, could provide far more insight than gene synthesis into questions of gene regulation, the interaction between DNA and mRNA, and the intergenerational transmission of genetic information. Artificial synthesis, by its very nature, merely mimicked human genetic information for the purpose of protein expression, rather than providing a vantage point for analysis of human genetic information. Researchers who were interested in exploring broader questions of biology, then, were biased in favor of cDNA methods. Not surprisingly, the Harvard and UCSF groups' insulin research revolved around cDNA methods, while Genentech, through contracts with City of Hope Medical Center, explored the possibilities of gene synthesis.

The first major milestone in the research was achieved by the Rutter-Goodman lab in early 1977. The goal of their research project was to insert the rat insulin gene into E. coli. The first task was to acquire a small amount of rat derived, purified pancreatic RNA.20 The rat pancreas RNA, laboriously distilled by a postdoctoral researcher in Rutter's lab, John Chirgwin, provided the UCSF researchers with the important intermediate genetic material that provides a transcription of the rat's genetic code. Alex Ullrich, a postdoctoral researcher in Goodman's lab, then "backwards-engineered" the mRNA into the original DNA strands utilizing reverse transcriptase.21 These DNA strands were then spliced into a plasmid utilizing the Cohen-Boyer technique. The vector was then inserted into E. coli. Finally, the plasmid DNA of the E. coli was sequenced, indicating that a portion of the E. coli colony did, indeed, possess the genetic material for rat insulin (Ullrich et al., 1977).

To put the experiment in perspective, the insertion of a eukaryotic gene into E. coli had been achieved as early as 1974 (Morrow et al., 1974), and the procedure for doing so would be commonplace by the early 1980s. On the other hand, Ullrich and Chirgwin achieved substantial innovation in terms of the sophistication and reliability of the cDNA cloning procedure. According to Hall (1988, p. 140), "The suite of techniques they put together while cloning insulin instantly became a how-to manual for molecular biologists all over the world." Further, the experiment represented the first time a medically useful gene was successfully inserted into bacteria. The experiment highlighted the potential of the new techniques while simultaneously focusing upon insulin as a studiable hormone.

Nearly six months after the experiment's publication in Science (Ullrich et al., 1977), it was disclosed that the UCSF researchers had broken NIH guidelines during their research (Wade, 1977). In particular, while the Science paper reported research that utilized the vector pMB9 in April 1977, the research team had previously attempted the experiment in January of the same year with another vector, pBR322.22 Both vectors, products of Herbert Boyer's lab, were required to be reviewed, approved, and certified by the Recombinant DNA Advisory Committee (RAC) prior to their use in experiments. While pBR322 had been approved by RAC, it had not yet been certified, the necessary condition for using the plasmid for the experiment. Whether or not Ullrich, the principal cloner in the experiment, recognized his use of pBR322 as a violation of NIH policy is unclear; contemporary discussion allowed for the possibility of an honest mistake (Wade, 1977). When the violation was brought to the attention of lab director William Rutter, the work was halted, though not officially disclosed to NIH.23 Two months later, when pMB9, another acceptable plasmid for the experiment, was certified, the experiment was repeated and was successful. On the basis of this later work, the group published their findings in Science, holding a press conference three weeks before publication. The Science article that disclosed the violation gives a negative view of the UCSF researchers: "Capitalism sticking its nose in the lab has tainted interpersonal relations—there are a number of people who feel rather strongly that there should be no commercialization of human insulin" (David Martin, quoted in Wade, 1977, p. 1342). Further, the article noted that "The UCSF team was in competition with a group at Harvard which was known to be working with a better source material" (Wade, 1977, p. 1342). The exposé highlights both competition between labs for scientific primacy and the potential commercial applications of the insulin research. The UCSF experiment was of interest to both the academic and commercial community. True, the UCSF researchers were still far from expressing human insulin in bacteria; but the experimental steps towards that goal were now far more clear. Conversely, understanding of gene regulation and expression was still developing, and the insertion experiment provided important tools for this basic research agenda.

While both the Goodman-Rutter collaborators and the Gilbert group experimented with insertion and expression of the rat insulin gene, Genentech contracted with researchers for the purpose of refining and expanding upon methods of gene synthesis and expression. Genentech contracted with Herbert Boyer's own gene cloning and plasmid construction lab at UCSF, as well as with Arthur Riggs and Keiichi Itakura of City of Hope National Medical Center just outside of Los Angeles. Upon the advice and requests of Riggs and Itakura, Genentech funded experiments whose goal was the expression of synthetic somatostatin, a simpler human hormone than insulin.24 While somatostatin was not perceived to have any direct commercial value, it was viewed by its Genentech funders as an acceptable first step toward the synthesis of the insulin gene. Itakura's goal in synthesizing somatostatin was to improve upon, refine, and expand the Khorana technique announced in early 1976. Itakura's technique reduced the time necessary for the synthesis procedure from years to weeks. While building upon the group's earlier research,25 the somatostatin project provided an opportunity to standardize procedures as well as build up a library of "codons," nucleotide triples that are translated into the intracellular production of an amino acid.26

By early 1977, Itakura, refining his method, was able to produce purified somatostatin DNA. To clone and express the gene, artificial somatostatin genes were sent to Boyer's UCSF lab. Herb Heyneker, a postdocoral researcher in the lab, spliced the gene into pBR322, the same vector as used in the rat insulin experiment. The vector was then inserted into E. coli. By August 1977, the team had expressed somatostatin in bacteria. The experiment marked the first instance of expression of a human protein in bacteria. For this alone, the somatostatin experiment represented a milestone in rDNA research. However, Itakura's technique and tools proved to be widely applicable to other problems within biochemistry.27

Once made practical by Itakura, genetic synthesis possessed an important advantage over cDNA cloning: cloning with synthetic material was not covered by RAC guidelines.28 Thus, in contrast to cDNA research, which often required expensive and time-consuming precautions, the Genentech researchers only had to obey minimal safety precautions. For research on nonhuman genetic material, this difference was mostly one of convenience. The rat insulin experiments, for example, were conducted in a lab that required close containment of air, changing one's clothes, and a closely monitored waste disposal process. Research on natural human genetic material, however, required extraordinary precautions. Military biowarfare labs were the only facilities with adequate biological and physical containment to pass RAC guidelines for experiments with human genetic material (P4 facilities). Access to these facilities was extremely rare and difficult to obtain. Complementary DNA cloning of human genetic material could be halted, in essence, because of the strict RAC guidelines. Genentech researchers, pursuing the synthetic approach, faced no such regulatory constraints.

Much of the scientific insight from cDNA cloning did not require the use of natural human genetic material, however. The Gilbert group at Harvard continued to pursue expression of rat insulin in bacteria. The Gilbert group's goal was to modify the insertion procedure of Ullrich (of the Rutter-Goodman group) to achieve expression of rat insulin. Argiris Efstratiadis, a postdoctoral researcher in the lab of Fotis Kafatos, attempted to produce the cDNA gene. Efstratiadis worked with RNA from pancreatic rat tumors developed by William Chick of the Harvard Medical School. Efstratiadis' goal was to purify the RNA extracted from the tumors, and "backwards-engineer" the rat insulin DNA strands. Forrest Fuller, of Gilbert's lab, attempted to develop a method of inserting the cDNA strand into bacteria in such a way that the bacteria would start manufacturing insulin.

The insulin research supervised by Gilbert was less successful than the Rutter-Goodman researchers for meeting the goal of successful insertion of the gene. Further, Fuller, a graduate student, worked unsuccessfully with globin29 as Efstratiadis could not yet provide the rat insulin gene. Perhaps more telling, Fuller was forced, at the request of Gilbert, to leave the Gilbert lab at the end of 1977, just six months after the UCSF insulin research was published in Science.30 Fuller's approach to gene expression was discarded, and Gilbert recruited Lydia Villa-Komaroff, a more experienced postdoctoral researcher in Fotis Kafatos' lab, to be responsible for cloning and expression in the insulin research. In addition, Stephanie Broome, another graduate student in Gilbert's lab, provided a detection technique that allowed the Harvard researchers to recognize even minute amounts of protein expression.

Within six months of formulating the new research plan,31 the Harvard group achieved expression of rat insulin in bacteria. Moreover, the bacteria secreted the rat insulin. The secretion of protein from bacteria was an unintended by-product of the experimental strategy, but it was not without value. More specifically, the expression of mammalian protein in bacteria by cDNA cloning represented an important scientific contribution, while the resulting secretion of protein provided clear-cut possibilities for commercial application. The paper reporting the results (Villa-Komaroff et al., 1978) is widely cited as a seminal work in gene expression, while Harvard's patent is a broad claim over techniques that involve protein secretion in bacteria.

Consequently, by the summer of 1978, two alternative methods for expressing human insulin in bacteria seemed feasible: cDNA cloning and chemical synthesis. Moreover, each research group designed a strategy that, if successful, would result in expression. Expression of human insulin in bacteria represented the first important commercial application of rDNA technology. Not surprisingly, then, the level of interest by the pharmaceutical and investment communities increased as these strategies were formulated. The Rutter-Goodman researchers, the Gilbert group, and the Genentech researchers funded human insulin experiments with private (corporate) funds. Genentech, by construction, utilized venture capital to fund research. Not surprisingly, the success of the somatostatin experiment had considerably increased the level of financing available to Genentech. For expression of human insulin, Genentech was able to establish its own laboratory in South San Francisco and hire its first full-time researchers, Dennis Kleid and David Goeddel. The Harvard researchers, in contrast, reached a funding agreement with Biogen, a new, internationally oriented biotechnology start-up firm. Gilbert, as a celebrated scientist, had been courted by Biogen's founders and had agreed to chair its scientific advisory board.32 Within weeks of the announcement of the rat insulin expression experiment, a funding agreement had been reached between Gilbert's lab and Biogen for research on expression of human insulin. Finally, in August 1978, the Rutter-Goodman labs finalized a complex funding arrangement with Eli Lilly. The Lilly research contract included research on insulin and human growth hormone, among other products. The UCSF researchers involved in the rat insulin insertion experiment, most notably Alex Ullrich, agreed with Lilly to pursue research on expression of human insulin.33

By focusing on the expression of human genetic material, those researchers utilizing cDNA cloning methods were required to satisfy far more stringent RAC guidelines than were necessary for the rat insulin experiments. As mentioned earlier, very few locations possessed sufficient safety precautions to satisfy the regulation, and most of the acceptable locations were military biowarfare labs. Indeed, both the Gilbert group and Alex Ullrich were forced to leave the United States in order to conduct experiments on human insulin. As discussed below, Lilly arranged for Ullrich to work in France, where regulation concerning containment was slightly more lax; Gilbert, utilizing Biogen connections, secured a month-long research stay at England's Porton Down Microbiology Research Labs, a military research center. Genentech researchers, in contrast, were able to operate their cloning operations out of a leased warehouse in South San Francisco.

The UCSF insulin effort was the one least focused on direct commercial application. During the summer of 1978, Ullrich continued to work with rat genetic material to achieve two aims: expression of rat insulin in bacteria through cDNA cloning and the isolation and sequencing of the chromosomal rat gene. The first goal reflected an interest in mastering and refining the techniques for expression and secretion demonstrated by the Harvard group earlier that year. The isolation of chromosomal insulin DNA, on the other hand, would result in the ability to read the entirety of the insulin gene, even those strands that are not transcribed into RNA.34 Ullrich's insulin research was aligned with Lilly's funding goals. Thus, shortly after the UCSF-Lilly funding agreement was finalized, Ullrich traveled to a Lilly facility in France where experiments with human genetic material could be conducted. Ullrich's research efforts in France were unsuccessful, in part because of accidental contamination of his experimental materials during travel. However, even if Ullrich had been successful in reaching his short-term research goals during the France trip, he would not have achieved expression of human insulin before the Genentech researchers.

During his time in France, Ullrich decided to disassociate himself from the Goodman lab. After the insertion experiment, the hostility and antagonism between Ullrich and his lab director, Howard Goodman, had steadily increased. The sources of the antagonism were numerous: disagreements over scientific credit, conflicts over patent recognition, the provision of adequate lab support, the insertion experiment's "plasmid" episode, and personal antipathy.35 Along with his lab colleague, Peter Seeburg, Ullrich accepted a long-standing offer of employment from Bob Swanson of Genentech. Genentech thus hired two of the most talented postdoctoral researchers of the Rutter-Goodman labs.

Gilbert's group, in contrast to the UCSF researchers, was intent on the expression of human insulin during the summer and early autumn of 1978. Four Harvard researchers, Gilbert, Efstratiadis, Villa-Komaroff, and Broome,36 secured a one-month research stay at Porton Down Microbiology Research Labs in England. The trip and research were funded by Biogen. The goal of the trip was to express human insulin in bacteria utilizing the same experimental strategy that had been successful with the rat gene. Like Ullrich, unfortunately, there was a serious contamination of the group's source material during their preparations for travel. The contamination was only discovered after substantial experimentation at Porton Down. Moreover, their research stay was not long enough to salvage the experiment. If the research at Porton Down had been successful, the Gilbert group may have had a chance of expressing insulin in bacteria roughly contemporaneously with the Genentech team. Instead, the Gilbert group, and Biogen, were unsuccessful in their goal of commercializing the insulin research of the late 1970s.

This is not to say that additional scientific research was not pursued by the Harvard lab on the insulin gene; insulin, as a "studiable" gene, remained an important focus of inquiry for molecular biologists. Nor did Gilbert, or Biogen, cease to attempt commercial exploitation of the insulin gene. In fact, in 1981, Biogen contracted with Novo Industry, the second largest insulin manufacturer in the world, to develop a process for cloning human insulin. While this research was not entirely successful, it did provide Biogen with income before its public offering. However, the lack of success at Porton Down, and Genentech's contemporaneous success in California, substantially reduced the financial benefits to Harvard and to Gilbert of the insulin research.

Genentech, on the other hand, achieved a large measure of corporate security with the expression of human insulin in the summer of 1978. A research contract, ensuring a steady stream of income, was negotiated with Eli Lilly after the successful achievement of expression of human insulin in bacteria. The research strategy utilized mirrored the somatostatin project, though additional players were brought onto the team, such as David Goeddel and Dennis Kleid, for the actual cloning. The experience with somatostatin also provided important solutions to problems facing the team: "When we started the insulin project, DNA synthesis was not the risky part. We were improving on … work on the somatostatin project" (telephone interview on February 18, 1993, of Keiichi Itakura, Department of Molecular Genetics, City of Hope National Medical Center, February 18, 1993).

Insulin, however, was a more complicated gene than somatostatin. Achieving expression required the development of a series of innovative techniques. The first difficulty involved the construction of the separate chains of DNA that together form insulin. Pancreatic production of insulin involves three chains, the A, B, and C chains, where only amino acids from the A and B chains are actually present in insulin. The C chain links the A and B chain and is "snipped" off as it links the two chains. Goeddel, working with Itakura's team at City of Hope, first had to assemble the A and B chains separately and insert them into appropriate plasmids. The B chain, however, was cleaved into two sections, to allow for easier manipulation and reduce risk of failure during synthesis or expression. The second difficulty was reconstitution, the process by which the A chain and B chain were to be linked together. The reconstitution procedure, regarded as somewhat doubtful in and of itself, required extensive purification of the insulin chains. The purification was achieved by the repeated use of HPLC (high-performance liquid chromatography), an efficient distilling method which had gained increased popularity during the 1970s. Reconstituted, the A and B chains formed insulin. The last hurdle overcome by the Genentech researchers was the requirement that the insulin produced be efficiently "harvested" for commercial use. This was not easy, as the method of inserting the insulin gene into bacteria required that the insulin expressed be bonded with another protein, beta-galactosidase (beta-gal). Arthur Riggs constructed a breakable methionine link between the beta-gal molecule and the insulin molecule which allowed for the separation of insulin and beta-gal. This innovation allowed the harvesting of pure insulin. While this technique was obviously refined during scale-up, Riggs' procedure significantly advanced the commercialization prospects of the Genentech effort. In contrast, the Gilbert group, even if they had been successful in the expression experiment, would have been much farther away from commercialization than the Genentech effort. Once these experimental barriers were overcome, a radioactive assay was utilized to detect the presence of human insulin within bacteria. On August 24, 1978, the Genentech team, working at City of Hope, successfully expressed human insulin in bacteria.

Obviously, the Genentech result did not mark the end of research on insulin, either in a scientific or in a commercial sense. Both the Rutter-Goodman labs, as well the Gilbert lab, continued to explore the scientific properties of insulin. One of the principal results of that research was the isolation and sequencing of the chromosomal human insulin gene in late 1979. Isolation of the rat insulin gene provided key insights into the interaction between introns, exons, and the modes of intergenerational transmissions of genetic information.

On the commercial side, Genentech and Eli Lilly signed a research and royalty agreement shortly after the Genentech expression experiment. Each contributed toward the scale-up program associated with the production of rDNA insulin. Genentech researchers spent much of 1979 increasing the insulin yields from E. coli. One of the most important Genentech innovations of this period was the introduction of the trp system. When insulin was first expressed, the experimental procedure required each of the insulin chains to be linked to beta-gal. Unfortunately, beta-gal was an extremely large molecule, and for each molecule of insulin produced, a molecule of beta-gal had to be produced. The large size of beta-gal decreased the potential yield of insulin from each bacterial cell. The trp system, in contrast, utilized a much smaller bacterial enzyme, trp E, which had the additional advantage of being easier to manipulate.

Once the Genentech yields were sufficiently high, Lilly began the construction of pilot plants in expectation of commercial production. Lilly also managed the complicated process of FDA approval. Extensive pharmacological experience with beef and pork insulin, however, decreased the regulatory burden. Humulin, Lilly's trade name for synthetic insulin, was first sold in 1983. To put perspective on the importance of the introduced method, Humulin sales now account for over 60 percent of the sales in the domestic U.S. insulin market ($300 million).

In sum, this case study provides an intimate view of the university-industry interface with respect to medical innovation. Humulin, while the most visible, is only one of the end products of the research recounted here. An array of techniques, refined tools, and scientific insights emerged as a result of this research and have been steadily built upon for both scientific and commercial exploitation.

The Role Of Interdisciplinary Research

The insulin research was not only interdisciplinary; the experiments reshaped the boundaries of existing disciplines. While genetic engineering is most often linked with molecular biology, the insulin research involved organic chemists, a diverse group of pancreatic researchers, and biochemists. Moreover, the insulin research, in conjunction with contemporaneous rDNA research, transformed the practice of biochemistry in general, aligning it more closely with the methodology and techniques of molecular biology.

Biochemistry and Molecular Biology

The links between biochemistry and molecular biology are the most salient place to begin analysis of interdisciplinary forces in this case study. The critical differences between biochemistry and molecular biology are actually a matter of debate and have certainly changed over time. As mentioned above, Paul Berg claims that molecular biology was an approach and philosophy toward bioresearch adopted by an increasing number of experimenters during the 1970s, and that the increase in this adoption was partially a function of the increasing availability of tools and technology (telephone interview with Paul Berg, Professor of Biochemistry, Stanford University, February 17, 1993). Gilbert, on the other hand, notes that while particular distinctions do not do justice to the set of skills possessed by a researcher in either field, biochemists might be more interested in activities such as the purification of proteins while molecular biologists would be more interested in the manipulation of DNA through the use of enzymes (telephone interview with Walter Gilbert, Professor, Department of Molecular Biology and Biochemistry, Harvard University, February 18, 1993).

Irrespective of definition, the insulin research sat clearly between the two disciplines. Moreover, the insulin research37 provided a framework for interaction between researchers. The alliance between the labs of Rutter and Goodman, discussed earlier, amply demonstrates the point. Rutter's lab was more specialized in biochemistry, while Goodman's lab contained a set of talented molecular biologists. Members of each lab provided critical inputs into the research process. For example, Ullrich's success in the insertion experiment was crucially dependent on the isolation and purification of mRNA insulin from rats. The purification process, which could not have been adequately achieved by any member of the Goodman lab, was beautifully handled by John Chirgwin, a member of Rutter's lab. These interactions led to the transmission of information between researchers from the two labs. Thus, for future research, the biochemists in Rutter's lab were able to leverage the new techniques of genetic engineering, while molecular biologists such as Ullrich became more acquainted with the techniques of purification. The application of molecular biology to insulin research resulted in an expansion of the skill base of both biochemists and molecular biologists.

Organic Chemistry and Molecular Biology

Organic chemistry played an important, though not clearly foreseeable, role in the insulin research. Keiichi Itakura, the principal organic chemist working for Genentech, claims (somewhat exaggeratedly) that ''99.9 percent of organic chemists work on other and different topics; 0.1 percent of organic chemists work on DNA synthesis" (telephone interview with Keiichi Itakura, Department of Molecular Genetics, City of Hope National Medical Center, February 18, 1993). In other words, the proper role for an organic chemist in genetic engineering research was not obvious before the insulin research. Itakura perceived the opportunity to make synthesis a common tool in genetic engineering. As a consequence, the impact of Itakura's research was felt most pointedly in molecular biology and biochemistry, rather than organic chemistry. This research focus for Itakura involved professional risk. If the synthetic approach had not been susceptible to improvement, Itakura's reputation as an organic chemist would have been marginal (as DNA chemistry was only a small part of organic chemistry) while his work would have failed as a useful application of organic chemistry to another discipline. As it was, Itakura's success ensured his standing within the organic chemistry community as well as the more general scientific community.

In contrast to biochemists who pursued the insulin research in order to acquire experience with the emerging tools of molecular biology, Itakura specialized in chemical synthesis and did not deal directly with gene splicing or expression. Itakura recalls that molecular biologists were responsible for "handling DNA molecules" (telephone interview, K. Itakura, 1993). This contrast highlights the direction of interaction. Where Itakura's chemistry served as an input into genetic engineering, molecular biologists and biochemists informed each other by expanding each others' knowledge and providing novel contexts for applications.

Interactions with the Wider Medical and Scientific Community

Researchers into rDNA insulin also interacted with the more general medical community (and pancreatic researchers in particular). Expression of insulin in bacteria required the knowledge and resources of individuals who were knowledgeable about insulin. Both the Harvard and UCSF researchers acquired this information through the formation of a research relationship with a researcher in their respective medical schools.

The Gilbert group teamed up with William Chick, a researcher at the Harvard Medical School. Chick's research on the inducement of rat insulin tumors yielded rich source material for the Harvard researchers. Chick's contribution was quite critical—Harvard's "better source material" is mentioned in the Science article that exposed the use of an uncertified vector in the UCSF insertion experiment (Wade, 1977).

Rutter and Goodman allied themselves with John Baxter, who was a member of both the biochemistry department and the medical clinic at UCSF. While Baxter's role within the insulin research reviewed in this case study was small, he collaborated with the Rutter-Goodman labs on human growth hormone. Growth hormone, like insulin, was being aggressively pursued by molecular biologists at this time (McKelvey, 1993). Moreover, Baxter's lab was included in the research agreement that was reached with Eli Lilly in August 1978. Baxter provided the most direct link from the Rutter-Goodman labs to the more general medical community.

Finally, the role of multidisciplinary, as opposed to interdisciplinary , researchers should be highlighted. While most components of the research involved collaboration between researchers from different fields, there is a notable example within the insulin research of a researcher crossing fields. Walter Gilbert's graduate school training was in physics, as was his first faculty appointment. His entree into molecular biology was the result of persistent encouragement by James Watson during the late 1950s. While the migration of physicists into other scientific fields has been examined elsewhere (Rosenberg, 1992), it is almost ironic that Gilbert, the ex-physicist, would enter a field, molecular biology, that itself would dramatically redefine the boundaries and priorities of a wide range of disciplines.

The Nature of Interdisciplinary Research

Defining disciplinary goals and demonstrating the interactions between researchers from different disciplines provides a pedagogically sound mapping of the terrain. However, the deeper insight is best stated by Walter Gilbert: "When one is working on the frontier, nearly anything that one puts one's fingerprint on will be interdisciplinary" (telephone interview with Walter Gilbert, Professor, Department of Molecular Biology and Biochemistry, Harvard University, February 18, 1993). Only a small number of researchers had mastered the emerging techniques of genetic engineering at the time of this case study. The general excitement within this small community and the feeling by researchers that they were breaking important new scientific ground with commercial application diminished the importance of disciplinary boundaries. Moreover, successful research resulted from the mastering of techniques and tools. Indeed, advances in molecular biology led to potential consequences in scientific sectors distant from molecular biology. The difficulty of acquiring the skills to mine this potential, along with the prestige to be had by mining it, resulted in a curious but consequential collection of collaborations and crossovers.

The Role Of Interinstitutional Research

Interinstitutional collaboration and linkage is ubiquitous in the insulin research programs. In particular, each group utilized private funding during their research as well as forming collaborations between researchers at different academic institutions. Additionally, Genentech's willingness to form these types of partnerships was markedly more pronounced than that of the Harvard or UCSF groups.

Academic Collaboration

The Gilbert and Rutter-Goodman groups seem to have shared a common model for collaboration and interinstitutional interaction. With respect to academic collaborations, each group attempted to limit the number of links formed during the research project. This is not to say that links were not formed. Instead, links were formed only after it became apparent that a particular set of characteristics were present in a proposed collaboration. William Rutter notes three of these characteristics: "complementing your own research objective," "believing you will get an answer faster," and introducing "specific expertise" (telephone interview with William Rutter, Professor of Biochemistry, University of California, San Francisco, and President, Chiron Corporation, June 14, 1993). For example, the UCSF group pursued, for a while, a collaboration with researchers at the University of Texas (most notably Peter Lomedico). The University of Texas researchers were supposed to assist with the acquisition of purified bovine insulin mRNA (telephone interview, W. Rutter, 1993). The UCSF research was stalled owing to the lack of source material for experiments and the link with the University of Texas was aimed at overcoming this fundamental obstacle in pursuing the research.

The Gilbert team's makeup also highlights the trade-offs associated with collaboration, albeit in a more subtle fashion. As discussed earlier, Gilbert first pursued the insulin research with Forrest Fuller, a graduate student in Gilbert's lab. While Fuller refined his cloning approach, the progress of the group's insulin research lagged. Gilbert, while supportive of Fuller's lack of experience during early stages of the research, eventually withdrew support for him. Argiris Efstratiadis then recruited Lydia Villa-Komaroff, a more seasoned postdoctoral cloner, to complete the project. This change in personnel clearly benefited the pace of the research, as the group achieved expression and secretion of rat insulin within six months. More significantly, however, the incident highlights the trade-off between the training of graduate students and the successful execution of important research. Gilbert preferred to align his own research goals with the training of his graduate students but, when this arrangement turned out to be unsuccessful, Gilbert collaborated with a more experienced specialist.

Genentech's willingness to cross institutional boundaries provides a striking contrast to the tactics of the other researchers. Indeed, the corporation was formed for exactly that purpose. Thus, the Genentech strategy relied on research at City of Hope, Boyer's UCSF lab, and its in-house lab consisting of Dennis Kleid and David Goeddel. While this increased willingness to cross institutional boundaries is not surprising, it is a distinctive feature of the successful Genentech strategy (and other small start-up firms). For example, the final tasks of the insulin research were divided between the City of Hope researchers and the Genentech researchers, Kleid and Goeddel. For a period of time, material was being shipped back and forth between northern and southern California. While this constituted a logical division of labor, Genentech's corporate funders believed the work could be speeded up by merging the two teams in one location. The two Genentech scientists were told, "You're going down to the City of Hope. Don't come back until it's done" (Bob Swanson, paraphrased by David Goeddel and quoted in Hall, 1988, p. 246). Indeed, the corporate form provides clear incentives to coordinate research among institutions with the goal of producing an appropriable product or process.

Financial Linkages

In addition to interinstitutional research collaborations, each of the research groups arranged for private funding of the insulin research. Both the Harvard and UCSF researchers arranged research funding agreements during the summer of 1978 (with Biogen and Eli Lilly, respectively). In contrast, the Genentech effort, by definition, was founded upon the infusion of private funding.

Once again, there are strong similarities between the Harvard and UCSF teams. Both pursued private funding relationships when their experimental strategy began to focus on human genetic material and the possibility of patentable and appropriable products and processes. In contrast, Genentech funders, whose sole goal was the development of appropriable products and processes, funded the somatostatin experiments, which had negligible economic value.

Further, one of the motivations for pursuing private funding by the university-based researchers was the possibility of acquiring access to facilities in which experiments with human genetic material was approved (P4 laboratories; see footnote 7). Indeed, as discussed above, the Gilbert group arranged their stay at Porton Down through Biogen connections, and Alex Ullrich completed a stay at Eli Lilly's French research facility. These connections were critical for these researchers, as there were very few sites at which experiments with human genetic material could be conducted. For the Harvard and UCSF researchers (and their funders), the RAC guidelines were burdensome and costly. By comparison, the Genentech strategy, which evaded RAC guidelines through the use of synthetic material, not only allowed the researchers greater freedom but reduced the cost of the research to the funders.

In sum, the types of interinstitutional links that were formed reveal sharp distinctions between the approaches of the Genentech researchers and the academic researchers. First, the willingness to form research collaborations was more prevalent among Genentech researchers. Secondly, the goals of Genentech researchers and their funders were more tightly aligned. In addition, strategies that aided in maintaining this close alignment of goals were spelled out early on (during the somatostatin experiments). As a consequence, the Genentech research strategy required few adjustments as the expression of human insulin became more feasible. In contrast, the Harvard and UCSF researchers needed to alter their funding arrangements as well as their base of operation as soon as they began experiments with human genetic material.

Analysis And Conclusions

This case study has focused upon rDNA insulin research in the late 1970s. In particular, the first three sections outlined the context and history of the insulin research. The next two sections examined the interdisciplinary and interinstitutional links which were formed during the experiments. These links were focused upon because they reveal sharp distinctions between the research strategies of the different research groups. In this concluding section, this heterogeneity is analyzed more systematically. The incentives to form these links are characterized, the cluster of characteristics that mark the research groups are identified, and the phenomena of interlab competition are addressed. Finally, the role of the insulin research in shaping future university-industry relationships is explored. This concluding section highlights our underlying query: Why do biotechnology firms and universities regard each other as complements, and is this relationship the result of the particular historical sequence that marked the early days of biotechnology?

The Existence of Heterogeneous Incentives

The three teams examined chose substantively different strategies for pursuing their research goals. More precisely, there are more similarities between the Gilbert and Rutter-Goodman groups than between these groups and the Genentech researchers. This dichotomy is apparent in the research strategies that were pursued, as well as in the interdisciplinary and interinstitutional links that were formed. One can understand these differences as the consequences of variation in the incentives faced by the different researchers. Simply put, Genentech faced very different incentives than either the Gilbert or Rutter-Goodman teams. Moreover, the two academically based teams faced quite similar incentive structures.

The primary difference between the incentives of the Genentech group and the academic groups is the degree to which trade-offs existed between the development of an appropriable product or process and other goals of the lab and its researchers. Genentech's very existence depended upon the successful completion of the insulin research project. While the development of novel scientific knowledge was a benefit of the research, the researchers were consistently focused on the expression of insulin in bacteria for the purpose of acquiring a royalty contract with a large pharmaceutical manufacturer. As the researchers closed in on this goal, ever greater organizational resources (both personal and financial) were invested with the sole purpose of bringing the project to fruition. Genentech, by its very design, was relatively unfettered by concerns other than the expression of insulin.

The Gilbert and Rutter-Goodman groups, in contrast, performed the research in an academic environment. Consequently, the research organizers needed to balance a multitude of goals. As academic researchers, they had incentives to utilize the insulin research for financial gain, but they also had responsibilities and incentives to train graduate and postdoctoral students as well as place the insulin research within the context of their broader scientific agenda. For example, while the Harvard effort was pushed by Walter Gilbert in particular directions, Gilbert notes that "we were limited by the practicalities of what graduate students and postdocs were interested in" (telephone interview with Walter Gilbert, Professor, Department of Molecular Biology and Biochemistry, Harvard University, February 18, 1993). Additionally, the insulin research at the academic labs was conducted with a set of biological questions in mind.38 These restrictions did not bind Genentech. Instead, Swanson and Boyer were free to search for a lab where the insulin project could be carried out efficiently and without concern about tailoring the experimental strategy to reveal information of scientific importance.

The contrast between the incentives faced by Genentech and the academic researchers increased as the research neared completion. As mentioned above, Genentech had incentives to become consistently more focused on the sole goal of insulin expression as the research proceeded. The Harvard and UCSF groups had difficulties achieving a comfortable balance. At Harvard, Gilbert decided to discontinue support for his graduate student, Forrest Fuller, in order to maintain the pace of his insulin research. The UCSF group experienced even greater difficulties in balancing the competing objectives of different researchers. Howard Goodman, in particular, was widely credited with assembling a talented group of molecular biologists, but with failing to adequately support and reward his researchers during their research (Hall, 1988, p. 206). In fact, as mentioned earlier, shortly after Genentech successfully cloned insulin two of Goodman's best postdoctoral researchers accepted employment offers from Genentech.39 These difficulties at UCSF are attributable, at least in part, to the need by the lab to balance a complex set of competing objectives.

The Consequences of Heterogeneous Incentives

Because the research groups faced different incentives, their experimental strategies, as well as their willingness to form interdisciplinary and interinstitutional links, were quite different. In fact, by contrasting researchers with dissimilar incentives, clusters of strategic characteristics can be identified which highlight the manner in which researchers responded to the idiosyncratic set of incentives that they face.

The most salient features of the Genentech strategy were the pressure to achieve expression of insulin quickly (and publicly), the desire to avoid costly regulatory barriers, and the willingness to form an eclectic set of interdisciplinary and interinstitutional links. The decision to pursue chemical synthesis of the gene (rather than cDNA cloning) is an important manifestation of this strategy. Though important scientifically, chemical synthesis "was a chemical challenge, not a biological challenge" (telephone interview with William Rutter, Professor of Biochemistry, University of California, San Francisco, and President, Chiron Corporation, June 14, 1993). More subtly, the type of scientific information revealed by synthesis was qualitatively different (and of less interest to biologists) than that which might be revealed by cDNA cloning. In other words, the very technique chosen by Genentech reveals important insights into their motivations. Further, the decision to pursue chemical synthesis was made easier by Genentech's willingness to cross interdisciplinary and interinstitutional boundaries. Keiichi Itakura's input as an organic chemist increased as a result of Genentech's focus on expressing a particular hormone, insulin. In sum, Genentech's strategy consisted of a distinct cluster of strategic characteristics which, in concert, increased the team's ability to express insulin in bacteria for the purpose of obtaining a royalty contract with Eli Lilly.

In contrast, the academic researchers' strategies were characterized by the importance of performing experiments that provided novel scientific information or techniques. This focus resulted in the use of cDNA methods by the Harvard and UCSF researchers. While the cDNA method provided scientific focus, this experimental strategy forced the academic researchers to overcome significant regulatory barriers. However, these barriers were not insurmountable except when the research required the use of human genetic material. The choice of cDNA strategies highlights the trade-off between scientific and commercial goals: human genetic material was relatively unnecessary for the scientific agenda, but it was paramount for success in the commercial arena. In conjunction with the focus on nonhuman genetic material, the cDNA strategy leveraged the disciplinary focus of these labs: molecular biology and biochemistry. The synthetic process, while "a quick way to get a result" (telephone interview, W. Rutter, 1993), would have required the input of organic chemists such as Itakura or Khorana. The Gilbert and Rutter-Goodman teams chose the insulin project in order to explore the novel techniques and emerging processes of rDNA research, as this research related to their own discipline. These researchers were not necessarily interested in expressing insulin, per se. Instead, the isolation of the insulin gene and the expression and secretion of insulin in bacteria reflected the discipline-specific priorities of biochemists and molecular biologists. Insulin was studied, not because of its commercial importance, but because it was a "studiable gene" (telephone interview, W. Rutter, 1993). Thus, in contrast with the Genentech strategy, the cluster of characteristics that identify the academic researchers (cDNA cloning, the focus on nonhuman genetic material, and relative disciplinary insularity) reflect the fundamentally different incentives and goals of these researchers.

Interlab Competition

In the previous two subsections, we characterized the incentives and strategies of the three research teams. These characterizations can now be employed to understand the often-cited role of interlab competition in the progress of science and in commercialization. Stephen Hall, among others, views the insulin research as a classic science "race," where different researchers focusing on the same goal attempt to establish experimental precedent for the purpose of scientific prestige and commercial appropriability.40 In contrast, the evidence seems to suggest a somewhat more subtle and indirect form of competition between these labs. For most of the period studied, the academic researchers and the Genentech researchers were not directly in competition, although the academic researchers were in competition with each other. Only during the summer of 1978, when both the Harvard and UCSF lab chiefs pursued explicit linkages with commercial firms, did the existence of Genentech change the approach and strategies of the academic researchers. Indeed, the existence of divergent goals between the academic and commercial researchers precluded the possibility of a three-way race.

The crucial distinction here is between the competition that existed between the academic teams and the possibility of a race between academic and commercial researchers for the purpose of financial gain. The scientific content of the insulin research dictated that the research would contain competition. In particular, the set of important experimental questions was well-defined and agreed upon by the biochemistry community. For example, successful insertion of the rat insulin gene (performed by the UCSF researchers) was an agreed-upon milestone in recombinant research. Primacy in the achievement of insertion ensured scientific prestige. This type of competition is not necessarily wasteful or unproductive, however. Walter Gilbert remarks that "it is hard to do science in isolation … the fact that other groups are working on a project validates your own research plan" (telephone interview with William Gilbert, Professor, Department of Molecular Biology and Biochemistry, Harvard University, February 18, 1993). Moreover, the existence of clear research goals allows lab chiefs to devote their efforts to successful experimental design rather than the development of research questions. Further, there is strong evidence of this type of competition between the UCSF and Harvard lab groups. William Rutter, in fact, distinguishes between the achievements of Genentech, which he describes as a "chemical challenge, not a biological challenge," and his lab's competition with the Harvard group: "There wasn't a race [for commercial appropriability] … but it certainly was a race between Wally Gilbert and myself … we were engaged in an intense competition to get that gene" (telephone interview with William Rutter, Professor of Biochemistry, University of California, San Francisco, and President, Chiron Corporation, June 14, 1993). Members of the Harvard group, including Gilbert and Efstratiadis, agree (telephone interview, W. Gilbert, 1993), noting the fundamental similarities in approach between the two academic groups. The chemical synthesis approach pursued by Genentech did not noticeably heighten the level of competition at the labs of the two academic teams.

Bob Swanson at Genentech, in contrast, exploited the competitive nature of his researchers to focus organizational resources on the successful expression of human insulin in bacteria. "Definitely the name Wally Gilbert was in Swanson's mouth all the time. That we had to beat him. But I think Swanson used that as a management tool to keep pressure on Goeddel, knowing that Goeddel was so competitive" (Roberto Crea, quoted in Hall, 1988, p. 219). Ironically, by the time that Gilbert started to seriously explore the possibility of expression of human material for commercialization, Genentech had nearly completed its mission. Moreover, the Harvard group was at a tremendous disadvantage from the viewpoint of commercialization. The Gilbert team had to conduct experiments under extremely restrictive conditions (P4), their expression system was not designed for the promotion of high yields, and they did not possess a reliable method for "harvesting" the insulin once it was produced by bacteria. In contrast, the Genentech effort, having focused on commercialization from the outset, had systematically overcome these hurdles before the summer of 1978. In sum, to the extent that there was a race for commercialization between Genentech and the Gilbert team, the race was entirely uneven. Genentech's organizational focus provided it with clear advantages over Gilbert's group, which resulted in commercial success and financial appropriability.

Conclusions: Consequences of Insulin Research for Future University-Industry Interactions

The patterns highlighted by this case study must be interpreted cautiously. In particular, the case study approach makes it difficult to distinguish between historical events that are idiosyncratic to the particular situation and the presence of more general trends and behavioral phenomena. This distinction is important here as the insulin experiments and the development of synthetic insulin are landmark events in the history of both biotechnology and biochemistry. The insulin case is "special" as it occurred at a particularly unique juncture within the rDNA revolution and both the scientific and commercial goals were highly focused. Moreover, the insulin research, in conjunction with contemporary research activities,41 conditioned and guided the development of university-industry interaction within biotechnology. Because of the idiosyncratic nature of the insulin research, the focus of the analysis in this subsection will be upon the set of phenomena in the insulin research that seem to have played a role in the future development of biotechnology and university-industry organization and interaction.

Perhaps the most important lesson to be learned from this case study is that firms and university labs represent fundamentally different organizational structures. Most of the important differences between firms and universities can be understood by examining the differences in their goals. As examined above, university researchers face trade-offs in pursuing the commercialization of basic research results. These trade-offs, by and large, are not present for biotechnology firms (in this case, Genentech). This case study has highlighted various observable consequences of the heterogeneity in incentives. This heterogeneity highlights the following features of the university-industry frontier:


Firms will organize their scientific strategy and recruit their scientific personnel with the intention of focusing their resources on very specific goals. Genentech's insulin strategy was consistently focused on the goal of appropriability. While the Genentech team was composed of scientists with university research backgrounds, Swanson and Boyer were careful to choose researchers whose goals were aligned with Genentech's: the expression of insulin in bacteria. The evasion of cumbersome regulation, the pursuit of techniques that would provide high protein yield, and the premium on the rapidity of execution (regardless of cost) were all components of Genentech's focused strategy. The university researchers, on the other hand, attempted to exploit the commercial properties of their research in conjunction with the production of important biological information. This resulted in the use of cDNA methods, which required compliance with safety regulations, more difficult challenges in potential scale-up, and a longer time horizon for the purpose of commercial production.


University departments will often be ''second-place" finishers in competitive, time-dependent innovation. One of the most important consequences of heterogeneity is the increase in the probability that appropriable rents will, in fact, accrue to firms rather than university labs. While a university lab might eventually be able to focus enough resources to produce appropriable information, a firm will often be able to establish the important commercial result precedentially. Obviously, this is not always true (as innovation success has an important random component). But, given the existence of heterogeneous incentives, it will be true on average. Within the insulin research, Genentech had neared the completion of its initial research program by the time that the Harvard and UCSF researchers had started to seriously organize their commercial activities.


Despite the existence of a strong tradition of scientific purity and commercial disinterest, there existed little difficulty in persuading university basic researchers to attack applied commercial problems. However, the actual transition takes time. All of the researchers involved in the insulin experiments had been trained in and had operated in noncorporate research environments. Bob Swanson, however, was able to persuade a critical number of researchers (first, and most notably, Herbert Boyer) to attempt commercial problems. However, the development of the Genentech corporate research culture took time to cultivate. Indeed, the somatostatin experiment represents an important transition period for the Genentech researchers as they performed an experiment with little direct commercial value in the context of a corporate funding relationship. The same can be said for the other research teams examined here. Both the Harvard and UCSF researchers developed corporate affiliations by the end of the summer of 1978. While these corporate relationships conditioned future research strategies of these labs, they were of little import in the determination of the strategy for exploring insulin. In an important sense, the Biogen and Lilly funding relationships with the Gilbert and UCSF labs, respectively, needed time to develop before commercially relevant work could be achieved. This transition precluded the ability of these labs to achieve the level of appropriability for their insulin research that Genentech achieved.42


There exists a social trade-off between the level of knowledge spillovers and the degree of appropriability. University research is organized around the former, biotechnology firms around the latter . The desire for appropriability competes with the desire for openness in science and innovation, The long tradition of openness in university research diminishes the organizational capability of universities to financially capitalize on innovations occurring within their sphere. Firms such as Genentech, in contrast, are created explicitly to ensure the researcher's property rights over innovations.

These four features of the university-industry relationship result from the heterogeneity in incentives and are highlighted in this case study of insulin. Contemporary debate of the insulin research, in fact, explored some of these issues or at least noted the difference between the activities of firms and university labs. Indeed, the emerging biotechnology industry seemed to have grasped some of these features in their subsequent organization of knowledge production and transfer. For example, despite proposals to create university-owned corporations that would manage biotechnology research (e.g., at Harvard), no major university pursued this strategy. Instead, molecular biology and biochemistry departments have become important training sites for industrial researchers as well as serving as a nexus for information dissemination. Moreover, university labs have continued to push forward the boundaries of fundamental scientific knowledge (perhaps best exemplified by the Human Genome Project). Thus, the division of labor that emerged out of the early years of the rDNA revolution resulted, at least partially, from the comparative advantages of industrial and university researchers. Moreover, this organization of labor augmented the possibility for complementary interaction along the university-industry frontier.


This paper was prepared for a conference sponsored by the Institute of Medicine's Committee on Technological Innovation in Medicine. I would like to thank Cathy Fazio, Joshua Gans, Annetine Gelijns, Deval Leshkari and, most especially, Nathan Rosenberg, for their discussions and comments on this work. I would also like to thank the numerous participants in the insulin research who agreed to be interviewed and provided me with their time and thoughtfulness. All errors and omissions, of course, are my own.


  • Arora, A., and Gambardella, A. 1989. Complementarity and external linkages: The strategies of large firms in biotechnology. CEPR Discussion Paper 167. Stanford, Calif.: Stanford University.
  • Braithwaite, A., and Smith, F.J. 1985. Chromatographic Methods, 4th ed. New York: Chapman and Hall.
  • Cohen, S. 1982. Gene expression in heterospecific hosts. In: W. Whelan, editor; and S. Black, editor. , eds. From Genetic Experimentation to Biotechnology—The Critical Transition. New York: John Wiley and Sons.
  • Fudenberg, D., Gilbert, R., Stiglitz, J., and Tirole, J. 1983. Preemption, leapfrogging, and competition in patent races. European Economic Review 22:3–31.
  • Hall, S.S. 1988. Invisible Frontiers: The Race to Synthesize a Human Gene. London: Sidgwick & Jackson.
  • Johnson, J. S. 1983. Human insulin from recombinant DNA technology. Science 219(February 11):632–637. [PubMed: 6337396]
  • Kenney, M. 1986. Biotechnology: The University-Industry Complex. New Haven, Conn.: Yale University Press.
  • Kolata, G. B. 1976. DNA sequencing: A new era in molecular biology. Science 192(May 14):645–647. [PubMed: 17819992]
  • Krimsky, S. 1982. Genetic Alchemy: The Social History of the Recombinant DNA Controversy. Cambridge, Mass.: MIT Press.
  • Lewin, R. 1978. Profile of a genetic engineer. New Scientist 79(September 28):924–926.
  • Maugh, T.H. 1976. The artificial gene: It's synthesized and it works in cells. Science 194(October 1):44. [PubMed: 11643334]
  • McKelvey, M. 1993. Exploring University-Industry Relations Through the Case of rDNA Human Growth Hormone. Unpublished manuscript. Sweden: University of Linkoping.
  • Morrow, J., Cohen, S., Chang, A., et al. 1974. Replication and transcription of eukaryotic DNA in Escherichia coli . Proceedings of the National Academy of Sciences, USA 71(May):1743–1747. [PMC free article: PMC388315] [PubMed: 4600264]
  • Rosenberg, N. 1992. Scientific instrumentation and university research. Research Policy 21:381–390.
  • Sylvester, E., and Klotz, L. 1983. The Gene Age. New York: Scribner's.
  • Ullrich, A., Shine, J., Chirgwin, J., et al. 1977. Rat insulin genes: Construction of plasmids containing the coding sequences. Science 196(June 17):1313–1319. [PubMed: 325648]
  • Villa-Komaroff, L., Efstratiadis, A., Broome, S., et al. 1978. A bacterial clone synthesizing proinsulin. Proceedings of the National Academy of Sciences, USA 75(August):3727–3731. [PMC free article: PMC392859] [PubMed: 358198]
  • Wade, N. 1977. Recombinant DNA: NIH rules broken in insulin gene project. Science 197(September 30):1342. [PubMed: 11643399]
  • Watson, J. 1968. The Double Helix. New York: Atheneum Press.
  • Watson, J. 1976. Molecular Biology of the Gene, 3rd ed. Menlo Park, Calif.: W. A. Benjamin.



In contrast to many other innovation studies, there are detailed accounts of the early years of recombinant deoxyribonucleic acid engineering, in both academic and popular publications. In particular, there is a well-written popular account of the insulin research (Hall, 1988). Further, the insulin research was widely reported contemporarily, most notably in Science, Nature, and Cell.


Gene expression refers to the intracellular production of a particular sequence of amino acids, which are joined into proteins such as insulin. The sequence is determined by the deoxyribonucleic acid sequence of the gene.


The classic popular reference concerning the discovery of the structure of DNA is James Watson's The Double Helix (1968), where Watson describes the intensity of the academic competition in the pursuit of the genetic structure. See also the classic textbook Molecular Biology of the Gene (Watson, 1976).


Bacteriophage are an especially studiable type of virus which can attack bacteria. Many of the early advances in recombinant DNA occurred within the context of the study of bacteriophage.


The definition and boundaries of each of these disciplines have changed over time, but the textbook definition or the relative scope of each has not changed nearly as much as that of molecular biology.


Restriction enzymes are the critical material in "cutting" up genetic material for the purpose of extracting specific DNA strands. These strands are then spliced into plasmids for the purpose of insertion into bacteria.


An excellent review of this debate and details of the regulation has been performed by Krimsky (1982). While the rDNA controversy will not be reviewed here, the role of regulation in the insulin research is examined. The chief regulator of rDNA experimentation in the late 1970s was the Recombinant DNA Advisory Committee (RAC), which operated under the auspices of the National Institutes of Health. The most important component of regulation was the issuance of standards of safety care for different classes of experiments. P1, the most lax standard, required little adjustment from standard laboratory procedure. P4, the strictest standard, was reserved for experiments with dangerous or human genetic material and could only be performed in the equivalent of a military biowarfare lab.


Adenine, cytosine, thymine, and guanine, respectively.


Gel electrophoresis involves the separation and identification of molecules in gels. Separation is achieved through the application of an electric field upon ion molecules in an electrolyte solution. Differential rates of migration of the ion molecules allow for identification (Braithwaite and Smith, 1985, p. 70).


Gilbert is one of the lab chiefs within this case study of insulin. The insulin research was a particular application of the overall research direction of Gilbert's lab after the sequencing breakthrough.


The base elements of DNA can be obtained commercially rather inexpensively.


This is by no means an exhaustive review of the major advances that took place in the mid-1970s. Utilization of X-ray detection techniques, radioactive assays, and high-pressure liquid chromatography (HPLC) are just a few examples of the expanding array of tools and techniques available to the genetic researcher.


It should be noted that the expression of commercially useful proteins in bacteria was not the direction pursued by most researchers. Further, university labs pursuing insulin expression in bacteria perceived the project as a "particular application" of the broader research program of the lab.


The concept of a "race" is introduced here to reflect the tone and style of much of the published literature on the insulin research. The definition of a race, however, is ambiguous, and most analysis that examines investment races is imprecise as to what is driving the "overinvestment" in research (see, for example, Fudenberg et al., 1983).


Other labs, for example, one at the University of Toronto, pursued rDNA insulin research, but Harvard, UCSF, and Genentech represent the most heavily invested and successful participants in this research.


These projects included methods of cloning and expression, understanding of the enzymes involved in transcription, and the role of transfer ribonucleic acid in the regulation of intron/extron expression and recombination (telephone interview with Rutter, 1993).


The story of the founding of Genentech and the early interactions of Boyer and Swanson are well documented (e.g., Lewin, 1978).


Cetus Corporation, located across the San Francisco Bay, had been in existence for nearly five years when Genentech was formed. Cetus' focus, however, was broader than that of Genentech, and Ronald Cape, Cetus' president, predicted that genetically engineered insulin was at least five years away as late as 1977.


It may be instructive to note that cDNA cloning, the most elegant method for isolation, begins with messenger RNA (mRNA) as source material and, utilizing reverse transcriptase, "backs out" the original DNA sequence. Shotgunning, a more cumbersome method useful for obtaining simple genetic structures from lower organisms, requires cutting up an organism's entire genome, inserting plasmids into bacteria, and isolating those bacteria that ended up receiving plasmids that contain the sought-after DNA. Synthesis, as described above, entails the construction of a gene by stringing together nucleotides in the appropriate order. Sylvester and Klotz (1983) provide a readable introduction to the techniques of genetic engineering.


More precisely, the cDNA technique utilized required the isolation of mRNA, the intermediate genetic material that transports instructions from DNA to a ribosome for the purpose of protein production.


Reverse transcriptase is a type of enzyme discovered by David Baltimore that catalyzes the production of DNA strands in interaction mRNA.


Vectors, such as pMB9 or pBR322, were named by their "creators," in this case, Mary Betlach (MB) and Bolivar and Rodriguez (BR), respectively.


There was informal contact with NIH concerning the violation of RAC rules. However, there was no formal procedure with NIH, nor did the group volunteer the information in the original Science paper or in the press conference that announced the results of the research (Hall, 1988, pp. 134–144).


Riggs and Itakura, while negotiating with Boyer and Swanson, also applied to NIH for funding of the same project. The grant request was turned down on the basis of its lack of applicability (Hall, 1988, p. 83).


The group's earlier "non-Genentech-funded" collaboration involved synthesis and insertion of the lac operator, a small strand of genetic material that regulates expression. The research team, headed by Itakura and Riggs at City of Hope, included Boyer, Herbert Heynekker, John Goodman, and John Shine of UCSF, as well as researchers from the California Institute of Technology and the University of Ottawa. Ironically, the experiment was inspired by Walter Gilbert, who suggested a critical experimental technique. Thus, the lac operator experiment involved collaboration by researchers in all three of the distinct research teams involved in the insulin research.


Similar to Khorana's research, Itakura's utilized commercially available chemicals to construct DNA strands. The speed of the synthetic method greatly increased as the number of ready-made codons increased. To provide context, the expression of a protein (such as the insulin hormone) involves the production of a specific sequence of amino acids that "fold" to create the desired protein.


Itakura's results are widely cited in the explanation of the emerging techniques (Cohen, 1982, p. 25).


See footnote 7.


Globin is an essential protein in the construction of red blood cells and had been utilized by the Harvard researchers in earlier research.


Stephen Hall's account of the dismissal of Forrest Fuller highlights the trade-off that Gilbert faced between the successful execution of the insulin research and the successful training of Fuller as a graduate student (Hall, 1988, pp. 176-178).


Their research plan included a novel proposal for the placement of the gene along the plasmid pBR322, which allowed for easier identification of the bacteria colonies where the rat insulin would be expressed.


Gilbert would later leave his post at Harvard to become president of Biogen. Because of difficulties in bringing products to market, Gilbert was replaced by the board of Biogen, and he returned to Harvard in 1984 (Hall, 1988, pp. 315-316).


John Seeburg, another postdoctoral researcher in Goodman's lab, was the primary researcher on human growth hormone.


The important distinction here is between the identification of exons, genetic material that is expressed, and introns, genetic material that is spliced out during translation into RNA. While this distinction is not critical for the analysis here, the importance of the intron/exon distinction was a "hot" topic in molecular biology at the time and Ullrich's shotgunning approach to finding insulin provided a method for reading both the intron and exon regions of the insulin gene.


Hall provides detailed evidence of personal antagonisms within the UCSF lab (Hall, 1988, pp. 204-207, 277-283).


Efstratiadis and Villa-Komaroff, having been hired as assistant professors, were in the process of moving to Harvard Medical School and the University of Massachusetts Medical Center, respectively.


Along with other work (e.g., human growth hormone).


Rutter, Gilbert, and Efstratiadis all mention scientific goals as their motivations. For example, how is gene expression regulated? What were the roles of introns and exons? For William Rutter, a pancreatic researcher, one of the principal research queries was "what are the intracellular processes of pancreatic beta cells which allow the body to produce insulin?" Note that these questions are only peripherally related to the expression of human insulin in bacteria, particularly when the experiments are being guided by an eventual goal of commercial production.


Alex Ullrich and Peter Seeburg.


Hall's book is subtitled The Race to Synthesize a Human Gene.


Some of the more public contemporary work included the human growth hormone research, as well as the highly publicized debate at Harvard University concerning the university's potential exploitation of genetic engineering. Briefly, Harvard proposed that a corporation be set up which would manage the commercial exploitation of technology discovered by Harvard researchers. The proposal, which received extensive national coverage, was abandoned because of opposition from faculty and alumni.


Both the Harvard and UCSF labs received important patents based upon their insulin research. The level of income derived from these patents is small, however, compared to Genentech's exploitation of its insulin success to transform itself into a billion-dollar pharmaceutical company.

Copyright 1995 by the National Academy of Sciences. All rights reserved.
Bookshelf ID: NBK232052


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.3M)

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...