Scientists love abbreviations and acronyms. What better way to peacock one’s expertise than by mystifying others with your area’s secret codes? When I go to a lecture given by a scientist working in a field different from my own I often get lost among the acronyms. And we immunologists are as afflicted with this tendency as anyone.
IFN is the abbreviation for interferon—a family of natural proteins produced in an organism, usually in response to an infection. First identified in the late 1950s by the London-based British virologist Alick Isaacs and his Swiss colleague Jean Lindenmann, interferons play important roles in the defense against viruses and other infectious agents, and in the regulation of immune functions.
TNF—which stands for tumor necrosis factor—was identified by Lloyd Old and his colleagues at the Memorial Sloan Kettering Cancer Center in New York City in the mid-1970s as a protein produced in experimental animals injected with bacteria or bacterial components. The name derives from the observation that the factor appeared to cause the death of tumor tissues, or, put more scientifically, to produce “tumor necrosis.”
The work that led to the identification of TNF was an outgrowth of older studies showing that bacterial infections in humans or in laboratory animals would sometimes lead to a shrinking and, in very rare cases, even complete disappearance of malignant tumors. A similar shrinking of tumors was seen in tumor-bearing experimental animals injected with low doses of toxins derived from some bacteria. However, these earlier observations left unanswered the question of whether the shrinking of tumors was a direct result of the action of the bacterial toxins or whether it was perhaps mediated by something made in the body in response to the toxins. Lloyd Old’s study suggested that TNF, a protein produced mainly by white blood cells, was the mediator responsible for the regression of tumors. The implication was that TNF was part of the body’s defense system against tumors. The study also raised the prospect that—when isolated and properly defined—the TNF protein might one day become useful as a therapeutic agent in the fight against cancer.
I first met Lloyd—then a rising star in the emerging tumor immunology field—shortly after I had joined the NYU School of Medicine as an assistant professor in the Department of Microbiology in 1965. I was intrigued by TNF from the outset because it was a natural protein produced in the body that like interferon—a protein I had worked on since the late 1950s—appeared to have a role in the immune system’s array of natural defenses.
By the mid-1970s it was becoming apparent that there existed a large number of secreted proteins that were important in the fine-tuning of the body’s immune responses. In 1974 immunologists agreed that secreted proteins whose primary function is to regulate immune responses, such as interferon and TNF, be called cytokines. The first half of the term, originating from the Greek kýtos, meaning “cell,” was inspired by the fact that these proteins are both derived from cells and act on cells. The latter half of the word—“kine,” as in “kinetic”—implies that the function of these proteins is to move the immune system into action.
A quick computer search of the published biomedical literature on PubMed—a comprehensive database of the US National Library of Medicine—reveals more than six hundred thousand printed scientific publications where the word “cytokine” has been used. I commiserate with the medical and science students who have to learn about the hundreds of cytokines that have been discovered in recent decades. Cytokines tend to be identified by acronyms (like TNF, IFN, and many others) or by the abbreviation IL—for interleukin—followed by a serial number, starting with IL-1. At this point we are up to IL-38, but the actual number is much larger because many interleukins consist of several molecular variants.
Some years after the original publication by Lloyd Old, my interest in TNF became more tangible. In the early 1980s, we were in my NYU laboratory using cells isolated from human blood to generate a type of interferon called IFN-gamma. Soon we realized that some other unknown cytokines were produced together with IFN-gamma in the same test tubes.
The methods available for the identification of cytokines in those days were still cumbersome and relied on the use of indirect biological assays. With my colleagues Donna Stone-Wolff, Hanna Kelker, and others, we eventually established that the fluids harvested from cultures of white blood cells, which served as the source of IFN-gamma, also contained two other cytokines: one of these we identified as a protein known among immunologists as lymphotoxin, and the other—though at the time defying definitive identification—we suspected of being identical to TNF.
In the summer of 1982, while attending a meeting on cytokines held on the campus of Haverford College in Pennsylvania, I ran into Michael Wall, a biotechnology entrepreneur whom I had known since the late 1970s. I had first met Michael when he visited me at my NYU laboratory. He was a principal at a tissue culture supply company called Flow Laboratories—a company he had founded but later sold. Remaining with Flow Laboratories after its sale, Michael was looking for opportunities to expand into biotechnology. He came to see me because he heard about our work with interferon and was interested in establishing a collaboration with my laboratory.
Michael, an MIT-trained electrical engineer, impressed me with his grasp of the biomedical field. We also hit it off personally. He struck me not only as a man with a passion for entrepreneurship, but also as someone who cared deeply about science, in addition to being a lively and charismatic person with wide-ranging interests. We agreed to strive to establish a collaboration. I had several subsequent meetings with him and his professional colleagues, but before the collaboration could get off the ground Michael decided to leave Flow Laboratories in order to pursue other opportunities. One opportunity Michael seized was the establishment of Centocor—a company that would come to play an important role in my work and life.
The story of Centocor stands out in the history of the biotechnology industry. The company was built around one technology—monoclonal antibody production. The undisputed original creators of this technology are Georges Köhler and César Milstein, for which they would earn the Nobel Prize in Physiology or Medicine in 1984. Invention of the monoclonal antibody technology represented the realization of the German immunologist Paul Ehrlich’s dream of a “magic bullet,” a compound that could selectively target a harmful agent. Before this groundbreaking development, antibodies could be generated only in live animals, and the resulting “polyclonal” antibodies found in the blood would represent a mixture of thousands of molecules with different specificities and properties.
Even though the potential commercial relevance of the original work by Köhler and Milstein done at Cambridge University was apparent to many from the beginning, the British National Research Development Corporation failed to file a patent application for the technology with the justification that it was impossible to identify immediate commercial uses.
Others were not so shy. One person who recognized the potential of the monoclonal antibody technology was Hilary Koprowski, a colorful, prominent Polish-born virologist and longtime director of the Wistar Institute in Philadelphia. Koprowski, with some of his colleagues, adapted the Köhler and Milstein technology for the production of antibodies directed against viruses and tumors, filing patent applications in the process. Within a short time, in 1979, Koprowski joined forces with Michael Wall to create the company that became Centocor. The company established its headquarters near Wall’s home in Malvern, Pennsylvania.
When I ran into Michael again at Haverford College in the summer of 1982, the company had just moved into its first laboratories and offices. Characteristically, Michael was full of enthusiasm and optimism about his new venture. Would I want to come to visit his new place and meet some of his collaborators? I did. At the time Centocor was still a very small enterprise—I estimate that they had some fifty employees, including a handful of scientists. I remember meeting Hubert Schoemaker, a Dutchborn molecular-biologist-turned-entrepreneur with a PhD from MIT who had become Centocor’s president and CEO. Michael was the company’s chairman.
My first visit to Centocor was largely ceremonial, but I did speak to Michael and Hubert about their aspirations for the company. They envisioned the venture to be fully based on applications offered by the monoclonal antibody technology, with the aim of developing diagnostic products, which took less time to bring to market than therapeutics.
To keep their costs down, Hubert and Michael said, the company would not be depending on in-house research for the development of its products. Instead, Centocor would license new technologies from universities or other partners. In implementing this strategy, Centocor, along with the rest of the biotechnology industry, was helped by the passing of the Bayh-Dole Act in 1980, which allowed universities and research institutions to patent and commercialize US government–funded research without having to pay royalties to the government. The US biotechnology industry owes much of its worldwide success to this one legislative act.
“By the way,” Michael asked, “is there something your laboratory is doing that would be of interest to us?” Indeed there was. I told him about our work on IFN-gamma. With the help of Junming “Jimmy” Le, who was about to join my lab, we were planning to generate monoclonal antibodies to IFN-gamma. Although we were thinking of producing the antibodies mainly as a tool for our own research into the nature of IFN-gamma, the antibodies could also be used for the detection of IFNgamma and its quantitative measurement. Having such a test (referred to as an “assay”) for the identification and quantification of IFN-gamma in biological samples would have potential diagnostic applications; for example, the presence of IFN-gamma in some body fluids or tissues might indicate immunity or sensitization to a component of a microbial agent. Michael and Hubert seemed to like the idea and we agreed to stay in touch.
My visit to Malvern was followed by more detailed discussions about a joint project between my lab at NYU and Centocor. Soon the discussions progressed to an actual collaboration. As I had proposed, the collaboration initially centered on the development of an assay for IFN-gamma. The basis for the assay were two monoclonal antibodies specific for IFN-gamma, generated by Jimmy Le in my laboratory. By the early 1980s, the technology of monoclonal antibody production was well established and quite widely employed. Jimmy, then a recent arrival from Shanghai, had mastered the technology prior to joining my laboratory.
Once we established that the IFN-gamma assay was working in principle, we proceeded to discuss the terms of an agreement between NYU School of Medicine and Centocor. In those days, collaborative agreements between academic institutions and pharmaceutical or biotechnology companies were still relatively rare. Today NYU Medical Center has a large technology transfer office that routinely handles these types of negotiations. In 1983, there was no specialized office and the negotiations were conducted by the associate dean, Dr. David Scotch, a physician-turned-manager who bore most of the administrative burden of the entire medical center.
To provide legal advice during the negotiations, NYU engaged Peter Ludwig, a patent attorney based at a private law firm. (As happens to be the case with many of my professional contacts, Peter has become a lifelong friend.) Hubert Schoemaker—business-savvy, but eminently fair and delightful to deal with—represented Centocor in the negotiations.
NYU and I wished to accomplish two goals. First, we wanted to secure financial support for our research that would allow completion of the work we were planning to do jointly with Centocor and also provide my laboratory with some extra funds that we could ferret away for new, more adventurous projects. Second, we wanted to make sure that if a commercially successful product emerged from the project, NYU would be paid appropriate royalties on the sales of the product. (I should mention that NYU—like other universities in the US—had and continues to have rules in place for sharing a portion of the royalty payments with faculty members or other employees who contribute to the invention.)
The first goal was relatively easy to accomplish. Centocor agreed to provide funding for research in my laboratory, based on a simple research agreement signed by the two parties. A more complicated and detailed license agreement did not get signed until mid-1984, though we had already begun our scientific collaboration under the terms of the research agreement. One of the issues was the scope of the project. The initial plan was to cover only the use of monoclonal antibodies to IFN-gamma, but I was arguing for a broader agreement.
The agreement was being negotiated at a time when my colleagues and I had come to realize that several cytokines were cogenerated with IFN-gamma in the human white blood cell cultures we used for the preparation of IFN-gamma. We had just completed the publication of a paper showing that a factor termed lymphotoxin and another cytokine, likely related to TNF, were produced in the cultures along with IFN-gamma. I argued that to understand the biological functions of IFN-gamma it was important to pay attention to its interactions with these other agents. We were eager to try to generate monoclonal antibodies to some other cytokines because we believed they would be invaluable tools for the dissection of cytokine functions.
To justify the broadening of the project, in a proposal submitted to Centocor titled “Monoclonal Antibodies to Interferons and Cytokines,” I wrote:
The availability of MoAbs [abbreviation for “monoclonal antibodies”] to Lymphotoxin and Monocyte Cytotoxin would be useful for laboratory studies and would help to determine the relationships among Lymphotoxin, Monocyte Cytotoxin and TNF. In addition, since some of these molecules might play a role in autoimmune disorders as well as in natural resistance to malig-nancies, such MoAbs could become useful for diagnostic and other medical applications.
Arguing that “such MoAbs could become useful for diagnostic and other medical applications” without offering any specific examples of how the antibodies could be utilized was pretty vague. Had I submitted a similar proposal to an established pharmaceutical company, it would very likely have been dismissed with a chuckle.
Licensing agreements, especially for potential therapeutic drugs, are usually signed when there is clear supporting evidence for the utility of a specific product, such as extensive data from studies in experimental animals. With the exception of the antibodies to IFN-gamma, we not only did not have the products, we did not even know precisely what the products were going to be or what they might be used for.
It is a testament to Michael Wall and Hubert Schoemaker’s trust and risk-tolerance that they accepted my proposal and agreed to sign a licensing agreement between NYU and Centocor, stipulating that my laboratory would provide Centocor with monoclonal antibodies to several cytokines, including IFN-gamma, lymphotoxin, and TNF. In return, Centocor agreed to provide research support for my laboratory for three years (eventually Centocor ended up supporting our research for fifteen years), to exert their best effort to develop products, and to pay royalties to NYU on the sales of any products based on monoclonal antibodies originating in my laboratory.
In December 1984, I attended a workshop, the first of its kind, for a small group of devotees interested in TNF and related factors, organized by Lloyd Old and his colleagues at the Memorial Sloan Kettering Cancer Center in New York City. In those days, the group of people actively interested in this field was still small. In fact, the approximately twenty-five attendees from America, Europe, and Asia present at the workshop represented most of the world’s scientists working on TNF at the time. Only a few years later, similar meetings would be attended by hundreds of participants.
One important reason why interest in TNF was so modest in 1984 was that the factor was still poorly characterized and difficult to produce and identify. However, the New York workshop marked the beginning of a dramatic change when, seemingly out of nowhere, Bharat “Bart” Aggarwal, then a young protein chemist working at the biotechnology company Genentech, reported the purification of the human TNF protein and unveiled its complete amino acid sequence to the workshop participants. Each of the many thousands of proteins in the body, including the TNF protein, is made up of chains of tightly linked amino acids. Twenty different types of amino acids are known and their positioning in the chain—referred to as the amino acid sequence—determines the shape and function of each protein.
Less than a year after the 1984 workshop, Genentech scientists Aggarwal, David Goeddel, and their colleagues published not only the complete amino acid sequence of the TNF protein but also the sequence of the DNA encoding the TNF protein, along with details of the organization and chromosomal location of its gene. This information—in addition to having broad scientific interest—formed the basis for the production of human TNF protein by recombinant DNA technology, thus for the first time making pure TNF protein available for scientific studies.
To explain, recombinant DNA is created by combining the genetic sequence encoding a protein—such as the TNF protein—with some other DNA sequences that, upon insertion into a living cell (bacterium, yeast, or animal cell), will direct the synthesis of the desired protein. The protein can then be isolated from the producing cells or their environment, purified, and “bottled.” Much of the biotechnology industry has been built on advances in the production of proteins by recombinant DNA technology.
Progress in the molecular characterization of the TNF protein and of the corresponding DNA sequence transformed a small, insular field of investigation into an exciting discipline that almost overnight became accessible to rigorous scientific inquiry. Genentech’s scientists were generous in providing investigators at academic institutions with free samples of pure recombinant TNF protein (meaning the TNF protein produced by recombinant DNA technology) for experimental studies. Of course, before giving the protein away, patent attorneys representing Genentech would make sure that the rights for commercial applications of TNF were properly secured.
In 1985, when we received the first gift of recombinant human TNF from Genentech, virtually nothing was known about the spectrum of TNF’s biological actions and its molecular underpinnings. The two biological activities then known to be associated with TNF were those that had originally been identified by Lloyd Old’s group: the ability to shrink tumors in animals and to kill some tumor cells grown outside the body in test tubes.
There was one additional function that came to light immediately after the publication of TNF’s protein sequence. Anthony “Tony” Cerami’s lab at the Rockefeller University in New York City (located across the street from Lloyd Old’s lab) had independently been studying a factor dubbed “cachectin,” suspected of causing the wasting in animals infected with the parasitic agent Trypanosoma brucei, the cause of African sleeping sickness. In a fascinating turn of events, when the cachectin protein had been isolated and sequenced, it became clear that it was identical to TNF.
Thus two laboratories, located a few hundred yards from one another, and for years focusing their work on separate projects, had come to realize that the same protein was the cause of the two seemingly unrelated phenomena they studied. Cerami’s evidence that cachectin/TNF can act as a mediator of disease symptoms presaged many subsequent demonstrations of TNF’s role in the genesis of disease.
Having received a supply of recombinant TNF from Genentech—the very first time we could lay our hands on pure TNF protein—we needed to decide what experiments we would use it for. We felt like kids in a candy store—what should we try first?
The project we had decided to embark on initially was aimed at solving the quandary of why TNF was selectively toxic for tumor cells while sparing normal cells. It was known that cytokines generally act on cells by binding to “receptors,” meaning that cytokine proteins contain “keys” that fit specific “keyholes” on the surface of responsive cells. Was the selectivity of TNF action on tumor cells perhaps caused by the fact that appropriate receptors existed only on tumor cells, but not on normal ones?
To answer this question a visiting scientist from Japan, Masafumi “Masa” Tsujimoto, compared the presence of TNF receptors on the surface of tumor cells that were known to be susceptible to killing by TNF with normal human FS-4 fibroblasts (cells derived from tiny foreskins removed from newborn baby boys by circumcision that we had for many years used as a source of interferon), in which TNF was not known to produce cell damage. Using radioactively labeled TNF, Masa found that specific cell surface TNF receptors were present on both tumor cells and FS-4 fibroblasts. This finding indicated that the selective killing of tumor cells by TNF could not be explained by a difference in the presence of receptors on tumor cells and normal cells.
Now that we had identified TNF receptors on normal fibroblasts, it would be logical to expect that TNF elicited some actions in these cells. But what were they? We did not have to look for long. After exposing cultures of FS-4 fibroblasts grown in test tubes to infinitesimally small amounts of TNF, we readily observed a change in the shape of the cells under an ordinary light microscope. In the presence of TNF, cells became elongated and they also grew faster, and as a result they became more “crowded” inside the test tubes. I remember how stunned I was by this observation, because, until then, I knew of no other natural protein that would cause such a striking change in the appearance of cells.
One of my graduate students, Vito Palombella, then took on the task of showing that TNF could indeed promote the growth of normal human fibroblasts—an unexpected finding at a time when TNF was thought to be a protein that caused selective killing of tumor cells.
More surprises followed. Masayoshi “Yoshi” Kohase, another visiting scientist from Japan, and Luiz Reis, a graduate student from Brazil, were instrumental in establishing that under some circumstances TNF inhibited virus replication in a manner somewhat similar to interferon. Today we know that TNF, along with interferon, is important in the defense against virus infections.
So surprising was the multitude and breadth of TNF actions that Jedd Wolchok, an MD-PhD student in the laboratory (whose last name would sometimes get confused with mine), proposed—only half-jokingly—that the abbreviation TNF should stand for “too numerous functions.” All of these findings, together with parallel findings made by colleagues in other laboratories, led to our present understanding of TNF as a cytokine with a broad range of activities affecting the immune system and other functions in the body.
Another project with surprising results was initiated by Tae Ho Lee, a graduate student from Korea. At the time it was already known that exposure of cells to TNF leads to gene activation resulting in the synthesis of a number of proteins that are not produced in the absence of TNF. Tae Ho, who received training in molecular biology while working for a Korean biotechnology company, agreed to attempt to identify yet unknown TNF-activated genes.
Today, such a project would be quite routine, as there are now established tools and automated equipment available for this type of work, and it is common to engage a service laboratory to carry out the relevant analyses. At the time, such tools did not yet exist, and Tae Ho had to employ laborious manual methods to identify, one by one, DNA sequences corresponding to cellular genes that are turned on by TNF. Tae Ho’s hard work has paid off: two sequences he identified represented previously unidentified genes. For a molecular biologist to find a new gene is, I imagine, a thrill akin to identifying a new plant or animal species, or a new planet.
We decided to focus our subsequent efforts on one of the two newly identified genes, which we termed TNF-stimulated gene 6 (or TSG-6) because it was the sixth consecutive genetic sequence found to be TNF-inducible in Tae Ho’s experiments. After establishing some fundamental properties of the TSG-6 gene and the protein encoded by it—referred to as TSG-6 protein—we tried hard to define the function of TSG-6 protein.
Before completing his PhD training, Tae Ho was joined in the TSG-6 project by Hans-Georg Wisniewski. Georg came to my lab as a postdoctoral fellow from East Germany just a few months before the fall of the Berlin Wall and collapse of the East German state. When Tae Ho returned to his native Korea, Georg took over the TSG-6 project. He has now spent twenty-five years analyzing the properties and functions of TSG-6 protein, a project he is still actively pursuing. Perhaps not unexpectedly, TSG-6 protein has turned out to be important in understanding innate immunity and inflammation—processes that are intrinsic to TNF actions. Unexpectedly, however, TSG-6 has also been found to be important in female fertility. We are still hoping that one day there will be practical medical applications stemming from our research on TSG-6.
One important reason why scientists, including our colleagues at Genentech, had so eagerly pursued the purification and characterization of TNF was the hope that the TNF protein—believed to selectively kill cancer cells while inflicting no harm on normal cells—might prove useful in the treatment of cancer. When pure recombinant TNF became available, several leading medical centers started to make preparations for the clinical evaluation of TNF’s possible worth in treating cancer.
Disappointingly, the very first clinical studies revealed that—even at doses too low to produce tumor regression—humans were exceedingly susceptible to TNF’s toxic effects that included a severe drop in blood pressure, blood clot formation, and adverse impacts on the heart muscle. Eventually, these findings put an end to plans for TNF’s use as a therapeutic agent. Genentech’s investment in the study of TNF brought the company prestige and visibility, but no marketable products. Fortunately, Genentech has since developed many successful therapeutic products to offset this—and some other—failures.
As one set of studies provided information about the toxicity of TNF given to patients by injection, other investigations showed that TNF produced within the body—for example in response to bacterial infection—played a role as a mediator of disease. The first condition in which the disease-producing role of TNF was clearly demonstrated was in the pernicious effect of a bacterial toxin in experimental animals. Bruce Beutler and Tony Cerami at Rockefeller University showed that animals injected with a lethal dose of the toxin could be protected and kept alive by injecting them with specific antibodies to TNF, thus showing that the generation of TNF in the animals’ bodies was responsible for the toxin’s deadly effect. Soon it became apparent that TNF was also acting as a mediator of disease in some forms of malaria and in a complication that can occur after bone marrow transplantation—graft-versus-host disease—in which the transplanted donor cells attack the recipient’s organs.
In the mid-1980s, when these findings were being made and reported, the news of TNF’s harmful effects came as a surprise to the scientific community. Until then, TNF—and cytokines in general—were considered to be essential for the regulation of immune responses and for boosting host defenses against infectious agents and malignant tumors. It was now becoming clear that, even though TNF plays a useful role in host defenses, when produced in excess amounts for extended periods of time, it would become harmful. As William Shakespeare noted in As You Like It, there can indeed be “too much of a good thing.”
The realization that excess production of cytokines can be harmful, even deadly, should not have come as a complete surprise. Much earlier, Ion Gresser, an American scientist working in Paris, had conclusively shown that interferon production can have deleterious effects in animals during some virus infections.
With the indication that TNF was too toxic to be used as a therapeutic agent in humans and the findings indicating TNF caused diseases in some cases, the plans for the possible clinical exploitation of TNF had to be thoroughly rewritten. Instead of considering the administration of artificially produced TNF to patients with cancer, there was a gradual shift toward the belief that it might be more productive to consider developing agents that block the harmful disease-producing actions of TNF generated inside the patient’s body. At this point I realized that the licensing agreement NYU had signed with Centocor could become more valuable than I had originally anticipated.