FAMILY BACKGROUND AND EDUCATION ( Parents / Childhood Interests and Activities / Interest in Science )

• Hughes: Dr. Cohen, I'd like to begin at the beginning, namely your birth on February 17, 1935, in Perth Amboy, New Jersey. Perhaps you could start by telling me something about your parents.
• Cohen: My father was Bernard Cohen and my mother was Ida (Stolz) Cohen. My father had a particularly important influence on my life as a scientist. He had always been interested in science and, in fact, at one point had started a post-high school education at the Pratt Institute of Technology in New Jersey, but for financial reasons couldn't continue that education. During World War II, he worked in a defense plant near Perth Amboy and after World War II, he established a small business. He had an innate curiosity about all kinds of things and was especially interested in understanding how things worked. He was employed as an electrician for part of his career, and during my childhood was involved in a number of entrepreneurial enterprises to supplement our income. I provided help in some of these. When fluorescent lighting was first commercialized, he assembled and sold fluorescent fixtures and I wired the transformers to the “starter” and transformer components in our basement. At one point, he sold electric fans, and I assembled the fans in the basement of our home. I guess I was around ten or twelve years old at the time. My mother and father were both graduates of Perth Amboy High School. My mother worked as a secretary during my early childhood. We were not well off financially, but somehow we always managed to do things that needed to be done and to buy things we needed, and to take family vacations. In her younger years, my mother was active in several community organizations. There was a social service organization called the Golden Chain that she was particularly involved with, and both of my parents had many, many friends. My father was viewed as Mr. Nice Guy, and as I got older, I realized that because he was seen in this way, sometimes people took advantage of him. My mother was ambitious for her family and for herself, and both of my parents were very hard working people throughout most of their lives.
• Hughes: That ethic was instilled in their children?
• Cohen: Yes, in both children. I have a sister, Wilma Probst, who is almost ten years younger than I. Since the age difference is so great, we grew up in very different environments. I was almost like a third parent to her. After the Second World War, my father started a small electrical supply business. Shortly after that, his mother died, leaving a retail yarn business in Perth Amboy to my father and his brother. The two brothers became partners in both the yarn business and the electrical business. The two businesses didn’t fit very well together, and neither did well, although both families made a living. At one point, I became interested in hydroponics. Do you know about hydroponics?
• Hughes: Very little.
• Cohen: It's the science of growing plants using nutrient solutions. Throughout most of my childhood, we lived half a block from the Raritan River and near a small park along the river. We lived in a fourplex house, and I spent a lot of time with my friends hanging out around the water. There was a sand beach not far down the river, and I built some wooden boxes, filled them up with sand from the beach, and started ordering chemicals to make solutions to grow plants hydroponically. The boxes were set up on our flat-roofed garage. I think my parents took a lot of kidding that their son was growing tomatoes on the garage roof. But they encouraged me anyway, and I enjoyed it. It was one of a number of science-related activities I was involved in.
• Hughes: And did the hydroponics work?
• Cohen: Many of the plants died, but I did get some tomatoes. They weren't the largest or juiciest tomatoes I've seen.
• Hughes: Were you in high school at this point?
• Cohen: No, that was, maybe, just before high school
• Hughes: Was it a disciplined household?
• Cohen: Not really. It took a lot to get my father angry, and to discipline his kids. When he became angry, he sometimes became really angry and on more than one occasion, took off his belt and gave me a whop on the backside with it. But, fundamentally he was a gentle person. My mother was much more emotional. Neither parent was a strict disciplinarian, although there certainly were times when they disciplined me. When I used language that my mother thought wasn’t appropriate, she would sometimes force a cake of soap against my teeth to “wash out” my mouth. I suppose that overall I wasn't much of a wayward kid, so there really wasn't a lot of need for discipline. At one point, a group of other boys and I went down to the river and were fooling around with some of the boats that were tied up there. We accidentally set ourselves adrift and were drifting out towards Raritan Bay, which empties into the Atlantic Ocean. It was in the late winter or early spring, and we knew that if we drifted out further, it would be a while before we would be found. So we jumped overboard—I guess that was in March, and the water was really cold— and we swam to the shore. I came home dripping wet. I don't remember the excuse I gave to my parents, but it was clear to them that I hadn't told them just what had gone on. However, they were willing to let it pass.
• Hughes: You essentially were raised as an only child?
• Cohen: I was until I was ten or so.
• Family, and Family Religion, Politics, and Ambitions
• Hughes: What about religion and politics?
• Cohen: My father came from a very religious family. In Jewish tradition there is a group called the Kohanim, who are descended from Moses’ brother Aaron, and who served as the priests of the Jerusalem temple. Although we were descendants of this line, my father was not an observant Jew. My grandfather, Samuel Cohen, was a strict disciplinarian, and when his father died, my father rejected a lot of the religion associated with his traditional upbringing. Although overall my parents, sister, and I weren't highly observant religiously, we went to the synagogue on the High Holy Days, Rosh Hashanah and Yom Kippur, and on some other occasions. I had a Bar Mitzvah at age 13. In fact, somewhere in that file [which I lent to you] is my bar mitzvah talk, which my mother kept a copy of until her death. I remember both of my grandparents on my father's side. My grandfather died when I was still quite young, around four or five. He worked as a butcher and had been raised in England. Some of his brothers had moved to South Africa, where they pursued medical careers. My grandfather and his wife, Bertha Samuels, who was my father's mother, had not divorced but they lived separately for many years, and she ran the small yarn business I've mentioned. My grandmother on my father's side lived until I was nine or ten, so I remember her well. On my mother's side, my grandmother, Sarah Wolf Stolz, had died in the influenza epidemic of 1918, and my mother was raised by my grandfather with the help of a neighboring family. A child of the neighboring family was my mother's lifelong friend, and I grew up thinking of her as my aunt and her children as my cousins. My mother had a brother, Michael Stolz, who lived in Pennsylvania and whom we saw occasionally. Politically, my parents were Democrats, and they were involved in small-town New Jersey politics in the sense that everyone knew just about everyone else in town and people with similar political views would band together to support their favorite candidates. My mother worked on election days to help supporters of the party get out the vote, but I don't think of my parents as political people.
• Hughes: You said your mother was ambitious. Was she also ambitious for you?
• Cohen: Oh yes.
• Hughes: Your parents wanted you to rise beyond their level?
• Cohen: From my mother it was obvious. I think that my father wanted that also, but he was more subtle about it. Their hopes were apparent to me, but at the same time they never were pushy. I did well academically throughout public school in Perth Amboy, and during that time became interested in writing. I won some writing contests and other awards, and my parents were always very pleased when this happened. You could see their pleasure and pride, but they weren't ambitiously aggressive about this. I can remember only one occasion of open parental ambition much later, after I established my laboratory here at Stanford, and my mother visited. I brought her to my lab and she looked at the door. There was a sign saying “Stan Cohen,” and she said that it should read “Dr. Stan Cohen.”
• Hughes: In the interview you did for MIT you said that it was in high school that your interest in science switched from the physical to the biological sciences.4
• Cohen: That's true. When I was ten, the first atom bomb was exploded. As an eighth grader, I had entered an essay contest and had written an essay on atomic energy, which won first prize. As a result of doing research for the essay, I became interested in atomic energy, and during the next year I read a lot about atoms. I thought at that time that I wanted to be a physicist. When I got to high school, I took the first year course in general science that was part of the normal curriculum, and during my second year took a biology course. There was a very stimulating high school biology teacher named Mrs. Florence Eggemann, who made biology exciting to me. I thought it would be more interesting to work with living things rather than with cyclotrons. So at that point my focus morphed to the biological sciences. To my high 4 Interview with Stanley Cohen by Rae Goodell, May 19, 1975, Stanford, California. Project on the Development of Recombinant DNA Research Guidelines, MIT Oral History Program. school career advisor, that meant being a physician. Later in high school, I decided to become a premed student and applied to college as a premed. It wasn't clear to me whether I wanted to practice medicine, but I was convinced that premedical curriculum would give me flexibility to move to other areas of biology. Whether I wanted to do basic science or take care of patients was something that I hadn’t determined, and as you'll see, this uncertainty existed for a long time afterward.
• Hughes: In high school, you were already thinking about biological research as a possible career?
• Cohen: Yes, I was. But I was almost equally interested in writing. I was Editor of the high school paper, and Associate Editor for the yearbook. I enjoyed writing and was repeatedly told that I wrote well. For a while I thought, well, maybe I want to do scientific writing.
• Hughes: Were you running with a group of friends that was planning for future careers?
• Cohen: Some were; some were not. We were a bunch of high school kids fooling around, going to movies on weekends, doing sports and just having fun in various other ways. I never excelled at sports, although I played baseball, basketball, and football, touch tag stuff, and was good at sprinting. I tried out for the high school track team, and did reasonably well. I became a member of the team but didn’t run fast enough to win races. I remember having to miss one particular match because I developed a wart on the bottom of my foot and had it removed. When I brought in a note from my doctor saying that I couldn't run for a couple of weeks, I remember our coach making a big thing out of it, joking loudly, “Cohen, have you been walking on toads?” And although I was tempted to tell him that I didn't think toads had anything to do with warts in humans, I decided to keep my mouth shut.

UNDERGRADUATE, RUTGERS UNIVERSITY, 1952-1956 ( Choosing Rutgers / Extracurricular Activities at Rutgers / Interest in Music )

• Hughes: Why did you choose Rutgers?
• Cohen: Well, for a number of reasons. The principal one was that they offered me the most scholarship support.
• Hughes: Which was the only way you could go to college?
• Cohen: Well, that was one reason, but there were also others. I was involved at that point with a synagogue-based youth organization [United Synagogue Youth]. I was supporting myself partly by leading multiple youth groups and getting paid for that, and these groups were in Northern New Jersey. More importantly, my father had developed diabetes and neuritis and his health was poor, so I wanted to remain in the area. Overall, I was happy at Rutgers, although not initially.
• Hughes: Were you living at home?
• Cohen: I lived in a dormitory at the college. The way I got involved with synagogue youth group activities is sort of interesting in retrospect. Although my family wasn’t steeped in religious practices, as a teenager I had joined a youth group at a local synagogue and, in my last year of high school, was sent to a convention aimed at forming a national organization of such groups. I ended up being elected national vice president and was asked by the organization to visit various cities around the country, essentially giving pep talks to other teenagers that were starting synagogue-based youth groups. I realized that I needed to learn more about Jewish history and tradition, and I traveled into New York from Rutgers weekly for a few months to take a course at the Jewish Theological seminary. For a while people thought I might become a rabbi.
• Hughes: Was that ever an idea that you had?
• Cohen: No, but some others thought that.
• Hughes: Why were you unhappy that first year?
• Cohen: Initially, I didn't feel a bond with many of the other students. But Rutgers is a fine university, and I found that by the end of the first year I was with a group of friends that had common interests, and in fact I still have close relationships with some of them.
• Hughes: How strong was the premed curriculum?
• Cohen: It was a very good premed curriculum, and I learned a lot there.
• Hughes: Did you know that before you applied?
• Cohen: I did, yes. But one of the things that bothered me is that many of the premeds were grinds. I worked hard at my studies as well, but I also liked to do other things. I found that wasn't the case with my premed classmates. But it sorted itself out. I ended up becoming heavily involved in extracurricular activities there as well, and was especially involved with the Rutgers debating team. Even though I ran fast enough for the high school track team, I didn't think my speed as a runner would make it on the college track team. I tried out for the debating team, and found that I was good at it. I enjoyed debating enormously. Our debating team had some excellent members during the time I was at Rutgers and we did extremely well in national tournaments. One of the things about debating is that it's important to anticipate the opposing arguments. A debater must be prepared to debate either the negative or positive side of an issue. Sometimes you don't know until you get to a tournament which side of the issue you're going to argue, and during the same day you could be arguing for or against the same proposition. Debating helps one see both sides of an issue more clearly. During the recombinant DNA controversy, when I had to make arguments to support my position, I think that my debating background made it easier to see the opposing point of view.
• Hughes: During your undergraduate years, you made quite a successful foray into music. Did your interest in music begin in college?
• Cohen: I had learned to play the piano as a child, although not very well. I also taught myself to play the ukulele and then subsequently picked up the guitar, which was more fun for a college student than the ukulele. I had written a few songs that friends thought sounded pretty good, and in college recorded two of them with a classmate, a guy named Bob Sileo, who had a very big and nice voice, as the vocalist, together with a group of other college musicians. We did this at the recording studio of a local radio station, and we actually had some vinyl records pressed, using a “label” I called Stanton Records. The recording quality wasn’t great, which I knew at the time, but I was foolish enough to continue with the project anyway. Maybe 25 records were sold to our friends and families. I think that I still have some of the remaining discs. Subsequently I decided to try to get one of my songs, which was called “Only You”, recorded by a professional vocalist and was able to find a music publisher who liked the song. A New York photographer named Jimmy Kriegsman, who was a leading photographer of pop music recording artists, also became interested in the song. With Kriegsman as a co-author, the song was published and recorded by Billy Eckstine, who was a very well-known vocalist at the time, and by two other groups. It did reasonably well.
• Hughes: Well, from those news clippings that you let me look at this morning, I understand that you used at least some of the royalties from that record to fund your schooling.
• Cohen: Yes, that's true. The song actually began to take off on the Hit Parade. However, the rise in popularity was aborted because of another song, initially called “Only You and You Alone,” which was recorded and released about the same time. The title of the other song was then shortened, and it also became “Only You.” Both songs were in the same style and it was confusing to radio disc jockeys. The other “Only You” was a better song, and even though the recordings of my song did reasonably well, the other “Only You” recorded by the Platters was a number one hit. Those days were a lot of fun, although I spent many hours walking the halls of the Brill Building trying to peddle my song. The building, I still remember the address, 1619 Broadway, housed most of the major music publishers of the time. George Levy from Lowell Music, which published my “Only You,” tried to interest me in staying in the music publishing business, but I didn't at all consider that.
• Hughes: Do you continue your interest in music?
• Cohen: When I was in medical school, I worked during one of the summers at a resort singing and playing the banjo. My daughter, Anne, is a solid musician who has perfect pitch. She points out to me that I sing a bit off key and I know that. However, I've learned that if you play the banjo loudly enough, people don't notice off-key singing.
• Hughes: I know that you worked under a pseudonym, which was Norman Stanton, when you were writing songs. Why did you choose to do it that way?
• Cohen: Well, not for any sound reason. Lots of songwriters had pseudonyms at the time and it was sort of fun to do. Norman is my middle name and Stanton is from Stan.

MEDICAL STUDENT, UNIVERSITY OF PENNSYLVANIA, 1956-1960 ( Choosing Penn / Research with Charles Breedis / Research in Peter Medawar's Laboratory, 1959 )

• Hughes: The next step is medical school. Why did you choose the University of Pennsylvania?
• Cohen: Well, I was influenced a lot by the feeling I got about a medical school when I went for an interview. I liked the feeling at the University of Pennsylvania.
• Hughes: What was there about it?
• Cohen: I liked the quality of the students I talked with during the interview day. I liked the way that the administrators and the interviewing faculty interacted with me. I was also attracted to some other medical schools, but there were some that didn’t appeal to me at all. I had an interview at one medical school where the person who interviewed me happened to be a psychiatrist. I knocked on the door of his office, and heard, “Come in.” I entered the room and he was standing with his back towards me, looking out the window. After standing there for a few moments, I said, “Dr. —,” whatever his name was. No response. Then after half a minute or so he spun around and said, “Well, sit down. What the hell are you waiting for?” I didn't particularly like this style of interviewing and quickly decided that I would not go there. Penn was a school that I liked and wanted to attend, and also they offered me the very substantial scholarship support that I needed. During medical school, I also received financial assistance from the Robert Wood Johnson Foundation, which is associated with the Johnson and Johnson Company in New Jersey. With the scholarship I received from Penn plus the funds from Robert Wood Johnson, I had most of my expenses paid for nearly all of medical school. I liked being in Philadelphia, although there were all kinds of bad jokes about the city. I had a few rocky times during my first year in medical school. On my first examination in biochemistry, I wrote answers to the essay questions up to the time limit, and at the end of the exam the instructor came around to collect the blue notebooks containing the student responses. I scribbled my name quickly on a notebook and handed it in. A few days later, the instructor came around to my lab bench and told me that the book I handed in was blank. I went running down to my locker to retrieve the lab coat I had been wearing and found the correct blue book crumpled up in the pocket. I brought up my lab coat with the crumpled notebook and handed the book to him apologetically. He said, “Well, I don't know whether we can accept it at this point.” I said, “Well, I understand, but I hope that you can.” A few days later the instructor told me jokingly with a deadpan face that the faculty had discussed the matter and concluded that if I were trying to cheat on the exam, this would be a very clever way of cheating and that I probably wasn't smart enough to concoct such a scheme, and so they accepted the exam book. The instructor later became a friend and mentor. This was a very supportive faculty. I liked the school and I liked the students. I enjoyed Penn.
• Hughes: Were you still considering research?
• Cohen: I considered it as a possibility, but didn’t get involved in a serious way in research until my second year of medical school, when I worked as a student researcher in the laboratory of Dr. Charles Breedis in the Department of Pathology. I had become interested in transplantation immunity and was intrigued by reports in the literature about transplanted tumors being rejected for immunological reasons.
• Hughes: How had you picked up on that subject?
• Cohen: In the second-year Pathology course lectures, I learned about immunity to skin grafts that had come from foreign sources. I also learned that foreign cells and tissues implanted into the cheek pouch of the Syrian hamster survived better than implants made elsewhere in the animal— especially when cortisone was given to the recipient—but conflicting results had been reported by different groups. I thought this observation was interesting, and I wanted to learn whether the hamster cheek pouch was really an immunologically privileged environment, and if so, why. I designed experiments that showed that normal adult skin grafts from rabbits could survive and grow in the hamster cheek pouch, while rabbit skin implanted elsewhere was promptly rejected, even in cortisone-treated animals. These experiments resulted in my first scientific publication.5 Then, I planned a simple experiment to try to understand the basis for the observation. I put rabbit skin grafts into cheek pouches and later grafts from the same rabbits onto the backs of the same hamsters that had received the cheek pouch implants. I saw that not only were the grafts on the backs of the animals rejected, but in some animals, the cheek pouch grafts—which had been growing well up until then—were also rejected. This suggested that once animals were sensitized by an orthotopic graft, the rejection mechanism acted throughout the animal. But the number of animals was small and the results weren't 5 Cohen, SN. Comparison of autologous, homologous and heterologous normal skin grafts in the hamster cheek pouch. Proceedings of the Society of Experimental Biology and Medicine. 1961; 106: 677-680. definitive enough to publish them.
• Hughes: Had you come to Breedis with this specific research project in mind?
• Cohen: I was interested in immunological rejection when I first approached him, but at that point hadn’t actually worked out the details of how I would investigate this. Breedis’ lab had been studying the Shope papilloma virus in rabbits, but didn’t work in the area of transplantation immunity. I proposed the specific experiments after he agreed to let me work in his lab.
• Hughes: How unusual was it for a second-year medical student to be doing laboratory research?
• Cohen: There were many students at Penn doing that, even before the medical scientist training programs that are now so prevalent. I continued that project during the summer between my second and third years of medical school. Some of my Penn classmates were also working on research projects and spending the summer in Philadelphia. We were given modest stipends, and together rented a small house near the medical school. During warm summer evenings, we sat on the front porch of the house drinking gin with tonic after a day of lab work, and discussed our experiments. Both the science and the social interaction were a lot of fun. That was the first serious scientific research that I did. Across the street from Penn Medical School were the Wistar Labs. Rupert Billingham, who worked at Wistar and was an expert in the field of transplantation immunity, became an additional source of advice. Billingham had trained with Peter Medawar, who had done pioneering work at University College in London on how animals react to implants from foreign sources, and Billingham himself had become well recognized as a leader in the field. I thought that Medawar's lab in London would be a great place to do further work on the cheek pouch project during my last summer as a medical student, and Breedis was supportive of this idea. I had never been outside of the U.S., except for a couple of childhood trips to Canada with my family, and wrote to Medawar in early 1959 asking whether he would accept me as a summer student. Although initially he said that he didn't have space, I persisted, and Billingham wrote a letter supporting my request. Medawar decided to take me on to work in his lab and I did that at the end of my third year at Penn medical school, extending the stay into the first part of my final year [May to September 1959]. Medawar's lab was very active scientifically at the time, and the work he had carried out earned him a Nobel Prize.
• Hughes: A year later in 1960. Did you have much contact with him?
• Cohen: Yes, he was very accessible, although he wasn't the person in the lab most directly involved in mentoring me. Peter Brent, an associate of Medawar's, was the primary scientist that supervised my project. While I was in Europe, I spent some time traveling around the British Isles and the continent. I supported myself by playing the banjo and singing off-key in cafes. It was a wonderful time.
• Hughes: And did the research go well?
• Cohen: It didn't go as well as I would have liked. I got results, but they were still not definitive enough for an additional publication. The questions I was trying to answer in Medawar’s lab were answered by Billingham and his co-workers a few years later, who established the mechanism underlying the failure of the cheek pouch grafts to initiate immunity. But my time in London was a great learning experience and I enjoyed it.

EARLY PROFESSIONAL CAREER ( Decision to Go to NIH / Clinical Associate, National Institute of Arthritis and Metabolic Diseases, 1962-1964 )

• Cohen: Early in my senior year in medical school, I applied for an internship. I had decided that I was more suited to an academic career involving medical research than to the clinical practice of medicine. I was quite interested in immunology because of my work with skin grafts, and I imagined myself going on and taking an internship and residency in internal medicine, and then doing immunologically related research and teaching in a university department of medicine. But another event affected my career very substantially. The Berlin Wall crisis occurred during my last year in medical school. Physicians were being drafted (the “Selective Service”) by the Army to care for troops that were stationed in Germany. I had decided that the clinical practice of medicine was not my career goal and was able to arrange to serve instead at the NIH [National Institutes of Health]. At Penn, I had encountered Colin MacLeod, who was one of three scientists (Oswald T. Avery, Maclyn McCarty, and MacLeod) who discovered 15 years earlier that DNA, rather than proteins, contain the genetic information of cells.6 It has been puzzling to many people why the group did not win a Nobel Prize for this enormously important discovery. Avery died a few years after that discovery and MacLeod died some years later. McCarty is still alive and in fact just…
• Hughes: Got the [Albert] Lasker [Award in Medical Research].
• Cohen: Yes, and I think it's been long overdue.
• Hughes: Do you think that because the paper was couched in conservative terms, there might have been some doubt as to whether they recognized the significance of their discovery?
• Cohen: No, I don't think so. The data are absolutely convincing and the conclusions were clearly stated. Conservatively stated, yes, but unequivocally. It's a classic paper and much has been written subsequently about the Nobel committee’s decision not to recognize its importance. In any case, MacLeod was a Research Professor at Penn and a friend of Joseph Bunim, who was head of the clinical branch of the Arthritis Institute [National Institute of Arthritis and Metabolic Diseases], and there was a lot of immunological research going on at the NIH related to arthritis. So, with a recommendation from MacLeod, who knew about my hamster cheek pouch work, I had the opportunity to do research at the NIH as a Public Health Service officer to satisfy my military obligation. I made arrangements to do that following an internship at Mount Sinai Hospital [1960-1961] and a year of residency in internal medicine at the University of Michigan [1961-1962]. My plan was to continue immunological work at the NIAMD. But a couple of months before I was scheduled to arrive, the person that I had been assigned to work with decided to temporarily leave the NIH. My appointment at the NIH was for a specific two-year period [1962-1964] as a Clinical Associate in the Arthritis and Rheumatism Branch, so I looked around at other labs. Clinical Associates at the NIH spent most of their time in the lab, but also took care of patients that were brought to the NIH Clinical Center for investigations of new therapeutic approaches. A number of young scientists who were Clinical Associates or Research Associates at the NIH during the time I worked there later accepted university faculty positions and made some very important scientific discoveries. I ended up in the laboratory of K. Lemone Yielding, who had done beautiful work with a more senior NIH scientist whom you may know of, Gordon Tomkins. [ Gordon M. Tomkins, M.D., Ph.D., [1926-1975] was chief of the Laboratory of Molecular Biology at the National Institute of Arthritis, Metabolism and Digestive Diseases from 1962 to 1969. In 1969 he became professor and vice chairman of the Department of Biochemistry and Biophysics, UCSF. ]
• Hughes: Oh yes.
• Cohen: Everyone knew that Gordon was one of the smartest people around at the NIH and I thought Lemone was pretty smart, too. Lemone and Gordon were working collaboratively on allosteric enzymes. These are enzymes that change their molecular conformation and substrate specificity. Lemone and Gordon were studying glutamic/alanine dehydrogenase. I wasn’t especially interested in allosteric enzymes but Lemone was willing to give me a place to work in his lab and to support work on a project that wasn't along the main lines of his research, but which I was eager to carry out. I will always be grateful for that. I had become interested in the mechanism of action of chloroquine, an anti-malarial drug that also was being used to treat arthritis. I ended up studying the interaction between chloroquine and DNA, the specificity of the interaction, what promoted it, what inhibited it. I found that chloroquine affects the functions of DNA and RNA polymerases, which were newly discovered enzymes at the time, as a result of its ability to bind to the DNA template used by these enzymes.
• Hughes: Was this your first taste of molecular biology?
• Cohen: Yes it was. In fact, “molecular biology” was a relatively new term then. I think I first became aware of the term when the first issue of Journal of Molecular Biology appeared in 1959, just a few years prior to my appointment to the NIH position. The NIH was an idyllic environment to work in. I was a novice, but could walk down the corridor and find people that I could readily get scientific advice from. Some of the major researchers of the period were there, people like Leon Heppel who helped to educate me about RNA biochemistry. In the next lab was Victor Ginsberg who was a polysaccharide chemist but was always ready to talk about any area of science and give advice when he could. Further on down the hall was a scientist named Art Weissbach, who was a bona fide DNA polymerase maven. He had a lot of experience working with DNA and I used to depend a lot on Art for advice and guidance. A Research Associate training in his lab, David Korn, subsequently came to Stanford as Chairman of Pathology and is currently Dean of the School of Medicine here. We first became friends at the NIH. It’s funny the kinds of things you remember: The first time that I isolated DNA, I used a protocol I had gotten from David. And, in order to help the DNA precipitate at one particular step, the protocol said to scratch the tube. I held the tube in one hand and scratched the outside of the tube with the other, but no precipitate formed, and I went to discuss this with David. He said, “Stan, you're supposed to scratch it on the inside with a glass rod.” That’s how inexperienced I was, but I subsequently got my DNA preparation. The research that I did in Lemone's lab was quite productive. It led to a paper in the Journal of Biological Chemistry [Cohen, SN, Yielding, KL. Spectrophotometric studies of the interaction of chloroquine with deoxyribonucleic acid. Journal of Biological Chemistry. 1965; 240: 3123-3131.], which was the premier biochemistry journal, and to a Proceedings of the National Academy of Sciences paper [Cohen, SN, Yielding, KL. Inhibition of DNA and RNA polymerase reactions by chloroquine. Proc Natl Acad Sci USA. 1965; 54: 521-527.], and then a couple of less important papers. But I realized I needed to learn more biochemistry. I had taken a biochemistry course as a medical student but didn’t have any serious training in the field. I had read a lot and had learned from attending seminars at the NIH, and certainly had picked up some practical knowledge about DNA through experiments I did in Lemone's lab and by talking with other scientists at the NIH about my experimental results.
• Hughes: Was there such a thing as a practical course in molecular biology?
• Cohen: Not at the time. But, it was clear that if I wanted to pursue work in this area, I would need to know much more biochemistry and genetics. My interests were taking me further and further from clinical medicine and yet my formal training had been as a physician. I had taken an internship and residency in internal medicine and enjoyed the challenge of making the right diagnosis and the satisfaction of helping sick people. I found that the satisfaction that I got from clinical medicine was complementary to the satisfaction I got out of research. In research, there is nothing better than the high that comes from discovering something new and important, and also nothing more depressing than times when things aren't going well. If I were to plot out satisfaction from a research career over time, the curve would resemble the profile of mountain peaks in the Pinnacles National Monument, which is located about 70 miles south of here. There were no highs in clinical medicine as satisfying to me as when things are exciting in the lab. But for me at least, clinical medicine provided a steady level of satisfaction. Through Art Weissbach, I was able to arrange to train as an American Cancer Society postdoctoral fellow [1965-1967] in the laboratory of Jerry Hurwitz who was a young biochemist focusing on RNA polymerase and other enzymes that interact functionally with DNA. His lab was at the Albert Einstein College of Medicine in New York. But notwithstanding my increasing interest in basic research, I planned to also use my training as a physician and decided to complete my clinical training by having a senior residency year in internal medicine at Duke University Hospital [1964-1965] before going to Jerry’s lab. Duke was a place that was quite flexible in wanting to support individualized career plans and I was able to arrange to spend two-thirds of my senior residency year doing clinical work, while spending the remainder of the year beginning postdoctoral training in Jerry’s lab. So I moved from the NIH to Duke in late June 1964, and then left for New York at the end of February 1965.
• Hughes: Did Duke have a policy that encouraged physicians to take basic science training?
• Cohen: Good question. Duke had a very strong focus on basic science training for its physician trainees. The chairman of Medicine was Eugene Stead, who was known as a strict disciplinarian who expected a lot from his students and medical housestaff. Many of the housestaff found him intimidating, but I felt that he was a really warm person, and I liked him very much. We worked for six days a week and had every other Sunday off. Some weeks we worked seven days a week. Even on the Sundays that we had off, Dr. Stead held “Sunday School,” which meant that we would all arrive at the hospital at 8 A.M. Dr. Stead—everyone called him “Dr. Stead” and we used to joke that we thought even his wife probably called him “Dr. Stead”—conducted “Sunday School,” and one of the residents or clinical fellows would be assigned to give a talk about a new scientific advance. We would be there from eight until eleven or so, and at the end of Sunday School if it was your day off, you'd have the rest of the day away from the hospital, probably to sleep. Dr. Stead expected directness. Of course he knew the realities of medical practice, but he simply did not tolerate excuses for sub-optimal performance. For example, if he asked about a lab test result for a patient, and if an intern said that he didn’t have a chance to do the test, Dr. Stead would look at the intern and in a soft, southern drawl, he'd say something like, “Well son, what you're telling me is that life is difficult. You don't have to tell me life is difficult; I already know that. What you're trying to tell me is that it's hard work being a good doctor.” He was tall—and sometimes it seemed as though he was about seven feet tall. The hospital beds at Duke had circular metal curtain supports around the top, and Dr. Stead would extend his arms upward and sometimes reach up to those railings. I enjoyed Dr. Stead, and learned a lot from him about life as well as about rigorous thinking in clinical medicine. A number of years later, when Herb Boyer and I received the City of Medicine Research Award [1988], [Gene Stead] also won that year’s Lifetime Achievement Award for his clinical accomplishments and contributions to medical education. I admired him enormously and it was a thrill for me to be getting an award along with my old Chairman of medicine.

POSTDOCTORAL RESEARCH FELLOW, ALBERT EINSTEIN COLLEGE OF MEDICINE, 1965-1967 ( Research on Lambda Phage Development / Developing an Interest in Antibiotic Resistance / Decision to Study Plasmids / Initial Postdoctoral Plans / Decision to Move to Stanford )

• Cohen: At the end of February 1965, I left Duke to begin postdoc training in Jerry Hurwitz's lab. One of the first people I encountered there was a young graduate student named Lucy Shapiro, who is now a colleague here at Stanford and is chairperson of the Department of Developmental Biology. Lucy has always been very outspoken and was quick to say that she had told Jerry that she felt that he should not have accepted me to his lab. She expected that because I am a physician, I would not actually be using the scientific training I would receive in his lab, and it would be wasted. There's not a whole lot one can say in response. But, soon after that rocky beginning, Lucy and I became good friends and have remained close friends over many years. I've teased her occasionally about that conversation. Most people in Jerry’s lab were working on enzymatic methylation of nucleic acids or on other biochemical projects. Possibly because of my limited background in biochemistry, Jerry assigned me to a partly genetic project that wasn't mainstream in his lab. He wanted me to try to learn something about the basis for transcription selectivity by RNA polymerase during development of bacteriophage lambda. Jerry was one of the discoverers of RNA polymerase. The project was related to Jerry’s interest in factors that affect the RNA polymerase interactions with DNA, but the genetic component was new for Jerry and no one else in the lab had been working on anything similar. It had been found earlier by several labs that fragments of lambda DNA produced by mechanical shearing could be physically separated by centrifugation in cesium salt gradients. The genes responsible for the phage functions expressed early in the life cycle mapped genetically to approximately one half of the lambda genome, the left half on a genetic map, whereas the genes involved in later functions mapped to the right side of the genome. Jerry was interested in learning whether there was differential transcription of these two sets of genes by purified RNA polymerase in vitro. And so I set out to mechanically shear lambda DNA and separate its two halves using the centrifugation approach that had been reported previously, and I tested the ability of purified Escherichia coli [E. coli] RNA polymerase to differentially transcribe the genes on the two lambda DNA fragments. The hypothesis from the genetic experiments done in vivo was that the bacterial RNA polymerase might be able to transcribe only the “early” genes and that proteins encoded by these genes would then facilitate transcription of the “late” genes. My experimental results showed that the early genes were, in fact, preferentially transcribed by the polymerase.
• Hughes: Did your previous experience in molecular biology at the NIH give you the tools that you needed for this research?
• Cohen: No. I had isolated DNA before and had done work with RNA polymerase and DNA polymerase at the NIH, but I had never purified any protein myself. In Jerry's lab I spent time in the cold room and learned to actually purify enzymes. I learned a lot, not only from Jerry, but also from the other postdocs and students that were in his lab. My experiments showed that transcription was initiated preferentially at promoters located on the left half of lambda DNA, and then set out to ask questions about strand specificity and directionality of transcription on lambda DNA. Results published a short while earlier by others showed that DNA strands could be physically separated by gradient centrifugation using a particular reagent [polyguanilic acid] that can bind preferentially to the two strands and enables their separation in cesium chloride density gradients. And so I set out to do that with lambda DNA. I learned additional DNA separation techniques and carried out experiments that produced a map of transcripts made in vitro on the bacteriophage DNA template. My results showed that transcription of some lambda genes is initiated on one DNA strand while some lambda genes are transcribed from the other strand. This work yielded publishable results that I was happy about, but similar experiments were being done concurrently by other groups of scientists and I was scooped on the publication of some of the findings. Since the genetic and biochemical techniques I was using were totally new to me and most were also new to Jerry's lab, I needed a lot of advice from people outside of the lab. Some of the advice came from Julius Marmur, who was a faculty member in the Department of Biochemistry at Albert Einstein, and from Carl Schildkraut in that department. Advice on lambda phage genetics came from Betty Burgee, who was at Cold Spring Harbor and had worked for many years with Al Hershey, who had done pioneering work on the exchange of genetics information by viruses. A former student of Jerry's named Anne Skalka, who was a close friend of Lucy Shapiro and also has become a good friend of mine, was working at Cold Spring Harbor, collaborating with Waclaw Szybalski at the University of Wisconsin in studies of lambda gene expression. It was an area of very active investigation.
• Hughes: You liked the activity?
• Cohen: Well, yes and no. I felt that the competition in the area of lambda biology was a bit too intense, but I certainly liked the excitement. I was invited to meetings to present my results and was invited to give seminars at universities. It was during one of these seminars that I first met Jim Watson. Mark Ptashne, who was working with lambda and lambda repressor, invited me to give a talk at Harvard, where Watson was a department chair, and Watson came to the seminar. During my entire talk, he sat in the first row and read the New York Times. Presenting my work in that setting as just a postdoc was a big event for me, and I was depressed that Watson seemed to find the work boring. But then at the end of the seminar he asked a number of insightful questions, so it was clear that he had been listening. I suppose that one of his minds was on the New York Times and another was focusing on my presentation. Because there was little expertise in viral genetics in Jerry’s lab, he arranged for me to take courses at Cold Spring Harbor during the summer of 1966. Each course offered total immersion in lectures and lab work for a few weeks. I spent essentially the entire summer at Cold Spring Harbor taking two courses sequentially, one in phage genetics and one in bacterial genetics. During both, there were visiting speakers. One of the speakers in the bacterial genetic course was Richard Novick, who had started an independent lab at the Public Health Research Institute in New York City after completing postdoctoral fellowship training with Rollin Hotchkiss at Rockefeller University. Richard was studying staphylococcal plasmids. At that time, there was general awareness that antibiotic resistance was becoming a serious problem. In fact, during my training at Penn, I had learned about a medical resident who died from antibiotic resistant staphylococcal pneumonia; the microbe that caused his death was resistant to every known antibiotic that was available at the time and his infection was not treatable. But there wasn’t much known about the genetic basis for resistance. Richard’s seminar made the connection between antibiotic resistance and plasmids. About the same time, two papers were published in the Journal of Molecular Biology on the molecular nature of antibiotic resistance plasmids: one by Stanley Falkow and his collaborators [Falkow, S, Citarella, RV, Wolhheiter, JA. The molecular nature of R-factors. J Mol Biol. 1966; 17 (1): 102-116.] and a second by Bob Rownd’s group [Rownd, R, Nakaya, R, Nakamura, A. Molecular nature of the drug-resistance factors of the Enterobactericeae. J Mol Biol. 1966; 17 (2): 376-393.]. These papers were published in succeeding issues of the journal. What interested me especially about those papers was that the plasmids that Falkow and Rownd groups were studying could be physically separated from chromosomal DNA in some species of bacteria that they had been transferred to, using differences in buoyant density in cesium chloride gradients. A few years before then it was discovered that resistance traits could be transferred between closely related bacteria. The work by Falkow and Rownd showed that multiple new bands of DNA were sometimes detectable in the recipient bacteria after transfer of resistance. What led to the occurrence of multiple bands wasn’t known, although it had been hypothesized that the bands were resistance gene components and transfer gene components of plasmids. Antibiotic resistance was an important medical problem, and I thought that some of the background and tools that I was using in my lambda studies might be applicable to studying plasmids. The approaches I had worked out to separate the halves of mechanically-sheared bacteriophage lambda DNA might be used to separate plasmids and plasmid DNA fragments from each other. My interest was in learning how resistance plasmids had evolved and how the resistance and transfer components of plasmids interacted functionally. Doing this would require identifying and mapping the genes that determine different plasmid functions.
• Hughes: Was anybody else taking that particular approach?
• Cohen: Well, I found out later that a couple of other groups were, but overall, plasmid biology was a very quiet area. Much of the molecular biology world was focused on phage. An important reason was that by using phage, it was possible to make identical copies—clones—of the progeny of a single DNA molecule: the phage genome. A cell infected by a phage makes thousands of replicas of the infecting virus during the normal viral life cycle. And so, it was possible to study the effects of a mutation in a single virus by producing a large population of viruses identical to the mutated one. But it wasn't possible to make clones of individual plasmids, and there weren’t many scientists interested in plasmids anyway. The fact that plasmid research was a sort of backwater of molecular biology was to me an attractive aspect of working on plasmids. I had been trained as a physician and had spent years learning clinical medicine, and I planned to look for a job in a Department of Medicine. I thought if I tried to compete with the hotshot labs working on phage, it would be difficult to do because I expected to also have clinical responsibilities. Antibiotic resistance was certainly a medically relevant area, and I thought that with only a few labs working on plasmids, and only a few papers being published every year, I could contribute something meaningful in an area that was very quiet—at least at that time.
• Hughes: In your M.I.T. Oral History, you say you got in touch with Falkow, which was an obvious thing to do. He was interested in plasmid epidemiology as much as he was in their molecular biology and I gather that was an unusual combination of interests [Interviews with Stanley Falkow by Charles Weiner, May 20, 1976 and February 26, 1977. MIT Oral History Program.].
• Cohen: Right. His overall interests were largely in understanding how bacteria cause disease. He had interests in molecular biology but he viewed himself principally as a microbiologist. And he told me that he was planning to end his molecular studies of plasmids. Falkow was encouraging and helpful to me in entering the field. We'll talk in a little while about how I went about a job search, but Jerry Hurwitz, my advisor, advised me not to move to a Department of Medicine. He thought it would be difficult to do serious research in a clinical department and tried his best to persuade me to take a job in a basic science department. Of course I considered his advice seriously, but decided in the end that I had invested so much of my life being trained in clinical medicine that I would try to combine clinical activities with basic research. I also had the concern that my experience in basic genetics and biochemistry was relatively limited, but I knew that I was a competent physician.
• Hughes: One could argue that you could have had a basic science appointment and then practiced medicine.
• Cohen: Not really. It just doesn't work that way in medical schools. Faculty in basic science departments usually do basic science research and teaching full time. If someone wants to also treat patients and teach clinical medicine, it usually means having a primary appointment in a Department of Medicine, Pediatrics, or another clinical department. But the academic environment at that time was very conducive towards doing basic research in clinical departments, and many clinical departments were trying to attract young physicians who had been trained scientifically. The hope was that these faculty members would provide a connection between clinical medicine and the basic sciences and would introduce more science into medical practice. So it turned out that my career goals were consistent with what many leaders in medical education were thinking.
• Hughes: Did the NIH support that model?
• Cohen: Definitely. And subsequently that model morphed to the medical scientist training programs implemented at many or most medical schools. Many physicians who received training in the basic sciences at the NIH did move to faculty appointments in clinical departments, but some have not. In 1967, about a year before I left Jerry’s lab, Falkow organized a symposium at Georgetown on antibiotic resistance plasmids. I attended the symposium and afterwards asked Stanley for some of the bacterial strains that I would need to begin my work. He was very generous and that was important in getting my plasmid experiments going. I suppose that I should say something more about the job hunt that brought me to Stanford. When I was at Duke as a senior resident in medicine, I got to know Jim Wyngaarden, who subsequently became Chair of the Department of Medicine at the University of Pennsylvania. Jim knew of my career plans and recruited me for an assistant professor position in Medicine at Penn. The prospect of returning to Penn was very attractive to me. As I've already said, I liked being a medical student there and I had good feelings about the place. Being a member of the Penn faculty would be sort of like “going home.” I accepted Jim’s offer, and while still a postdoctoral fellow in Jerry’s lab, traveled on weekends from New York to Philadelphia with my wife Joan to look for a house to live in. After several months, Joan and I found one that we were interested in buying in a suburb of Philadelphia. I phoned Jim at his office to let him know, and he said that he had just then been trying to reach me in New York tell me that he had decided to return to Duke. My appointment at Penn had been approved, I could still go there, but Jim said he hoped that I would join him in the Department of Medicine at Duke, which he would be leading. I hadn’t especially liked living in North Carolina during my residency at Duke, but Duke was an excellent place medically and scientifically, and the offer was attractive. I didn't think it made sense to move to a chairmanless department at Penn. I visited Durham to look at the lab that Jim was offering and to make a decision. There was a heavy smell of freshly harvested tobacco, and the heat and humidity of the Durham area were particularly oppressive at the time of my visit. When I opened the door to the air-conditioned car I was traveling in, my eyeglasses fogged up. I realized that I just wasn't happy about the prospect of returning to North Carolina. But deciding not to accept Jim’s offer was difficult, since I liked and respected him, and I didn't have another job. Jerry Hurwitz offered to let me stay on as a postdoc and generously arranged for an interim appointment as an Assistant Professor of Developmental Biology and Cancer at Albert Einstein. I held this position for most of a year [1967-1968], while I searched for a permanent job.
• Hughes: You didn't consider staying at Albert Einstein?
• Cohen: I did briefly, but felt that it wasn’t wise to continue with a career at the same institution as my mentor. Jerry was a very well known scientist, and I thought that it was important for me to be in a separate place and work independently. If I stayed at Einstein we would continue to publish together and my research program wouldn’t be viewed as being independent. I was ready to start my own laboratory. Jerry has a lot of friends in the field of biochemistry. He had been in the same department at Wash U. [Washington University in St. Louis] as Arthur Kornberg and he was a friend of Paul Berg. Paul was working on RNA polymerase, partly in competition with Jerry, but he and Paul interacted very amicably and on a relatively frequent basis. Jerry also knew Dale Kaiser and other former members of the Biochemistry Department at Wash U. that Kornberg had brought to Stanford. One day Dale was visiting Jerry’s lab, and Jerry told him that I was looking for a job. Dale and I knew each other because of my work with lambda. Dale was one of the leaders in the lambda field; he and I had been at several scientific meetings together and had talked pretty extensively about lambda biology. Dale suggested that I consider moving to Stanford, and I thought that was an interesting idea. Subsequently, Jerry had a discussion about this with Paul Berg, and the Stanford possibility progressed a little further.
• Hughes: Berg was chairman?
• Cohen: No, Kornberg was Chairman of Biochemistry. But, because of Paul’s interest in RNA polymerase, he also knew of my work. He offered to speak with Halstead Holman who was Chair of Medicine. Shortly afterwards, I received an invitation to give a seminar at Stanford, essentially a job seminar. This was an interview for a possible appointment in the Department of Medicine here.
• Hughes: Stanford was one of the places where the clinical departments were interested in a basic science orientation?
• Cohen: You bet. Holman was strongly focused on that notion. So I traveled to Stanford, and gave two seminars, one for the Department of Biochemistry in its library, and one for the Department of Medicine. The Department of Biochemistry seminar was very well attended, in fact the room was packed, and it was clear that a lot of faculty, students and postdocs were interested in the work that I had been doing with lambda. There were good questions and enthusiastic discussion, and I enjoyed that very much. It was apparent that I had passed this biochemistry test; I knew that Paul and his colleagues were going to support my appointment. I also gave a seminar in the Department of Medicine, in what I think was possibly the smallest lecture room at Stanford. There were only ten or so members of the Department of Medicine faculty that attended, and even though the room was small, it seemed empty. I think that the few people who were there had come because they had been asked, or felt obliged, to appear. Most of them were members of the Department of Medicine division that I was being interviewed for. I was disappointed that there were not more people in Medicine that had an interest in the work, since this was said to be a department that was strongly basic science oriented. Anyway, after my visit I was offered a Stanford position as Assistant Professor of Medicine. Hal Holman, who was recruiting me, knew that I had not been to the West Coast before coming for the initial interview. He proposed that I come out for a second visit and said that he would provide a rental car and several days of expenses for my wife and me to travel around Northern California and decide whether this was a place where we wanted to live. We drove to Yosemite and up the coast, and the beauty of California helped to attract us to come here.

JOINING THE STANFORD FACULTY ( Starting a Research Program and Trying to Become a Hematologist / Interactions with Faculty in Departments of Biochemistry and Medicine / Clinical versus Research Activities / Starting the Division of Clinical Pharmacology / Computer-Based Research On Drug Interactions and Antimicrobial Therapy / Joining the Department of Genetics / Congruency of Research Interests with Lederberg )

• Cohen: The job opportunity itself had its pluses and minuses. I knew many of the biochemistry faculty and they were interested in me and my work, and I felt that I would have strong scientific support from them. I expected that I would be able to discuss science with the Biochemistry Department faculty readily, and I liked that idea. The negative aspect was that the Department of Medicine position I was being recruited for was in the Division of Hematology and I had not been trained at all as a hematologist. At the NIH, where I had been a Clinical Associate, I learned something about clinical arthritis and immunology, but not hematology. And yet the department was interested in having a molecularly oriented hematologist and they wanted me to become one. Despite my misgivings about this plan, the overall attractiveness of the offer convinced me to accept. But after a couple of months of participating in clinical hematology rounds and conferences, I began to have concerns as to whether I had made the right decision. Looking at blood cells under the microscope and deciding whether the granules they contained were “big” or “small” or were stained pink or blue, which was an important and necessary part of hematological diagnosis at the time, was not something that I enjoyed. But I had made a commitment to try to learn to be a hematologist and I expected to do that.
• Hughes: Had you also made provisions to protect your research time?
• Cohen: Yes, and I was very hard-nosed about this issue. In fact, I had a discussion with Arthur Kornberg about this shortly after coming to Stanford. I had known Arthur from his visits to Jerry’s lab in New York, and I visited him soon after moving here. Arthur, who has always been very direct said, “Well, I hear you're going to be a hematologist.” I said, “Yes, I’m planning to try.” He said, “Well, that's a very demanding medical subspecialty,” and said that he didn't imagine that I would do significant research if I had a lot of clinical responsibilities. I told him that I was eager to protect my research time and I expected to seriously pursue studies of plasmids. He then said, “Well, if you're going to be seriously involved in research, then you're going to neglect your clinical responsibilities, and that's not appropriate either.” Arthur was pointing out, and it's true, that it is difficult to pursue a career in clinical medicine and basic science at the same time unless the basic science is very closely related to the clinical activities. Some of my colleagues on the Stanford faculty—Hugh McDevitt is one—have done that very successfully. Hugh has made major contributions in the area of immunology, and his clinical involvement has been in rheumatology and clinical immunology. But I was proposing to work on plasmids and at the same time was trying to learn to be a hematologist, and perhaps that wasn't a very smart way to proceed.
• Hughes: Kornberg was probably also speaking from his own personal experience, was he not?
• Cohen: Arthur had been trained as a physician. After some difficult clinical experiences during World War II, following his graduation from medical school, Arthur decided to abandon clinical medicine. But I've never discussed that decision with him. Arthur also told me he thought that plasmids were not a very interesting area to be working in. He said that most of the things that were important to know about plasmids were probably already known, and that I should study something meaningful, like phage. So this wasn't a very comforting introduction to Stanford. I should say, in fairness to Arthur, that he was making points that he felt strongly about, and his style is to leave little room for uncertainty. In retrospect, he was right about the difficulties of successfully pursuing a career in both clinical medicine and basic science, but wrong about plasmids. The other faculty in the Biochemistry Department were quite helpful to me in many ways. When I began my laboratory here, I continued for a while to study bacteriophage lambda gene expression, essentially extending some experiments I had started in Jerry's lab, while beginning work on plasmids. The laboratory that was assigned to me by Holman wasn't ready to be occupied at the time of my arrival, and Dave Hogness, who also worked with lambda, generously provided temporary space in his laboratory for me to use for a few months. As a result, I became friends with additional people in the Biochemistry Department, especially with the postdoctoral fellows and students. Most of the Biochemistry faculty were considerably older than I was, or at least it seemed that way, and I viewed the students and postdocs more as my contemporaries. Out of those interactions came relationships that were important in my later work.
• Hughes: So perhaps it was a blessing in disguise that you did not immediately get lab space in the Department of Medicine.
• Cohen: That's right. I interacted heavily with two separate groups of people during my first years at Stanford. One group included the biochemistry postdocs and students, and Peter Lobban was one of those students. We'll talk about his work in a little while. Lou Reichardt, another biochemistry student that I talked with a lot, shared a lab with Peter. Fred Welland was a young physician who had been trained in oncology and was receiving research training in Hogness' lab during the time that Hogness let me use one of his lab benches. Fred was especially helpful to me during my early months here. Just a short while later, Fred tragically developed an incurable cancer and, as an oncologist, he knew the prognosis and killed himself. I also became close to a group of young faculty in the Department of Medicine, all of whom had very solid clinical training and also significant training in basic science. Holman had recruited all of us within a relatively short space of time. Our laboratories were located near each other on the first floor of the S-wing of the medical school. They were Hugh McDevitt, who had discovered the genetic basis for the immune response and has become a world-class leader in the field of immunology, Tom Merigan, who became the head of the Infectious Disease Division shortly after his arrival at Stanford and has done important work on interferon and HIV, Frank Stockdale, who was a medical oncologist as well as a first-rate basic scientist who now has shifted his appointment primarily to the biology department, Bill Robinson, who was an excellent virologist, and me. We all had received rigorous basic science training and were here in clinical department appointments. We became friends and colleagues in the enterprise that Holman was trying to create: a Department of Medicine that was oriented towards the basic sciences.
• Hughes: Now were any of these people that you just mentioned in medicine interested in a molecular approach?
• Cohen: Yes, they all were in different ways.
• Hughes: Were you doing such things as attending lectures and seminars in the Department of Biochemistry?
• Cohen: Well, I was, but Merigan and McDevitt and Robinson were more closely aligned with the Department of Microbiology and Immunology. Bacterial plasmids were certainly related to microbiology, but my basic science connections were largely with the Department of Biochemistry because of my earlier research, and that department was where the lambda people were. A lot of work with viruses was being done in the Biochemistry Department, and you might ask, why not in microbiology? Well, the department name isn’t necessarily related to the area of faculty scientific interest.
• Hughes: I have heard it said, and it's been from UCSF scientists who may have an axe to grind, that Stanford is more insular than UCSF, that the departments are more hierarchically structured than is true at UCSF.
• Cohen: I don't know that the correct term is “hierarchically structured.” I wouldn't necessarily agree with that, but it’s true that departments at Stanford historically have existed as discrete units. That's probably part of the Kornberg legacy. Kornberg brought his colleagues from Wash U. to Stanford and created an elite Department of Biochemistry. The department faculty was a very distinguished one, and frankly, some of the faculty had an elitist attitude. I was accepted as a young Department of Medicine colleague and my science was respected by the Biochemistry department, but I think that some members of that department had different standards to evaluate science being done in a medical department and expected less. I had initially thought that a joint appointment in Biochemistry might be a reasonable way to formally recognize the scientific relationships that existed de facto, but soon recognized that this was not likely to happen.
• Hughes: Joint appointments were not very common?
• Cohen: They were quite uncommon at the time at Stanford. In fact, I wasn't aware of anyone who had a joint appointment in a clinical department and a basic science department when I first came here, and soon realized that the biochemistry department especially wouldn’t be likely to make joint appointments, at least at that time.
• Hughes: Is that largely an issue of power?
• Cohen: I can’t really assess that. I was a assistant professor just trying to get my research program started, and I appreciated whatever scientific help and advice I could get from members of the Biochemistry Department. We interacted amicably and closely. And whether there was some formal relationship with the biochemistry department was not an issue for me. The question I was trying to deal with was whether I wanted to continue to do hematology and make clinical rounds on patients who had hematological diseases. Caring for hematology patients was challenging. Most of them were very ill, and while I had been well trained in clinical medicine, I hadn't treated patients during the years I spent in the Hurwitz lab, and some relearning was necessary. Secondly, I found that even though I had assurance that I would have a large fraction of my time protected for research for at least the first year or two, there were increasing clinical incursions and requests to spend more and more time clinically. I knew that if I did this, my research program wouldn’t get off the ground. I wasn't inclined to be more involved with patient care, and yet I wanted to do a good job clinically. By early 1969, my plasmid work had progressed to the point where I had evidence that multiple molecular classes of resistance plasmids can exist concurrently in E. coli as DNA circles. These results were published in Nature in the last issue of the year [Cohen, SN, Miller, CA. Multiple molecular species of circular R-factor DNA isolated from Escherichia coli. Nature. 1969; 224: 1273-1277.]. Chris Miller, a young Berkeley graduate I had hired as a research assistant, was included as a co-author. Some months later, Chris and I extended these findings in a long paper in the JMB, and in October 1970 reported the isolation of the separate transfer unit of plasmid DNA [Cohen, SN, Miller, CA. Non-chromosomal antibiotic resistance in bacteria. II: Molecular nature of Rfactors isolated from Proteus mirabilis and E. coli. Journal of Mol. Biol. 1970; 50: 671-687.] . My reading of the literature was mostly in molecular biology, biochemistry, and microbiology and not so much in clinical medicine, except during the time when I was assigned to the attending or consulting service. I found that continued clinical involvement was becoming increasingly difficult. As an attending physician on the medical service, I phoned the medical resident the night before to learn what patients would be presented to me by medical students for discussion on the following day, so I could refresh my knowledge of those diseases and have something meaningful to teach to the students.
• Hughes: I think this is the time to bring in the Division of Clinical Pharmacology.
• Cohen: In 1969, after trying for more than a year to become interested in clinical hematology, I decided that it just wouldn’t happen. I discussed this with Hal Holman. He said, “Well, we'd like to keep you here on the faculty. What would you like to do medically?” I thought it was a remarkably open and generous response. I said I was potentially interested in the emerging field of Clinical Pharmacology. There was a need for better research on drug effects in patients, and for better teaching of rational drug therapy, and I had some ideas about developing computerbased systems to provide advice to physicians in this area. Particularly about drug interactions.
• Hughes: Now was that an idea whose time had come in medicine?
• Cohen: Well, I don’t know that its time had come, but it was an idea that I had at the time. Possibly it was a little premature. At the NIH I had studied the mechanism of action of chloroquine, and during my clinical work at Stanford, I became interested in the ability of concurrently administered drugs to affect the actions of another drug. Hal Holman said, “Okay, if you want to start and lead a Clinical Pharmacology division, go ahead.” Ken Melmon, whom I knew from the NIH and who at that time was head of Clinical Pharmacology at UCSF, was a great help in letting me visit UCSF to observe the workings of the Clinical Pharmacology program there. I spent several weeks traveling back and forth to UCSF. During that time, Ken and I, who had only been casual friends previously, got to know each other much better. Ken subsequently came to Stanford as Chairman of Medicine when I was Chair of Genetics and we worked closely as department chairpersons. We have continued to have a close friendship for many years. Holman’s go-ahead enabled me to leave Hematology and start the Division of Clinical Pharmacology in the Department of Medicine. An important factor in establishing my credibility in Clinical Pharmacology was that I received a Burroughs Wellcome Scholar Award, which was a very prestigious award in the field. The Burroughs Wellcome Fund was, and still is, a charitable foundation that supports the development of clinical pharmacology programs at universities in the U.S. I was nominated by Stanford as a candidate for the 1970 Burroughs Wellcome award, and I guess on the basis of my research work and the clinical pharmacology program that I proposed to establish, I was selected. That award provides funds intended to free up more time for laboratory research by recipients. I also applied for and received a Career Development Award from the NIH, which is also intended to provide salary support to enable recipients to focus on their research. So Stanford didn’t have to pay my salary, and with the approval of Burroughs Wellcome, I was able to use the funds I received from them to help support my research program more directly. So I got off to a good start in clinical pharmacology, as well as in my lab. But the Pharmacology Department here did not like the notion of having a Division of Clinical Pharmacology within the Department of Medicine, and was not especially supportive during those early years.
• Hughes: They saw you as a competitor?
• Cohen: Well, that’s an inference I wouldn't want to make. I think that perhaps they felt that I really didn't know a whole lot of pharmacology. They saw me as a biochemist, molecular biologist, or microbiologist, rather than a “card-carrying” pharmacologist; I was working with phage lambda and on plasmid biology, and it was hubris for me to begin a program in clinical pharmacology. In reality, I was more of a Clinical Pharmacologist than hematologist, but without the credibility that resulted from the Burroughs Wellcome Scholar Award, it would have been impractical to make the transition. Holman and I both thought that I could contribute to the field of clinical pharmacology, and the Burroughs Wellcome Fund thought that also.
• Hughes: And how long did that last?
• Cohen: The Division of Clinical Pharmacology that I started continues to exist. Initially, I was the only member of the Division. Then, I was able to recruit the support of Leo Hollister who had a lab at the VA [Veterans Administration hospital] at the time. Leo was a bona fide clinical pharmacologist who was doing very nice research on psychoactive drugs. His research contributions were well recognized outside of Stanford, but hadn't been adequately acknowledged at this university. He had an appointment in the Department of Medicine, but wasn’t part of the “mainstream.” I think he didn’t even have a tenure line appointment at that time. Yet, he was more senior than I was and certainly was more experienced as a clinical pharmacologist.
• Hughes: Do you think that lack of acknowledgment had something to do with his location at the VA?
• Cohen: I think in part it did. I also think that Leo was a very low-key person, and many people had the impression that he was less capable than he really was. I thought he was a smart and very able guy and was happy to have his participation in the development of a Clinical Pharmacology Division, and he was happy to be brought into the mainstream of Stanford faculty. Together, we were the nucleus of what became a successful Division of Clinical Pharmacology. Soon afterwards, I recruited another faculty person, Terry Blashke, to the Division faculty, and the Division continued to expand. I remained head of Clinical Pharmacology until I shifted my principal appointment to the Department of Genetics in 1978. In my role as a Clinical Pharmacologist, I worked on developing a computer-based reporting system to alert physicians about possible drug interactions. The drug interaction project attracted an extraordinarily capable first year medical student, Ted Shortliffe [Edward H. Shortliffe], who was the driving force in still another project in my lab: the development of a computer-based expert system to provide advice about antimicrobial therapy. Ted had done undergraduate work in medical computing at Harvard, and after considering different research options at Stanford, asked to work with me, and I was very happy about this. Ted applied to the MSTP (Medical Science Training Program), and although it took some persuasion from me to convince the selection committee that computer-based medical research was really “science,” he was selected as an MSTP student. Ted’s research was interdisciplinary and didn’t fit into any particular departmental program, so an interdepartmental committee was put together to eventually determine whether he qualified for a Ph.D. degree. That worked, and was in fact necessary because I did not have an appointment in a Ph.D.-granting department. Ted subsequently has had a remarkable career in academic medicine, and has become one of the leaders in the medical use of artificial intelligence methods. He’s now at Stanford as a Professor of Medicine. We established a collaboration with Tom Merigan and Stan Axline [Stanton G. Axline], another young faculty member in the Division of Infectious Diseases, and began to develop a system that Ted named MYCIN. We used a rule-based approach to provide medical advice about antimicrobial therapy. This meshed well with my interest in resistance to antimicrobial drugs. MYCIN and the drug interaction reporting system, which was called MEDIPHOR [Monitoring and Evaluation of Drug Interactions by Pharmacy-Oriented Reporting], were both very successful projects, and over a period of years I had substantial grant support for both projects from the NIH. As the size of the group I put together to do computer-based research in clinical pharmacology increased, we outgrew our space. I was able to get funds from the NIH to rent a small, prefabricated building, which was installed in a Medical Center parking lot. It included offices, a little conference room, and a room that contained several computers. This was before the days of PCs [personal computers], and we used small mainframe computers to do our analyses. So in the early 1970s, I was heavily committed to both my basic research on plasmids and to clinically-related, computer-based research. All of this provided a lot of intellectual stimulation and fun, but it became increasingly difficult for me to concurrently pursue two careers, one as a clinical pharmacologist and another as a basic scientist working on plasmids.
• Hughes: How did you balance those two lives?
• Cohen: With difficultly.
• Hughes: Did one or the other suffer?
• Cohen: I don’t know for certain. In principle, yes, one or the other probably suffered, but it wasn't apparent. Things seemed to be going well in both areas. I was working very hard and getting data that I was being invited to present at scientific meetings, and I was publishing our findings in first-tier journals. There was also some time spent writing a book on drug interactions [Cohen, SN, Armstrong, MF. Drug Interactions. Baltimore: Williams & Wilkins, 1974.]. That came about through a situation where I had agreed to write the book in collaboration with a clinical pharmacology postdoctoral fellow, Marsha Armstrong. I subsequently became so involved with my research activities that I wasn’t eager to proceed, but she was very insistent. I ended up deciding to do the book, and it turned out to be a useful contribution to the field at the time. Other than the work I was doing in Clinical Pharmacology, my research activities took me further and further from clinical medicine. But before we talk about my lab research, I should tell you that by early 1975, it was becoming increasingly clear that I couldn't continue to spend so much time in clinically-related activities and still pursue the scientific opportunities that my lab research had opened up. I left on sabbatical leave in mid 1975, and this postponed the need to deal with the issue. But returning to clinical responsibilities in after the sabbatical year ended brought it to the fore again. At that point Hal Holman was no longer chairman of the Department of Medicine. So, I went to speak with his successor, Dan Federman (Daniel D. Federman), to ask whether I could, at least during the next couple of years, be assigned fewer clinical responsibilities. He said that he was unwilling to do this and told me that if I wanted to have a full time appointment in the Department of Medicine, it would be necessary for me to continue to pull my weight clinically. Joshua Lederberg, who was Chairman of the Genetics Department, suggested that I consider a joint appointment in Genetics, and I was very attracted to this idea. I liked and admired Josh enormously. Josh was also very much involved with computer-based research and had been one of the developers of DENRAL, a computer-based expert system for analysis of DNA data obtained by mass spectrometry, and he was familiar with my computer-based research in clinical pharmacology. We had multiple common research interests, and Josh was very supportive of my science. Having an appointment in a basic science department as well as one in Medicine was very appealing, and I thought that taking on teaching responsibilities in Genetics might allow me to reduce my clinical role. The joint appointment in Genetics began in mid 1977.
• Hughes: A primary appointment has a stipulation about time commitment?
• Cohen: No, but it defines which department the faculty member has a principal relationship with, which department the research space comes from, which department sets the salary, which department provides administrative support, et cetera.
• Hughes: Is there a difference in salary between the clinical and basic science appointments?
• Cohen: There is.
• Hughes: So that was a consideration as well?
• Cohen: It was certainly a fact. But the salary difference wasn't something that I was considering in making a decision.
• Hughes: Was the disparity considerable? I ask that knowing the history at UCSF where the disparity between clinical and basic science salaries had been a bone of contention.
• Cohen: I think that the disparity is very considerable. There’s even a significant disparity between clinical departments in the surgical specialties vs. medical specialties. The Department of Medicine and the Department of Pediatrics are two of the lower paying clinical departments. That has to do with the compensation that physicians receive in private practice in the different areas of clinical medicine.
• Hughes: Yes, it's all competitive.
• Cohen: Right. Although I continued to hold a Department of Medicine appointment and did not take a salary cut immediately when I switched my primary appointment, I expected that I would not have an increase in salary for a while until the basic sciences salary level caught up.
• Hughes: The context of your basic science research up until this point was biochemistry.
• Cohen: And scientifically, also in genetics and microbiology.
• Hughes: Well, we'll develop that because we haven't heard much about genetics.
• Cohen: Formally, the disciplines of genetics and biochemistry are quite different. However, faculty in biochemistry departments and genetics departments often carry out similar research. Molecular biology was a child of both disciplines, and a reason that we've been talking so much here about biochemistry is that the Biochemistry Department at Stanford had people doing molecular biology in areas related to my scientific interests. Aside from Lederberg himself and another scientist named A.T. Ganesan, who worked on the bacterium B. subtilis and is now deceased, there was no faculty person in Genetics working on bacteria. There was a faculty person named Luca Cavalli-Sforza, who was, even then, one of the world’s leading human geneticists. Although Luca began his career in microbial genetics, his scientific interests had shifted. There was, Len Herzenberg, an immunogeneticist that Lederberg had recruited soon after he started the department, and Eric Shooter, whose work was primarily in neurobiology and who later became the first chair of a newly established Neurobiology Department at Stanford. There were also a couple of non-tenure-line appointments, but Genetics was a very small department at that time. Lederberg's continuing interest and involvement in plasmids went back for many years. His early work on bacterial conjugation and recombination, which earned him the Nobel Prize, was dependent on plasmids, and Luca Cavalli had independently worked with conjugation in the 1940s. They began their interactions in those days. Lederberg is the person who invented the term “plasmid”; he coined the term in a Physiological Reviews article in 1952 [Lederberg, J. Cell genetics and hereditary symbiosis. Physiological Reviews. 1959; 32: 403-430.]. As a member of the U.S. National Academy of Sciences, he later communicated to the PNAS [Proceedings of the National Academy of Sciences] two of the papers describing the early DNA cloning experiments from my lab. Lederberg’s scientific interests have always been extraordinarily broad. During the 1970s, he wrote a weekly science column for the Washington Post, and he also had played a major role in the development of artificial intelligence activities here at Stanford. Lederberg had established a computer system called ACME to carry out of his activities his research in computer sciences. It was the predecessor of the SUMEX-AIM “Artificial Intelligence In Medicine” system. When Lederberg left Stanford to become president of Rockefeller University, because of my own work with expert systems on the MYCIN project, I agreed to server as the Principal Investigator for the SUMEX-AIM grant for a year. After that, Ted Shortliffe, whose primary interest was in computer-based research, joined the Stanford faculty and took on that responsibility.
• Hughes: Another connection of Lederberg's with your subsequent work on recombinant DNA was an application that he made in the late 1960s to NIH for support on joining DNA from different sources [Wright, S. Molecular Politics: Developing American and British Regulatory Policy for Genetic Engineering, 1972-1982. Chicago: University of Chicago Press, 1994, p. 71. Hereafter, Molecular Politics.]. Do you remember talking about it early on with Lederberg?
• Cohen: I don’t. But, he had interests in a lot of different areas and that wouldn’t surprise me. After the initial experiments, Lederberg was one of the first people I discussed the results with.

EARLY LABORATORY RESEARCH AT STANFORD: SCIENTIFIC BACKGROUND AND INITIAL EXPERIMENTS ( Plasmid History / Role of Plasmids in Antibiotic Resistance / Molecular Nature of R Factors / Isolation of Circular R-factor DNA and Resistance Transfer Factor (RTF) / Hiring Lab Personnel: Annie Chang and Chris Miller / Expanding the Lab Group / Organization of Lab Activities During the Early Years / )

• Hughes: We mentioned plasmid research in past sessions, but I thought today we should talk in greater detail and perhaps you'd like to start with the history of the field.
• Cohen: Of plasmids?
• Hughes: Of plasmids.
• Cohen: Okay. Much of molecular biology in its early stages had to do with phage for reasons that I mentioned the last time we spoke, primarily because it was possible to make identical copies of individual phage particles. But a key aspect of Lederberg’s early work on recombination in bacteria in the 1940s involved a genetic element that was then called a fertility (F) factor and promoted gene transfer. Additional studies showed that the F factor is a plasmid that can sometimes integrate into the bacterial chromosome. Work with the F factor was also important in the development of the concept of the operon by Jacob and Monod. The E. coli lac gene, which provided the model for these experiments, had been picked up by the F plasmid when F inserted itself into the chromosome. I don't know the extent that…
• Hughes: Yes.
• Cohen: …I should go into the technical details here.
• Hughes: No this is fine.
• Cohen: So, plasmids had an important role in the early years of molecular biology. They were also important to the concept of the “replicon” developed by Jacob and his colleagues. Autonomously replicating extrachromosomal elements were sometimes called “episomes” at that time. As I’ve mentioned, Lederberg coined the term “plasmid” in his Physiological Reviews article in 1952.
• Hughes: Where did he get—why plasmid?
• Cohen: Because, as he mentioned in the article, they contribute to the genetic fluidity of the organism. For many years the molecular nature of plasmids was not at all clear. Some of the earliest work on this was done in the laboratory of Paul Doty at Harvard in the early 1960s. Stanley Falkow and Bob Rownd, whose work with R factor DNA a few years later was important in exciting my interest in plasmid biology, were collaborating with Julius Marmur and Doty to study DNA isolated from bacteria containing F factors, and the group found that F-factor DNA could be detected as a discrete entity. They transferred the F factor into bacterial species that contain a chromosome having a different nucleotide composition, and showed that the F-factor DNA formed a band at a different position in cesium chloride gradients [Marmur, J, Rownd, R, Falkow, S, Baron, LS, Schildkraut, C, Doty, P. The nature of intergeneric episomal infection. Proc Natl Acad Sci USA. 1961 July; 47 (7): 972-979.]. After the discovery of antibiotics in the 1940s, there was the prevalent view that these drugs would end infectious diseases caused by bacteria. Of course, that has not happened, and the reason was the advent of antibiotic resistance. Initially, the question of how resistance develops was controversial, and some workers in the field proposed that exposure to an antibiotic induced resistance. Others argued that bacteria acquire spontaneous mutations that make them insensitive to the antibiotics, and that resistant bacteria are given an advantage, in a Darwinian sense, by the widespread clinical use of antibiotics—which kill or restrict the growth of antibiotic-sensitive bacteria in microbial populations. But the resistant bacteria survive and propagate themselves. That controversy was resolved experimentally, and Lederberg and his first wife, Esther [Lederberg], did a crucial experiment using a procedure called replica plating [Lederberg, J, Lederberg, E. Replica plating and indirect selection of bacterial mutants. Journal of Bacteriol. 1952 March; 63 (3): 399-406.]. Would you like me to go into the details of the procedure?
• Hughes: Well, yes.
• Cohen: Essentially, a population of bacteria was plated on a petri dish that lacked any antibiotic. Then, a replica of this population, made by taking a sterile piece of velvet and pressing it against the top of the petri dish and then against the top of another petri dish, was created. But the second petri dish contained an antibiotic. This prevented the growth of most of the bacterial cells into colonies, but the Lederbergs found that there were some antibiotic-resistant colonies that grew on the second dish. They went back to the initial antibiotic-free petri dish and found antibiotic resistant bacteria at locations corresponding to the positions of these colonies, whereas the bacteria in other locations were sensitive. These experiments, plus others, helped to establish the mutational basis for antibiotic resistance. Early on, it was also thought that antibiotic resistance was entirely a chromosomal phenomenon. If the frequency of mutation to resistance to a single antibiotic in a bacterial population was 10-6, one in a million, then the chance of having the cell become concurrently resistant to two different antibiotics was 10-12. This led to the notion of circumventing resistance by using multiple antibiotics to treat infections. But in the mid 1950s there began to appear, initially in Japan and in England, instances of bacteria that were resistant to not just one antibiotic but were concurrently resistant to two or three or four drugs. As work continued with these resistant microorganisms, another very important phenomenon was discovered: resistance could be transferred between bacteria. Some resistant bacterial cells not only expressed resistance to multiple antibiotics, but also could transfer the multidrug resistance to other bacteria by cell-to-cell contact. The recipients of resistance could reproduce and generate a population of resistant offspring. In this way, resistance can spread rapidly through bacterial populations, and be transferred from a less pathogenic bacterial host to one that is more pathogenic. The genetic element responsible for antibiotic resistance in bacteria was termed an “R factor.” And, studies in the laboratory of Dr. Tsutomu Watanabe, in Japan, and by other scientists as well, particularly in Japan and in England—Naomi Datta and Guy and Eleanor Meynel, and E. S. Anderson—genetically mapped the locations of R factor genes. It was found that one locus of the R factor contains genetic information that enables transfer, and this was called the “RTF” or “resistance transfer factor,” whereas genes encoding antibiotic resistance were mapped genetically to another segment. Studies from the labs of Falkow and Rownd showed that R factors could be detected in cesium chloride gradients as bands located at a different position from chromosomal DNA. This work implied that the R factors are discrete units, but it was not known at that time that they consist of circular DNA molecules.
• Hughes: Is this idea of the extrachromosomal nucleic acid something that's happening on a broad basis or is this unique to—I'm thinking of [Barbara] McClintock's work with the jumping genes, for example.
• Cohen: Yes. Extrachromosomal DNA occurs in many types of organisms, but what I’m describing is different from jumping genes.
• Hughes: Which, if I understand the history right, was not a concept that was readily accepted. And I guess what I'm really asking is how acceptable is it to be thinking about genetic material that is outside the chromosome?
• Cohen: Oh, I think that was well recognized, at least genetically. Work by Lederberg and others had provided genetic evidence of “episomes.”
• Hughes: And we are now talking about the sixties?
• Cohen: Now we're up to the mid-sixties. In 1967 and 1968, other observations relevant to my planned work with plasmids were reported. One was the finding by Radloff, Bauer and Vinograd at Caltech that closed circles of DNA from a virus that infects mammalian cells can be separated from noncircular DNA using the dye ethidium bromide. The dye molecules insert themselves between coils of the DNA helix and change the spacing between those coils. If the duplex DNA exists as a closed circle, its ability to adjust to the change in spacing is constrained, and this causes the DNA circle to form a tightly twisted coil. The change in conformation alters the buoyant density of DNA when it is centrifuged in cesium salt gradients, so that coiled circular DNA can be separated from non-circular DNA. Vinograd’s lab originally did this with polyoma virus, and were able to separate polyoma virus circular DNA from noncircular DNA molecules [Radloff, R, Bauer, W, Vinograd, J. A dye-buoyant-density method for the detection and isolation of closed circular duplex DNA: the closed circular DNA in HeLa cells. Proc Natl Acad Sci USA. 1967 May; 57 (5): 1514-1521.]. This advance was important. It provided me and others with a method for separating circular plasmids from bacterial chromosomes efficiently. Other relevant discoveries were made in Don Helinski's lab at UC San Diego. This work, which was published by Helinski and his collaborators, Michael Bazaral and Don Clewell, in late 1968 and 1969, provided the first molecular evidence that the DNA of small colicinogenic plasmids is circular. [Clewell, DB, Helinski, DR. Supercoiled circular DNA-protein complex in Escherichia coli: purification and induced conversion to an open circular DNA form. Proc Natl Acad Sci USA. 1969 April; 62 (4): 1159- 1166. ] [ Bazaral, M, Helinski, DR. Circular DNA forms of colicinogenic factors E1, E2 and E3 from Escherichia coli. J Mol Biol. 1968 September 14; 36 (2): 185-194.]
• Hughes: May I ask you a question about that though? Are you emphasizing the circularity of it?
• Cohen: Yes. It had been well established that plasmids consist of DNA, and genetic evidence had suggested that F factors were circular. But Helinski’s work provided the first physical evidence of circularity, least for some small bacterial plasmids. My subsequent work and the work of others showed that large R factors are also circular DNA molecules, but the circularity of plasmids remained somewhat controversial. After I had been working on plasmids at Stanford for a little more than a year, I presented my lab’s evidence for the existence of R factor circles at a scientific meeting sponsored by the American Society for Microbiology in Miami Beach, I guess in mid-1969. A postdoc from Bob Rownd’s laboratory, which argued that R factors were not circular, stood up in the discussion period to question my conclusions and said, “Well, you know, Dr. Rownd believes that R-factor circles are an experimental artifact.” I was taken a little aback by this put-down; it was my first talk as an independent scientist.
• Hughes: But what is the real significance, though, about worrying about that fact?
• Cohen: Well, there was a fundamental uncertainty at that time about the structure of R factors. The circularity of R-factor DNA also had practical significance in terms of my subsequent work. By 1969 it was accepted that R factors are extrachromsomal elements, and that autumn, Helinski and his co-workers reported that colicinogenic factors are circular DNA; but colicinogenic factors are small genetic elements and there was evidence that R factors are much larger. There were disparate views about the molecular nature of R factors, and they were viewed by some as being linear DNA molecules. Even as late as the early 1970s, arguments were being made that R-factor circles were an artifact of the isolation procedure: that they became circularized during the course of isolation. But let’s go back a step. When I began at Stanford in March 1968, I was a young assistant professor eager to begin my research. I had received an NIH grant and wanted to start the research. The goal of the project was to elucidate how R factors had evolved and to learn whether they were formed by association of independent sub-units, as had been suggested by genetic data. I needed access to an analytical ultrcentrifuge to proceed with my experiments but hadn’t requested funds to purchase this instrument, which even in the late 1960s cost about $40,000, in the budget I had submitted to the NIH. Jerry Hurwitz had advised me to limit the amount of support I requested in what was my first research grant application; $40,000 was a lot of money for a single piece of equipment, and certainly a lot for an assistant professor to be asking for. There were a few analytical ultracentrifuges in the labs of other Stanford faculty, and Holman had indicated that I could make part-time use of those instruments in my research. However, it turned out that these centrifuges weren't as readily available as we had hoped. So I needed additional research funds. As I’ve already mentioned, when I arrived at Stanford I found that my assigned space was still occupied by others that Hal Holman had loaned it to on a temporary basis. Holman pointed out that I didn’t yet have funds to purchase an analytical ultrcentrifuge and wondered whether I actually needed to occupy the lab prior to obtaining the centrifuge. Through the generosity of several pharmaceutical companies that I made appeals to, I soon pulled together enough funds to purchase the centrifuge and move into my assigned space. But in the interim, I was able to begin doing experiments using borrowed laboratory space in the Department of Biochemistry. In a surprisingly short time after my move to Department of Medicine space in mid-1968, Chris Miller, who was a technician in my lab, and I worked out methods to isolate circular R-factor DNA using nitrocellulose. The nitrocellulose method enabled us to isolate circular plasmid DNA and analyze it by electron microscopy and by centrifugation. The results showed that E. coli bacterial cells carrying an R factor named RI contain multiple molecular species of circular DNA. The nitrocellulose method proved to be less efficient than the ethidum bromide method, but the data we obtained resulted in my first paper on R-factor DNA structure [Cohen, SN, Miller, CA. Multiple molecular species of circular R-factor DNA isolated from Escherichia coli. Nature. 1969; 224: 1273-1277.]. My next goal was to try to isolate the individual components of R factors. If R factors truly come apart, perhaps there was a way to isolate the resistance transfer factor [RTF] as a separate circular DNA molecule. I worked out a scheme for doing this, which involved multiple rounds of plasmid transfer. The notion was that by using multiple rounds of transfer, together with a screen that did not select for antibiotic resistance, it might be possible to isolate cells that contain only the RTF unit. And the strategy worked. Chris and I showed in a paper published in the PNAS in 1970 that the RTF unit of resistance plasmids could be isolated as a discrete, autonomously replicating genetic element [Cohen, SN, Miller, CA. Non-chromosomal antibiotic resistance in bacteria. III: Isolation of the discrete transfer unit of the R-factor RI. Proc Natl Acad Sci USA. 1970; 67: 510-516.]. Okay. So at that point we went on to study, in some detail using cesium chloride centrifugation methods, the conditions that affect the relative amounts of DNA components of large antibiotic resistance plasmids. Eventually these experiments led to a long paper in the Journal of Molecular Biology on the molecular nature of resistance plasmids isolated from E. coli and Proteus mirabilis [Cohen, SN, Miller, CA. Non-chromosomal antibiotic resistance in bacteria. II: Molecular nature of Rfactors isolated from Proteus mirabilis and E. coli. Journal of Mol. Biol. 1970; 50: 671-687.]. Similar studies of R-factor DNA by electron microscopy were going on in the laboratory of Roy Clowes, who was a professor at the University of Texas at Dallas and who subsequently became a good friend. It seemed to me that in order to further understand the functions of R-factor components, we needed to have a way of introducing the plasmid DNA into bacterial cells. Up until then, we were taking plasmid DNA out of bacteria and had isolated the transfer unit as well as the whole R factor. But we also wanted to learn what the other resistance-associated DNA bands we saw in gradients were. We had isolated the RTF as an independent replicon, but didn’t know whether the resistance gene component(s) would also be able to exist as separate replicons. We wanted to correlate each of the DNA bands with particular biological functions.
• Hughes: So in essence, were you moving from studies that emphasized structure into ones that are now emphasizing structure and function?
• Cohen: Yes. I think that's a fair way of putting it. During the first year or so of my work at Stanford on antibiotic resistance plasmids, I also was following up some of the earlier results that I had gotten in Jerry Hurwitz’s lab on bacteriophage lambda. This became a sort of “bread and butter” project, while I was beginning the new research. I’d like to backtrack and say something now about the people working in my lab early in my career at Stanford; they were doing much of the actual bench work for the experiments I’ve been describing. A few days after my arrival here in March 1968, I started interviewing candidates for a research assistant position. The first person I interviewed was Annie Chang. Annie was born in China and received her undergraduate degree from McGill University. She had a reasonable background in biology, but she knew nothing about DNA isolation or about plasmids. She previously had worked on a protein-related project, but in the course of the interview, I began to feel that she might be very suitable for the job I had available.
• Hughes: Why?
• Cohen: Well, she was very direct and she asked good questions. When she didn't know the answer to a question I asked her, she said so. It was clear that she had worked hard in the past, as I had, and I thought that she would be motivated to learn. When I began to tell her about plasmids and why I was interested in them, she made the point that plasmid research seemed to be an obscure area of biology. I agreed with her, but pointed out that what may be considered obscure to one person can be exciting and interesting to someone else. As an example, I noted that someone might be interested in the mechanism of joint articulation in an insect's knee, and I would find that area a bit obscure. And she smiled and didn't say anything. I learned later that her brother was an entomologist and had research interests somewhat akin to the example that I had chosen. I was advised by colleagues to interview several other candidates before filling the job opening, and I did that. But, I offered Annie the position, and she accepted and helped me to set up my laboratory. She still works in my laboratory, having gotten her Ph.D. degree subsequently, and is now in a Senior Research Associate position here at Stanford. Another person I hired a few months later—I had funds for two technicians—was Chris Miller, who also still works in my laboratory. I met Chris in May or June of 1968 when she was about to graduate from Berkeley. She started in my lab soon after her graduation and worked here for several years before moving to Chicago to be with someone she was seeing at that time. Periodically I stopped in Chicago en route to or back from the East Coast, and invited Chris to lunch or dinner, and tried to persuade her to come back to work in my lab at Stanford. Eventually, she did that, persuading her husband to accept a position in the Bay Area.
• Hughes: So Annie, at least originally, came with a biochemical background?
• Cohen: Yes, she came with some biochemical background. Chris also had been a biochemistry major at Berkeley. Both of them had undergraduate degrees but little or no subsequent training or research experience. I didn’t anticipate being able to find anyone who had experience working specifically with plasmids, but my feeling was that I would search for smart and motivated people and train them in the area I was interested in. During those first years, I was still able to work at the lab bench myself a fair amount of the time, and was able to provide hands-on training on how to do things.
• Hughes: Were you thinking at that stage of the work you were doing as essentially biochemistry?
• Cohen: Yes, that’s basically correct. There really was not a lot of genetics involved. The genetics of R factors had been investigated by Japanese scientists and others in England, and I was interested mainly in trying to learn about the molecular nature of resistance plasmids.
• Hughes: Is there any story behind the concentration—from what you were saying—of the research in three countries, namely in Japan, Britain, and the United States?
• Cohen: Historically, the phenomenon of multi-drug resistance was discovered in Japan in the mid 1950s and then was observed in England. But I hadn’t read those papers, at least not at the time of their publication. There was a later article in Scientific American by Tsutomo Watanabe in 1963, which brought antibiotic resistance plasmids, or antibiotic resistance factors, as they were called then, to wider attention in Western countries. I also hadn’t seen that article, but after hearing the seminar by Novick at Cold Spring Harbor and reading the papers in the JMB by Falkow and Rownd, I went back and read Watanabe’s Scientific American paper, and my reaction was, “Wow!” Collectively, all of these things made me want to learn more about bacterial antibiotic resistance. I’d like to say something else about my lab group during the early years. As a faculty member starting work in the Department of Medicine in 1968, I had no access to graduate student trainees. There was no graduate student training program in the Department of Medicine, and as I’ve mentioned, I didn’t have a joint appointment in a basic science department. So there was no prospect of having graduate students working in my lab, except through ad hoc programs like the one established later for Ted Shortliffe, and I didn’t yet have the publication record I thought would be needed to attract postdocs. But relatively soon after my arrival at Stanford, two things happened that expanded the size of my lab group. One was an inquiry by a young physician named Arnold Brown. He said that he was interested in pursuing a career that includes research and, wanted to receive basic science training. He had attended a seminar talk that I had given to a Department of Medicine group, and liked what he heard. He knew that it was relatively late in his career to start to learn molecular biology, but asked nevertheless to train in my lab. Although Arnie had no prior experience at all in laboratory research, and I knew that a lot of effort from me would be required to provide him with lab bench training, he was very motivated and I thought that he would learn. It was an opportunity to expand my lab group, and that was fine with me. I offered Arnie a postdoc position, and he accepted. As I began to report some of my lab’s initial observations on plasmids, one of Stanley Falkow's graduate students named Richard Silver decided that he wanted to come to my lab for postdoctoral training. Rich was interested in my work and in the approaches I was using. He had been trained in Falkow's lab, and he wanted to learn more about the molecular structure of plasmids. Falkow had told him that he thought my lab would be a good place to do this. So Rich applied for a postdoctoral position, and I was happy to accept him into my lab. So during those first few years, I was able to put together a laboratory group of four people— Rich, Arnie, Chris, and Annie—and I was working at the bench myself for part of the time, so it was a five-person lab group, and it was very closely knit. The group was small, but we worked on several different projects. Arnie and Annie were finishing up some of the work I had been doing with bacteriophage lambda. Rich was working on a plasmid project. Later, when Chris moved to Chicago, I hired Annette McCoubrey as a replacement. After the work with lambda was completed, I also switched Annie to a plasmid project. It was a good start for my lab and I enjoyed those years. I had clinical responsibilities also, as I’ve mentioned, but my research progressed nicely.
• Hughes: Was there any pattern to your day or your week? Were there times when you could count on being in the lab?
• Cohen: It depended on my clinical assignments. During my initial stint as a hematologist, I participated in a weekly hematological patient review session, which I think was on Wednesday afternoons. It lasted from noon or one o’clock until six o’clock in the evening. We reviewed the clinical status of each patient that had been on the hematology service during the week. I also saw hematology patients at other times, and made attending rounds on the general medical service. Most Fridays, in the mid-afternoon, I went out to the local Baskin-Robbins store to get ice cream for the people in my lab. I brought back ice cream and we would sit around together for a while and eat and chat. It kind of became a lab tradition.
• Hughes: Do you carry it on?
• Cohen: No, I don’t. I can do without the calories and cholesterol, and now the laboratory is much bigger and I have stopped working at the bench. But, I have tried to maintain a small lab atmosphere and we still have a mom-and-pop-shop-type lab. I should say something more about this before we get too far away from the point. These days in molecular biology, many larger labs, and even medium-sized labs such as mine, often have specialization within the lab, so that one person performs the same type of work, for example DNA sequencing, for a number of projects. Someone else is doing the plasmid construction and someone else is doing whatever. My lab has always been a place where a person works on multiple aspects of a project, learning the various techniques and concepts necessary for it. I think that this is probably not the most productive way to organize a lab, but I have always felt that it is the best way to train young scientists to do research. Research assistants in my lab almost always have been assigned to their own projects, serving as my “hands,” and have been authors on papers. When a paper reports work that a research assistant has had the primary role in carrying out, the RA has been listed as first author. Over the years, this has sometimes been problematical, when research assistants and postdocs have together collaborated on experiments. For example, I can remember one instance where a postdoc complained that first authorship was justified for her because she would soon be looking for faculty position, whereas the research assistant on the project didn’t have that pressing need. But the research assistant had made and recognized the importance of most of the key observations. Authorship should be determined by scientific contributions.
• Hughes: Do you want to say something now about communication? I mean, what means did you have to stay in touch with these various projects that were going on in the lab?
• Cohen: When my office was a little cubbyhole off of a single laboratory that contained only five or six benches, I could talk to people in the lab about their experiments everyday. We planned experiments together and students and postdocs usually would come into my office right after they got a result and discuss the data with me. It wasn’t a large operation.
• Hughes: That’s less possible now with a larger lab?
• Cohen: Well, it’s less convenient. That’s true. I don’t have the day-to-day contact with everyone in my lab that I did in those days, but I stay in pretty close touch. In addition to the general lab meetings we have on a weekly basis, I try to schedule individual meetings with people in my lab every few weeks. The timing depends on the person and the stage of the project. I’ve learned over the years that if it’s been a while since a student or postdoc has stuck their head through the doorway of my office to talk to me about results, this can indicate a problem with the way the project is going. When students are getting good results in the lab, they’re usually eager to communicate those results. If students haven’t stopped in to talk for a while, I’ll arrange a time to speak with them. The office that I currently have is located across the corridor from my lab. Before I moved into this lab space, it was used by the medical center to house small animals. It’s in a part of the building surrounded by a three-foot thick windowless “shearwall” that I was told was designed to withstand earthquakes. When the space was assigned to me, I redesigned it for use as a lab. I was able to get approval to put three windows in the thick outer wall facing the courtyard. I had to make the decision about whether the bench space or my office would have the windows, and I decided it was more important for the windows to be in the lab. I put my office, which was a little cubbyhole about half the size of the office that we’re in now, off of a corridor in the lab. That worked well for communication because lab people had to pass my office to go almost anywhere, and would stick their heads into my office regularly, sometimes multiple times during the course of the day when my door was open. I eventually became a little claustrophobic in my small, windowless office and persuaded the Dean to give me some additional space so I could have a window. But this office is located across the corridor from my lab space and one of the issues I had to face was being physically separated from my laboratory. I didn’t like that idea, but I don’t think it has significantly affected communication with my students and postdocs. Walking across the corridor is not quite as easy as passing my office door 10 times a day, but it works. Okay, let’s get back to the 1970s, unless we want to stop here.
• Hughes: I’m willing to go.

RESEARCH FINDINGS BY VARIOUS LABS PRIOR TO THE INVENTION OF RECOMBINANT DNA ( Uptake of Bacteriophage DNA by E. coli: the Work of Mandel and Higa / Cohen Lab’s Development of a System for Genetic Transformation for E. coli Using Plasmid DNA / End-To-End Joining of DNA Molecules by DNA Ligase / Work on DNA end Joining in the H. Gobind Khorana Lab / Lobban’s Priority / Gene Splicing versus Recombinant DNA )

• Cohen: Okay, let’s go on. My lab’s analysis of R-factor structure had gone well, but I realized that to make further progress I needed a way to introduce plasmid DNA into bacteria and have the plasmid genes expressed there. Bacterial transformation by DNA wasn’t a novel idea. It goes back to the work of Avery, McCarty and MacLeod who showed, by transforming Pneumococcus, that genetic information resides in DNA. After the Avery work, genetic transformation by DNA was shown also in other species of bacteria: Haemophilus influenza, and Bacillus subtilis, for example, but at that point no one had been able to genetically transform E. coli, which was the organism that I was working with. Around 1960, Dale Kaiser and David Hogness were introducing fragments of lambda DNA into E. coli to learn whether the genetic map is co-linear with the lambda genome. To get lambda DNA into the bacteria, they needed a live “helper” virus. A few years later, Rich Calendar at Berkeley found that using a buffer containing calcium ions increased lambda DNA uptake in the helper phage system. Then in 1970, [Morton] Mandel and [Akiko] Higa, who were working at the University of Hawaii in Honolulu, reported an important observation: treatment of E. coli with calcium chloride enabled the uptake of lambda DNA by E. coli even in the absence of helper phage, and the bacteria produced viable phage particles. They also tried to transform calcium chloride-treated E. coli genetically with chromosomal DNA but they did not get transformants, even though viable phage particles were made by the bacteria. Their work was described in a Note in the Journal of Molecular Biology. Peter Lobban was a graduate student in Dale Kaiser’s lab in the Department of Biochemistry, and he was using Mandel and Higa’s calcium chloride procedure to try to get uptake of DNA of the bacterial virus, P22 by E. coli. Peter got viable P22 phage when he transfected the DNA, as Mandel and Higa had found for lambda DNA. P22 DNA was taken up by the bacteria and virus particles were formed. I knew about Peter’s results, and wondered whether calcium chloride treatment would also work to get R-factor DNA into E. coli. But, unlike phage production by transfected bacteria, which requires only that the bacterial cells serve as a bag of enzymes to assist the phage in proceeding through its reproductive cycle, genetic transformation requires that transfected bacteria make copies of themselves, and of the genes that have been taken up. Mandel and Higa’s efforts to genetically transform calcium chloride-treated E. coli had failed, so why did I think that I might be able to get genetic transformation by genes carried by R factors? Well, to be propagated in calcium chloride-treated bacteria, the incoming chromosomal DNA that Mandel and Higa used had to recombine genetically with the resident chromosome. But my work had shown that R factors were autonomous replicons. Like phage, R factors can replicate on their own, and I thought that genes carried by these plasmids might be able to transform bacteria without entering the chromosome. It seemed worth trying. There was another event that occurred around this time, and that was the addition of a new person to my research group: Leslie Hsu. Leslie had graduated from Stanford in 1970 with an undergraduate degree in biology. She had done very well academically and had been accepted to Harvard as a graduate student in the Biolabs, but decided to delay her entry into graduate school for a year and to travel in Europe for with friends for several months. In January 1971, she returned to Palo Alto to look for a temporary job for five or six months before leaving for graduate school in Boston. I needed a secretary at that time. Up until then, I shared a secretary with several other junior faculty, but my research activities and backed-up manuscripts that I wanted to submit for publication were growing to the point where I needed additional office help. Funds to get this help were available from my Burroughs Wellcome Fund Clinical Pharmacology Scholar Award. So, I hired Leslie as my secretary. And she was absolutely great. First of all, she had been a pianist and had very adept hands and could type at an unbelievable speed. She was also very smart. Thirdly, she was interested in the work we were doing because she had been trained as a biologist. And after a few weeks, she asked if she could attend my lab meetings. That was an unusual request from someone working in an office, but certainly she was welcomed at our lab meetings. So she became even more interested in my lab work. As I’ve mentioned, in addition to the bench research going on in my lab, I had non-bench work going on with the computer-based drug interaction reporting system I was trying to develop. Leslie was interested in not only the research going on in the lab but also in the clinicallyrelated drug interaction project. After a few months, she told me that she was thinking about going to medical school and pursuing a career in both clinical medicine and basic research. She thought that what I was doing was a good model. To her, this was a more appealing plan than going to graduate school and pursuing a career fully in basic research. So by the end of May or the beginning of June, Leslie decided to apply to medical school. She had great grades as an undergraduate and excellent recommendations from her undergraduate advisors. And it was clear to me also that she was an intellectually gifted person whom I could recommend highly for admission to medical school and I did that. There was a space available in the entering class, and in June or early July, she was accepted to Stanford Medical School to begin classes a couple of months later, in September. Leslie wanted to pursue a research project in my laboratory as a medical student. The project that I assigned to her was to try to transform E. coli genetically with R-factor DNA. She and I got very helpful technical advice from Peter Lobban, who had the Mandel and Higa calcium chloride procedure working for P22 DNA, and Leslie found that she could introduce R-factor DNA into bacteria, and that cells taking up these plasmids could reproduce and could express antibiotic resistance genes carried by the R factor. Antibiotics were used selectively to allow the growth of cells that had taken up and were propagating the plasmids, and bacteria that expressed the antibiotic resistance genes grew into colonies. Initially, the efficiency of genetic transformation was low, but we made procedural modifications that improved it. The discovery that bacteria could be genetically transformed by R-Factor DNA was very exciting to me because it made possible the cloning of individual plasmid DNA molecules. Later, [S.] Cosloy and [M.] Oishi27 found that the ends of linear chromosomal DNA fragments, which Mandel and Higa were using in their unsuccessful transformation experiments, are degraded by exonucleases present in E. coli cells. When mutants of E. coli lacking exonuclease activity are used, chromosomal DNA can also transform E. coli. So a key factor underlying the success of our transformation experiments was the circularity of the R-factor DNA we were using. I think that our paper, which was published in the Proceedings of the National Academy of Science in August 1972, was viewed as “interesting,” [Cohen, SN, Chang, ACY, Hsu, L. Nonchromosomal antibiotic resistance in bacteria: Genetic transformation of Escherichia coli by R-factor DNA. Proc Natl Acad Sci USA. 1972; 69: 2110-2114.] but most people working in molecular biology at the time didn’t realize that we could now do with plasmids what could be done previously only with phage: namely clone entire extrachromosomal genomes. Although scientists working with plasmids were turned on by our publication, the most important aspect of this work---the ability to make clones of cells containing individual autonomously replicating DNA molecules, was missed by most of the scientific community. And the lack of greater realization of the implications of the R factor transformation work was fine with me. It gave me time to proceed further without the pressure of intense scientific competition.
• Hughes: Why did they miss that?
• Cohen: Well, I don’t really know. I suppose partially because there weren’t many researchers working with plasmids at the time. Genetic transformation had been demonstrated previously in other organisms using chromosomal DNA, so the introduction of genetic traits into bacteria wasn’t new. However, in the genetic transformation that had been shown for B. subtilis, the introduced DNA had to recombine into the chromosome to be propogated, and of course that requires homology between the introduced DNA and the chromosomal DNA. It wasn’t immediately apparent to many people that transformation by plasmid DNA was different from previous genetic transformation because plasmids were autonomous replicons that could be stably inherited without being integrated into the chromosome. No homology with chromosomal DNA is required. Also the title of our paper was, “Nonchromosomal antibiotic resistance in bacteriogenetic transformation of Escherichia coli by R-factor DNA,” and there wasn’t a lot of interest in R factors. I think most scientists didn’t grasp the significance of being able to take a plasmid out of a cell, introduce it into another cell, and than select cells that contain plasmids that are the progeny of a single DNA molecule taken up by a particular cell. But, being able to clone individual plasmid replicons was important to me because, as I explained earlier, I wanted to learn the genetic contents of the multiple R factor DNA bands that were being detected. By isolating the extrachromosomal DNA and using it to transform a population of bacteria, I hoped to obtain bacterial clones that contained different plasmid DNA species. I was very excited about our results, and, as you see, I’m still excited about the finding twenty years later.
• Hughes: Yes.
• Cohen: One of the questions I wanted to answer was whether the large R factors that we were working with were composed of multiple replicons that had been joined together. The R factors were very large DNA molecules—up to 100 kilodaltons. I thought that if we could mechanically shear these large DNA molecules and then introduce the fragments into bacteria, we might get recircularization of some of the fragments of the plasmid intracellularly. I should say something at this point about DNA ligases. In fact, let’s stop here.
• Hughes: Dr. Cohen, when we stopped last time, we were just about to enter the subject of ligases. So do you want to start there this time?
• Cohen: Yes, let’s do that. DNA ligases were discovered in the late 1960s. In fact, Jerry Hurwitz’s lab, where I was a postdoctoral fellow at the time, was one of several labs competing in a search for enzymes that can join together pieces of DNA end-to-end. The first report of discovery of a DNA ligase activity in E. coli came from the lab of Martin Gellert. Marty was someone I had known for several years. When I was at the NIH in Lemone Yielding’s lab, Marty, who was then working with Gary Felsenfeld at the NIH, helped me with technical advice during my spectrophotometric studies of chloroquine interactions with DNA. Anyway, Marty had the insight to use the complementary ends of bacteriophage lambda in his search for DNA-joining activity. It had been known for some time that the linear DNA of bacteriophage lambda has ends that can join together to make covalently closed DNA circles during the lambda life cycle. This occurs because lambda DNA ends are single-stranded and the nucleotides at each end are complementary to the nucleotides at the other end. Because of the complementarity, the two ends of lambda DNA can pair with each other, and be held together by hydrogen bonds. At a particular stage of the lambda life cycle, the lambda DNA ends come together to form circular molecules containing nicks, which are then sealed in vivo by an enzyme that forms covalent bonds between the nucleotides at the DNA ends. Marty used hydrogen-bonded lambda DNA circles to search for an activity in E. coli extracts that converted hydrogen-bonded circles to covalently closed ones, and he found it. Using lambda DNA circles, Malcolm Gefter, a graduate student in the Hurwitz lab, confirmed Gellert’s results and highly purified the E. coli DNA ligase to almost homogeneity. Other laboratories, including Bob Lehman’s laboratory and Arthur Kornberg’s laboratory here at Stanford, isolated ligase from E. coli about the same time. The fact that cohesive-ended molecules had been used as a substrate in these experiments was important in causing the scientific community to focus on using complementary nucleotides held together by hydrogen bonding to join together DNA ends.
• Hughes: Had there been talk before that date of the benefits of joining different types of DNA?
• Cohen: Not in the Hurwitz lab. If there was such talk elsewhere at that time, it’s unlikely that I’d have known about it.
• Hughes: The discovery of ligases prompted thinking along those lines?
• Cohen: Well, ligases certainly did provide a tool for linking DNA ends together. One of the first uses of ligase for DNA end joining was by [H. Gobind] Khorana and his collaborators. Khorana is a biochemist who in the mid 1960s developed methods for synthesizing small oligonucleotides that have a defined sequence. By linking together synthetic deoxyribonucleotides, he produced an intact gene for alanine transfer RNA, and in 1968 won the Nobel Prize for his role in elucidating the genetic code and establishing its function in protein synthesis. The sequence of the tRNA was known, and Khorana and his colleagues synthesized short DNA oligonucleotides containing overlapping ends to re-create the sequence encoding the tRNA. The sequence at the end of the deoxyribonucleotide chain was complementary to the sequence at the end of an adjacent one. Khorana used DNA ligase to covalently join DNA oligonucleotides that were held together by hydrogen bonding of base pairs in the overlapping regions and, in a laborious way, synthesized a gene. That work involved both DNA ligation and complementarity between DNA ends. Vittorio Sgaramella, a postdoc in Khorana’s lab, made the observation together with J.H. van de Sande that a DNA ligase encoded by the genome bacteriophage T4 can join together not only cohesive-ended molecules but even blunt-ended synthetic DNA molecules. That work was reported in a paper in the PNAS in 1970 [Sgaramella, V. Enzymatic oligomerization of bacteriophage P22 DNA and of linear simian virus 40 DNA. Proc Natl Acad Sci USA. 1970 November; 67 (3): 1468-1475.]. But, the scientific community was so focused on cohesive-ended DNA molecules that for a while, probably until 1974 or so, some scientists working on DNA end joining expressed doubt that DNA ends that are not complementary could actually come together and be joined. But the data in the Sgaramella paper are clear, and they showed that complementarity at the ends of DNA chains is not necessary for joining. Work by Peter Lobban, by Jackson, Symons, Berg, and by Jensen et al. on DNA End Joining
• Cohen: In 1969, Peter Lobban, who was a graduate student in Dale Kaiser’s lab, proposed, as part of an examination to qualify him for the next stage of his PhD training, a strategy to add complementary nucleotides to the ends of DNA so that different DNA fragments could be joined together. Instead of laboriously adding nucleotides one at a time to create a complementary DNA sequence as Khorana had done, Peter proposed using the enzyme terminal transferase which added a series of identical nucleotides to the 3’ terminus of a nonduplexed DNA strand or of a strand of a DNA duplex, adding a stretch, for example, of polyAs [adenines] to one population of DNA molecules. He then would add polyTs [thymidines] to another population of DNA molecules. By mixing the two DNA species, Peter expected that hydrogen bonding between the As and Ts would hold the DNAs together, and that he could then use E. coli DNA ligase to covalently link the molecules. Although this strategy was initially proposed for a hypothetical project, he then decided to develop the method as an actual thesis project. I was able to obtain a copy of Peter’s proposal in the mid 1970s when I wrote a Scientific American article about this overall technology [Cohen, S.N. The manipulation of genes. Scientific American. 1975; 233: 25-33.].
• Hughes: That idea was unique?
• Cohen: Well, it was very clever but not actually unique. There was another group that had, so far as I can determine, independently used the same approach. These were three scientists who worked for an industrial organization, the International Minerals and Chemicals Company, and most others in molecular biology probably are not aware of this group of scientists. And in mid 1971, these scientists, Jensen, Wodzinski, and Rogoff, published a paper [Jensen, RH, Wodzinski, RJ, Rogoff, MH. Enzymatic addition of cohesive ends to T7 DNA. Biochem Biophys Res Commun. 1971 Apr 16; 43 (2): 384-392.] reporting use of the same strategy; they added a stretch of Ts to one batch of DNA molecules and a stretch of As to another. I think they used the DNA of bacteriophage T7 or another phage as a substrate. They showed by sedimentation in gradients that DNA molecules were held together by dA-T [deoxyadenine-thymidine] tails. Then, they incubated the molecules with DNA ligase, but when they then heated the mixture, the DNA molecules came apart, so their attempts to join the DNA molecules covalently were not successful. So by 1971 there were three groups using complementary dA-T tails added biochemically to hold DNA fragments together: Jensen and his colleagues, Peter Lobban and Dale Kaiser, and the Jackson, Symons and Berg group. Lobban was trying to join together segments of the genome of a bacteriophage, P22, and Jackson, Symons and Berg were trying to join lambda dv, which is a circular variant of bacteriophage lambda, to DNA of the mammalian virus SV40. How Jackson, Symons, and Berg began using the dA-T joining approach has never been entirely clear to me. When I wrote my Scientific American article in 1975, I tried to determine who in the Stanford Biochemistry Department first had the idea for using that method to link DNA molecules together. Peter Lobban presented this strategy in his qualifying exam proposal in 1969. Paul Berg has said that he didn’t learn of Lobban’s proposal or thesis project until after Dave Jackson had begun his SV40 experiments, which was not until sometime in 1970, according to Jackson’s account in his M.I.T. oral history [Jackson, D. Recombinant DNA Oral History Collection. MIT.]. Paul signed Peter’s final thesis dissertation as a member of the examining committee, but has indicated that he was not a member of the faculty committee that reviewed and evaluated Lobban’s qualifying exam proposal. In any case, the work on dA-T joining was carried out in the Department of Biochemistry concurrently by David Jackson and Bob Symons in Paul Berg’s lab, and by Peter Lobban in Dale Kaiser’s lab using different kinds of DNA. Both groups have said that they shared information. The Jackson, Symons, and Berg [Jackson, DA, Symons, RH, Berg, P. Biochemical method for inserting new genetic information into DNA of Simian Virus 40: circular SV40 DNA molecules containing lambda phage genes and the galactose operon of Escherichia coli. Proc Natl Acad Sci USA. 1972; 69 (10): 2904-9.] paper credits Peter Lobban for having discovered two crucial steps in the dA-T joining procedure that allowed Berg and his colleagues to link the molecules covalently, and which presumably prevented Jensen et al. from being successful in their earlier attempts at in vitro ligation. One of these steps increased the efficiency of adding tails to DNA; terminal transferase adds nucleotides to 3’ extensions and if the extensions are longer, the enzyme works better. Peter used lambda exonuclease to trim back the 5’ ends of the duplex DNA, which increased the single-strand length of the 3’ extension and made it easier to add tails. The second step, which I think is probably the more crucial one, was Peter’s discovery that to get good ligation, phosphates had to be removed from the terminus of the DNA molecules that were to be linked together, and he used a specific E. coli exonuclease [exonuclease III] to do this. My understanding from the publications is that there was a contaminant in the terminal transferase so that the DNA fragments had mixed phosphate and hydroxyl termini. A way was needed to remove the 3’ terminal phosphate, and this was important in being able to seal the nick and get covalent joining of DNA ends. Lobban was successful in splicing together segments of DNA from bacteriophage P22; and using information that they said was received from Lobban, the Jackson, Symons, and Berg group linked SV40 DNA to lambda dv DNA using the dA-T method.
• Hughes: Were you following closely what was happening in the two respective labs?
• Cohen: Well, I certainly knew about Peter Lobban’s work, and I also knew more generally about Dave Jackson’s work with SV40. I knew that Peter had joined monomers of P22 using dA-T tails, and that Berg’s group was doing similar experiments to join lambda dv and SV40. I think that most people at Stanford viewed the method as Lobban’s and considered the lambda dv-SV40 work to be another application of the approach that Peter was using for his thesis project.
• Hughes: Which is not the tilt that one gets by reading Lear’s book on the history of recombinant DNA, where he puts Lobban forward as being the neglected student who didn’t receive proper credit.
• Cohen: Well, I think that Lear’s take is correct. Peter’s role was clearly recognized at Stanford, but he hasn’t received proper credit outside of Stanford. That’s exactly the view that those of us who were here have, and it’s different from the view of most of the rest of the world. I’ve talked about this point with Lou Reichardt, who worked at a bench in the same lab as Peter, and with others in the Department of Biochemistry. All of us knew that Peter had worked out the technical problems in dA-T joining to make the procedure work, and this was acknowledged in the Jackson, Symons and Berg paper. Although Peter proposed his project the year before Dave Jackson started working on dA-T joining and Peter worked out the bugs in the procedure, it’s sad that Peter was viewed by much of the outside world as having simply having used a method designed and developed by Berg group. After his postdoctoral fellowship, Peter was not able to get a suitable faculty position and he ended up leaving the biological sciences and studying engineering. I think that he currently works as an engineer and lives in the Bay Area.
• Hughes: Do you think that his status as a graduate student at the time of this work made a difference?
• Cohen: No, the authors of his paper were Lobban and Kaiser; Dale Kaiser was a senior faculty person at the time. Just as Jackson and Symons were postdocs in Berg’s lab, Lobban was a graduate student in Kaiser’s lab. In recognition for Dale Kaiser’s role, the year [1980] that Boyer and Berg and I received the [Albert] Lasker [Basic Medical Research] Award, Dale shared the award with us. So there are some in the scientific community who are aware of this history. So as you see, splicing of synthetic and natural DNAs had been done in various ways before my collaboration with Boyer. The covalent linkage of deoxribonucleotide chains had first been accomplished by Khorana and his collaborators. The linkage of separate pieces of natural DNA by adding complementary “tails” was reported first by Jensen and his colleagues, although this joining was not covalent. Peter Lobban and Dale Kaiser had achieved covalent joining of duplex DNAs of a phage, and Jackson, Symons, and Berg had done this with SV40 and lamda dv. In the Jackson, Symons and, Berg work, the DNA came from different sources; one was an animal virus and one was a bacterial virus. But chemically, the same in vitro procedure, which depended on the actions of terminal transferase, lambda exonuclease and exonuclease III was used for both. I think there is the notion among the general public, and maybe even among some scientists, that recombinant DNA is equivalent to gene splicing. But the splicing together of DNA molecules in vitro is only part of “recombinant DNA”. A key additional step is the in vivo propagation and cloning of the DNA molecules that have been biochemically joined.
• Hughes: From what I understand, Berg’s group had conceptually carried the research a bit further in that by the summer of 1972, they were thinking of introducing the chimera into E. coli.
• Cohen: I think that occurred even prior to 1972. He was planning on trying to use SV40 to introduce DNA into animal cells.
• Hughes: My point is that Berg indeed seemed to be thinking about what you are calling gene cloning. As we know, he decided not to do that experiment because of the potential biohazards. Am I right?
• Cohen: Berg’s writings about this point indicate that his goal was to see whether SV40 could work as a mammalian version of a transducing bacteriophage virus, and you could certainly say that transduction is a biological way of cloning genes. Actually, the concept of gene cloning by viral transduction goes back to the early work of Lederberg, and was also considered by Peter Lobban in his 1969 proposal.
• Hughes: Ah. So Lobban was interested in more than just splicing?
• Cohen: I think that his interests were more general. Peter was working with P22, which was the bacteriophage that Norton Zinder and Joshua Lederberg had used in their discovery of atransduction, which involves the picking up of chromosomal genes during the normal phage growth cycle [Zinder, ND, Lederberg, J. Genetic exchange in salmonella. J. Bacteriol. 1952; 64 (5): 679-699.]. Peter’s 1969 proposal and the paper that Lobban and Kaiser published [Lobban, PE, Kaiser, AD. Enzymatic end-to-end joining of DNA molecules. J. Mol. Biol. 1973; 78: 453-471.] on their dA-T joining work with P22 DNA recognized that it might be possible to generate transducing phages biochemically by inserting blocks of genes from other organisms into the phage genome. And Lederberg and others have indicated that they also had been thinking about gene cloning. This was a natural outcome of the Zinder and Lederberg transduction experiments. So, the concept of gene cloning wasn’t novel at that point; it preceded my own work, Berg’s work with SV40, and Lobban’s work. At a Cold Spring Harbor Lab course in 1971, discussion with Berg’s student, Janet Mertz, led Robert Pollack, a CSH scientist, to raise concerns about possible biohazards of such hybrid molecules, and as you’ve just said, Berg decided not to continue working with them [Berg, P. Dissections and Reconstructions of Genes and Chromosomes. Nobel lecture. 1980 December 8; Department of Biochemistry, Stanford University School of Medicine, Stanford, California 94305.]. But putting aside any such biohazard concerns, there wasn’t, so far as I know, yet a way to infect mammalian cells with SV40 DNA—so it would have been necessary to develop such a method before being able to actually use SV40 to propagate and clone genes in mammalian cells. Berg indicated later [Lear, J. Recombinant DNA: The Untold Story. New York: Crown Publishers, 1978, p. 44-46.] that he also planned to try to introduce the SV40 DNA into E. coli to determine if the SV40 DNA genes would be expressed there. But how he planned to do this isn’t clear: Mandel and Higa had reported their inability to genetically transform E. coli cells using the calcium-chloride procedure, and prior to Leslie’s Hsu’s experiments with R-factor DNA, which I talked about earlier, it wasn’t known that genetic transformation of E. coli was achievable.
• Hughes: Scientifically, it wouldn’t be terribly interesting to splice pieces of DNA together?
• Cohen: Oh sure it would be.
• Hughes: Why?
• Cohen: Well, Khorana spliced together deoxyribonucleotide chains to create a whole gene, and these experiments helped to provide an understanding of the genetic code and of mechanisms of protein production. In fact, there were lots of people interested in mechanisms of DNA repair and DNA end joining per se, including Gellert, Hurwitz, and others. Two papers by Sgaramella around this time were about DNA end joining. [Sgaramella, V. Studies on polynucleotides: A novel joining reaction catalyzed by T4 polynucleotidal ligase. Proc Natl Acad Sci USA. 1970; 87: 1468-75.] [Sgaramella, V. Enzymatic oligomerization of P22 DNA in a linear simian virus SV40 DNA. Proc Natl Acad Sci USA. 1972 November; 69 (11): 3389-3393.] The Jackson, Symons, and Berg paper focuses on the biochemical joining reaction. Even today, there is a lot of scientific interest in the biochemistry of DNA end joining and, more generally, in DNA repair and in the enzymes that do this. I think that there’s an important point here that is often overlooked: It’s sometimes stated that the reason the Berg lab wasn’t the first to show that DNA cloning is possible is that biohazard concerns led Berg to abandon further experiments with SV40 and lambda dv hybrids. But if Paul believed he had an approach for actually propagating foreign DNA in E. coli using lambda dv, why not use another piece of DNA instead of SV40? There was no general concern about propagating DNA in bacteria at the time, and the issues of concern to Robert Pollack and some others related specifically to the fact that SV40 is a tumor virus.
• Hughes: Right.
• Cohen: I posed this question at different times to both Bob Symons and Dave Jackson; I don’t think I’ve ever asked Paul. Bob and David said they weren’t thinking in those terms at the time. But after it was shown by Leslie Hsu that E. coli can be genetically transformed, and we published this in August 1972, the Berg group could, in principle, have tried to use lambda dv to introduce some other DNA fragment in bacteria without having to worry about the biohazard issues that were raised by SV40. But if Berg had tried such an experiment, the experiment wouldn’t have worked. The reason is that they were adding dA-T tails to lambda dv DNA that had been cut with the EcoRI restriction enzyme to linearize the DNA. The EcoRI cleavage site in lambda dv is in the O gene, which is required for replication of the bacteriophage, and constructs containing inserted DNA fragments at that site would not replicate in E. coli, and therefore can’t be propagated. That was not known in 1971 or 1972. So it would have been problematical to clone any DNA in bacteria using the system that Berg and his colleagues described.
• Hughes: The line of research using SV40 was temporarily stopped because people were thinking in terms of potential biohazard, not in terms of generalizing this technique and using viruses that did not have a potential biohazard.
• Cohen: Or not using mammalian viruses.
• Hughes: Right.
• Cohen: Anyway, if Jackson et al. had tried DNA cloning in bacteria using lambda dv, they would have gotten a negative result, and this might have been interpreted as indicating that DNA hybrids made in vitro can’t be propagated.
• Hughes: In terms of acceptance by the scientific community, did it make a difference that Peter Lobban was a graduate student and his research findings appeared in his dissertation? A dissertation does not normally have wide readership.
• Cohen: It’s not that Peter’s findings weren’t accepted. They were published in a leading scientific journal, the Journal of Molecular Biology [Lobban, P, Kaiser, D. Enzymatic End-to-End Joining of DNA Molecules. J Mol. Biol. 1973; 78: 453-471.]. But by the time Lobban’s work was published (August 1973), the complementary nature of DNA ends generated by cleavage with the EcoRI endonuclease had been shown by several labs, and making complementary DNA ends this way was much easier than adding dA-T tails. Also, my work with Annie Chang, Herb Boyer, and Bob Helling had shown that EcoRI-generated DNA fragments could be cloned in bacteria using plasmids. Boyer mentioned these experiments at a Gordon conference on nucleic acids in June 1973 and word of the results had spread. So interest began to turn from phage to plasmids as possible vehicles for propagating DNA, and from using dA-T tail addition, to using restriction enzymes to generate complementary DNA ends. Anyway, my point was that given Paul’s statements about his intent [Berg, P. Dissections and reconstructions of genes and chromosomes. Nobel lecture. 1980 December 8; Department of Biochemistry, Stanford University School of Medicine, Stanford, California 94305.], it’s puzzling that he didn’t try DNA other than SV40.
• Hughes: That was a conceptual block?
• Cohen: Conceptual block in what sense?
• Hughes: That line of research was not pursued in the sequential way that it might have been if the biohazard issue hadn’t come up.
• Cohen: Well, maybe the block was conceptual, or possibly experimental. The biohazard concerns raised were about just tumor viruses and the concerns didn’t preclude cloning of other DNAs.

DNA CLONING: THE INVENTION OF RECOMBINANT DNA ( Leading Up to the First Cohen-Boyer Experiment / The Species Barrier Issue / Scientific Goals in the Development of Recombinant DNA Methodology / Restriction Enzyme History / Inviting Boyer to the Honolulu, Hawaii Meeting on Plasmid Biology, November 1972 / Work by Sgaramella and by Mertz and Davis Showing that the EcoRI Restriction Enzyme Generates Complementary DNA Termini / Caveats About the Feasibility of DNA Cloning / Initial DNA Cloning Results / pSC101: the First Vector for Recombinant DNA / Measuring Success in the Experiments / Contributions of Individual Team Members / Recognizing Potential Industrial Applications )

• Hughes: Well, could I clarify the connection between this work that you’ve been describing—the Lobban and Kaiser, and Jackson, Symons and Berg work—and your own? Am I understanding correctly that you were pursuing your interest in plasmid science which made it of great importance to you to be able to clone the materials that you were interested in studying and in a sense it was almost incidental that this other line of research was going on? To put it very simplistically, you didn’t look at what Lobban and the Berg group were doing and decide, “I’m going to pursue a different approach.”
• Cohen: What you’ve said is exactly correct. What they were doing was incidental to my work. As I’ve said, my goal was to isolate plasmid DNA molecules and re-introduce them into E. coli, and we worked out a procedure for doing this. The next step was, okay, can we take plasmid DNA molecules apart and isolate the replication region? We were trying to do this by shearing plasmid DNA molecules into pieces mechanically, and I hoped to get rejoining of some of these fragments by DNA recombination in cells after the fragments were introduced into calcium chloride-treated bacteria. The work by Lobban and Kaiser, by Berg’s group, and by Jensen was focused on the biochemical joining of DNA ends rather than on separating or isolating genes. My focus was as much on taking plasmids apart as well as on putting them together in order to identify individual plasmid genes. Restriction enzymes offered a possible way for me to do both. Mechanical shearing broke different DNA molecules differently; restriction enzymes cut DNA molecules in the population uniformly. Early on, my lab showed that multiple resistance plasmids can coexist in bacteria as separate pieces of circular DNA [ Cohen, SN, Miller, CA. Multiple molecular species of circular R-factor DNA isolated from Escherichia coli. Nature. 1969; 224: 1273-1277. ] [ Cohen, SN, Miller, CA. Non-chromosomal antibiotic resistance in bacteria. III: Isolation of the discrete transfer unit of the R-factor RI. Proc Natl Acad Sci USA. 1970; 67: 510-516. ] [ Cohen, SN, Miller, CA. Non-chromosomal antibiotic resistance in bacteria. II: Molecular nature of Rfactors isolated from Proteus mirabilis and E. coli. J Mol Biol. 1970; 50: 671-687. ] The need to separate and isolate different R-factor DNA species was the driving force that led me to try to genetically transform E. coli with this DNA. Once we found that we could transform E. coli cells with R-factor DNA and showed that transformed bacteria acquire DNA circles having all of the properties of the parent R factor, I thought, well okay, here is an autonomously replicating DNA molecule that we can take out of a bacterial cell, put back into another cell, and propagate and clone it. [Cohen, SN, Chang, ACY, Hsu, L. Nonchromosomal antibiotic resistance in bacteria: Genetic transformation of Escherichia coli by R-factor DNA. Proc Natl Acad Sci USA. 1972; 69: 2110-2114.] Would it be possible to attach other plasmid DNA fragments to the plasmid replication region so that segments containing specific plasmid genes can be identified? I knew from the heteroduplex experiments that Phil Sharp, Norman Davidson, and I had reported that large plasmids probably evolved in nature by genetic recombination events that joined resistance genes to replication regions. At that time, I was thinking just about studies of E. coli plasmids. It was known, as Herb Boyer has probably told you in discussing restriction enzymes [See the oral history with Herbert W. Boyer in this Bancroft Library series], that bacteriophage propagated on one E. coli strain can be restricted in its ability to grow on a different strain. So propagation of phage between strains of even the same species of bacterium is sometimes difficult, and there was a general belief that “natural barriers” would preclude DNA exchange between unrelated biological species.
• Hughes: Jumping ahead, the obstacle you confronted with the Xenopus work, even if you were successful in transforming bacteria, was: Would a gene function in a foreign host?
• Cohen: Right, would it function? And could you propagate it?
• Hughes: And the feeling was that it probably wouldn’t?
• Cohen: Well, there was evidence that the mechanisms and signals governing gene expression and DNA replication in prokaryotes and eukaryotes are different. And, eukaryotic mRNA molecules contain stretches of A nucleotides at their 3’ ends, and these hadn’t been thought to occur, at least then, in bacteria. DNA isolated from different species was often different in nucleotide composition. There were multiple reasons to doubt that simply linking a DNA fragment biochemically to an E. coli plasmid replicon would enable propagation of the foreign DNA in E. coli and that chimeric DNA molecules made biochemically would be viable and functional in cells. Certainly, many of my colleagues at Stanford originally thought that the biological crossing of species barriers would not be successful.
• Hughes: Another point, which is inherent in what we’ve been saying but perhaps should be stated explicitly, is that you were really not focusing on the methodology. I mean, it was not your idea to develop what later became recombinant DNA technology.
• Cohen: That’s correct.
• Hughes: What you were trying to do was to pursue your science, and for your science you needed this particular method.
• Cohen: Right.
• Hughes: Is that the order of priority?
• Cohen: Yes.
• Hughes: I think we now look back and tend to see the technology as being the dominant thing, where in actuality it was the science.
• Cohen: I think that’s an important point, Sally, and that was also probably true for at least some of the other people that were working [in the field] as well. In my case, the technology was developed out of necessity so that we could study antibiotic resistance plasmids.
• Hughes: We started this session with the idea of pulling together the different strands that went into what eventually became recombinant technology. According to your Harvey lecture [Cohen, SN. The transplantation and manipulation of genes in microorganisms. The Harvey Lectures. New York: Academic Press. 1980; 74: 173-204.], there were four elements that you felt were necessary.
• Cohen: Right.
• Hughes: We’ve got the ligases and we’ve got the cloning vehicle.
• Cohen: Yes, which was the plasmid.
• Hughes: And we’ve got the procedure for introducing hybrid molecules into a cell.
• Cohen: Transformation.
• Hughes: Right.
• Cohen: And we’ve got the joining [ligation].
• Hughes: And we talked a little about the restriction enzymes. Do you want to say more?
• Cohen: Okay, let me tell you my understanding of the history of these enzymes, which were discovered almost a decade prior to the experiments we’ve been discussing. As I’ve mentioned, restriction of bacteriophage growth by some bacterial strains had been known for some time, largely from the early work of Werner Arber. But some phage escape the restriction mechanisms. It was found that specific enzymes restrict phage growth by cleaving the phage DNA, and these are called “restriction enzymes.” Other enzymes can modify the phage DNA to make it unsusceptible to cleavage by the cognate restriction enzymes, and these are called “modification enzymes.” Restriction and modification enzymes commonly work in pairs. This was the phenomenon that Herb was interested in studying. An early worker in the field of restriction/modification was Daisy Dussoix, who as a graduate student in Arber’s lab in Switzerland, had made observations central to the discovery of the restriction phenomenon. Later, Dussoix, whose name had become Roulland-Dussoix, moved to UCSF and collaborated with Boyer. Also at UCSF was a graduate student named Robert Yoshimori who, as I understand it, was initially a student of Dussoix. When Dussoix left UCSF in the early 1970s, Herb inherited Yoshimori as a student. Yoshimori had identified an enzyme that came from an E. coli strain isolated from a patient hospitalized at UCSF. It was encoded by an antibiotic resistance plasmid [Yoshimori, R, Roulland-Dussoix, D, Boyer, HW. R factor-controlled restriction and modification of deoxyribonucleic acid: Restriction mutants. J Bacteriol. 1972 December; 112 (3): 1275-9.], so the history forms a circle, in a sense. And as I’ve mentioned, Tsutumo Watanabe in Japan, who had done some of the early major work with antibiotic resistance plasmids, had also found that certain resistance plasmids encode restriction/modification systems. The plasmid that Yoshimori had identified and isolated encoded a restriction enzyme called EcoRI (E. coli restriction enzyme I). This was the enzyme used in the initial experiments that Boyer and I did. As I’ve mentioned, Don Helinski and Watanabe and I organized an NSF (National Science Foundation)-sponsored plasmid DNA meeting in Hawaii in November 1972 [November 13-15, 1972]. Watanabe was quite ill in the months prior to the meeting and died just a few days before it began. Don phoned me a couple weeks before the meeting to tell me he had learned about some recent work that Herb Boyer had been doing with restriction enzymes encoded by plasmids. I didn’t know Boyer personally and I knew relatively little about his work, although I had seen some of his papers. He had published a review on restriction enzymes and several other papers in that area. At that point he hadn’t published anything on EcoRI. Since this was a meeting about plasmid biology, Don suggested that we invite Herb as a participant, and I thought that was a good idea. So, I wrote to Herb extending a formal invitation on behalf of the two of us as the co-organizers, and also on behalf of Watanabe. So Herb showed up at the meeting. About the time that the meeting was held, some additional relevant papers were published. These were all in the November 1972 issue of the PNAS. There was a paper by Joe Hedgepeth, Howard Goodman, and Herb Boyer in which they showed that the EcoRI enzyme-generated DNA ends have a unique sequence and that DNA ends generated by EcoRI cleavage were complimentary [Hedgpeth, J, Goodman, HM, Boyer, HW. The DNA nucleotide sequence restricted by the RI endonuclease. Proc Natl Acad Sci USA. 1972; 69: 3448-3452.]. They did this by using DNA sequencing of the ends. There were two other papers on the complementarity of EcoRI generated ends that were published in the same PNAS issue: one was Vittorio Sgaramella’s paper on covalent joining of DNA molecules [Sgaramella, V. Enzymatic oligomerization of bacteriophage P22 DNA and of linear Simian virus 40 DNA. Proc Natl Acad Sci USA. 1972 Nov; 69 (11): 3389-93.] and the other was by [Janet] Mertz and [Ronald] Davis. [Mertz, JE, Davis, RW. Cleavage of DNA by R 1 restriction endonuclease generates cohesive ends. Proc Natl Acad Sci USA. 1972 November; 69 (11): 3370-4.] I’d like to back up a bit here and provide some background information. Sgaramella, who earlier had trained in Khorana’s lab and had participated there in studies of the chemical joining of DNA segments, had come to Joshua Lederberg’s lab as a postdoctoral fellow, I think in 1970. He was working in Josh’s lab with bacteriophage P22 and was using EcoRI enzyme to cleave P22 DNA. To examine the DNA fragments generated by cleavage, Sgaramella needed an electron microscope, and there was no electron microscope in the Department of Genetics at the time. But the Biochemistry Department had an E.M. that was being used primarily by Ron [Ronald W.] Davis, who just had been recruited to Stanford as an assistant professor after completing training in Norman Davidson’s lab at Caltech. Ron is an extraordinarily creative scientist, and already had made major contributions in working out heteroduplex techniques to identify regions of similarity in different DNA molecules. Vittorio was given use of the Biochemistry Department’s E.M. The department had generously also allowed me to use that electron microscope to examine R-factor DNA molecules. I had learned how to prepare DNA preparations for examination by electron microscopy from Phil [Phillip A.] Sharp and others in Norman Davidson’s lab at Caltech, and I came back here and used the Biochemistry Department’s electron microscope to look at plasmid DNA structure. My understanding of this part of the history is that while Sgaramella was using the biochemistry department’s electron microscope, he found that P22 DNA fragments generated by EcoRI cleavage joined together end-to-end to form oligomers when E. coli DNA ligase was added. Unlike the T4 ligase, which Sgaramella had previously shown, while in Khorana’s lab, can join blunt-ended DNA molecules, E. coli ligase was known to require complementarity to join DNA ends. This led Sgaramella to conclude that EcoRI generates complementary ends during its cleavage of DNA. Sgaramella also observed that EcoRI-cleaved SV40 DNA, which John Morrow, a graduate student in Berg’s lab, had found is cleaved by the EcoRI enzyme at a single site [Morrow, JF, Berg, P. Cleavage of simian virus 40 DNA at a unique site by a bacterial restriction enzyme. Proc Natl Acad Sci USA. 1972 November; 69 (11): 3365-9.], also forms oligomers. Morrow’s discovery of the ability of EcoRI to cleave SV40 DNA at a single site was in contrast to what had been observed for a Hemophilus influenzae restriction enzyme [Kelly, TJ, Smith, HO. A restriction enzyme from Hemophilus influenzae. II. J of Mol Biol. 1970; 51 (2): 393-409.], which [Kathleen] Danna and [Daniel] Nathans had shown cleaves SV40 into multiple fragments [Danna, K, Nathans, D. Specific cleavage of simian virus 40 DNA by restriction endonuclease of Hemophilus influenzae. Proc Natl Acad Sci USA. 1971 December; 68 (12): 2913-2917.]. After learning of Sgaramella’s results, Janet Mertz, who of course knew also of Morrow’s finding, tested whether the ends generated by EcoRI cleavage of SV40 DNA could join together to regenerate duplex covalently-closed DNA molecules when E. coli ligase was added. The molecules were examined by her and Ron Davis on the E.M., and they found that cleaved SV40 could in fact recircularize. Janet and Ron published a paper showing that EcoRI cleavage generates complementary ends in SV40 [Mertz, JE, Davis, RW. Cleavage of DNA by R 1 restriction endonuclease generates cohesive ends. Proc Natl Acad Sci USA. 1972 November; 69 (11): 3370-4.], in the same issue of the PNAS that reported Vittorio Sgaramella’s conclusion that complementary ends are generated in phage P22 and SV40 DNA by this enzyme. This PNAS issue also contained the report of the sequence of EcoRI-generated DNA ends by Hedgepeth, Goodman, and Boyer. [Hedgpeth, J, Goodman, HM, Boyer, HW. DNA nucleotide sequence restricted by the RI endonuclease. Proc Natl Acad Sci USA. 1972 November; 69 (11): 3448-52.] So who first made the discovery that EcoRI generates complementary DNA ends that can be joined by E. coli ligase? I think that most people credit Mertz and Davis for that finding, even though the papers by Mertz and Davis and Vittorio Sgaramella were published in the same issue of the PNAS. Paul Berg, who credits Mertz and Davis for this discovery, indicates [Paul Berg, Ph.D., "A Stanford Professor's Career in Biochemistry, Science Politics, and the Biotechnology Industry," an oral history conducted in 1997 by Sally Smith Hughes, Regional Oral History Office, The Bancroft Library, University of California, Berkeley, 2000.] that his name was not included as an author of the Mertz and Davis paper because the PNAS does not allow an author’s name to appear on more than one paper in a single issue of the journal, and he already had authored a paper with John Morrow in that same issue. The paper by Sgaramella, which was communicated to the PNAS by Lederberg and was published under his tutelage, cites the similar findings by Mertz and Davis and acknowledges the use of Biochemistry Department facilities and expertise. There is no mention of Sgaramella’s findings in the Mertz and Davis paper, which was communicated to the PNAS by Berg about a week later. But some years ago, Vittorio showed me a statement that he said had been drafted, he thought by Paul, to acknowledge Sgaramella’s priority in the discovery. Vittorio told me that the statement was intended for inclusion in the Mertz and Davis paper. However, the published paper does not contain it. The Sgaramella and the Mertz and Davis papers, along with the Hedgpeth, Goodman, and Boyer paper, were pub At the Honolulu Meeting: Beginning the Collaboration with Boyer. I went to the meeting in Honolulu and saw the sequence data that Joe Hedgepeth and Herb Boyer had obtained for EcoRI cleavage sites. From the six base pair sequence that Herb disclosed, I estimated that the large antibiotic resistance plasmids I was studying, which were about 100,000 nucleotides in length, would be cleaved into perhaps 20 fragments, on average once every 5000 nucleotides. The plasmids would be cut specifically and reproducibly, and each of these fragments would likely contain only a few genes. This would certainly be better than the mechanical shearing methods I had been using for taking plasmids apart, and the number of DNA fragments would probably be low enough to separate them by centrifugation and determine the size. And because the sequences at the ends of the multiple plasmid DNA fragments likely to be generated by EcoRI would be complementary, I thought that individual plasmid DNA fragments in the mixture might join to each other in different combinations. If cleavage left the replication functions of the plasmid intact, the replication region might join to different antibiotic resistance genes in the mix and form DNA circles containing different fragment combinations. Ligase could be added to seal the circles, and we could try to genetically transform calcium chloride-treated E. coli cells with the ligated DNA. Maybe we could isolate cells containing plasmids containing different combinations of antibiotic resistance genes. During an evening walk along the street that parallels Waikiki Beach, Herb and I had a lengthy discussion about the experiments that his lab and mine were doing. The people present were Stanley Falkow, Charles Brinton, a University of Pittsburgh microbiologist who was on sabbatical leave in my lab at the time, and Charlie’s wife, Ginger. Charlie had been working with pili, which are hairlike projections on the surface of bacteria; they’re encoded by plasmid genes and are involved in plasmid transfer. Charlie and I were doing some collaborative experiments at the time at Stanford. His wife Ginger was with him here at Stanford and she came along to Hawaii. So Stanley, Charlie, Herb, Ginger, and I were taking this long walk and chatting. We ended up at a delicatessen and continued to talk, over sandwiches and beer. Herb initially was not very interested in looking at plasmid genes, and offered to provide the enzyme as a gift for the experiments I wanted to do. He said he had given EcoRI to various other people at Stanford, and he’d be willing to give some to me. I said, “Well, that doesn’t seem quite fair. Your lab has spent a lot of time isolating the enzyme and we should really do this as a collaboration.” And that’s the way we decided to do it. Something that’s often missed by people looking at this episode through a “retrospectroscope” is there was no assurance that any of this experimentation would work. We knew even from the Khorana lab’s experiments published several years previously that pieces of DNA could be linked together biochemically: biochemical joining wasn’t revolutionary. And we knew, because we had done it in my lab, that we could genetically transform E. coli with plasmid DNA and could use antibiotic resistance genes to identify cells that acquire the plasmids. We expected from the sequence at the EcoRI cleavage site that this restriction enzyme would cut the DNA of our large plasmids reproducibly into multiple fragments. And we knew at that point that EcoRI generates cohesive DNA ends. So these components were there. But the crucial question was whether biochemically linked DNA fragments could be propagated in living E. coli cells and would function there, and the answer was not known. The joining of fragments at EcoRI sites would be bringing together DNA sequences nonbiologically, whereas transduction and other forms of genetic recombination occurring in cells were biological processes that had evolved in nature. As Falkow said at the time, “If it works, let me know.” [Interview with Stanley Falkow by Charles Weiner, May 26, 1976 and February 26, 1977, MIT Oral History Program.] There are some people who think that once a method of biochemical joining DNA ends was worked out, it was obvious that the chimeric DNA could be cloned. That’s easy to say in retrospect, but in actuality it was not the case—especially for DNA molecules that contain components derived from different biological species. I’ll say more about this a little later. In the first experiments of my collaboration with Boyer, we took a large plasmid, the R6-5 plasmid, which carries multiple resistance genes, and cut it up into pieces using EcoRI. An experimental procedure that became available at the time facilitated experiments, and that was the ability to analyze DNA fragments by agarose gel electrophoresis. [Joe] Sambrook, [Phillip] Sharp, and [William] Sugden had worked this out [Sharp, PA, Sugden, B, Sambrook, J. Detection of two restriction endonuclease activities in Haemophilus parainfluenzae using analytical agarose-ethidium bromide electrophoresis. Biochemistry. 1973; 12 (16): 3055-3063.]. Herb had learned about the procedure from Joe Sambrook. The procedure made it convenient to fractionate and characterize DNA fragments by size. The fragments could be stained by ethidium bromide, which was the same DNA-binding dye that we had been using to separate circular plasmid DNA from chromosomal DNA. The fluorescent dye makes the fragments visible when they’re exposed to UV [ultraviolet] light so the DNA can be seen in the gels. The cleaved DNA was electophoresed on a gel and stained. We saw that EcoRI had cut the plasmid into fragments of defined sizes; we could see eleven of them, rather than the 20 that I had estimated. We introduced the EcoRI cleaved plasmid DNA into E. coli, both with and without ligation and isolated bacteria expressing resistance to different combinations of antibiotics. We wanted to see whether we could get reconstituted plasmids that express only some of the resistance genes present on R6-5 and include only some of the eleven DNA fragments. One goal of the work was to identify the fragment that contains the replication machinery of R6-5, so we also wanted to determine whether any DNA fragment was common to all of the plasmids. We also wanted to see whether certain fragments were correlated with specific resistance traits. And, in fact, we did see new combinations of resistance genes and DNA fragments. But, we also saw new combinations of EcoRI-generated fragments in bacteria receiving DNA that hadn’t been treated with ligase. This showed that single-strand nicks in the DNA can be sealed by ligation in vivo after introduction of the DNA into bacteria.
• Hughes: Now were you an original observer of the phenomenon?
• Cohen: Which phenomenon?
• Hughes: That when plasmid fragments were introduced into a cell the ligation occurred naturally.
• Cohen: Yes. That was an unexpected observation. We initially had thought that it would be necessary to ligate the DNA molecules in vitro. DNA splicing occurred in vivo at a lower efficiency, but it occurred. This [finding] became important a few years later in some of the biohazard controversy issues. By transforming cells with a mixture containing all of the EcoRI-generated fragments of R6-5 and selecting for kanamycin resistance, we recovered a smaller kanamycin resistance plasmid, and when we isolated that plasmid and cleaved it with EcoRI, we found that it contained several fragments identical in size to the EcoRI-generated fragments generated by cleavage of[ ?]
• Hughes: Now, is that pSC101?
• Cohen: No, not yet. pSC101 was not yet part of the experiments. I need to explain here that pSC101, which we had isolated earlier after mechanical shearing of R6-5 DNA, [Cohen, SN, Chang, ACY. Recircularization and autonomous replication of a sheared R-factor DNA segment in Escherichia coli transformants. Proc Natl Acad Sci USA. 1973; 70: 1293-1297.] didn’t actually originate from R6-5, as we had originally thought. We showed, a couple of years later, that it was a separate Salmonella panama plasmid that had contaminated the transformation mix [Cohen, SN, Chang, ACY. Revised interpretation of the origin of the pSC101 plasmid. J. Bacteriol. 1977; 132: 734-737.]. We had originally named the plasmid “Tc6-5,” and changed it to “pSC101,” in keeping with impending recommendations about plasmid nomenclature [Novick, RP, Clowes, RC, Cohen, SN, Curtiss, R, Datta, N, Falkow, S. Uniform nomenclature for bacterial plasmids: A proposal. Bacteriological Reviews. 1976; 40: 168-189.]. The heteroduplex investigations that Sharp, Davidson and I had done suggested that large R factor plasmids may have been formed in nature by the addition of antibiotic resistance genes to regions for replication and transfer. The kanamycin resistance plasmid [named pSC102] we recovered in our initial DNA cloning experiments included three fragments from R6-5. We wanted to learn which one of these contained the kanamycin resistance gene, and one way to find out was to try to clone the resistance gene fragment. To do this, we wanted a small plasmid replicon carrying a resistance gene that would allow us to select cells that contain it, but which did not express kanamycin resistance—and we wanted it to be cleaved only once by EcoRI. We tested several plasmids and found that pSC101 had these properties. We cleaved both pSC101 DNA and pSC102 DNA with EcoRI, mixed the two DNAs together, and added the mixture to calcium chloride-treated E. coli cells. By including tetracycline in the growth medium, we could select cells that harbor the pSC101 backbone, and we than tested these cells for resistance to kanamycin to identify bacterial clones that expressed both resistance genes. We isolated plasmids from these bacteria and found one, pSC105, that included the pSC101 DNA fragment plus one of the three fragments of pSC102. pSC102 had been formed by the rearrangement in vitro of fragments of the same DNA molecule, the R6-5 plasmid, but pSC105 was the first DNA ever to be propagated that contained fragments of different DNA molecules that had been joined together outside of cells. We later found how close the EcoRI cleavage site is to a location where insertion of another DNA fragment would have prevented the experiment from being successful. The promoter for the tetracycline resistance gene on pSC101 begins just 35 or 40 base pairs away from this cleavage site. If cleavage had occurred in the gene or its promoter, the selection for tetracycline resistance wouldn’t have worked. Our backup plan was to use pSC102 to clone other fragments of R6-5, but it was very convenient to have a small tetracycline-resistance replicon that worked as a vector. We then joined pSC101 to a plasmid we had obtained from Stanley Falkow, RSF1010, which we found also contained a single EcoRI cleavage site, making a two-replicon plasmid that we showed could be propagated in E. coli. We published these experiments in the first of the three PNAS papers [Cohen, SN, Chang, ACY, Boyer, HW, Helling, RB. Construction of biologically functional plasmids in vitro. Proc Natl Acad Sci USA. 1973; 70: 3240-3244.] that reported the cloning of DNA from different sources.
• Hughes: There must have been a moment when you realized that the experiment had worked. What was that moment?
• Cohen: That moment was elation. [Laughter.]
• Hughes: Based on which particular part of the experiment?
• Cohen: Actually, it was a series of moments. The first was when we found that the large R6-5 plasmid was cut into multiple fragments. Second was when we found that the plasmid fragments could be propagated in tranformants, and that different bacterial cell clones expressed different combinations of antibiotic resistance. Third was when we actually analyzed the plasmid DNA from these cells and found that the different cells contained different plasmids, and that the DNA fragments were all derived from the R6-5 parent. Fourth was when we found that pSC101 was cleaved only once by the enzyme. The most crucial moment was, I suppose, when we linked the kanamycin resistance fragment of pSC102 to pSC101, which showed that pSC101 would actually work as a carrier to propagate another DNA fragment in bacteria. The fact that we could take a non-replicating piece of DNA, link it to a plasmid vector, and replicate it in bacteria was especially exciting. So there were a series of exciting discoveries. I suppose that’s the best way of putting it.
• Hughes: All right, the experiment worked. Then where did your thinking go?
• Cohen: Well, in part my thinking went to, “Hey, now I can study plasmids in the way that I’ve wanted to study plasmids.” That was the original motivation for doing these experiments. [Interruption to find reprint of the paper.] But in the summary of our paper reporting these experiments we also said, “The general procedure described here is potentially useful for insertion of specific sequences from prokaryotic or eukaryotic chromosomes or extrachromosomal DNA into independently replicating bacterial plasmids. The antibiotic resistance plasmid pSC101 constitutes a replicon of considerable potential usefulness for the selection of such constructed molecules, since its replication machinery and its tetracycline resistance gene are left intact after cleavage by the EcoRI endonuclease.” So yes, at the time, we certainly realized the potential utility for using this method to clone DNA from other sources. But we were really quite cautious about what we said in the paper because the generality of what we had done was not yet determined. Yes, the restriction modification systems of E. coli did not destroy the new E. coli plasmid constructs we had made, but we had no idea what would happen if we tried taking DNA from another species and putting it into E. coli.
• Hughes: Is this the time to talk about who was doing what? The work was going on in your lab and Boyer’s lab, and you were obviously doing different things.
• Cohen: Well, that’s right. At that time, Annie Chang, who was a research technician in my lab, lived in San Francisco. And that was very convenient because she was able to be a courier between Boyer’s lab and mine in addition to doing many of the day-to-day experiments. She had been a co-author on the transformation experiments with Leslie Hsu. In the collaboration with Herb, the plasmids were isolated and the DNA was purified in my lab. The DNA was then taken up to UCSF by Annie where Herb and/or Bob Helling, who was on sabbatical leave in his lab, cut it with EcoRI and did the ligation, and then sent the DNA back to us. We did the transformation and selection for the cells that expressed the resistance phenotype in my lab; we isolated the plasmid DNA from these cells and it was characterized in various ways. The characterization by electron microscopy and centrifugation was done here. The characterization by gel analysis was done largely by Bob Helling in Boyer’s lab. So it was really a project in which both labs were contributing very substantively towards the experiments. There were skills from both of our labs that were important.
• Hughes: Was this research a central focus in each lab or was it just one of several projects?
• Cohen: It was one of several projects going on in each lab at the same time. My lab was also studying the role of tranposons in plasmid evolution. And, we were studying plasmids in other ways as well. Annie was the one person in my lab who was working on the DNA cloning project. Other people in the lab were working on other projects. I felt I could afford to put a research technician on a project that had a high risk of not leading anywhere. If it wasn’t successful, okay, well that’s the way it goes. The postdocs who needed papers to get them faculty jobs had projects that I felt had a higher likelihood of producing publishable results; but Annie had a permanent job in the lab and if the project didn’t work out, she wasn’t at risk. And there were also a number of projects going on in Herb’s lab at the time. You’ve talked with Mary Betlach and others about these.
• Hughes: Yes. [Oral histories with Mary Betlach and with Axel Ullrich are available online in this Bancroft Library series..] Did Boyer consider it a risky experiment as well?
• Cohen: As far as I know, he did. From the discussion that we had in Hawaii, he also felt that these experiments had exciting potential but, again, it was very uncertain whether they would work.
• Hughes: Over what period of time did these experiments go on?
• Cohen: We began the work just shortly after the New Year in 1973, and by early March, two months later, we had shown the basic feasibility of the method. So the experiments went very quickly.
• Hughes: Were there any surprises?
• Cohen: Well, one of big surprises was that normally non-replicating fragments of DNA could actually be propagated in this way.
• Hughes: But other than that?
• Cohen: I was also surprised to find that non-ligated fragments could be joined together in vivo after they were taken up by cells. But other than that, the experiments were basically planned out at the beginning of the collaboration and we were happy to find that they progressed according to plan. As you know, even experiments that are carefully planned can run into unforeseen obstacles, but these just worked out extremely well. Bob Helling and Herb had the EcoRI cleavage conditions worked out and the agarose gel technique going; I had the electron microscope heteroduplex methods, transformation methods, and other plasmid procedures worked out. The experiments proceeded quickly, and it was an extremely exciting time. We often wished that the bacteria would grow faster so that we could get the result of an experiment sooner. We came into the lab each morning to look at the [culture] plates. We would hurry to isolate the plasmid DNA and Annie would carry some of it up to Herb’s lab where it would be analyzed by agarose gel electrophoresis. At the same time, we’d look at it here by centrifugation and by EM. It was really a continual high.
• Hughes: Was there any thought at this point that there might be industrial applications? It’s certainly not
• in the paper, but was it in your thinking?
• Cohen: It’s hard to think back and pinpoint the moment that this thought first occurred. To me, it probably was not before the initial positive results, but certainly very soon after that. But industrial applications were dependent on the ability to clone DNA from other species, and as I’ve said, we hadn’t shown that yet. That’s why we didn’t go beyond the conservative statement we made in the discussion: “...potentially useful for insertion of specific sequences from prokaryotic or eukaryotic chromosomes....” But even if it turned out that animal cell genes couldn’t be propagated in bacteria, the point of this statement was that it might be possible to take genes that were native to other bacterial species and introduce them into more-easily grown E. coli.
• Hughes: The idea was in the wind. I haven’t read Peter Lobban’s thesis. Apparently, he made specific reference to the fact that this procedure might have industrial applications.
• Cohen: I read Peter’s thesis dissertation when I wrote my Scientific American article in 1975, but don’t remember any mention of that. Jensen et al., the industrial group I mentioned earlier, probably were thinking in terms of such applications in undertaking the work they published in 1971, since they worked for a company. But in starting the initial gene cloning experiments, we weren’t thinking, or at least I wasn’t, about putting animal cell DNA into bacteria. And when I did start thinking about this, most colleagues I discussed it with thought that it was unlikely that animal cell DNA could survive the restriction systems of prokaryotes. There’s another point that I should make about the cloning of animal cell DNA. The problem of how to select bacteria that contain foreign DNA fragments was, at that time a formidable one. In our initial experiments, it was possible to identify the bacteria containing recombinant plasmids because the fragments that we joined to the pSC101 vector included a resistance gene that we could test for, or select for. I knew that even if eukaryotic DNA could in fact be propagated in bacteria, it would be necessary to work out ways to identify the bacteria that acquired recombinant plasmids, versus those bacterial clones that taken up the vector only.