How Escherichia coli Became the Flagship Bacterium of Molecular Biology

ABSTRACT

Escherichia coli is likely the most studied organism and was instrumental in developing many fundamental concepts in biology. But why E. coli? In the 1940s, E. coli was well suited for the biochemical and genetic research that blended to become the seminal field of biochemical genetics and led to the realization that processes already known to occur in complex organisms were conserved in bacteria. This now-obvious concept, combined with the advantages offered by its easy cultivation, ultimately drove many researchers to shift from the complexity of eukaryotic models to the simpler bacterial system, which eventually led to the development of molecular biology. As knowledge and experimental tools amassed, a positive-feedback loop fixed the central role of E. coli in research. However, given the vast diversity among bacteria and even among E. coli strains, it was by many fortuitous events that E. coli rose to the top as an experimental model. Here, we share how serendipity and its own biology selected E. coli as the flagship bacterium of molecular biology.

INTRODUCTION

As described in detail by Friedmann (1), Theodor Escherich was a prominent 19th century physician-scientist who brought Robert Koch’s methods to pediatrics and, as one of the first to probe the microbiome, founded the field of intestinal pathology. Despite the contemporary international acclaim that he enjoyed, Escherich is probably best remembered today for his discovery of a bacterium he called Bacterium coli commune. As described in his first book, which he published at the age of 29 (2), this bacterium was found in the colon of milk-fed newborns at much higher levels than in adults. By 1894, B. coli commune had already caught the attention of researchers because it was considered a common inhabitant of the gut of humans and many domestic animals and also had the potential of causing disease when present at other body sites (3). We now know that Escherichia coli is not very abundant in the human gut but was readily isolated at the time because most other members of the gut microbiome are not easily culturable. Indeed, it was noted early on that B. coli commune is an exceedingly easy organism to grow under different conditions, has a short generation time, is prototrophic and motile, and can grow on lactose, properties that were soon noticed by enzymologists and physiologists. Moreover, the bacterium first called B. coli mutabile by Rudolph Massini, which was a Lac⁻ strain that throws off Lac⁺ revertants, provided the first contribution from bacteriology to the theory of mutation (4). Owing to its increasing popularity among bacteriologists, Castellani and Chalmers in the 1919 third edition of their Manual of Tropical Diseases (5) suggested that the name be changed to Escherichia coli. Friedmann (1) notes that E. coli is featured prominently in all three editions of the standard textbook for generations of bacteriologists, Bacterial Metabolism, although the name E. coli does not replace B. coli until editions published in 1939 and 1949 (6). As a side note, in 1928, the author of those texts, Marjorie Stephenson, was the first to isolate a bacterial enzyme (lactate dehydrogenase from E. coli, of course) and one of the first two women, along with crystallographer Kathleen Lonsdale, elected a Fellow of the Royal Society (7, 8). We refer to Stephenson’s textbook again below.

One of the protagonists of our story, E. coli K-12, was isolated from the stool of a recovering diphtheria patient in 1922 and was deposited in the strain collection of the Department of Bacteriology and Experimental Pathology at Stanford University in 1925. It was used there for many years in the teaching laboratories of medical bacteriology and bacterial physiology courses since it fulfilled all of the standard identification tests for an E. coli strain (9). The designation “K,” for capsular antigen (Kapsel in German), is often used for serotyping E. coli strains, but that is not true for K-12. There really is a K12 capsule antigen (10), but K-12 strains produce a colanic acid exopolysaccharide instead of a K capsule (11). Unlike K capsules, colanic acid is not associated with pathogenesis but is thought to be beneficial in many environments outside the host (12). The reason why it was called K-12 has apparently been lost to history (13). Curiously, we found that E. coli K-13 and the Flexner bacillus I-13 (now Shigella flexneri) were also used by members of the Stanford University Department of Bacteriology and Experimental Pathology in the 1920s and 1930s, suggesting that the K-12 designation might result from cataloging strains (14, 15).

One of the early instructors in the medical bacteriology course at Stanford was Charles E. Clifton, who was interested in bacterial growth and physiology. One of Clifton’s early studies explicitly stated that E. coli was used because of the “more extensive information available on its metabolism and the ease with which contaminants may be detected” (16). Indeed, in the 1930s, E. coli was used in many physiology and biochemical studies presumably because it grows easily and has diverse metabolic capacities. Although we do not know how many E. coli strains were available in the Stanford collection, Clifton used E. coli K-12 for most of his studies, including those that linked bacterial metabolism, growth, and nutrients, which ultimately led to the conclusion that the availability of nutrients controls the growth rate and the amount of growth in a population (16, 17). Whatever Clifton’s reason was for selecting E. coli K-12, as early as 1933, it turned out to be crucial for the development of bacterial genetics and molecular biology, as we describe below (18).

It is commonly, and rightfully, noted that the landmark paper of Salvador Luria and Max Delbrück (19) signaled the dawn of molecular biology. It showed that Darwin’s theory of natural selection following random mutation applied even to bacteria. Although Luria and Delbrück used E. coli, it was not E. coli K-12 but rather another strain, which Max Delbrück christened B, for bacterium (20). Delbrück was introduced to bacteriophages by Emory Ellis and his wife Marion. They hoped that bacteriophages might be simple model systems that would provide insights into the basic properties of viruses such as the Rous sarcoma virus. They were using E. coli, because it grew well and was available, and phages isolated from the sewers of Pasadena, California. Delbrück was impressed by the stepwise growth curves showing that the phage multiplied in the bacterium and not in the solution (21).

We think that E. coli B is the second most important E. coli strain in research. For example, it was the strain used by Ellis Englesberg in his studies of the arabinose operon that established positive control by transcription activators (22), by Bill Studier and Barbara Moffatt for their T7 expression systems for the production of proteins that generated the widely used BL-21 strain (23), and by Richard Lenski in his long-term evolution experiments (24). E. coli B and its radiation-resistant derivative E. coli B/r (referred to as “B bar r”), isolated by Evelyn Witkin (25), were the preferred strains of many physiologists and have been the subject of many comprehensive studies of growth and metabolism (26).

Like many other bacteriophage researchers, Delbrück viewed E. coli simply as a growth medium for phage. The ability to grow easily and in chemically defined media were desirable qualities of his phage hosts (27). E. coli fit the bill. Luria and Delbrück spent summers at the Cold Spring Harbor Laboratory, where in 1944 Delbrück cemented the role of E. coli in research by convincing phage workers that to make real progress they had to concentrate on a set of seven phages (T1 to T7) that infected E. coli B. The following year, Delbrück established the famous phage course that eventually morphed into the Advanced Bacterial Genetics course, which is still being offered each summer. Many who took this course, including Seymour Benzer, went on to make significant contributions, often using E. coli (20). In 1950, Al Hershey accepted a position at the Cold Spring Harbor Laboratory, uniting the three future Nobel laureates and firmly establishing this location as the home of the Phage Group. One of the seminal studies from the Phage Group at the Cold Spring Harbor Laboratory was that of Al Hershey and Martha Chase (28), who labeled phage proteins of phage T2 with ³⁵S and DNA with ³²P and showed that only the DNA was injected into infected E. coli cells. This confirmed the results of Avery, MacLeod, and McCarthy (29) and established that hereditary information is carried in the DNA. This experiment is sometimes called “the Waring blender experiment” because Hershey and Chase used this kitchen appliance to remove empty phage heads from the infected cells. Another famous Waring blender experiment is presented below.

While the Phage Group was popularizing the use of E. coli in studies on the nature of nucleic acids and mutagenesis, E. coli attracted other key players who demonstrated that cells sense and adapt to environmental conditions. Two major characters in this story were Jacques Monod and André Lwoff, who would later share the 1965 Nobel Prize in Physiology or Medicine with François Jacob for their work on gene regulation. Monod was interested in studying growth, and Lwoff convinced him that, rather than focusing on the growth of the Tetrahymena protozoa, he should study bacterial growth, mentioning E. coli specifically; to help make his case, he gave Monod a copy of the aforementioned textbook by Marjorie Stephenson and a few other reprints and assured him that it was not pathogenic (30). Despite Monod’s considerable contribution to the French resistance (30, 31), he published his thesis during the darkest period of World War II, in which he introduced the phenomenon of diauxie (32). His thesis describes work with Bacillus subtilis but, following Lwoff’s advice, he soon turned to E. coli. As described in textbooks (33) and monographs (20), if E. coli is given both glucose and lactose, it first consumes the glucose, and then, after a lag in which β-galactosidase is induced, it resumes growth utilizing the lactose. This diauxic growth pattern focused Monod’s attention on the mechanism of enzyme induction. However, Monod’s early work on bacterial physiology was done not with E. coli K-12 but rather with strains such as ML30, which, for example, was used in the work that discovered and characterized the lactose permease (34). The original ML strain was isolated from human intestines (35); rumor has it that the ML designation stands for “merde Lwoff” (36)!

The studies from the Phage Group, Lwoff, Monod, and Jacob did not employ E. coli K-12 or even the same strains but shared the desire to understand bacterial and phage physiology using mutants. By chance, Stanford University launched E. coli K-12 into prominence by connecting Clifton and Tatum through the convergence of bacterial physiology and genetics. That story begins in the Biology Department of Stanford with the one gene-one enzyme hypothesis of George Beadle and Edward Tatum using the fungus Neurospora (37). Our experience is that students today are so far removed from this work that they cannot appreciate its significance. We now know that genes can specify multifunctional enzymes! What they do not appreciate is that, at the time this work was done, it was not at all clear what genes actually did. Beadle and Tatum showed that genes specify proteins. Looking back, they should have said one gene-one protein.

In their study, Beadle and Tatum developed methods to isolate mutants of Neurospora that could grow on rich medium but not on synthetic minimal medium (37). Initially, they isolated three auxotrophs that required vitamin B₆, vitamin B₁, and p-aminobenzoic acid. In subsequent work, they isolated more mutants that required many different vitamins or amino acids. In some cases, it could be shown that the mutants lacked the ability to carry out a specific reaction; in all cases, they could show genetically that a single gene was affected (38). These studies launched the field of biochemical genetics (38). Metabolic pathways could be analyzed by isolating mutants and then adding back pathway intermediates to restore growth through chemical complementation. Biochemical genetics was crucial in the rapid development of both bacterial genetics and physiology in the 1940s to 1950s.

Because isolating Neurospora mutants was cumbersome, and to demonstrate similar findings in another organism, Tatum decided to look for bacterial auxotrophs generated by X-ray treatment using the well-behaved E. coli K-12 strain from the Stanford collection (39, 40). At the time, it was unclear whether bacteria contained genes like those already described in eukaryotes; the fact that Tatum was able to isolate double mutants that required two different growth factors allowed him to conclude that changes in nutritional requirements originated from single mutant parents and therefore were heritable, analogous to gene mutations (40). However, since there was no way to analyze these mutants genetically, he had no way to prove that.

The genetic analysis of mutants derived from phage and biochemical genetics supported the existence of genes, but it was unclear whether bacterial genes were analogous to the nuclear genes that followed Mendelian inheritance, as studies were limited to clonal strains. Avery, MacLeod, and McCarthy had demonstrated that DNA was the transforming principle in bacteria, but it was achievable only under specific conditions and was limited to capsular phenotype in streptococci (29). Therefore, it was recognized that showing recombination of traits between different strains would be important for characterizing the nature of genes and mutations in bacteria and expanding the genetic methods that were already available in eukaryotic systems (39). This feat was finally accomplished by Lederberg and Tatum in 1946 (41).

As described by Brock (20), in 1944 Joshua Lederberg, a medical student at Columbia University, conceived of a way to test for bacterial mating. He would mix two different auxotrophs and then plate on minimal medium to identify prototrophic recombinants. Since he plated billions of cells, he could detect even rare events. From today’s viewpoint, this seems rather simple, but he was the first to propose using conditional mutants in such a selection. Lederberg was using strains of E. coli, but these experiments failed because of the high frequency at which a single auxotrophic mutation reverted. In 1945, Tatum moved from Stanford University to Yale University and, realizing that double mutants would solve the reversion problem, Lederberg contacted Tatum and a collaboration was arranged. From Yale, Tatum wrote to Clifton asking him to send a number of strains from the Stanford collection (42). Naturally, Clifton sent him E. coli K-12 since it was the strain that he mostly used in his physiology studies. Unbeknown to everyone, including this strain in the package was critical for the success of the Lederberg and Tatum experiments.

In their Nature paper (41) (which is one-half page long!) and later in a more detailed paper in the Journal of Bacteriology (43), Lederberg and Tatum describe mixing two triple mutants, one, Y-10, requiring threonine, leucine, and thiamine and the other, Y-24, requiring biotin, phenylalanine, and cystine. In pure cultures, the triple mutants could occasionally give rise to revertants of one of the auxotrophies. However, if they mixed the two triple mutants, they could recover wild-type and double mutant strains. Cleverly, they also obtained T1 phage from Luria and Delbrück (19) and used it to isolate a resistant derivative of one of the triple mutants, thus creating a marker that they could screen for without selecting it in their matings. Using this T1-resistant strain, Lederberg and Tatum could show that some of the wild-type recombinants they isolated were T1 resistant while others were T1 sensitive. As they stated, “These experiments imply the occurrence of a sexual process in the bacterium Escherichia coli.” Although they did not understand the mechanism of conjugation, they recognized that being able to obtain different mutants suggested “very strongly that hybridization and segregation take place” (43). They even suggested that the nonrandom segregation could be the result of linkage between the markers and proposed that the method could be generalized and used with other selectable markers such as resistance to antibiotics.

Lederberg and Tatum reported that recombinants were rare, but the fact that they occurred at all is remarkable. They noted that “washed cells were inoculated heavily into synthetic agar medium” (41). Thus, mating pairs that formed were left undisturbed, allowing time for the transfer of the entire chromosome (44). In addition, we now know that the F (fertility factor) episome, or plasmid, present in E. coli K-12 contains an insertion sequence (IS) element insertion in a regulatory gene that derepresses conjugation functions (45). If the strains contained a wild-type version of F, it is not at all clear that their cleverly designed experiments would have worked. Moreover, they were extremely lucky that one of the triple mutant strains, Y-10, had lost the F plasmid during construction, so that mating could occur with the F⁺ strain Y-24 (46).

There is another fact about E. coli K-12 that made this strain so special, and it requires a brief diversion. As shown by Esther Lederberg when she was a graduate student, K-12 is lysogenic for a bacteriophage that she christened λ (47). This phage has had major impacts on molecular biology, some of which are noted below. The discovery of this phage is just one of Esther’s many accomplishments. As a woman in a male-dominated field, living in the shadow of her husband and Nobel laureate Joshua Lederberg, Esther Lederberg struggled for professional recognition. Her foundational discoveries in the field of microbiology were often overlooked (48). They impacted the work of many scientists, including Benzer’s studies on the so-called T-even phages (T2, T4, and T6). These phages make small plaques when plated on E. coli B. Hershey identified mutants that make large plaques, which were designated “r” for rapid lysis (49). It turns out that mutations in several genes can confer this phenotype. Focusing on T4, Benzer discovered that T4 rII mutants do not form plaques on E. coli K-12 strains that are λ lysogens but they do form plaques on K-12 strains that have been cured of λ, apparently reflecting the ongoing biological warfare between these two phages (50). Thus, Benzer could identify rII mutants using E. coli B and then map them genetically by selecting wild-type recombinants on E. coli K-12. Benzer exploited this system “to run the genetic map into the ground” (51, 52). Indeed, Benzer opened the door for detailed analysis of mutagenesis and the elucidation of gene structure and the genetic code (e.g., reference 53).

Let’s return to the famous mating experiments of Lederberg and Tatum. Since both triple mutants used (Y-10 and Y-24) were E. coli K-12, Lederberg and Tatum, and probably most other scientists as well, assumed that this bacterium was homothallic (i.e., two identical cells not differing in mating type). They also assumed that the two strains fused to form a diploid zygote, which would then resolve following recombination, perhaps very quickly, into a haploid daughter cell. A simple but beautiful set of experiments by William Hayes destroyed this view. As described in another short Nature paper (54), Hayes was interested in the kinetics of conjugation, so he planned to add streptomycin at various intervals to interrupt the mating process. He used two auxotrophs, which we call A and B. A was a double mutant, and B was a triple mutant. He isolated streptomycin-resistant variants of both A and B to use in reciprocal mating experiments. He discovered that, when A was streptomycin sensitive and B was resistant, the drug had little effect on the mating process regardless of the time at which the drug was added. In contrast, when A was streptomycin resistant and B was sensitive, no recombinants were obtained. Clearly, E. coli K-12 is not homothallic and mating was unidirectional. He concluded that A can serve as a donor even in the presence of the drug but B accepts genes and incorporates them and thus must survive. Subsequently, Hayes (55), with the help of Luca Cavalli and the Lederbergs (46), showed that strain A carries what they called the fertility factor (F⁺). Strain B lost this factor (F⁻) and consequently its ability to function as a donor, so it became a more efficient recipient.

As evidenced by the experiments described above, wild-type E. coli K-12 (F⁺) transfers chromosomal markers at very low frequencies. Luca Cavalli (56) described a strain that transfers chromosomal markers at high frequency, and a similar strain was later isolated by William Hayes (55); these strains were termed Hfr for high frequency of recombination, and the two strains are eponymously named HfrC and HfrH, respectively. HfrH, in particular, opened the door for a genetic analysis of conjugation when Francois Jacob and Elie Wollman obtained it from Hayes. They too were interested in the kinetics of mating, and they reasoned that the Waring blender, mentioned previously in the Hershey-Chase experiment (28), would likely disrupt the mating process. They discovered that HfrH transferred genetic markers in a linear fashion from a fixed starting place (57). Markers close to the “origin” were transferred at high frequency, while markers distant from the origin were transferred at ever decreasing rates, a phenomenon they called the gradient of transmission.

Jacob and Wollman (57) also discovered that a mutational event converted an F⁺ strain to an Hfr strain, a fact they proved using the fluctuation analysis of Luria and Delbrück (19). With the different Hfr strains that they isolated, they could show that each of them transferred different markers early, all markers are linked in a circularly permuted arrangement, and, accordingly, the resulting map places genes on a circular chromosome. The circular arrangement was so unexpected and controversial that Jacob and Wollman state, “It seems unnecessary to emphasize that this diagrammatic representation, which is the simplest one that will account for the observed results at this time, is not meant to imply that the bacterial chromosome is actually circular” (57). But it is, as was shown directly by autoradiography by John Cairns when he visualized the circular chromosome of E. coli K-12 undergoing replication 5 years later (58).

Using their interrupted mating techniques, Jacob and Wollman mapped the location of the λ prophage in the E. coli K-12 chromosome, and what they saw was again surprising (59). If the Hfr strain carries the λ prophage, then lytic induction occurs immediately after the phage DNA enters the recipient. However, if the λ prophage is in the F⁻ recipient, then nothing happens when the corresponding region of the chromosome is transferred. In other words, when prophage DNA enters the cytoplasm of a nonlysogen, induction occurs. They termed this phenomenon zygotic induction. As we describe below, we now know that the cI repressor encoded by the prophage, which is absent in the recipient, prevents induction in the donor strain (60–62). Through these experiments and those we describe next, E. coli K-12 showed the world the concept of transcriptional control and, in a broader sense, demonstrated that the central dogma is dynamic and subject to regulation.

By the late 1950s, three different kinds of lac mutants were known, and all mutations were tightly linked on the chromosome, i.e., mutants that lacked β-galactosidase (lacZ⁻), mutants that lacked lactose permease (lacY⁻), and mutants that produced LacZ and LacY constitutively (lacI⁻ [for inducibility]). It was also known that the production of LacZ and LacY was induced by the presence of lactose (inducer), which, as mentioned previously, intrigued Jacques Monod. Together with Arthur Pardee and François Jacob, Monod realized that insights into the mechanism of enzyme induction could be obtained using diploid (i.e., dominance) analysis, which they described in their enormously influential paper, often called the PaJaMo paper or experiment (63).

At the time, there was no way to make stable diploids, so Pardee et al. (63) used Hfr crosses to make unstable, transient diploids. Because the various lac mutations were so tightly linked, wild-type recombinants were very rare and could not affect the results. These experiments established that lacZ and lacI were separate genes. Using the Waring blender, the authors could show that lacZ expression began as soon as a functional gene was transferred into a lacZ⁻ recipient. They also discovered that the lacI⁻ mutation, causing constitutive lacZ expression, was a recessive mutation. In other words, Lac was a system under negative control. The inducer inactivates a repressor, which turns genes off. The striking similarity between zygotic induction and the PaJaMo experiment did not escape the authors, who proposed that, when either a λ prophage or a lac operon was injected into the cytoplasm of a strain that lacked a repressor, induction occurred. Indeed, as stated by Alexander Gann on the occasion of the 50th anniversary of the completion of these experiments (31), “Many observations reported since in studies of gene regulation and developmental biology are in essence re-runs of their experiments, in different ways and in a variety of systems.”

Jacob and Monod were convinced that they had discovered a very general regulatory mechanism (61). The PaJaMo experiment also launched the search for mRNA. Since in eukaryotic cells the DNA is in the nucleus and protein synthesis occurs on ribosomes in the cytoplasm, it was proposed that a messenger, likely RNA, was the intermediate in the transfer of information from the DNA to protein. It was also known that ribosomes contain RNA, so it was further assumed that each ribosome carried the information to make one protein. The fact that β-galactosidase synthesis commenced as soon as the lacZ gene entered the cytoplasm and was shut off as soon as the repressor accumulated did serious damage to this hypothesis (20, 36).

Together, the studies we have described placed E. coli at the center of molecular biology. Much of the central dogma was eventually deciphered using E. coli, logic, and ingenuity. Francis Crick and colleagues used E. coli K-12 and B strains to elucidate the genetic code (53). Matthew Meselson and Franklin Stahl defined semiconservative replication with E. coli B (64, 65). As more knowledge in physiology and molecular biology accumulated and more genetic tools were developed, it was only natural that researchers would continue to work in E. coli. However, there were other fortunate qualities of this bacterium, and of K-12 strains, that also helped maintain E. coli as a model organism. For example, as research on the composition of the bacterial envelope intensified in the 1960s, it became clear that it was critical to separate the different layers and compartments in the cell. In 1968, Takashi Miura and Shoji Mizushima described the separation of the inner and outer membranes from E. coli (strain K-12) using density gradient separation, a technique that does not work for some other Gram-negative bacteria (66, 67). After Norton Zinder and Joshua Lederberg first described generalized phage transduction in Salmonella (68), Ed Lennox demonstrated it in E. coli K-12 using the P1 phage (69). By demonstrating cotransduction of markers, Lennox showed that genetic mapping using P1 transduction was possible. Since Lennox was a student in Luria’s laboratory at the University of Illinois, it was probably natural that he used E. coli K-12. What he did not know was that E. coli K-12 carries an IS element in the wbbL gene that makes it unable to produce O antigen. It was later shown that a wbbL⁺ E. coli K-12 strain is resistant to P1 and even difficult to transform (70).

Qualities like those described above and its full genome sequencing in 1997 (71) surely ensured that E. coli K-12 remained the premier model organism for decades. However, through genomics, we have learned that there is not only an incredible diversity among bacteria but also a great deal of diversity even among E. coli strains, which was first noted by Escherich through phenotypic characterization (72, 73). Although some of what is true in E. coli K-12 might not be applicable to other bacteria or E. coli strains, especially given the extensive laboratory cultivation and even radiation that K-12 strains have undergone (74), this flagship bacterium revolutionized our understanding of biology, at incredible speed, because of its metabolic versatility, easy cultivation, amenability to genetic manipulation, and serendipity. Indeed, as Monod foresaw, E. coli, even its imperfect K-12 strains, has taught us a lot about elephants (1).

Biographies

Natividad Ruiz received a B.A. in Microbiology and Chemistry from the University of Kansas in 1993 and a Ph.D. from Washington University in St. Louis, Missouri, in 1998 for her thesis work with Dr. Michael Caparon on Streptococcus pyogenesis pathogenesis. She was a postdoctoral fellow and research scientist in Professor Thomas J. Silhavy’s laboratory at Princeton University before joining the faculty of the Ohio State University in 2010, where she is a Professor of Microbiology. The Ruiz laboratory studies the biogenesis of the cell envelope of Gram-negative bacteria. Using E. coli as a model, their research focuses mainly on the transport of envelope components across cellular compartments.

Thomas J. Silhavy received his B.S. in Pharmacy (summa cum laude, 1971) from Ferris State College and his M.S. (1974) and Ph.D. (1975) in Biological Chemistry from Harvard University. As a graduate student with Winfried Boos, he helped characterize the role of periplasmic binding proteins in sugar transport. As a postdoctoral fellow with Jonathan Beckwith at Harvard Medical School, he helped establish gene fusions as an experimental tool. He came to Princeton University in 1984 as a founding member of the Department of Molecular Biology and is currently the Warner-Lambert Parke-Davis Professor of Molecular Biology. He is best known for his work on protein secretion, membrane biogenesis, and signal transduction.