Chapter 78

The Operon: an Historical Account^†

JON BECKWITH

INTRODUCTION

One of the landmark events in the history of molecular biology was the proposal of the operon model of genetic regulation. The combination of brilliant intuition and experimentation of Jacob, Monod, and their collaborators produced a theory in 1961 which had enormous impact on the directions that molecular biology took in the subsequent decades. As with the elaboration of any paradigm in science, subsequent research has revealed complexities to gene regulation never imagined in the early 1960s. Further, many of the apparent rules for operon control outlined in those early days have since been violated by the detailed analysis of numerous systems of gene control. As the discovery that reverse transcriptase can copy RNA into DNA did not reduce the importance of the central dogma of gene expression, the newer findings on operon complexity do not vitiate the model itself. However, I believe that it is an instructive exercise in the history of science to follow the development of the operon model and the ways in which our concept of the operon has changed over the years. This analysis reveals once again both the generative importance of paradigms in science and the ways in which paradigms constrain the thinking of those working under them.

Beginning in 1961, a series of review articles written by Jacob and Monod outlined a a general model of genetic regulation which was based on their studies on the lac genes of Escherichia coli and of the genes of bacteriophage lambda (26, 27, 28, 39). A number of aspects of this model were fundamental in laying the groundwork for subsequent research in molecular biology. First, it was proposed that regulation of protein synthesis operated at the genetic level. Up to that time, the gene had seemed "inviolate" (31), immune to anything which could regulate its functioning. Secondly, studies in the lac system led to the concept of an intermediate in gene expression, mRNA. Experiments followed immediately, establishing the existence of such messengers. Control of protein synthesis then could be simply explained as a matter of whether or not genes were being transcribed.

In this model, operons were defined as "units of transcriptive activity . . . coordinated by a genetic element or operator" (28). Further, it was proposed that "each operon would be, by the intermediary of an operator, submitted to the action of a repressor" (29). In other words, Jacob and Monod suspected that the repressor-operator mode of control might account for genetic regulation in general. This model seemed so explanatory to them that they extended it to suggest that in higher organisms, "differentiation operates at the genetic level using elements basically similar to those found in bacteria . . ." (28). On the basis of this proposal, they described the kinds of circuitries involving repressors and operators for a series of genes which would allow a simple explanation for the mechanism of differentiation (28, 39). Today we know that the control mechanism elaborated for the lac operon is only one of many different types of regulation seen for bacterial genes. In fact, in a recent review of genetic texts, Cove (9) suggested that "perhaps the most important principle to emerge out of the study of the regulation of gene expression is that general principles do not exist." It is these aspects of the evolution of thinking about operon structure and function that I will review here.

Why was it that, from these early studies, a universal model of genetic regulation involving repressors and operators was deduced? In fact, most biologists studying regulation over the next 7 or 8 years did use the operon model to explain their own data on gene expression in bacteria and higher organisms. For those who suggested that there were genes controlled by different mechanisms, acceptance was very hard to come by. The case of Ellis Englesberg described below illustrates this atmosphere. One reason for the success of the operon model was the strong personality of Monod in particular. "He had an air of assurance, he was immensely articulate, he was very convincing" (52). He was persuaded by the beauty and simplicity of the ideas, did not like proposals of alternative models, and forcefully and effectively argued against them. (A number of biologists working on gene regulation at that time have described to me the impact of Monod’s personality on the response to alternative proposals.) However, another reason I would propose for this general acceptance of a universal model of genetic regulation is the series of major findings in biology that indicated the unity of mechanism in the biological kingdom. From pathways of metabolism to the structure of DNA, from the genetic code to the mechanism of protein synthesis, it appeared that organisms whether eukaryote or prokaryote used the same approaches to carry out their cellular functions. It seemed that in evolution only one out of many possible pathways or mechanisms had been chosen for most important cellular processes. The pathways chosen must have been the most efficient ones for solving particular problems. Given the extreme simplicity and economy of the repressor-operator model and its verification both in the lac operon and in lambda, it seemed reasonable to propose this as yet another example of the universality of biological mechanisms. Furthermore, the comprehensive view of gene expression that the operon model offered as a whole must have made it tempting to accept the details of it as universal. However, there was no objective reason to assume this universality. Of course, the role of objectivity in the scientific process has been greatly exaggerated (4).

THE OPERON: THEN AND NOW

The proposal of the operon model of genetic regulation opened up the field of study of gene expression and control. The intellectual processes which led to these breakthroughs are described by Kenneth Schaffner (46), by Horace Freeland Judson in The Eighth Day of Creation (31), by Jacob and Monod in their Nobel acceptance speeches (25, 38), by some of their coworkers in The Origins of Molecular Biology (36) dedicated to Jacques Monod, and by Melvin Cohn in his introduction to The Operon (8). The approaches used by Jacob, Monod, and coworkers—the isolation of regulatory mutants, the use of diploids to study the interaction of alleles of regulatory genes, etc., and the conceptual breakthroughs such as the insights leading to the negative control model and the formulation of a cis-acting site for regulation, the operator—made the task of elaborating control mechanisms seem much more approachable. Rather than the vague formulations of the previous decade, it now appeared possible from their work to suggest concrete molecular mechanisms for these processes and to develop precise tests for them. In fact, the molecular mechanism proposed for the lac operon was shown by work of the subsequent decade to be correct in its important features.

The term "operon" was first proposed in a short paper in the Proceedings of the French Academy of Sciences in 1960 (29). In that paper, it was shown that two genes, the lacZ gene coding for β-galactosidase and the lacY genes coding for the galactoside permease, could be coordinately controlled by a genetic element located adjacent to them, termed the operator. The operator at that time was thought to be the site where both the initiation of gene expression took place and control of regulation by the lactose repressor was effected. This economic model of gene expression and regulation allowed a single site to serve both as a receiver of regulatory signals and as what we would call today the promoter. However, subsequent studies with the lac operon revealed that the promoter and operator were distinct elements, the order of sites being P-O-Z-Y- (23). Yet, even this early modification of the operon model turns out not to have provided a general picture of the relationship between promoters and operators. As suggested by "Cove’s principle," the relative positions of operator and promoter vary from system to system. For instance, it has been shown that the earlier, more economic model of overlapping promoter and operator holds for those genes regulated by the trp repressor. This repressor controls three separate operons or genes, the aroH gene, the trp operon, and the structural gene for the repressor itself, trpR. In each case, the binding site for the repressor coincides with a portion of the promoter (19). The overlap occurs in the –35 region for the aroH gene, in the –10 region for the trp operon, and in the –10 and transcription initiation regions for the trpR gene.

One of the first indications that the repressor-operator model might not hold for all systems came from studies on the operon determining histidine biosynthesis (his) in Salmonella typhimurium. This operon was composed of 10 genes determining the structure of enzymes of the pathway. From their genetic analysis, Ames, Hartman, and coworkers had found classes of regulatory mutations in the his operon similar to those found with lac (2). Constitutive mutations mapping adjacent to the operon, with cis-acting effects, were termed operator-constitutives. There were also constitutive mutations in trans-acting genes with properties analogous to those in the lacI gene. However, a greater complexity in the regulation was indicated by the existence of several such regulatory genes, in contrast to the single lacI gene. Yet, in 1963, Ames and Hartman did refer to operator-constitutive mutants and "mutations in one or more repressor gene(s)." By 1966, however, they had discovered that many of the so-called "repressor" genes were genes determining either the structure of histidyl-tRNA or its ability to be charged (43, 44). This led them to modify their original proposal and to suggest that the his operon might be controlled at the translational level. As we know now, their proposal of translational level effects was, in part, correct as shown by the demonstration of the attenuation mechanism acting in this system (32). There were no "repressor" genes, and the "operator-constitutive" mutations were actually alterations of the attenuator region leading to reduced transcription termination (30).

Another attempt to "stretch" the negative control model to explain complex data is seen in studies of certain amino acid biosynthetic pathways. In the case of the isoleucine-valine and threonine pathways, several amino acids were required for effective end product repression. This phenomenon was termed multivalent repression (14, 15), and a model was presented in which a repressor was allosterically transformed by the concerted action of several aminoacyl tRNAs (6a). Umbarger recalls, ". . . (we) went off to our own biosynthetic and catabolic systems in bacteria, fungi and even developing embryos to invoke operators, repressors . . . with complete abandon . . ." (51a). Recently it has been shown that the dependence on multiple amino acids for control is due to attenuators in the thr and ilv operons. These attenuators include coding regions for peptides which are rich in the codons for the amino acids involved in multivalent repression (16, 34, 37). There is, at this point, no evidence for operators or repressors in these systems!

Some of the features of the research in this field during the early period are reminiscent of the work that goes on under a scientific paradigm as described by Kuhn (33). The operon model was enormously generative of a new research field. For it to have this effect, however, people working under it had to accept many of its rules. Otherwise, the research would have become anarchic. As data began to accumulate which appeared to cause complications for the paradigm, the tendency was to invent more and more complex explanations which allow continued acceptance of the paradigm. Eventually the accumulation of contradictory information was too great to permit unwavering adherence to the paradigm. Finally, new explanatory theories became acceptable. Of course, in this case, the new theories did not replace the old one, but simply illustrated the rich diversity of mechanism.

It was not until the work of Ellis Englesberg and his coworkers on the genes determining arabinose metabolism that a true challenge to the universality of repressor-operator control appeared. While others presented genetic evidence that was consistent with positive control operating in their systems (17, 48), Englesberg was the first to hold tenaciously to such views until they were accepted. In their early work, Englesberg and his group, like others, assumed that their system was subject to negative control. In one paper they suggested that certain mutations, which later proved to be in a positive regulatory gene, were, in fact, operator mutations (11). However, the genetic data they had accumulated on the ara operon led them to suggest, beginning in 1963, that the product of the regulatory gene, araC, was a positive control factor rather than a repressor (21). The proposal was stated most forcefully in 1965 when they described positive control of the ara operon as being "in sharp contrast to the negative or repressor control . . . primarily elaborated in the β-galactosidase system" (12). Using the same general approach of mutant analysis and diploid studies developed by Jacob and Monod, Englesberg came up with as strong evidence for positive control in this system as had been obtained for negative control of the lac operon. Yet, the idea that a universal mechanism had already been found made it very difficult indeed for the positive control model to be accepted. As late as 1969, the Englesberg group was publishing papers loaded with extensive tables, pleading for acceptance of the model. The tone of these papers indicates the criticism to which this proposal was subject. In one, Englesberg and coworkers introduce an extensive study on constitutive mutants with the sentence "It is argued that we have not searched hard enough for such mutants [repressor mutants]" (13). The idea of positive control may finally have become acceptable only when evidence accumulated from other systems such as the maltose regulon (20) and gene N of bacteriophage lambda (50) indicated that a similar mechanism was involved.

In retrospect, Englesberg’s accomplishment and tenacity are extraordinary considering the atmosphere of that period. If we reread papers from his laboratory, we find that, in every respect, the model he proposed, including very complex details, was correct. He was the first to provide substantial evidence that positive control might also be an important regulatory mechanism in bacteria. Today we know that positive control is widely used in both prokaryotes and eukaryotes to regulate gene expression. But perhaps because of the long period in which Englesberg’s proposals were not accepted, he has never received the recognition for this achievement that he deserves.

Ironically, the further elaboration of the functioning of the lac operon led to the discovery that it too was subject to positive control. The phenomenon of catabolite repression was explained by the dependence of operons such as lac on the CAP protein for their transcription (10, 53). Findings such as these, that an operon can be controlled by more than one regulatory protein, have led to an even more complex picture of controlling elements. In some cases, operons have been studied which are subject to at least three different regulatory mechanisms, and accordingly, their controlling element region has several components to it. For example, the operon determining histidine utilization in Klebsiella aerogenes, hut, is controlled by a repressor, by CAP, and by a positive control factor involved in the regulation of genes determining the utilization of nitogen sources (42). These controls appear necessary since histidine can be used as either a nitrogen or carbon source. In the controlling element region, there must be an operator, a CAP-binding site, and a site of action for the other positive control factor. Why the trp operon should be regulated both by a repressor-operator interaction and by attenuation is still an open question (5).

As these examples illustrate, the recognized types of control mechanism have expanded far beyond the original repressor-operator control. In addition to positive control and attenuation, regulation can occur at the translation level (7) and at the level of mRNA degradation (retroregulation) (18, 47). This is not to mention those cases in which regulation involves the rearrangement of genetic material (49). Rather than nature having found one most efficient mechanism, it appears that each system has evolved its own approach. Whether these can be explained as the best possible mechanisms for the system to be controlled or are simply the results of independent evolution leading to different solutions is not clear. Certainly arguments can be made that in the case of the regulation of ribosomal proteins, a shorter evolutionary process was required to evolve the mRNA for those ribosomal proteins which bound to rRNA so that those proteins would act as translation repressors, rather than evolving an independent regulatory mechanism. In other words, the mechanism used may be no more efficient or satisfactory than another; it may simply have been an easier one to evolve. On the other hand, arguments have been put forth as to why positive control functions for some catabolic systems and negative control applies to others (45). At any rate, it appears that any reasonable regulatory mechanism one can imagine may in fact exist; new variations on gene regulation are found regularly.

The coordinate control of gene expression in an operon was a central feature of the original model. The finding by Ames and Garry in 1959 of coordinate control of the enzymes of the histidine operon (1) played an important role in the development of this concept. This property of operons followed from the fact that the regulation of control was imposed on the initiation of transcription of the entire operon. During the period that this model was being formulated, a fly in the ointment appeared when it was found that under certain conditions the synthesis of the galactoside transport system could be induced independently of the synthesis of β-galactosidase. While inducers such as lactose or isopropyl-β-d-thiogalactoside induced the synthesis of both proteins simultaneously, galactinol appeared to induce the transport system only. Pardee, who discussed this problem with Monod while working at the Institut Pasteur, describes (40) how to him this result seemed to violate the operon model. However, Monod felt that the evidence was so strong for the model that this one apparent aberration could be ignored. As it turns out, Monod was correct; years later the transport system induced by galactinol was shown not to be the lacY gene product but a separate system involved in the transport of galactose (41). This episode illustrates how important it was for Monod to rely on intuition and conviction. As Irving Zabin puts it, Monod, when learning of an experimental result that didn’t fit an idea of his, would say "I don’t believe it for a minute" (52). In fact, it is often the case in science that findings are made which appear to destroy a particular theory, but which are set aside and ignored until satisfactory explanations are found. The reasons for the choices made are not based on any objective criteria; they are made because of the strong conviction of the researcher in the validity of the theory.

As the study of operons continued it became clear that in certain operons it was possible to have noncoordinate expression. This occurs in situations in which there are regulatory signals in addition to those operating on the promoter at one end of the operon. These can include internal promoters, transcription termination signals between genes in an operon, or mRNA degradation signals within or at the end of an operon. The first example of an internal promoter was presented by Bauerle and Margolin (3), who discovered that the last three genes of the trp operon in S. typhimurium were expressed at higher rates than the first two genes under conditions in which the operon was repressed. This disparity was due to the presence of a weak promoter located near the terminus of the second gene of the operon. As a result, when derepression took place, the expression of the first two genes would increase faster relative to the basal level than that of later genes.

Other examples discovered since that time have often been operons which were composed of genes whose products were not in the same pathway. In these situations, the cell must under certain conditions regulate the synthesis of the gene products independently. For instance, an operon has been described which includes the genes rpsU, coding for ribosomal protein S21, dnaG, coding for DNA primase, and rpoD, coding for the sigma subunit of RNA polymerase (6, 35). Each of these products is involved in the initiation of synthesis of macromolecules; they may be linked in an operon so as to allow the cell to regulate simultaneously key steps in growth. However, there are conditions under which the synthesis of the gene products must be regulated independently. As a result, the operon contains a variety of regulatory signals. These include at least seven promoters, some of which are internal to the operon, an internal transcription termination site, an RNase processing site, and a site which may play a role in antitermination. As a consequence, despite the description of this cluster of genes as an operon, under many conditions there is no coordinate control of their expression. For example, while the expression of the rpsU and dnaG genes is unaffected by the heat shock response, the synthesis of sigma protein is induced by a shift to high temperature (51). This increased expression is due to the presence of a heat shock-responsive promoter located within the dnaG gene.

An increasing number of examples are being reported in which an operon is composed of genes whose products do not function in the same pathway. Perhaps most surprising among these are operons which include genes for proteins and genes for stable RNA species. In one case, an operon, the first gene of which codes for N-formyl-methionyl tRNA, also contains genes for the NusA protein, involved in transcription termination, and initiation factor 2 of protein synthesis (24). In another case, an operon was shown to code for both a tyrosinyl tRNA and the TufB protein (22).

CONCLUSION

What this brief review of the evolution of the concept of the operon has shown is that (i) the proposal of the operon model opened up an enormously fruitful field of research and (ii) operons are structured and regulated in ways never imagined when the first simplified model was proposed. The beauty of the original model and its apparently powerful explanatory qualities not only generated this field, but also for a period constrained thinking about alternative models. This may well be inevitable in any case of powerful new concepts. Part of their strength ironically lies in their ability to channel research in a way that restricts new speculation.

ACKNOWLEDGMENTS

I thank Bruce Ames, Harrison Echols, Francois Jacob, Maxime Schwartz, and Charles Yanofsky for important and useful comments. I thank Ann McIntosh for excellent assistance in the preparation of this manuscript. I am also grateful to Ethan Signer for pointing out to me the tone of the Englesberg papers. This work was supported by a research grant from the National Institute of General Medical Sciences and by an American Cancer Society Research Professorship. Much of the writing was done while I was on leave at the Department of Microbiology and Immunology, University of California, Berkeley.

† Reprinted from the first edition.