Transcriptional control of polarity in Escherichia coli by cAMP

sRNA-mediated regulation of gal mRNA in E. coli: Involvement of transcript cleavage by RNase E together with Rho-dependent transcription termination

In bacteria, small non-coding RNAs (sRNAs) bind to target mRNAs and regulate their translation and/or stability. In the polycistronic galETKM operon of Escherichia coli, binding of the Spot 42 sRNA to the operon transcript leads to the generation of galET mRNA. The mechanism of this regulation has remained unclear. We show that sRNA-mRNA base pairing at the beginning of the galK gene leads to both transcription termination and transcript cleavage within galK, and generates galET mRNAs with two different 3'-OH ends. Transcription termination requires Rho, and transcript cleavage requires the endonuclease RNase E. The sRNA-mRNA base-paired segments required for generating the two galET species are different, indicating different sequence requirements for the two events. The use of two targets in an mRNA, each of which causes a different outcome, appears to be a novel mode of action for a sRNA. Considering the prevalence of potential sRNA targets at cistron junctions, the generation of new mRNA species by the mechanisms reported here might be a widespread mode of bacterial gene regulation.

A fortunate journey on uneven grounds

I was surprised to be invited to write a prefatory chapter for the Annual Review of Microbiology. Indeed, I did not feel that I belonged to that class of eminent scientists who had written such chapters. Perhaps it is because I am a kind of mutant: In spite of having experienced war, both German and Soviet occupations, repeated bombardments, dictatorships, and a revolution, I managed nonetheless to engage in scientific research, thus realizing a childhood dream. After having obtained my Doctor Rerum Naturalium degree in Budapest, Hungary, I was fortunate to meet Jacques Monod at the Pasteur Institute, and this became a turning point in my scientific career. In his laboratory, I contributed to the definition of the lactose operon promoter, uncovered intracistronic complementation in β-galactosidase, and investigated the role of cAMP in Escherichia coli. In my own laboratory, together with many gifted students and collaborators, I studied the role of adenylate cyclase in bacterial virulence. This allowed the engineering of recombinant adenylate cyclase toxin from Bordetella pertussis for the development of protective and therapeutic vaccines.

In vivo transcription dynamics of the galactose operon: a study on the promoter transition from P1 to P2 at onset of stationary phase

Quantitative analyses of the 5′ end of gal transcripts indicate that transcription from the galactose operon P1 promoter is higher during cell division. When cells are no longer dividing, however, transcription is initiated more often from the P2 promoter. Escherichia coli cells divide six times before the onset of the stationary phase when grown in LB containing 0.5% galactose at 37°C. Transcription from the two promoters increases, although at different rates, during early exponential phase (until the third cell division, OD₆₀₀ 0.4), and then reaches a plateau. The steady-state transcription from P1 continues in late exponential phase (the next three cell divisions, OD₆₀₀ 3.0), after which transcription from this promoter decreases. However, steady-state transcription from P2 continues 1 h longer into the stationary phase, before decreasing. This longer steady-state P2 transcription constitutes the promoter transition from P1 to P2 at the onset of the stationary phase. The intracellular cAMP concentration dictates P1 transcription dynamics; therefore, promoter transition may result from a lack of cAMP-CRP complex binding to the gal operon. The decay rate of gal-specific transcripts is constant through the six consecutive cell divisions that comprise the exponential growth phase, increases at the onset of the stationary phase, and is too low to be measured during the stationary phase. These data suggest that a regulatory mechanism coordinates the synthesis and decay of gal mRNAs to maintain the observed gal transcription. Our analysis indicates that the increase in P1 transcription is the result of cAMP-CRP binding to increasing numbers of galactose operons in the cell population.

Timing of gene transcription in the galactose utilization system of Escherichia coli

In the natural environment, bacterial cells have to adjust their metabolism to alterations in the availability of food sources. The order and timing of gene expression are crucial in these situations to produce an appropriate response. We used the galactose regulation in Escherichia coli as a model system for understanding how cells integrate information about food availability and cAMP levels to adjust the timing and intensity of gene expression. We simulated the feast-famine cycle of bacterial growth by diluting stationary phase cells in fresh medium containing galactose as the sole carbon source. We followed the activities of six promoters of the galactose system as cells grew on and ran out of galactose. We found that the cell responds to a decreasing external galactose level by increasing the internal galactose level, which is achieved by limiting galactose metabolism and increasing the expression of transporters. We show that the cell alters gene expression based primarily on the current state of the cell and not on monitoring the level of extracellular galactose in real time. Some decisions have longer term effects; therefore, the current state does subtly encode the history of food availability. In summary, our measurements of timing of gene expression in the galactose system suggest that the system has evolved to respond to environments where future galactose levels are unpredictable rather than regular feast and famine cycles.

Gene regulation in continuous cultures: a unified theory for bacteria and yeasts

During batch growth on mixtures of two growth-limiting substrates, microbes consume the substrates either sequentially (diauxie) or simultaneously. The ubiquity of these growth patterns suggests that they may be driven by a universal mechanism common to all microbial species. Recently, we showed that a minimal model accounting only for enzyme induction and dilution, the two processes that occur in all microbes, explains the phenotypes observed in batch cultures of various wild-type and mutant/recombinant cells (Narang and Pilyugin in J. Theor. Biol. 244:326-348, 2007). Here, we examine the extension of the minimal model to continuous cultures. We show that: (1) Several enzymatic trends, attributed entirely to cross-regulatory mechanisms, such as catabolite repression and inducer exclusion, can be quantitatively explained by enzyme dilution. (2) The bifurcation diagram of the minimal model for continuous cultures, which classifies the substrate consumption pattern at any given dilution rate and feed concentrations, provides a precise explanation for the empirically observed correlations between the growth patterns in batch and continuous cultures. (3) Numerical simulations of the model are in excellent agreement with the data. The model captures the variation of the steady state substrate concentrations, cell densities, and enzyme levels during the single- and mixed-substrate growth of bacteria and yeasts at various dilution rates and feed concentrations. This variation is well approximated by simple analytical expressions that furnish deep physical insights. (4) Since the minimal model describes the behavior of the cells in the absence of cross-regulatory mechanisms, it provides a rigorous framework for quantifying the effect of these mechanisms. We illustrate this by analyzing several data sets from the literature.

Establishment of an mRNA gradient depends on the promoter: an investigation of polarity in gene expression

We found six mRNA species specific to the galactose operon of Escherichia coli. Analyses of both ends of the mRNAs indicated that while the 5' ends are fixed at the promoter region, the 3' ends vary along the operon. The resulting gal mRNA map suggests generation of an mRNA concentration gradient that is higher in the promoter-proximal region and lower toward the distal region. Real-time RT-PCR analyses of the amount of each mRNA species confirmed the existence of the gradient. This gradient of mRNA concentration could serve as an underlying mechanism for the long known phenomenon "natural polarity." Further analyses of the 3' ends of the mRNAs showed that they are generated by either an unknown mRNA processing/transcription termination mechanism(s) or Rho-dependent intra-cistronic transcription termination. The results showed also that transcription from the P2 promoter can yield a more severe mRNA gradient than that from the P1 promoter, suggesting that the slope of the mRNA gradient depends on which promoter the transcription has initiated from. These results led us to suggest a novel gene regulation model in which transcription initiation is tightly coupled to mRNA processing and/or transcription termination.

Three-stage Regulation of the Amphibolic gal Operon: From Repressosome to GalR-free DNA

The gal operon of Escherichia coli is negatively regulated by the Gal repressosome, a higher order nucleoprotein complex containing a DNA loop that encompasses two gal promoters. In the repressosome structure, Gal repressor (GalR) dimers are bound to the two operator sites, flanking the promoter region, thus generating a DNA loop. The DNA loop is stabilized by binding of the architectural HU protein to the apex of the loop, and negative supercoiling. The gal promoters are also regulated in opposite directions by GalR without DNA looping. The repressosome-mediated as well as looping-independent transcription regulation of the two promoters is lifted in the presence of the inducer D-galactose. We tested the effect of D-galactose on various DNA-protein and protein-protein interactions of different regulatory complexes and on transcription repression in vitro. We found that the inducer breaks up the repressosome with clear intermediates in a concentration-dependent manner. The sequential disassembly generates different stages of regulation of the gal operon. The D-galactose-dependent switch from one stage of regulation to another satisfies the amphibolic requirement of the gal operon.

Relief of catabolite repression in a cAMP-independent catabolite gene activator mutant of Escherichia coli

We isolated and characterized a new catabolite gene activator mutant (crp*) of Escherichia coli that confers cAMP-independent expression and total relief of catabolite repression of beta-galactosidase and tryptophanase synthesis. The two mutations responsible for this phenotype change the amino acids at codon 72 from Glu to Ala and at codon 144 from Ala to Thr in the corresponding CAP* protein.

Spot 42 RNA mediates discoordinate expression of the E. coli galactose operon

The physiological role of Escherichia coli Spot 42 RNA has remained obscure, even though the 109-nucleotide RNA was discovered almost three decades ago. Structural features of Spot 42 RNA and previous work suggested to us that the RNA might be a regulator of discoordinate gene expression of the galactose operon, a control that is only understood at the phenomenological level. The effects of controlled expression of Spot 42 RNA or deleting the gene (spf) encoding the RNA supported this hypothesis. Down-regulation of galK expression, the third gene in the gal operon, was only observed in the presence of Spot 42 RNA and required growth conditions that caused derepression of the spf gene. Subsequent biochemical studies showed that Spot 42 RNA specifically bound at the galK Shine-Dalgarno region of the galETKM mRNA, thereby blocking ribosome binding. We conclude that Spot 42 RNA is an antisense RNA that acts to differentially regulate genes that are expressed from the same transcription unit. Our results reveal an interesting mechanism by which the expression of a promoter distal gene in an operon can be modulated and underline the importance of antisense control in bacterial gene regulation.

Synthesis and degradation of the mRNA of the Tn21 mer operon

The mercury resistance locus encoded by Tn21 on the monocopy IncFII plasmid R100 (merTn21) consists of a metal-responsive activator/repressor, merR, which controls initiation of a polycistronic message that includes genes for the uptake (merTPC) and reduction (merA) of Hg2+ and merD, which may also play a minor regulatory role. Comparison of the relative abundance of the 5' and 3' ends of the merTPCAD transcript revealed a strong transcriptional gradient in the operon, consistent with previous observations of lower relative abundance of the more promoter-distal gene products. In vivo mRNA degradation rates varied only slightly for the different genes: however, the rates of mRNA synthesis varied considerably from the beginning to the end of the operon. Specifically, mRNA corresponding to the promoter-proximal genes, merTPC, achieved a maximum in vivo synthesis rate between 60 and 120 seconds after induction; this rate was maintained for approximately ten minutes. In contrast, the synthesis rates of mRNA corresponding to the promoter-distal genes merA and merD, were initially fivefold lower than the rates of the promoter-proximal genes for the first five minutes after induction, and then rose gradually to approximately 50% of the merTPC synthesis rates. These data suggested that early after induction only 20% of the transcripts initiating at merT proceed beyond merC. At later times after induction approximately 50% of the transcripts proceed beyond merC. Nuclease end mapping did not reveal any discrete termination events in the merPCA region, thus, premature termination may occur at many sites.

Autogenous control is not sufficient to ensure steady-state growth rate-dependent regulation of the S10 ribosomal protein operon of Escherichia coli

The regulation of the S10 ribosomal protein operon of Escherichia coli was studied by using a lambda prophage containing the beginning of the S10 operon (including the promoter, leader, and first one and one-half structural genes) fused to lacZ. The synthesis of the lacZ fusion protein encoded by the phage showed the expected inhibition during oversynthesis of ribosomal protein L4, the autogenous regulatory protein of the S10 operon. Moreover, the fusion gene responded to a nutritional shift-up in the same way that genuine ribosomal protein genes did. However, the gene did not exhibit the expected growth rate-dependent regulation during steady-state growth. Thus, the genetic information carried on the prophage is sufficient for L4-mediated autogenous control and a normal nutritional shift-up response but is not sufficient for steady-state growth rate-dependent control. These results suggest that, at least for the 11-gene S10 ribosomal protein operon, additional regulatory processes are required to coordinate the synthesis of ribosomal proteins with cell growth rate and, furthermore, that sequences downstream of the proximal one and one-half genes of the operon are involved in this control.

Mutation of the promoter and LexA binding sites of cea, the gene encoding colicin E1

Three mutations were introduced into the cea promoter using oligonucleotide directed mutagenesis. The resulting mutant promoter has the Escherichia coli consensus sequences at its - 35 and - 10 positions, separated by the optimal spacing. In addition, a plasmid with a mutation in one of the two LexA repressor binding sites in the cea regulatory region was isolated that decreases homology with the consensus LexA binding site. The effects of these mutations on cea expression were studied in cea-lacZ protein fusions. The promoter-up mutant, when present in a multicopy plasmid, showed a shorter induction lag when compared to the wild-type cea gene, and there was less of an effect of the catabolite repression system on cea expression. However, when present in a single copy in the bacterial chromosome, catabolite repression and an induction delay were observed, despite the increased strength of the promoter. The operator mutant showed a slightly higher basal level of expression, but was still repressible. Induction occurred with a shortened lag period, but the effects were not as great as with the promoter mutant. These results support the idea that tight repression by LexA contributes to the delay in cea induction.

Interaction of the CRP-cAMP complex with the cea regulatory region

Analysis of the induction of expression of cea-lacZ fusions in cya and crp mutants showed that catabolite repression affects the kinetics of induction and the rate of induced synthesis. In a cya mutant, addition of cAMP reduced the induction lag and increased the amount of beta-galactosidase produced. The CRP-cAMP complex was found to bind to two sites 5' to the cea promoter, but deletion analysis showed that only one of these was involved in the control of cea. Deletion of this site resulted in a loss of the stimulatory effects of cAMP in a cya mutant.

Temporal control of colicin E1 induction

The expression of the gene encoding colicin E1, cea, was studied in Escherichia coli by using cea-lacZ gene fusions. Expression of the fusions showed the same characteristics as those of the wild-type cea gene: induction by treatments that damage DNA and regulation by the SOS response, sensitivity to catabolite repression, and a low basal level of expression, despite the presence of the fusion in a multicopy plasmid. Induction of expression by DNA-damaging treatments was found to differ from other genes involved in the SOS response (exemplified by recA), in that higher levels of DNA damage were required and expression occurred only after a pronounced delay. The delay in expression following an inducing treatment was more pronounced under conditions of catabolite repression, indicating that the cyclic AMP-cyclic AMP receptor protein complex may play a role in induction. These observations also suggest a biological rationale for the control of cea expression by the SOS response and the cyclic AMP-cyclic AMP receptor protein catabolite repression system.

Translation of galE and coordination of galactose operon expression in Escherichia coli: effects of insertions and deletions in the non-translated leader sequence

Using gene-manipulation techniques, we made a set of short insertions and deletions in the Escherichia coli galactose operon between the transcription start site and the Shine-Dalgarno sequence of the first gene of the operon, galE. Translation initiation is severely reduced when the distance between the 5' end of the message and the Shine-Dalgarno sequence drops below 15 bases. Transcription of the gal operon can start at two distinct sites, S1 and S2, separated by 5 bp, situated 16 and 21 bp upstream of the galE Shine-Dalgarno sequence, respectively. When transcription starts at S2, gal operon expression is discoordinate as the galE gene is better translated than promoter-distal genes. Here we report that gal operon expression is discoordinate even when message starting at S2 is shortened. This shows that the better translation of galE from transcripts starting at S2 is not simply due to the fact that they are longer than transcripts starting at S1.

Interaction between the cyclic AMP receptor protein and DNA. Conformational studies

The binding of the cyclic adenosine 3',5' monophosphate receptor protein (CRP or CAP) of Escherichia coli to non-specific DNA and to a specific lac recognition sequence has been investigated by circular dichroism (c.d.) spectroscopy. The effect of cAMP and cGMP on the co-operative non-specific binding was also studied. For the non-specific binding in the absence of cAMP a c.d. change (decrease of the intensity of the positive band with a shift of its maximum to longer wavelength) indicates that the DNA undergoes a conformational change upon CRP binding. This change might reflect the formation of the solenoidal coil previously observed by electron microscopy. The amplitude of the c.d. change increases linearly with the degree of saturation of the DNA and does not depend on the size of the clusters of CRP bound. From the variation of the c.d. effect as a function of the ionic strength, the product K omega (K, the intrinsic binding constant and omega, the co-operativity parameter) could be determined. The number of ion pairs involved in complex formation between CRP and DNA was found to be six to seven. Experiments performed with several DNAs, including the alternating polymers poly[d(A-T)] and poly[d(G-C)], demonstrated that the conformational change does not depend on the DNA sequence. However, in the presence of cAMP the c.d. spectrum of the DNA shows only a small variation upon binding CRP. In contrast, in the presence of cGMP the conformational change of the DNA is similar to that observed when non-liganded CRP binds. For the specific lac operon binding, the c.d. change is different from those observed for non-specific binding in the presence or absence of cAMP. These results emphasize the high variability of the DNA structure upon binding the same protein.

2-Ketoglutarate as a possible regulatory metabolite involved in cyclic AMP-dependent catabolite repression in Escherichia coli K12

2-Ketoglutarate--unlike any other derivative of the citric acid cycle--was found to strongly repress catabolite-sensitive genes, such as the lactose operon (lac) or the tryptophanase gene (tna), when added to cells grown in glycerol. 2-ketoglutarate affects the expression of these genes by decreasing cyclic AMP synthesis. Such inhibition of cyclic AMP synthesis requires the presence of enzyme III, a component of the phosphoenol pyruvate:sugar phosphotransferase transport system (PTS). Thus, it is proposed that 2-ketoglutarate is one of the catabolite repressors postulated by Magasanik in 1961. In addition, by studying the effect of 2-ketoglutarate in various mutants, we show the existence of a cyclic AMP-independent catabolite repression mechanism whose mediator is synthesized from 2-ketoglutarate.

Transcription analysis of the sucAB, aceEF and lpd genes of Escherichia coli

Transcript mapping of the Escherichia coli sucAB, aceEF and lpd genes, encoding the five components of the pyruvate and 2-oxoglutarate dehydrogenase complexes, was carried out using single-stranded M13 probes. The sucA and aceE genes encode the specific dehydrogenase components (E1o, E1p), and the sucB and aceF genes encode the specific dihydrolipoamide acyltransferases (E2o, E2p). The common lipoamide dehydrogenase (E3) component is encoded by a single lpd gene adjacent to the aceEF genes. The sucAB, aceEF and lpd genes were all expressed on independent transcripts, and the promoters and terminators were identified. In addition, readthrough transcription from the sucAB genes to a downstream gene designated sucC, and from the aceEF genes to the adjacent lpd gene, was found. The relative levels of transcription of the suc, ace and lpd genes, and of the three different transcript types covering the ace-lpd region, were quantified using RNA from cells grown on different substrates. Most of the E3 components supplying the pyruvate dehydrogenase complex appear to be synthesised from approximately 6415-base aceEF-lpd readthrough transcripts, but additional approximately 4640-base aceEF transcripts terminating after the aceF gene provide a transcriptional basis for the observed stoichiometric excess of E1p and E2p relative to E3 in the assembled complex. Conversely most of the E3 components required for the 2-oxo-glutarate dehydrogenase complex appear to be synthesised from the independent 1670-base lpd transcripts.

[Mode of action of cyclic amp in prokaryotes and eukaryotes, CAP and cAMP-dependent protein kinases]

cAMP is an ubiquitous compound which is involved in the regulation of many biological processes. In bacteria such as E. coli, cAMP mediates the activation of catabolic operons via the CAP protein. The CAP-cAMP complex, whose tridimensional structure has recently been established, binds to the promoter regions of catabolic operons at a specific site, and activates their transcription by inducing RNA polymerase to bind and initiate transcription at the correct site. Various phenomenons including protein-protein interactions or CAP-induced DNA bending or kinking could be involved in the process of forming the open transcription complex. In eukaryotes, cAMP activates cAMP dependent protein kinases which covalently modify proteins by phosphorylation on serine or threonine residues. The catalytically inactive holoenzyme is generally a tetramer containing two regulatory subunits, each capable of binding two molecules of cAMP, and two catalytic subunits. In mammalian cells, two types of cAMP dependent protein kinases (I and II) can be distinguished on the basis of their regulatory subunits; their relative proportion varies from tissue to tissue. Binding of cAMP to the regulatory subunits induces the dissociation of the holoenzyme and releases the free and active catalytic subunits. Phosphorylation of proteins occurs at sequences containing two basic residues in the vicinity of the phosphorylated serine or threonine. A heat-stable protein, present in most eukaryotic cells, specifically interacts with the catalytic subunit and inhibits its activity. The amino-acid sequence of cAMP dependent protein kinases has recently been determined. It is interesting to note that the domains responsible for cAMP binding by the regulatory subunits of mammalian cAMP dependent protein kinases and CAP share important sequence homologies. The same phenomenon is observed concerning the domain responsible for ATP binding to the catalytic subunit of cAMP dependent protein kinases and that of tyrosine-specific protein kinases from oncoviruses. Other eukaryotic proteins such as S-adenosyl-L-homocysteine (SAH) hydrolase are also capable of binding cAMP. The latter is involved in the regulation of S-adenosyl-L-methionine dependent methylations, and its activity could be affected by cAMP. Besides its role as an effector of enzymatic activity via phosphorylation, such as in the regulation of glycogen metabolism, cAMP has recently been shown to activate the transcription of a number of eukaryotic genes. This process probably also involves protein phosphorylation, but its precise mechanism remains to be understood.

Catabolite repression 1985

The present status of catabolite repression is summarized with respect to the involvement of cyclic AMP and other mediators. A model is presented which may account for the relationship between positive control of gene expression exerted by cAMP and its receptor, CAP, and negative control of catabolite repression mediated by specific metabolites.