To access the private area of this site, please log in.

Username: 

Password:  

 

Chapter 94

The Leucine/Lrp Regulon

E. B. NEWMAN, R. T. LIN, and R. D’ARI

 

INTRODUCTION

The leucine/Lrp regulon is a recently described global response governed by a transcriptional regulator called the leucine-responsive regulatory protein, or Lrp. This regulon has been the subject of several recent reviews (18, 22, 72, 172). Lrp, specified by the lrp gene immediately downstream of trx at 20 min (116), affects the transcription of a large number of genes, increasing the expression of some and decreasing that of others (22, 59, 71, 83). At some promoters, Lrp action is greatly modified by the presence of l-leucine in the growth medium, whereas at others, leucine has little or no effect.

Most global regulators are recognized by their physiological and environmental roles. For example, the catabolite activator protein (Crp) regulates the expression of genes involved in carbohydrate degradation, and the RelA protein, by varying the ppGpp concentration, adjusts the rate of ribosome formation to the availability of aminoacyl-tRNA. In many cases, the physiological role of the response is clear because a unifying stimulus was imposed by the screening process itself, as in the cataloguing of proteins induced by passages between aerobiosis and anaerobiosis (100, 101). No corresponding general statement can be made for the leucine/Lrp regulon.

Lrp was discovered in a number of separate quests as a transcriptional regulator of specific operons, many of which were known to be turned on or off by exogenous leucine. It thereby received several designations, reflecting the specialized interests of the various laboratories: Lrs, for "leucine resistance" of ilvIH expression (83, 109); Ihb, for "ilvIH-binding" protein (90); LivR, for "regulator of leucine-isoleucine-valine" transport (4); OppI, for regulator of the OppABCD "oligopeptide permease" (8); Mbf, for "methylation-blocking factor" in the pap operon (15); and Rbl, for "regulation by leucine" when it was demonstrated to be a global regulator governing the synthesis of l-serine deaminase (l-SD) and two other enzymes (59). All groups actively working on this gene or protein agreed to call it Lrp, for "leucine-responsive regulatory protein."

The lack of physiological focus has oriented research on Lrp differently from that on other global regulators. To date, work has mainly addressed two types of question: physiological (what genes does Lrp affect, and what are the physiological consequences of its action?) and biochemical (what kind of protein is Lrp, and how does it interact with DNA?).

Estimates of the total number of genes in the leucine/Lrp regulon range from 35 to 75. The lower estimate comes from a comparison of two-dimensional protein gels from extracts of wild-type and lrp mutant strains, grown with and without leucine. Some 30 proteins were clearly affected—up or down—by the absence of Lrp (24). The higher estimate is from a study of random λplacMu insertions in the Escherichia coli genome, screened for leucine effects and shown to be Lrp dependent (61). The first 22 fusions analyzed showed few duplicates, suggesting that they do not come close to revealing all Lrp-regulated genes.

 

Lrp

Lrp is a transcriptional regulator (activator or repressor) of a large number of operons. It is a small basic protein (pI 9.2) composed of two identical 18,800-Da subunits (119). Its sequence resembles that of only one other E. coli protein, AsnC, a positive regulator of asnA (the structural gene of asparagine synthetase A), with which it has 25% amino acid similarity. The mature protein has 163 residues, including all amino acids except tryptophan. A sequence centered at position 40 may represent a helix-turn-helix motif (119). Wild-type Lrp and Lrp-1, a slightly altered protein, were purified by means of phosphocellulose, heparin-agarose, and DNA-cellulose columns, locating the Lrp-containing fractions by their ability to retard a DNA fragment immediately upstream of the ilvIH operon, which is positively regulated by Lrp (119). The DNA sequence agrees with the first 38 N-terminal amino acids of the protein (83, 119).

Lrp is a moderately abundant protein (119). By assays of protein reacting with anti-Lrp antibodies, it was estimated to make up 0.1% of cellular protein in cells grown in glucose minimal medium. Comparisons with known proteins on two-dimensional gels suggested a similar figure (0.1 to 0.2%). This was calculated to correspond to about 3,000 molecules per cell, an estimate close to the number of Crp molecules per cell.

Purified Lrp binds well to double-stranded DNA containing an appropriate operator site, even in the presence of a 1,000-fold excess of calf thymus DNA, but it does not bind to single-stranded DNA of the same sequence (114). Lrp binding has major structural effects. As judged by the circular permutation assay, it bends the DNA an estimated 52° for a single Lrp molecule and 135° for two molecules bound to adjacent sites (114). Wang and Calvo (114) suggest that Lrp forms an architectural element, facilitating the assembly of a nucleoprotein complex regulating transcription.

 

REGULATION OF Lrp SYNTHESIS

Like many transcriptional regulators, Lrp is autogenously regulated: in minimal glucose medium, expression of an lrp::lacZ fusion is lowered 2- to 3-fold by the presence of a single chromosomal lrp ⁺ gene and 10-fold by the presence of a functional lrp ⁺ gene on a multicopy plasmid (61, 116). The presence of leucine in the medium has no effect on expression. The lrp promoter has been located by primer extension 267 bp upstream of the translational start codon (116). Lrp binds to a site between 32 and 80 bp upstream of the transcription start, as indicated by gel mobility shifts, DNA footprinting, and deletion studies. Mutations in that area alter the response to Lrp, strongly suggesting that autogenous regulation is a direct effect of Lrp binding there (116).

The lrp gene is subject to a second powerful regulation: growth in rich medium causes striking repression. In LB broth, there is a 10-fold, Lrp-independent decrease in lrp transcription (61). This decrease was also observed in medium containing 1% Casamino Acids, with or without glucose. A large part of this repression could be produced by adding either the α-ketoglutarate or the oxalacetate family of amino acids to glucose minimal medium. Similar repression was seen in cells grown in minimal medium with either pyruvate or acetate (but not glycerol) as a carbon source (R. Sears, R. T. Lin, and E. B. Newman, unpublished results). Thus, the level of Lrp may vary with the relative external abundance of amino acids and the type of carbon source supplied but thus far has been found to be lowest in rich medium.

Considering its widely differing binding affinities for different promoters (discussed below), Lrp may well be an important metabolic regulator in both rich and poor media. The different levels of lrp expression may indicate different settings of the cell’s metabolic web, whereby systems with low affinity for Lrp are significantly regulated by it only during growth in minimal medium (25).

 

PHYSIOLOGICAL CONSEQUENCES OF Lrp ACTION

There are major differences between the transcriptional pattern in an lrp mutant and that of a wild-type strain (discussed below). Indeed, these differences provide one basis on which the regulon has been defined. In this section, we first consider the overall metabolic effects of total loss of Lrp, i.e., the physiology of an lrp mutant. Given the magnitude of changes consequent upon a total loss of Lrp, variations in the amount of Lrp would be expected to alter gene transcription considerably. The largest variation of the amount of Lrp presently known is the 10-fold decrease in rich media. We then briefly consider the physiology of wild-type cells growing in LB broth. One of the principal effects of an lrp mutation is a reorganization of the cell’s metabolism, making it highly sensitive to further perturbations, as discussed in the section on other metabolic operons. Finally, a special instance of this reorganization, the partial suppression of metK alleles by Lrp, is discussed.

 

Metabolism of lrp Mutants

As might be expected from the very different quantitative effects of Lrp at different promoters (discussed below), the phenotypic manifestations of loss of Lrp are many and varied. The principal effects now known involve the metabolism of serine and one-carbon units, nitrogen assimilation, biosynthesis of some amino acids, transport capabilities, and the synthesis of external appendages. However, other phenotypes may well appear as the study of the leucine/Lrp regulon continues.

Sensitivity to l-Leucine and l -Serine.

Much of the physiological work on the lrp mutant was done with strains carrying an ilvA deletion, and so isoleucine and valine were routinely added to all media. Surprisingly, in an ilvA ⁺ genetic background, the lrp mutant proved to be sensitive to serine and leucine, either of which blocks growth unless isoleucine and valine are added as well (2). Serine is known to cause transient isoleucine-valine starvation in wild-type E. coli (110), and leucine also slows the growth of wild-type strains (12). It is not presently known how Lrp helps E. coli overcome the stress caused by addition of serine or leucine to the medium.

Limited Availability of l-Serine and l-Leucine.

Under standard laboratory conditions, the lrp mutant is surprisingly robust, considering the large number of genes with altered expression. It is a prototroph, although growth in glucose minimal medium is slower than that of wild-type strains (in the presence of isoleucine and valine, a doubling time of 84 min, compared with 58 min for the wild type). This slow growth is alleviated by adding serine and leucine to the medium (2, 59), suggesting that synthesis of these amino acids is limiting in the absence of Lrp. The lrp mutant tends to accumulate partial suppressors on subculturing and should be frequently retransduced for continued study (E. B. Newman, unpublished observations).

The decreased ability of the lrp mutant to synthesize serine and leucine probably reflects the fact that Lrp activates the transcription of genes involved in serine and leucine biosynthesis, serA and the leuABCD operon (59, 61, 107). The increased l-SD activity of the lrp mutant enables it to use l-serine as a sole carbon and energy source (59). High l-SD activity probably does not aggravate serine limitation during growth on glucose because of the high K_m of the enzyme for l-serine (104, 105).

In glucose minimal medium (containing isoleucine and valine), the lrp mutant is unable to grow either aerobically at 42°C or anaerobically at 37°C (2). This behavior again seems to reflect effects on serine synthesis since in both cases, growth is restored by the addition of serine to the medium or by increasing the number of copies of the serA gene (2).

Increased Serine Degradation.

Not only does the lrp mutant have reduced synthesis of l-serine, but it also has increased serine degradation, owing to a sevenfold derepression of l-serine deaminase activity (59). This permits it to utilize l-serine as a sole carbon source (in the presence of isoleucine and valine), unlike wild-type E. coli K-12.

Synthesis of One-Carbon Units.

Wild-type E. coli has two routes for formation of N-5,10-methylenetetrahydrofolate (mTHF), which is the source of one-carbon units used in methylation reactions, either directly or via S-adenosylmethionine (73). The first route involves the enzyme serine hydroxymethyltransferase (SHMT), the glyA gene product, which catalyzes the formation of glycine and mTHF from serine and THF (Fig. 1). This is also the major pathway of glycine synthesis during growth on glucose. The second route involves the glycine cleavage enzyme complex (Gcv), forming mTHF from the alpha carbon of glycine and simultaneously releasing CO₂ and NH₃.

Accordingly, in wild-type cells growing in glucose minimal medium, the glycolytic intermediate 3-phosphoglycerate is the direct precursor of serine, glycine, and one-carbon units. The requirement of the cell for one-carbon units is greater than its need for glycine, and the Gcv system permits it to meet this requirement by overproducing glycine and cleaving the excess (73, 74).

In the lrp mutant, this scheme is radically altered because the gcvTHP operon absolutely requires Lrp activation for physiologically significant expression (61). The lrp mutant, like gcv mutants (70), cannot derive nitrogen from glycine, and all mTHF must be formed by SHMT (61). Indeed, an lrp glyA mutant, which cannot make SHMT, cannot grow in a glycine-supplemented minimal medium presumably because it cannot make one-carbon units for biosynthesis (61). This should result in overproduction of glycine, a problem which may be further compounded by the increased ability of the lrp mutant to form glycine from threonine because of derepression of a threonine cleavage system (Fig. 1) (59, 89; see below). Without the glycine cleavage system to equilibrate glycine metabolism, one would expect the lrp mutant to excrete glycine, as has been observed in gcv mutants (19), but glycine has not been detected in culture fluids (R. T. Lin, unpublished results).

Nitrogen Metabolism.

The lrp mutant assimilates ammonia or ammonium ion readily, at least at high concentrations, but it has difficulty using organic nitrogen owing to its deficiency in glutamate synthase (25). Like gltB and gltD mutants, it cannot use arginine or ornithine (and perhaps other amino acids as well) as a nitrogen source, and owing to lack of expression of gcv (see above), it cannot use glycine either.

Appendages and Transport.

The relation of the lrp mutant with the outside world is greatly changed, with respect both to its appendages and to its transport capabilities. Almost all operons coding for fimbriae seem to be affected by Lrp (see below and chapter 11). In most cases studied, the lrp mutant is unable to make the appendage, suggesting that adhesion to animal tissues would be severely handicapped in the absence of Lrp. Relatively few transporters have been studied with respect to Lrp regulation (discussed below). In the presence of leucine, Lrp represses two high-affinity leucine uptake systems, one of which also transports isoleucine, valine, threonine, and alanine. Lrp also represses an oligopeptide uptake system, activates a serine transporter, and slightly stimulates expression of the maltose and lactose permeases. So far, the only phenotype associated with these alterations in transport ability is resistance of the lrp mutant to toxic tripeptides (5, 6).

Absence of Effect on Timing of Initiation of DNA Synthesis.

Lrp, known as methylation-blocking factor (Mbf) in fimbrial operons, is able to distinguish the methylation state of certain GATC sequences. It is known that specific sequestration of hemimethylated GATC sequences in the replication origin oriC is important for proper regulation of initiation, preventing reinitiation of a given oriC sequence in the same generation (77). Lrp was therefore a potential candidate for regulating initiation by binding to freshly replicated (hemimethylated) oriC sequences and guiding them to the membrane. However, the timing of replication initiation is not affected in the lrp mutant (99), indicating that Lrp does not play this role.

 

Growth of E. coli with Reduced Lrp Levels

In LB broth, E. coli has a low Lrp level and is exposed to a high-leucine environment, which for many operons lowers the efficiency of Lrp activation or represssion (see below). LB broth-grown E. coli therefore must differ from its lrp mutant only in the expression of those operons with the lowest affinity for Lrp. Consequently, an accurate assessment of the role of Lrp in LB broth requires a quantitative evaluation of the affinities of the various target promoters for Lrp with and without leucine. This information is not yet generally available in detail (25). However, using a plasmid-carried lrp gene under the control of the araBAD promoter, we have shown that the affinity of the gcv and gltD promoters for Lrp is much higher than that of the sdaA gene; therefore, that expression of the former two genes may not be affected by the change in Lrp concentration in LB (Liang Tao and E. B. Newman, unpublished results).

 

Reduced Tolerance of lrp Mutants for Other Mutations

Lrp deficiency reorganizes metabolism considerably but does not prevent growth in glucose minimal medium at 37°C; indeed, in the presence of serine and the three branched-chain amino acids, the lrp mutant grows almost as well as its parent (2, 61). However, loss of Lrp leads to diminished metabolic flexibility. The lrp mutant is not adaptable, a characteristic which is seen clearly in its reduced tolerance of secondary mutations. For example, an lrp relA double mutant has an absolute requirement for leucine and requires serine as well for a normal growth rate (2). Similarly, an lrp pnt double mutant, lacking pyridine nucleotide transhydrogenase as well as Lrp, requires either glutamate, glutamine, aspartate, or asparagine to grow, although both single mutants are prototrophic (2). An lrp glyA double mutant cannot grow in minimal medium and grows very poorly in LB broth, presumably because it cannot make one-carbon units. These observations suggest that the changed metabolic pattern in the lrp mutant, although internally consistent, is not able to compensate for changes brought about by further mutation, a property that we call synergistic inefficiency.

The metabolic relationship between lrp and relA mutants is complex and not yet understood. The relA single mutant, like the lrp mutant, is sensitive to l-serine and l-leucine, either of which blocks growth unless isoleucine and valine are added as well (1, 110). Loss of RelA, like loss of Lrp, somehow exacerbates the natural sensitivity of E. coli to these amino acids. It is possible that Lrp activates relA expression, although the relA gene must be expressed at a significant level in the lrp mutant since a relA mutation alters the strain’s phenotype.

 

Improved Growth of lrp metK Mutants

The metK gene product, S-adenosylmethionine synthetase, catalyzes the formation of S-adenosylmethionine, the direct methyl donor in most methylation reactions. One of the surprising activities of Lrp is that it makes metK mutants almost inviable: metK lrp ⁺ strains grow exceedingly slowly, whereas metK lrp double mutants grow well, and indeed metK strains tend to accumulate lrp mutations (61). These observations suggest that Lrp represses the synthesis of a substitute for MetK. However, it is unclear at this time what that substitute might be.

 

Lrp AS A CHROMOSOME ORGANIZER

Soon after protoplasm was enclosed in a membrane, the resulting cells must have evolved a DNA-packaging mechanism, without which their large mass of DNA would be unmanageable. In eukaryotes, histones carry out this function. In bacteria, too, packaging probably involves basic DNA-binding proteins —some usually thought of as specific regulators rather than as chromosome organizers—such as Crp, integration host factor (IHF), H-NS, AraC, and Lrp, among others. Each of these proteins is thought to bind specifically to certain sites and to bend DNA (28, 44, 63, 94, 95, 114). Other basic proteins, such as HU and Fis, also bind to DNA, with no apparent site specificity, and affect its structure (23, 91). Indeed, there may be very little uncoated DNA in the cell. The shape of the chromosome must be the result of the effects of large numbers of protein molecules binding to the DNA and bending it. This bending may be a major determinant of gene transcription (16, 79, 122, 123).

It is thus possible that Lrp, by its nature as a small basic DNA-bending protein and the large number of Lrp molecules per cell (approximately 3,000), is one of the determinants of chromosome structure. It is clear that Lrp also has a strong effect on the expression of a number of operons. Lrp probably binds DNA at a large number of sites of varying affinity. If there is a high-affinity site in the promoter region of a given gene (e.g., ilvIH or serA), Lrp binding is likely to have a strong effect on expression. If there is a low-affinity site in the promoter region, Lrp binding might have less effect. Thus, Lrp may play a dual role in cell physiology: as a specific regulator of certain operons, where its binding affects expression significantly, and as a less specific DNA-wrapping and -organizing protein. Changes in expression in mutants may not indicate a direct regulatory role of the missing protein but rather be due to structural changes in less thoroughly wrapped DNA.

Crp is considered to be an important regulatory protein because its action depends on an effector cyclic AMP, which varies in concentration according to the physiological conditions of the cell. Even the concentration of Crp varies (46). So far, there is no evidence for a cofactor for the many non-leucine-dependent Lrp-regulated genes. The importance of Lrp for these operons may be less in its regulation of their expression than in its general role in helping to determine the folding of the chromosome.

As the cell evolved, chromosome structure likely evolved too. All of the proteins that bind to DNA, and change its shape as by bending it, must determine the overall structure, and the controls on gene expression must have been involved in this structured DNA. The establishment of a new binding protein during evolution might then have changed the characteristics of the cell greatly. It has been suggested that Lrp converts the cell from metabolism needed inside a host to metabolism needed in independent growth in impoverished media (24, 59, 61). This change could have happened quite abruptly with the advent of Lrp during evolution.

The various DNA-organizing proteins may be able to replace each other in their nonspecific roles. Thus, an lrp mutant overproducing Crp might be in better physiological shape than the lrp mutant itself: it perhaps could not turn on the glt operon, for instance, but it might be able to maintain a more normal chromosome structure. A striking example of protein interchange of this sort is the ability of the lymphoid enhancer binding factor LEF-1 to replace IHF in bending E. coli DNA, provided that an appropriate binding sequence is available (35). If Lrp, Crp, IHF, and H-NS fill similar DNA organizational functions, overproduction of Lrp in a mutant lacking one of the other proteins might improve the cell’s physiology. Conversely, a double mutant lacking two of the proteins might be worse off than one would predict from the characteristics of each single mutant. The synergistic inefficiency observed when lrp is coupled with various other mutations (see above) may reflect overlapping organizational functions of this sort.

This point of view implies that proteins like Lrp may bind in a physiologically significant but nonregulatory manner at DNA sites other than promoters. Suggestive evidence for this possibility comes from the observation that Lrp retards 5 of the 11 HindIII fragments from pBR328 (R. T. Lin and E. B. Newman, unpublished observations). In vivo DNA protection studies are generally done with promoter DNA from suspected target operons. It would be of interest to do such studies with coding DNA, particularly genes containing sequences similar to that for the proposed Lrp consensus binding site (see below).

 

KNOWN OPERONS THAT ARE REGULATED BY Lrp

A number of specific members of the leucine/Lrp regulon have been identified. They have been found among genes for which expression is known to be affected by exogenous leucine, among genes specifying proteins that are formed at altered rates in an lrp mutant, as determined on two-dimensional gels, among genes downstream of unmethylated GATC sequences (38), and among λplacMu insertions in Lrp-regulated genes that are identified by the phenotype of the insertion mutation or by direct determination of the flanking DNA sequences. A few more Lrp-regulated operons (lrp, gcvTHP, malT, and lacZYA) were established by good guesses. All known genes that had been shown to be regulated by Lrp as of February 1995 are listed in Table 1 and discussed below. With recent interest in Lrp and the availability of an easily transduced lrp::tet mutation, many investigators are beginning to screen genes of interest for regulation by Lrp. This tool has added to the list of genes known to be regulated by Lrp and also to a list of genes that have been shown to be unresponsive to Lrp (Table 1).

 

Biosynthetic Operons

ilvIH (Isoleucine/Valine Biosynthesis).

The ilvIH operon specifies acetohydroxy acid synthase III (AHAS III), one of three isoenzymes catalyzing the carboligase step in branched-chain amino acid synthesis. Transcription of ilvIH is activated 30-fold by Lrp, and exogenous leucine almost completely abolishes activation (83, 90, 102, 118). A strain lacking Lrp is physiologically deficient in the enzyme, although this has little phenotypic effect, presumably because the isoenzyme AHAS I, specified by ilvBN, is expressed at a sufficient level in the absence of Lrp. (AHAS II is not functional in E. coli K-12 because of a frameshift mutation in ilvG [55].) The interaction of Lrp with the ilvIH promoter is by far the best studied of the Lrp-DNA interactions (see below).

serA (Serine Biosynthesis).

The serA gene codes for phosphoglycerate dehydrogenase, the first enzyme specific to serine biosynthesis. Its transcription is activated sixfold by Lrp, and leucine reduces expression twofold. The sixfold decrease in an lrp mutant does not make the strain auxotrophic for serine in aerobic growth at 37°C. However, SerA activity seems to be limiting since addition of serine to the medium increases the growth rate (2). Furthermore, serine auxotrophy is observed in several double mutants such as lrp relA mutants (2; see above).

leuABCD (Leucine Biosynthesis).

Leucine synthesis seems to be limiting in lrp mutants since adding leucine to the medium increases the growth rate (2, 61). Evidence that the leuABCD operon is activated by Lrp comes from studies of Lrp-regulated λplacMu insertions that confer leucine auxotrophy and are located in the leuABCD operon at 2 min (59, 107). These fusions show 11-fold stimulation by Lrp (59).

The leuABCD operon also has an attenuator that results in higher expression when leucine is limiting (117), and leucine limitation stimulates leu::lacZ expression, even in the absence of Lrp (61). As a result, it is difficult to evaluate possible effects of exogenous leucine on the Lrp component of leu regulation. This is the only clearly documented case of a transcriptional effect of exogenous leucine in the lrp mutant.

glyA (Glycine Biosynthesis).

The glyA gene specifies the enzyme SHMT, forming glycine and mTHF from l-serine. In wild-type E. coli growing in minimal glucose medium, this is the principal source of glycine, as indicated by the absolute glycine requirement of glyA mutants (81). Studies with a glyA::lacZ fusion indicate that Lrp represses glyA transcription (M. San Martano and E. B. Newman, unpublished results).

gltBDF (Glutamate Synthase).

The gltBDF operon codes for a regulatory protein (GltF) and the two subunits of glutamate synthase, an enzyme that, together with glutamine synthetase, is responsible for assimilation of ammonia at low external concentrations. In the lrp mutant, little or no glutamate synthase activity is detected, and the spot corresponding to GltD is absent in two-dimensional protein electrophoretograms (24). This deficiency causes a typical Glt^– phenotype (see above). The expression of a gltBDF::lacZ fusion was decreased 2.2-fold in the presence of 10 mM leucine and 44-fold by Lrp deficiency (25). Regulation of this promoter in vitro by Lrp and leucine has been considered in detail (25).

 

Degradative Operons

sdaA (l-Serine Deaminase).

E. coli K-12 makes two quite similar l-SDs, which convert l-serine to pyruvate and ammonia (106). l-SD I, product of the sdaA gene, has complex regulation, with expression stimulated by heat shock, anaerobiosis, UV irradiation, and leucine and repressed 7- to 10-fold by Lrp in glucose minimal medium (45, 69). In the lrp mutant, the higher level of l-SD permits growth with l-serine as sole carbon source. l-SD II is the product of sdaB, the second gene of the sdaCB operon. It is expressed only in rich medium, in which it is regulated primarily by Crp (96, 97). Its expression is not significantly affected by exogenous leucine or by an lrp mutation. It is curious, however, that transcription of the upstream sdaC gene is activated by Lrp (see below). The apparent lack of effect on sdaB may be due in part to its very poor ribosome-binding site, with enzyme levels too close to background to detect Lrp activation (97).

kbl-tdh (Threonine Degradation).

E. coli can degrade threonine to acetyl coenzyme A and glycine in two steps, by oxidation to 2-amino-3-ketobutyrate followed by cleavage (Fig. 1). The two enzymes, threonine dehydrogenase and 2-amino-3-ketobutyrate coenzyme A lyase, are specified by the tdh and kbl genes, respectively, which form an operon. The kbl-tdh operon is repressed about 20-fold by Lrp, and leucine partially relieves repression (59, 89). This operon is not normally expressed in minimal medium, as evidenced by direct enzyme assay and by the fact that glyA mutants, lacking SHMT and thus unable to form glycine from serine, have an absolute glycine requirement which cannot normally be satisfied by threonine (27). Overexpression of the kbl-tdh operon permits formation of glycine and serine from threonine and enables the cell to grow with threonine as carbon source (19, 88).

lacZ (β-Galactosidase).

Transcription of the lacZ gene is decreased by about 30% in the lrp mutant (107). However, the growth rate on 0.2% lactose is not decreased in the absence of Lrp. It is not known whether this modest transcriptional stimulation reflects a direct interaction of Lrp with the lac promoter or an indirect effect.

 

Other Metabolic Operons

gcvTHP (Glycine Cleavage).

The gcvTHP operon codes for the glycine cleavage system, which converts glycine to mTHF, NH₃, and CO₂, producing both ammonia and one-carbon units from glycine. The operon is activated by Lrp: an lrp mutant makes 20 times less enzyme and is physiologically defective in glycine cleavage, i.e., unable to obtain either one-carbon units or ammonia from glycine (61). Exogenous leucine has no significant effect on gcv expression. Lack of glycine cleavage activity in the lrp mutant has profound effects on its physiology (see section above).

pntAB (Pyridine Nucleotide Transhydrogenase).

Pyridine nucleotide transhydrogenase, product of the pnt gene, catalyzes the reaction nH⁺ _in + NADPH + NAD ↔ nH⁺ _out + NADP + NADH, permitting E. coli to replenish its NADPH pool independently of the pentose phosphate shunt or, alternatively, to extrude protons at the expense of NADPH. Transcription of the pntAB operon is decreased four- to fivefold by leucine and to a lesser degree by alanine and methionine (36). In the lrp mutant, it is decreased four- to sixfold (2; see above). Thus, the pntAB operon is activated by Lrp, and activation is reversed by leucine. The decreased Pnt level in the lrp mutant has no known effect on the phenotype. However, lack of both Pnt and Lrp causes unexpected auxotrophies (see above).

lysU (Lysyl-tRNA Synthetase).

E. coli has two lysyl-tRNA synthetases, LysU and LysS (42). This is unusual since each of the 19 other amino acids has a single synthetase. The two genes are regulated very differently: lysS expression is constitutive and Lrp independent, whereas the lysU gene is highly regulated, with expression stimulated by heat shock, anaerobiosis, low external pH, and leucine and repressed ninefold by Lrp (33, 47, 56, 60, 68). Leucine relieves Lrp repression, and the stimulation by leucine is accompanied by a fivefold increase in lysU mRNA (47, 56). Alanine also relieves repression (42). Heat induction of lysU is independent of Lrp but requires a small mRNA immediately downstream of the initiation codon (68). Under most laboratory growth conditions, the constitutive synthetase LysS seems to be sufficient for normal growth; LysU alone (in a Δ lysS mutant), on the other hand, does not support wild-type growth rates, especially at low temperatures (33).

 

Transport Operons

livJ and livKHMGF (Leucine Transport).

E. coli has multiple transport systems for leucine. Two of these have high affinity for leucine, involve periplasmic binding proteins of different specificities, and are regulated by Lrp. The livJ gene product binds leucine, isoleucine, and valine, whereas the livK gene product is specific for leucine (3, 4, 52). These two systems share a set of membrane components, products of the livHMGF genes (in the livKHMGF operon). The two operons, adjacent on the genetic map, are both turned off in the presence of high leucine concentrations, and the turnoff requires Lrp. This regulation has been established by direct transport measurements (51, 52, 78), by analysis of proteins on two-dimensional gels (24), by a study of livJ::lacZ and livK::lacZ fusions (40), and by identification by inverse PCR and sequence analysis of the Lrp-regulated insert in strain CP36 (61, 107). This pattern of regulation—repression only in the simultaneous presence of leucine and Lrp—suggests that the active repressor is effector-bound Lrp. Biochemical studies are needed to determine whether Lrp action on these two operons is direct.

Early work indicated that the LivJ transport system can also be repressed by exogenous methionine (37). Similarly, in studies of homoserine transport (apparently via the LivJ system), it was observed that exogenous methionine or alanine could repress expression of the transport system, albeit to a lesser degree than exogenous leucine (108).

sdaC (Serine Transport).

E. coli also has multiple transport systems for l-serine. One, highly specific to serine, is induced in the presence of leucine (39). The sdaC gene product has not been identified. However, its nucleotide sequence, together with the altered transport characteristics of an sdaC mutant, are consistent with its being an l-serine transporter (96). Expression of sdaC is stimulated two- to threefold by exogenous leucine, and this stimulation requires Lrp. Thus, sdaC may well specify the leucine-inducible, serine-specific permease. As in the case of the liv operons, the pattern of regulation is unusual—activation only in the simultaneous presence of leucine and Lrp—and suggests that the activator is effector-bound Lrp. Again, biochemical data are needed to determine whether Lrp action is direct.

oppABCD (Oligopeptide Transport).

E. coli and Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium) have several oligopeptide permeases that are able to take up dipeptides, tripeptides, and larger peptides (5). The system specified by the oppABCD operon has a periplasmic binding protein and transports a wide range of tripeptides. In E. coli (but not in S. typhimurium), its expression is increased by the presence of leucine in the growth medium; it is also increased by the presence of alanine (6). In an lrp mutant, the operon has high constitutive expression, suggesting that it is repressed by Lrp (8).

malEFG , malK-lamB-malM, and malT (Maltose Transport).

Maltose uptake in E. coli requires four proteins, products of the malEFGK genes, which are arranged in two operons, malEFG and malK-lamB-malM, and transcribed divergently from a 297-bp intergenic region (9, 21). Transcription of the mal genes is under the control of several factors, including Crp and a specific transcriptional activator, MalT. Transcription of the two uptake operons is decreased 50 to 70% in the lrp mutant grown in glycerol (107). Although it exhibits slightly lower expression of the maltose transport proteins, the lrp mutant grows normally in 0.2% maltose. It is possible that Lrp activation is important only at low (noninducing) maltose concentrations, when it may help the cell to scavenge the sugar and form maltose triose, the endogenous inducer (85).

The transcription of malT is also decreased 50% in the lrp mutant. This in itself should reduce expression of the other mal operons. However, there is probably a further Lrp effect at the individual malE and malK promoters, as suggested by two observations. First, loss of Lrp does not reduce the expression of the MalT-regulated malPQ operon, which specifies maltodextrin phosphorylase and amylomaltase, involved in maltose catabolism. Second, leucine decreases the expression of the malE and malK operons some 35% but does not affect malT transcription (107).

ompF and ompC (Outer Membrane Porins) and micF.

The outer membrane of gram-negative bacteria is relatively permeable to molecules of molecular weight less than about 600, allowing passage into the periplasmic space. This permeability is primarily via porins, outer membrane proteins that form hydrophilic channels of broad specificity. In E. coli and S. typhimurium, the principal porins are the ompC and ompF gene products. Their synthesis is under complex control, including a response to the osmolarity of the medium: at high osmolarity, the synthesis of OmpC increases and that of OmpF decreases (49). This decrease of OmpF, at least at some levels of osmolarity, is thought to involve micF RNA, an antisense RNA that reduces the translation of ompF mRNA (87). A comparison of two-dimensional protein gels of proteins from wild-type and lrp mutant cells suggests that synthesis of OmpC is repressed by Lrp whereas that of OmpF is stimulated (24). The effect of Lrp at ompC seems to be direct; however, the effect on ompF stems from a negative effect of Lrp on the expression of micF (26).

 

Operons Specifying Fimbriae

E. coli can express many types of external appendages known as fimbriae, a term used interchangeably with pili (see chapter 11). Which appendages are produced in particular conditions is subject to an astonishing variety of regulatory mechanisms, including Lrp in most cases studied. A number of common laboratory strains do not express these genes, many of which are carried on plasmids.

Of the many operons specifying fimbrial components, the following have already been shown to be regulated by Lrp: the E. coli chromosomal operons pap (15, 75), sfa (112), daa (10, 112), fim (11, 31), and csg (chapter 11), the E. coli plasmid operons fan (15) and fae (43), and the S. typhimurium plasmid operon pef (30; chapter 11). The involvement of Lrp in several others has not yet been tested. In an lrp mutant, expression of a fanABC::lacZ fusion was reduced 70-fold, making this among the operons subject to the strongest Lrp activation (15). Addition of leucine (or alanine) decreases transcription of the fan operon 10-fold but has no effect on pap transcription.

With respect to Lrp action, the best-studied fimbrial system is the pap operon, which is the basis for an elegant model of Lrp action, described briefly later in this chapter and in detail in chapter 11.

 

Other Operons

osmY.

Lrp is involved in the regulation of the osmY gene in a rather complex manner. Expression of osmY is normally induced on entry into stationary phase, particularly in LB broth-grown cells, under the control of a stationary-phase σ factor, σ ^S. It is also induced in media containing high salt (98). During growth in LB broth, expression of an osmY::lacZ fusion is induced earlier and to a greater extent in the lrp mutant than in the wild type. The effect in minimal medium, where osmY expression is not detectable during exponential phase, was even more dramatic, an osmY::lacZ fusion being induced fivefold in an lrp mutant (53). Strains defective in the gene coding for the stationary-phase sigma factor, rpoS, had very little expression of osmY. However, an lrp rpoS double mutant expressed osmY constitutively throughout the growth cycle. This was interpreted as indicating that Lrp represses expression from osmY during exponential phase, and σ ^S overcomes this repression (53). The interaction of Lrp with other factors is discussed below.

o489.

An open reading frame, o489, of unknown function has been shown to be regulated by Lrp. It is located immediately downstream of the uspA gene, which is not regulated by Lrp (76). This open reading frame has been shown by inverse PCR to carry the Lrp-regulated insert in strain CP59 (61, 107). Lrp regulation of o489 has also been shown in a screening for environmentally regulated methylation of GATC sites (38). The deduced o489 product is exceptionally hydrophobic, suggesting that it may be membrane bound, as a channel or porin.

 

MOLECULAR ASPECTS OF Lrp FUNCTION

 

Definition of Lrp Domains

The Lrp molecule has been suggested to consist of three domains: a DNA-binding domain in the N-terminal 40% of the protein, a transcription activation domain in the region constituting 40 to 80% of the protein, and overlapping this domain, a leucine response domain in the C-terminal third (82). This structure was suggested in a detailed study in which a plasmid-carried lrp gene was subjected to in vitro mutagenesis and introduced into a strain carrying a functional chromosomal lrp gene and a chromosomal ilvIH::lacZ fusion (82). Mutations affecting expression of the fusion, selected for loss of response to exogenous leucine (seven alleles), defined a leucine response domain in the C-terminal third of the molecule. Mutants with decreased expression of the ilvIH::lacZ fusion (15 alleles) were divided into two classes, according to whether crude extracts could retard ilvIH DNA. Those that retarded were considered activation mutants; those that did not defined a DNA-binding domain in the N-terminal third of the molecule, many in a putative helix-turn-helix region.

 

Interactions of Lrp with Specific Promoters

Certain Lrp-regulated promoters have been studied in some detail, giving rise to quite specific models. This work is only beginning. Nonetheless, it is perhaps surprising that none of the models seems to be generalizable in detail to all Lrp-regulated promoters. The interaction of Lrp with the promoter of its own structural gene was described above. Here, we review in detail the Lrp interactions in two operons, ilvIH and pap, studied by the groups of Calvo and Low. We then compare the much less intensely studied interactions at gcv, serA, gltBDF, lysU, and tdh.

ilvIH in E. coli.

Normal regulation of ilvIH requires a long sequence upstream of the promoter: a 331-bp sequence suffices to give normal regulation of transcription initiation and activation by leucine, but a 200-bp sequence does not (41, 113). Two in vitro transcription start sites have been located by primer extension: P1, located 31 bp upstream of the ATG codon that begins the coding sequence, and P2, some 60 bp further upstream, both within the upstream region defined by binding studies (118).

In vitro, P2 is repressed and P1 is activated 2.7-fold by Lrp (half-maximal activation at 15 nM Lrp) (118). In vivo, the parental strain grown in glucose minimal medium transcribes ilvIH from P1 only. Adding leucine during growth decreases transcription four- to sevenfold. In the lrp mutant, transcription from P1 is decreased even further (7- to 14-fold). No in vivo transcription from P2 has been reported.

Lrp binding has been studied both by gel retardation and by footprinting (90, 113, 114, 118). Lrp binds cooperatively to several sites and is thereby thought to activate transcription (113). The Lrp-binding sites comprise two regions upstream of the ilvIH promoter: an upstream region between –255 and –215, which contains two high-affinity Lrp-binding sites, and a lower-affinity downstream region (–101 to –56) which contains four Lrp-binding sites (90, 113). In gel retardation studies, the downstream region is 50% saturated at 45 nM Lrp, and the upstream region is saturated at a much lower concentration, 8 nM. Although the upstream promoter is not used for in vivo transcription, the upstream sites must be present and saturated for transcription to occur (113). Detailed studies showed cooperative binding of Lrp to the two upstream sites (1 and 2) and to three of the four downstream sites (3, 4, and 5) but not to the sixth site. In vivo footprinting studies similarly indicated that Lrp binds to sites 1 through 5 but not to site 6 during growth in glucose (67). Palindromic motifs were found in sites 2 and 6, and a consensus for the six sites suggested the sequence 5'-AGAATtttATTCT-3' (but see below).

The functional importance of the various binding sites was assessed by mutating each site in turn and determining the effect on DNA binding and on in vivo expression from an ilvIH::lac fusion (113). Mutations in either site 1 or site 2 result in lowered binding of Lrp and decrease cooperativity between sites 1 and 2, and they reduce β-galactosidase activity from ilvIH::lacZ to about 65%. This observation is difficult to interpret, however, because a mutation in the area between the upstream and downstream regions reduces transcription to a similar extent.

A similar analysis was made for the downstream region. Mutations in sites 3, 4, and 5 decrease in vivo expression much more than upstream mutations; indeed, mutations in these sites reduced expression to 15%, the level seen when the entire control region is deleted. This interesting study will lead to more definite conclusions with more mutations in each site. Isolation of these mutants suggests that Lrp regulation directly activates transcription at the ilvIH promoter. Results of similar studies of two mutations in the upstream region of serA which also show altered Lrp binding and altered expression of serA::lacZ support the conclusion that Lrp acts by binding to the target promoter (see below).

Lrp has been shown by circular permutation studies to bend DNA carrying a single binding site (site 2) by 52°, with the bending center at TTTT flanked by a four-base palindromic motif (114). It bends a DNA fragment carrying two sites (1 and 2) by 135° (114). Combined with the results of footprinting studies, these results suggest that DNA might be wrapped around multiple Lrp proteins, or that Lrp binding to multiple sites might loop the DNA. The requirement for a stereospecific alignment between the promoter and various cis-acting sequences supports the idea of a multiprotein complex at this promoter (93).

ilvIH in S. typhimurium.

An Lrp-like protein regulating ilvIH is also present in S. typhimurium (118). The ilvIH operon is cryptic in S. typhimurium, producing ilvIH mRNA but no functional enzyme (102) owing to a TGA stop codon at position 12 of the ilvI gene(E. Ricca, cited in reference 118). That an Lrp-like protein may be involved in its regulation was indicated by the fact that growth with leucine repressed ilvIH mRNA 2-fold, compared with 4- to 10-fold in E. coli. The likely existence of an S. typhimurium Lrp was indicated by the retardation of DNA fragments carrying either S. typhimurium or E. coli ilvIH DNA by S. typhimurium crude extracts, forming a similar pattern.

An upstream and a downstream region defined with E. coli Lrp retardation of S. typhimurium DNA were subdivided into six binding sites by footprint analysis. Two of three upstream sites are placed as in E. coli, with a third slightly further upstream. Downstream sites roughly coincide with E. coli sites. It therefore seems that the ilvIH promoters in both species are organized with multiple binding sites.

The papBA and papI Promoter: GATC Methylation.

Expression of pyelonephritis-associated fimbriae is subject to phase variation, with two relatively stable epigenetic states, ON and OFF. Regulation of the two divergent pap operons involves Lrp, the PapI and PapB proteins, Crp, and the methylation state of two sequences in the intergenic region, GATC1 and GATC2, separated by 102 bp (13, 15, 75). In the ON phase, GATC1 is unmethylated (and GATC2 is methylated), whereas in the OFF phase, GATC2 is unmethylated (and GATC1 is methylated). Lrp binds to a region including GATC2. When the DNA is in the OFF state, this binding is responsible for keeping GATC2 unmethylated (15). If GATC1 is unmethylated, then addition of PapI leads to protection of this site, whereas GATC2 now becomes exposed enough to be methylated. However, binding of Lrp and PapI to GATC1 is inhibited if the site is methylated (14, 75). An interesting model suggests that in phase ON cells, a multiprotein complex bends the DNA and positions RNA polymerase such that both papBA and papI can be transcribed (112; see also chapter 11).

This elegant regulatory mechanism has been extended to the fimbrial operons sfa and daa (112). It may also hold for the afa, prf, and (in S. typhimurium) pef operons, which have similarly placed GATC sequences and PapI-like proteins (chapter 11). However, it is not general, even for all of the fimbrial operons: the fim and fan operons, although both activated by Lrp, lack GATC motifs and a PapI-like protein (15, 32). The fim genes are regulated by DNA inversion, under Lrp control (32). There is, however, another gene regulated by both Lrp and methylation, the uncharacterized o489 (38, 107; see above).

The serA Promoter.

Two in vivo transcription start sites have been located upstream of the serA gene: P1, 45 bp from the translation start site, and P2, 93 bp further upstream (58). In contrast, ilvIH uses only one of its two promoters in vivo. P2 is repressed by Lrp and therefore used only in the lrp mutant or in growth medium in which Lrp is made at a low level, e.g., in LB broth. P1 is activated by Lrp and therefore used in the wild type grown in glucose minimal medium. Two Lrp-binding sites have been established by gel retardation and footprinting. An upstream high-affinity site (–155 to –81) is more or less in the position corresponding to that of the downstream low-affinity ilvIH site of E. coli. The serA low-affinity site is less well localized, starting somewhere downstream of –82. Binding of Lrp at the upstream site would be expected to block P2, leaving P1 free to be transcribed. This is similar to the situation for ilvIH, in which binding at the low-affinity site would block P2.

The serA promoter has been subcloned both as a single fragment incorporating both promoters and as two fragments, a downstream fragment (–140 to +10) and an upstream fragment (–370 to –70), all fused to lacZ (J. Zhang and E. B. Newman, unpublished results). The fragment carrying both promoters contains all signals needed for regulation, as judged by a 46% decrease in expression by leucine and a 72% decrease in an lrp-deficient cell, similar to the 30 and 93% decrease seen in the intact chromosome. Expression from the downstream promoter alone is reduced to 15% of that of the two-promoter fragment, and that of the upstream promoter is reduced to only 3%; therefore, the entire sequence is clearly needed for full expression of either promoter. Lrp represses the downstream promoter (expression increased 3.5-fold in the lrp mutant) and slightly activates the upstream promoter (63% expression in the mutant); in both cases, leucine partially alleviates Lrp action.

Other Promoters: gcvTHP, gltBDF, lysU, and tdh.

Lrp activation of the gcv operon depends on a sequence that lies between –466 and –169 (103). Between –92 and –229, multiple sites for Lrp binding were demonstrated by both mobility shifts and DNase I footprinting (120).

At the gltBDF promoter, Lrp binds to a 666-bp fragment from –322 to +344 (25). Binding studies with a somewhat longer fragment indicate that leucine does not abolish binding but increases the apparent dissociation constant. At high concentrations of Lrp, leucine has no effect.

At lysU, Lrp binds to a fragment 51 to 158 bp upstream of the translation start site, as judged by retardation and DNase I protection (60). Leucine reduces this binding at concentrations similar to those effective at the ilvIH upstream, high-affinity site (60). DNase I and hydroxyl radical footprinting experiments indicated that several Lrp molecules bind in a highly cooperative manner to a DNA region of over 110 bp which encompasses the –35 box of the lysU promoter (34). A series of mutations which resulted in parallel effects on the strength of Lrp-DNA association in vitro and on the degree of repression of lysU expression by Lrp in vivo all fell in this region, particularly in AT-rich runs which the authors considered as indicative of Lrp binding (34).

Deletions upstream of the tdh promoter define an area from –69 to –44 bp as responsible for leucine regulation (89). Deletion of this area makes the operon unresponsive to Lrp. A mutation in a putative consensus sequence just upstream of this area (–72 to –61) causes a loss of leucine regulation.

 

Lrp Interactions with Other Regulatory Factors

Particular models have been proposed for interactions at ilvIH and pap promoters (93, 111, 114). It seems unlikely that any factor acts alone at an E. coli promoter. Binding factors interact with each other and with RNA polymerase, and the effect of these factors changes drastically with slight changes in spacing. Lrp is probably no exception to this rule (23), the more so since it bends DNA (114). Upstream of sdaC there are three IHF-binding sites, not yet shown to be functional, that may be involved in forming such complexes. It was suggested that H-NS modulates Lrp activity at ilvIH (57) and at both kbl-tdh and gcv (50). The gcv promoter seems to be exceedingly complex, responding to several regulators (103, 120). Regulation of the osmY gene seems to involve displacement of Lrp by RNA polymerase associated with σ ^S, the stationary-phase sigma factor (53). Lrp probably acts together with a variety of other regulatory proteins to form elaborate transcription complexes that differ at different target promoters, depending on the regulatory responses required at the particular promoter and probably also on evolutionary happenstance.

 

Lrp Consensus Sequence

In a study of the regulation of the kbl-tdh operon by leucine and Lrp, Rex et al. (89) showed that the region from 44 to 69 bp upstream of the transcription start was involved. Comparing the sequence of this region with that of the ilvIH upstream region, they proposed an asymmetric 12-bp Lrp consensus binding sequence, TTTATTCtNaAT, which they also found, in both orientations, upstream of other Lrp-regulated genes. Deletion of this sequence upstream of the kbl-tdh operon resulted in high-level constitutive expression, independent of leucine and Lrp, strongly suggesting that the sequence is indeed part of the operator site recognized by Lrp (89).

Wang and Calvo (113), in a detailed analysis of the six Lrp- binding sites upstream of the E. coli ilvIH operon, found that those with strongest affinity for Lrp have a symmetric sequence of putative consensus AGAATTTTATTCT (with a TTT spacer in the middle). The 8 bp on the 3' end of this sequence match the 5' end of the consensus proposed by Rex et al. (89). The remaining ilvIH sites contain the central spacer together with a half palindrome, in either orientation (TTTATTCT or AAAATTCT). A comparison with the Lrp-binding sites of the S. typhimurium ilvIH operon suggested that the half motifs do not include the first base in this organism (114). Using chemically synthesized sites, Wang and Calvo (113) showed that Lrp binds well to the perfect palindrome in vitro, less well to an imperfect palindrome, and not measurably to half palindromes, whereas in natural DNA, half motifs near a full palindrome show strong cooperative Lrp binding. It seems reasonable to conclude that Lrp action in vivo involves more than a single half site.

Using an efficient method for detection of common motif in unaligned DNA sequences, Lisser and Margalit analyzed a set of 23 gene sequences transcribed under Lrp control and identified the putative consensus sequence (g/a)(g/c)nnnTTTATtCTgG (62). The core of this consensus sequence, TTTATtCT, is compatible with the two consensus sequences described above; however, there are differences in flanking regions.

In those promoters for which Lrp protection studies have been reported, the putative consensus sequence is generally (but not always) within the protected area. It would also be interesting to know how frequently this sequence can be found in genes not regulated by Lrp. In any case, further experimental studies are required to establish the function of these potential Lrp-binding sites.

 

Leucine Effects In Vitro

Lrp was discovered as the mediator of leucine effects, and not surprisingly, workers have systematically looked for effects of leucine on Lrp in vitro. Although purified Lrp binds leucine with a dissociation constant of 1.22 × 10^–5 M(65), leucine has at best a modest effect on the DNA-binding capacity of Lrp. Ricca et al. (90), in ilvIH binding experiments with partially purified Lrp, showed that 38 mM leucine reduced binding by only 55%. Willins et al. (119), with purified Lrp, showed that at low Lrp concentrations, binding to the ilvIH regulatory region was reduced (but not abolished) by the presence of 30 mM leucine, whereas at high Lrp concentrations 30 mM leucine had no effect. Lin et al. (60) showed that Lrp protection of lysU DNA was reduced about twofold in the presence of 20 mM leucine; again, at high Lrp concentrations, leucine had essentially no effect. In an in vitro transcription system with ilvIH DNA, Willins and Calvo (118) showed that purified Lrp stimulates transcription from the p1 promoter about threefold, and 37 mM leucine did not reduce this stimulation. Ernsting et al. (25), in careful Lrp-binding studies, found that 30 mM leucine reduces the affinity of Lrp 3.3-fold for gltBDF and 2-fold for ilvIH DNA.

The overall picture is that although leucine reduces the ability of Lrp to bind to DNA in vitro, the effect is modest, seen only at low Lrp concentrations and high leucine concentrations, well above physiological levels. Taken at face value, the data suggest that the true Lrp effector is not leucine itself. In E. coli, in which leucine is essentially a metabolic dead end, the only leucine derivatives known are the various leucyl-tRNA^Leu species; in fact, leucyl-tRNA synthetase is required for leucine-mediated repression of the LivJ high-affinity transport system (84). It will be interesting to see similar studies on other Lrp target operons, to test a possible role of leucyl-tRNA^Leu inLrp action. Determination of whether both free and effector-bound Lrp can be active, as suggested by the in vivo data, will require positive identification of the effector and further in vitro experiments.

 

Lrp AND LEUCINE EFFECTS IN OTHER ORGANISMS

The global nature of Lrp regulation in E. coli leads one to wonder whether it exists in organisms. Four non-E. coli lrp-like genes with nucleic acid sequence identity over 87%, and even higher protein sequence identity, have been reported: in Enterobacter aerogenes, Serratia marcescens, Klebsiella pneumoniae, and S. typhimurium (29). Various physiological effects of the non-E. coli molecules have been shown. In S. typhimurium, an Lrp-like protein has been shown to activate transcription of the cryptic ilvIH operon and, in crude extracts, to retard a DNA fragment carrying this operon (115; see above). Further indication of Lrp activity in S. typhimurium comes from the observation that opp::lacZ fusions introduced from E. coli still show induction by exogenous leucine (5), although the homologous S. typhimurium opp locus is not leucine inducible (48). Interestingly, a second oligopeptide transport system in S. typhimurium, coded for by the tppB locus, is induced by leucine (48).

Two less similar lrp-like open reading frames have been described. The Rhizobium species NGR324 carries a symbiotic plasmid, pNGR324a, involved in establishing symbiosis with a broad range of legumes. One symbiotically active plasmid open reading frame, of as yet unknown function, is highly homologous to Lrp (with 55% identity over a 241-nucleotide sequence) (80).

The least conserved of the known Lrp-like proteins is a Pseudomonas putida protein, BkdR, with 37.5% protein sequence identity (65, 66). It stimulates the expression of enzymes required to use branched-chain amino acids as carbon sources. A P. putida bkdR mutant is unable to grow on branched-chain amino acids, and growth is restored by plasmids carrying either the P. putida bkdR gene or the E. coli lrp gene. Functional complementation between Lrp and BkdR in these rather distantly related species suggests conservation of Lrp throughout the gram-negative bacteria. However, the very low level of BkdR protein has been interpreted to mean that it may not be a global regulator in P. putida (65). By using a bkdR DNA probe, a positive signal was detected in Pseudomonas cepacia, Agrobacterium tumefaciens, Rhodobacter sphaeroides, and Rhizobium leguminosarum but not in the gram-positive species Streptococcus mutans (65).

Serratia marcescens secretes several proteases across both membranes to the external medium. One of these is inducible. Its synthesis, first found to be induced in the presence of external protein, was later shown to be induced rather specifically by exogenous leucine (17, 64), which is presumably produced from protein either by the basal level of the inducible protease or by one of the other proteases secreted by this species. It is not known whether the induction by leucine involves an Lrp-like protein. Existence of an Lrp-like protein in Neisseria lactamica has also been suggested (54).

In eukaryotes, leucine is an allosteric effector of several enzymes, but excluding enzymes involved in leucine metabolism, we have not found any reports of enzyme synthesis being regulated by leucine. On the positive side, let us add that for higher eukaryotes at least, we have also failed to find any reports of proteins whose synthesis is not regulated by leucine.

For most of the members of the leucine/Lrp regulon, it is not clear why exogenous leucine should be taken as a signal. It is possible that the choice is to some degree arbitrary, leucine having been adopted simply as an indicator of active proteolysis in the immediate environment. Molecules used in this way to signal a general set of conditions via an "integrator" protein have been found in other organisms. For example, in Aspergillus nidulans, β-alanine induces the synthesis of enzymes needed for the degradation of acetamide, in addition to those needed to degrade β-alanine (7). Here, it is assumed that β-alanine is taken as a general indicator of decay in the environment and that acetamide is rarely if ever encountered without β-alanine. From this point of view, higher organisms may well have responses similar to the bacterial leucine/Lrp regulon but use a different effector.

Looking at the question slightly differently, it would also be of interest to know whether bacteria have additional Lrp-like regulons responding to other signals. E. coli is known to have a complex response to exogenous serine (110) and to 2-ketobutyrate and related compounds (20). The possibility that these or other molecules govern global responses has not, to our knowledge, been explored.