Module 3.6.1.8_2004_archived

Biosynthesis of Phenylalanine and Tyrosine

JAMES PITTARD* AND JI YANG

[SECTION EDITOR: GEORGES COHEN]

Posted November 15, 2004

Archived November 7, 2008

Department of Microbiology and Immunology, University of Melbourne, Victoria 3010, Australia

*Corresponding author. Phone: +61 3 8344 5696, Fax:+61 3 9347 1540, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

( see updated version 3.6.1.8 )

HISTORY

The pathway of biosynthesis of the aromatic amino acids is shown, for convenience, in three parts (Fig. 1 to 3). Figure 1 shows the "common pathway" leading to the synthesis of the branch point compound chorismate, and Fig. 2 and 3 show the two terminal pathways in which chorismate is converted to phenylalanine and tyrosine, respectively. The pathway to tryptophan biosynthesis is dealt with separately in EcoSal. The common pathway has sometimes been referred to as the shikimate pathway. An additional four terminal pathways lead from chorismate to the so-called aromatic vitamins folate, ubiquinone, menaquinone, and enterochelin, but these are not considered here.

Fig. 1Common pathway of aromatic amino acid biosynthesis.

Source: Adapted from reference 173a with the kind permission of the publisher.

Fig. 2Terminal pathway of phenylalanine biosynthesis.

Source: Adapted from reference 173a with the kind permission of the publisher.

Fig. 3Terminal pathway of tyrosine biosynthesis.

Source: Adapted from reference 173a with the kind permission of the publisher.

Identification of the various intermediates in the aromatic pathway was completed by the early 1960s. As with other pathways, studies of auxotrophic mutants were of key importance. The first intermediate to be identified was shikimate; its position in the common pathway was established when Davis (59) showed that this compound alone of 55 aromatic and hydroaromatic compounds tested could fulfill the requirement of certain aromatic auxotrophs of Escherichia coli for tyrosine, phenylalanine, and tryptophan. Other mutants, which failed to respond to shikimate, accumulated this compound in their culture fluids. In the same study, it was established that anthranilate could be substituted for tryptophan, phenylpyruvate could be substituted for phenylalanine, and under appropriate conditions, 4-hydroxyphenyl pyruvate could be substituted for tyrosine. The accumulation by auxotrophs of intermediates preceding the blocked reaction led to the identification of 3-dehydroquinate (DHQ), 3-dehydroshikimate, and shikimate 3-phosphate, all intermediates in the common pathway (63, 189, 229, 231).

The identification of a phosphorylated compound (shikimate 3-phosphate) in the culture fluids was an exception to the general experience with other phosphorylated intermediates, in which the compounds that accumulated in culture fluids were predominantly in the dephosphorylated form. Such was the case with enol-pyruvyl shikimate (EPS) and the intermediates in the tryptophan pathway, indoleglycerol (InG) and 1-(o-carboxyphenylamino)-I-deoxyribulose (62, 74, 93, 183). Neither the phosphorylated intermediates nor their dephosphorylated derivatives serve as growth factors for any mutant blocked prior to their formation in a pathway, since the phosphorylated compounds cannot enter the cell, and the dephosphorylated derivatives are generally not true intermediates. Shikimate is a notable exception. The branch point compound chorismate fails to act as a growth factor, presumably because it is not able to enter the cell.

The early reactions in the common pathway were more difficult to establish. A mutant strain of E. coli that accumulated shikimic acid was grown on glucose variously labeled with ¹⁴C. The shikimate was isolated and chemically degraded to establish the distribution of specific carbon atoms of glucose in the shikimate molecule. The results showed that three of the carbon atoms of shikimate were derived from glycolysis and four were derived from the pentose phosphate pathway (208).

The nature of these precursor compounds was established when erythrose 4-phosphate (E4P) became available (15) and it was demonstrated that cell extracts of E. coli could convert E4P and phosphoenolpyruvate (PEP) to DHQ. Furthermore, on fractionation, an enzyme preparation could be obtained that converted E4P and PEP to 2-keto-3-deoxy-d-arabo-heptonic acid 7-phosphate (206). This compound was later to be termed 3-deoxy-d-arabino-heptulosonic acid-7-phosphate (DAHP).

The identification of chorismate, the branch point compound, required the use of mutant strains blocked in each of the three terminal pathways. A strain of Aerobacter aerogenes (62-1) that was blocked in the second reaction of the tryptophan pathway and the first reactions of both the phenylalanine and the tyrosine pathways was isolated. Extracts of this strain were able to convert shikimate to anthranilate. When, however, glutamine was omitted from the reaction mixture, a new compound was formed. This compound could be extracted from the reaction mixture and used as a substrate for anthranilate production by extracts of a mutant blocked after the EPS phosphate (EPSP) step (94). Subsequent studies showed that the compound could be accumulated by strain 62-1 and that it could be converted enzymically into anthranilate, prephenate, 4-hydroxyphenylpyruvate, phenylpyruvate, and 4-hydroxybenzoate. The compound was called chorismic acid (92, 94). After the structure of chorismate had been established, it was shown that it could be formed from EPSP by cell extracts of E. coli (158). The position of EPSP in the pathway was further supported by the demonstration that cell extracts produced EPSP from shikimate 3-phosphate and PEP and could convert EPSP to phenylpyruvate (138).

All the reactions shown in Fig. 1, 2, and 3 have been confirmed in cell extracts; many of these experiments used purified enzyme preparations. Although much of the early work establishing the pathway used E. coli and, in some cases, A. aerogenes, the reactions apply equally to Salmonella enterica serovar Typhimurium. Recent interest in the pathway has been stimulated by the recognition that some of the enzymic reactions may be excellent targets for new antimicrobial compounds (200). There has also been extensive engineering of the pathway in E. coli, in an attempt to produce commercially viable processes for the production of aromatic amino acids, aromatic vitamins, or important nonaromatic intermediates (16, 89, 168). Research has also been stimulated by the realization that the attenuation of some pathogenic strains to create potential vaccines resulted from blocks in the pathway of biosynthesis of the aromatic amino acids and vitamins (122, 212, 219). Furthermore, the role of the EPSP synthase enzyme in glyphosate sensitivity and resistance in plants (79) has drawn attention to this particular reaction and others in the pathway in the search for new herbicides.

ENZYMES OF THE COMMON PATHWAY

The DAHP Synthases

Both E. coli and Salmonella have three isofunctional DAHP synthase enzymes (23, 32, 73, 152, 181, 196, 204). Each one is inhibited by a different amino acid, that is, DAHP synthase (TYR) by tyrosine, DAHP synthase (PHE) by phenylalanine, and DAHP synthase (TRP) by tryptophan. Whereas both DAHP synthase (PHE) and (TYR) can be inhibited to greater than 95% by the cognate amino acid, inhibition of DAHP synthase (TRP) by tryptophan does not exceed 60% (175, 182). The partial inhibition of DAHP synthase (TRP) ensures that even in the presence of all three aromatic amino acids there is a residual biosynthetic capacity able to provide enough chorismate for synthesis of the aromatic vitamins (222). Expression of each of the genes coding for the different isoenzymes is also subject to transcriptional control. The expression of aroH [DAHP synthase (TRP)] is repressed by the TrpR protein in the presence of tryptophan. One report (161) states that TyrR protein also influences the expression of aroH, but extensive studies in our laboratory have failed to confirm this (unpublished data). Expression of aroF [DAHP synthase (TYR)] is repressed by the TyrR protein in the presence of tyrosine. On the other hand, the expression of aroG [DAHP synthase (PHE)] shows only small changes in response to the aromatic amino acids, but its transcription is repressed by increased levels of TyrR protein (24, 26, 31, 58, 119, 175, 222, 223).

Throughout this chapter we have retained the genetic nomenclature that has been used in reporting these studies in E. coli and Salmonella. Recently, it has been argued (28) that, to compare what is happening in a wide range of organisms, a single system of nomenclature in which aroA indicates a gene for the first reaction of the pathway, aroB the second, and aroC the third and so on is required. Although the merits of this proposal are clear, its adoption for this chapter would only add unnecessary confusion; hence, we have retained the published nomenclature for the genes.

The different regulation of each of these three DAHP synthases by the three aromatic amino acids was used in a strategy to recover mutants lacking each one of these activities (220, 221). Each of the DAHP synthase enzymes has a requirement for divalent cations which can be met to different degrees by several divalent metals, including Mn²⁺, Cd²⁺, Fe²⁺, Co²⁺, Ni²⁺, Cu²⁺, or Zn²⁺ (210). Each of the enzymes has a high affinity for PEP (approximately 10 μM), and it has been proposed that within the cell PEP is normally bound to the enzymes. PEP has been shown to stabilize DAHP synthase during purification and storage and to protect against heat inactivation (152, 162, 196). In the absence of PEP the instability of the enzyme can be prevented by EDTA, and it has been shown that, in the absence of PEP, redox metal ions that normally activate the enzyme cause oxidation of the two active-site cysteinyl residues Cys⁶¹ and Cys³²⁸ (167). The various DAHP synthases show different rates of decay when cells enter stationary phase that may reflect differences in sensitivity to oxidation under conditions in which PEP may become limiting.

Whereas DAHP synthases (PHE) and (TRP) are both dimers, DAHP synthase (PHE) exists as a tetramer. The monomers are approximately the same size and show a great deal of sequence homology, indicating that the various forms have arisen as a result of gene duplication and divergence (182). These isoenzymes in E. coli and Salmonella belong to the A_1α group of DAHP synthases widely represented in gram-negative bacteria (211). The crystal structure of DAHP synthase (PHE) complexed with PEP and Pb²⁺ has been determined. The protein is a tetramer and each of the monomers is a (β/α)₈ barrel with an amino-terminal extension and a two-stranded β-sheet inserted between helix α5 and strand β6. The active site and the inhibitor-binding site have been identified and shown to be separated by at least 15 Å. The inhibitor-binding site comprises a cavity that lies outside the core of the barrel. Various residues, whose substitution affects feedback inhibition in one or other of the three isoenzymes, also cluster on the outer side of the β core of the barrel behind the β3 and β4 strands (199). Further studies involving the crystal structure of the enzyme complexed with Mn²⁺ and a substrate analogue 2-phosphoglycolate have revealed details of the active site of the enzyme (218). This has been further refined using a dimeric form of the protein and high resolution (199). Studies of crystals complexed with Mn²⁺, PEP, and phenylalanine have revealed the conformational changes transmitted from the inhibitor-binding site to the active site upon the binding of phenylalanine (201). Studies of the structures of the tyrosine-inhibitable and phenylalanine-inhibitable DAHP synthase enzymes of Saccharomyces cerevisiae revealed that they were almost identical in structure to the E. coli DAHP synthase (PHE). Furthermore it was established that a single amino acid residue at the base of the inhibitor-binding cavity determined whether the enzyme was inhibited by tyrosine or by phenylalanine. Although there were 61 amino acid residues that differed between the tyrosine- and phenylalanine-inhibitable enzymes, changing only one of these, a serine at the base of the inhibitor-binding cavity to a glycine, converted a phenylalanine-inhibitable enzyme into a tyrosine-inhibitable enzyme. An examination of DAHP synthase sequences from Candida, Aspergillus, Escherichia, and Haemophilus confirmed the relationship between a serine residue and phenylalanine inhibition and a glycine residue and tyrosine inhibition (104).

When cells are grown in minimal medium, DAHP synthase (PHE) is the major isoenzyme present (approximately 80%). Under conditions in which the aromatic amino acids limit growth, as in minimal medium supplemented with all amino acids except phenylalanine, tyrosine, and tryptophan, DAHP synthase (TYR) becomes the major enzyme. DAHP synthase (TRP) always remains a minor component even when expression of the gene is derepressed (68, 213).

DHQ Synthase

DHQ synthase, formerly known as 5-dehydroquinate synthetase, catalyzes the conversion of DAHP to DHQ. DHQ was first identified as an intermediate in aromatic biosynthesis in 1953 (229). It was isolated from the culture fluids of an aromatic auxotroph of E. coli. The active compound had previously been shown to support the growth of mutants blocked very early in the aromatic pathway. Its structure was determined, and it was shown to be chemically convertible to dehydroshikimate by heat at an acid pH (63). The enzymic conversion of DAHP to DHQ was first reported in 1963 with a partially purified enzyzme preparation from E. coli. The enzyme required both NAD⁺ and Co²⁺ for activity (207). Studies by Rotenberg and Sprinson (186, 187) with DAHP specifically labeled with tritium established that the cyclization step involves an interaction between the enol formed in phosphate elimination and the carbonyl at C-2 in an aldolase-type reaction. Kinetic isotope effects were observed at C-5, which supported a mechanism involving oxidation at C-5 by NAD. The finding that all the tritium of labeled DAHP is conserved in DHQ established that hydride transfer in the subsequent reduction of C-5 involved the same hydrogen atom that was taken from DAHP. NADH is enzyme bound and reduced the keto group at C-5 with the regeneration of NAD.

DHQ synthase has been purified to homogeneity from E. coli B and shown to be a single polypeptide with a molecular weight of 57,000. The native enzyme is a monomer, and the K_m for DAHP is 33 μM (145). The structural gene for this enzyme (aroB) was cloned and sequenced; the protein was overexpressed, and its amino acid composition and amino-terminal sequence were determined. From the sequence, the monomer is deduced to be a protein of 362 amino acids with a calculated M_r of 38,800 (154).

Frost et al. (82) used gene-cloning techniques to produce high levels of DHQ synthase for purification. They purified the enzyme to homogeneity and showed it to be a monomeric protein with an M_r between 40,000 and 44,000. In microbial eukaryotes, DHQ synthase activity is carried on a pentafunctional protein called AROM, which comprises the five central enzymes of the shikimate pathway. Carpenter et al. (34) have determined the crystal structure of the DHQ synthase domain of this protein and provided much structural detail relevant to the molecular reactions catalyzed by this enzyme. The extensive homology between this domain and the DHQ synthases of E. coli and Salmonella make it likely that enzymic mechanisms deduced from this structure also apply to these enzymes.

As with the other remaining enzymes of the common pathway (with the exception of shikimate kinase and EPSP synthase), DHQ synthase appears to be synthesized constitutively. Its synthesis is not repressed by any of the aromatic amino acids or by chorismate, nor is it induced by DAHP. The specific activity of DHQ synthase in extracts of E. coli is 25 mU/mg of protein. It has been calculated that this is about fivefold greater than would be required to meet the aromatic needs of cells growing with a doubling time of about 1 h (213). Strains of E. coli with feedback-resistant DAHP synthase (TYR) accumulate DAHP, indicating that under these conditions DHQ synthase activity has become rate limiting (164). DHQ synthase activity is inhibited by high concentrations of tyrosine (0.6 mM or greater) (16). Unexpectedly, the aroB gene is found to be part of a single transcription unit comprising seven genes of diverse functions. Although three of these, aroB, aroK, and trpS, have functions relevant to the synthesis or use of aromatic amino acids, there seems to be no particular advantage in this grouping. Detailed studies to identify transcriptional patterns under different conditions will be required before the proposed term of supraoperon is really justified (141).

DHQ Dehydratase

DHQ dehydratase, commonly known as dehydroquinase, catalyzes a cis elimination of water to convert DHQ to 3-dehydroshikimate and to introduce the first double bond of the aromatic ring (27, 103, 215). The enzymic reaction was first studied by Mitsuhashi and Davis (156), who partially purified the enzyme from both A. aerogenes and E. coli. They separated DHQ dehydratase from dehydroshikimate reductase (now known as shikimate dehydrogenase) and demonstrated the reversibility of the reaction. The K_m for DHQ was 44 μM. Among plants, fungi, and microorganisms there are two distinct classes of dehydroquinate dehydratase enzymes, each possessing different structures and utilizing different enzymic mechanisms (125, 197, 236). The enzymes present in E. coli and Salmonella belong to type I, whose members are exclusively involved in the synthesis of shikimate. Class II enzymes, found in plants and other microorganisms, are sometimes involved in biosynthetic reactions and sometimes in the catabolic pathways involving quinic acid. The DHQ dehydratase from E. coli was shown to be a simple dimeric protein with a subunit M_r of 29,000 (37). The structure of the DHQ dehydratase from S. typhi has been solved and the monomers have been shown to have an overall topology involving an eight-stranded α/β barrel (99). Important residues involved in proton abstraction and in the formation of the Schiff’s base intermediate have been identified (137). As previously mentioned for DHQ synthase, DHQ dehydratase appears to be synthesized constitutively. When E. coli and serovar Typhimurium are grown in minimal medium, DHQ dehydratase specific activities range from 50 to 100 mU/mg (95, 213). Examination of the E. coli genome sequence leads to the prediction that the gene for dehydroquinate dehydratase, aroD, is the third gene in a transcription unit, b1691, ydiB, aroD. The gene ydiB is a paralogue of aroE, the gene for dehydroshikimate dehydrogenase, and will be discussed in the next section. The function of b1691 is unknown. When overexpressing the aroD gene, Leech et al. (137) reported that removing the flanking chromosomal DNA by PCR and cloning it back into the vector increased expression about 10-fold, so that, although it seems that neither the aromatic amino acids nor the TyrR or TrpR regulator protein influences aroD expression, other systems may well play a part.

Shikimate Dehydrogenase

The enzyme that converts 3-dehydroshikimate to shikimate was first studied in 1954 with partially purified extracts of E. coli W. The reaction was studied in the reverse direction (shikimate to dehydroshikimate) in a coupled reaction with glutathione reductase to oxidize the NADPH that is formed. The reaction depends on NADP and is specific for shikimate. Shikimate dehydrogenase activity was not detected in extracts of aromatic auxotrophs that accumulated dehydroshikimate. The K_m of the enzyme for shikimate was 55 μM (251). The reaction is stereospecific, involving transfer of hydrogen from the A side of NADPH (56). Chaudhuri and Coggins (36) purified shikimate dehydrogenase from E. coli to homogeneity and reported that it is a monomeric protein with an M_r of 32,000. The structural gene (aroE) was cloned, and its sequence was determined. The amino acid composition and amino-terminal sequence of the purified protein were determined. These results indicate that the monomer is a polypeptide of 272 amino acids with a calculated M_r of 29,380 (6). The constitutive level of this enzyme in cells grown in minimal medium is about 60 mU/mg (213).

A search of the E. coli genome sequence revealed a second gene that encodes a protein YdiB, which shares 25% sequence identity with AroE throughout its entire length and is regarded as a paralogue of AroE. Both AroE and YdiB have been overexpressed and purified, and the structures of the purified enzymes have been determined (19, 143, 153). Whereas the aroE-encoded enzyme oxidizes shikimate with NADP⁺, it will not function with NAD⁺ as cofactor, nor is it able to use quinic acid as a substrate. On the other hand, YdiB can use either NAD⁺ or NADP⁺ and can utilize either shikimic acid or quinic acid as a substrate. Although the affinities of both enzymes for their respective cofactors and for shikimate are comparable, YdiB has a much lower catalytic activity and its physiological role in the cell is at the moment an enigma. Its presence in a transcription unit with aroD, which encodes the enzyme for the previous reaction in the pathway, would perhaps favor a role in the biosynthetic pathway that does not seem to be supported by the enzyme studies. The original E. coli mutant used to locate the aroE gene on the chromosome has a complete block in the pathway, requiring the aromatic vitamins and the aromatic amino acids for growth, but the status of ydiB in this strain has not been determined (177). The detailed studies of the structures of both AroE and YdiB have provided a great deal of information about the catalytic site and molecular transformations carried out by these enzymes (19, 153).

Shikimate Kinase

Shikimate kinase catalyzes the formation of shikimate 3-phosphate from shikimate and ATP. Shikimate 3-phosphate was first identified as a possible intermediate in aromatic biosynthesis by Davis and Mingioli (62). Its chemical structure was established by Weiss and Mingioli (231). No mutants blocked in the shikimate kinase reaction were found among aromatic auxotrophs of either E. coli (177) or serovar Typhimurium (95). The reasons for this became apparent when it was shown that extracts of serovar Typhimurium yield two separable peaks of shikimate kinase activity on DEAE-cellulose (159).Using ultracentrifugation in sucrose density gradients, Berlyn and Giles (20) confirmed the observation made in Salmonella spp. and established that E. coli also had two shikimate kinase enzymes. Further study of the E. coli enzymes established that the synthesis of one of these (called shikimate kinase II, because it was the second shikimate kinase to be eluted from DEAE-Sephadex) was subject to specific control involving the regulator gene tyrR. In particular, when cells were starved for tyrosine and tryptophan or when they carried an inactive tyrR regulator gene, synthesis of shikimate kinase II was derepressed about 10-fold, levels varying between 5 and 55 mU/mg. By studying the levels of constitutive synthesis by a tyrR mutant growing at different growth rates, predictions were made about the possible chromosomal location of the aroL gene. This was confirmed by the use of selective F' strains, and localized mutagenesis was used to isolate a mutant lacking shikimate kinase II activity (78). The isolation of this mutant facilitated the subsequent cloning and sequencing of the aroL gene (65, 67, 155). The protein translated from the sequence has an M_r of 18,998. The gene encoding shikimate kinase I (aroK) was identified as part of a transcription unit including aroB (140). The aroK sequence shows a 30% homology to shikimate kinase II over its entire length (235).

An examination of the kinetics of the enzymic reaction with fully purified shikimate kinase II and partially purified shikimate kinase I revealed important differences between the two enzymes. The K_m of shikimate kinase II for shikimate is 200 μM. On the other hand, the K_m of shikimate kinase I for shikimate appears to be in excess of 5 mM (67). This apparent low affinity of shikimate kinase I for shikimate explains why aroL mutants that lack shikimate kinase II activity will grow in minimal media only if the levels of DAHP synthase are high. Neither enzyme is inhibited by any of the end products. A further marked difference between the two enzymes is that mutants lacking aroK activity are resistant to the antibiotic mecillinam, whereas aroL mutants show no change in mecillinam sensitivity (217). The structures of the aroK-encoded enzyme from E. coli and shikimate kinase II from Erwinia chrysanthemi have been determined and shown to be structurally similar (130, 184).

The existence of isofunctional enzymes has been observed most frequently in the first reaction of a pathway that later branches to multiple end products, for example, aspartokinases for lysine, methionine, and threonine (45) and DAHP synthases for the aromatic amino acids. It is also true that specific regulation of gene expression most frequently affects genes specifying the first enzyme in a pathway. It is noteworthy that of all the genes of the common pathway of aromatic biosynthesis, only the three specifying the three DAHP synthase enzymes and the gene for shikimate kinase II are subject to control. Does this mean that in the evolutionary past aromatic biosynthesis started with shikimate, or does it indicate an as yet undiscovered pathway diverging from shikimate? The very low affinity of shikimate kinase I for shikimate raises the possibility that this enzyme has another major function unrelated to biosynthesis of the aromatic amino acids.

EPSP Synthase

EPSP synthase converts PEP and shikimate 3-phosphate into EPSP. It was the dephosphorylated form of this compound that was first identified in culture fluids of aromatic auxotrophs blocked in the last step of the common pathway (62). Levin and Sprinson (138) partially purified the enzyme activity from E. coli and showed that the product of the reaction was EPSP. The mechanism of this reaction was studied with partially purified enzyme from Salmonella spp. and shown to involve the transfer of the enolpyruvyl grouping unchanged to the acceptor molecule (22). This reaction introduces the three-carbon fragment that is destined to become the side chain of phenylalanine and tyrosine but to be replaced in the synthesis of tryptophan. The enzyme from E. coli has been purified to homogeneity (139). The subunit M_r was estimated to be 49,000 by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate, and the native molecular weight was estimated to be 55,000 by gel filtration, confirming that the enzyme is a monomer.

The gene for this enzyme, aroA, has been cloned, and strains carrying this gene on a multicopy plasmid have been used to produce EPSP synthase (77). The K_ms for PEP and shikimate 3-phosphate are 16 and 2.5 μM, respectively. The amino-terminal sequence and the amino acid composition of the purified protein have been determined. This information, in conjunction with the nucleotide sequence of the aroA gene, indicates that the primary translation product is a polypeptide of 427 amino acids with a calculated M_r of 46,112 (76). Intensive studies of the active site and the molecular details of enzyme catalysis have been prompted by the importance of this enzyme in plants where it is the target for the herbicide glyphosate. The three-dimensional structure of the enzyme has been determined in the presence of its substrate and also in the presence of the inhibitor glyphosate (195, 209).

An analysis of the DNA sequence of the region containing aroA coupled with polarity studies, in both E. coli and Salmonella, revealed that aroA is part of a transcription unit with serC (pdxF) in which serC (pdxF) is the first gene of the operon (111). A rho-independent terminator between serC (pdxF) and aroA results in an eight times greater amount of the single gene serC (pdxF) transcript than the cotranscript, also including aroA. Previous reports had suggested that expression of aroA was unaffected by either of the aromatic regulators TyrR and TrpR (95, 213). However, when E. coli was grown in media containing 20 amino acids, levels of EPSP synthase were found to be repressed about fivefold (213). Subsequently, a careful examination of expression of the serC (pdxF) promoter in a variety of media and involving regulator mutants crp and lrp has shown that the serC (pdxF)-aroA operon is activated by Lrp and repressed by Crp. The combined effects of these regulators provide maximum expression in glucose minimal medium and lowest expression in rich media like Luria broth (147).

Chorismate Synthase

The chorismate synthase reaction introduces a second double bond into the aromatic ring system and forms the branch point compound chorismate, which serves as the starting substrate for the three terminal pathways of aromatic amino acid biosynthesis and for the pathways to ubiquinone, menaquinone, folate, and enterochelin (92, 94). The enzyme has been partially purified from E. coli. It is oxygen sensitive and requires a reducing environment that can be established with an FADH₂- and NADH-regenerating system in an anaerobic environment. Activity is inhibited by iron chelators and by high levels of Fe²⁺. The basis for the latter effect is unknown, but it is postulated that Fe²⁺ is a cofactor for chorismate synthase. It is suggested that the reaction is irreversible. A tentative molecular weight for the enzyme of somewhere between 70,000 and 100,000 has been proposed (158).

Two reports of the cloning and sequence analysis of the gene for chorismate synthase (aroC) of E. coli have been published (35, 237). A difference between the two nucleotide sequences results in different predicted amino acid sequences for the 36 amino acids at the carboxyl end of the protein. Removal of these amino acids does not destroy enzyme activity, and so the reported difference may be a true polymorphism. The calculated M_r for the monomer from E. coli is either 39,138 or 38,183; the M_r for the monomer from serovar Typhimurium is 39,108. Gel filtration experiments show that the native enzyme is a tetramer (237). The secondary structure of the enzyme has been investigated using far-UV circular dichroism (CD) spectroscopy and is predicted to be an α-β barrel (142). Substrate and cofactor analogues have been used to probe the mechanism of the enzymic reaction (166).

Synthesis of the enzyme is believed to be constitutive in both serovar Typhimurium and E. coli. The specific activity of chorismate synthase in cell extracts is only 10 to 20% of that of the prior enzyme, EPSP synthase (95, 213). It is puzzling that the last reaction of the pathway appears to be rate limiting, unless such a situation facilitates subsequent controls directing chorismate along the various terminal pathways. Clearly, more work needs to be done on this important enzyme and its role in vivo in aromatic biosynthesis. An analysis of the E. coli genome sequence indicates that aroC may be expressed as part of a transcription unit comprising a number of putative genes, that is, yfcB, aroC, mepA, yfcA, b2326, and b2325. No detailed studies on this transcription unit have yet been reported.

ENZYMES OF THE TERMINAL PATHWAYS

The Chorismate Mutases

The first reaction of both the phenylalanine and the tyrosine pathways involves the conversion of chorismate to prephenate. The structure of prephenate was determined by Weiss et al. (230). This relatively unstable compound can be converted to phenylpyruvate at an acid pH, resulting in some puzzling results when phenylalanine auxotrophs were first studied. Simmonds (203) described a phenylalanine auxotroph that appeared on prolonged culture to provide its own auxotrophic requirements. Davis (61) and Katagiri and Sato (124) investigated this phenomenon and suggested that the substrate of the blocked reaction was being converted chemically to a precursor in phenylalanine biosynthesis. Davis showed that the accumulated substrate is prephenate and that the phenylalanine precursor is phenylpyruvate. The enzymes that carry out the synthesis of prephenate have been referred to as chorismate mutases. Cotton and Gibson (51) showed that chromatography of cell extracts of A. aerogenes and E. coli on DEAE-cellulose gave two well-separated proteins with chorismate mutase activity. Prephenate dehydratase activity was associated with the first component to be eluted, and prephenate dehydrogenase activity was associated with the second. Prephenate dehydratase carries out the second reaction of the phenylalanine pathway, and prephenate dehydrogenase carries out the second reaction in the tyrosine pathway (51). Subsequent purification of these enzymes from E.coli has confirmed that, in each case, both activities are the product of a single bifunctional enzyme (72, 90, 114, 127, 128).

The phenylalanine enzyme is referred to as chorismate mutase-prephenate dehydratase (encoded by pheA), and the tyrosine enzyme is referred to as chorismate mutase-prephenate dehydrogenase (encoded by tyrA). Serovar Typhimurium also contains a monofunctional chorismate mutase that is expressed in the periplasm of the cell. However, the role of this enzyme in aromatic amino acid biosynthesis has yet to be established (28).

Chorismate Mutase-Prephenate Dehydratase

Chorismate mutase-prephenate dehydratase is a homodimer with a subunit molecular weight of about 40,000 (13, 57, 90). Several studies indicate that chorismate mutase-prephenate dehydratase has two independent catalytic sites (13, 64, 177). The two enzymic activities can be inhibited differentially by chemical modifying reagents (91), by phenylalanine (72), and by substrate analogues (14). Kinetic studies and the use of radioactively labeled chorismate suggest that when the reaction is carried out with purified enzyme in vitro, prephenate dissociates from the mutase site and equilibrates with the bulk medium before combining at the dehydratase site (75). Inhibition of prephenate dehydratase by phenylalanine approaches 90% in both E. coli and serovar Typhimurium, and the associated mutase activity in both cases is inhibited 55% (72, 193, 194). The use of genetic engineering to express discrete domains from the overall protein has demonstrated that both mutase and dehydratase activities are present in a polypeptide comprising residues 1 to 285, but neither activity is inhibited by phenylalanine. Polypeptides comprising residues 101 to 300 retain full dehydratase activity but lack mutase activity and fluorescence emission spectra, and binding assays involving the wild-type protein indicate that residues 286 to 386 are crucial for phenylalanine binding (253). Site-directed mutagenesis coupled with the isolation of mutant proteins has identified four residues important for dehydratase activity (T278, N160, Q215, and S208) (254). Isothermal titration calorimetry, in conjunction with the study of certain selected mutant proteins, has identified two highly conserved regions, GALV (residues 309 to 312) and ESRP (residues 329 to 332), as important for phenylalanine binding and feedback inhibition (178). Earlier studies had reported that mutations located within codons 304 to 310 resulted in complete loss of feedback inhibition (163).

Chorismate Mutase-Prephenate Dehydrogenase

Chorismate mutase-prephenate dehydrogenase also forms homodimers with a subunit molecular weight of about 40,000 (114, 128). Chorismate mutase-prephenate dehydrogenase mutants that have lost dehydrogenase activity but retained mutase activity have been isolated (185). Although chemical modification of the purified enzyme causes parallel loss of both activities (114, 127), tyrosine differentially inhibits prephenate dehydrogenase activity (113), NAD activates chorismate mutase activity, and chorismate stimulates prephenate dehydrogenase activity (110). Kinetic and computer simulation studies support a single active site (109). However, the identification of an inhibitor that inhibits prephenate dehydrogenase activity without affecting chorismate mutase activity may indicate discrete binding pockets within separate or partially overlapping active sites (216). Tyrosine can cause up to 95% inhibition of prephenate dehydrogenase activity and, in the presence of NAD, up to 45% inhibition of the associated chorismate mutase activity (113). Site-directed mutagenesis has been used to identify residues critical for either mutase or dehydrogenase activity. Substituting lysine 37 with glutamine destroys mutase without affecting dehydrogenase activity, whereas changing histidine 197 to asparagine destroys dehydrogenase without affecting mutase activity (40). Attempts to identify discrete domains in the T protein by engineering peptide fragments, as had been done for the P protein, indicated that neither the mutase nor the dehydrogenase activity could be expressed in a fully functional form as a discrete contiguous subregion of the T protein. Tyrosine binding and feedback inhibition could not be attributed to a structural domain separate from the mutase and dehydrogenase domains, but it was possible to show that two carboxy-terminal sequences lacking approximately one-quarter of the T protein from the amino terminus had lost all mutase activity while retaining significant levels of dehydrogenase activity (38).

The gene for chorismate mutase-prephenate dehydrogenase (tyrA) is part of an operon with aroF and its expression is coordinately regulated by TyrR protein with tyrosine as a corepressor. The expression of the gene for chorismate mutase-prephenate dehydratase (pheA) is regulated by a system of attenuation that senses the availability of charged phenylalanyl-tRNA^Phe and is discussed later.

Because each of these enzymes and anthranilate synthase compete with each other for the same substrate, chorismate, it is of interest to compare the affinities of each for this substrate. The K_ms of highly purified enzyme preparations are as follows. Chorismate mutase-prephenate dehydratase from E. coli has a K_m for chorismate of 45 μM (72), whereas chorismate mutase-prephenate dehydrogenase from the same organism has K_ms for chorismate of 92 μM and for prephenate of 50 μM (114). In contrast, the K_m of anthranilate synthase for chorismate is 1.2 μM (12). This very marked difference in affinities implies that, under conditions of chorismate limitation, this compound would be preferentially directed down the tryptophan pathway rather than toward phenylalanine or tyrosine. This difference may explain why aromatic auxotrophs with incomplete blocks in any of the common pathway reactions seem able to meet their requirement for tryptophan before their requirement for phenylalanine and tyrosine (60).

The Aminotransferases

The last reaction in both the phenylalanine and the tyrosine pathways involves transamination of the respective α-ketoacids, phenylpyruvate, and 4-hydroxyphenylpyruvate, with glutamate as the amino donor. Early studies on aminotransferases in E. coli indicated that there were a number of aminotransferase enzymes and that these enzymes had rather broad specificities (39, 46, 150, 151, 157, 188, 202). Phenylalanine and tyrosine could be formed by at least two enzymes, one that was termed the aromatic aminotransferase and the other that was termed the aspartate aminotransferase. Synthesis of the aromatic aminotransferase could be repressed by tyrosine, but synthesis of the aspartate aminotransferase appeared to be constitutive. There was at least one more enzyme that could function in the conversion of phenylpyruvate to phenylalanine. The analysis of this rather complex situation has been greatly assisted by the isolation of mutants lacking individual aminotransferase enzymes (87, 88, 214) coupled with the development of methods for the separation of aminotransferase activities (150, 151, 179).

Three enzymes can contribute to aromatic amino acid biosynthesis: the so-called aromatic aminotransferase coded for by tyrB, the aspartate aminotransferase coded for by aspC, and the branched-chain-amino-acid aminotransferase coded for by ilvE. Double mutants inactivated in both tyrB and aspC require tyrosine but not phenylalanine for growth, whereas mutants lacking all three activities require both phenylalanine and tyrosine (88). The purification of the aspartate and the aromatic aminotransferase enzymes provided an opportunity to compare their properties. This has been done by Mavrides and Orr (151), by Powell and Morrison (179), and, more recently, by Hayashi et al. (105). Both enzymes are homodimers with subunit molecular weights of approximately 43,000.

The aspartate aminotransferase has a much higher affinity for aspartate than does the aromatic aminotransferase. On the other hand, the affinity of the aromatic aminotransferase for phenylalanine and tyrosine and their α-ketoacids is much greater than that of the aspartate aminotransferase. The aminotransferase reaction is freely reversible, and it seems that under normal physiological conditions phenylalanine and tyrosine syntheses are primarily carried out by the aromatic aminotransferase, the product of the tyrB gene. Only when the pools of phenylpyruvate and 4-hydroxyphenylpyruvate become very large will the aspartate aminotransferase contribute to the synthesis of these two amino acids. The converse is true when one considers the synthesis of aspartate. It is not known whether the ilvE-encoded aminotransferase is ever involved in phenylalanine biosynthesis apart from that seen in tyrB, aspC mutant strains. The nucleotide sequences of aspC, ilvE, and tyrB have been determined (81, 131, 132). The coding regions of aspC and tyrB encode 396 and 397 amino acid residues, respectively. These sequences have 169 residues in common, and it seems reasonable to assume that gene duplication played a role in their evolution. The deduced amino acid sequence for the aspartate aminotransferase also agrees with the published amino acid sequence for this enzyme purified from E. coli B (129).

By changing six residues in E. coli AspC to the corresponding residues in E. coli TyrB (i.e., V39L/K41Y/T47I/N69L/T109S and N297S), Onuffer and Kirsch (165) were able to produce mutant enzymes, which had acquired the broadened substrate specificity of the TyrB enzyme. X-ray crystallographic analysis of key inhibitor complexes of this hexamutant revealed that the tyrosine aminotransferase is able to undergo a conformational switch in which an active-site arginine can move aside to allow access of aromatic ligands (146). A mutant form of the TyrB enzyme in which the six substitutions change to the AspC sequence (i.e., S109T/S297N/L39V/Y41K/I47T and L69N) shows a nearly ninefold preference for aspartate over phenylalanine (198).

Preliminary crystallographic analysis of the TyrB enzyme shows a dimer in which each monomer binds a molecule of pyridoxal phosphate via a covalent bond linked to the epsilon NH₂ of Lys258 (126). The gene for the branched-chain aminotransferase, ilvE, encodes a smaller protein of 309 amino acids and does not show homology to either aspC or tyrB (132). The availability of extensive sequence data on aminotransferases from a large number of different organisms has allowed an investigation of their phylogeny. There is an aminotransferase superfamily comprising four families. The TyrB and AspC proteins are closely linked in subfamily Iα, and the IlvE aminotransferase is located in the more distant family III (121).

Using highly purified preparations of the aromatic aminotransferase, chorismate mutase-prephenate dehydrogenase, and chorismate mutase-prephenate dehydratase, Powell and Morrison (180) demonstrated protein-protein interactions between the aromatic aminotransferase and each of the other two enzymes. This interaction did not occur if the aspartate aminotransferase was substituted for the aromatic aminotransferase. The complex with mutase-dehydrogenase dissociated in the presence of tyrosine, and the interaction with mutase-dehydratase required the presence of phenylpyruvate. The interactions therefore are quite specific. The extent to which these interactions occur in vivo and the role that they play in synthesis have yet to be determined.

PHENYLALANINE AND TYROSINE TRANSPORTERS

The capacity of a microorganism to efficiently utilize amino acids that may be present in the environment is as important as many of the controls regulating endogenous biosynthesis. There are at least three transport systems that efficiently take up phenylalanine or tyrosine from the environment. Each of these systems involves a polytopic integral membrane protein whose activity is energized by the proton motive force. The first to be identified in both serovar Typhimurium and E. coli was the so-called general aromatic transport system, which transports each of the three aromatic amino acids phenylalanine, tyrosine, and tryptophan with high affinity (K_m, 0.4 μM) and is encoded by the aroP gene (1, 2, 25, 41). The AroP protein is a protein of 457 amino acids and has been shown to contain 12 transmembrane spans with the amino-terminal and carboxy-terminal regions in the cytoplasm (50). The transcription of the gene encoding this protein is regulated by the TyrR regulator that represses expression in the presence of either phenylalanine, tyrosine, or tryptophan (234). The AroP protein is a member of the amino acid transport (AAT) family within the superfamily of amino acid-polyamine-organocation transporters (APC) (120a, 252). A second protein encoded by a gene, pheP, also transports phenylalanine (233). The PheP and AroP proteins are closely related, sharing 61% identical residues. However, PheP, which is principally a transporter of phenylalanine with a K_m of 2 μM, is expressed at low levels in the cell. When the level of expression is increased by changing the translational start codon from GTG to ATG or by increasing the copy number, the protein now exhibits significant transport of tyrosine even though the K_m for tyrosine is significantly higher (30 μM) than that for phenylalanine (49). The basis for this change is not understood, but the low level of PheP expression in wild-type cells may well favor specificity for a single amino acid (i.e., phenylalanine). Extensive studies have been carried out identifying critically important residues in the protein, establishing topology and some aspects of a putative tertiary structure (49, 71, 169, 170, 171, 172, 173, 255). A recent study has established that the LIV-I/LS branched-chain-amino- acid transporter can also transport phenylalanine, with K_m values in the LIV-I and LS systems being 19 and 30 μM (129a). It remains to be seen whether this finding will explain the earlier reports of a locus azaB, mutations of which impaired active transport of phenylalanine in aroP mutants (238).

The tyrP gene encodes a protein that specifically transports tyrosine. Along with two tryptophan-specific transporters, Mtr and Tna, TyrP belongs in a small family of transporters (HAAP), which appear to have 11 transmembrane spans with the carboxy terminus in the periplasm (191, 241). The expression of the tyrP gene is also regulated by the TyrR protein, which in the presence of tyrosine represses expression and in the presence of phenylalanine or tryptophan causes enhanced expression (5, 234).

In addition to the transporters of the aromatic amino acids, E. coli has another integral membrane protein, ShiA, which transports the common pathway intermediate shikimic acid (232). This protein shows significant homologies with a number of proteins in cluster 3 of the major facilitator superfamily (148), but its overall role in the physiology of the cell is not clear.

The Genes of the Aromatic Pathway

The chromosomal locations of the structural genes for the enzymes of the pathway of biosynthesis of phenylalanine and tyrosine, their transport systems, and the pathway's two major regulator genes in E. coli are shown in Fig. 4. Full details of genes and enzymes are now available from a number of electronic sources, such as EcoCyc at www.EcoCyc.org or the E. coli Genetic Stock Center (CGSC) database at cgsc.biology.yale.edu/cgsc.html.

Fig. 4Linear map showing chromosomal location of various genes referred to in the text. The arrows indicate the directions of transcription. The transcriptional units that are controlled by the TyrR protein are shown in boldface.

Source: Adapted from reference 174a with the kind permission of the publisher.

Regulation of the Pathway.

The flow of substrates along the pathway is controlled by feedback inhibition of each of the DAHP synthase enzymes and the three major enzymes utilizing chorismate, that is, anthranilate synthase, chorismate mutase-prephenate dehydrogenase, and chorismate mutase-prephenate dehydratase. In addition, the synthesis of a number of enzymes and transporters is regulated in response to fluctuations in the intracellular concentrations of the three aromatic amino acids. The major regulatory protein involved in these controls is the TyrR protein, although the TrpR protein, which controls expression of the tryptophan operon, plays an ancillary role. The expression of the pheA gene, which encodes the chorismate mutase-prephenate hydratase enzyme, is regulated by an attenuation mechanism that senses the availability of charged phenylalanyl-tRNA^Phe.

The TyrR Regulon.

The TyrR regulon comprises at least eight separate transcription units (24, 26, 29, 30, 65, 78, 97, 106, 107, 112, 118, 119, 149, 167, 174, 174a, 190, 223, 234, 250). The TyrR protein can act as a repressor and, in some cases, as an activator of transcription. The effector molecules ATP, tyrosine, phenylalanine, and, to a lesser extent, tryptophan are able to affect its biological activity. The next section summarizes what is currently known about this protein and its various activities.

The TyrR Protein.

The tyrR gene has been cloned, and its sequence has been determined (47, 48, 243). From the sequence, the TyrR protein comprises 513 amino acids with a calculated subunit molecular weight of 57,640, compared with a measured molecular weight of 110,000 + 5,000 for the homodimer, in which form the protein normally occurs in the absence of effectors (8, 55).

The protein has been overexpressed and purified in several laboratories (5, 8, 55). In vitro studies indicate that the protein has three functional domains. The amino-terminal domain involves residues 1 through 190, the central domain involves residues 206 through 433, and the third domain involves residues 444 through 513 (55; J. Belcher, unpublished results) (see Fig. 5).

Fig. 5Diagram of the TyrR protein, showing the three domains and the key residues currently identified as important for various TyrR functions.

Source: Adapted from reference 174a with the kind permission of the publisher.

Each of the domains is important for a different function of the TyrR protein. They have all been independently expressed, and the amino-terminal domain has been crystallized (69, 70, 133, 144).

The Amino-Terminal Domain.

Deletions and specific amino acid substitutions have identified the region of the protein between amino acids 2 and 19 as playing a critical role in activation of the expression of the genes tyrP and mtr. A number of mutations affecting this region abolish TyrR-mediated activation without affecting repression of the other members of the regulon (53, 54, 243, 248). Alanine-scanning mutagenesis has identified residues 9 and 10 as critical for activation and also identified a second critically important region involving D103 (244). A glycine at position 37 also appears to play an important role. Iterative database searches have identified a ligand-binding domain (ACT) based on the carboxy-terminal domain of 3-phosphoglycerate dehydrogenase. In the case of TyrR this ACT domain occupies the region between positions 2 and 72 of the amino-terminal domain (7, 80). A second domain referred to as PAS and implicated in signal transduction has been identified in the region between positions 80 and 114 (80).

In vitro transcription studies have established that activation of both tyrP and mtr requires an interaction between the TyrR protein and the carboxy-terminal region of the α-subunit of RNA polymerase, as has been shown for other class I activators (120, 135, 246). Substitutions DN250 and RE310 in the α-subunit of RNA polymerase destroy TyrR-mediated activation without affecting the interaction of the mutant α with UP sequences and may constitute a TyrR-specific patch (H. Camakaris, personal communication). In vitro studies with supercoiled templates of either mtr or tyrP have shown that TyrR-phenylalanine-mediated activation can reverse transcriptional inhibition caused by small DNA-binding proteins such as HU or IHF (245).

The failure to find any mutations specifically affecting activation and not repression in any region other than that coding for the amino-terminal domain, coupled with the demonstration that a mutant deleted for much of the central domain (residues 226 to 419) shows wild-type levels of activation (unpublished results), suggests that both a binding pocket for the aromatic amino acids and a region of the protein able to interact with the α-subunit will be found in the amino-terminal domain. Studies with the purified protein have identified ATP-independent phenylalanine binding, in addition to the ATP-dependent tyrosine binding previously associated with the central domain (239). Neither the binding site nor the consequences of this binding have been identified, and it still is possible that both the amino-terminal and the central domain may have binding sites that recognize the three aromatic amino acids with different affinities.

When overexpressed, the amino-terminal domain exists as a dimer, indicating that it also contains a dimerization motif (133).

The Central Domain.

The central domain of TyrR is homologous to the central domains of several activator proteins, including NtrC, NifA, DctD, XylR, and FhlA, which activate expression from σ⁵⁴ promoters (discussed in reference 174). Mutational studies of the NtrC protein showed that the central domain of this protein is essential for activation (11, 228). A key component of this central domain is an adenylate-type ATP-binding site. Mutations that cause amino acid substitutions in site A of the ATP-binding site and a number of other mutations in this central domain destroy the ability of NtrC protein to activate gene expression. Furthermore, in the case of NtrC, ATP hydrolysis is involved in converting the RNA polymerase-promoter complex from a closed to an open state. The situation with regard to TyrR is quite different. Mutations altering site A of the ATP-binding site and three other mutations affecting the highly conserved residues E-274, G-285, and E-302 of the central domain have no effect on activation but abolish TyrR-tyrosine-mediated repression of aroF, aroL, tyrP, and tyrB (248). It has been demonstrated (8) that purified TyrR protein binds ATP with a K_d of 5 to 7 μM. In the presence of excess ATP, TyrR protein binds tyrosine with a K_d of 50 μM. Tyrosine binding cannot be detected in the absence of ATP. It has also been shown that in the presence of the nonhydrolyzable ATP analogue, ATPγS, and tyrosine (500 μM) or phenylalanine (25 mM), TyrR protein self-associates to form a hexamer (240). This hexameric form is postulated to play a major role in tyrosine-mediated repression. In the central-domain mutant EQ274, both tyrosine binding and tyrosine-dependent hexamerization are significantly impaired, although binding of ATP is unaltered (134). The role of ATP hydrolysis in repression has not yet been resolved. The purified TyrR protein does have ATPase activity that is stimulated by tyrosine (55, 240). Cui et al. (52) estimate that the level of this activity is similar to that reported for the unphosphorylated form of the NtrC protein. Weiss et al. (228) reported that activation of NtrC by phosphorylation increases its ATPase activity about 11-fold. Although mutants of TyrR with amino acid substitutions in site A of the ATP-binding site have been isolated and shown to have lost the ability to carry out TyrR-tyrosine-mediated repression (248), these mutant proteins have not yet been purified and tested for ATP binding and hydrolysis. The finding by Wilson et al. (240) that the nonhydrolyzable ATP analogue ATPγS can substitute for ATP in the self-association of TyrR dimers into hexamers might indicate that only binding is required for repression. If this is the case, then perhaps ATP hydrolysis functions to facilitate the reverse reaction of hexamer to dimers, hence freeing operator sites. In the case of NtrC protein, Austin and Dixon (10) have reported that ATPase activity is stimulated by DNA-containing NtrC-specific binding sites. In the case of XylR, oligomerization caused by the binding of ATP is necessary both for the activation of the σ⁵⁴ promoter Ps and for the repression of its own σ⁷⁰ promoter Pr (21). The TyrR protein has also been shown to have phosphatase activity that is stimulated by Zn²⁺ and inhibited by tyrosine and its analogues and by ATP and its analogues. The phosphatase active site has been localized within the central domain, but its role in TyrR-mediated activities has not yet been elucidated (256).

When the central domain fragment 188 to 467 is expressed it forms a monomer. Tyrosine in the presence of ATPγS promotes its oligomerization to a hexamer (70), confirming that the ATP-dependent binding site for tyrosine, the ATP-binding site, and the hexamerization motif are all found in the central domain. When the amino acid sequence of the central region of TyrR is compared with those of other members of the NtrC family, it can be seen that a small sequence of seven amino acids (EKGAFTG) is missing within the C3 region immediately after residue 279 (160). This deletion is not present in PhhR, a homologue of TyrR present in Pseudomonas aeruginosa and able to activate transcription from σ⁵⁴ promoters (205). Mutational analysis of the C3 region of NifA has established that, although such mutants are unable to activate expression, they still bind DNA and are able to form oligomers, leading to the hypothesis that the GAFTGA subregion may be involved in recognition of the σ⁵⁴ promoter complex (98). On the other hand, mutants in the same region of the enhancer-binding protein DctD have been shown to retain the ability to cross-link to σ⁵⁴ and the β-subunit of RNA polymerase and to hydrolyze ATP. It has been proposed that the role for the C3 region may be the coupling of ATP hydrolysis and open complex formation (227).

The Carboxy Domain.

The carboxyl domain contains a classic Cro-like HTH DNA-binding motif. Alanine-scanning experiments have identified several residues in both helix 1 and helix 2 that are essential for binding. The mutant analyses combined with consideration of the known interactions of Cro and CAP have led to the suggestion that R484 and H494 both form critical bonds with the invariant bases (G·C)(C·G)8 (116, 117). Deletion of 10 of the 11 amino acids between helix 2 and the carboxy terminus also destroys activation and repression functions (T. Betteridge, unpublished results). Similarly, a truncated TyrR protein missing the HTH motif can neither repress nor activate members of the regulon (53). An alanine-substituted mutant TyrR protein, HA494, is completely defective in binding to the tyrP operator, whereas another substituted mutant protein, TA495, shows reduced binding. Missing contact probing using this mutant protein suggests that T495 makes specific contacts with adenine and thymine bases at ±5 positions in the TyrR boxes. This analysis also reveals the importance for binding of the AT-rich sequences between the palindromic arms of strong but not the weak TyrR boxes (117) (see below).

When overexpressed, the carboxy-terminal domain is also shown to exist as a dimer (133). The TyrR protein showing the three domains and the critical amino acids that have been discussed in this section is represented in Fig. 5, and a discussion of the DNA sites to which the protein binds is included in the next section.

The TyrR Boxes and Their Arrangement

The target sites on the DNA that are recognized by the 'I'yrR protein are all related to the palindromic sequence TGTAAAN₆ TTTACA. The underlined bases (G and C) are invariant, and changing them to any other base largely destroys the function of the box (3, 42, 44, 83, 116, 123, 136, 190, 192, 242, 248). These palindromic sequences, together with the two flanking nucleotides at each end, are referred to as TyrR boxes. Their distribution with regard to the putative RNA polymerase-binding sites for each of the transcription units is shown in Fig. 6 along with relevant TrpR-binding sites for aroL and mtr. The sequence of each box is shown in Fig. 7. All the transcription units repressed by tyrosine, that is, aroF-tyrA, aroL, aroP, tyrB, and tyrP, have two adjacent TyrR boxes separated by a single base. In every case, the box showing the closest agreement to the consensus sequence is bound by the TyrR protein in the absence of effectors (4, 5, 136, 174, 192, 242, 247). For convenience, these boxes are referred to as strong boxes. Strong boxes, with the exception of aroG, are also AT rich in the central six positions (Fig. 7). The adjacent weak boxes are bound by TyrR protein only in the presence of effectors ATP and tyrosine, which in some cases can be substituted with phenylalanine, and only if there is a strong TyrR box nearby on the same face of the helix. In the case of tyrP, this cooperative binding between strong and weak boxes still occurs if the boxes are separated by one turn of the helix but not if the separation involves three turns of the helix (4). In general, weak boxes have a weaker agreement with the consensus and contain two or more G’s and C’s within the central six bases. In the case of aroF, tyrP, aroL, and tyrB the weak box is closest to or overlapping the RNA polymerase-binding site.

Fig. 6Distribution of TyrR boxes for various transcription units of the TyrR regulon.

Source: Adapted from reference 174a with the kind permission of the publisher.

Fig. 7Nucleotide sequences of TyrR boxes. Bases that are identical with the consensus sequence are shown in boldface. The central six bases of each box are shown in lowercase letters.

It has been proposed that the hexameric form of the protein formed in the presence of tyrosine and ATP is able to bind to two or more boxes (240). In some other cases, where there are two boxes, such binding involves two dimers and can be facilitated by the presence of phenylalanine and ATP (247). One exception to the general rule that the weak box plays a major role in tyrosine-mediated repression is the finding that box 2 of aroL apparently plays no part in repression. Like the other weak boxes, it is bound by TyrR protein only in the presence of tyrosine and ATP. However, mutating the invariant G·C pairs in this box has no effect on repression, which appears to involve a TyrR-mediated loop between strong boxes 1 and 3, which are separated by 54 bp (136). The presence of two strong boxes separated by three turns of the helix is also found in the case of the aroF promoter, where DNA looping between the strong boxes is again a strong possibility.

Diverse Strategies within a Regulon

A major advantage of the regulon structure is that each transcription unit is unique and has been able to evolve transcriptional controls best suited to the physiological role of the proteins encoded by that transcription unit. In the case of the TyrR regulon these various responses depend on subtle differences in TyrR box composition, their number and location, and, in some cases, the additional action of other transcription factors. These different strategies will be illustrated by a more detailed consideration of the regulation of some of the genes of the regulon.

The TyrB protein is an aminotransferase essential for the biosynthesis of tyrosine and important for the biosynthesis of phenylalanine. In minimal medium the tyrB gene is derepressed, but when tyrosine or phenylalanine is added to the medium expression is repressed fourfold and threefold, respectively. As shown in Fig. 6, tyrB has two TyrR boxes located downstream of the promoter. In the absence of aromatic amino acids, TyrR protein selectively binds to the strong box. In the presence of either tyrosine or phenylalanine this binding is extended to include the weak box. When TyrR is bound to both these boxes, RNA polymerase can still bind to the tyrB promoter. However, in the presence of tyrosine, when it is presumed that a hexamer binds across both boxes, RNA polymerase is unable to proceed to the formation of an open complex. In the presence of phenylalanine, when it is presumed that two dimers bind cooperatively to the two boxes, open complexes can be formed but the RNA polymerase is prevented from leaving the promoter and actively initiating transcription. By inserting bases between the TyrR boxes and the promoter it could be shown that both of these effects disappeared if the boxes were moved more than 19 bases away. So the dependence of TyrR on tyrosine or phenylalanine, to bind the weak box and the positioning of that box close to the transcription initiation site, ensures effective transcriptional control by tyrosine and phenylalanine. The fact that the TyrR boxes are located near the transcription start site rather than overlapping the –35 region of the promoter also ensures that the aromatic amino acids will not cause activation of tyrB expression (247).

The gene aroP also has two TyrR boxes located downstream of the promoter. However, in this case, the arrangement of strong and weak boxes is the reverse of what would be expected, with the strong box closest to the promoter. Furthermore, both boxes are separated from the promoter by a distance, which, in the case of tyrB, has been shown to nullify repression. The resolution of this dilemma came with the discovery of a promoter on the opposite strand, which was nonproductive in transcription initiation in vivo but which could nevertheless bind RNA polymerase when TyrR protein bound the TyrR boxes in the presence of one or other of the aromatic amino acid effectors. When an RNA polymerase molecule binds to this promoter, RNA polymerase is no longer able to occupy either of the promoters that would lead to transcription of the aroP gene. This effect depended not on TyrR protein’s ability to repress but on its activation functions, as demonstrated by the fact that activation-minus mutants of TyrR (RQ10) could not repress aroP, whereas the mutant TyrR EQ274, which is impaired in other cases of tyrosine-TyrR-mediated repression, is able to repress aroP. The AroP protein is a transporter that actively transports each of the three aromatic amino acids into the cell. The system that has evolved to control its synthesis utilizes the activation functions of TyrR, which can respond to each of the three aromatic amino acids (224, 225, 226).

The gene aroL encodes an important enzyme in the common pathway of aromatic amino acid biosynthesis, and its expression is repressed by TyrR-tyrosine (fivefold). The addition of tryptophan further increases repression (66). In this case, the tryptophan effect depends on the TrpR protein and the presence of a TrpR-binding site downstream of the TyrR boxes (108). The TrpR-tryptophan effect is observed only in tyrR⁺ strains and disappears if the TrpR and TyrR boxes are moved apart by half a turn of the helix, suggesting that interactions between these two proteins may play a role in the tryptophan effect (136). Furthermore, mutational studies show that, in this case, although there is a weak TyrR box overlapping the transcription start site, repression depends on the two strong boxes lying at either end of the promoter region and probably involves TyrR-mediated looping of the DNA (136).

Another gene whose expression is repressed by tryptophan and involves the TrpR protein is the gene for the tryptophan-specific transporter mtr. This gene is also subject to TyrR-mediated control, but in this case the TrpR effect is clearly dominant (107, 190, 192). As can be seen in Fig. 6, the TrpR-binding site overlaps the promoter whereas the two TyrR boxes are located upstream of the –35 sequence. In the absence of tryptophan and in the presence of either phenylalanine or tyrosine, TyrR protein bound to the TyrR boxes can activate expression of the mtr gene. When tryptophan is missing from the aromatic amino acid pool, the cell increases its efforts to capture any tryptophan in the environment.

The gene tyrP, which encodes a tyrosine-specific transporter, has evolved a similar control system, in that it is fully repressed in the presence of tyrosine but activated to higher levels in the absence of tyrosine and the presence of either phenylalanine or tryptophan. This is achieved by positioning adjacent strong/weak TyrR boxes so that the weak box overlaps the –35 sequence. In the presence of tyrosine the TyrR hexamer binds to both boxes and prevents the RNA polymerase from binding to the tyrP promoter (249). Both the TyrR protein and the σ-subunit of RNA polymerase compete for binding to critical bases in the region of the weak box (33, 117). In the presence of phenylalanine and, to a lesser extent, tryptophan, TyrR protein dimers bind to the strong box and interact with the α-subunit of RNA polymerase to increase binding and enhance open complex formation (135, 246). Phenylalanine can also facilitate the cooperative binding of two TyrR dimers to a strong/weak box combination, and if the level of TyrR protein is increased or an activation-defective mutant is used, then phenylalanine, like tyrosine, represses tyrP expression. These results indicate that the intracellular levels of TyrR protein are delicately balanced to achieve particular regulatory outcomes. The strong TyrR box is not optimally placed for activation, which is improved by moving it three or four bases upstream; however, its position is optimal with regard to the placement of the adjacent weak box and effective repression, which would appear to be the dominant response (4).

Three other transcription units show a very significant response to changes in the level of TyrR protein. These are the aroF-tyrA operon and aroG, each of which encodes a gene for one of the DAHP synthase isoenzymes and tyrR, the gene for the TyrR protein. Regulation of aroG and tyrR appears to be unaffected by the presence or absence of aromatic amino acids, although early studies in a chemostat had suggested a fourfold derepression of aroG in cultures of an aromatic auxotroph starved for phenylalanine or tryptophan. On the other hand, growth of wild-type cells in the presence of all aromatic amino acids caused only a 20% drop in activity, and similar small changes were observed in studying the expression of an aroG-lac construct (17, 18, 26, 30, 58, 118). Changing the levels of TyrR protein either by introducing additional copies of the gene on a multicopy plasmid or by mutating the tyrR gene has a marked and unambiguous effect on expression from the aroG and tyrR promoters. The aroG gene has a single strong TyrR box in a position identical with that of the weak box in tyrP. It is an unusual box in that, although the palindromic arms exactly fit the consensus, the sequence between them is GC rich. Nevertheless, it is bound by the TyrR protein in the presence of ATP without any phenylalanine or tyrosine (18). In light of the results with tyrP, it would seem that repression of aroG is a consequence of direct competition between TyrR protein and RNA polymerase binding to the DNA. The situation with tyrR is less clear, because the strong TyrR box and a second weak box are located upstream of the –35 hexamer (9). No studies to investigate the binding of both the TyrR protein and RNA polymerase to this promoter have yet been carried out.

The promoter of the aroF tyrA operon shows about a 10-fold repression in minimal medium, which is due to normal levels of the TyrR protein alone. When tyrosine is added, expression is repressed a further 10- to 20-fold. If levels of TyrR protein are increased, phenylalanine can also cause repression. As can be seen in Fig. 7, the aroF promoter has three TyrR boxes. As is the case with tyrP, separating boxes I and 2 by half a turn of the helix completely destroys repression (4). On the other hand, inactivating box 3 or moving it up to 110 bases upstream only partially reduces repression (43, 44). Box 2 is a very AT-rich sequence, and its position corresponds to that of known UP sequences that are recognized by the α-subunit of RNA polymerase and can greatly enhance promoter activity (100). Although this has not yet been tested, it seems likely that the strong minimal medium repression of aroF-tyrA is a consequence of competition between the α-subunit and TyrR protein for binding to this region of the DNA. Whatever the mechanism, it is of particular interest that both the aroF and aroG promoters have reserve strengths that are realized only when levels of TyrR protein in the cell are diminished. These two enzymes compete with other enzyme systems for the substrates PEP and E4P and under certain physiological circumstances may need to be synthesized in greatly increased amounts. For example, when strains of E. coli are grown in minimal medium supplemented with all amino acids, except the three aromatic amino acids, the expression of aroF is fully derepressed (213). The complementary studies on the levels of TyrR protein under these conditions have not yet been carried out.

The pheA Gene

The expression of the pheA gene is controlled by a system of transcriptional attenuation that senses the availability of charged phenylalanyl-tRNA^Phe (257). Early reports, which appeared to identify operator mutations in pheA and unlinked mutations in a putative regulator pheR, have now explained these as mutations in the leader region of the pheA gene or as mutations affecting the availability of charged phenylalanyl-tRNA^Phe (84, 85, 86, 96, 101, 102, 176).

Acknowledgments

This work was supported by the Australian Research Committee.

We thank colleagues for permission to quote unpublished results and Helen Camakaris for reading the manuscript.