Chapter 24

Ammonia Assimilation and the Biosynthesis of Glutamine, Glutamate, Aspartate, Asparagine, l-Alanine, and d-Alanine

LAWRENCE J. REITZER

INTRODUCTION

This chapter discusses the synthesis of glutamine, glutamate, aspartate, asparagine, and d- and l-alanine in Escherichia coli, Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium), and Klebsiella aerogenes. Because the enzymes of glutamine and glutamate synthesis also participate in ammonia assimilation, the first section considers ammonia assimilation. Many references that support the generalizations of this section are provided elsewhere in this review. Table 1 is a summary of the enzymes discussed in this review and their proposed function or functions.

Table 1Synthesis of glutamine, glutamate, aspartate, asparagine, L-alanine, and D-alanine: enzymes, genes, reactions, and additional functions other than biosynthesis

ASSIMILATION OF AMMONIA

Cellular Requirements for Nitrogen and Ammonia

For bacteria grown in minimal medium with an ample supply of an ammonium salt, the nitrogen for almost all nitrogen-containing compounds is derived from the primary products of ammonia assimilation, glutamate and glutamine. Directly or indirectly, glutamate provides α-amino groups for all of the amino acids, half the nitrogens for the pyrimidine, purine, and imidazole ring, and the amino group of adenine. Glutamine, which can be considered a high-energy nitrogen donor, provides the nitrogen for amino sugars, NAD, and p-aminobenzoate and the remaining nitrogens for purines, pyrimidines, histidine, tryptophan, and (sometimes) asparagine. One kilogram (dry weight) of E. coli contains 11 to 12 g-atoms of total nitrogen. It has been estimated that synthesis of glutamate and compounds that obtain their nitrogen from glutamate requires about 10 g-atoms of nitrogen. Synthesis of glutamine and the nitrogen from the glutamine amide that is incorporated into other compounds requires about 1.3 g-atoms of nitrogen (205). Ignoring for the moment that glutamate can be synthesized from glutamine, the cell requires much more glutamate than glutamine for biosynthesis.

Cells absolutely require ammonia for glutamine synthesis, and, depending on the growth medium, they may require ammonia for synthesis of all nitrogenous compounds. The requirement for ammonia depends on the nitrogen source in the medium and the products of its catabolism. Three situations must be considered. First, if ammonia is present at high enough a concentration, it is assimilated directly into both glutamine and glutamate by glutamine synthetase (GS, reaction 1 below) and glutamate dehydrogenase (GDH, reaction 3 below), respectively. Second, for nitrogen sources that produce glutamate as a degradation product, such as proline, aspartate, or γ-aminobutyrate, ammonia is needed only for glutamine synthesis. For the degradation of these amino acids, the reaction that generates ammonia has not been determined. (Chapter 23 of this volume describes this unsolved problem in more detail.) Finally, for nitrogen sources such as serine, which generate ammonia but not glutamate, the ammonia assimilated into glutamine provides the nitrogen for glutamate synthesis by glutamate synthase (GOGAT, reaction 2). In this case, the reactions catalyzed by GS and GOGAT form an ammonia assimilatory cycle, which results in the energy-dependent assimilation of ammonia into glutamine and glutamate (Fig. 1). In other words, all cellular nitrogen is derived from ammonia and the amide of glutamine which, in turn, has been derived from ammonia. In summary, nitrogen source catabolism must produce ammonia. The requirement for ammonia can be reduced but not eliminated if glutamate is also a product of catabolism.

Fig. 1Ammonia assimilatory cycle. The reactions catalyzed by GS and GOGAT result in ATP-dependent ammonia assimilation and glutamate synthesis. Such a cycle is essential for entry into nitrogen-limited growth and for steady-state growth with some sources of nitrogen.

Source:

Nitrogen Sources, Nitrogen-Limited Growth, and the Ntr Response

Surveys of compounds that can serve as sole nitrogen sources have been presented for S. typhimurium, E. coli, and K. aerogenes (66, 180). E. coli and S. typhimurium grow aerobically with only one inorganic nitrogen compound, ammonia. However, some Klebsiella species can utilize nitrate, nitrite, or atmospheric diatomic nitrogen. E. coli and S. typhimurium can use nitrate or nitrite as nitrogen sources during anaerobic growth, when these compounds are also terminal electron acceptors. A limited number of organic nitrogen compounds, about 25, can support growth as sole sources of nitrogen. These nitrogen sources are usually macromolecular precursors, such as amino acids or nucleotide bases. Mutants that grow on an extended range of nitrogen sources can be isolated (89).

Ammonia is considered the preferred nitrogen source because it supports the highest growth rate and its assimilation results in repression of proteins that catabolize other nitrogen sources. Growth with another nitrogen source is invariably slower than growth with ammonia and is said to be nitrogen limited. Nitrogen limitation induces synthesis of GS, the primary enzyme of ammonia assimilation, and other proteins, which are said to be nitrogen-regulated (Ntr) proteins. Known proteins of the Ntr response are involved in degradation of arginine, ornithine, agmatine, putrescine, and γ-aminobutyrate in E. coli (169, 209), degradation of histidine, proline, and urease in K. aerogenes (147, 180), transport of γ-aminobutyrate, glutamine, and ammonia in E. coli (76, 77, 203), and transport of glutamine, arginine, aspartate, lysine, ornithine, glutamate, and histidine in S. typhimurium (55, 96). A more complete discussion of nitrogen source catabolism is presented in chapter 23.

The central participants of the Ntr response are nitrogen regulator I (NR_I or NtrC), nitrogen regulator II (NR_II or NtrB), and RNA polymerase complexed to σ ⁵⁴. NR_I is a transcriptional activator of σ ⁵⁴-dependent promoters, and NR_II is a bifunctional protein that can either transfer phosphate to NR_I or control the dephosphorylation of NR_I-phosphate. Nitrogen limitation results in the phosphorylation of NR_I by events described in Control of GS Synthesis, below. NR_I-phosphate stimulates expression of the glnALG operon, which codes for GS, NR_II, and NR_I. Such expression results in a higher level of NR_I, which becomes sufficient to stimulate transcription of Ntr genes with weak NR_I-binding sites. Table 2 summarizes the genes and proteins involved in glutamine synthesis and the Ntr response; alternative designations are provided. This review discusses the regulators that control the activities of NR_I and NR_II within the context of expression of the glnALG operon. Chapter 23 considers expression of other Ntr genes.

Table 2Proteins and genes that regulate glutamine synthesis, and alternative designations

Ammonia Assimilatory Functions of GS, GDH, and GOGAT

The availability and quality of energy and nitrogen are the major factors in determining when GS, GDH, and GOGAT participate in ammonia assimilation. For cells grown in energy-poor, nitrogen-rich (ammonia-containing) media, GDH-deficient cells are at a significant disadvantage relative to wild-type cells, which suggests that GDH is important for glutamate synthesis and ammonia assimilation (71a).

For growth in energy-rich (glucose-containing), nitrogen-rich (ammonia-containing) media, the level of GDH is high and that of GS is low. Nonetheless, in this situation a GDH deficiency does not put cells at a competitive disadvantage, which suggests that GOGAT, not GDH, synthesizes glutamate (71a), which in turn implies that GS, not GDH, assimilates ammonia. It should be noted that a GOGAT deficiency does not cause a growth defect for cells grown with 1 mM ammonia or more, which implies that GDH can synthesize glutamate and therefore can assimilate ammonia in these media.

For growth in an energy-rich, nitrogen-poor environment (i.e., with less than 1 mM ammonia or with another nitrogen source), the intracellular concentration of GS is high and GS assimilates ammonia. GDH does not participate in ammonia assimilation, presumably because the concentration of ammonia is well below the K_m of GDH for ammonia. In such media, GOGAT is essential; the precise function of GOGAT in such media will is discussed in Function of GOGAT in Nitrogen Assimilation, below.

GLUTAMINE

The GS-catalyzed synthesis of glutamine from ammonia and glutamate is at a major branch point in metabolism. Glutamine is a constituent of proteins and a nitrogen donor for synthesis of purines, pyrimidines, and other compounds. In addition, the reaction catalyzed by GS is the major mechanism of ammonia assimilation during nitrogen limitation. The importance of these multiple functions can be surmised by recognizing the layers of regulation and the multiplicity of effectors that control the activity and synthesis of GS.

Control of the Catalytic Activity of GS

Feedback Inhibition and Its Relation to Covalent Adenylylation.

GS purified from cells grown in nitrogen-rich medium is inhibited by alanine, glycine, serine, histidine, tryptophan, CTP, AMP, carbamyl-phosphate, and glucosamine 6-phosphate (65, 175, 177, 206, 207). These inhibitors can be divided into two classes. Class I inhibitors—alanine, glycine, and serine—acquire their nitrogen from glutamate, which may or may not obtain nitrogen from glutamine. The other six inhibitors, which make up class II, are compounds whose synthesis absolutely requires the amide of glutamine. Class I and II compounds appear to inhibit by different mechanisms. It has been proposed that each class II inhibitor binds to a separate allosteric site (65, 175, 177) and that class I inhibitors bind to a single site (65). These earlier results suggested that the single class I site was an allosteric site, although the evidence was not considered conclusive (175). However, direct evidence from X-ray diffraction analysis, together with kinetic evidence, shows that the class I inhibitors bind to the substrate site for glutamate (see reference 100 and references therein). The binding site for one class II inhibitor, AMP, has been determined—it binds to the substrate site for ATP and not to an allosteric site as suggested by some earlier results (99a).

Susceptibility to feedback inhibition is altered when cells are grown in nitrogen-limited medium (81). When compared with GS from cells grown in nitrogen-rich medium, GS from nitrogen-limited cells is less sensitive to feedback inhibition. This differential inhibition results from covalent adenylylation, i.e., the addition of an AMP group to a specific tyrosine—adenylylated GS is more sensitive to feedback inhibition than unadenylylated GS (80, 170, 171). The next sections describe the adenylylation system.

Overview of Cyclic Adenylylation.

The effect of adenylylation on GS activity is sufficiently dramatic that adenylylated and unadenylylated GS can be considered different proteins with different functions. For cells grown in nitrogen-rich medium, GS is adenylylated and is present at low concentration. The fact that adenylylated GS is feedback inhibited suggests that its function is primarily for glutamine synthesis and not for ammonia assimilation. For cells grown in nitrogen-limited medium, GS is not adenylylated and is present at a high concentration. That unadenylylated GS is not subject to feedback inhibition suggests that ammonia assimilation is an important function of this form of GS.

The proteins of the adenylylation cascade, the bifunctional uridylyltransferase (UTase)/uridylyl-removing enzyme (UR), P_II, and adenylyltransferase (ATase), have been purified and characterized. The open arrows in Fig. 2 depict the events that lead to adenylylation. A high intracellular concentration of glutamine activates UR, which causes the deuridylylation of the regulatory protein P_II. The interaction of unmodified P_II with ATase results in adenylylation of GS. The solid arrows in Fig. 2 illustrate the events that lead to deadenylylation. A high intracellular concentration of α-ketoglutarate activates UTase, which transfers a UMP group to subunits of the regulatory protein P_II to form P_II-UMP. P_II-UMP interacts with ATase, which in turn catalyzes the removal of AMP from GS.

Fig. 2Cyclic adenylylation of GS. The reactions that control the adenylylation of GS are shown. The open arrows depict events leading to adenylylation of GS. This cascade is triggered by a high ratio of glutamine to ?-ketoglutarate, which results in UR being more active than UTase. The dark arrows illustrate events that result in removal of the adenylyl groups. This cascade is triggered by a low ratio of glutamine to ?-ketoglutarate. Adenylylation inactivates the modified subunit and subjects the unmodified subunits to cumulative feedback inhibition.

Source:

Effects of Adenylylation on GS Activity.

Adenylylation affects GS in a number of ways. First, adenylylation inactivates the modified subunit only (175). Purified GS is a dodecamer of identical 55,000-Da subunits. An enzyme with 9 of its 12 subunits adenylylated has three active subunits. Second, partial adenylylation sensitizes the unadenylylated subunits to feedback inhibition (170, 171). Third, adenylylation affects metal specificity and the pH activity profile (80, 175). The activity of GS can be measured by what is referred to as the glutamyl transferase reaction. Both adenylylated and unmodified GS catalyze the transfer of a glutamyl residue from glutamine to hydroxylamine in the presence of Mn²⁺, arsenate, and ADP. Sixty millimolar Mg²⁺ completely inhibits the activity of the adenylylated subunits but has no effect on the unadenylated subunits (9, 65, 97). The glutamyl transferase assay with and without Mg²⁺ has been exploited for the rapid assessment of the degree of adenylylation from crude extracts (9).

ATase.

ATase, a monomer with a molecular weight of about 130,000, catalyzes the ATP-dependent addition of AMP to a subunit of GS, with the release of PP_i:

GS + nATP → GS-AMP_n + nPP_i

Each subunit can be adenylylated, so that GS can have 12 adenylyl groups. ATase also catalyzes the phosphate-dependent removal of AMP from each subunit of GS:

GS-AMP_n + nP_i → GS + nADP

Clearly, the deadenylylation reaction is not a reversal of the adenylylation reaction (41, 65, 175, 177). Purified ATase can catalyze both adenylylation and deadenylylation without the regulatory proteins, P_II and UTase/UR (65, 177). Therefore, the catalytic potential for both reactions resides in ATase. Biochemical and genetic evidence indicates that the two activities of ATase are on separate sites on the same polypeptide (65, 157). Without P_II, glutamine stimulates ATase-dependent adenylylation and phosphate stimulates ATase-dependent deadenylylation (2, 41). The rate of adenylylation without P_II is about threefold lower than with P_II, whereas the rate of the deadenylylation reaction without P_II-UMP is exceedingly low but detectable (2). Subsequent genetic results have shown that these P_II-independent reactions of ATase occur in cells. In P_II-deficient strains (glnB mutants), GS is highly adenylylated in nitrogen-replete cells and not as highly adenylylated in nitrogen-poor cells(23, 51). The binding sites on ATase for the two factors that stimulate adenylylation, glutamine and P_II-UMP, are separate, providing further confirmation that glutamine can directly affect ATase without P_II (65). In summary, biochemical and genetic results show that P_II-independent regulation of ATase is somewhat similar to the overall regulation of the cascade system with P_II and UTase/UR.

UTase/UR and PII , Sensors of Glutamine Availability.

P_II-UMP stimulates the deadenylylating activity of ATase; unmodified P_II promotes adenylylation (21). The modification of P_II is controlled by the bifunctional UTase/UR, which catalyzes the covalent uridylylation or deuridylylation of P_II (1, 60). Glutamine stimulates UR; α-ketoglutarate stimulates UTase (1). Therefore, the ratio of glutamine to α-ketoglutarate controls net uridylylation of P_II and, through ATase, adenylylation of GS.

Purified P_II has an apparent molecular weight of 44,000 and was thought to be a tetramer of identical subunits (1, 174). However, recent cross-linking studies show that P_II is actually a trimer (A. J. Ninfa, personal communication). Each subunit can be uridylylated at a specific tyrosyl residue (158). Purified UTase/UR has a molecular weight of about 95,000 (60). It has a tendency to oligomerize; the most active species is a monomer. As shown below, the deuridylylation reaction is not a reversal of the uridylylation reaction.

By a number of criteria, both UTase and UR activities are on the same polypeptide. The two activities are copurified, and the ratio of activities remains constant during purification (60). Furthermore, strains with point mutations in glnD, which are deficient in UTase, also lack UR; revertants acquire both activities (6, 49).

Evidence that UTase/UR, Acting through PII, Controls GS Adenylylation.

Mutational loss of UTase/UR; results in overadenylylation of GS in mutants strains of E. coli, K. aerogenes, and S. typhimurium (6, 15, 49). This result is expected, since these cells contain unmodified P_II, which should stimulate ATase-dependent adenylylation. Overadenylylation can be suppressed by mutations in glnB, which result in loss of P_II (23, 51). In glnB strains, GS is slowly adenylylated and deadenylylated, confirming the role of P_II in these reactions. Members of an unusual subclass of glnB (P_II-encoding) mutants of S. typhimurium and K. aerogenes have essentially the same phenotype as do strains without UTase/UR; i.e., GS is overadenylylated (56, 147). In the K. aerogenes mutant, the phenotype was shown to result from an altered P_II that could stimulate ATase-dependent adenylylation but not ATase-dependent deadenylylation. It was suggested that P_II in these strains could not be uridylylated (50, 51). (The growth phenotype of these mutants is discussed in Genetics of GS Synthesis [below], because these mutants also are altered in their ability to phosphorylate NR_I and express Ntr genes.)

The genetic evidence just discussed provides powerful evidence that UTase/UR and P_II control the activity of ATase in vivo. Such evidence does not establish the relevance of particular physiological effectors. To verify that the overall adenylylation cascade was controlled by the same effectors, glutamine and α-ketoglutarate identified by analysis of purified proteins, Mura et al. (133, 134) characterized adenylylation of GS in situ. It was possible to make cells of E. coli permeable so that all the proteins of the adenylylation cascade were still active. In such cells, the ratio of glutamine to α-ketoglutarate controlled the degree of adenylylation: a high ratio resulted in adenylylation, and a low ratio resulted in deadenylylation. These responses were affected by Mn²⁺, CMP, UTP, and a high pH, factors that affect UTase/UR but not P_II or ATase (65, 176). It was concluded that the ratio of glutamine to α-ketoglutarate controls the activity of UTase/UR and, through P_II and ATase, controls the adenylylation of GS.

Conditions were found in which such treatment of cells resulted in inactivation of UTase/UR but not of other enzymes of the adenylylation cascade. In such cells, the degree of uridylylation of P_II is frozen. Therefore, changes in adenylylation should be due solely to low-molecular-weight effectors of ATase, such as glutamine and phosphate. Deadenylylation required phosphate as a substrate and was positively regulated by α-ketoglutarate and ATP (133, 134). The effect of ATP had been previously observed with purified ATase (2), but that of α-ketoglutarate had not. The effect of α-ketoglutarate may be P_II dependent (see Control of GS Synthesis, below). Adenylylation was stimulated by glutamine, consistent with previous biochemical and genetic evidence that suggested that glutamine binds ATase. In summary, these results confirmed the behavior of UTase/UR-independent control of ATase deduced from analysis of purified proteins.

The Complexity of the Adenylylation System Enhances Its Sensitivity.

The entire adenylylation system is called a bicyclic cascade, because two reversible modifications, uridylylation-deuridylylation and adenylylation-deadenylylation, control the activity of GS. ATase without UTase/UR and P_II can catalyze the cyclic adenylylation of GS. This cycle is a monocyclic cascade because only one reversible modification controls GS activity (177). Biochemical and genetic studies (discussed above) indicated that the monocyclic and bicyclic cascades respond to the same or similar environmental cues. This raises the issue of the advantages of the more complex bicyclic cascade over a monocyclic cascade, which would appear to be sufficient. One advantage of such complexity arises from the fact that this regulatory pathway is branched. UTase/UR and P_II also control the activities of proteins that regulate gene expression (discussed below). A second advantage of the bicyclic cascade is that the response is faster and more sensitive to changes in the glutamine/α-ketoglutarate ratio (133, 134). This enhanced sensitivity has been termed signal amplification and has been reviewed (33, 177).

Function of the GS Adenylylation System.

Covalent adjustments to changes in nitrogen availability are undoubtedly faster than adjustments that could result from alterations in gene expression. Cells that respond faster to nitrogen deprivation because of activation of GS activity (deadenylylation) have an obvious competitive advantage over cells that can respond only by increasing the rate of GS synthesis. Enzymatic inactivation of GS during the reverse transition, from a nitrogen-poor to an ammonia-containing medium, is also important in a competitive situation. Kustu et al. (93) examined the effect of loss of ATase in S. typhimurium. A glnE (ATase-deficient) mutant grew normally with ammonia and a variety of compounds as sole nitrogen sources. The responses of the mutant and a wild-type strain were compared when ammonia was added to each growing in nitrogen-limited medium. For the glnE mutant, ammonia addition resulted in a long growth lag, a 10-fold drop in the glutamate pool, and an increase in the glutamine pool. The ATP pool was not affected. For a wild-type strain, ammonia addition had virtually no effect on growth rate or on pools of glutamate, glutamine, and ATP. It was concluded that adenylylation, which both lowers the specific activity of GS and sensitizes GS to feedback inhibition, prevents glutamate depletion during the transition from a nitrogen-poor to an ammonia-containing environment (93). It can also be concluded that the class I inhibitors of GS (alanine, glycine, and serine), which should be present during the transition, do not appear to be physiologically important inhibitors of unadenylylated GS.

Structure and Expression of the GS-Encoding glnALG (glnAntrBC) Operon

Genes of the Operon and Their Products.

The glnA gene is a member of the glnALG, or glnAntrBC, operon; glnA, glnL (or ntrB), and glnG (ntrC) code for GS, NR_II (NtrB), and NR_I (NtrC), respectively (4, 30, 119, 129, 143). The glnA gene is the only gene specifying a GS in E. coli, K. aerogenes, and S. typhimurium. Its inactivation results in an absolute requirement for glutamine (97, 107, 118). This is not so for other bacteria, which may have two or three genes specifying different GSs (172). Sequence comparisons indicate that there are only two major families of GSs: GSI is found in most prokaryotes, including the enteric bacteria; GSII is found in eukaryotes and members of the families Rhizobiaceae, Frankiaceae, and Streptomycetaceae (90). A major difference between the two GSs, when they exist in the same organism (as they do for Rhizobium species), is that GSI can be adenylylated but GSII cannot (53a). Phylogenetic analysis suggests that genes for GSI and GSII diverged 3.5 billion years ago, or 1.7 billion years prior to the divergence of prokaryotes and eukaryotes (90).

Structure of the glnALG (glnAntrBC) Operon.

Most aspects of the structure of the operon are conserved among the enteric bacteria. The gene arrangement, the number and arrangement of promoters, and the regulators that control their activity are conserved. The glnA gene has two promoters: glnAp ₁ and glnAp ₂, which have transcriptional start sites about 190 and 70 bp, respectively, upstream from the translational start site (73, 153). The operon also contains an internal promoter preceding glnL (87, 142, 182). A rho-independent terminator is present between glnA and glnL (68, 87, 161, 182). The glnA-glnL intercistronic region in E. coli contains two repetitive extragenic palindromic sequences that are not found in S. typhimurium (161). It is not clear whether their presence has any role in expression of the operon.

Expression of the glnALG (glnAntrBC) Operon: Promoter Utilization.

During nitrogen-limited growth, glnA is transcribed from glnAp ₂, the strongest promoter of the operon (73, 153). The glnL and glnG genes are also transcribed from this promoter as a result of readthrough past the terminator between glnA and glnL (87, 142). In E. coli, expression from glnAp ₂ during nitrogen-limited growth results in approximately 1,000 molecules of dodecameric GS (i.e., 12,000 subunits) and 70 molecules of dimeric NR_I (i.e., 140 subunits) (152). Transcription from glnAp ₂ requires the phosphorylated form of NR_I and RNA polymerase complexed with the minor σ ⁵⁴ (73, 75, 79, 138, 139, 153). The mechanism of transcriptional activation from the glnAp ₂ promoter is unusual for a prokaryotic gene. NR_I activates from sites that are analogous to eukaryotic enhancers (94, 154, 196), and σ ⁵⁴ is unlike other sigma factors (see references 125 and 131 for recent reviews). A detailed description of the activation mechanism is beyond the scope of this review and is presented elsewhere (94, 95, 108, 196).

During carbon-limited growth, glnA transcription initiates from glnAp ₁ (153). Expression requires cyclic AMP complexed to cyclic AMP receptor protein (CRP) and RNA polymerase complexed with the major σ subunit of the cell, σ ⁷⁰ (153). (As discussed below, CRP-dependent control can lead to strong catabolite repression when transcription from glnAp ₂ is impossible.) When glnA is transcribed from glnAp ₁, transcription of glnL and glnG is initiated from glnLp. Transcription from this promoter does not require cyclic AMP-CRP or any other activator (152, 182). Transcription initiated from glnAp ₁ is completely terminated at the end of glnA (this conclusion is based on combining results from references 153 and 182). Both glnAp ₁ and glnLp are minor promoters that are not expressed during nitrogen-limited growth; both are repressed by NR_I (142, 152, 181, 182). The function of glnAp ₁ and glnLp is to ensure a basal level of the three products of the operon when growth is not nitrogen limited. When transcription is initiated at glnLp, glnL and glnG mRNAs are translated from the glnLG mRNA to the same extent. However, when transcription is initiated from glnAp ₂, glnL mRNA is translated considerably less than glnG mRNA (T. P. Hunt and B. Magasanik, personal communication). In this case, the upstream region of the glnL mRNA probably interferes with ribosomal binding.

Control of GS Synthesis.

This section presents an overview of the factors that control GS synthesis. This subject has been reviewed previously (107, 109, 155). The central feature of control is that UTase/UR and P_II, the regulatory proteins that control GS adenylylation, also control the phosphorylation of NR_I, as depicted in Fig. 3. Nitrogen limitation induces not only the synthesis of GS but also the synthesis of other proteins that transport and degrade nitrogenous compounds (96, 107, 147). The same regulators induce both GS and many other Ntr proteins.

Fig. 3Overview of the transcriptional control of glnA by nitrogen availability. The events that result in regulation of transcription from the glnAp2 promoter of the glnALG operon are shown. UTase/UR and PII, the regulators of the adenylylationcascade, also control the phosphorylation of NRI, the transcriptional activator of the glnAp2 promoter. The ratio of glutamine to ?-ketoglutarate is an indicator of nitrogen starvation. A low ratio (dark arrows) results in net modification of PII, which does not interact with NRII. In this case, NRII phosphorylates itself and transfers the phosphate to NRI, which in turn activates transcription. A high ratio, the indicator of nitrogen sufficiency, results in net deuridylylation of PII. The interaction between unmodified PII and NRII prevents the net phosphorylation of NRI. In this case, there is no transcription from glnAp2.

Source:

The dark arrows in Fig. 3 depict the sequence of events that lead to transcription from glnAp ₂ and induction of GS synthesis during nitrogen-limited growth. A low ratio of glutamine to α-ketoglutarate, the signal of nitrogen deprivation, stimulates UTase, which uridylylates P_II. P_II-UMP does not interact with NR_II. Left to itself, NR_II phosphorylates itself and transfers the activated phosphate to an aspartate of NR_I (138, 197). NR_I-phosphate interacts with RNA polymerase complexed with σ ⁵⁴, which is already bound at the start site of transcription, and catalyzes open complex formation.

The open arrows in Fig. 3 show the events that occur when the ratio of glutamine to α-ketoglutarate is high, that is, when nitrogen availability does not limit growth. Glutamine stimulates UR, which prevents uridylylation of P_II and removes UMP from P_II-UMP. By an unknown mechanism, P_II together with NR_II not only blocks phosphorylation of NR_I but also stimulates the dephosphorylation of NR_I-phosphate (79, 138). Recently, the entire signal transduction cascade (UTase/UR, P_II, NR_II, and NR_I) has been reconstituted with purified components (E. S. Kamberov, M. R. Atkinson, and A. J. Ninfa, Abstracts of the 1994 Molecular Genetics of Bacteria and Phage Meeting, p. 32). It was shown that α-ketoglutarate also binds P_II and greatly accelerates the P_II-NR_II-stimulated dephosphorylation of NR_I-phosphate. It is not entirely clear whether α-ketoglutarate is a cofactor that is always present at saturating amounts or an allosteric effector. If α-ketoglutarate is an allosteric effector of P_II activity, part of the signal of nitrogen limitation is simultaneously a signal of nitrogen excess. It is also not known whether the binding of α-ketoglutarate to P_II participates in the control of ATase. If it does so, aspects of control of the adenylylation cycle may have to be reevaluated.

Because UTase/UR and P_II control the activities of both ATase and NR_II, the response to changes in nitrogen availability is potentially highly coordinated. One factor that may modulate this coordination is the level of P_II, which is controlled by PurR and purines (70). A second factor may be the amount of P_II required to affect ATase and NR_II. In vitro, 100 times more P_II is required to inhibit the phosphotransfer activity of NR_II than to stimulate the adenylylation activity of ATase (79), although, as described above, α-ketoglutarate greatly stimulates dephosphorylation of NR_I-phosphate.

Genetics of GS Synthesis.

This section describes the effects of mutations in the genes involved in GS synthesis. The defects can almost always be understood in terms of the regulatory cascade that controls the activity of NR_II and in terms of the specific controls of the promoters of the glnALG operon.

   glnD (UTase/UR) mutants. Mutants with low or no UTase/UR require glutamine for good growth (6, 15, 49). As described above, GS is overadenylylated in these mutants and therefore has less biosynthetic activity (6, 15, 49). In addition, such strains have moderately less GS (6, 23, 49). Although GS can still be induced to some extent when glutamine is the sole nitrogen source (6, 23, 49), these strains fail to induce the Ntr response and cannot utilize a variety of nitrogen compounds (6, 15, 23, 49). The fact that P_II is not modified without UTase/UR accounts for these defects. Unmodified P_II is frozen in the mode that stimulates adenylylation of GS and dephosphorylation of NR_I-phosphate (Fig. 2 and 3).

The effects of mutations in glnD mutation can be suppressed by two general mechanisms: an increase in biosynthetically active GS or an increase in the synthesis of GS. Loss of ATase increases the activity of GS (6, 23, 49); loss of P_II increases the activity and level of GS (23). Suppression of mutations in glnD can also result from alterations in NR_I or NR_II that make them insensitive to regulation by UTase/UR and P_II (23, 195). The former explanation probably accounts for the glnA-linked suppressors that were isolated before the regulators of GS synthesis were known (6, 49). These conclusions can be stated in different terms—mutations in glnE (ATase), glnB (P_II), glnL (NR_II), and glnG (NR_I) are epistatic to those in glnD, consistent with the mechanism of action of UTase/UR.

   glnB (P_II) mutants. Without P_II, cells have a high level of GS whether grown with or without ammonia (23, 51). Such a consequence is expected upon removal of the negative regulator of the kinase activity of NR_II. However, there is still some residual regulation. Mutants lacking P_II but containing UTase/UR can increase GS to a level higher than that found in mutants lacking both P_II and UTase/UR (23). The possibility was suggested that UTase interacts with NR_II and slowly stimulates the phosphorylation of NR_I. Although most glnB mutants contain elevated levels of GS, a glnB mutant can be a glutamine auxotroph. The P_II in such a strain was shown to be insensitive to UTase/UR and thus was frozen as a signal of nitrogen sufficiency (51). In essence, these strains are very similar to strains without UTase/UR. Mutations in glnL and glnG are epistatic to those in glnB, which shows that P_II acts through NR_II and NR_I (23).

   glnL (NR_II) or ntrB (NtrB) mutants. NR_II is a bifunctional kinase dephosphorylation-stimulating enzyme. Genetic evidence suggests that these two functions can be separated. Some glnL mutants have a high level of GS even in the presence of ammonia. These mutants contain a P_II-insensitive NR_II (138). In other words, only the kinase activity of NR_II is functioning. Mutants have also been isolated which have a phenotype that suggests that only the dephosphorylation-stimulating activity of NR_II is active (3, 105). These strains are glutamine auxotrophs, because unphosphorylated NR_I can repress expression from glnAp ₁ but cannot stimulate transcription from glnAp ₂.

Strains with nonpolar loss-of-function mutations in glnL have an unexpected phenotype. In such E. coli and S. typhimurium mutants, GS synthesis is still responsive to changes in nitrogen availability (5, 119). For either species, loss of NR_II results in failure to utilize arginine as a nitrogen source and slow responses to changes in nitrogen availability (119, 153; L. J. Reitzer, unpublished observation). Since NR_I-phosphate is absolutely required for transcription from glnAp ₂ (79, 138) and transcription is initiated from glnAp ₂ in glnL mutants (153), NR_I must be phosphorylated by an NR_II-independent route. It has been shown that some small, high-energy phosphate-containing compounds, such as acetyl-phosphate and carbamoyl-phosphate, can phosphorylate NR_I without NR_II in vitro and in vivo (48).

   glnG (NR_I) or ntrC mutants. Loss of NR_I results in an Ntr^– Gln⁺ phenotype—that is, cells fail to utilize a number of nitrogen sources but are not glutamine auxotrophs (92, 143). In such mutants, glnA is transcribed from glnAp ₁ (153). The Ntr^– phenotype results from failure to induce GS and NR_I to sufficiently high levels and the subsequent failure to induce other Ntr proteins (92, 143, 153).

   rpoN or ntrA (σ⁵⁴) mutants. The rpoN (ntrA) gene codes for σ ⁵⁴, which is required for transcription from glnAp ₂ (57, 59, 73, 75). Expression of rpoN is not regulated by nitrogen availability (25). An rpoN mutant cannot utilize most nitrogen sources because of the low level of GS and the inability to express other σ ⁵⁴-dependent Ntr genes. An rpoN mutant is a glutamine auxotroph when grown in glucose-containing medium but a prototroph when grown with a poor carbon source, such as succinate. The difference results from the activity of the glnAp ₁ promoter, which is cyclic AMP dependent. A mutation in glnG (ntrC) suppresses the glutamine requirement of an rpoN grown in glucose-containing medium, since the loss of NR_I results in transcription from glnAp ₁ (73, 75, 153).

GLUTAMATE

Two enzymes, GDH and GOGAT, can synthesize glutamate from α-ketoglutarate for cells grown in glucose-ammonia minimal medium. The discovery of GOGAT, which catalyzes the NADPH-dependent transfer of the glutamine amide to α-ketoglutarate, was based on the observation that the intracellular pool of glutamate was normal in nitrogen-limited K. aerogenes, even though GDH was repressed (122, 178). It is worth noting that nitrogen limitation does not repress GDH as strongly in E. coli or S. typhimurium, so that this discovery was facilitated by studying K. aerogenes. This section discusses the enzymes of glutamate synthesis and their functions.

Phenotypes of Strains Lacking GDH, GOGAT, or Both

Strains lacking both GDH and GOGAT are glutamate auxotrophs (11, 19, 20, 106, 180). The auxotrophy can be satisfied by either glutamate or a compound, such as aspartate, that can donate its nitrogen via transamination to α-ketoglutarate. GDH-deficient strains of E. coli, K. aerogenes, and S. typhimurium have no discernible growth defect in glucose-containing (energy-rich) media (19, 180, 187). E. coli gdhA strains are, however, at a competitive disadvantage in an energy-limiting environment (71a). A second known effect of GDH deficiency is enhanced sensitivity to glutamate analogs (17). In contrast, GOGAT deficiency is more severe. GOGAT-deficient strains can synthesize glutamate (via GDH) only if the ammonium ion concentration is greater than about 1 mM (180). As discussed below, GOGAT deficiency also affects the ability to induce the Ntr response. Not surprisingly, GOGAT-deficient strains fail to utilize a variety of nitrogen sources (19, 135, 144, 180).

Glutamate Dehydrogenase

Properties of GDH.

GDH has been purified from E. coli and S. typhimurium. The enzyme is a hexamer of identical subunits with a molecular weight of about 300,000. The K_ms for ammonia and α-ketoglutarate are about 1 mM (35, 36, 165, 188). The amino acid sequences of GDH from K. aerogenes and E. coli are highly homologous to other bacterial GDHs and to GDHs from Neurospora crassa and Saccharomyces cerevisiae (115, 121, 127, 132, 184, 185).

gdhA Gene.

A careful mapping of E. coli gdhA, which codes for GDH, shows that the locus is at 38.6 min (71), which corrects a previous determination (144). The gene is at 27 min on the S. typhimurium map (163), which corresponds to the E. coli map position after a chromosomal rearrangement is taken into account (71). The gene for GDH has been cloned from E. coli, K. aerogenes, and S. typhimurium (37, 115, 204). The E. coli gene, which has been the most extensively characterized, is monocistronic and contains a single promoter (159). The region downstream from the structural gene contains two repetitive extragenic palindromic sequences (7). Their deletion reduces the half-life of the gdhA mRNA (124) but otherwise does not appear to participate in regulation.

Regulation of GDH Synthesis.

In E. coli, GDH is high for cells grown in glucose-ammonia minimal medium and is repressed by exogenous glutamate or aspartate (18, 20, 67, 186, 187). Glutamate represses as the sole carbon source or the sole nitrogen source or in medium containing ammonia. In S. typhimurium, exogenous glutamate does not repress GDH, although aspartate does (18). The mechanism of repression has not been studied, although it probably involves an elevated pool of glutamate or a compound derived from glutamate. The glutamate-dependent repression reduces gdhA transcription, although posttranscriptional regulation may also be involved (159).

Carbon limitation in E. coli reduces the steady-state level of GDH (67, 159). A CRP-binding site overlaps the –35 RNA polymerase-binding site, suggesting that CRP-dependent modulation probably involves steric blocking of RNA polymerase binding. This form of regulation may serve to prevent removal of excessive α-ketoglutarate from the citric acid cycle during carbon-limited growth.

Nitrogen limitation represses GDH synthesis strongly in K. aerogenes (10, 20), moderately in E. coli (159; although reference 159 should be consulted for a summary of reports to the contrary), and not at all in S. typhimurium (18). In K. aerogenes, the Nac (nitrogen assimilatory control) protein mediates this control. Nac is a transcriptional regulator that represses GDH synthesis and activates expression of a subset of Ntr genes, such as those for histidine and proline utilization. Nac itself is synthesized as part of the Ntr response (167; for a review, see reference 8). However, the activity of Nac does not appear to be controlled. The function of Nac may be to buffer the control of some Ntr genes from the exquisitely sensitive regulators of the Ntr response, P_II and UTase/UR. Therefore, even if nitrogen-limited cells enter an environment richer in nitrogen, Nac-dependent regulation would persist until Nac is diluted out. Such insensitivity may be desirable once a commitment to induce the Ntr response has been made.

Glutamate Synthase

Properties of Purified GOGAT and Its Subunits: an Unusual Amidotransferase.

GOGAT has been purified from E. coli W and K. aerogenes (128, 179). The enzyme has two nonidentical subunits in equimolar amounts. Although GOGAT has not been purified from S. typhimurium or E. coli K-12, the sizes of the subunits have been determined. The M _r of the small subunit is about 50,000 in all organisms, whereas that for the large subunit shows some variability, ranging from 135,000 to 175,000 (106, 128, 179). A variety of methods suggest that purified GOGAT is an octamer, although other results suggest that active GOGAT is a dimer that aggregates during purification (128, 179). Purified GOGAT contains flavin (both flavin adenine dinucleotide and flavin mononucleotide), iron (mostly ferrous), and labile sulfide; a stoichiometry of 1:4:4 was suggested from the data of Miller and Stadtman (128) and by analogy with other iron-sulfide-containing flavoproteins. This suggested stoichiometry is at some variance with that proposed by Trotta et al. (179). Flavin adenine dinucleotide, not flavin mononucleotide, has been implicated as the active species of flavin, but the data presented were considered inconclusive (128). The references provide more detailed kinetic, physiochemical, and enzymological information (16, 112, 128, 156, 179).

The large subunit of the K. aerogenes enzyme binds flavin, an iron sulfide, and glutamine (111, 112, 179). Several researchers have commented on the large size of the glutamine-binding subunit, which probably results from its participation in electron transfer, an unusual property for an amidotransferase. A glutaminase activity, a common property of amidotransferases (22), is associated with the large subunit (62, 111, 112, 179). The glutaminase activity of amidotransferases may be a damage-induced artifact (62, 198). This activity is not affected by the removal of the flavin, which implies that glutamine binding is flavin independent (62, 111, 179).

The purified small subunit of GOGAT can catalyze the ammonia-dependent formation of glutamate (111). Furthermore, a variety of treatments that inactivate or damage the glutaminase activity of the large subunit have no effect on the ammonia-dependent activity (62, 111, 112, 113, 179). In S. typhimurium, mutational loss of only the large subunit is sufficient to cause the characteristic phenotype of GOGAT-deficient strains (106). A gltB gdh strain of S. typhimurium, which has the small subunit of GOGAT but neither the large subunit of GOGAT nor GDH, requires glutamate for growth (106). Therefore, the small subunit has GDH-like activity only in vitro, not in vivo.

The reaction catalyzed by GOGAT occurs in two steps: reduction of the enzyme-bound flavin by NADPH, followed by the reaction of the reduced flavin with α-ketoglutarate and glutamine to generate oxidized flavin and two glutamates (62, 113, 128, 156). From the analysis of reaction mechanism and the biochemistry of the subunits, the following scheme for the overall reaction has been proposed. NADPH binds to the small subunit and transfers electrons to the large subunit, which reduces the flavin. α-Ketoglutarate binds the small subunit, and glutamine binds the large subunit. The glutamine amide is transferred to α-ketoglutarate, and the reduced flavin reduces a proposed iminoglutarate intermediate to glutamate. This reaction mechanism implies that both subunits are necessary for catalytic activity and thus is consistent with the genetic evidence cited above.

gltBDF Operon.

The gltBDF operon is located at 69 min on the E. coli chromosome and corresponding positions on the K. aerogenes and S. typhimurium chromosomes (54, 57, 144). The genes have been most extensively studied in E. coli (61, 104, 106). gltB and gltD specify the large and small subunits of GOGAT, respectively (37, 106). The operon also contains gltF, which may encode a transcriptional regulator, and a fourth open reading frame, which may specify a protein that has some effect on expression of the operon (26, 27). The major promoter of the operon precedes the first gene (140). Minor promoters may exist between gltB and gltD and after gltF (27). Two possible transcriptional terminators exist in the large intercistronic region between gltD and gltF; a terminator may also be present between gltF and DNA containing the fourth open reading frame (27).

Regulation of GOGAT Synthesis.

GOGAT catalyzes a reaction at a crossroads of carbon, energy, and nitrogen metabolism. Therefore, it is not surprising that a number of conditions affect the steady-state level of GOGAT. Two aspects of control of GOGAT synthesis are highly conserved. First, GOGAT is high in ammonia-containing minimal medium (18, 20, 128). Second, glutamate or glutamate-generating nitrogen sources, such as arginine, aspartate, histidine, and proline, repress GOGAT synthesis if growth is nitrogen limited (10, 18, 20, 128, 162). Apparently, this repression requires induction of the Ntr response for efficient transport and subsequent catabolism of these amino acids. Glutamate-dependent repression also accounts for the lower level of GOGAT in hisT mutants of S. typhimurium. The mutation results in undermodification of uridines in many anticodon loops of tRNA. Such mutants grow faster with arginine or proline as nitrogen sources, an effect attributed to enhanced arginine catabolism, which results in a postulated elevation in the glutamate pool (162).

GltF, the product of the third gene of the gltBDF operon, has been proposed to participate in glutamate-dependent repression (27). E. coli mutants without GltF, a potential transmembrane protein kinase, have elevated GOGAT in glutamate-containing medium. A alternative hypothesis has postulated that GltF is involved in Ntr-independent regulation of genes of proline, arginine, or glycine catabolism (88). If loss of GltF also impairs glutamate transport, pools of glutamate may be much lower in the mutant than in a wild-type strain. In such mutants, GOGAT levels should be higher. Resolution of this issue requires further work.

Growth in broth represses GOGAT in the enteric bacteria (18, 43). It would be reasonable to suppose that glutamate-dependent repression could account for this regulation. However, at least for E. coli, this regulation involves the leucine-responsive protein, Lrp, which controls the expression of at least 30 proteins (42, 43). For GOGAT synthesis, Lrp is considered a leucine-insensitive activator, although Lrp mediates a twofold repression by leucine (43). Lrp binds upstream from the start site of gltBDF transcription (43). The low level of Lrp in cells grown in Luria broth accounts for the low level of GOGAT in this medium (102). The physiological significance of Lrp-mediated regulation of GOGAT is discussed in Function of GOGAT in Nitrogen Assimilation, below.

Carbon limitation moderately represses GOGAT synthesis. A number of observations suggest that such regulation is physiologically significant. First, K. aerogenes grown in a carbon-limited chemostat has very low GOGAT levels (122). Second, exogenous cyclic AMP moderately represses GOGAT in E. coli grown in glucose-containing medium (148). Finally, a CRP-binding site overlaps the –35 RNA polymerase-binding site of the gltBDF operon in E. coli, suggesting a mechanism for CRP-dependent repression (140). Such a pattern of regulation may prevent GOGAT from draining α-ketoglutarate from the citric acid cycle when carbon or energy is limiting.

The stringent response also appears to control GOGAT synthesis. Such a possibility was first suggested from the phenotype of E. coli with gltX351, which encodes a temperature-sensitive glutamyl-tRNA synthetase. This strain was initially reported to have 10 times more GOGAT and GS than did a wild-type strain (99). A careful reexamination, which took into account the rapid accumulation of suppressor mutations, showed that GOGAT levels were actually lower in this mutant than in a wild-type strain (141). Changes in the intracellular pool of glutamate could not account for the diminished GOGAT level. Since a defect in a glutaminyl-tRNA synthetase also resulted in a low GOGAT level, it was suggested that the stringent response may participate in control of GOGAT levels (141). Consistent with this possibility is the observation that a relA mutant of S. typhimurium does not respond as well as a relA ⁺ strain to shifts in nitrogen availability (166). Stringent control of energy-consuming nitrogen assimilation may be selectively advantageous when macromolecular synthesis is inhibited.

In summary, the physiological factors that reduce or prevent formation of GOGAT are glutamate, growth in broth or carbon-limited media, the stringent response, mutations in hisT, and the presence of leucine. Factors that increase GOGAT are growth in ammonia-containing minimal media and in gltF mutants. Two DNA-binding regulators, Lrp and CRP, are known to control GOGAT synthesis. The mechanism of the glutamate-dependent repression has yet to be determined. In addition to these controls, the mechanism and significance of induction by phosphate limitation, which has not been discussed, has yet to be explored (71a, 126).

Functions of GDH and GOGAT

Relative Contributions of GDH and the GS-GOGAT Pathway for Glutamate Synthesis.

The most highly conserved feature of the regulation of GDH and GOGAT synthesis is that both are present at high levels in ammonia-containing medium (18, 20, 67, 128, 162, 186, 187). Furthermore, both enzymes in E. coli are regulated in parallel with various degrees of nutrient limitation (71a). For example, both enzymes are present in carbon- and energy-limited cells; however, because cyclic AMP and its binding protein moderately repress both enzymes, these enzymes are not at their highest level in such media (67, 122, 140, 148, 159). Their simultaneous presence raises the issue of their relative contributions to glutamate synthesis. GDH-deficient strains have no observable growth defect in energy-rich (glucose-containing) media (19, 180, 187). However, E. coli gdhA strains are at a competitive disadvantage relative to wild-type strains in carbon- and energy-limited media (71a). These results suggest that GDH is important for glutamate synthesis during energy-limited growth and that GOGAT is the primary enzyme of glutamate synthesis during energy-rich growth (71a). The advantage of such regulation is obvious: GS-GOGAT-dependent synthesis of glutamate consumes ATP, but GDH-dependent synthesis does not.

Although a mechanism to account for which enzyme synthesizes glutamate under different growth conditions was not presented (71a), a simple mechanism can be proposed. Because GS catalyzes the synthesis of glutamine, a substrate for GOGAT, the rate of the GS-catalyzed reaction necessarily limits the rate of glutamate formation by GOGAT. Therefore, the relative activities of GS and GDH, the major ammonia-assimilating enzymes of the cell, determine the relative contributions of GOGAT and GDH to glutamate synthesis. This implies that the substrates for GS and the numerous regulators of GS activity also control which pathway of glutamate formation will be utilized. GDH will be the primary enzyme of glutamate formation when the activity of GS is low, i.e., in energy-poor (low ATP), nitrogen-rich (ammonia-containing) minimal medium. GOGAT will synthesize glutamate when GS is activated, i.e., when nitrogen availability is limited. Nitrogen availability is sensed by the ratio of glutamine to α-ketoglutarate, which is also an indicator of relative energy sufficiency. Such a mechanism to account for glutamate synthesis under different conditions has not been tested. However, if this mechanism is correct, it implies that cells with GDH and GOGAT present at the same time have the capacity to adjust rapidly to changes in relative energy and nitrogen availability.

Nitrogen Source Utilization by GOGAT-Deficient Strains.

GOGAT-deficient strains grow poorly in a variety of nitrogen-limited media (20, 39, 135, 144; for a review, see reference 180). The only nitrogen sources that such strains can utilize are ammonia (if greater than 1 mM), aspartate, asparagine, glutamate, glutamine, and d-serine. Such mutants grow well in ammonia-containing media, since GDH can synthesize glutamate. They can probably utilize d-serine for the same reason; that is, ammonia production may be sufficient for glutamate synthesis by GDH.

GOGAT-deficient strains can utilize aspartate, asparagine, glutamate, and glutamine as nitrogen sources, presumably because these amino acids can generate glutamate. However, such strains cannot utilize other glutamate-producing nitrogen sources, such as proline or arginine. Whether a GOGAT-negative strain can utilize a particular glutamate-producing nitrogen source depends on whether its utilization requires induction of the Ntr response. Even though growth with aspartate and glutamine as nitrogen sources induces the Ntr response, glnG (Ntr^–) mutants can degrade these nitrogen sources (130, 143). In other words, utilization of these nitrogen sources does not require the Ntr response. In contrast, glnG mutants cannot utilize any other glutamate-producing amino acid as a sole nitrogen source (180), which implies that their utilization requires proteins of the Ntr response. In summary, GOGAT-negative strains fail to utilize any nitrogen source that absolutely requires induction of the Ntr response.

Function of GOGAT in Nitrogen Assimilation.

There are two views concerning the function of GOGAT during nitrogen assimilation. One view holds that GOGAT is essential for induction of the Ntr response because it prevents accumulation of glutamine, the primary product of ammonia assimilation, during nitrogen limitation. An alternative hypothesis proposes that the metabolic activities of GOGAT are neither necessary nor sufficient for induction of the Ntr response. Instead, it has been proposed that GltF, a regulatory protein formed in parallel with GOGAT, is necessary for control of the Ntr response. This section weighs the evidence for these assertions.

For growth with limiting ammonia, GOGAT is synthesized at a high level and is required (18, 122). In contrast, for cells grown with glutamate-generating nitrogen sources, such as arginine, GOGAT is actually repressed (18, 162). Although GOGAT is repressed in wild-type strains, GOGAT-deficient strains cannot utilize arginine as the sole nitrogen source (180). These apparently contradictory observations can be reconciled, if GOGAT is necessary for induction but not maintenance of the Ntr response. This conclusion is strongly supported by the observation that Ntr-constitutive, GOGAT-deficient mutants can utilize arginine and other glutamate-producing amino acids. In other words, for these nitrogen sources, Ntr constitutivity suppresses a possible requirement for GOGAT-dependent induction (20, 144, 180).

To understand the role of GOGAT in inducing the Ntr response, it is necessary to consider metabolism during the transition to nitrogen-limited growth. This transition can be sudden—for example, when alanine is added to cells growing in ammonia-containing medium (130). Whether it is sudden or gradual, as ammonia becomes unavailable, glutamine is depleted and the lower ratio of glutamine to α-ketoglutarate triggers the Ntr response. GS is deadenylylated, resulting in increased catalytic activity and more efficient ammonia assimilation. Most of the glutamine amide must then be used for synthesis of glutamate. Two independent observations support the proposition that GOGAT is the major enzyme that metabolizes glutamine. First, GOGAT deficiency results in a glutamine pool that is 46 times higher than that found in a wild-type strain (88). Second, GOGAT-negative strains grow very slowly with glutamine as the sole nitrogen source (144; L. J. Reitzer, unpublished observation). In summary, these results suggest that GOGAT prevents an accumulation of glutamine to maintain a physiological ratio of glutamine to α-ketoglutarate so that the Ntr response can be induced.

A nonmetabolic explanation for the Ntr^– phenotype of GOGAT-deficient strains postulates that GltF, not GOGAT, is necessary for the Ntr response (27). This hypothesis is based on the following results. Ntr^– GOGAT-deficient strains have been shown to have polar mutations that prevent expression of gltF, a member of the GOGAT-encoding gltBDF operon (26, 27). In addition, an insertion in gltF results in cells with an apparent Ntr^– phenotype, even though GOGAT activity is not affected (27). These cells could not induce synthesis of histidase, the product of an Ntr gene. Overexpression of gltF ⁺ suppressed two aspects of this Ntr^– phenotype of gltF strains: failure to utilize arginine as a nitrogen source and inability to induce histidase (26, 27). (Optimal suppression also required overexpression of gltD [small subunit] but not gltB [large subunit] [27].) The deduced amino acid sequence of GltF suggests that it is a transmembrane protein kinase. Therefore, it was proposed that GltF phosphorylates either NR_I or NR_II and that such phosphorylation is essential for induction of the Ntr response (27).

GltF is clearly an interesting and important protein that may be involved in nitrogen source utilization. However, overexpression of gltF only partially suppressed the Ntr^– phenotype of GOGAT-deficient strains. The level of GS, an Ntr gene product, was still low, and GS was overadenylylated (26, 88). These properties suggest that the suppressed cells still contain a high level of glutamine and that suppression does not result from induction of Ntr genes (88). An alternative explanation for the apparent suppression might be that overexpression of gltF ⁺ bypasses the requirement for induction of the Ntr response. It was suggested that enhanced catabolism of some amino acids could account for the suppression (88). In this case, the enhanced production of glutamate would suppress the need for GOGAT and induction of the Ntr response. This is possible since an NR_I-deficient, Ntr^– strain can utilize aspartate as the sole nitrogen source (130, 143). Therefore, the available evidence favors the hypothesis that the glutamine-depleting activity of GOGAT is required for induction of the Ntr response.

The involvement of GOGAT in the induction of the Ntr response could account for Lrp-dependent regulation of GOGAT synthesis and the low level of GOGAT in broth-grown cells. It has been proposed that Lrp is a sensor of general nutritional sufficiency (see reference 137 and chapter 94 of this volume for reviews). In a nutritionally rich environment, when Lrp is repressed, it would probably be disadvantageous to respond to the small or temporary fluctuations in nitrogen availability that trigger the sensors of the Ntr response. Therefore, without Lrp, which is required for GOGAT synthesis, the cell would not respond to such fluctuations. In contrast, for cells in minimal medium, rapid responses to changes in nitrogen availability would be advantageous. This explanation accounts for the high level of GOGAT from cells grown in ammonia-containing (nitrogen-rich) minimal medium, even though GOGAT is an enzyme required when nitrogen becomes limiting.

ASPARTATE

Aspartate is synthesized from oxaloacetate by pyridoxal 5'-phosphate-dependent transamination with glutamate as the amino donor:

oxaloacetate + l-glutamate

Bacteria contain transaminases with broad specificities (12, 164). However, because it has been possible to isolate an aspartate auxotroph and to clone the genes involved, the enzymes of aspartate synthesis have been definitively identified. In fact, these enzymes have become model systems for analysis of pyridoxal 5'-phosphate-dependent enzymes.

Genetics of Aspartate Synthesis

There has been only one genetic study of aspartate auxotrophy in the enteric bacteria (64). In E. coli K-12, aspartate auxotrophy results from mutations in aspC and tyrB. In addition to the requirement for aspartate, the double mutant requires tyrosine. Neither an aspC ⁺ tyrB nor an aspC tyrB ⁺ strain requires any amino acid for growth, although the latter strain produces small colonies on minimal medium agar plates, suggesting that AspC is the predominant aspartate transaminase.

Genes of Aspartate Synthesis and Their Expression

Expression of aspC, which specifies what has been called aspartate transaminase, is constitutive (34, 52, 64, 114, 116). Although AspC is the primary enzyme of aspartate formation, it can also synthesize phenylalanine and tyrosine in vivo (52). Numerous biochemical characterizations of the nonrepressible aspartate transaminase activity support this conclusion (31, 32, 34, 64,114, 116, 117, 146, 208). The tyrB gene encodes what has been called the aromatic amino acid transaminase. Tyrosine represses TyrB synthesis (34, 173, 189). TyrB catalyzes the synthesis of phenylalanine and tyrosine in vivo, although it can also catalyze the synthesis of aspartate (64). Biochemical studies confirm this specificity (69, 117, 146). Both aspC and tyrB have been cloned, and their sequences and map locations have been determined (52, 63, 91, 110). Both genes are monocistronic (52).

Properties of the Transaminases That Synthesize Aspartate

AspC and TyrB are dimers with virtually identical molecular weights (52, 146). Each subunit binds one molecule of pyridoxal 5'-phosphate. The deduced amino acid sequences indicate that AspC has 25% sequence identity and 40% sequence homology with aspartate transaminases of higher animals; residues essential for catalytic activity are absolutely conserved (52, 78, 85, 86). The aspartate transaminase from E. coli has lower specific activity and broader substrate specificity than the enzyme from animals (85, 208). Two structural features have been proposed to account for these substrate differences. The active site of AspC is more hydrophobic than that of the vertebrate aspartate transaminases (78, 82). This hydrophobicity may permit binding of the keto acid precursors for phenylalanine and tyrosine. Minor rearrangements of the backbone structure also contribute to substrate specificity (82). AspC and TyrB have 40% sequence homology (52). The active site pocket of TyrB is even more hydrophobic than that of AspC and could account for its substrate specificity (69).

ASPARAGINE

Genetics of Asparagine Synthesis

In E. coli and K. aerogenes, asparagine auxotrophs have mutations in two unlinked genes, asnA and asnB, which are located on the E. coli chromosome at 84 and 16 min, respectively, and in analogous positions on the K. aerogenes chromosome (47, 74, 145, 151). Neither an asnA ⁺ asnB nor an asnA asnB ⁺ strain of K. aerogenes has an observable defect when grown in an ammonia-containing medium; an asnA asnB ⁺ strain has no detectable defect in any medium (151). In contrast, an asnA ⁺ asnB mutant of K. aerogenes cannot grow on a variety of other nitrogen sources, such as proline, aspartate, nitrate, or glutamate, although it grows normally with ammonia or asparagine as the nitrogen source. This suggests that the function of the asnB product is asparagine synthesis during nitrogen-limited growth (151). As described in the next section, assay of enzyme activities from asnA and asnB strains suggested the presence of two asparagine synthetases (ASs) with different substrate specificities.

Properties of the Asparagine Synthetases

E. coli and K aerogenes, and probably other bacteria, have two ASs: a universally distributed glutamine-dependent enzyme and an ammonia-dependent enzyme, which is confined to prokaryotes (see the references cited in reference 151; for a review, see reference 150). These enzymes are encoded by asnB and asnA, respectively (74, 151). The reactions catalyzed by these enzymes are shown below; AMP and PP_i are products of all known ASs (123, 150). These enzymes do not share components.

glutamine + ATP + aspartate → asparagine + glutamate + AMP + PP_i

NH₃ + ATP + aspartate → asparagine + AMP + PP_i

The substrate specificities are sufficient to account for the phenotypes of strains that lack either AS. Strains with only the glutamine-dependent AS can synthesize asparagine with glutamine as a nitrogen donor and have no observable phenotype. However, strains containing only the ammonia-dependent AS cannot grow in nitrogen-limited media, apparently because of insufficient ammonia.

The ammonia-dependent AS has been partially purified from E. coli and K. aerogenes, and its kinetic properties have been described (28, 29, 151). The K_m for ammonia is about 0.3 mM at neutral pH. Asparagine is a potent inhibitor. The deduced amino acid sequence suggests a subunit molecular weight of 36,688 (24, 136). The native enzyme has a molecular weight of about 80,000, suggesting that it is a dimer (28, 151). In ammonia-containing medium, an asnA ⁺ asnB strain is a prototroph at 37°C but is an asparagine auxotroph at 42°C (40). This observation is consistent with the thermosensitivity of the ammonia-dependent enzyme (74). An analysis of the amino acid sequence of the ammonia-dependent AS showed that it has a protein motif found in class II aminoacyl-tRNA synthetases. It is most homologous to the aspartyl-tRNA synthetase, which catalyzes a similar reaction. Alteration of residues in this motif inactivated the enzyme. On the basis of these observation, it was suggested that the ammonia-dependent AS evolved from an ancestral aspartyl-tRNA synthetase (72).

The glutamine-dependent AS has been purified to apparent homogeneity from K. aerogenes (151). Asparagine inhibits the activity of the glutamine-dependent enzyme, as it does for the ammonia-dependent enzyme (151). It has properties typical of other amidotransferases: it shows substrate-independent glutaminase activity, and a high level of ammonia can replace glutamine as the nitrogen donor (151). Despite these similarities, the mechanism of catalysis of AS may differ from that of other amidotransferases (160). The purified enzyme is a tetramer of identical subunits; each subunit has a molecular weight of about 57,000 (151). The deduced molecular weight of AsnB subunits of E. coli is 62,666 (168), which agrees reasonably well with that of the K. aerogenes protein.

Regulation of Synthesis

Asparagine represses both ASs in E. coli and K. aerogenes (28, 74, 151). The mechanism of asparagine-dependent repression of asnA is understood to some extent. The product of the asnC gene of E. coli controls the expression of asnA. The asnC gene, adjacent to and divergently transcribed from asnA, encodes a 17,000-Da DNA-binding protein (24, 40, 84). AsnC activates transcription of asnA, and asparagine antagonizes this activation. AsnC also controls the expression of its own gene, although asparagine does not influence this autoregulation (40, 84). There is evidence to suggest that AsnC may also regulate expression of other genes and that this regulation may be posttranscriptional (40, 83).

Asparagine-dependent repression is not the only form of regulation observed. In nitrogen-limited K. aerogenes, the level of the glutamine-dependent AS is high and that of the ammonia-dependent enzyme is low. The converse is true for cells grown in nitrogen-rich medium (151). Nitrogen limitation results in an elevated level of NR_I, the glnG gene product. In a glnG mutant, the ammonia-dependent enzyme was not repressed, which suggested the possibility that NR_I represses asnA expression (151). However, there are no likely NR_I-binding sites upstream of asnA or asnC of E. coli (40). In summary, the mechanistic relationship between nitrogen availability and synthesis of the ASs is not understood.

l-ALANINE

E. coli (and undoubtedly other bacteria) contains transaminases with broad specificities (12, 164). Therefore, the starting point for understanding l-alanine synthesis should be a genetic analysis. However, an alanine auxotroph has yet to be identified. This section considers probable pathways and genes of l-alanine synthesis.

Enzymes of l-Alanine Synthesis

In most organisms, a glutamic-pyruvic transaminase (GPT) catalyzes l-alanine formation. Such an activity has been found in crude bacterial extracts (31, 45, 149, 183), although no attempt has been made to purify this protein. Nonetheless, this enzyme is likely to be involved in l-alanine synthesis in E. coli. Transaminases B and C, which were first identified in the initial survey of transaminases in E. coli (164), are probably also involved in alanine synthesis. Transaminase C (see below) catalyzes alanine formation, with valine as the amino donor. This reaction, together with the activity of transaminase B (see below), can lead to net synthesis of l-alanine.

The labeling of pyruvate-derived amino acids from specifically labeled glucose can be predicted from the known chemistry of glucose metabolism. Such predictions accurately accounted for the labeling of valine and leucine but not for the labeling of alanine, which was significantly lower than expected. This observation was consistent with much older observations that suggested a pyruvate-independent pathway of alanine synthesis (see reference 38 and references cited therein). A plausible alternative pathway has yet to be proposed.

Genes of l-Alanine Synthesis and Control of Their Expression

A tight alanine auxotroph has yet to be isolated, suggesting that more than one enzyme is responsible for l-alanine synthesis. The identification of leaky alanine requirers has been possible only for strains with requirements for other amino acids. In this section, the genes that have been implicated in l-alanine synthesis are discussed.

avtA.

The avtA gene encodes transaminase C, which can catalyze synthesis of either alanine or valine (12, 13, 46, 191, 199). E. coli and S. typhimurium strains with mutations in avtA have no observable phenotype (13, 199). The avtA strain is a valine prototroph, since transaminase B, the product of ilvE, also catalyzes valine synthesis. An avtA ilvE double mutant requires valine and isoleucine but not alanine for growth (13, 199). The avtA gene has been cloned from E. coli, and its map position has been determined (192).

Alanine but not valine represses AvtA synthesis, suggesting that AvtA is an alanine biosynthetic enzyme (13, 46, 199). In addition, leucine or starvation for any of a number of amino acids represses AvtA synthesis (13, 120, 200). It was suggested that amino acid starvation results in elevation of alanine or leucine concentrations (200). A structural similarity with alanine has been proposed to account for leucine-dependent repression (200). However, because of the control by leucine, the possibility of Lrp-dependent regulation should be seriously considered.

alaA.

Strains with mutations in alaA have no observable growth defect (12). The effect of a such a mutation can be detected only in an alaA ilvE double mutant, which requires either valine or alanine for good growth (12, 191). Without IlvE, AvtA catalyzes valine synthesis with alanine as the amino donor. The defect in AlaA probably reduces the synthesis of alanine, thereby resulting in a requirement for either alanine or valine (191). GPT activity is normal in an alaA strain, which is consistent with other evidence that alaA does not encode GPT (12, 191). Although alaA has been cloned and mapped on the E. coli chromosome, it has not been characterized (12).

alaB.

The alaB gene on a multicopy plasmid results in a high level of GPT (191), suggesting that it encodes a GPT. The plasmid with alaB suppresses the leaky alanine growth requirement in an alaA ilvE double mutant (191). The gene has not been further characterized.

d-ALANINE

Functions of the Two Alanine Racemases

For the enteric bacteria, d-alanine is synthesized by racemization of l-alanine. E. coli and S. typhimurium contain two distinct alanine racemases (194, 202). The control of their synthesis suggests that one is an anabolic enzyme and the other has a catabolic function. However, these distinctions are misleading. The so-called catabolic enzyme is the major alanine racemase under all growth conditions (190, 194, 202). In other words, the catabolic racemase is actually bifunctional, since it has an anabolic and a catabolic function.

Genetics of d-Alanine Formation

The alr gene encodes the anabolic alanine racemase in both organisms. The S. typhimurium dadB gene, called dadX in E. coli, encodes the catabolic enzyme. Mutations in both genes are required for d-alanine auxotrophy (194, 202). Loss of the catabolic racemase results in an inability to utilize l-alanine as the sole carbon, energy, and nitrogen source (194, 202). Mutants without the anabolic enzyme have no observable growth defect (194, 202).

The genes coding for the two alanine racemases are not linked (58, 194, 201). Both have been cloned from E. coli and S. typhimurium, and their sequences have been determined (14, 58, 101, 103, 193, 194). The anabolic and catabolic enzymes are highly homologous (58, 101). The dadX gene (catabolic alanine racemase encoding) in E. coli is part of the bicistronic dadAX operon (103, 202). DadA is probably the small subunit of d-amino acid dehydrogenase, a membrane-bound protein required for l- or d-alanine catabolism. The corresponding operon in S. typhimurium has not been characterized.

Regulation of Synthesis

Synthesis of the anabolic alanine racemase is constitutive (190, 202). Synthesis of the so-called catabolic alanine racemase of E. coli and S. typhimurium is induced by dl-alanine and repressed by glucose (53, 98, 194, 201, 202). l-Alanine appears to be the actual inducer in E. coli (202). Since a Δ cya strain cannot induce synthesis of the catabolic racemase, it was concluded that cyclic AMP and CRP are required for induction and mediate the glucose effect (103).

The dadAX operon has two transcriptional start sites separated by 5 bp. The weaker downstream promoter is used primarily when cells are grown without alanine. Its function appears to be for synthesis of basal alanine racemase for biosynthesis of d-alanine. The stronger upstream promoter is used when cells are grown with alanine. A cyclic AMP-CRP site is centered 60 bases upstream from the start site of the strong promoter. CRP complexed with cyclic AMP has been shown to bind to the promoter region of this operon (103).

Properties of Purified Alanine Racemases

Both alanine racemases are present in very low abundance. Their isolation required cloning of the genes and overproduction of the proteins. Only the racemases from S. typhimurium have been purified (44, 58, 190, 193). For both enzymes, the active species is a monomer of 39,000 Da with one pyridoxal 5'-phosphate per monomer. Both enzymes were purified from the soluble fraction of cell extracts, which implies a cytoplasmic localization. These results are consistent with the earlier fractionation studies (54, 201). Curiously, the specific activity of the purified biosynthetic enzyme is only about 1.5% that of the catabolic enzyme (44, 193).

ACKNOWLEDGMENTS

My work in this area has been funded by Public Health research grant GM38877 from the National Institute of General Medical Sciences. I am grateful to Catherine Bailey for preparation of the figures.