Biosynthesis of Glutamate, Aspartate, Asparagine, l-Alanine, and d-Alanine

Larry Reitzer

[SECTION EDITOR, GEORGES COHEN]

Posted July 6, 2004

Department of Molecular and Cell Biology, FO 31, The University of Texas at Dallas, 2601 N. Floyd Rd., Richardson, TX 75083-0688

Phone: 972-883-2502 or -2524, Fax: 972-883-2409, E-mail: mailto: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

Glutamate, aspartate, asparagine, l-alanine, and d-alanine are derived from intermediates of central metabolism, mostly the citric acid cycle, in one or two steps. While the pathways are short, the importance and complexity of the functions of these amino acids befit their proximity to central metabolism. The flux of nitrogen through glutamate is enormous. Inorganic nitrogen (ammonia) is assimilated into glutamate, which is the major intracellular nitrogen donor. Glutamate is a precursor for arginine, glutamine, proline, and the polyamines. Glutamate degradation is also important for survival in acidic environments, and changes in glutamate concentration accompany changes in osmolarity. The metabolic flux of carbon through aspartate is also large. Aspartate is a precursor for asparagine, isoleucine, methionine, lysine, threonine, pyrimidines, NAD, and pantothenate; a nitrogen donor for arginine and purine synthesis; and an important metabolic effector controlling the interconversion of C₃ and C₄ intermediates and the activity of the DcuS-DcuR two-component system. Finally, l- and d-alanine are components of the peptide of peptidoglycan, and l-alanine is an effector of the leucine-responsive regulatory protein and an inhibitor of glutamine synthetase (GS). Given the functional complexity of these amino acids and their proximity to central metabolism, perhaps it is not surprising that redundant enzymes synthesize them. Table 1 summarizes the genes and enzymes of glutamate, aspartate, asparagine, l-alanine, and d-alanine synthesis and the regulators and environmental factors that control the expression of these genes.

Table 1Genes of glutamate, aspartate, asparagine, L-alanine, and D-alanine synthesis and their regulation.

BIOSYNTHESIS OF GLUTAMATE

One gram of Escherichia coli cells contains 250 μmol of glutamate in protein. However, because glutamate is a major intracellular nitrogen donor, the cell requires the synthesis of ~8,200 μmol of glutamate per g (136). One gram of E. coli cells contains ~10,300 μmol of nitrogen, which means that ~80% of cellular nitrogen passes through glutamate during growth in an ammonia-containing minimal medium.

Enzymes of Glutamate Synthesis

In E. coli and other bacteria, either glutamate dehydrogenase (GDH) (reaction 1) or glutamate synthase (also called glutamine oxoglutarate amidotransferase [GOGAT]) (reaction 2) can synthesize glutamate. Both catalyze the NADPH-dependent reductive amination of α-ketoglutarate. GOGAT was discovered based on the observation that Klebsiella aerogenes had a normal intracellular pool of glutamate, even though GDH was completely repressed (108, 161). Glutamate auxotrophy requires loss of both GDH and GOGAT, and aspartate can satisfy the auxotrophy (5, 165). Loss of either GDH or GOGAT does not result in auxotrophy but nonetheless results in distinctive phenotypes, which are discussed below.

NH₃ + α-ketoglutarate + NADPH glutamate + NADP⁺ (1)

Glutamine + α-ketoglutarate + NADPH 2 glutamate + NADP⁺ (2)

NH₃ + glutamate + ATP glutamine + ADP + PO₄ ²⁻ (3)

Alternate pathways can synthesize glutamate if the medium contains certain organic nitrogen sources. Glutamate is the final product of arginine, proline, and probably ornithine catabolism (150). Furthermore, transaminases can transfer the amino group of aspartate, γ-aminobutyric acid, putrescine, or N-succinylornithine (an intermediate in arginine catabolism) to α-ketoglutarate (150, 151). These alternate routes of glutamate synthesis often require special growth conditions, such as nitrogen limitation (79, 151).

GDH and GOGAT not only synthesize glutamate, they also contribute to ammonia assimilation. The mechanism of GDH-dependent ammonia assimilation is obvious (reaction 1). In contrast, GOGAT is the second enzyme of a cyclic two-step pathway in which GS (reaction 3) assimilates ammonia and GOGAT (reaction 2) regenerates glutamate consumed by GS. The two pathways differ in their K_ms for ammonia—that of the GS-GOGAT pathway is much lower—and the requirement for ATP. The energy difference between the GDH and GS-GOGAT pathways is >10% of the cell’s ATP requirement (1 g [dry weight] of E. coli cells requires ~8,200 μmol of glutamate and ~72,000 μmol of ATP in an energy-rich environment [136].)

GDH

GDH and Its Gene.

GDH from E. coli and Salmonella enterica serovar Typhimurium is a hexamer with an M _r of ~300,000. The K_ms for ammonia and α-ketoglutarate are ~1 mM (27, 28, 145, 169). The E. coli gdhA operon is monocistronic and contains a single promoter (140). The region downstream from the structural gene contains two repetitive extragenic palindromic sequences (2). Their deletion reduces the half-life of the gdhA mRNA (111) but otherwise does not appear to participate in regulation.

Phenotype of gdhA Mutants.

GDH-deficient strains of E. coli, K. aerogenes, and S. enterica serovar Typhimurium grow normally in glucose-containing (energy-rich) minimal medium (17, 165, 168) but are at a competitive disadvantage in energy-limited medium (61, 62, 63). GDH deficiency also enhances sensitivity to glutamate analogs, which presumably reflects their effects on GOGAT activity (15).

GDH Synthesis and Degradation.

GDH activity is generally low for cells grown in broth and high for cells grown in glucose-ammonia minimal medium (16, 18, 167, 168). In ammonia-containing minimal media, glutamate and a variety of other amino acids may moderately repress GDH activity (up to threefold) (16, 98, 167, 168). Perhaps the various supplements make the ammonia-containing minimal media more like broth.

Growth in nitrogen-limited minimal medium represses GDH synthesis strongly in K. aerogenes (4, 18), moderately in E. coli (reference 140, which also contains reports to the contrary), and not at all in S. enterica serovar Typhimurium (16). Repression by Nac appears to account for both the repression and the species differences. Nitrogen limitation induces Nac in E. coli and K. aerogenes (3, 152). S. enterica serovar Typhimurium lacks the nac gene (116). Nac activity is itself not controlled, which implies that the level of Nac is important for regulation (51). Nac is stable in K. aerogenes and is rapidly degraded in E. coli, which presumably results in a low effective concentration (116). Nac binds to two sites of the K. aerogenes gdhA regulatory region (52). Strong and weak Nac-binding sites are centered at −89 and +57, respectively, relative to the transcription start site, and Nac appears to bind cooperatively to these sites. It has been proposed that Nac blocks the binding of an unidentified transcriptional activator near the upstream site. Growth-limiting nitrogen sources do not repress GDH to the same extent (167). The expression of nac varies with the nitrogen source (B. Schneider and L. Reitzer, unpublished observation), which may partially account for this differential expression.

Carbon limitation also reduces GDH activity in E. coli (55, 140). Two mechanisms have been proposed to mediate this control. A putative cyclic AMP receptor protein (CRP)-binding operator site overlaps the −35 RNA polymerase-binding site (140). However, the proposed site has not been inactivated. An alternative mechanism is carbon starvation-induced GDH degradation (104). GDH was the most rapidly degraded of 18 metabolic enzymes during nitrogen and carbon limitation. NADPH stimulates this degradation, which ATP, GTP, and guanosine 3' diphosphate 5' diphosphate (ppGpp) allosterically block. Chloramphenicol prevents this degradation, which suggests the involvement of an unstable protein. Lower activity does not imply a reduced significance for GDH during carbon- or energy-limited growth. It should be remembered that a GDH-deficient strain is at a competitive disadvantage in such an environment.

Despite the ever-increasing number of microarray analyses, few have suggested conditions that alter gdhA expression in E. coli. In most cases, the sequence of events that leads to altered expression is not clear. Microarray analyses have confirmed decreased gdhA expression for cells grown in broth and in nitrogen-limited medium (160, 195). Loss of the leucine-responsive regulatory protein (Lrp) increases gdhA transcripts threefold (69). This regulation is probably indirect. Loss of Lrp prevents gltBD expression, which in turn prevents induction of the Ntr response and the repressor Nac. These events are discussed below. Acivicin, a glutamine analog that inhibits glutamine-dependent amidotransferases, decreases gdhA transcripts fourfold (155). An early event in osmotic shock for cells in minimal medium is diminished gdhA expression, despite an increase in intracellular glutamate (179).

GOGAT

The Enzyme.

GOGAT has been purified from E. coli W and K. aerogenes (114, 164). The enzyme has two nonidentical subunits in equimolar amounts. The M _r of the small subunit is ~50,000 in these and related organisms, while that for the large subunit shows some variability, ranging from 135,000 to 175,000 (97, 114, 164). A variety of methods suggest that purified GOGAT is an octamer (114).

The purified small subunit of GOGAT can catalyze a GDH-like reaction, i.e., ammonia-dependent glutamate formation, which is not affected by inactivation of the large subunit (47, 99, 100, 101, 164). However, a gltB gdhA strain of S. enterica serovar Typhimurium, which has the small subunit of GOGAT but neither the large subunit of GOGAT nor GDH, requires glutamate for growth (97). Therefore, the small subunit alone is not sufficient for glutamate synthesis in vivo.

The large subunit of the K. aerogenes enzyme contains flavin, iron (mostly ferrous), and labile sulfide and binds glutamine (99, 100, 164). Purified GOGAT contains both flavin adenine dinucleotide and flavin mononucleotide—the former was suggested to be the active species (114). A stoichiometry of 1:4:4 was suggested for flavin to iron to sulfide (114). The large subunit can hydrolyze glutamine without amide transfer to an appropriate substrate (99, 100, 164). However, this activity is probably a damage-induced artifact (47). The references provide more detailed kinetic, physiochemical, and enzymological information (14, 100, 114, 139, 164).

gltBD Operon.

The adjacent gltB and gltD genes code for the large and small subunits of GOGAT, respectively (29, 97). A major promoter precedes gltB (122), although evidence has been presented for a minor promoter between gltB and gltD (21). Two possible transcriptional terminators are just downstream from gltD (21). It has been proposed that gltF, which is downstream from gltD, and the next downstream gene may be part of the gltBD operon because mutations in these downstream genes affect gltBD expression (20, 21). However, 460 bases separate gltD and gltF, and microarray analyses do not provide evidence for coexpression of gltB and gltD with downstream genes. Finally, compelling evidence against a regulatory function for gltF has been provided (53). It seems likely that the operon contains only two genes.

Phenotype of gltBD Mutants.

GOGAT-deficient strains fail to utilize a variety of nitrogen sources and cannot grow in medium with low concentrations (<1 mM) of ammonium ion (18, 34, 117, 125, 165). These two properties are related, since the slow catabolism of most organic nitrogen sources does not produce sufficient ammonia for GDH-dependent glutamate synthesis. Such strains can utilize ammonia (if the concentration is >1 mM), since GDH can synthesize glutamate. They 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. The difference between these groups of nitrogen sources is that utilization of the former group does not require the Ntr response whereas utilization of amino acids of the latter group does. GOGAT-deficient strains fail to induce the nitrogen-regulated (Ntr) response, which includes many transport systems for nitrogen-containing compounds (165). The basis for this phenotype is complex (53, 136) and is discussed elsewhere.

GOGAT Synthesis and Degradation.

Lrp is required for GOGAT synthesis in E. coli and K. aerogenes (37, 71). The gltBD operon is considered a leucine-insensitive Lrp-activated operon, although leucine actually reduces gltBD expression 2.2-fold (38). Purified Lrp binds to three upstream regulatory sites, which are centered at −152, −215, and −251 from the transcription start site (182). Alteration of these sites impairs expression in vivo (182). Leucine causes an 8.8-fold decrease in the binding of Lrp to the most proximal site (182). To account for the discrepancy between the effect of leucine on Lrp binding (an 8.8-fold decrease) and gltBD expression (a 2.2-fold decrease), it was proposed that an Lrp-leucine complex activates better than Lrp without leucine (182). The level of Lrp in cells appears to be optimal for gltBD expression. Higher levels decrease expression (182). IHF also binds the regulatory region, ~90 bases upstream from the transcription start site, and is required for expression in vivo (127). It is not known whether Lrp and IHF are sufficient for gltBD expression. If they are, then the gltBD promoter is unusual for a σ ⁷⁰-dependent promoter with respect to the distance between the activator and RNA polymerase-binding sites.

Guanosine tetraphosphate controls Lrp synthesis (88), which explains several aspects of the control of GOGAT synthesis. First, a relA mutant does not induce GOGAT as well as a relA ⁺ strain during a nutritional downshift (from broth to minimal medium) (146). This is presumably the result of diminished guanosine 3' diphosphate 5' triphosphate (pppGpp) or ppGpp. Second, GOGAT activity is low in broth and highest in glucose-ammonia minimal medium (16, 18, 38, 114). Presumably, (p)ppGpp and Lrp account for this control. Third, for S. enterica serovar Typhimurium grown in glucose-ammonia medium, glutamate does not repress GOGAT, although aspartate, an amino acid mixture, and Casamino Acids do repress it (16). A possible explanation is that cells in the more broth-like media (with more supplementation) have less (p)ppGpp and Lrp.

Nitrogen limitation represses gltBD expression, but only with glutamate or glutamate-generating nitrogen sources, such as arginine, aspartate, histidine, and proline (16, 18, 114), not with growth-limiting ammonia (53). Nac, which nitrogen limitation induces, represses GOGAT synthesis and appears to mediate this control in E. coli (53). Nitrogen source-dependent differences in nac expression (Schneider and Reitzer, unpublished observation) may explain the nitrogen source-dependent differences in gltBD expression. However, Nac-binding sites in the gltBD regulatory region and their relevance to control have yet to be established. Nac cannot account for repression by nitrogen limitation in S. enterica serovar Typhimurium (16), which lacks Nac. An unidentified factor may account for this control in S. enterica serovar Typhimurium. The nitrogen source-dependent repression may explain the reduced GOGAT activity in hisT mutants of S. enterica serovar Typhimurium (143). These mutants have undermodified uridines in many tRNA anticodon loops. For unknown reasons, these mutants grow faster with arginine or proline as the nitrogen source. The effect on growth was attributed to enhanced arginine catabolism and the resulting elevation in the glutamate pool. Nitrogen limitation has little effect on the rate of GOGAT degradation (104).

A number of observations suggest the physiological relevance of repression of GOGAT by carbon limitation. First, exogenous cyclic AMP moderately represses GOGAT for E. coli grown in glucose-containing medium (132). Second, K. aerogenes grown in a carbon-limited chemostat has very low levels of GOGAT (108). Finally, a CRP-binding site overlaps the −35 RNA polymerase-binding site of the gltBDF operon in E. coli, suggesting a possible mechanism for CRP-dependent repression (122). In addition to this control, the rate of GOGAT degradation is ~15% per h for carbon-limited cells (104). The effects of carbon limitation are easily rationalized. GOGAT consumes α-ketoglutarate. When carbon or energy is limiting, GOGAT could potentially drain α-ketoglutarate from the citric acid cycle, which is obviously detrimental.

Several aspects of the control of GOGAT synthesis have yet to be explained. First, phosphate limitation induces GOGAT synthesis (112). Second, an E. coli gltX351 mutant encodes a temperature-sensitive glutamyl-tRNA synthetase. This strain was reported to have 10 times more GOGAT and GS than a wild-type strain (89). A careful reexamination showed that these strains rapidly acquired suppressor mutations and that levels of GOGAT were actually modestly lower (124). It is not clear whether the effect of this mutation involves a known regulatory mechanism.

Several microarray analyses have indicated changes in gltBD expression. Growth in broth and loss of either Lrp or IHF predictably diminish expression (1, 69, 160). The mechanisms that alter gltBD expression from other microarray analyses are not readily apparent. Hydrogen peroxide, diauxie, and entry into stationary phase diminish gltB expression (24). A bifunctional alkylating agent in a mutS strain diminishes gltD expression (147). Acivicin, a glutamine analog, diminishes gltD expression (155). Biofilms have dramatically lower gltBD expression than cells in stationary phase (149). Loss of SeqA and elevated dam expression decrease gltBD expression (94), whereas a dam mutant has increased gltBD expression (123). An H-NS deficiency increases gltB expression (67). An early event in osmotic shock is diminished gltBD expression (179). This was unexpected, since glutamate accumulates during osmotic shock. Furthermore, this change is seen despite an increase in himD (IHF) expression (179). 4,5-Dihydroxy-2-cyclopenten-1-one, an antibiotic, stimulates gltB expression fourfold in broth (128). Finally, ingestion of E. coli by normal and phagocyte oxidase-deficient neutrophils diminishes gltB expression twofold (156). Unless otherwise mentioned, none of these changes is correlated to changes in Lrp or IHF synthesis.

BIOSYNTHESIS OF ASPARTATE

It has been estimated that 230 μmol of aspartate is present in proteins per g cells. However, aspartate is also the precursor for several amino acids and is a nitrogen donor in arginine and purine synthesis. The cell must synthesize ~2,770 μmol of aspartate to meet these requirements (136). In other words, ~27% of cellular nitrogen passes through aspartate.

Genetics of Aspartate Synthesis

There has been only one genetic study of aspartate synthesis in the enteric bacteria (48). In E. coli K-12, aspartate auxotrophy requires mutations in both aspC and tyrB. In addition to the requirement for aspartate, this double mutant also requires tyrosine, but not phenylalanine. (Two transaminases, TyrB and AspC, catalyze the last reaction in tyrosine synthesis, while three transaminases, TyrB, AspC, and IlvE, catalyze the last reaction in phenylalanine synthesis.) Both AspC and TyrB catalyze the pyridoxal 5'-phosphate-dependent transamination of oxaloacetate with glutamate as the amino donor.

oxaloacetate + glutamate ↔ aspartate + α-ketoglutarate

Therefore, this is the only route of aspartate synthesis in E. coli and probably in most organisms. aspC ⁺ tyrB and aspC tyrB ⁺ strains are aspartate prototrophs. However, an aspC ⁺ tyrB strain produces larger colonies on minimal-medium agar plates than an aspC tyrB ⁺ strain, which suggests that AspC is the predominant aspartate transaminase (48).

Properties of AspC and TyrB

Although AspC is the primary enzyme of aspartate formation, it can also synthesize phenylalanine and tyrosine in vivo (48) and in vitro (57, 106, 131, 189). Conversely, although TyrB preferentially catalyzes synthesis of phenylalanine and tyrosine, it can also synthesize aspartate in vivo and in vitro (48, 57, 131). In addition to these reactions, TyrB also participates in methionine recycling in Klebsiella pneumoniae (59). This is consistent with an early observation that E. coli TyrB can deaminate methionine (25). Kinetic parameters have been determined for purified AspC and TyrB for many of the substrates (57, 106, 131, 189). The aminotransferases have been classified into four groups based on protein sequence alignment, hydropathy plot patterns, and theoretical secondary structure (109). AspC and TyrB constitute the α subgroup of group I (73). The group I aminotransferases catalyze reactions with small or aromatic amino acids (109).

Both AspC and TyrB are dimers with virtually identical molecular weights (43, 131). Each subunit binds one molecule of pyridoxal 5'-phosphate. AspC has 25% sequence identity and 40% sequence homology with aspartate transaminases of higher animals (43, 83, 84). Residues essential for catalytic activity are absolutely conserved (83). The aspartate transaminase from E. coli has lower specific activity and broader substrate specificity than those from animals. The structure of AspC from E. coli has been determined, and differences in the active site and minor rearrangements of the backbone structure probably account for these differences (76, 80). The active-site pocket of TyrB is even more hydrophobic than that of AspC and could account for its substrate specificity (57).

Genes of Aspartate Synthesis and Their Expression

Both aspC and tyrB are monocistronic (43). An apparent transcription start site for aspC has been determined (43). Aspartate and the aromatic amino acids do not affect AspC synthesis (25, 105). Such regulation is not surprising, considering that aspartate is the precursor for cofactors and other amino acids. Tyrosine represses TyrB synthesis (25, 105, 153), and the mechanism of this control is considered elsewhere. Despite the absence of specific regulation of AspC, 3 (out of ~50) microarray analyses suggest that there is regulation of AspC synthesis. Levels of aspC transcripts are three-fold higher in minimal medium than in rich medium (160). Levels of aspC transcripts are twofold higher in cells with a low concentration of Mg²⁺ and twofold lower in cells without PhoP, the response regulator of the PhoQ/PhoP two-component regulatory system that senses extracellular Mg²⁺ (115). Acivicin, a glutamine analog that inhibits many glutamine-dependent reactions, induces aspC expression 7.5-fold (155).

BIOSYNTHESIS OF ASPARAGINE

The only need for asparagine appears to be as a protein constituent, and the requirement is for ~230 μmol per g of E. coli cells, or ~4.5% of the cell’s nitrogen requirement (136).

Genetics of Asparagine Synthesis

Asparagine auxotrophs in E. coli and K. aerogenes are double mutants with lesions in two unlinked genes, asnA and asnB (42, 68, 138). An asnA asnB ⁺ strain has no observable phenotype (138). In contrast, an asnA ⁺ asnB mutant of K. aerogenes is an asparagine auxotroph during nitrogen-limited (ammonia-limited) growth, since it can utilize asparagine but not other nitrogen sources, such as proline, aspartate, nitrate, or glutamate (138). The substrate specificities of AsnA and AsnB, which are asparagine synthetases, account for these mutant phenotypes and are described below. In addition, an asnA ⁺ asnB mutant is a temperature-sensitive asparagine auxotroph (35), because AsnA is thermolabile (68).

Properties of Asparagine Synthetases

E. coli and K. aerogenes have glutamine-dependent and ammonia-dependent asparagine synthetases, which are encoded by asnB and asnA, respectively (68, 138). They catalyze the following reactions:

(AsnB) glutamine + ATP + aspartate → asparagine + glutamate + AMP + PPi

(AsnA) NH₃ + ATP + aspartate → asparagine + AMP + PPi

AMP and PPi are products of all known asparagine synthetases (19, 110, 135, 190). The ammonia-dependent enzymes appear to be confined to prokaryotes (19, 190). The nitrogen donor substrate accounts for the unusual phenotype of the asnA ⁺ asnB strain (138). Strains with only AsnA require ammonia for asparagine synthesis, which becomes a serious problem when the ammonia level is low, i.e., during nitrogen-limited growth. Therefore, AsnB is required for nitrogen-limited growth. (It should be noted that the synthesis of glutamate is also catalyzed by separate ammonia- and glutamine-dependent enzymes, i.e., GDH and GOGAT, respectively. In this case, loss of the glutamine-dependent reaction also results in failure to utilize a variety of alternate nitrogen sources, whereas loss of the ammonia-dependent activity results in cells with only subtle defects.)

AsnA has been purified from E. coli and partially purified from K. aerogenes (66, 138, 159). The kinetic properties of this enzyme have been described elsewhere (23, 138). The K_m for ammonia is ~0.3 mM at neutral pH, and asparagine is a potent inhibitor. AsnA is a homodimer (138, 159) that appears to be sensitive to oxidative stress (36), and as already mentioned, it is thermolabile (68).

AsnB has been purified from K. aerogenes (138) and E. coli (10). The purified enzyme is a homotetramer (138), and asparagine inhibits its activity (138). AsnB has properties typical of other amidotransferases: substrate-independent glutaminase activity and a high level of ammonia can replace glutamine as a nitrogen donor (10, 138). The kinetic parameters of AsnB (10, 138) and its structure (90) have been determined, substrate binding sites have been probed (12, 126), and the mechanism of nitrogen transfer has been examined (10, 141, 157). The steady-state kinetic mechanism has also been explored (11, 162).

AsnA and AsnB are not homologous, and it has been argued that AsnB is not derived from AsnA (190). Instead, AsnA is homologous to aspartyl tRNA synthetase, which catalyzes a similar reaction (65). It has been proposed that AsnA evolved from an ancestral aspartyl tRNA synthetase, which is consistent with their structural similarities (65, 118).

Regulation of Synthesis

Asparagine represses both asparagine synthetases in E. coli and K. aerogenes (22, 68, 138). AsnC mediates the repression of asnA. The asnC gene is adjacent to and divergently transcribed from asnA, and AsnC is a DNA-binding protein (35, 82). AsnC activates the transcription of asnA, and asparagine prevents this activation. AsnC also controls its own synthesis, although asparagine does not influence this autoregulation (35, 82, 129). There is suggestive evidence that AsnC may also regulate the expression of other genes (35, 81), and this regulation may be posttranscriptional (81). Nothing is known about the asparagine-dependent repression of AsnB or the proteins that control asnB expression.

Nitrogen limitation (utilization of a nitrogen source that limits growth) represses AsnA (129, 138, 195), and Nac mediates this repression. Nitrogen limitation induces NtrC (also called NR_I), the ntrC (also called glnG) gene product, which in turn activates nac expression. Nac represses asnC, which prevents asnA expression (129). Purified Nac binds the asnC regulatory region, and putative Nac sites have been identified (129).

Only two microarray analyses indicate changes in asnA and asnB transcripts. In the first study, neutrophil ingestion of E. coli repressed asnA (2.2-fold) and asnB (3.5-fold) (156). This situation also repressed gltB, which is poorly expressed in rich medium. A plausible explanation is that the neutrophil environment is rich in amino acids, including asparagine, which can repress both asnA and asnB. In the second study, variation of the carbon source also affected asnA and asnB expression (121). Relative to growth with glucose, growth with acetate reduced asnA and asnB expression ~2-fold, whereas growth with glycerol increased the expression of both enzymes ~3-fold.

BIOSYNTHESIS OF l-ALANINE

Alanine, an essential component of proteins and cell wall peptidoglycan, is at a branch point in metabolism. The requirement for protein alanine is 488 μmol per g (dry weight) (119). The requirement for peptidoglycan alanine is 55 μmol per g (dry weight) (119), which implies that major perturbations that affect cell wall synthesis should only modestly affect net alanine synthesis. The total requirement for alanine is 5.3% of the cell’s nitrogen (136).

Possible Pathways and Enzymes of l-Alanine Synthesis

An l-alanine auxotroph has yet to be isolated, which means that the genes and enzymes of alanine synthesis are not known. It seems likely that multiple enzymes catalyze alanine synthesis. The simplest pathway of alanine synthesis is transamination of pyruvate, catalyzed by glutamic-pyruvic transaminase (GPT). However, this is not known with certainty, since the anomalous labeling of alanine from labeled glucose is consistent with the possibility of a pyruvate-independent pathway (30). Furthermore, GPTs might not be the only enzymes that catalyze alanine synthesis. GPT activity can be a side reaction of alanine racemase (85). The cysteine desulfurases convert cysteine to alanine and sulfane sulfur (S⁰), which is required for the formation of Fe-S clusters, thiamine, thionucleosides in tRNA, biotin, lipoic acid, and molybdopterin (113). It is difficult to estimate the relative contributions of the transaminase and desulfurase pathways, but it is suspected that the latter is not a major pathway of alanine synthesis. (The total elemental sulfur in E. coli is ~210 ± 70 μmol per g [dry weight] [60], which is essentially the sulfur content of protein cysteine plus methionine, which is ~233 μmol [119]. The amount of sulfur in specific S⁰-requiring components is generally not known, except for the sulfur in iron-sulfur clusters. E. coli contains ~2.5 nmol of iron per g [dry weight] [120]. If all of this iron is complexed to S⁰, which is an overestimate, then this component of the desulfurase pathway synthesizes 0.0005% of cellular alanine.)

Transaminases of Alanine Synthesis

GPT activity has been found in crude bacterial extracts (40, 134). However, no attempt has been made to purify GPT. Transaminase C (see below) catalyzes alanine formation with valine as the amino donor. This reaction, together with transaminase B (see below), can lead to net synthesis of l-alanine. Transaminases B and C were first identified in the initial survey of transaminases in E. coli (144).

Pyruvate + glutamate l-Alanine + α-ketoglutarate

Valine + pyruvate l-Alanine + α-ketoisovalerate

Glutamate + α-ketoisovalerate α-Ketoglutarate + valine

avtA : Transaminase C.

The avtA gene encodes transaminase C (6, 7, 41, 180). E. coli and S. enterica serovar Typhimurium avtA mutants have no observable phenotype (7, 180). The avtA strain does not require valine, since transaminase B, IlvE, also catalyzes valine synthesis. The loss of avtA is apparent only in an avtA ilvE double mutant, which requires valine and isoleucine, but not alanine, for growth (7, 180).

Alanine, but not valine, represses AvtA, which suggests that AvtA synthesizes alanine (7, 41, 180). In addition, leucine or starvation for any of a number of amino acids also represses AvtA synthesis (7, 107, 181). It was suggested that amino acid starvation results in elevation of the alanine or leucine concentration and subsequent repression (181). A structural similarity with alanine has been proposed to account for leucine-dependent repression (181). An alternative explanation is based on the observation that both leucine and alanine bind Lrp, which suggests that Lrp might control avtA expression. However, a microarray analysis of an lrp mutant does not support this conjecture (69). Several microarray analyses of E. coli have detected changes in avtA expression. Cells fermenting glucose had a higher level of avtA transcripts than cells fermenting xylose (50). E. coli cells treated with acivicin, an inhibitor of glutamine-dependent amidotransferases, have two-fold fewer avtA transcripts (155). Perhaps the accumulation of alanine or leucine, which do not require glutamine for their synthesis, accounts for this effect. Third, a dam mutant had 2.5-fold more avtA transcripts than a wild-type strain (123). Finally, exposure to paraquat, which generates superoxide and induces the soxRS regulon, increases avtA expression 1.7-fold (130).

alaA .

Mutagenesis of an ilvE mutant resulted in isolation of a strain with a leaky requirement for either alanine or valine (172). The leaky requirement results from the action of AvtA, which is still present and which can synthesize valine or alanine if the other is provided. The second lesion was designated alaA. An alaA mutant has no observable growth defect (6). It was proposed that AlaA is not a GPT, based on normal GPT activity in an alaA mutant and other evidence (6, 172). alaA has been cloned and mapped on the E. coli chromosome (6); however, it has not been characterized and it has not been identified with a gene of the sequenced genome.

alaB .

Plasmid-borne alaB suppresses the leaky alanine-valine requirement of an ilvE alaA strain (172). Strains with this plasmid have unphysiologically high levels of GPT, which suggests that AlaB is a GPT (172). alaB has not been further characterized.

BIOSYNTHESIS OF d-ALANINE

d-Alanine is required for peptidoglycan synthesis. Peptidoglycan contains d- and l-alanine in a ratio between 2 and 1. The total peptidoglycan alanine is 55 μmol per g of E. coli cells. Therefore, the total requirement for d-alanine is between 27 and 36 μmol, or ~0.3% of the cell’s nitrogen (136).

Genetics of d-Alanine Formation

d-Alanine auxotrophy in E. coli and S. enterica serovar Typhimurium requires mutations in two genes, both of which specify an alanine racemase (176, 183). Therefore, the only route of d-alanine synthesis is racemization of l-alanine. The alr gene encodes one of the alanine racemases in both organisms (171, 183). The S. enterica serovar Typhimurium dadB gene, called dadX in E. coli, specifies the other racemase. (The E. coli dadB designation had been previously assigned to a gene that appears to code for the large subunit of d-amino acid dehydrogenase [45].) The DadB(X) racemase is the predominant racemase in both organisms (171, 183). Loss of this racemase results in an inability to utilize l-alanine as a sole carbon, energy, and nitrogen source (176, 183).

The alanine racemase genes are not linked (176, 183). Both have been cloned from E. coli and S. enterica serovar Typhimurium (46, 95, 175, 176). dadB(X) is the second gene of a dadAB(X) operon (95, 183). DadA is the small subunit of d-amino acid dehydrogenase, a membrane-bound protein required for l- or d-alanine catabolism (176, 184, 185). An operon structure has not been definitively established in either S. enterica serovar Typhimurium or K. aerogenes, but dehydrogenase and racemase activities are coregulated in the former (176). Loss of the dad operon in K. aerogenes results in an alanine-sensitive strain (70). The basis of this phenotype results from alanine inhibition of GS, which blocks glutamate synthesis by the GS-GOGAT pathway. Inhibiton of GS also induces the Ntr response. The resulting induction of Nac completely represses gdhA expression, which blocks the GDH pathway of glutamate synthesis. Nac does not completely repress gdhA in E. coli, which suggests that an E. coli dad mutant should not be alanine sensitive.

The Alanine Racemases

The racemases have been purified and characterized from S. enterica serovar Typhimurium but not from other enteric bacteria (39, 175). The racemases appeared to be monomeric, with 1 mol of pyridoxal 5'-phosphate per monomer (39, 175). However, the existence of dominant-negative alleles implies multimers, and the activity of mixed oligomers suggests that active racemases in E. coli and other organisms are homodimeric (158). Both S. enterica serovar Typhimurium enzymes were purified from the soluble fractions of cell extracts, which implies a cytoplasmic localization (39, 175). Curiously, the V _max of purified DadB is 60 times greater than that of purified Alr (171). The basis for this difference is not clear, considering the 52% identity of the amino acid sequences (46). The alanine racemases are targets of antibiotics, such as d-cycloserine. The properties of the alanine racemases have been reviewed elsewhere (171).

Regulation of Synthesis

Expression of alr appears to be constitutive in S. enterica serovar Typhimurium (171), and it is not affected by the carbon source, l-alanine, or d-alanine in minimial medium in E. coli (183). Of ~50 microarray analyses examined, none reported a change in alr transcription.

Expression of dadAX is induced by dl-alanine and repressed by glucose (44, 87, 176, 184, 185). l-Alanine appears to be the actual inducer in E. coli (183). Lobocka et al. suggested two promoters, one for basal expression and a second for inducible expression (95). Zhi et al suggest at least three promoters (191). The mechanism of induction is complex and has been most thoroughly studied in E. coli both in vitro and in vivo (103, 191, 192). Lrp binds to 11 sites in the promoter region (192). Binding to the promoter-proximal sites represses expression, which alanine or leucine relieves (192). The binding of Lrp-alanine to the upstream promoter-distal sites activates expression (192). Similarly, Lrp both represses and activates the dad operon of K. aerogenes (71). Lrp also represses the dad operon in S. enterica serovar Typhimurium (58), although the details of regulation differ from those of E. coli (discussed in reference 103). An E. coli Δ cya strain cannot induce dadAX (95). Mobility shift experiments identified an upstream cAMP-CRP site (95), which mutational alteration and footprinting more precisely located (191). The CRP-dependent control accounts for the catabolite repression. In addition to these control mechanisms, dadAB expression in K. aerogenes requires Nac, unlike expression in E. coli and S. enterica serovar Typhimurium (70). Only two E. coli microarray analyses suggested changes in dadX expression. First, a dam mutant had about fivefold more dadX transcripts than a wild-type strain (123). Second, a strain that overproduces threonine has a twofold reduction in dadX transcripts compared to a wild-type strain (91).

TRANSPORT

Glutamate, aspartate, asparagine, l-alanine, and d-alanine have multiple transport systems. The properties of the known systems are summarized in Table 2. The genes for most of the major amino acid transport systems have been identified, although it would not be surprising if more systems are subsequently identified. The regulation and functions of many of these systems have not been thoroughly studied. Microarray analyses have provided the first information on the control of some of these systems.

Table 2Summary of transport systems

Glutamate Transport

E. coli has three kinetically and genetically characterized glutamate transport systems, which have been called GltI, GltII, and GltIII. Wild-type E. coli cannot utilize glutamate as a carbon source. However, mutants with elevated GltI or GltII activities can utilize glutamate as a carbon source (32, 102). gltS (called gltC in early publications) specifies the GltI system (32, 33, 75, 102). This system transports glutamate and α-ketoglutarate, but not aspartate. GltS-dependent transport requires sodium and is inhibited by d-glutamate, glutamine, α-methylglutamate, and homocysteate (reference 148 and references therein). Two microarray analyses suggest changes in gltS expression. First, xylose fermentation results in higher gltS expression than glucose fermentation (50). This might reflect growth rate-dependent control. Second, gltS expression was affected more than that of any other gene in a seqA mutant and a strain that overproduced the Dam methylase (94). The mechanism of this change is not obvious, since the gltS promoter region does not contain Dam sites.

gltP encodes the GltII system (13, 32, 148, 163, 170). This system transports glutamate and aspartate with nearly equal affinities (32, 148). Glutamate transport is inhibited by d-glutamate, aspartate, asparagine, glutamine, and the analogs cysteate and β-hydroxyaspartate (32). Transport is driven by respiratory substrates (i.e., the proton motive force) and is independent of Na⁺, K⁺, and Li⁺ (32). Carbon-limited growth (with either aspartate or glycerol as a carbon source) increases glutamate transport in a mutant that can utilize glutamate as a carbon source (74). This result can be reinterpreted to suggest that carbon limitation induces GltP but not GltS. The GltII system is the major glutamate and aspartate carrier, at least when cells are grown with lactate as the carbon source (148). I am unaware of a microarray analysis that suggests changes in gltP expression.

The GltIII system has a periplasmic binding protein (148, 187), which appears to be the product of gltI (also called ybeJ) (137). gltI is coexpressed with three downstream genes, which specify proteins of an ABC transport system. The purified binding protein has a high affinity for glutamate and aspartate (187). Cysteate inhibits this transport system (148). Nitrogen limitation induces GltI (the protein, not the transport system) in S. enterica serovar Typhimurium (86) and the gltIJKL genes in E. coli (195). An unusual feature of the gltIJKL operon is an insertion element between the first structural gene and the likely σ⁵⁴-dependent promoter (137). In addition to induction by nitrogen limitation, an alkylating agent increases gltJ expression twofold in a mutS strain (147), and acivicin, a proposed inhibitor of glutamine amidotransferases which probably perturbs metabolism, represses gltJ expression 2.5-fold (155).

A fourth glutamate transport system, XasA/GadC, is required for glutamate-dependent survival in extremely acidic environments (pH 2.5) (64, 178). GadC appears to catalyze glutamate–γ-aminobutyrate exchange (177). The two substrates are linked by glutamate decarboxylase (GadB), which generates the γ-aminobutyrate from the imported glutamate. σ ^S-dependent expression is activated by GadX and GadW and repressed by GadW, H-NS, and CRP (178). In the absence of H-NS, σ ⁷⁰ can activate expression (178).

Aspartate Transport

There are at least six aspartate transport systems. As mentioned above, biochemical characterizations suggest that the GltP (GltII) and GltIJKL (GltIII) systems transport aspartate. The only genetic evidence consistent with the biochemical evidence is that a GltP up mutant grows faster with aspartate as the sole nitrogen source (148).

The Ast system is a high-affinity aspartate-specific system that does not transport either glutamate or dicarboxylic acids (77, 148). Loss of Ast results in significantly slower growth with aspartate as the nitrogen source, which suggests that it is a major aspartate transport system (77). However, Schellenberg and Furlong estimate that the gltP, gltIJKL, and ast systems account for 52, 25, and 23% of aspartate transport, respectively—that is, the ast system is not a major system (148). Although ast mutants exist, the gene has not been mapped.

C₄-dicarboxylic acids inhibit aspartate transport in an ast mutant, which suggested that systems that transport C₄-dicarboxylic acids also transport aspartate (78). The aerobic dicarboxylic acid system, Dct, transports aspartate, albeit with low affinity (77). Loss of the Dct system alters the kinetics of aspartate uptake but has no effect on growth with aspartate as the nitrogen source (77). The Dct system consists of two integral membrane proteins, DctA and DctB, and a binding protein, Cbt. Curiously, purified DctA, DctB, and Cbt do not bind aspartate at the concentrations tested (92, 93). The nucleotide sequence for dctA is known, but those for dctB and cbt are not known, even though the last two genes have been mapped (reference 31 and references therein). DctA is homologous to GltP (31). An electrochemical proton gradient drives Dct-dependent aspartate transport (54), which is similar to GltP-dependent transport. dctA expression is activated by CRP and is weakly activated by the DcuS-DcuR two-component regulatory system in response to C₄-dicarboxylic acids, aspartate, and citrate (31). The CRP-dependent control may account for the elevated expression of dctA during xylose fermentation relative to glucose fermentation (50). dctA is repressed by ArcA during anaerobic growth and apparently by the DctA protein itself (31).

The DcuA, DcuB, and DcuC systemstransport dicarboxylic acids during anaerobic growth in E. coli (154, 194). A dcuA dcuB double mutant cannot utilize aspartate anaerobically, whereas dcuA and dcuB mutants utilize aspartate normally, which implies that DcuA and DcuB, but not DcuC, transport aspartate (154). One function of these systems is to transport dicarboxylic acids, which can be converted to fumarate, a terminal electron acceptor for anaerobic respiration. dcuA expression is constitutive (49). CRP, FNR, and DcuS-DcuR (which responds to dicarboxylic acids and aspartate) activate dcuB expression, whereas NarL complexed with nitrate represses dcuB (49, 193).

Asparagine Transport

Asparagine transport has not been extensively studied. Willis and Woolfolk defined two kinetic components of asparagine uptake in E. coli with apparent K_m values of 3.5 and 80 μM (188). Asparagine represses the high-affinity component (188). The gene or genes for this system have not been identified. Jennings et al. cloned the S. enterica gene for an asparagine permease, ansP, which appears to correspond to the low-affinity system in E. coli (72). Growth on asparagine as a nitrogen source appears to depend on this low-affinity system, since the high-affinity system should be repressed. An antibiotic represses ansP expression (128). The mechanism and function of this regulation are unclear, although it does suggest that there is regulation. Asparagine can be a source of fumarate during anaerobic respiration. Therefore, it would not be surprising if FNR or ArcA/ArcB regulated ansP expression.

l- and d-Alanine Transport

Two major systems of l- or d-alanine transport have been characterized. CycA (also called the glycine-alanine system) transports d-alanine, d-serine, glycine, and d-cycloserine (142, 174). CycA is also a minor system for l-alanine transport (26, 142, 173). A survey of amino acid transport in membrane vesicles suggests that the proton motive force energizes the glycine-alanine system (96). cycA mutants are resistant to d-cycloserine (173), unable to utilize d-alanine as a carbon source (142, 173), and resistant to d-serine in a strain lacking d-serine deaminase (26). Although cycA regulation has not been extensively studied, it appears that l- and d-alanine optimally induce; aspartate, succinate, and serine as carbon sources moderately induce; and glucose represses (174). These results suggest control by catabolite repression, a catabolic function, and alanine-specific control. In addition, nitrogen limitation moderately activates cycA (195).

The second l-alanine transport system is inhibited by serine, threonine, and leucine, which led to its identification as the LIV-I system for branched-chain amino acids (133, 142). This is the major system for l-alanine transport, since loss of cycA has only a minor effect on l-alanine transport (26, 173). LIV-I is sensitive to osmotic shock, which implies a binding protein (142). The purified binding protein binds alanine (133). The LIV-I system includes a periplasmic binding protein, LivJ; two membrane-associated proteins, LivH and LivM; and cytoplasmic proteins that bind the membrane and hydrolyze ATP, LivG and LivF (166). Lrp with leucine represses the livJ and livHMGF operons of the LIV-I system (56). However, Lrp without leucine activates livHMGF but has no effect on livJ (9). This explains why loss of Lrp diminishes livHMGF expression but increases livJ expression (69). l-Alanine, but not d-alanine, represses the major l-alanine transport system (174). Since Lrp binds alanine, it is possible that Lrp with alanine accounts for the repression. In addition to this regulation, charged leucyl-tRNA appears to cause premature rho-dependent termination of livJ transcription (166, 186). Growth in broth represses this system, which is consistent with both regulatory mechanisms (160).

ACKNOWLEDGMENTS

Grant MCB0323931 from the National Science Foundation supported my research during the preparation of this review.