Regulation of Serine, Glycine, and One-Carbon Biosynthesis

George V. Stauffer

[SECTION EDITOR: GEORGES COHEN]

Posted May 19, 2004

Department of Microbiology, The University of Iowa, Iowa City, IA 52242

Phone: 319-335-7791, Fax: 319-335-9006, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

The biosynthesis of serine, glycine, and one-carbon (C₁) units constitutes a major metabolic pathway in Escherichia coli and Salmonella enterica serovar Typhimurium. During growth on glucose, as much as 15% of the carbon assimilated in E. coli involves serine and its metabolites (158, 170). Serine, in addition to its role in protein synthesis, is used in the synthesis of cysteine, tryptophan, phospholipids, and glycine (14, 94, 123, 158). Glycine, in addition to its role in protein synthesis, is used in the synthesis of purines and heme-containing compounds (14, 94, 158). C₁ units derived from serine and glycine are used in the synthesis of purines, histidine, thymine, pantothenate, and methionine and in the formylation of the aminoacylated initiator fMet-tRNA^fMet used to start translation in E. coli and serovar Typhimurium (14, 93, 123, 135). Methionine is converted to S-adenosylmethionine (SAM), a major C₁ donor in numerous other cellular methylation reactions. For example, DNA methylation regulates gene expression, enhances recombination, and plays a role in the initiation of DNA replication (122); protein methylation plays a role in chemotaxis (213); and stable RNA modification plays a role in structure and function (13). In addition, there is a link between C₁ limitation and an altered heat shock response (56). Thus, the need for serine, glycine, and C₁ units in many cellular functions makes it necessary for the genes encoding enzymes for their synthesis to be carefully regulated to meet the changing demands of the cell for these intermediates.

PATHWAY

The serine, glycine, and C₁ pathway in E. coli and serovar Typhimurium is shown in Fig. 1 (156, 222, 223). During growth on glucose, the glycolytic intermediate 3-phosphoglycerate is converted to serine in three steps. The serA gene product, 3-phosphoglycerate dehydrogenase (3-PGDH), oxidizes 3-phosphoglycerate to 3-phosphohydroxypyruvate, with the concomitant reduction of NAD⁺ to NADH (156, 214, 215, 219). This is the first committed step in the phosphorylated pathway of serine synthesis. The serC gene product, 3-phosphoserine aminotransferase, converts 3-phosphohydroxypyruvate to 3-phosphoserine by adding an amino group to 3-phosphohydroxypyruvate (156, 222, 223). This enzyme is also required for the third step in pyridoxal 5'-phosphate synthesis (33, 34, 104-106, 118, 187). Since 3-phosphoserine aminotransferase requires pyridoxal 5'-phosphate as a coenzyme, sufficient pyridoxal 5'-phosphate must be made to sustain 3-phosphoserine transaminase activity for both pyridoxal 5'-phosphate and serine synthesis (34, 37). Extensive studies of 3-PGDH showed that the enzyme actually favors the reverse-reductase reaction of 3-phosphohydroxypyruvate to 3-phosphoglycerate (214, 215, 250). However, the 3-phosphoserine aminotransferase pulls biosynthetic intermediates in the forward direction and contributes to the control of the flow of intermediates through the serine pathway (214, 250). 3-Phosphoserine phosphatase, the serB gene product, dephosphorylates 3-phosphoserine to l-serine (156, 222, 223). The glyA gene product, serine hydroxymethyltransferase (SHMT), converts serine to glycine (189, 197). During this reaction, a methyl group is transferred to the cofactor tetrahydrofolate (THF), forming 5,10-methylene THF (5,10-mTHF). This reaction is the major source of C₁ units in cell metabolism (14, 135). SHMT is a zinc-binding protein and requires pyridoxal 5'-phosphate as a coenzyme (98). Thus, sufficient pyridoxal 5'-phosphate must also be sustained for glycine and C₁ synthesis. The final reaction is the oxidative cleavage of glycine to CO₂, NH₃, and 5,10-mTHF by the glycine cleavage (GCV) enzyme system (128, 142, 148, 159, 201, 204). The GCV systems of several microbial, plant, and animal systems have been characterized and shown to be multienzyme complexes consisting of four protein components, the P protein, H protein, T protein, and L protein (51, 100, 102, 133). P protein catalyzes the pyridoxal phosphate-dependent liberation of CO₂ from glycine, leaving a methylamine moiety (53, 83). The methylamine moiety is transferred to the lipoic acid group of the H protein, which is bound to the P protein prior to decarboxylation of glycine (54, 83, 147, 233). The T protein catalyzes the release of NH₃ from the methylamine group and transfers the remaining C₁ unit to THF, forming 5,10-mTHF (55, 145, 146). The L protein then oxidizes the lipoic acid component of the H protein and transfers the electrons to NAD⁺, forming NADH (134). Although the GCV enzyme complex has not been as well characterized in E. coli or in serovar Typhimurium, genetic and biochemical analyses indicate that the GCV enzyme systems in these organisms also consist of four protein components (145-148, 159, 202, 211). The T, H, and P proteins are coded by the gcvTHP operon (15, 159). The L protein is encoded by the lpdA gene and is not part of the gcvTHP operon (15, 211). The lpdA-encoded L protein is also common to the pyruvate and 2-oxoglutarate dehydrogenase multienzyme complexes (71, 212).

Fig. 1Serine, glycine, and C1 pathway in E. coli and serovar Typhimurium.

Mutations that result in an enzymatic deficiency in the serine pathway (serA, serB, or serC) result in a requirement satisfied by either serine or glycine (156, 223). Mutations that result in an enzymatic deficiency in glycine synthesis (glyA), however, result in a requirement for glycine that is not satisfied by serine (155). These results demonstrate that serine is the normal precursor of glycine. Mutations that result in an enzymatic deficiency in the GCV enzyme system (gcvT, gcvH, and gcvP) do not result in auxotrophy (159). However, mutations that result in an enzymatic deficiency in both the serine pathway and the GCV enzyme system can no longer use glycine as a serine source (142, 159). Under these conditions, the GCV enzyme system is unable to provide the necessary C₁ units from glycine to condense with a second molecule of glycine to form serine via the reverse SHMT reaction. E. coli mutants with no measurable GCV enzyme activity excrete glycine (62, 159). Thus, when E. coli is grown with glucose as a carbon source, the SHMT reaction appears to produce an excess of glycine relative to the C₁ units required for various biosynthetic and methylation reactions. Mutants that overproduce the GCV enzyme complex are partial glycine auxotrophs due to rapid glycine catabolism (61, 78). These results are consistent with the role of the GCV enzyme system being to balance the cell’s requirements for glycine used in protein and purine biosynthesis and C₁ units used in various biosynthetic and methylation reactions.

THREONINE UTILIZATION AS A SECONDARY PATHWAY FOR SERINE AND GLYCINE

Threonine utilization (the Tut cycle) constitutes a secondary pathway for serine and glycine biosynthesis. Threonine is converted to glycine and acetyl coenzyme A by a two-step pathway (Fig. 2) (49, 107, 167, 173, 174). The tdh gene product, threonine dehydrogenase, oxidizes threonine to α-amino-β-ketobutyrate. The kbl gene product, α-amino-β-ketobutyrate lyase, then cleaves α-amino-β-ketobutyrate to generate glycine and acetyl coenzyme A. The GCV enzyme system converts glycine to a C₁ unit, and SHMT catalyzes the condensation of a C₁ unit with a second molecule of glycine to produce serine. The genes encoding the enzymes for threonine utilization are members of the Lrp regulon (46, 221). Normally, in cells grown in glucose minimal (GM) medium, threonine dehydrogenase levels are low due to repression by Lrp (176, 221). The addition of leucine to the growth medium, or inactivation of Lrp, induces the tdh operon, explaining the increased efficiency of the Tut cycle in the presence of leucine (8, 139, 221). In addition, glycine serves as the precursor of serine when fructose or acetate is used as the carbon source (180), likely from threonine via the Tut cycle (173).

Fig. 2Threonine utilization (Tut) pathway for serine and glycine biosynthesis.

E. coli grown with lactate as a carbon source and with elevated levels of leucine, isoleucine, threonine, and methionine has reduced 3-PGDH activity (127). This medium is similar to media shown to increase the efficiency of the Tut cycle (173). It is possible that the two pathways for serine biosynthesis operate in a mutually exclusive fashion. It would be advantageous to the cell to derive serine from threonine under conditions where 3-phosphoglycerate levels are low, as this would spare the intermediate for gluconeogenesis (112, 173).

SERINE TOXICITY

l-Serine inhibits the growth of E. coli cells in GM medium, and isoleucine releases this growth inhibition (25, 144, 230, 231). Mutations that result in a defective l-serine transport system also reverse growth inhibition in the presence of l-serine (144). Growth inhibition by l-serine is increased in the presence of glycine, methionine, and leucine or by elevated temperature (25, 31, 76, 230), due to an elevated rate of serine catabolism to pyruvate induced by these conditions (89, 124, 125, 221). High pyruvate levels lead to isoleucine restriction, since the acetohydroxy acid synthase isozyme I, in the presence of high levels of pyruvate, synthesizes almost entirely 2-acetolactate (on the pathway to valine-leucine) and only 1 to 2% 2-aceto-2-hydroxybutyrate (on the pathway to isoleucine) (9). In addition, leucine represses ilvIH (32), which encodes the acetohydroxy acid synthase isozyme III. This enzyme is more effective in isoleucine synthesis than the isoenzyme I normally present in E. coli K-12 strains. This at least partially explains the l-serine sensitivity. l-Serine has also been reported to inhibit homoserine dehydrogenase I (73, 76). Since homoserine dehydrogenase I is involved in threonine-isoleucine synthesis, its inhibition would also contribute to l-serine sensitivity.

FEEDBACK INHIBITION

Like many enzymes that catalyze the first committed step in a biosynthetic pathway, 3-PGDH is feedback inhibited by l-serine through an allosteric mechanism (38, 39, 65, 67, 156, 188, 214, 215, 250). Because serine binding affects the velocity of the reaction and not the binding of the substrate or cofactor, the enzyme is classified as belonging to the V _max type (67, 214, 215). 3-PGDH is also feedback inhibited by glycine and l-alanine, but much higher concentrations are required for inhibition (214, 250). 3-PGDH is a tetramer of identical subunits, each with a molecular weight of 44,000 and each with three structural domains referred to as the cofactor binding domain, the substrate binding domain, and the regulatory domain (188). When two of the serine binding sites are occupied, ~85% inhibition of activity is attained; when all four of the serine binding sites are occupied, 100% inhibition of activity is attained (67, 188). The concentration of serine required for 50% inhibition is ~40 μM (214, 250).

The crystal structure of 3-PGDH was solved in the presence of bound serine (188). Each subunit contacts two other subunits through adjacent cofactor binding domains and through adjacent regulatory domains. l-Serine binds to 3-PGDH at the interface between two regulatory domains of adjacent subunits, forming hydrogen bonds with adjacent domains that result in a hydrogen bond network across the noncovalent interface (67, 188). This network appears to tether the domains together and results in inhibition of the catalytic activity (2, 3, 11, 66, 67, 188). Site-directed substitution mutagenesis of specific amino acids at the regulatory domain interface directly affect serine’s ability to inhibit 3-PGDH, and mutations that specifically alter amino acids involved in cross-linking the adjacent regulatory domains through hydrogen bonding with the bound effector cause a complete loss of inhibitory capacity for ligands such as serine, glycine, and alanine (3).

Mutants producing 3-PGDHs that are resistant to serine inhibition overproduce serine (220). Furthermore, the intracellular concentrations of 3-phosphoglycerate and serine remain relatively constant even with a 10-fold difference in 3-PGDH activity (127). Thus, end product inhibition of 3-PGDH activity is an effective form of control of the metabolic flow of carbon through the serine-glycine pathway. In addition, high levels of α-ketoglutarate inhibit anaerobic growth of E. coli on GM medium, and this inhibition is reversed by serine (250). 3-PGDH has been shown to reduce α-ketoglutarate to 2-hydroxyglutarate (250). If serine synthesis competes with the 3-PGDH α-ketoglutarate reduction activity, the α-ketoglutarate concentration may play a role in regulating serine biosynthesis by modulating 3-PGDH activity (250).

There is no evidence for significant inhibition of 3-phosphoserine aminotransferase activity, and the inhibition of 3-phosphoserine phosphatase by l-serine is 1,000 times less sensitive than is the inhibition of 3-PGDH (156, 223).

Feedback inhibition of SHMT activity could not be demonstrated in either E. coli or serovarTyphimurium by compounds involved in C₁ metabolism (d-serine, α-methylserine, l-methionine, l-threonine, SAM, purines, and thymine) (120, 190, 195). The structure of the E. coli SHMT in ternary complex with glycine and 5-formyl tetrahydropteroylglutamate has been determined (184). This structure provides the basis for a thorough investigation of the enzyme mechanism through characterization of site-specific mutants.

REGULATION OF serA, serB, AND serC

The serA, serB, and serC genes are not regulated by repression or activation mechanisms in response to the levels of serine, seryl-tRNA, or seryl-tRNA synthetase (127, 156, 157, 220, 223). The serA gene, which is transcribed from two promoters (Fig. 3) (247), is activated by the leucine-responsive regulatory protein (Lrp), and this effect is partially reversed by leucine (20, 88, 112, 176, 221). Promoter P1 is activated by Lrp and is the dominant promoter in GM medium. Lrp represses promoter P2, consistent with the Lrp binding region covering the promoter P2 −35 and −10 regions (Fig. 3). Expression of serA is known to be reduced when cells are grown under one of three conditions: with a poor carbon source supplemented with threonine, methionine, leucine, and isoleucine; in GM medium supplemented with various amino acids; and in Luria-Bertani broth (LB) (21, 127). The low levels of serA expression are likely due to reduced levels and activity of Lrp under these growth conditions (19, 140). An lrp mutant is not auxotrophic for serine, however, but grows more slowly than the lrp⁺ parent strain, and normal growth occurs by serine supplementation (5). A basal level of transcription from promoter P2 is likely sufficient to allow growth in an lrp mutant without serine supplementation (20).

Fig. 3serA control region. The −35 and −10 regions for promoters P1 and P2, the Lrp binding region, and a putative CRP binding site are indicated ( ). The sites are approximately to scale.

Promoter P2 is activated by the cyclic AMP receptor protein-cyclic AMP complex (CRP cAMP) (247). Although a putative CRP binding site occurs in the promoter P2 region (Fig. 3), a direct role for CRP-cAMP has not been demonstrated. An earlier report suggested that 3-PGDH levels are negatively regulated either directly or indirectly by CRP-cAMP (127). The discrepancy possibly lies in the experimental systems. The earlier study (127) measured serA expression from a single-copy chromosomal insert, whereas the later study (247) measured serA-lacZ expression from multicopy plasmids.

The serC gene is upstream from aroA in a two-gene operon (41, 86, 118), probably reflecting the need to coordinately regulate serine, pyridoxal 5'-phosphate, and chorismate biosynthesis in certain bacteria (41, 106, 154). The aroC transcript level, however, is five- to eightfold lower than that of the serC transcript, likely due to the action of an attenuator between serC and aroA (118). The serC gene is activated by Lrp at the transcriptional level, and this activation is dependent on the presence of an upstream Lrp recognition sequence, suggesting that Lrp has a direct role in serC transcription (118). The serC gene is repressed by CRP-cAMP (118). Deletion of a putative upstream CRP-cAMP binding site in the serC control region, however, had no effect on CRP-cAMP repression, suggesting that CRP-cAMP acts by an indirect mechanism, possibly controlling the amount or activity of an alternative regulator (118). An earlier report suggested that serC is activated slightly by CRP-cAMP (111). The discrepancy possibly lies in the experimental systems, as the earlier study (111) measured expression from an aroA-lacZ fusion from a multicopy plasmid rather than a single-copy chromosomal insertion (118). Since Lrp expression is high in GM medium (19, 108, 140) and CRP-cAMP levels are low (103, 151), the effects of these two global regulators on serC would be to maximize serC expression in cells growing in GM medium. Growth in LB, however, results in maximal repression of serC by CRP-cAMP and negligible activation by Lrp (118). Since the 3-phosphoserine aminotransferase pulls biosynthetic intermediates in the forward direction in the serine pathway (214, 250), this regulation of serC likely plays a significant role in the overall carbon flow through the serine and pyridoxal 5'-phosphate pathways.

The serB gene does not appear to be regulated by an induction-repression mechanism that involves 3-phosphoglycerate, serine, seryl-tRNA, or seryl-tRNA synthetase (127, 156, 157, 220, 223). The serB gene is divergently transcribed from smp, encoding a membrane protein of unknown function (126, 137). Selection for mutations that increased smp promoter activity resulted in a serB promoter −35 region down mutation, suggesting that smp and serB are inversely related (137). The serB gene is cotranscribed with sms, whose inactivation caused a slight sensitivity to methyl methanesulfonate (138). The sms gene was later identified as the rad gene (193). The sms-radA gene product plays a role in recombination and DNA repair (10). However, it is unknown why there is transcriptional linkage between serB and sms-radA. It is possible that Sms/RadA functions in the repair of DNA damage induced by end products of serine metabolism (137). Despite a lack of information as to the mechanism of serB regulation, the transcriptional linkage between serB and sms-rad allows coordinate regulation of these different cellular functions.

REGULATION OF THE glyA GENE

The glyA gene product, SHMT, catalyzes the interconversion of serine, glycine, and C₁ units and is the cell’s major source of glycine and C₁ units for cell metabolism (123, 135). Because the SHMT reaction plays a central role in cell physiology, the glyA gene is highly regulated so as to be able to respond to the needs of the cell for these intermediates for either protein synthesis or various biosynthetic and methylation reactions. Glycine represses glyA expression, but not through the functional state or glycyl-tRNA synthetase of glycyl-tRNA (48, 132). When cells are grown under conditions where glycine limitation is produced, C₁ compounds still repress glyA (197, 199). In addition, growth of a serine-glycine auxotroph in a serine-limited chemostat results in derepression of glyA, but the addition of purines to the culture reduces glyA expression (35). These results demonstrate that glycine is not solely responsible for glyA regulation.

Purine supplementation of GM medium and purine limitation result in repression and derepression of glyA, respectively, in both E. coli and serovar Typhimurium (35, 132, 196, 200). PurR, a repressor protein for purine nucleotide synthesis (101, 182), plays a direct role in glyA regulation in E. coli (209). PurR binds to a 24-bp sequence from positions −72 to −49 upstream of the glyA transcription start site (Fig. 4) (208). The PurR binding site in E. coli and its location relative to the transcription start site are conserved in serovar Typhimurium (210), and the mechanism of glyA regulation by purines in this organism is most likely the same as in E. coli. Hypoxanthine and guanine, corepressors for PurR regulation of the pur regulon, increased binding of PurR to the glyA operator sequence (117, 208). Furthermore, when PurR was bound to the glyA operator, it prevented RNA polymerase (RNAP) binding to the promoter (117). Thus, the mechanism of PurR repression of glyA appears to be steric hindrance. The PurR binding site for glyA, with the sequence 5'-AGGTAATCGTTTGCGT-3', has a 13-of-16 nucleotide match with the PurR consensus binding sequence 5'-ACGCAAACGTTTGCGT-3' (131, 181). Mutations that changed the PurR binding sequence away from the consensus sequence increased glyA expression twofold (208), whereas mutations that changed the binding sequence toward consensus had no significant effect on either PurR binding or purine-mediated repression. Thus, it is likely that the position of the PurR binding site relative to the glyA promoter is the major determinant of PurR-mediated repression for the glyA gene. The narrow range of PurR regulation is not unexpected. Since the products of the SHMT reaction are used in many other metabolic pathways, this would allow sufficient levels of C₁ units and glycine produced by the SHMT reaction, even in the presence of repressing levels of purines.

Fig. 4glyA and hmp control region. The promoter −35, −10, and +1 elements for glyA and hmp ( ) and the binding sites for MetR ( , , ) and PurR ( ) are indicated. The RNAP binding site for glyA extends from bp −68 to −75 (lightly shaded bar) in the presence of MetR ( ). MetR binding site 3 was observed only in the presence of RNAP and has not been shown genetically to be involved in glyA regulation ( ). The sites are approximately to scale.

MetR, a lysR family protein (185), is a positive regulator for the metA, metE, metF, and metH genes (26, 121, 227, 228). Homocysteine, a methionine pathway intermediate, serves as a coregulator for MetR regulation of these genes and has a negative effect on metA and metH expression (121, 227) and a positive effect on metE expression (227, 228). MetR also activates glyA, with homocysteine serving as the coactivator (163). MetR binds to two sites in the E. coli glyA promoter region from bp −109 to −158 upstream of the transcription start site (114, 161, 164, 210) (Fig. 4). The MetR binding sites in E. coli and their locations relative to the promoter region are conserved in the serovar Typhimurium glyA promoter (210, 224), and the mechanism of MetR regulation of glyA in this organism is most likely the same as for E. coli. MetR binding site 1, with the sequence 5'-TGAANNANNTGCA-3', has 8 of 9 nucleotides that match the MetR consensus binding sequence 5'-TGAANNA/TNNTTCA-3', and site 2, with the sequence 5'-TGAANNGNNATCC-3', has 6 of 9 nucleotides that match the consensus sequence (N is A, G, C, or T). Site 1, with a closer match to the MetR consensus sequence, has a higher affinity for the MetR protein than binding site 2 (114, 115). The coactivator homocysteine causes a minor increase in the binding of MetR to site 1 but results in a significant increase in MetR binding to the low-affinity site 2 (114). Point mutations in either MetR binding site toward the MetR consensus binding sequence increased MetR activation of a glyA-lacZ fusion (114). Point mutations in either MetR binding site away from the MetR consensus binding sequence, however, reduced glyA-lacZ expression below the level observed in a metR mutant (114). Thus, MetR bound to a single site has a negative effect on glyA expression. Furthermore, when site 2 was inactivated, the ability of MetR bound to site 1 to repress glyA-lacZ expression was shown to be dependent on the face of the helix relative to the RNAP binding site (116). Although the mechanism by which MetR represses glyA when bound only to site 1 is unknown, the repression likely plays an important role in glyA regulation. MetR levels are low under conditions where there are adequate levels of the methionine and C₁ units required for cell metabolism (226). Although MetR binds cooperatively to sites 1 and 2, at a low concentration of MetR, only site 1 is occupied in vitro (114, 115). If this reflects the in vivo situation, the role of MetR might be to bind to site 1 and function as a repressor of glyA under the condition where presumably low levels of SHMT are required to produce glycine, primarily for protein synthesis. MetR levels are high under conditions where there are inadequate levels of the methionine and C₁ units required for cell metabolism (226). MetR then binds to both sites 1 and 2 and functions as an activator for glyA expression, since higher levels of SHMT are necessary to produce glycine and C₁ units for other cellular biosynthetic and methylation reactions.

The met genes that are regulated by MetR have binding sites that are adjacent to or that overlap the promoter sequence (18, 26, 121, 224, 225). Evidence suggests that there is an interaction of the carboxy-terminal domain (CTD) of the α subunit of RNAP and the MetR protein for MetR activation of the E. coli metE and metH genes (52, 91, 92). Because the MetR binding sites for glyA are considerably further upstream than the MetR binding sites for the met genes, this would not allow a direct interaction between MetR and RNAP unless DNA bending occurs. MetR binding to site 1 or 2 alone, or to both sites, results in bend angles of ~30 and 33°, respectively, for the glyA control region, and the bending is independent of homocysteine (114). The degree of bending appears to be insufficient to allow a MetR-RNAP interaction at glyA. At high concentrations of MetR, however, a third low-affinity MetR binding site was detected by gel mobility shift assay (114). When MetR and homocysteine were prebound to the glyA promoter at a low concentration, followed by the addition of RNAP, no significant effect on the footprint pattern of the glyA control region was observed when MetR and RNAP were tested separately (117). However, when MetR and homocysteine were prebound to the glyA promoter region at a high concentration, followed by the addition of RNAP, there were significant changes in both the MetR and RNAP footprint patterns. The upstream RNAP protected region was extended from bp −68 to about bp −75, and the downstream MetR protected region was extended from bp −109 to about bp −77 relative to the glyA transcription start site (Fig. 4) (117). Thus, it is possible that a third MetR binding site exists between site 2 and the RNAP binding site and that, when occupied by MetR, could contact RNAP. However, additional quantitative DNase I footprint assays and genetic experiments are required to verify whether a third MetR binding site is involved as part of the MetR regulatory mechanism. It is also possible that MetR binding at sites 1 and 2 blocks the binding of a negative-acting regulator to the glyA promoter region. PurR is known to negatively regulate glyA. Furthermore, in a purR mutant, MetR is not required for high levels of glyA-lacZ expression, and in a metR mutant, PurR has its greatest effect on repressing glyA-lacZ expression (208). However, footprint analysis showed that MetR and PurR bind independently to the glyA control region (117). Thus, although MetR activation and PurR repression might be parts of a single mechanism for glyA control, the mechanism does not appear to involve competition of the two proteins for their respective target sites. However, it is still possible that at high levels MetR binds to site 3 and interferes with PurR binding to the glyA control region (Fig. 4).

It was reported previously that metK and metJ mutations prevent complete reduction of SHMT levels in response to C₁ compounds, but in GM medium, only metK mutations and not metJ mutations result in elevated SHMT levels (70, 128, 198). The metJ gene encodes the repressor protein for the methionine pathway, and metK encodes SAM synthetase, which converts methionine to SAM, the corepressor for MetJ (69). Since MetJ and SAM repress metR, metJ mutants have elevated levels of MetR (226). Although the effects of metK mutations on metR expression were not tested, it is likely that such mutations also result in elevated levels of MetR. However, since metK mutants are blocked in the conversion of methionine to SAM, methionine intermediates likely accumulate. Thus, elevated levels of both homocysteine and MetR result in activation of glyA. In metJ mutants, however, all of the met genes are derepressed, and levels of pathway intermediates are likely low due to rapid conversion to methionine and SAM. Thus, the homocysteine levels are likely too low to allow for activation of glyA by MetR, but full repression of glyA by C₁ compounds is prevented. In addition, the glyA gene is derepressed in metE and metF mutants, but not in metA and metB mutants, during methionine limitation (35, 36, 70, 120). Homocysteine is produced only at low levels in metA and metB mutants by a circular pathway from SAM (40). These levels are kept low by subsequent conversion of homocysteine to methionine, since the mutants have adequate levels of 5-methyl THF and homocysteine transmethylase activities (69). On the other hand, homocysteine accumulates in metE and metF mutants, since the nonfolate branch of the methionine pathway is operational (69). Thus, higher levels of SHMT are expected in the metF and metE mutants than in the metA and metB mutants due to higher levels of the coactivator homocysteine.

Trimethoprim supplementation also results in derepression of glyA in both E. coli and serovar Typhimurium (35, 196, 200). However, the derepressed levels of SHMT observed in a wild-type strain by trimethoprim supplementation is not seen in a metR mutant (163). Trimethoprim reduces the levels of THF and its derivatives (12, 68) and blocks synthesis of 5-methyl THF from the folate branch of the methionine pathway, but not that of homocysteine from the nonfolate branch of the methionine pathway. Thus, part of the increase in SHMT synthesis when trimethoprim is added to the growth medium is likely due to homocysteine accumulation, resulting in MetR activation of glyA. In addition, trimethoprim addition results in a limitation of the supply of purines (35), and part of the observed increase in SHMT synthesis by trimethoprim supplementation is likely due to a loss of PurR repression of glyA.

The starvation-induced csgD gene encodes a positive transcriptional regulator of the extracellular matrix components curli fimbriae and cellulose (59). In multiple copies, csgD also suppresses glycine auxotrophy of a strain lacking dihydrofolate reductase activity by up-regulating glyA and increasing glycine synthesis (22). It was proposed that up-regulation of glyA is an integral response to signals for curli formation, increasing the cell’s ability to make glycine, which makes up ~20% of the major protein of curli. However, the mechanism of up-regulation of glyA in the multicopy csgD strain is unknown. CsgD could function directly as an activator for glyA or indirectly by activating metRor repressing purR. Also unknown is the effect of a single-copy chromosomal csgD gene on glyA expression.

In E. coli and serovar Typhimurium, the glyA gene is divergently transcribed from hmp (Fig. 4) (27, 161, 210, 234). The hmp gene product is a flavohemoglobin that plays a role in protection from nitrosating agents, nitric oxide-related species, and oxidative stress (27, 129, 166). The hmp gene is up-regulated by S-nitrosoglutathione (GSNO) and sodium nitroprusside and decreased by elevated homocysteine levels (130). Up-regulation of hmp by GSNO and sodium nitroprusside is abolished in a metR mutant (130). Because homocysteine is nitrosated by GSNO (130), such nitrosating agents likely deplete homocysteine pool sizes. Modulation of homocysteine levels would modulate binding of MetR to the glyA-hmp intergenic region. Interestingly, the high-affinity MetR site 1 alone is sufficient for MetR induction of hmp (130), whereas MetR bound only to MetR site 1 represses glyA (Fig. 4) (116). When homocysteine levels are high, MetR is bound to both sites 1 and 2, activating glyA (114, 163) and repressing hmp (130). In addition, purine supplementation represses glyA (209) but results in induction of hmp (130). Thus, glyA and hmp appear to be reciprocally regulated, but the significance of this regulatory pattern awaits further study.

In E. coli, the glyA gene produces a monocistronic mRNA with a 182-nucleotide-long 3' nontranslated region that contains two repetitive extragenic palindromic (REP) sequences and a Rho-independent transcription terminator (164). REP sequences are highly conserved inverted repeats present in up to 1,000 copies on the E. coli chromosome and have been shown to stabilize upstream mRNA by blocking the processive action of 3'-to-5' exonucleases (80). In a number of operons, mRNA stabilization by REP sequences plays an important role in the control of gene expression (80). In the glyA mRNA, either of the two REP sequences or the transcription terminator is essential to maintain normal glyA mRNA stability by blocking the 3'-to-5' exonucleolytic activities of polynucleotide phosphorylase and RNase II (160, 162). However, in serovar Typhimurium, there is an absence of REP sequences between the glyA translation termination site and the proposed transcription termination site (210). Thus, despite the role of REP sequences in glyA mRNA stability in E. coli, it is likely that this is not the primary or the sole reason for the presence of the REP sequences.

Mutants in both E. coli and serovar Typhimurium with increased levels of SHMT have been isolated (113, 199). Although these mutants were not characterized completely, evidence suggests that the changes are not in genes encoding known regulators of glyA (metR and purR) (116, 163, 208, 209) or a possible regulator of glyA (csgD) (22). Thus, regulation of the glyA gene is likely to be more complex than presented, with additional factors either directly or indirectly involved in controlling its expression.

REGULATION OF THE gcvTHP OPERON

The GCV enzyme complex is important for balancing the cell’s requirements for glycine used in protein and purine biosynthesis and C₁ units used in various biosynthetic and methylation reactions (61, 78, 159). Thus, it is not surprising that regulation of the gcvTHP (gcv) operon is complex and both gcv-specific and globally acting regulatory proteins are involved, enabling the cell to respond to changing environmental stimuli. To date, five proteins (PurR, CRP, GcvA, GcvR, and Lrp) have been shown to be involved in gcv regulation. The first indication that synthesis of the GCV enzyme complex is regulated was the observation that an E. coli serine auxotroph grown in medium containing serine and a purine base resulted in a temporary inability of the cells to grow when transferred to medium containing only glycine (141). Thus, formation of the GCV enzymes appeared to be repressed under these conditions, resulting in an inability to generate C₁ units from glycine via the GCV enzyme complex. The addition of glycine to the repressing growth medium overcame the repressive effect of the purine base, allowing growth immediately upon transfer to medium containing only glycine and suggesting that glycine induces synthesis of the GCV enzyme complex (141). Later, it was shown that induction of the GCV enzyme system by glycine does occur in the presence of a purine base (128).

GcvA, a member of the LysR-type transcriptional regulator (LTTR) family (185, 240), plays a dual role in regulation of the gcv operon. GcvA is required for a six- to sevenfold activation of gcv in the presence of exogenous glycine (205, 241, 242). When glycine is limiting, GcvA represses gcv, and this repression is more severe in the presence of purines, resulting in a fivefold repression below the noninduced levels observed in GM medium (241, 242, 245). Repression by GcvA requires an accessory protein, GcvR (described below) (61). GcvA binds to three sites in the gcv control region from bp −271 to −242 (site 3), from bp −242 to −214 (site 2), and from bp −69 to −34 (site 1) relative to the transcription start site (Fig. 5) (243). All three of these sites are required for GcvA repression of the operon in the absence of glycine, whereas only sites 3 and 2 are required for GcvA activation of the operon in the presence of glycine (243, 245). GcvA binds by way of a helix-turn-helix motif (97). Amino acid substitutions V32A in the turn of the helix-turn-helix motif and S38P in the recognition helix result in a loss of DNA binding and prevent both GcvA activation and repression (97).

Fig. 5gcv control region. The promoter −35, −10, and +1 elements ( ) and the binding sites for GcvA ( ), Lrp ( ), PurR ( ), and CRP ( ) are indicated. The sites are approximately to scale.

Most LTTRs bind to their promoters in the absence of their respective small coregulatory molecules (185). Coregulatory molecule binding by LTTRs usually results in altered DNA binding and DNA bending, which facilitates LTTR activation or repression of the respective operons (58, 87, 143, 177, 232, 235, 239). DNase I footprint studies with GcvA showed hypersensitive cleavage sites in the region between the upstream sites 3 and 2 and the downstream site 1 when GcvA was bound to these sites (243), and gel mobility shift assays showed that GcvA bound to sites 3 and 2 induces a bend angle of ~100° (78). The coinducer glycine, however, did not alter either GcvA binding affinity to the three sites or the bend angle induced by GcvA binding to the DNA (78).

Many activator proteins, including LTTR proteins, recruit RNAP to promoters via protein-protein interactions with the subunits of RNAP (1, 43, 90, 183, 217, 218). Amino acid substitutions L30 and F31 in the stabilizing helix of the GcvA helix-turn-helix DNA binding motif are positive control mutants, decreasing the ability of GcvA to activate gcv transcription but having no affect on its ability to bind gcv DNA and repress gcv (97). In addition, deletions of the αCTD of RNAP and several amino acid substitutions in the αCTD of RNAP result in an inability of GcvA to activate, but not to repress, gcv (95). These results suggest that the αCTD and GcvA interact to activate gcv but that GcvA repression does not require interaction with the αCTD (95). Since activation of gcv requires GcvA bound to sites 3 and 2, located from bp −271 to −214 upstream of the transcription start site (Fig. 5), it is reasonable to assume that a looped nucleoprotein structure is required for a GcvA-RNAP interaction. The interaction of GcvA and RNAP from these sites is somewhat unusual for sigma 70-activated promoters (23). Activation generally occurs through a mechanism that involves at least one target site, located near the promoter, that allows the activator to touch RNAP in such a way that activation is achieved. A GcvA binding site does occur from bp −69 to −34 relative to the transcription start site (Fig. 5), but this site functions only in GcvA repression (243, 245). Although the site is within what is considered a normal activation location (23), presumably the activator domain of the GcvA protein is unexposed or unable to make an appropriate contact with RNAP from this site.

Lrp, a global regulator involved in the control of the transcription of numerous genes involved in amino acid metabolism (19, 140), is required for GcvA activation and repression of the gcv operon (205). Lrp binds to multiple sites in the gcv control region from bp −229 to −92 relative to the transcription start site (Fig. 5) (205). There is some flexibility in the absolute distance of the Lrp binding sites (and the GcvA binding sites) relative to the gcv promoter, but there is a strict orientation dependence of these sites, consistent with a possible protein-protein interaction of Lrp, GcvA, or both of these proteins with RNAP (207). Lrp binding to the gcv promoter region has been shown to induce a bend of ~89° (207). When the Lrp binding region was replaced with the phage λ I1A site for integration host factor (IHF) binding, which bends DNA ~140° (246), it resulted in IHF-dependent activation and repression of gcvT expression that was dependent on the face of the helix (206). Activation and repression were still dependent on GcvA and GcvR, suggesting that the role of Lrp in gcv regulation is structural, binding and bending the DNA to facilitate the formation of appropriate activation and repression complexes (206, 207). However, because the level of IHF-dependent activation was not as high as that of the Lrp-dependent activation observed in a wild-type strain, it is still possible that Lrp also activates transcription through interactions with RNAP or a second regulatory protein.

GcvR is necessary for negative regulation of the gcv operon (61). In the absence of GcvR, there is constitutive expression of a gcvT-lacZ fusion and of the gcv operon, even under conditions where the cell is starved for the small coeffector molecule glycine (61, 78). However, GcvR does not appear to regulate gcv expression directly like a typical repressor protein, as it fails to bind to the gcv control region in gel mobility shift assays. In addition, GcvR does not regulate gcv expression indirectly by regulating the expression of gcvA (63). Rather, repression of the gcv operon depends on the relative levels of GcvA and GcvR. Overproduction of GcvR results in superrepression of gcv, and overproduction of GcvA results in constitutive expression of gcv (61). Recently, it was shown both in vivo and in vitro that GcvA interacts directly with GcvR (60, 78). Mutations C169R and R197G in the C-terminal half of GcvA, or deletion of the last 14 amino acids from GcvA, result in a loss of GcvA-GcvR heteromultimerization in vivo and in constitutive expression of gcv (60). In addition, the coeffector molecule glycine was shown to bind to GcvR rather than to the activator protein GcvA (78). Glycine binding to GcvR either blocks GcvA/GcvR heteromultimerization or causes a dissociation of the complex once formed (78). Mutations in gcvR that produce proteins that do not bind glycine were isolated (78). These mutations result in a superrepression phenotype (78), presumably because the GcvA/GcvR complex, once formed, can no longer be inactivated by the coeffector glycine. Since GcvR has no repressor capabilities in the absence of GcvA (61), the results suggest that GcvR/GcvA interaction is required for repression. This is supported by the observation that overproduction of only the C-terminal half of GcvA in a gcvA⁺ gcvR⁺ strain interferes with GcvR repression, likely due to the C-terminal GcvA fragment interacting with and titrating out available GcvR protein, leading to constitutive activation of the operon by GcvA.

The above-mentioned results led to a model for repression and activation of the gcv operon by GcvA, GcvR, Lrp, and glycine. In the absence of glycine, GcvR binds directly to GcvA (78). Since all three GcvA binding sites are required for normal repression of gcv (243, 245), it is likely that the GcvA/GcvR complex, with the help of Lrp, results in the formation of a repression loop (Fig. 6A). This either prevents RNAP from binding to the promoter or prevents a GcvA/RNAP contact that is necessary for activation. In the presence of glycine, GcvA/GcvR interaction is blocked. Bending of the DNA by Lrp and GcvA allows GcvA bound to sites 3 and 2 to interact with the αCTD of RNAP, permitting activation of transcription (Fig. 6B). Whether it is GcvA bound at site 3 or site 2 that contacts RNAP is unknown. It should be noted that GcvA binding site 1 overlaps the RNAP binding region, but site 1 is required only for GcvA repression of the operon (243, 245). The model predicts that transcription of the gcv operon is inversely proportional to the level of GcvA/GcvR complex formed, which is inversely proportional to the level of glycine present in the cell. Thus, glycine does not act as a classical coactivator in this system but rather through a mechanism of derepression by binding to GcvR and preventing GcvR from interacting with GcvA to block the GcvA activator function. The GcvR protein binds the effector molecule glycine at a K_d of ~200 μM (79). This correlates well with the estimated intracellular concentration (~41 μM) of glycine in mid-exponential-phase cells grown in GM medium (172). This level of glycine is insufficient to allow full activation of the gcv operon, as seen in a gcvR mutant, but is sufficient to bind some GcvR and to allow activation above the fully repressed level (61, 78). Since genetic experiments suggest that there are relatively constant levels of GcvA and GcvR in the cell (63, 96, 240), this would allow the cell to respond rapidly to changes in intracellular glycine concentrations due to either transport of exogenous glycine or cellular metabolism.

Fig. 6Model for repression and activation of the gcv operon. See the text for details.

Threonine supplementation of GM medium also induces expression of gcv (62). However, this is most likely an indirect effect resulting from the conversion of threonine to glycine (173) rather than from threonine interacting with GcvA or GcvR. This hypothesis is supported by the observation that there was a further increase in induction of the gcv operon when both leucine and threonine were added to the growth medium. Leucine is known to increase the conversion of threonine to glycine via the Tut cycle (8, 139, 176, 221).

Full repression of the gcv operon requires the addition of purines to GM medium (61, 241), but the mechanism of the purine involvement is unknown. Since one molecule of glycine and two C₁ units from 10-formyl THF are incorporated into the purine ring structure (248) and purines repress glyA expression (35, 132, 196, 200), it is possible that part of the purine effect is to lower the intracellular glycine levels. This would increase GcvA/GcvR heterodimerization and repress gcv expression. The addition of both glycine and purines to the growth medium results in derepression of the gcv operon (205, 241). The dominance of glycine over purines is consistent with purines having an indirect effect in the regulatory mechanism. Additional studies are necessary, however, to determine if purines play a more direct role in the GcvA/GcvR repression mechanism.

PurR, a repressor for many genes involved in nucleotide metabolism (101, 182), mediates a twofold decrease in gcv transcription in the presence of exogenous purines (202, 241). This second mechanism of purine repression of gcv is independent of the GcvA and GcvR proteins (241). PurR binds to the gcv control region from bp −3 to +17 relative to the transcription start site (Fig. 5) (241) and likely interferes with RNAP binding to the gcv promoter. The PurR binding site in gcv, with the sequence 5'-AAGAGAACGATTGCGT-3', has only 12 of 16 nucleotides that match the PurR consensus sequence 5'-ACGCAAACGTTTGCGT-3', which probably results in the narrow twofold range of PurR regulation.

CRP is required for a fourfold activation of the gcv operon in GM medium (244). Since CRP does not significantly affect gcvA-lacZ or gcvR-lacZ expression (244), the effect is not indirect by CRP controlling GcvA and GcvR levels. CRP binds to a site from about bp −303 to −324 relative to the transcription start site (Fig. 5) (244). For most well-studied CRP-regulated systems, CRP binds to sites located three, five, or six helical turns upstream of the −10 promoter region (103) and either interacts with RNAP to facilitate transcription initiation or bends the DNA (17, 42, 103, 178), which possibly is involved in the activation of gene expression. Since the CRP binding site for gcvA is far upstream of the RNAP binding site (Fig. 5), it is unlikely that CRP directly contacts RNAP to activate transcription of gcv unless DNA looping occurs at the gcv promoter. This is possible since, as described above, DNA looping is likely part of the GcvA/GcvR activation-repression mechanism. However, a direct CRP/RNAP interaction seems unlikely, since mutations in known CRP-activating domains are able to complement a crp deletion and restore gcvT-lacZ expression. A second CRP binding site occurs in the gcv control region centered at bp −140 relative to the transcription start site (244). However, mutation analysis indicated that this site is not required for CRP activation of a gcvT-lacZ fusion. In addition, this site is totally within the region shown to be protected from DNase I digestion by Lrp (205) and is probably not accessible to CRP in vivo due to Lrp binding to this sequence. Although the specific mechanism of the CRP effect is unknown, the effect is dependent on the repressor function of GcvA (244). This suggests that the role of CRP is to function as an antirepressor to antagonize GcvA/GcvR-mediated repression.

REGULATION OF THE lpdA GENE

The lpdA gene, encoding the L protein of the GCV enzyme complex (211), is not part of the gcv operon. Regulation of lpdA is complex, likely due to the lpdA-encoded lipoamide dehydrogenase functioning not only as a component of the GCV enzyme complex but also as a component of the pyruvate dehydrogenase (PDH) and 2-oxoglutarate dehydrogenase (ODH) multienzyme complexes (72, 153). The PDH enzyme complex is induced by pyruvate and partially repressed by growth on glucose or acetate and by anaerobiosis, whereas the ODH enzyme complex is induced during aerobic growth on acetate and citric acid cycle intermediates and is repressed by glucose, anaerobiosis, and cAMP deficiency (4, 16, 169, 191). The PDH enzyme complex is encoded by the pdh operon (pdhR-aceEF-lpdA) and includes lpdA, encoding the lipoamide dehydrogenase for all three enzyme complexes (169). The lpdA gene is expressed from two promoters. Promoter P_pdh transcribes the entire pdh operon, and promoter P_lpd transcribes only the lpdA gene (169, 194). Transcription from P_lpd is coregulated with the synthesis of the ODH complex so that lipoamide dehydrogenase can be synthesized under conditions where synthesis of the other components of the PDH complex from the P_pdh promoter is repressed but synthesis of the ODH complex is induced (28, 194). This ensures adequate levels of the lipoamide dehydrogenase for the two enzyme complexes under different growth conditions. The GCV enzyme complex is known to be reduced by ~50% when cells are grown in GM medium supplemented with acetate and succinate (211). The GCV enzyme assay measures only the overall specific activity of the GCV enzyme complex, and whether the reduction in activity was due to a reduced amount of lipoamide dehydrogenase or all components of the GCV enzyme complex is unknown. Nevertheless, the results suggest that the GCV enzyme system is likely regulated, at least partially, in conjunction with expression of the PDH and ODH complexes. However, whether the P_phd or P_lpd promoter responds to signals specific for gcv operon expression or whether the gcv operon responds to signals specific for the P_phd and P_lpd promoters has not been examined.

REGULATION OF THE gcvA AND gcvR GENES

The model for gcv regulation predicts that GcvA and GcvR levels are maintained within a narrow range in order for the cell to be able to respond appropriately to an inducing or repressing signal. GcvA binds to a single site in the gcvA control region from bp −28 to +20 relative to the transcription initiation site (240) and negatively regulates its own synthesis over a threefold range (Fig. 7) (240). However, the gcvA gene is neither induced by glycine nor repressed by inosine (240), suggesting that the ability of GcvA to activate the gcv operon in the presence of glycine and to repress the operon in the presence of purines is not mediated through regulation of the gcvA gene by these compounds. The known GcvA binding sites have a conserved 5'-CTAAT-3' sequence that has been shown to be important for GcvA binding at the gcv promoter (243). Two mutations that changed the conserved 5'-CTAAT-3' sequence in the gcvA promoter region away from consensus resulted in 9- and 24-fold overexpression of a gcvA-lacZ fusion compared to wild-type expression (96). This increase in gcvA-lacZ expression was due in part to loss of autoregulation by GcvA and in part to a GcvA-independent mechanism. Interestingly, the mutations did not significantly alter the ability of GcvA to bind to the target site in the gcvA promoter region or the DNase I footprint pattern when GcvA was bound (96). Thus, GcvA either does not bind the gcvA promoter region in vivo despite its ability to bind in vitro or when bound in vivo is unable to regulate gcvA-lacZ expression. When the mutations were introduced into the promoter region of the native gcvA gene, both resulted in constitutive expression of a gcvT-lacZ fusion. One of the mutations was shown by Western blot analysis to increase the intracellular concentration of GcvA. Overexpression of gcvR reestablished repression, supporting a model for gcv regulation that requires a relatively constant GcvA-to-GcvR ratio for appropriate regulation of gcv in response to glycine.

Fig. 7gcvA and gcvB control regions. The gcvA and gcvB promoter −35, −10, and +1 elements and the GcvA binding site ( , ) are indicated. The numbering is based on the +1 transcription start site for gcvB. The sites are approximately to scale.

Expression of a gcvR-lacZ fusion was neither induced by glycine nor repressed by inosine (63). In addition, a gcvR-lacZ fusion was not regulated by the GcvA protein or autogenously regulated (63). Rather, gcvR appears to be constitutively expressed. This likely keeps the GcvR levels relatively constant to maintain an appropriate GcvA/GcvR ratio so that the cell can respond to inducing and repressing growth conditions.

THE gcvB GENE

The E. coli gcvBgene encodes two small RNA transcripts of ~130 and ~200 nucleotides that are not translated in vivo (7, 229). These RNA molecules were discovered in the course of examining the promoter region for the gcvA gene (242). The gcvA gene, encoding GcvA, is autoregulated (243). In addition to this negative role at its own promoter, GcvA at the gcvA locus also activates the divergently transcribed and overlapping gcvB promoter when the cellular level of glycine is high and represses gcvB when glycine is limiting (Fig. 7) (229). Like the gcv operon, repression of gcvB by GcvA requires the negative-acting GcvR protein. Thus, qualitatively gcvB responds to the same signals as the gcv operon. However, unlike the gcv operon, activation of the gcvB promoter is not dependent on the Lrp protein. A strain carrying a chromosomal deletion of gcvB exhibits normal growth in LB and in GM medium compared to the wild type and normal regulation of the gcv operon and GCV enzyme activity. In addition, there appears to be no autoregulation of gcvB by its own RNA products, since a Δ gcvBstrain shows normal regulation of a gcvB-lacZ fusion. However, this mutant has high constitutive synthesis of OppA and DppA, the periplasmic binding protein components of the two major peptide transport systems normally repressed in cells growing in rich medium (6, 85, 149). Although the mechanism(s) involving the gcvB RNAs in the repression of these two genes is not known, oppA regulation appears to be at the translational level, whereas dppA regulation occurs at the mRNA level (229). The gcvB RNA is a target for Hfq, an RNA binding protein required for the action of many small RNAs that act by base pairing with target mRNAs (249). Thus, it is possible that the gcvB RNA interacts with the oppA and dppA mRNAs to carry out its regulatory function. Since oppA and dppA are involved in many cellular processes, there are likely dramatic downstream consequences of gcvB RNA action on these genes. For example, the OppABCDF transporter serves as the primary recycler of cell wall peptides (64), and DppA, in conjunction with the Tap signal transducer, is involved in peptide chemotaxis (119). In addition, some of the small molecules transported by these systems not only provide nutrients but serve as signals for intercellular communication and as toxins and antibiotics (81, 192).

At present, the role of gcvB in overall cell physiology is unknown. Although the two genes identified so far encode periplasmic proteins that bind oligopeptides of various lengths, other target genes likely exist that also respond to the gcvB RNAs (229). Of interest, both oppA-phoA and dppA-lacZ are highly expressed and unresponsive to gcvB in both GM medium plus glycine and GM medium plus inosine, exhibiting gcvB-mediated repression only in LB (229). Perhaps there are other target genes that can respond to the 25-fold difference in gcvB RNA levels in minimal media. It is likely that the gcvB-encoded RNAs in some way integrate the C₁ pathway with these other metabolic pathways. Understanding this network, how it is integrated into overall cell physiology, and why RNAs whose syntheses are in turn controlled by gcv-specific regulatory proteins regulate these other pathways will provide important insight into the basic strategies used by cells to coordinate the expression of diverse metabolic pathways. The analyses of the Sanger Center and National Center for Biotechnology Information databases predict that RNA homologues of the gcvB RNAs are produced in serovar Typhimurium, but whether the RNAs are produced and function as regulatory molecules in this organism has not been examined.

AMINOACYL-tRNA SYNTHETASES

The aminoacyl-tRNA synthetases are a group of enzymes that catalyze the attachment of amino acids to their cognate tRNA molecules (186). These enzymes can be divided into two classes on the basis of limited sequence similarities and types of active-site platforms (45). These two classes are the parallel β-sheet surrounded by α-helices (class I) and the antiparallel seven-stranded β-sheet surrounded by α-helices (class II). Glycyl-tRNA synthetase, encoded by glyS, is a class II aminoacyl-tRNA synthetase (50). In E. coli, glycyl-tRNA synthetase is an α₂β₂ dimer with subunit molecular weights of ~33,000 (α) and 80,000 (β) (99, 150, 186, 238). Interestingly, as a side reaction, this enzyme also synthesizes dinucleoside polyphosphates (50, 110). These compounds possibly participate in the regulation of cell functions, such as DNA replication and cell proliferation (165).

Seryl-tRNA synthetase, encoded by serS, is a class II aminoacyl-tRNA synthetase (30, 45, 77, 109, 186). The seryl-tRNA synthetase is an α ₂dimer with a subunit molecular weight of ~50,000 (77, 109). The E. coli seryl-tRNA synthetase has been crystallized, and the three-dimensional structure has been determined (29, 168). The main features of its structure are a protruding α-helical arm that is the N-terminal domain required for tRNA binding and a globular domain of seven antiparallel strands forming the C-terminal catalytic domain.

Knowledge about the regulation of glyS and serS expression is still poor. For glycyl-tRNA synthetase, studies suggest that the enzyme levels are subject to a type of metabolic regulation, with the number of molecules varying depending on the nature of the growth medium and the growth rate (136, 152, 175). For seryl-tRNA synthetase, a temperature-sensitive mutation in the structural gene was reported to result in altered rates of enzyme synthesis (82). However, it was shown that the steady-state levels of serS mRNA in the mutant and the wild type are similar and that the temperature-sensitive phenotype is likely due to a low level of specific activity of the mutant synthetase at the nonpermissive temperature rather than a decreased level of expression (47). A temperature-resistant revertant was found to have a change upstream of the −10 promoter sequence, increasing the steady-state level of serS mRNA and compensating for the low level of specific enzyme activity.

TRANSPORT MECHANISMS

The E. coli glycine transport system (Cyc) has been shown to transport glycine, d-alanine, d-serine, and the antibiotic d-cycloserine (24, 236, 237). This is the major route of glycine transport (K_m, 4 μM), although other routes of entry function at higher concentrations of amino acid (179). Mutations in cyc (also designated dag) result in d-cycloserine resistance and reduced uptake of glycine by >90% compared to the wild type (24, 62, 179, 237). The transport system appears to be constitutively expressed (24).

There are multiple transport systems for serine. A major threonine-serine system is an Na⁺-coupled symporter (74, 75, 144). A second l-serine-specific H⁺-coupled symporter system is induced by leucine (74). The gene for this system is under the control of the Lrp protein, and cells grown in the absence of leucine possess very little of this transport system. A third system is the leucine-isoleucine-valine system, a binding protein-dependent transport system that transports serine slowly (179). A final system, encoded by tdcC, is involved in the transport of threonine and serine during anaerobic growth (57, 216). This system is not functional in cells grown under aerobic conditions (84). The TdcC system is an H⁺-serine-threonine symporter (144).

Transport systems often play roles in the regulation of gene expression, usually by transporting effector molecules into the cell, where they are sensed by soluble or membrane-bound regulatory proteins (44). In other cases, membrane-spanning transport systems can exogenously regulate gene expression by producing an intracellular regulatory signal in response to the presence of extracellular effector molecules. This allows the regulation of gene expression by a compound in the medium when that compound is also a metabolic intermediate always present in the cell. The introduction of a second copy of the cyc gene into E. coli results in increased induction of a gcvT-lacZ fusion over that observed when only a single copy of cyc is present (62). This suggests that at least part of the increased induction of gcvT-lacZ by glycine results from an increase in glycine transport. However, several observations leave open the possibility that part of the induction signal is exogenous. Proline is transported into the cell by the putP transport system, where proline then induces both putA and putP (171). When wild-type and putP strains were grown in the presence of a proline-containing dipeptide, induction of putP and putA was the same in both strains and was not significantly different from that observed when the wild-type strain was grown in the presence of free proline (171). When wild-type and cyc strains were grown in the presence of a glycine-containing tripeptide, induction of gcv was the same in both strains but was significantly less than that observed when the wild-type strain was grown in an equimolar amount of free glycine (62). The difference in induction is possibly due to an exogenous induction signal by glycine, which would be absent in the case of induction by peptide-bound glycine. However, it is possible that the peptide-bound glycine is transported less efficiently than free glycine or that the conversion of tripeptide to free glycine is limiting. In addition, gcv mutants excrete glycine into the growth medium, presumably an indication that internal glycine pools are saturated (62, 159). Expression of a gcvT-lacZ fusion was significantly elevated in a gcv cyc⁺ mutant compared to a gcv cycdouble mutant grown in GM medium (62). Although the gcv cycstrain might be unable to maintain an internal glycine concentration equal to that of the gcv cyc⁺ strain, it is also possible that in the gcv cyc⁺ strain glycine excreted in the growth medium induces gcv exogenously. If exogenous induction occurs, these results would suggest that a functional glycine transport system is required.

ACKNOWLEDGMENTS

Studies from my laboratory cited in this review were supported by grant GM-26878 from the National Institute of General Medical Sciences and by award no. MCB 01-01918 from the U.S. National Science Foundation.