Ribose Utilization with an Excess of Mutarotase Causes Cell Death Due to Accumulation of Methylglyoxal

Abstract

Methylglyoxal (MG) is a highly reactive metabolic intermediate, presumably accumulated during uncontrolled carbohydrate metabolism. The major source of MG is dihydroxyacetone phosphate, which is catalyzed by MG synthase (the mgs product) in bacteria. We observed Escherichia coli cell death when the ribose transport system, consisting of the RbsDACBK proteins, was overproduced on multicopy plasmids. Almost 100% of cell death occurs a few hours after ribose addition (>10 mM), due to an accumulation of extracellular MG as detected by ¹H-nuclear magnetic resonance (¹H-NMR). Under lethal conditions, the concentration of MG produced in the medium reached approximately 1 mM after 4 h of ribose addition as measured by high-performance liquid chromatography. An excess of the protein RbsD, recently characterized as a mutarotase that catalyzes the conversion between the β-pyran and β-furan forms of ribose, was critical in accumulating the lethal level of MG, which was also shown to require ribokinase (RbsK). The intracellular level of ribose 5-phosphate increased with the presence of the protein RbsD, as determined by ³¹P-NMR. As expected, a mutation in the methylglyoxal synthase gene (mgs) abolished the production of MG. These results indicate that the enhanced ribose uptake and incorporation lead to an accumulation of MG, perhaps occurring via the pentose-phosphate pathway and via glycolysis with the intermediates fructose 6-phosphate and glyceraldehyde 3-phosphate. It was also demonstrated that a small amount of MG is synthesized by monoamine oxidase.

Sugar utilization in Escherichia coli is biased in favor of glucose, which is phosphorylated during uptake by the phosphotransferase system (15). Phosphorylated sugars are the major intermediates of the glycolytic pathway, although they are finally dephosphorylated into pyruvate. Intracellular levels of these phosphorylated compounds might be regulated by an appropriate level of glycolysis. A disturbance of glycolytic metabolism by an externally added phosphorylated sugar was shown to cause cell death with a production of methylglyoxal (MG), a toxic by-product generated from dihydroxyacetone phosphate (20). However, MG production was observed for only a subset of sugar phosphates which enter directly into the glycolytic pathway, including d-glucose 6-phosphate, d-fructose 6-phosphate, and d-mannose 6-phosphate (8).

The initial finding of the production of MG was made with E. coli cells when glycerol metabolism involving glycerol kinase was unmodulated (5). Cell death by MG occurs within a few hours. However, it remains largely unknown how this regulatory abnormality occurs, as well as why MG is produced. Recent evidence indicates that the accumulation of MG in various cell types is widely associated with cellular processes, including oxidative stress, apoptosis, and disease complications (19). The targets of MG modification, affecting both protein and nucleic acid, appear to be either specific or nonspecific, depending on the intracellular level of MG accumulation (10).

Ribose uptake in bacteria is mediated by a high-affinity transporter consisting of the extracytoplasmic binding protein (RbsB), membrane permease (RbsC), and cytoplasmic ATPase (RbsA) (14). Several low-affinity transporters can also serve as an alternative route for ribose uptake. These include the allose transporter (10), the xylose transporter (18), and the glucose phosphotransferase system with altered specificity and expression (13). It was previously known that the rbsD gene, the first component of the rbs operon, is required when ribose is utilized by the low-affinity transporters. Recently, RbsD was characterized as a ribose mutarotase that catalyzes the conversion between the β-pyran and β-furan forms of ribose. Since the majority of ribose exists as β-pyranose in solution, the intracellular supply of furanose may be limiting when the ribokinase with a preference for furanose is actively consuming ribose (17). This is particularly evident in cells transporting ribose at a lower affinity (13). It was found that monosaccharide mutarotases, including RbsD, are ubiquitous and conserved in both prokaryotes and eukaryotes (16).

Intracellular production of MG is achieved in E. coli and most bacteria by MG synthase encoded by the mgs gene (20). Although the corresponding gene in eukaryotic cells has yet to be characterized, alternative sources, such as P450, amine oxidase, and triose phosphate isomerase, have been implicated (19). The formation of MG from aminoacetone was reported for Staphylococcus aureus as part of a cycle for oxidation of l-threonine and glycine (11), while the compound can be generated nonenzymatically. Since MG is a strong electrophile known to modify a variety of macromolecules, cells have developed various methods of detoxification (9). The major pathway involves glyoxalase I, generating from MG S-d-lactoyl-glutathione, which is then converted to d-lactate and glutathione by glyoxalase II. The toxicity of MG in E. coli is affected by intracellular pH, which modifies the potassium efflux system activated by S-d-lactoyl-glutathione (20).

Here we report cell death caused by production of MG due to enhanced ribose uptake with an excess of mutarotase. A considerably higher level of MG was detected when cells were metabolizing ribose with a plasmid copy of RbsD. The unregulated production of MG appears to be due to a rapid increase in the glycolysis intermediates from ribose degradation. Such a metabolic burden may result in MG production by MG synthase. The majority of MG production is attributed to MG synthase, while the monoamine oxidase, being implicated in generating MG in other bacteria (11), contributed little MG.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

All strains used were derivatives of E. coli K-12 and are listed in Table 1. The MK1 and MK2 strains were constructed by transferring mgsA::kan from MJF 397 and mao::kan from RK2116, respectively, to CK281 by using P1. The MJF 397 and RK2116 strains were provided by I. R. Booth (University of Aberdeen) and Y. Murooka (Osaka University). To obtain the MK3 strain, maoA::kan from RK2116 was first transferred to SK317, which was then transferred to MK1, an mgsA::kan derivative of the CK281 strain. To detect the phenotype of monoamine oxidase, an M9 minimal plate was used with 3 mM dopamine as the sole nitrogen source and 0.4% succinate as a carbon source. The pJK5 plasmid was constructed by inserting the 490-bp XmnI/AvaI fragment of pYP22DA containing rbsD into pBR322 digested with EcoRV and AvaI. The 5,404-bp PvuII/EcoRV fragment of the rbs operon containing rbsACBK was cloned into the EcoRV site of pACYC184 to yield pJK10. To generate pJK12 containing rbsACB but not rbsK, a NaeI fragment of pJK10 was removed by partial digestion and ligation.

TABLE 1.

Strains and plasmids

Strain or plasmid	Relevant characteristicsa	Reference or source
Strains
OW1	leu thr his thi rpsL lacY xyl ara tonA tsx	13
CK281	OW1 rbsD-rbsR lacZ	This work
JM110	dam dcm supEA4 hsdR17 thi leu rpsL lacY galK galT ara tonA thr tsx Δ(lac-proAB) F−[traD36 proAB+laclqlacZΔM15]	Lab collection
SK317	JM110 trg2::Tn10	This work
MJF397	F−gal lacZ rha thi mgsA::kan	20
RK2116	recB recC sbcB maoA::kan	21
MK1	CK281 mgsA::kan	This work
MK2	CK281 maoA::kan	This work
MK3	CK281 maoA::kan mgsA::kan	This work
Plasmids
pACYC184	ori (p15A)Tcr Cmr	Lab collection
pBR322	ori (pMB1)Tcr Apr	Lab collection
pJK5	pBR322 rbsD+	This work
pJK10	pACYC184 rbsA+rpsB+rpsC+rpsK+	This work
pJK12	pACYC184 rbsA+rpsB+rpsC+	This work
pYP22DA	pBR322 rbsD+rpsA+	This work

^aAp, ampicillin; Cm, chloramphenicol; Tc, tetracycline.

M9 minimal medium (12) supplemented with 0.2% (13 mM) d-ribose (Sigma Chemical Co.) and/or 0.4% (wt/vol) glycerol as a carbon source was used. The M9 plate was prepared by addition of Bacto Agar (Difco) to a final concentration of 1.5%. For derivatives of the OW1 strain, supplements of 1 mM histidine, 1 mM leucine, 1 mM threonine, and 0.01% thiamine were added. Luria-Bertani (LB) medium (Difco) was prepared as described previously (12). When required, cultures were grown in the presence of ampicillin (100 μg/ml), chloramphenicol (30 μg/ml), or kanamycin (25 μg/ml).

Western blot analysis.

Cells were grown to stationary phase in M9 glycerol medium. After sonication, the samples were lysed by boiling in 5× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer. Equal amounts of proteins (20 μg) were separated by SDS-12% PAGE and transferred to a polyvinylidene difluoride membrane (Amersham Co.). The RbsD protein was detected by use of an anti-RbsD and a horseradish peroxidase-conjugated anti-rabbit antiserum, followed by enhanced chemiluminescence detection (ECL; Amersham Co.).

Uptake assay for d-[¹⁴C]ribose.

The uptake assay for d-[¹⁴C]ribose was carried out as described previously (6). Cells were grown in M9 medium with 0.4% glycerol, harvested at an optical density at 600 nm (OD₆₀₀) of 1.0, washed three times with the same volume of 10 mM phosphate buffer (pH 7.0), and resuspended in the same buffer to a final OD₆₀₀ of 0.2. The cell suspension was incubated at 30°C for 20 min. d-[¹⁴C]ribose (51.1 mCi/mmol; Du Pont Co.) was added at 2.0 μM to a 1.2-ml cell suspension. Samples of 0.3 ml were taken at 10, 20, 30, and 40 s and filtered through a 0.45-μm-pore-size nitrocellulose filter (Amicon). The filtered samples were washed twice with 0.3 ml of phosphate buffer. After the samples were dried, the radioactivity was measured in a scintillation cocktail.

Assays for cell growth, viability, and lethality.

Growth rates were determined by using cells cultured in M9 medium containing 0.4% glycerol and appropriate antibiotics. After overnight incubation in a shaking incubator at 37°C, the culture was divided into two batches of fresh M9 medium with 0.4% glycerol. After 2 h, 0.2% d-ribose was added to one of the batches. Growth was monitored at OD₆₀₀ with a Beckman DU-65 spectrophotometer. As a measure of cell viability, each strain was grown in M9 medium with 0.4% glycerol with appropriate antibiotics to an OD₆₀₀ of 1.0. After the addition of 0.2% d-ribose, samples were taken at intervals. Viability was scored by plating serial dilutions onto LB plates with appropriate antibiotics. The plates were incubated for 8 h at 37°C, and the numbers of colonies were counted. To observe growth on plates, a colony from each LB plate was streaked onto an M9 minimal plate with 0.2% d-ribose. After 36 h of incubation at 37°C, the colony sizes were compared.

NMR analysis.

The Bruker AVANCE-400 nuclear magnetic resonance (NMR) spectrometer, equipped with a temperature controller, was used for the NMR experiments. All NMR experiments were carried out with 500 μl of the supernatant, with 10% D₂O added as a locking substance. A 5-mm-diameter NMR tube was used, and the sample was kept at 28°C. As a measure of MG released from cells, growth media were taken at intervals, and the cells were removed by centrifugation for 1 min at 16,000 × g. The supernatant was stored at −20°C until the measurement was performed. For quantitative analysis, the ¹H-NMR experiment was performed with a 30° pulse with long relaxation delay. An assignment of the MG peak in the ¹H-NMR spectra was made with reference to the NMR spectrum for commercially available MG (Sigma Chemical Co.).

In order to detect ribose 5-phosphate, cells were prepared as described for the uptake assay for ribose, but with 5% ribose added. After 10 s of incubation, the cells were harvested by centrifugation for 3 min at 4,000 × g. The pellet was resuspended in 1 ml of phosphate buffer, followed by sonication and centrifugation for 5 min at 16,000 × g. The supernatant was immediately frozen and saved in a freezer until NMR measurements were done. The ³¹P-NMR assay was performed with a 30° pulse with a relaxation delay of 6 s. About 1,000 transients were accumulated for each sample (RbsD⁺ and RbsD⁻). For identification of ribose 5-phosphate, the RbsD⁺ sample was supplemented with ribose 5-phosphate (Sigma Chemical Co.) and used as a standard.

Quantitation of MG by HPLC.

The amount of MG was measured by high-performance liquid chromatography (HPLC) of growth medium with 0.2% d-ribose, which was prepared by a method similar to that used for the NMR experiment. As described previously (3), 0.1 ml of 5 M perchloric acid (Sigma Chemical Co.) was added to a 0.9-ml sample. Samples were incubated on ice for 10 min and centrifuged at 16,000 × g for 10 min. The samples were derivatized at 20°C for 4 h with 500 nmol of o-phenylenediamine (o-PD; Aldrich Chemical Co.), and loaded with an internal standard, 2.5 nmol of 5-methylquinoxaline (5-MQ; Aldrich Chemical Co.). Solid-phase extraction was performed with a C₁₈ SPE cartridge (Waters Sep-Pak C₁₈ plus cartridge; Millipore Corp.), previously activated with 8 ml of acetonitrile, followed by 8 ml of 10 mM KH₂PO₄ (pH 2.5). The cartridge was rinsed with 3 ml of 10 mM KH₂PO₄, pH 2.5, and the retentate was eluted with 3 ml of acetonitrile. Eluates were filtered through a 0.2-μm-pore-size Gelman polyvinylidene difluoride filter (Fisher Scientific).

The quinoxaline derivative of MG (2-MQ) was detected with an internal standard (5-MQ) at 315 nm by HPLC (HP1100 system; Agilent Technologies). The mobile phase was a 68:32 (vol/vol) solution of 10 mM KH₂PO₄ (pH 2.5)-HPLC-grade acetonitrile with flow rate of 1.0 ml/min. A volume of 150 μl was injected. Here, the retention times of 2-MQ and 5-MQ were approximately 6.3 and 9.7 min, respectively. Peak integrality ratios of 2-MQ to 5-MQ were used for quantitative analysis. Measurements were made in triplicate.

RESULTS

Overproduction of the ribose mutarotase RbsD inhibits growth on d-ribose.

In a previous study (13), it was shown that the RbsD protein was required for growth on ribose when the sugar was transported through a mutated glucose transporter (PtsG). Recently, RbsD was characterized as a novel type of mutarotase involved in the anomeric conversion of ribose (16). Here, we report a growth inhibition phenotype of RbsD when the rbsD and rbsACBK genes were overproduced on separate plasmids (Fig. 1). The rbsD plasmid (pJK5) was derived from pBR322, and the pJK10 (rbsACBK) plasmid was derived from pACYC184; both were carried by the CK281 strain with all rbs genes deleted (ΔrbsD-rbsR) (Table 1). The ribose-induced lethality was not observed when we overproduced only RbsD with a chromosomally encoded rbs operon or when rbsDACBK was overproduced in a single plasmid, although slight growth retardation was detected (data not shown). This may have been due to overproduction of the membrane proteins. Since the strain we used initially was found to have a mutation in mlc (13), encoding a negative regulator for ptsG, we excluded the possibility of an mlc effect by repeating the experiment with a strain derived from MC4100 (mlc⁺), showing the same growth phenotype as that of CK281 (data not shown). RbsD is normally not required for ribose utilization, as shown by the growth of strains containing only the rbsACBK components on ribose (Fig. 1A, right). The growth inhibition was strictly dependent on ribose and RbsD, regardless of the presence of other sugars, such as glycerol (Fig. 1A, left), xylose, xylulose, ribulose, and glucose (data not shown). When RbsD was present, the inhibition started immediately upon ribose addition, and no increase in cell density was observed (Fig. 1B, right).

FIG. 1.

Effect of RbsD on cell growth. (A) Growth of various E. coli strains on M9 minimal plates with 0.2% d-ribose (Rib) and 0.4% glycerol (Gly) as carbon sources. The plates were incubated for 1.5 days. Plate quadrants 1, CK281/pACYC184/pJK5; plate quadrants 2, CK281/pACYC184/ pBR322; plate quadrants 3, CK281/pJK10/pBR322; plate quadrants 4, CK281/pJK10/pJK5. (B) Growth in M9 minimal medium with 0.4% glycerol as a carbon source. After 2 h of incubation, 0.2% d-ribose was added (•). The strains used were CK281/pJK10/pBR322 (RbsD⁻) and CK281/pJK10/pJK5 (RbsD⁺).

Expression of RbsD from the plasmid copy (pJK5 in CK281) was considerably enhanced compared to that of the chromosomal copy (OW1), even when it was fully induced by 0.2% ribose (Fig. 2, Western blot). Such overproduction significantly affected the uptake of sugar such that the plasmid copy of RbsD increased the uptake rate to about 60%, which is almost twice that of the chromosome-encoded transporter (Fig. 2). An expression of the whole rbs operon (the RbsDACBK genes) with its own promoter produced about 35% RbsD protein without any lethal phenotype, as analyzed by Western blotting (data not shown).

FIG. 2.

Effect of RbsD on ribose uptake. Inset, Western blot for RbsD, which was analyzed by SDS-12% PAGE. Cells were grown in M9 minimal medium with 0.4% glycerol and resuspended in 10 mM phosphate buffer. After d-[¹⁴C]ribose was added, radioactivity was measured as described in the text. Uptake rates of the following strains are shown: ΔRbs, CK281/pACYC184/pBR322; Rbs⁺, OW1/pACYC184/pBR322; pACBK, CK281/pJK10/pBR322; pACBK/D, CK281/pJK10/pJK5.

Growth inhibition leads to cell death and requires ribokinase.

Since the inhibition of growth may reflect cell death without lysis, we carried out a viability assay for cells grown in M9 glycerol medium until the OD₆₀₀ reached 1.0. As shown in Fig. 3, the strain containing both pJK5 (RbsD) and pJK10 (RbsACBK) exhibited a dramatic reduction in the number of viable cells upon ribose addition at 0.2%. More than 50% of the cells died within an hour after ribose treatment. In contrast, the strain lacking RbsD did not show any hint of cell death. The lethal effect caused by RbsD was not observed when we removed rbsK for ribokinase from the plasmid (Fig. 3, RbsD⁺ RbsK⁻), indicating that phosphorylation of ribose was necessary for cell death.

FIG. 3.

Involvement of RbsD and RbsK in cell death. (A) Assay of cell survival. Cells were grown in M9 medium with 0.4% glycerol, to which 0.2% d-ribose was added. Cell viability was determined by serial dilutions and spreading onto LB plates: ○, CK281/pJK10/pJK5; •, CK281/pJK12/pBR322; ▾, CK281/pJK10/pBR322; ▿, CK281/pJK12/pJK5. D, RbsD; K, RbsK. (B) Accumulation of ribose 5-phosphate in the presence of RbsD. Cells (RbsD⁺ and RbsD⁻) were grown in M9 medium with 0.4% glycerol as described in Materials and Methods. Ribose 5-phosphate was detected by ³¹P-NMR measurement. The peaks detected to the right of ribose 5-phosphate presumably originated from the buffer or from other metabolites. The compound used as a standard was obtained from Sigma Chemical Co. and added to the RbsD⁺ sample (Std). D⁻, CK281/pJK10/pBR322; D⁺, CK281/pJK10/pJK5.

In order to demonstrate that RbsD indeed increases the intracellular incorporation of ribose, we carried out a ³¹P-NMR experiment to detect ribose 5-phosphate. Figure 3B shows that the peak at 6.4 ppm, confirmed to be ribose 5-phosphate by the addition of a commercially available compound, appeared within 10 s in the strain with RbsD overproduced but not in the strain lacking RbsD. The results indicate that RbsD as a mutarotase enhances phosphorylation of ribose by facilitating a formation of the furanose form.

Cell death by RbsD results from the production of MG.

Since we noticed that the kinetics of ribose-induced cell death showed a similarity to the kinetics observed for MG toxicity (8), the levels of MG produced by the ribose-treated cells were measured. MG was detected by ¹H-NMR in comparison to the commercially available compound. As shown in Fig. 4, only the RbsD-positive strain produced MG, which was determined by either NMR (Fig. 4A) or HPLC via 2-MQ derived from the reaction with o-PD (Fig. 4B). Cells were grown in M9 medium with 0.4% glycerol and mixed with 0.2% ribose at an OD₆₀₀ of approximately 1.0. The level of MG rose sharply upon addition of ribose. After several hours, the MG level reached a steady state (Fig. 4B) at ca. 1 mM. In the strain lacking RbsD, the MG level was very low and did not cause any cell death. It was reported that the threshold level of MG leading to cell death was about 0.6 mM (2).

FIG. 4.

Accumulation of MG in strain with RbsD. Cells were grown in M9 minimal medium with 0.2% d-ribose and 0.4% glycerol as carbon sources. The culture medium was sampled at the indicated times after addition of 0.2% d-ribose (13 mM). The samples were centrifuged, and the supernatants were used for further analysis. (A) Detection of MG by ¹H-NMR. The MG peaks are indicated by arrows. The peak found at 1.9 ppm in the commercially available MG (Sigma Chemical. Co.) used as a standard is acetic acid. (B) Assay for MG by HPLC. Samples were prepared by derivatization with o-PD at 20°C for 4 h and by extraction with a C₁₈ SPE cartridge. The HPLC mobile phase was composed of 68% 10 mM KH₂PO₄ (pH 2.5) and 32% acetonitrile. 5-MQ (100 μM) was added as an internal standard. d-Ribose was added at a concentration of 0 mM (▿ and ○) or 13 mM (▾ and •). D⁻, CK281/pJK10/pBR322; D⁺, CK281/pJK10/pJK5.

MG is primarily generated from dihydroxyacetone phosphate by MG synthase.

To assess whether the MG production by RbsD is mediated by MG synthase (the mgs product), we constructed mgsA::kan derivatives of the strains by P1 transduction. In the CK281 mgs strain with both RbsACBK and RbsD, only a slight effect of d-ribose on cell growth was observed (Fig. 5A), far below the lethality observed in mgs-positive cells (Fig. 3). The mgs-negative strain no longer produced MG (Fig. 5B).

FIG. 5.

(A) Effect of a mutation in MG synthase on cell viability. The mgs-null strain (MK1) was derived from CK281. Cells were grown in M9 minimal medium with 0.4% glycerol as a carbon source with (closed symbols) or without (open symbols) the addition of 0.2% d-ribose. The strain used was MK1/pJK10/pJK5 (• and ○). (B) An involvement of monoamine oxidase (MA) in MG production. The maoA derivative (MK2) and the maoA mgs derivative (MK3) were from CK281. Cells were grown in M9 medium with 0.4% glycerol with supplements including 1 mM threonine, to which 0.2% d-ribose was added. Strains are CK281/pJK10/pJK5 (•), MK1/pJK10/pJK5 (○), MK2/pJK10/pJK5 (▾), and MK3/pJK10/pJK5 (▿).

Since MG was shown to be derived from aminoacetone in S. aureus (7), we investigated the involvement of monoamine oxidase in the formation of MG. Two strains with maoA single and maoA mgsA double mutations were constructed from CK281 (see Materials and Methods). As shown in Fig. 5B, the amount of MG produced in the maoA-negative strain was lower than that of the wild type, indicating that the monoamine oxidase was partially responsible (less than 20%) for the production of MG. However, MG was undetectable in the mgs-negative strain as in the maoA mgsA double mutant, presumably because the low level can be detoxified by cellular enzymes.

DISCUSSION

The metabolic production of MG was initially found in a strain with a mutation in the glycerol kinase gene which resulted in an unregulated incorporation of glycerol (5). Subsequent evidence indicated that other sugars, including d-xylose, l-arabinose, and even sugar phosphates, can elicit MG synthesis when uptake is amplified or unregulated (1). d-Ribose was also reported to accumulate MG to a lesser degree (20). In this study, we demonstrated that an enormous amount of MG was produced due to an enhanced initial catabolism of ribose, involving the ribose mutarotase. More than 90% of cells died after an hour, due to an accumulation of MG, which increased to 1 mM within a few hours. At this concentration of MG, almost all cells were killed. Thus, the threshold concentration for lethality would be much lower, as indicated earlier (2, 20). Since cell death begins to be seen at a low external MG concentration, the intracellular level of MG is likely to be much higher. Our attempt to measure intracellular MG detected a concentration of about 1.5 mM after 1 h of ribose addition, more than three times higher than the extracellular concentration (data not shown), which is consistent with a previous result (22).

Although an initial observation of cell death (Fig. 1) was made for the strain with both the transporter (RbsACBK) and RbsD overproduced, it was found that a strain without RbsD also accumulated some amount of MG (Fig. 4B). In addition, adding more ribose to the transporter-overproduced strain without RbsD enhanced MG production (data not shown), implying that the metabolic disturbance leading to MG production is due to a steep increase in intracellular ribose catabolism. However, without RbsD, the simple increase in sugar concentration did not result in persistent cell death (data not shown). Thus, the role of RbsD appears to be critical, presumably in rate enhancement towards the lethal production of MG. Since RbsD is involved in the anomeric conversion of ribose, the effect of RbsD on sugar uptake may be due to its activity on the early stage of ribose conversion after entry into the cell.

Ribose enters the glycolytic pathway after being converted to xylulose 5-phosphate via ribose 5-phosphate (Fig. 6). The entry is mediated by a concerted action of transketolase and transaldolase to generate fructose 6-phosphate and glyceraldehyde 3-phosphate. Enhanced uptake of xylose was reported to result in the accumulation of MG, perhaps by increasing the levels of the glycolytic intermediates. However, ribose was not as efficient in producing MG as was xylose (19), which might be due to differences in their metabolic rates. In our experiment, overproduction of RbsD was critical in facilitating the entry of ribose into the glycolytic pathway. It appears that this is due to a faster conversion of ribose into ribose 5-phosphate. Previously, an external addition of 2.5 mM ribose 5-phosphate had only a slight effect on the MG level (between 60 and 70 μM) (7). In our experiment, an intracellular level of ribose 5-phosphate would be much higher, since the ribose was added at a concentration greater than 10 mM.

FIG. 6.

Metabolic pathway leading to generation and detoxification of MG. Abbreviations: D-Rib, d-ribose; D-Rib-5-P, d-ribose 5-phosphate; D-Ru-5-P, d-ribulose 5-phosphate; D-Xu-5-P, d-xylulose 5-phosphate; F-6-P, fructose 6-phosphate; FDP, fructose diphosphate; G-3-P, glyceraldehyde 3-phosphate; DHAP, dihydroxyacetone phosphate; rpi, the ribose 5-phosphate isomerase gene; rpe, the ribulose 5-phosphate 3-epimerase gene; tkt, the transketolase gene; tal, the transaldolase gene; mgsA, the MG synthase A gene; maoA, the monoamine oxidase A gene; gloI,II, the glyoxalase I,II gene; GSH, glutathione.

It has been reported that the removal of MG synthase from E. coli completely abolished the production of MG when cells were grown with elevated xylose metabolism (1). This is also the case with the ribose-grown cells investigated here. However, it was shown that some MG may be derived from threonine by monoamine oxidase (Fig. 5B), which is presumably due to the threonine added as a growth supplement. As a matter of fact, externally added threonine enhanced the production of MG in the threonine prototroph (data not shown). The lack of MG accumulation in the mgs-defective strain could be due to MG detoxification. It is likely that the MG level obtained solely from threonine degradation can be tolerated by the detoxification system using glutathione. At any rate, our results suggest that threonine catabolism into MG exists in E. coli it does as in S. aureus (7).

The exact role of MG in cell growth is still poorly understood, although evidence of its signaling function in eukaryotic cells has been proposed previously (4). In prokaryotes, the Mgs pathway appears to be crucial in reducing the level of sugar phosphates by generating MG. Therefore, an mgs-negative strain may suffer growth impairment under physiological conditions with excess sugar phosphates. In our case, an addition of phosphates to the medium relieved the growth inhibition of the mgs-negative strain (I. Kim, unpublished data), implying that the MG pathway contributes to intracellular recycling of inorganic phosphate. It is conceivable that cells might have to develop the Mgs pathway during evolution in order to better cope with a physiological assault imposed by a constantly changing environment.