Three pathways for trehalose metabolism in Corynebacterium glutamicum ATCC13032 and their significance in response to osmotic stress


E-mail; Tel. (+49) 221 470 6464; Fax (+49) 221 470 5091.


Genome scanning of Corynebacterium glutamicum ATCC13032 revealed the presence of five different genes encoding enzymes belonging to three putative trehalose biosynthesis pathways (OtsAB, TreYZ, TreS). The function of the different pathways and of trehalose as an osmoprotectant was studied by characterizing several strains defective for individual trehalose biosynthetic routes. Trehalose synthesis was shown to increase upon hyperosmotic conditions. Cytoplasmic trehalose levels varied considerably depending on kind and accessibility of carbon and nitrogen sources. In contrast to other organisms, osmoregulated trehalose synthesis in C. glutamicum is mediated by the TreYZ and not by the OtsAB pathway. Irrespective of their significance for the osmotic response, otsA and treS were upregulated at the transcriptional level after hyperosmotic shock. In vivo, TreS-mediated trehalose synthesis only occurred if maltose was used as the carbon source. In vitro, TreS catalysed the conversion of maltose into trehalose and, conversely, trehalose into maltose. As the reaction seems to be near equilibrium, TreS appears to be important for trehalose degradation rather than synthesis because a 1000-fold excess of trehalose to maltose was detected in the cytoplasm. Also, evidence is given that both the OtsAB and the TreYZ pathways are involved, but not essential, in supplying trehalose for mycolic acid biosynthesis.


Trehalose, a non-reducing (α-1,1) glucose disaccharide isolated from bacteria, fungi, plants and mammals, can serve a variety of purposes, for example as a carbon source, storage carbohydrate or stress protection compound (Thevelein, 1984; Wiemken, 1990; Ranade and Vining, 1993; Strøm and Kaasen, 1993; Crowe et al., 1998; Argüelles, 2000). Its success in nature in the latter context is explained by its unique physical and chemical properties, including a high degree of chemical stability, hydrophilicity, lack of intramolecular hydrogen bonding and ability of non-hygroscopic glass formation (Crowe et al., 1998). At increased osmolalities, trehalose acts as a compatible solute (Giæver et al., 1988; Hounsa et al., 1998) by stabilizing the native conformation of proteins through preferential exclusion (Arakawa and Timasheff, 1985; Xie and Timasheff, 1997). However, unlike other compatible solutes, non-reducing disaccharides such as trehalose are able to stabilize macromolecules in the fully dehydrated state by direct interaction (Crowe et al., 1984a,b; 1998). Fundamental to this could to be the ability of non-reducing disaccharides to replace surface-associated water molecules by hydrogen bonding and to form glassy structures (Crowe et al., 1998).

The most common route of de novo trehalose synthesis is by UDP-glucose and glucose-6-phosphate forming trehalose-6-phosphate with subsequent dephosphorylation yielding free trehalose (Fig. 1). The reactions are catalysed by trehalose-6-phosphate synthase and trehalose-6-phosphate phosphatase respectively. The corresponding genes were first identified in Escherichia coli and termed otsA/B (osmotically regulated trehalose synthesis) on account of their involvement in osmoregulation. Subsequently, this pathway has been identified in a number of other organisms including several bacteria, yeast and other fungi as well as plants such as Arabidopsis thaliana (De Virgilio et al., 1993; Lippert et al., 1993; Vogel et al., 1998; De Smet et al., 2000)

Figure 1.

Known trehalose biosynthesis pathways.

Two additional, but less prominent, routes for trehalose synthesis have been discovered in the past decade without any evidence for an involvement in stress protection (Fig. 1). One catalyses the conversion of oligo/polymaltodextrins/glycogen into trehalose in a two-step reaction by maltooligosyltrehalose synthase (TreY) and maltooligosyltrehalose trehalohydrolase (TreZ). This pathway was identified in Rhizobium, Arthrobacter as well as several Sulfolobus and Mycobacterium species (Kobayashi et al., 1996; Maruta et al., 1996; De Smet et al., 2000). The other comprises a single transglycosylation reaction producing trehalose from maltose by the action of the trehalose synthase TreS. This enzyme was identified in Pimelobacter sp. R48, Thermus aquaticus (Tsusaki et al., 1996) and Mycobacterium sp. (De Smet et al., 2000).

Most organisms in which trehalose synthesis has been studied possess only a single synthesis pathway, whereas in mycobacteria, genes coding for the enzymes of all three pathways were detected (De Smet et al., 2000). These enzymes were shown to be active in crude extracts. However, no physiological functions were identified for the individual pathways. Although trehalose synthesis usually occurs in a non-constitutive, environment-dependent manner, certain mycobacteria were shown to exhibit a steady-state trehalose pool subject to a constant turnover (De Smet et al., 2000). Mycobacteria, along with corynebacteria and nocardiae, differ from other Gram-positive bacteria by their peculiar cell wall consisting, to a large extent, of so-called mycolic acids. A large fraction of mycolic acids exists in a trehalose-esterified form, thus requiring trehalose synthesis. In addition to acting as a cell wall component, trehalose may play a role as a stress metabolite in C. glutamicum which, as a soil bacterium, frequently faces changing conditions with respect to the water content of the surrounding environment. This paper presents evidence that C. glutamicum, similar to mycobacteria, possesses three active trehalose synthesis pathways. By characterizing the response of C. glutamicum to different hyperosmotic conditions, this paper shows that trehalose, along with several amino acids, is synthesized as a compatible solute becoming the predominant cytoplasmic osmolyte upon nitrogen limitation or starvation. In addition, a model is proposed ascribing different functions to the individual pathways of trehalose metabolism in C. glutamicum.


The impact of trehalose in the osmostress response of C. glutamicum

As a soil bacterium, C. glutamicum is frequently exposed to changes in the external osmolality that can occur on desiccation of the habitat. To circumvent the deleterious effects caused by hyperosmotic stress, C. glutamicum accumulates compatible solutes in the cytoplasm by either synthesis or uptake. These experiments investigate the biosynthesis of trehalose, both on long-term adaptation (Fig. 2A and B) to a constant high osmolality and on short-term adaptation to a sudden osmotic upshift (Fig. 2C and D). As only biosynthesis of compatible solutes was of interest, media free of compounds that could serve as compatible solutes through uptake were used in all experiments.

Figure 2.

Composition of the compatible solute pool of C. glutamicum wild-type cells under various conditions: (A) at constant osmolality of 0.9 osM; (B) or of 2.4 osM; (C) growth at 0.9 osM in the exponential phase; or (D) after a sudden upshift in osmolality from 0.9 osM to 2.4 osM at the same fermentation time. The upshift in osmolality was carried out by the addition of NaCl. Filled squares, trehalose; open inverted triangles, proline; asterisks, glutamate; open triangles, glutamine; open circles, bacterial growth.

Growing C. glutamicum at an osmolality of 2.4 osM decreases the growth rate from 0.43 h−1, as detected in the basal medium of 0.9 osM, to 0.23 h−1 (Fig. 2A and B). As an adaptation to osmotic stress, both the concentration and the pattern of compatible solutes in the cytoplasm were found to change during fermentation. During growth in basal medium, the amounts of trehalose, proline and glutamate did not exceed 50, 150 or 200 µmol g−1 cell dry weight (cdw) respectively (Fig. 2A). During growth at 2.4 osM (Fig. 2B), the internal concentration of glutamate did not change significantly, indicating that glutamate is not involved in the long-term adaptation of C. glutamicum to osmotic stress. The amount of trehalose in the cell was almost doubled, with highest concentrations found in the lag phase and in the stationary phase. The accumulation pattern and the low overall concentrations provide evidence that trehalose does not play an important role in the long-term osmostress response under the experimental conditions applied. However, significant changes were detected for cytoplasmic proline concentration, which increased about eightfold from ≈ 100–800 µmol g−1 cdw upon growth at high osmolality. After reaching maximum concentrations during exponential growth, proline levels remained constant and decreased again after the onset of stationary phase.

When facing a sudden shift to high osmolality, cells must react promptly in order to counteract the immediate consequences of upshift, i.e. loss of cytoplasmic water, decrease in turgor and cessation of growth. As a first rapid reaction to osmotic upshift, potassium was accumulated transiently (from ≈ 450 to 800 µmol g−1 cdw), immediately followed by glutamate, trehalose and glutamine synthesis (Fig. 2D). Although the accumulation of the latter solutes was only temporary, reaching maximum values within the first hour after osmotic shift, cytoplasmic proline concentrations increased steadily for at least 3 h, reaching values three- to eightfold higher than the content of glutamate, glutamine and trehalose respectively.

Dependence of trehalose synthesis on the availability of nitrogen sources

The experiments described above suggest that proline, and not trehalose, is the major de novo-synthesized compatible solute of C. glutamicum. A principal difference between trehalose and all other detected solutes is the absence of nitrogen in its chemical structure. The accumulation of different types of compatible solutes and their physiological significance may therefore depend on the availability of nitrogen. Thus, experiments were carried out examining the influence of nitrogen limitation and nitrogen starvation on the pattern of compatible solute synthesis (Fig. 3). To limit nitrogen, all nitrogen-containing compounds were replaced by 100 mM glutamine in the medium. The uptake of glutamine can be limited by providing trace amounts of sodium, the co-substrate for glutamine transport (Siewe et al., 1995). To this end, Na+ was excluded from all medium ingredients, resulting in residual sodium concentrations of ≈ 100 µM resulting from chemical contamination, which is 10-fold below the KM value of 1.4 mM for sodium (Siewe et al., 1995). To achieve nitrogen starvation, CgXII medium was used in the absence of a nitrogen source. Osmotic upshock was performed by adding 750 mM KCl instead of 750 mM NaCl. The results of these experiments show that, in the case of nitrogen limitation, a significant increase in the maximum trehalose content was observed in comparison with N-surplus conditions (Fig. 3B). This occurred at the expense of amino acid synthesis as demonstrated in the case of proline; diminished to < 50% of its original value (Fig. 3C). In the complete absence of a nitrogen source, trehalose was the only detectable compatible solute after osmotic upshift (Fig. 3B and C), reaching a maximum value of ≈ 600 µmol g−1 cdw, i.e. about six times the amount determined in the presence of sufficient nitrogen. As expected, neither proline nor any other amino acid was synthesized after osmotic upshock. Only a low steady-state glutamate concentration of ≈ 20 µmol g−1 cdw was still detectable.

Figure 3.

Effect of nitrogen limitation and nitrogen starvation on (A) growth, (B) trehalose and (C) proline synthesis in wild-type cells after a hyperosmotic shock by the addition of 750 mM KCl. Closed circles, N-surplus; open squares, N-limitation; asterisks, N-starvation. Please compare ‘Experimental conditions’ for the set up of N limitation and starvation. The trehalose concentration in the non-shocked control cells did not exceed 10, 30 or 75 µmol g−1 cdw in N-surplus, N-limitation or N-starvation medium respectively (compare Supplementary material, Fig. S1).

Dependence of trehalose synthesis on carbon sources

For characterizing the dependence on the availability of appropriate carbon sources, the trehalose content of cells grown on 4% glucose, sucrose or maltose in standard CgXII medium, i.e. under N-surplus conditions, was determined (Fig. 4). At constant osmolality of 0.9 osM, trehalose synthesis from maltose was highest, followed by glucose and sucrose (Fig. 4A and B). After osmotic upshift from 0.9 to 2.4 osM (Fig. 4C–E), amounts of trehalose synthesized from maltose and glucose were almost equal, exceeding those synthesized from sucrose by a factor of 2–3 (Fig. 4D). As mentioned above, proline synthesis responds to changes in trehalose synthesis. Consistent with this finding, proline synthesis was significantly reduced when maltose was used as carbon source (Fig. 4E).

Figure 4.

Effect of the carbon source on trehalose content in wild-type cells. (A) Growth and (B) trehalose content during a cultivation at constant osmolality of 0.9 osM. (C) Growth and (D) trehalose and (E) proline synthesis after a hyperosmotic shock from 0.9 to 2.4 osM induced by the addition of 750 mM NaCl in N-surplus CgXII medium. The carbon sources: filled squares, glucose; asterisks, sucrose; open triangles, maltose. The trehalose concentration in the non-shocked control cells did not exceed 30, 50 or 160 µmol g−1 cdw in the indicated growth phase with sucrose, glucose or maltose as carbon source respectively (compare Supplementary material, Fig. S1).

Function of the different trehalose biosynthesis pathways in the osmotic stress response of C. glutamicum

Genes belonging to the three trehalose synthesis pathways identified recently in mycobacteria are also present in the genome of C. glutamicum, a close relative of mycobacteria. According to those genes, the corresponding pathways were designated OtsAB, TreYZ and TreS (cf. Fig. 1). To study the physiological significance of these pathways, a number of strains defective in different biosynthetic routes were constructed. For pathways including sequential catalytic steps, the gene encoding the first enzyme was deleted in order to avoid the accumulation of potentially toxic intermediates.

In order to study the possible involvement of a given trehalose synthesis pathway in the long-term osmostress response of C. glutamicum, various strains defective in different trehalose synthesis pathways were grown in a high-osmolality medium (2.4 osM). Growth properties and cytoplasmic trehalose contents were then determined (Fig. 5). Although growth of strains CglΔotsAΔtreS and CglΔtreYΔtreS was impaired under elevated osmolality, similar to that of the wild type, strains CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY suffered a more drastic growth inhibition (Fig. 5C). However, at low osmolality, growth of these two strains was also considerably hampered (Fig. 5A), indicating a more general effect than an increase in osmosensitivity. Both wild type and CglΔotsAΔtreS exhibited similar internal trehalose concentrations, whereas CglΔtreYΔtreS possessed only half the trehalose content without any negative effect on growth (Fig. 5C and D). No trehalose was found in strains CglΔotsAΔtreY and CglΔotsAΔtreYΔtreS, indicating that trehalose is not synthesized by TreS under these experimental conditions (Fig. 5B and D). Furthermore, the absence of cytoplasmic trehalose seems to be the reason for the observed growth inhibition.

Figure 5.

Figure 5.

Osmolality-dependent ( A and C) growth and ( B, D and E) synthesis of trehalose and proline of C . glutamicum strains deficient for different trehalose biosynthesis pathways. Cultivations were carried out at 0.9 osM (A and B) or elevated osmolality of 2.4 osM (C–E). Filled squares, wild type ATCC13032; open circles, CglΔotsAΔtreS; open diamonds, CglΔtreYΔtreS; open inverted triangles, CglΔotsAΔtreY; crosses, CglΔotsAΔtreSΔtreY.

Figure 5.

Figure 5.

Osmolality-dependent ( A and C) growth and ( B, D and E) synthesis of trehalose and proline of C . glutamicum strains deficient for different trehalose biosynthesis pathways. Cultivations were carried out at 0.9 osM (A and B) or elevated osmolality of 2.4 osM (C–E). Filled squares, wild type ATCC13032; open circles, CglΔotsAΔtreS; open diamonds, CglΔtreYΔtreS; open inverted triangles, CglΔotsAΔtreY; crosses, CglΔotsAΔtreSΔtreY.

To study possible involvement of individual trehalose synthesis pathways in trehalose accumulation after a hyperosmotic shock, various strains defective in different trehalose synthesis pathways (single and double deletion strains) were subjected to a sudden osmotic shift from 0.9 to 2.4 osM by the addition of 750 mM NaCl. Cells were analysed for trehalose content and ability to recover from shock (Fig. 6). Although the trehalose content of wild-type cells reached a maximum of ≈ 100 µmol g−1 cdw, the strain possessing solely TreYZ (CglΔotsAΔtreS) accumulated 1.5-fold higher amounts (Fig. 6D). In contrast, OtsAB as the sole pathway of CglΔtreSΔtreY synthesized only minor amounts of the disaccharide. Despite this difference, no effects on the recovery from osmotic upshift, e.g. resumption of bacterial growth, were observed (Fig. 6C). This may be explained by the dominant role of proline as a compatible solute under these N-surplus conditions. As found during cultivation in media of constant osmolality, strains CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY were totally devoid of cytoplasmic trehalose. Although proline was still present in significant amounts (up to 800 µmol g−1 cdw), recovery from hyperosmotic shock was severely impaired (Fig. 6C) in those strains.

Figure 6.

Resumption of bacterial growth (A and C) and trehalose synthesis (B and D) after a hyperosmotic shock from 0.9 to 2.4 osM by the addition of 750 mM NaCl of strains defective in one or several trehalose synthesis pathways.
A and B. Filled squares, wild type ATCC13032; open squares, CglΔotsA; filled inverted triangles, CglΔtreY; asterisks, Cgl13032ΔtreS.
C and D. Filled squares, wild type ATCC13032; open circles, CglΔotsAΔtreS; open diamonds, CglΔtreYΔtreS; open inverted triangles, CglΔotsAΔtreY; crosses, CglΔotsAΔtreSΔtreY.

To exclude pleiotropic effects caused by multiple gene knock-outs, experiments were performed using strains defective in one synthesis pathway only (Fig. 6A and B). These results were in agreement with those obtained using double deletion strains. The loss of the TreYZ pathway led to very low cytoplasmic trehalose levels, whereas the absence of the OtsAB pathway resulted in a slightly decreased trehalose pool. None of these strains was inhibited in recovery from osmotic shock (Fig. 6A). Surprisingly, after the osmotic upshock, the treS deletion strain led to increased cytoplasmic trehalose concentrations (Fig. 6B) and a growth delay of ≈ 100 min (Fig. 6A). The increased trehalose concentrations in treS deletion strains (Fig. 6B and D and Fig. S2 in Supplementary material) are in apparent conflict with the assumed role of TreS as a trehalose synthase as reported for other organisms.

Taken together, osmoresponsive trehalose synthesis both under constant hyperosmotic condition and after hyperosmotic shock mainly depended on the TreYZ and only marginally on the OtsAB pathway. TreS-mediated trehalose synthesis was not measurable under any condition. Based on the observed growth inhibition of strains CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY, a fundamental metabolic effect conferred by the total absence of trehalose must be assumed.

Expression of trehalose biosynthesis genes in response to osmotic stress

To examine whether the change in TreYZ-mediated osmoresponsive trehalose synthesis was based on expression regulation, the time course of gene transcription after a hyperosmotic shock was analysed by RNA hybridization experiments using digoxygenin (DIG)-labelled antisense RNA probes. The experiments were carried out under conditions identical to those shown in Fig. 2. Surprisingly, the transcripts of genes belonging to the osmoresponsive TreYZ pathway were increased after an osmotic upshock at most by a factor of two (Fig. 7), indicating that treY and treZ are constitutively expressed to guarantee an immediate response under osmotic stress. Although under conditions of a sudden osmotic upshift, almost no OtsAB-dependent trehalose synthesis was found (cf. Fig. 6D), the otsA transcript was upregulated by a factor of five after 15 min. Furthermore, after 60 min, the treS transcript was also induced threefold. In a different approach, we tested whether the availability of nitrogen has an influence on the expression of the trehalose biosynthesis genes. After osmotic upshock, in both N-limiting and N starvation conditions, no change in the expression profile was detected. This indicates that the regulation of the internal trehalose concentration occurs mainly at the level of enzyme activity (data not shown).

Figure 7.

Changes in transcript amounts of treY, treZ, otsA and treS in response to a hyperosmotic shock from 0.9 to 2.4 osM by the addition of 750 mM NaCl in the wild-type ATCC13032. At the indicated times, cells were harvested. RNA was isolated, transferred on a nylon membrane and hybridized with the depicted antisense RNA probes. An antisense 16S rRNA probe was used as control. The signal intensity was measured by means of quantification of the emitted chemiluminescence. Highest signal intensity was set to 100%.

The physiological function of TreS

TreS was studied in more detail, as its physiological function was unclear on the basis of the results presented so far. Because of maltose being the presumptive substrate of TreS, the influence of this disaccharide on TreS-mediated trehalose synthesis was investigated. Strains CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY were grown in the presence of maltose as sole carbon source, and cytoplasmic trehalose was determined at different times during cultivation. Sucrose, the routine carbon source, was used in control experiments. Consistent with former experiments, the two strains did not synthesize trehalose if sucrose was the sole carbon source (Fig. 8A). In the presence of maltose, however, significant amounts of trehalose were found in CglΔotsAΔtreY cells (Fig. 8B) but not in CglΔotsAΔtreSΔtreY cells. This demonstrates that trehalose, detected in strain CglΔotsAΔtreY, resulted from TreS activity. Consequently, this indicates that, in principle, TreS can catalyse the conversion from maltose to trehalose. The presence of trehalose in strain CglΔotsAΔtreY could only partly release growth inhibition (not shown), yet its growth was still inferior to the wild type grown on maltose but proceeded noticeably faster than that of the triple deletion strain.

Figure 8.

Properties of TreS. Cytoplasmic trehalose in cells grown with sucrose (A) or maltose (B) as carbon source. TreS-dependent transformation of maltose to trehalose and vice versa in crude cell extracts (C). Time course of trehalose degradation in crude extracts (D).
A and B. Filled squares, wild-type ATCC13032; open inverted triangles, crosses, CglΔotsAΔtreY; (CglΔotsAΔtreSΔtreY.
C. Open squares, trehalose from maltose; asterisks, maltose from trehalose, CglΔotsAΔtreY extract. Open inverted triangles, trehalose from maltose; crosses, maltose from trehalose, CglΔotsAΔtreSΔtreY extract.
D. Asterisks, CglΔotsAΔtreY extract; crosses, CglΔotsAΔtreSΔtreY extract.

The described inactivity of TreS in terms of trehalose synthesis and the significantly increased cytoplasmic amount of trehalose in treS deletion strains (Fig. 6B and D) led to the hypothesis that TreS is involved in trehalose degradation rather than synthesis. This is supported by the trehalose-into-maltose converting activity that has been demonstrated with recombinant mycobacterial TreS (De Smet et al., 2000). To test whether this is also true for C. glutamicum, enzyme assays were carried out using crude cell extracts of CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY. As shown in Fig. 8C, maltose synthesis from trehalose, as well as the reverse reaction, could be detected in ΔotsAΔtreY extracts. Both conversions did not occur in ΔotsAΔtreSΔtreY extracts, indicating that TreS represents the only enzyme of C. glutamicum capable of converting maltose into trehalose and vice versa. Interestingly, in crude extracts of the triple deletion strain, no other trehalose-degrading activity is present (Fig. 8D and Fig. S3 in Supplementary material). This strengthens the hypothesis that TreS is responsible for trehalose degradation. That the TreS reaction is obviously near equilibrium in the in vitro assay (Fig. 8) and that trehalose in cells grown on sucrose is present at a trehalose–maltose ratio of 1350+/−300:1 in the exponential phase and of 4000+/−480:1 in the stationary phase strongly suggest that TreS is predominantly carrying out the conversion of trehalose to maltose in vivo.

Physiological function of the OtsAB pathway

Corynebacteria, mycobacteria and nocardiae differ from other bacteria in their peculiar cell wall, formed to a major extent by mycolic acids, present largely in trehalose-esterified form. This fraction of the cell wall mycolates is sensitive to saponification. The so-called trehalosyl mycolates represent the only trehalose-containing cell wall components described up to now. Trehalose-6-phosphate, the product of the OtsA reaction, was shown to be necessary for the production of trehalosyl mycolates and other types of cellular mycolates in Corynebacterium matruchotii (Shimakata and Minatogawa, 2000). As the OtsAB pathway obviously plays no crucial role in the osmotic stress response of C. glutamicum, we investigated whether OtsA is important for mycolic acid biosynthesis. To this end, cell wall extracts were prepared from C. glutamicum whole cells and, as a measure of the presence of trehalosyl mycolates, the trehalose content was determined after saponification of the extracts. Figure 9 shows the content of trehalose in cell wall extracts in a set of different deletion strains. With the exception of CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY, trehalose was detected in all cell wall extracts studied, which is indicative of the presence of trehalosyl mycolates. However, the loss of OtsA activity in the corresponding ΔotsA single deletion strain did not result in the absence of trehalose from the cell wall, i.e. in trehalosyl mycolate deficiency, which would be predicted on the basis of the in vitro data of Shimakata and Minatogawa (2000). In contrast, the trehalose content of the cell wall of strain CglΔotsA was only slightly decreased compared with the wild type, an effect that was also observed in the ΔtreY strain. Therefore, OtsA activity itself is not necessary for mycolic acid biosynthesis.

Figure 9.

Trehalose isolated from cell wall extracts of various strains by means of saponification. 1, wild-type ATCC13032; 2, CglΔotsA; 3, CglΔtreY; 4, CglΔtreS; 5, CglΔotsAΔtreY; 6, CglΔotsAΔtreSΔtreY.


Trehalose is a stress metabolite widely distributed in nature. For an immotile soil bacterium such as C. glutamicum, frequently confronting changing temperature and water activity in its habitat, such a substance could be of great importance. A number of studies have been done on the osmoregulation of this organism concerning uptake of osmoprotective substances (Peter et al., 1998; Rübenhagen et al., 2000; 2001). However, only limited knowledge is available on the synthesis of such substances, e.g. trehalose (Guillouet and Engasser, 1995). The presence of three putative pathways for trehalose synthesis suggests an important function of this metabolite in C. glutamicum, therefore we examined its function, with a special focus on osmoprotection.

The results presented in this paper indicate that the significance of trehalose as a compatible solute in C. glutamicum strongly depends on environmental conditions. In the presence of excess nitrogen, proline is the predominant compatible solute, and trehalose plays only a marginal role. These findings agree with previously published data (Guillouet and Engasser, 1995). Upon limiting or removing the nitrogen source, a marked shift from amino acids was observed, i.e. nitrogen-containing solutes to the carbohydrate trehalose. Soil bacteria are periodically confronted with nutrient limitations in addition to changes in osmolality; thus, the limiting conditions in situ may well reflect naturally occurring conditions. The ability to alter the internal composition of compatible solutes, depending on the surrounding environmental conditions, may indicate that C. glutamicum lives in moderate environments where a variety of compatible solutes would provide sufficient protection. Under nitrogen-limiting conditions, the halophilic bacterium Ectothiorhodospira halochloris, which synthesizes betaine and trehalose in response to osmotic stress, was unable to increase trehalose levels above 20% of the total content of compatible solutes, relying solely on betaine (Galinski and Herzog, 1990). Possibly, N-containing compounds such as betaine are essential for survival in extremely high osmolalities, because they have a higher potential to keep proteins in a native state (Youxing et al., 1998). In addition to nitrogen availability, the osmolyte pool was influenced by the kind of sugar added as a carbon source. Maltose-grown cells exhibited an almost threefold increase in trehalose content when compared with sucrose-grown cells, with an accompanying decrease in cytoplasmic proline. This observed result cannot result from specific pathways being activated by the kind of carbon source because, in all cases, trehalose synthesis was predominantly catalysed by the TreYZ pathway (unpublished data). The composition of the osmolyte pool largely depends on environmental conditions, which include type of carbon source and availability of nitrogen. Thus, osmoregulation by compatible solute synthesis does not follow a rigid scheme. Rather, it responds to a variety of parameters sensed by the cell in addition to osmolality, and transduces these signals to the internal osmoregulation network. The molecular sensing and signal transduction mechanisms are currently not known. However, a regulatory mechanism co-ordinating trehalose and proline synthesis must be postulated.

In contrast to other well-studied organisms (Kaasen et al., 1994), osmotically regulated trehalose synthesis in C. glutamicum is independent of the OtsAB pathway and almost entirely TreYZ mediated. The TreYZ pathway has not yet been implicated in osmoregulation in other organisms. Using this pathway for stress adaptation seems reasonable as maltodextrins (e.g. typical bacterial storage carbohydrates such as glycogen) represent rapidly accessible substrates that do not require an uptake mechanism. Such a transport mechanism might be hampered after hyperosmotic shock (Wood, 1999).

As TreYZ was found to be sufficient for trehalose synthesis in response to osmotic stress under virtually all conditions tested, the presence of three trehalose synthesis pathways in C. glutamicum seems paradoxical. The results obtained indicate that TreS should be considered as a substitute for a trehalase in the cellular context. Several observations favour this suggestion: (i) no open reading frame homologous to known trehalase genes is, as yet, identified in the genome of C. glutamicum; (ii) the deletion of treS in the genome of C. glutamicum leads to increased cytoplasmic trehalose levels; (iii) the treS transcript can be found independent of whether TreS-mediated trehalose synthesis is detectable or not; (iv) no trehalose-degrading activity was measured in crude extracts of CglΔotsAΔtreSΔtreY; and (v) TreS catalyses maltose synthesis from trehalose (as documented in an in vitro assay). The preferred conversion in vivo is likely to be trehalose-to-maltose as trehalose, compared with maltose, is found in 1000-fold higher concentrations in C. glutamicum cells. After stress adaptation, trehalose recycling through transformation to maltose may be of particular importance, as CglΔtreS cells suffered from a prolonged lag phase after shock until the trehalose concentration was decreased again. A similar observation was made with a trehalase-deficient yeast strain unable to degrade trehalose after heat shock. Like the C. glutamicum strain CglΔtreS, the trehalase-deficient yeast strain was severely impaired in its ability to recover from stress (Nwaka et al., 1995a,b), indicating that trehalose degradation may be fundamental for stress recovery. In the case of heat stress adaptation in yeast, this hypothesis was confirmed recently by Singer and Lindquist (1998). The surprising effect that trehalose amounts decreased after osmotic upshock, even when TreS was missing, can be explained by the complex trehalose metabolism in C. glutamicum. Trehalose acts as a compatible solute but, in addition, it is diluted in the cytoplasm as a result of (i) integration into the mycolic acid layer of the cell wall; (ii) a dilution effect when cells begin to grow again; and (iii) its excretion into the medium depending on the external osmolality (not shown).

In contrast to TreS, the role of the OtsAB pathway, the most prominent of the stress-regulated pathways in other organisms, remains obscure in the case of C. glutamicum. The OtsAB pathway exhibited low basic activity and did not significantly affect the cell's trehalose pool under any condition applied, nor did the lack of otsA greatly reduce cellular trehalose levels. Shimakata and Minatogawa (2000) ascribed a central role to this pathway in the context of mycolic acid synthesis of Corynebacterium matruchotii. Using an in vitro mycolic acid synthesis system, Shimakata and Minatogawa (2000) found that the conversion of corynomycolate to trehalosyl mycolates and free mycolate is dependent on trehalose-6-phosphate, the product of the OtsA reaction. However, the presence of trehalose in cell wall lipids, i.e. in trehalosyl mycolates, also in the otsA-deficient C. glutamicum strain contradicts this hypothesis. CglΔotsA possessed ≈ 75% of the amount detected in the wild type. The reduction indicates that OtsAB may, in fact, have some effect on trehalosyl mycolate synthesis. However, this is presumably not mediated via trehalose-6-phosphate and, therefore, OtsAB and TreYZ together could mediate trehalosyl mycolate synthesis. These results show that: (i) the physiological function of the OtsAB pathway remains unresolved; and (ii) TreYZ is indeed involved in the trehalosyl mycolate synthesis.

In summary, we suggest that C. glutamicum, like other bacteria, possesses one true trehalose synthesis pathway (TreYZ) and two additional pathways with low trehalose-synthesizing capacity. TreS apparently compensates for the absence of a classical trehalase by degrading internal trehalose to maltose when present in large amounts, thereby making possible a recycling of trehalose as a carbon source if the stress has passed. The physiological role of the OtsAB pathway needs further investigation because it is not essential for mycolic acid synthesis. It seems more likely that this pathway is important for stress response under particular physiological conditions not tested so far. For example, consider the simultaneous occurrence of osmotic stress and carbon limitation. Such a situation would rule out TreYZ activity because of the absence of glycogen, the substrate of TreYZ. In agreement with a stress-related function of OtsAB is the significant reaction of this pathway to hyperosmotic stress at the transcriptional level and the weak reaction in terms of trehalose synthesis under carbon-excess conditions. The metabolic and regulatory connection between trehalose metabolism and cell wall biosynthesis is complex and requires further detailed examination. It should be mentioned that the growth inhibition observed in strains CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY, both of which are devoid of trehalosyl mycolates, may thus well be a consequence of an altered cell wall, as those strains obviously differed from all other strains in the physical properties of their cell surface and in the excretion of cytoplasmic metabolites.

Experimental procedures

Strains and culture conditions

All cultivations were performed in shake flasks under aerobic conditions. Escherichia coli strains were kept at 37°C and C. glutamicum at 30°C. In both cases, LB medium was used as complex medium. CgXII medium (Keilhauer et al., 1993) was used as minimal medium for C. glutamicum. Sucrose was routinely used as carbon source except when indicated. The concentration of all carbon sources was 4%. High osmolality was adjusted by the addition of 750 mM NaCl (equivalent to an osmotic upshift of 1.5 osM) and checked by means of freezing point reduction with an Osmomat 030 (Gonotec). In all upshift experiments, sufficient nitrogen was present, and an osmotic upshift under nitrogen-limiting conditions was also performed (see below). For all strain characterizations, cells from LB precultures were washed once in phosphate-buffered saline (PBS) (Sambrook et al., 1989) and used to inoculate CgXII precultures, which were grown to exponential or early stationary phase. From those cultures, the CgXII main cultures were inoculated to an initial optical density (OD600) of 0.5–1. If the main culture differed from the preculture in any parameter, e.g. carbon source, nitrogen source, NaCl content, a washing step between pre- and main culture was included.

Construction of strains defective for trehalose synthesis genes

The coding regions of otsA, treY and treS flanked by upstream and downstream stretches of 200–500 bp were amplified by polymerase chain reaction (PCR) using genomic DNA as template and the oligonucleotides listed in Table 2. The amplified DNA fragments were ligated with SmaI-linearized pUC vectors (Yanisch-Perron et al., 1985). Ligation products were transformed in E. coli DH5Δmcr cells (Grant et al., 1990; Inoue et al., 1990). The resulting plasmids were designated pUC18otsA, pUC18treY and pUC19treS. To obtain deletion alleles, internal fragments of 200–700 bp were excised from the genes harboured in pUC18/19 using suitable restriction endonucleases (HpaI–PflM for otsA, AscI–AflII for treY and BssHII–Tth111I for treS). If necessary, blunt ends were created before the vectors were religated. To transfer the deletion alleles into the integration vector pK19mobsacB (Schäfer et al., 1994), the fragments were excised from pUC vectors, treated with Klenow and ligated with SmaI-linearized pK19mobsacB (Table 3). The obtained constructs were tested by restriction analysis.

Table 2. . Oligonucleotides used in this study.
DesignationOligonucleotide sequence
Size of amplified
fragment (kb)
otsAs5′-tct gcc agt gga tat gac tgt cc-3′1.808
otsAa5′-cgt tga cgt cgt ggg tat aga cc-3′ 
treYs5′-gca cgt cca att tcc gca ac-3′2.739
treYa5′-tca aaa ctc act atc ggg tac-3′ 
treSs5′-ggc ctg gag aat tcg gat acc-3′2.541
treSa5′-gct cca cat cgg ggt ttt gcc-3′ 
otsAgs5′-atg gat gat tcc aat agc ttt-3′0.386
otsAioa5′-ccg tgt gcc gcc act tgg-3′ 
treYgs5′-atg gca cgt cca att tcc gca-3′0.479
treYsoa5′-cca tca agc tcc gcg aat tcc-3′ 
treZgs5′-atg ctc aaa gac ttg acc ggc-3′0.494
treZsoa5′-gtc acg ccg agg tcg cgc-3′ 
treSgs5′-atg act gat acc tct ccg ttg-3′0.489
treSia5′-gga ttc ttg gaa cca tgc gtg-3′ 
Table 3.  Plasmids used in this study.
pUC18/19plac, ApRYanisch-Perron et al. (1985)
pUC18otsApUC18 carrying a 1.8 kb fragment containing the otsA ORFThis study
pUC18treYpUC18 carrying a 2.7 kb fragment containing the treY ORFThis study
pUC19treSpUC19 carrying a 2.5 kb fragment containing the treS ORFThis study
pK19mobsacBori pUC, KmR, mob sacBSchäfer et al. (1994)
pK19mobsacBΔotsApK19mobsacB carrying ΔotsAThis study
pK19mobsacBΔtreYpK19mobsacB carrying ΔtreYThis study
pK19mobsacBΔtreSpK19mobsacB carrying ΔtreSThis study
pDrive Qiagen
pDriveotsApDrive containing an intragenic 0.39 kb otsA fragmentThis study
pDrivetreYpDrive containing an intragenic 0.48 kb treY fragmentThis study
pDrivetreZpDrive containing an intragenic 0.49 kb treZ fragmentThis study
pDrivetreSpDrive containing an intragenic 0.49 kb treS fragmentThis study

In order to exchange the wild-type alleles for the deletion-harbouring alleles in the genome of C. glutamicum ATCC13032 (Abe et al., 1967), the method of allelic replacement described by Schäfer et al. (1994) was used. In short, competent cells were transformed with the pK19mobsacB derivatives by means of electroporation (Van der Rest et al., 1999). Plasmid integration in the genome was verified by selecting kanamycin-resistant and sucrose-sensitive colonies (the expression of the sacB gene encoding the levan sucrase is toxic in sucrose-containing media). To promote re-excision of the plasmid DNA, positive clones grown overnight in LB broth containing 7.5% glucose were plated on LB agar supplemented with 15% sucrose in different dilutions, usually between 10−3 and 10−5. Positive colonies obtained from the subsequent selection (KmS, SucR) were tested for allelic replacement by PCR. To generate strains carrying multiple gene deletions, single deletion strains were used instead of the wild type. CglΔotsAΔtreS and CglΔotsAΔtreY were constructed from CglΔotsA, CglΔtreYΔtreS from CglΔtreY and CglΔotsAΔtreSΔtreY from CglΔotsAΔtreS. All strains are listed in Table 1.

Table 1. . Bacterial strains used in this study.
E. coli
 DH5α-mcrsupE44hsdR17recA1endA1gyrA96thi1Grant et al. (1990)
C. glutamicum
 ATCC13032Wild typeAbe et al. (1967)
 CglΔotsAWild type carrying a chromosomal deletion in the otsA ORFThis study
 CglΔtreYWild type carrying a chromosomal deletion in the treY ORFThis study
 CglΔtreSWild type carrying a chromosomal deletion in the treS ORFThis study
 CglΔotsAΔtreYWild type carrying chromosomal deletions in the otsA and treY ORFsThis study
 CglΔotsAΔtreSWild type carrying chromosomal deletions in the otsA and treS ORFsThis study
 CglΔtreYΔtreSWild type carrying chromosomal deletions in the treY and treS ORFsThis study
 CglΔotsAΔtreSΔtreYWild type carrying chromosomal deletions in the otsA, treY and treS ORFsThis study

RNA hybridization experiments

For purification of RNA from C. glutamicum, cells were harvested in 1–5 ml samples depending on the cell density of the respective culture. Cells were sedimented by centrifugation at 30°C, suspended in buffer RA I (Macherey-Nagel) and broken by 300 mg glass beads (0.1–0.25 mm diameter) by two sequential passages of 30 s and 6.5 m s−1 in a FastPrep®120 instrument (Q-Biogene). RNA was purified from the supernatant with the Nucleospin® RNAII kit (Macherey-Nagel) according to the manufacturer's instructions. For the construction of RNA antisense probes, intragenic otsA, treY, treZ and treS fragments of a size of roughly 500 bp were amplified using ATCC13032 cells as template and the oligonucleotides shown in Table 2. The generated DNA fragments were ligated with pDrive (Qiagen) (Table 3). DIG-11dUTP-labelled antisense RNA was obtained by in vitro transcription (1 h, 37°C) from XbaI-linearized vectors using T7 polymerase. Analysis of gene transcription was monitored by RNA hybridization experiments using DIG-labelled antisense RNA probes. For that purpose, ≈ 6 µg of RNA was transferred to a nylon membrane using a Minifold Dot Blotter (Schleicher and Schuell). RNA was bound to the membrane by careful vacuum suction (15 mbar). RNA was cross-linked to the membrane by means of ultraviolet irradiation at 125 J cm−2. Hybridization and detection steps were carried out according to the DIG Application Manual (Roche Molecular Biochemicals). Chemiluminescence was detected either via commercially available X-ray films or, in the case of a densitometric quantification, via the CCD camera of the LAS 1000 CH system (Fuji). The signals were quantified with the program AIDA IMAGE ELIZA 2.11.

Hyperosmotic shock at nitrogen limitation and nitrogen starvation

Cells were deprived of nitrogen by growth in CgXII minimal medium in which the nitrogen sources ammonium sulphate (150 mM) and urea (80 mM) were substituted by 100 mM glutamine. Glutamine uptake was limited by providing only trace amounts of sodium, the co-substrate for the transporter as demonstrated by Siewe et al. (1995). To this end, Na+ was excluded from all medium ingredients as well as from the PBS used for washing. For nitrogen starvation, CgXII medium without any nitrogen source was used. The upshock of the limitation, starvation and control culture was performed by the addition of 750 mM KCl instead of the routinely used 750 mM NaCl. Owing to the different medium composition in the case of N-limitation and N-starvation, the basal osmolality of the medium was not 0.9 osM but rather 0.5 osM. The addition of 750 mM KCl shifted the osmolality about 1.5 osM to 2.4 osM in the control medium (N-surplus) and to 2.0 osM in the N- starvation and N-limitation culture. In order not to transfer any nitrogen source or NaCl from the preculture medium, cells from the CgXII precultures were washed twice before the N-limited/starved cultures were inoculated. The latter culture was kept in the nitrogen-starved state for 3–4 h before the upshock was carried out with KCl.

Analysis of cytoplasmic metabolites

For the analysis of cytoplasmic metabolites, 1 ml samples were taken from cultures. Cells were separated from the growth medium by vacuum filtration using glass-fibre membranes (Millipore) pre-equilibrated in fresh CgXII medium. The cells were washed twice with 1 ml of fresh, isoosmotic CgXII medium. The filters were transferred to 1.5 ml reaction vessels, and the adsorbed cells were permeabilized by incubation in 0.1% CTAB for 5 min. Subsequently, the filter was removed, and the remaining cell debris was sedimented at 20 000 g and 4°C. Samples were stored at −20°C until use. Cytoplasmic amino acids were determined using a reverse phase high-performance liquid chromatography (HPLC) system (HP 1090 Liquid chromatograph, HP1046A fluorescence detector; Hewlett Packard) with automated fluorescent precolumn derivatization. Hypersil ODS 5 columns (CS Chromatographie Service) were used. Primary amino acids were derivatized by ortho-phthaldialdehyd/mercaptopropionic acid, proline by means of 9-fluorenyl-methyl-chloroformiate/acetonitrile. Solvent A contained 100 mM sodium actetate, pH 7.2, 0.5% tetrahydrofurane, and solvent B was a mixture of acetonitrile–methanol−100 mM Na-acetate pH 7.2 (2:2:1). A hydrophilic-to-hydrophobic solvent gradient was used ranging from 7% solvent B in the beginning to 100% solvent B at the end of a run. L-Ornithine was used as internal standard. Trehalose and other sugars were determined by split injection capillary gas chromatography with a GC9000 series gas chromatograph (Fisons Instruments) using a fused silica capillary column (FS-SE-54-0.25; CS Chromatographie Service). Injection and detection devices were kept at 280°C. Separation was achieved by a linear temperature gradient from 160°C to 280°C with a heating rate of 12°C min−1, starting at 2 min after injection. The final temperature was kept constant for 3 min for complete elution. Myo-inositol was used as internal and trehalose as external standard.

Assay for TreS activity

Cells of CglΔotsAΔtreY and CglΔotsAΔtreSΔtreY were grown to exponential phase, harvested, washed twice in PBS and suspended in 2 ml of 50 mM potassium phosphate (pH 7.5), 10 mM MgCl2, 40 µg ml−1 DNase I and Complete Protease Inhibitor (Roche Diagnostics) before disrupting the cells by three passages through an Aminco French® pressure cell at 1.2 × 108 Pa. Cell debris was removed by centrifugation at 4°C. The enzyme assays were carried out in 100 mM sodium phosphate (pH 6) according to the method of De Smet et al. (2000) with a maltose or trehalose concentration of 20 g l−1. The protein content in the assay was in the range of 15–18 mg ml−1.

Preparation of cell wall extract and isolation of trehalose from fatty acids

Cell wall of C. glutamicum was prepared according to the method of Puech et al. (2000). One or two grams of cells (wet weight) were suspended in 20 ml of 50 mM KiPO4, 2 mM MgCl2 (pH 7.5). After the addition of DNase I (2 µg ml−1 buffer), cells were disrupted by four French press passages at 1.2 × 108 Pa, followed by a low-speed centrifugation at 4°C to separate unbroken cells. Subsequently, the cell wall was sedimented (12 000 g, 45 min, 4°C), washed once in 10 ml of H2O and centrifuged again. Afterwards, the pellet was suspended in 2 ml of H2O and freeze-dried under vacuum. The saponification of fatty acids was carried out as described by Shimakata et al. (1986). Dried cell wall extract was dissolved in 500 mM KOH−50% ethanol (60–200 µl mg−1 cell wall extracts) in a sonification bath (30 min, 40°C). To start saponification, the extract was incubated for 30 min at 70°C before extraction in 1–2 ml of hexan/diethylether. The aqueous phase was isolated and neutralized with one volume of 1 M HCl and stored at −20°C. The quantification of liberated trehalose was done by means of gas chromatography.

Nucleotide sequence accession numbers

Nucleotide sequence accession numbers in the database ( are: Cgl2624 (otsA); Cgl2626 (otsB); Cgl2303 (treS); Cgl2118 (treY); and Cgl2125 (treZ).


We thank B. Bathe and B. Möckel (Degussa AG, Hanau, Germany) for providing the otsA, otsB, treS, treY and treZ sequences, and Sharon Taylor for critical reading of the manuscript.

Supplementary material

The following material is available from

Fig. S1. A. Trehalose content in the cytoplasm of growing wild-type cells in media with different nitrogen concentrations at a basic external osmolality of 0.9 osM.

B. Trehalose content in the cytoplasm of growing wild-type cells in minimal media supplemented with sucrose (asterisks), glucose (filled squares) or maltose (inverted open triangles) at a basic external osmolality of 0.9 osM.

Fig. S2. The deletion of treS in the genome of C. glutamicum leads to an increase in internal trehalose concentrations.

Fig. S3. In connection with the hypothesis that TreS acts as a trehalose-degrading enzyme, one has to postulate that, in addition, a maltosidase activity (α-glycosidase activity cleaving maltose) is present that would be necessary for the complete degradation of trehalose after TreS-mediated maltose formation.