Howard Hughes Medical Institute Fellow of the Life Sciences Research Foundation. To whom correspondence may be addressed. Present address: School of Chemistry, University of Melbourne, Parkville, Victoria, Australia 3054.
* This work was supported by National Institutes of Health Grant AI51622 (to C. R. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. S1). § Received a predoctoral fellowship from the National Science Foundation.
Mycobacteria contain high levels of the disaccharide trehalose in free form as well as within various immunologically relevant glycolipids such as cord factor and sulfolipid-1. By contrast, most bacteria use trehalose solely as a general osmoprotectant or thermoprotectant. Mycobacterium tuberculosis and Mycobacterium smegmatis possess three pathways for the synthesis of trehalose. Most bacteria possess only one trehalose biosynthesis pathway and do not elaborate the disaccharide into more complex metabolites, suggesting a distinct role for trehalose in mycobacteria. We disabled key enzymes required for each of the three pathways in M. smegmatis by allelic replacement. The resulting trehalose biosynthesis mutant was unable to proliferate and enter stationary phase unless supplemented with trehalose. At elevated temperatures, however, the mutant was unable to proliferate even in the presence of trehalose. Genetic complementation experiments showed that each of the three pathways was able to recover the mutant in the absence of trehalose, even at elevated temperatures. From a panel of trehalose analogs, only those with the native α,α-(1,1) anomeric stereochemistry rescued the mutant, whereas alternate stereoisomers and general osmo- and thermoprotectants were inactive. These findings suggest a dual role for trehalose as both a thermoprotectant and a precursor of critical cell wall metabolites.
Trehalose (α-d-glucopyranosyl-(1,1)-α-d-glucopyranoside, Fig. 1) is a non-reducing disaccharide that is abundant in mycobacteria in free form and in glycoconjugates that are found in the cytosol and the cell wall (
). Analogous studies in other organisms have suggested several purposes for the abundance of trehalose in mycobacteria. In Escherichia coli and Saccharomyces cerevisiae, trehalose is known to be critical for thermotolerance (
). Many of these compounds may be biosynthetic precursors of closely related glycolipids found in the cell wall of this and other mycobacteria. Finally, there is a series of phosphorylated mannopyranosyltrehaloses of unknown function found in Mycobacterium bovis BCG (
). The most widespread trehalose-containing glycolipids are trehalose 6-mycolate (trehalose monomycolate) and trehalose 6,6′-dimycolate (trehalose dimycolate, Fig. 1A). These compounds are loosely associated with the cell wall and may be removed by extraction or simple mechanical treatment (
). Trehalose monomycolate has been implicated as a carrier of mycolic acid from the site of synthesis in the cytosol to the acceptor arabinogalactan in the cell wall and is ubiquitous among mycobacteria. Trehalose dimycolate has been detected in all mycobacteria with the exception of Mycobacterium leprae (
). otsA encodes trehalose-6-phosphate synthase, which catalyzes the condensation of glucose 6-phosphate and UDP-glucose to form trehalose 6-phosphate. Trehalose 6-phosphate is subsequently dephosphorylated by the action of OtsB, trehalose-6-phosphate phosphatase, to afford free trehalose. OtsA and OtsB have been studied in detail in both bacteria and yeast (
), α-(1,4)-glucose polymers are converted to trehalose by isomerizing the terminal α-(1,4)-linkage to an α,α-(1,1)-linkage. This is accomplished by the product of the treY gene, maltooligosaccharyltrehalose synthase. The trehalose disaccharide is subsequently hydrolyzed from the polymer by the product of the treZ gene, maltooligosaccharyltrehalose trehalohydrolase. In the third trehalose synthesis pathway, also found in Pimelobacter (
), the product of the treS gene isomerizes the α-(1,4)-linkage of maltose directly to the α,α-(1,1)-linkage of trehalose. Interestingly, the slow growing mycobacterium M. leprae possesses only two routes for trehalose biosynthesis, the OtsA-OtsB and TreS pathways. M. leprae does possess regions homologous to TreY and TreZ, but these are interrupted with frame shifts and stop codons (
Despite the widespread presence of trehalose in mycobacteria, little is known about the biological functions of this disaccharide and its conjugates. The presence of multiple biosynthetic pathways suggests a critical role for trehalose in mycobacteria. Indeed, Corynebacterium glutamicum, which also has multiple pathways, appears to be growth impaired when trehalose synthesis is completely disrupted (
). To investigate the importance of trehalose and trehalose-containing metabolites, we constructed a mutant strain of M. smegmatis lacking all three pathways. We demonstrate that trehalose is vital both for growth and for entry into stationary phase and is not simply an osmo/thermoprotectant as in other bacteria.
General—Pfu DNA polymerase was obtained from Stratagene. Restriction enzymes were from New England Biolabs or Amersham Biosciences. Calf intestinal alkaline phosphatase was from Amersham Biosciences. Unnatural isomers of trehalose (α,β and β,β) were from Sigma. The xylosyl (α-d-glucopyranosyl-(1,1)-α-d-xylopyranoside (Glc-(1,1-α,α)-Xyl)) and galactosyl (α-d-glucopyranosyl-(1,1)-α-d-galactopyranoside (Glc-(1,1-α,α)-Gal)) trehalose analogs were generated by chemical synthesis as reported previously (
). Plasmid miniprep kits and the QIAquick kit for DNA extraction from agarose gels were from Qiagen. T4 DNA ligase was from New England Biolabs. Plasmids used in this work are shown in Table I, and oligonucleotide primers are shown in Table II. Preliminary sequence data for M. smegmatis were obtained from The Institute for Genomic Research website at www.tigr.org. For E. coli, antibiotic concentrations used were the following: ampicillin, 100 mg liter–1; kanamycin, 50 mg liter–1; hygromycin, 200 mg liter–1; spectinomycin, 20 mg liter–1. For M. smegmatis, antibiotic concentrations used were the following: kanamycin, 20 mg liter–1; hygromycin, 50 mg liter–1; spectinomycin, 20 mg liter–1.
Table IBacterial strains and plasmids used in this study
). Oligonucleotide primers were used to amplify 1.6–2-kb regions upstream and downstream of the desired gene. The three delivery vectors contained deletions of the central portions of each gene. In the case of OtsA and TreS, the central portions of the genes were replaced with a hygromycin resistance cassette and a spectinomycin resistance cassette (from pUC19Ω), respectively. In the case of TreY, an in-frame deletion of the middle portion of the treY gene was constructed to generate an unmarked mutant. The upstream regions of OtsA, TreS, and TreY were generated using the primer pairs: OtsA5′,5′f and OtsA3′,5′f; TreS5′,5′f and TreS3′,5′f; TreY5′,5′f and TreY3′,5′f (Table II), which generated PmlI/HindIII, HindIII/PstI, and HindIII/KpnI fragments, respectively. The downstream regions were generated using primer pairs OtsA5′,3′f and OtsA3′,3′f; TreS5′,3′f and TreS3′,3′f; and TreY5′,3′f and TreY3′,3′f (Table II), which generated BamHI/NotI, BamHI/BamHI, and KpnI/PacI fragments, respectively.
The hygromycin resistance marker was inserted between the two OtsA fragments into the KpnI restriction site, and the spectinomycin resistance marker was inserted between the two TreS fragments into the BamHI restriction site. The final delivery vectors, p2NIL_MsOtsA, p2NIL_MsTreS, and p2NIL_MsTreY were generated by adding the PacI cassette (PAg85-lacZ Phsp60-sacB) from pGOAL17 to the vector bearing the mutated allele. The triple deletion mutant strain was prepared in three steps by sequentially transforming M. smegmatis mc2155 with the delivery vectors. In the first step the delivery vector p2NIL_MsOtsA was pretreated with UV light (120 mJ cm–2) and used to electroporate M. smegmatis mc2155. The deletion mutant was selected as described previously to afford mc2155::OtsA (
). In the second step this strain was transformed with UV light-treated p2NIL_TreS except with spectinomycin in place of hygromycin. After selection the genotype was confirmed as described below, to afford the double mutant containing deletions of OtsA and TreS (mc2155::OtsA/TreS). In the final step this strain was transformed with UV light-treated p2NIL_TreY, and transformants were selected for several rounds in the presence of 1 mm trehalose. The transformants were subjected to genotypic analysis to afford mc2155::OtsA/TreS/TreY.
Genotypic Analysis—Genomic DNA was prepared from M. smegmatis by standard methods. Southern blotting analysis was carried using the DIG detection kit (Roche Applied Science). Hybridization probes were generated by PCR using the following pairs of primers: for mc2155::otsA, 5′ probe (OtsAForScreen and OtsA3,5f) and 3′ probe (OtsA3′flank3′end and OtsaRevScreen); for mc2155::otsA treS,5′ probe (TreSScreenFor and TreS3,5f) and 3′ probe (TreS3For and TreSScreen-Rev); and for mc2155::otsA treS treY, 5′ probe (TreY5ProbeFor and TreY5ProbeRev) and 3′ probe (TreY3ProbeFor and TreY3ProbeRev) (Table II) with Expand polymerase (Roche Applied Science) using DIG-labeled dNTPs; one probe was specific for the upstream region of the gene (5′ probe), and one probe was specific for the downstream region (3′ probe). Genomic DNA was digested with the restriction enzymes as shown in the legend to Fig. 4, which generated unique bands for the wild type and mutant strains. The digested genomic DNA was separated by electrophoresis on two 0.8% agarose gels (20 V, 12 h), and each was transferred to a HyBond-N+ membrane (Amersham Biosciences) by capillary action. After cross-linking the DNA to the membrane (UV light, 120 mJ), each blot was hybridized (68 °C) with DIG-labeled probes specific for regions upstream and downstream of the interrupted gene. After washing and blocking, the hybridized bands were detected using an anti-DIG-antibody-alkaline phosphatase fusion and 3-bromo-4-chloroindolyl phosphate as substrate.
Identification of M. smegmatis Homologs of OtsA, TreS, and TreY and Design of Trehalose Biosynthesis Mutant—To disrupt the trehalose biosynthesis pathways in M. smegmatis, we chose to target the enzymes OtsA, TreY, and TreS. For the OtsA-OtsB pathway, we elected to delete the gene for the glycosyl-transferase OtsA, as a mutant strain lacking OtsB might exhibit a leaky phenotype due to the action of endogenous nonspecific phosphatases. Similarly, in the TreY-TreZ pathway, we targeted the transglycosylase TreY rather than the hydrolase TreZ as there was potential for compensation of the TreZ deletion by other α-glucosidases in the cell.
To confirm the presence of the three target genes in M. smegmatis, the sequences of OtsA, TreS, and TreY from M. tuberculosis were used as BLAST queries to search the unpublished M. smegmatis genome TIGR data base (www.tigr.org). Open reading frames corresponding to each gene were found in M. smegmatis. Additionally, homologs of the M. tuberculosis OtsB and TreZ genes were found, suggesting that these pathways for trehalose biosynthesis are intact in M. smegmatis.
Construction and Analysis of OtsA, TreS, and TreY Deletion Mutants—To investigate the importance of trehalose biosynthesis in M. smegmatis, we constructed a series of single, double, and triple deletion strains of genes in each of the pathways. The chromosomal deletion mutants were constructed in sequential order starting with OtsA, followed by TreS and then TreY, using the general method of Parish and Stoker (
). Southern analysis of genomic DNA was used to confirm the deletion of the genes. Fig. 3 shows the Southern blots for the OtsA single deletion strain, the OtsA/TreS double mutant, and the OtsA/TreS/TreY triple deletion mutant, hereafter denoted as the “Tre– mutant.”
The Tre–Mutant Is a Trehalose Auxotroph and Requires Trehalose for Entry into Stationary Phase—The Tre– mutant was tested for growth in liquid media in the presence and absence of trehalose. The requirement for trehalose was initially difficult to ascertain due to the fact that the mutant must be grown in trehalose-containing media prior to phenotypic analysis and appears to store trehalose in some form. Thus, to test the requirement of the Tre– mutant for trehalose, the bacteria were first cultured in a minimal amount of trehalose required for growth (5 μm) for 1 day to starve them of the disaccharide before monitoring growth in the absence of trehalose. As shown in Fig. 4A, the Tre– mutant is indeed a trehalose auxotroph.
The Tre– mutant also has a defect in its ability to enter stationary phase. Wild type and Tre– mutant strains were grown in the presence of 5 μm trehalose to starve them of trehalose and then used to inoculate fresh cultures in the presence or absence of trehalose. The strains were then plated at various dilutions over the next several days, and colony forming units (cfus) were quantified to assess the viability of the original culture. Wild type M. smegmatis showed an increase in cfu/ml through the exponential phase, which plateaus as the bacteria enter the stationary phase. Fig. 4B shows that while the wild type and supplemented Tre– mutant cells maintain their viability, the viability of the unsupplemented mutant drops off precipitously after 1 day.
Three Pathways for Trehalose Biosynthesis Are Functionally Redundant for Growth of M. smegmatis in Culture—Complementation vectors were constructed in the plasmid pMS3GS using the M. tuberculosis otsA, treS, and treY genes. This kanamycin-resistant vector contains the promoter region for the M. tuberculosis glutamine synthetase and allows for the constitutive expression of proteins cloned downstream of the promoter region. A complete description of this vector can be found in (
). Each complementation vector was independently transformed into the Tre– mutant, and transformants were grown on kanamycin-containing medium in the absence of trehalose. Each gene independently complemented the Tre– mutant to trehalose prototrophy (Fig. 5). In addition, all the single pathway mutants and all possible combinations of the double mutants were tested for growth. All strains showed trehalose-independent growth, and the growth rates of all of these strains was comparable with wild type bacteria (not shown). The level of free trehalose in each single pathway mutant was measured using a procedure similar to that reported by Elbein and Mitchell (
). No significant differences were observed compared with wild type M. smegmatis (data not shown). We also labeled the single mutants and the Tre– mutant (grown in the presence of 20 μm trehalose as required) with [14C]glucose and characterized the gross metabolite profile by two-dimensional TLC. The major labeled species were similar in all of the strains (see Fig. S1 in supplemental material).
The Tre–Mutant Displays a Temperature-sensitive Pheno-type Even When Supplemented with Trehalose—Some organisms up-regulate trehalose during thermal stress to prevent deleterious consequences such as protein aggregation. When trehalose biosynthesis is disrupted, organisms such as E. coli and S. cerevisiae show reduced viability under conditions of thermal stress (
). Interestingly, adding exogenous trehalose does not recover these mutants, suggesting that exogenous trehalose cannot be imported. To address the effects of trehalose biosynthesis disruption on thermal tolerance, we incubated the Tre– mutant at either 43 or 18 °C for 96 h and monitored growth by turbidity. Consistent with previous observations in other organisms, the Tre– mutant demonstrated severely reduced viability at these non-optimal temperatures, even when supplemented with concentrations of trehalose that promote its growth under optimal conditions (Fig. 6). Mutants expressing only one of the three trehalose biosynthesis pathways had similar viability to wild type cells. These results suggest that thermoprotection can only be achieved by endogenously biosynthesized trehalose, while recovery of the Tre– mutant under optimal conditions can be achieved by exogenously supplied trehalose.
Rescue of the Tre–Mutant with Trehalose Analogs—To investigate the importance of trehalose's structure for rescue of the Tre– mutant, we assayed growth in the presence of trehalose analogs including the commercially available α,β- and β,β-trehalose isomers and synthetic analogs in which one glucose residue was replaced with either xylose or galactose (
) (Fig. 7). As shown in Fig. 8A, both α,β- and β,β-trehaloses could not rescue the mutant at concentrations as high as 500 μm, indicating that the α,α-(1,1) linkage is critical to the function of trehalose in M. smegmatis. The xylosyl (Glc-(1,1-α,α)-Xyl) and galactosyl (Glc-(1,1-α,α)-Gal) trehalose analogs were both able to recover the mutant at concentrations of 500 μm (Fig. 8B) but with slightly delayed growth kinetics (not shown). For the xylosyl analog, concentrations as low as 10 μm recovered the mutant as well as native trehalose, whereas the galactosyl analog was inactive at this concentration. Taken together, these data show that structural changes to native trehalose affect its ability to rescue the Tre– mutant.
The results shown here demonstrate that trehalose biosynthesis is essential for the proliferation of M. smegmatis. In the absence of trehalose, the Tre– mutant shows severely impaired growth. However, upon supplementation with trehalose, the mutant strain grows similar to wild type. Independent restoration of intact pathways for trehalose biosynthesis, namely the OtsA-OtsB, TreY-TreZ, or TreS pathways, by the use of complementation plasmids bearing the M. tuberculosis otsA, treS, or treY genes, converts the M. smegmatis Tre– mutant from trehalose auxotrophy to prototrophy. These data suggest that the three pathways for trehalose biosynthesis in M. smegmatis are mutually redundant for growth in the exponential phase.
Of the three pathways in mycobacteria, the OtsA-OtsB pathway is the one shared by a majority of prokaryotes. Furthermore, the substrates used in this pathway, glucose 6-phosphate and UDP-glucose, are simpler and likely to be more abundant than the substrates of the other two pathways. These features invite speculation that the OtsA-OtsB pathway may be responsible for a majority of trehalose synthesis. However, the level of free trehalose we measured for the OtsA– mutant was comparable with wild type M. smegmatis. Two other groups have disrupted trehalose production in C. glutamicum. While Morbach and co-workers observed similar results (
). Thus it appears that mycobacteria can regulate trehalose levels in the absence of either OtsA or TreY. Both groups found that disruption of OtsA and TreY or TreZ in C. glutamicum, leaving the TreS pathway as the sole source of trehalose, produced a phenotype similar to the triple knock-out. In stark contrast, we observed that TreS alone, expressed from a complementation plasmid or from the endogenous gene, can support growth of M. smegmatis at wild type levels. This discrepancy may reflect differences in expression levels of TreS between corynebacteria and mycobacteria or abundance of the substrate, maltose.
After entry into stationary phase, wild type M. smegmatis is viable for an extended period of time, with some cells surviving for periods up to two years (
). By contrast, the Tre– mutant rapidly loses viability after only 1 day in the absence of trehalose, showing a major impairment in stationary phase survival. This defect can be overcome by restoration of any one trehalose biosynthesis pathway. Mechanisms of stationary phase survival of mycobacteria have attracted considerable attention due to the importance of asymptomatic infection in tuberculosis. These data indicate that a supply of trehalose is essential for stationary phase survival in M. smegmatis and is likely to be significant for M. tuberculosis as well.
At this stage the precise cause of trehalose auxotrophy in the Tre– mutant in unknown. Our data show that a functional trehalose biosynthesis pathway is required for thermoprotection, while exogenous trehalose does not suffice. In this case, trehalose is biosynthesized in the cytosol and can further be transported outside the bacterium for incorporation into cell wall metabolites. Additionally, M. smegmatis cannot use trehalose as a sole carbon source (data not shown) even though it possesses a cytosolic trehalase (based on genome analysis). Taken together, these results suggest that exogenous trehalose cannot gain entry into the cell. When grown under optimal conditions, the Tre– mutant does not require intracellular trehalose but still requires exogenous trehalose, presumably for critical secondary metabolites generated in the cell wall. As mentioned previously, mycobacteria produce a wide variety of trehalose conjugates and a subset of these are likely required for normal cell growth. Our studies with trehalose analogs support this notion. Derivatives with non-native glycosidic bond stereochemistry were unable to rescue the Tre– mutant despite their physical similarity to trehalose. Analogs in which a single functional group was perturbed were capable of restoring growth, although less efficiently than native trehalose. This is consistent with a scenario in which trehalose is enzymatically modified to more elaborated species that are required for cell proliferation.
Trehalose-containing metabolites can differ from species to species; however, both trehalose and trehalose monomycolate are common to all species of mycobacteria that have been investigated. Similarly, trehalose dimycolate has been detected in all mycobacteria with the exception of M. leprae. Thus, it is likely that one or more of these common compounds are the agents that are required for the survival of M. smegmatis. Other groups have reported disruptions of one (
) of the three mycolyl transferases involved the formation of trehalose dimycolate, but there has been no report of a functional deletion of trehalose dimycolate until this work. Interestingly, the Tre– mutant could not be rescued by trehalose dimycolate itself (data not shown), ruling out this compound alone, although it may be required in combination with other trehalose-containing metabolites.
An intriguing possibility is that trehalose is essential for cell wall biosynthesis in mycobacteria. It has been suggested that trehalose acts as a carrier for mycolic acid, transporting it from the point of synthesis in the cytosol to the cell wall where it is appended to the growing arabinogalactan chain (
). If so, then the trehalose auxotrophy of the Tre– mutant could be explained by the inability of this strain to construct a functional cell wall. Notably, demonstration of the true carrier of mycolic acid for cell wall biosynthesis has remained elusive. It has also been suggested that polyprenol mannose 1-phosphates may act as carriers for mycolic acid based on the isolation of 6-mycolyl derivatives (
). If trehalose is acting as a mycolic acid carrier, this provides an explanation for the loss of viability in stationary phase. Recent work on M. smegmatis has shown that stationary phase cultures do not consist of dormant quiescent cells but rather a dynamic population of cells undergoing active growth and cell division (
). In this case, active transport of mycolic acid is expected to be essential for the construction of new cell walls.
Interestingly, while most bacteria, plants, and insects synthesize trehalose, higher order vertebrates such as mammals do not. Mammals are capable of degrading trehalose through the action of trehalase, an enzyme present in the brush border of the small intestine and the kidney proximal tubule, which catabolizes trehalose into its glucose monomers (
). However, our work suggests that the three pathways are redundant, at least in culture, rendering the trehalose biosynthesis system a complex target for drug development. It would be interesting to study the effects of disruptions in each of the pathways in a virulence model to clarify this point.