OUP user menu

Functional analysis of a lipid galactosyltransferase synthesizing the major envelope lipid in the Lyme disease spirochete Borrelia burgdorferi

Yngve Östberg , Stefan Berg , Pär Comstedt , Åke Wieslander , Sven Bergström
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00728.x 22-29 First published online: 1 July 2007

Abstract

One of the major lipids in the membranes of Borrelia burgdorferi is monogalactosyl diacylglycerol (MGalDAG), a glycolipid recently shown to carry antigenic potency. Herein, it is shown that the gene mgs (TIGR designation bb0454) of B. burgdorferi encodes for the protein bbMGS that, when expressed in Escherichia coli, catalyzes the glycosylation of 1,2-diacylglycerol with specificity for the donor substrate UDP-Gal yielding MGalDAG. Related lipid enzymes were found in many Gram-positive bacteria. The presence of this galactosyltransferase activity and synthesis of a cholesteryl galactoside by another enzyme were verified in B. burgdorferi cell extract. Besides MGalDAG, phosphatidylcholine, phosphatidylglycerol, and cholesterol were also found as major lipids in the cell envelope. The high isoelectric point of bbMGS and clustered basic residues in its amino acid sequence suggest that the enzyme interacts with acidic lipids in the plasma membrane, in agreement with strong enzymatic activation of bbMGS by phosphatidylglycerol. The membrane packing and immunological properties of MGalDAG are likely to be of great importance in vivo.

Keywords
  • Borrelia
  • membrane lipid
  • glycosyltransferase
  • glycosylation
  • monogalactosyl-1,2-diacylglycerol

Introduction

The Lyme disease spirochete Borrelia burgdorferi is transmitted by the bite of Ixodes ticks, and infection results in a wide range of clinical manifestations (Steere, 1989). Borrelia spp. are regarded as Gram-negative bacteria; however, borrelial cell envelope composition differs from other Gram-negative bacteria, as it lacks a typical lipopolysaccharide (Takayama et al., 1987) and phosphatidylethanolamine (Belisle et al., 1994). However, the presence of glycolipid antigens other than lipopolysaccharide has been detected (Cinco et al., 1991; Eiffert et al., 1991; Wheeler et al., 1993; Belisle et al., 1994; Honarvar et al., 1994). Polar lipid composition seems to be essentially identical in the outer and cytoplasmic membranes, and borrelial antisera react with B. burgdorferi membrane glycolipids (Belisle et al., 1994; Radolf et al., 1995; Hossain et al., 2001; Schröder et al., 2003).

Mass-spectroscopic analysis has identified α-monogalactosyl diacylglycerol (α-MGalDAG, 36.1%), phosphatidylcholine (11.3%), and phosphatidylglycerol (10.5%) as the major lipids in B. burgdorferi (Hossain et al., 2001). Also, a cholesteryl galactoside with an immunogenic motif has been discovered in the B. burgdorferi membranes (Ben-Menachem et al., 2003; Schröder et al., 2003). Recently, mouse natural killer cells were shown to recognize the α-MGalDAG lipid leading to immune activation, including antibody production (Kinjo et al., 2006). Correspondingly, membranes of Borrelia hermsii, the etiologic agent of North American tick-borne relapsing fever, contain MGalDAG, phosphatidylcholine, phosphatidylglycerol, cholesteryl glucoside, and an acylated cholesteryl glucoside (Livermore et al., 1978). Diglyceride-based glycolipids, like the major MGalDAG of the B. burgdorferi envelope, are most widespread in Gram-positive bacteria including mycoplasmas, but essentially absent among Gram-negative bacteria with the exception of many photosynthetic species (including plant chloroplasts) and some Pseudomonas species (Shaw, 1970). Understanding of lipid biosynthesis and function in B. burgdorferi is limited, but the presence and activities of two enzymes involved in synthesis of phosphatidylcholine and phosphatidylglycerol have been revealed (Wang et al., 2004). Here, the monogalactosyl-1,2-diacylglycerol synthase from B. burgdorferi (bbMGS) have been identified, cloned, and functionally verified. This enzyme catalyzes synthesis of the major membrane lipid MGalDAG. The characterized enzyme is closely related to many Gram-positive analogs.

Materials and methods

Bacterial strains and growth conditions

Borrelia strains used were noninfectious B. burgdorferi B31 (ATCC 35210) (Burgdorfer et al., 1982) and an infectious Ixodes tick isolate, B. burgdorferi N40 (Barthold et al., 1988, 1990). Borreliae were cultivated in vitro at 32°C in BSK-H (Sigma) (Östberg et al., 2002). Escherichia coli TOP10 (Invitrogen) was used as a cloning host and cultivated in Luria (L)-broth supplemented with 50 µg mL−1 carbenicillin when required, or on nutrient agar containing 100 µg mL−1 carbenicillin, X-gal, and IPTG when required. Expression of recombinant protein was induced by the addition of 0.5 mM IPTG at a cell density of OD600 nm=0.6, and after 4 h of incubation the cells were harvested by centrifugation.

PCR amplification and cloning

The oligonucleotides used in PCR for cloning of mgs (TIGR designation bb0454) were forward primer GAL1 (5′-AATAAATTATACTTAAGCTAGGAGGAG-3′) and reverse primer GAL2 (5′-ATTTGTTTGTGCCTTTTGATAAGCTG-3′). An in-frame stop codon was introduced into the forward primer (TAA in bold), to prevent synthesis of a fusion protein with vector-encoded β-galactosidase. The PCR product was purified by the High Pure™ PCR Product Purification Kit (Boehringer Mannheim) and ligated into an SmaI-digested pPCR-Script Amp SK(+) vector (Stratagene). Transformed E. coli was screened by blue-white color selection using X-gal and IPTG, and plasmids of positive (white) clones were purified with a QIAprep® Spin Miniprep kit (QIAGEN). The vector containing the cloned mgs gene was sequenced using a DYEnamic ET Terminator Cycle Sequencing Kit (Amersham Pharmacia Biotech), the vector primers M13 Universal and M13 Reverse (Amersham Pharmacia Biotech), and the internal primers GAL3 (5′-CAGACATCATTCATACTCACTCTG-3′) and GAL4 (5′-ATTACTGTCATTGGATATACTTCGC-3′).

Isolation of RNA and reverse transcription-polymerase chain reaction (RT-PCR)

Total mRNA was isolated from Borrelia by the Ultraspec-II RNA isolation system (Biotex Laboratories) and treated with DnaseI (Invitrogen). In the RT reaction, first-strand cDNA reactions were primed with 5 pmol of gene-specific primer GAL2 or GAL4. RNA (500 ng) were incubated for 5 min at 65°C, followed by quenching on ice before the addition of primer, 500 µM of each dNTP, RNase out (Invitrogen), 5 × reaction buffer, and 25 U of AMV reverse transcriptase (Roche). The samples were incubated for 2 min at 37°C, 50 min at 42°C, and 15 min at 70°C for cDNA synthesis. Using PCR, 5 µL of the first-strand cDNA reaction was amplified in a total volume of 50 µL containing 20 nM of each primer (GAL3 and GAL4), 200 µM of each dNTP, 50 mM KCl, 1.5 mM MgCl2, and 1 U Taq DNA polymerase in 20 mM Tris-HCl (pH 8.4).

Solubilization of cells and lipids

Solubilized whole-cell extracts were prepared by mixing c. 1010B. burgdorferi or E. coli cells (~10 mg of protein) in 0.5 mL assay buffer [110 mM HEPES pH 8.0, 22 mM MgCl2 and 22 mM 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS; Boehringer Mannheim)], followed by vortexing for 10 s every hour during incubation on ice for 3 h. Mixed micelle dispersions were prepared by mixing the specific lipids [dissolved in chloroform/methanol (2 : 1 v/v)]. After evaporation under reduced pressure, the lipids were solubilized to homogeneity in assay buffer and incubated overnight at 4°C. Synthetic rac-1,2-dioleoyl-glycerol (1,2-DOG) and cholesterol were purchased from Sigma and synthetic dioleoyl-phosphatidylcholine and dioleoyl-phosphatidylglycerol from Avanti Polar Lipids.

In vitro assay of glycolipid synthesis

In all in vitro assays, 25 µL solubilized cell extract was added to 20 µL mixed micelle dispersion and incubated on ice for 30 min. The enzymatic reaction was started by adding 5 µL UDP-[14C]galactose or UDP-[14C]glucose to a final concentration of 1 mM (30 GBq mol−1) in 50 µL. Supplemented substrate lipids (diacylglycerol or cholesterol) and activator/matrix lipids (phosphatidylglycerol/phosphatidylcholine) were used at a final concentration of 1 mM (20 GBq mol−1) and 9 mM, respectively. 14C-diacylglycerol was labeled on the glycerol moiety. After incubating for 30 min at 28°C, the reaction was terminated by adding 375 µL methanol/chloroform (2 : 1 v/v). Lipids were extracted (Karlsson et al., 1994), and separated by thin-layer chromatography (TLC) (0.2 mm Silica Gel 60 from Merck) developed with chloroform/methanol/water [80 : 25 : 4 (v/v) or 65 : 25 : 4 (v/v)]. Enzymatically synthesized products were visualized and quantified by electronic autoradiography (Packard Instant Imager™) and identified by appropriate reference lipids on the TLC plates. All assays were performed in duplicate.

The glycolipid products were also identified in a spray assay. In that case, lipid extracts were developed by TLC (as described above), sprayed with sulfuric acid/methanol [1 : 1 (v/v)], and incubated at 170°C. Lipids containing sugar moieties turned purple after c. 2 min.

Analysis of synthesized lipids

One milliliter of exponentially growing Borrelia culture was inoculated into 100 mL BSK-H containing 5 mCi L−1 33ortho-phosphate (Amersham Pharmacia Biotech), 0.2 mCi L−1 [4-14C] cholesterol (DUPONT®, NEN Products), 1 mCi L−1d-[U-14C] glucose (DUPONT®, NEN Products), and 0.5 mCi L−1d-[1-14C] galactose (Amersham Pharmacia Biotech). After incubation for 7 days at 32°C, the spirochetes were counted in a Petroff–Hauser chamber and harvested by centrifugation. The cells were washed once with PBS-Mg (5 mM), followed by resuspension in 5 mL chloroform/methanol (2 : 1). Samples were stirred for 1 h and centrifuged at 20 000 g for 15 min. Supernatants were transferred to new tubes and pellets were resuspended in 5 mL methanol, stirred for 30 min, and centrifuged at 20 000 g for 15 min. The new supernatants were mixed with the first and evaporated under a stream of nitrogen. Dry lipids were dissolved in 200 µL chloroform/methanol (2 : 1). These synthesized lipids were separated by TLC, developed in chloroform/methanol/water (65 : 25 : 4 (v/v)), and visualized by autoradiography and with Rhodamine 6G under UV light.

Sequence analysis and structure prediction

The amino acid sequence of bbMGS was used to search for homologs by PSI-blast (Altschul et al., 1997) in the databases at NCBI. Potential membrane-binding segments were analyzed by the membrane protein explorer (MPEx) (http://blanco.biomol.uci.edu/mpex/). Secondary structure prediction of the bbMGS amino acid sequence was performed with tools available at the ExPASy Molecular Biology Server.

Results

Identification of mgs

A PSI-blast search against the databases at NCBI, using the amino acid sequence of monoglucosyl diacylglycerol (MGlcDAG) synthase (alMGS) from Acholeplasma laidlawii (Berg et al., 2001) as a query, revealed mgs (TIGR designation bb0454), a B. burgdorferi B31 gene encoding a protein (bbMGS) with high sequence identity to the probe. The same database search also identified numerous potential proteins in other bacterial species with high similarity/identity to bbMGS (Table 1). A predominance of the related sequences were from Gram-positive bacteria but not surprisingly, the proteins with the highest similarities to bbMGS were found in two other Lyme disease species: Borrelia afzelii strain PKo and Borrelia garinii strain PBi. In addition, proteins with high similarity were found in the spirochetes Treponema denticola and Leptospira interrogans. However, no homolog to mgs was present in T. pallidum (Nichols strain), in agreement with the absence of MGlc-DAG in this species (Matthews et al., 1979). Among the identified homologs were also proteins from thermophiles, Archaea, and photosynthetic bacteria (data not shown). The bbMGS amino acid sequence, with 383 residues, has 29% identity to that of alMGS from A. laidlawii and is annotated as a potential lipopolysaccharide biosynthesis-related protein (TIGR home-page; Fraser et al., 1997). Furthermore, according to the protein domain database at NCBI, both alMGS and bbMGS belong to pfam00534, the glycosyltransferase group 1 that transfers NDP-linked sugars, like glucose, galactose, and mannose, to a variety of acceptor substrates, such as glycogen and lipopolysaccharides. Like alMGS and related sequences in Table 1, bbMGS contains a characteristic EX7E motif located in the C-terminal half of the protein (data not shown) and this motif is proposed to be involved in nucleotide-sugar binding and catalysis (Abdian et al., 2000).

View this table:
Table 1

Analogs of bbMGS in other bacteria according to a blast search analysis

SpeciesProteinExpectIdentical aa (%)Similar aa (%)
Borrelia afzeliiBAPKO_04772e-1749497
Borrelia gariniiBG04635e-1719296
Bacillus clausiiABC18529e-473054
Listeria innocuaLIN27005e-463050
Clostridium acetobutylicumCAC25368e-463453
Treponema denticolaTDE20344e-443355
Lactobacillus johnsoniiLJ17726e-443053
Listeria monocytogenesLMO25567e-432949
Streptococcus pyogenesSPy05166e-412951
Streptococcus agalactiaeSAG07105e-392950
Acholeplasma laidlawiialMGS2e-382950
Streptococcus pneumoniaeSP10768e-352748
Thermotoga maritimaTM07446e-332852
Leptospira interrogansLIC117521e-222546
  • Similar amino-acid sequences as identified by psi-blast at NCBI. Only selected hits are shown. Note the predominance of analogs in Gram-positive bacteria.

  • MGlcDAG glycosyltransferase function verified experimentally.

Hence, these sequence features and the relatedness to alMGS suggested that bbMGS functions as a lipid glycosyltransferase, using a sugar-nucleotide for the donation of a sugar to an acceptor lipid, potentially 1,2-diacylglycerol, as this is the backbone of all chemically characterized polar lipids in Borrelia spp.

Expression of bbMGS in E. coli

The recombinant protein bbMGS was expressed in E. coli and functionally characterized after solubilization of the cells (see Materials and Methods). As it was suspected that bbMGS may have catalytic activity similar to that of alMGS, the cell extract was supplemented with potential acceptor and donor substrates, with or without activator/matrix lipids (Fig. 1). The combination of 1,2-diacylglycerol and radiolabelled UDP-Glc yielded no new lipid product compared with the E. coli reference strain carrying empty vector (Fig. 1, lanes 1 and 2). However, replacing UDP-Glc with UDP-Gal gave rise to an additional lipid product, which migrated identically to that of the galactolipid MGalDAG (Fig. 1, lane 3). The corresponding result was observed when the radiolabelled isotope was switched from the donor substrate to diacylglycerol, the acceptor lipid (Fig. 2, lane 1). This lipid glycosylation was also influenced by the local lipid composition, i.e. the type of polar lipid present in the mixed-micelle aggregates. Nearly 10-fold higher activity was detected when the anionic phosphatidylglycerol was present in the assay mixture instead of the zwitterionic phosphatidylcholine (Fig. 1, lanes 3 and 5). As other similar lipid glycosyltransferases usually have a requirement for divalent cations as a cofactor, Mg2+ was added to all assays. Some bacterial lipid glycosyltransferases of similar structure can also yield dihexose- and trihexose-diacylglycerol products, in a processive manner, in addition to monohexose-diacylglycerol (Hölzl et al., 2005). However, this was not the case here (data not shown). In an attempt to find other possible acceptor substrates for bbMGS, diacylglycerol was replaced with cholesterol in the assays with the two NDP-sugar donors. However, no product corresponding to a possible cholesterylgalactoside could be detected (Fig. 1, lanes 6 and 7).

Figure 1

In vitro galactolipid synthesis in lysates from Escherichia coli carrying cloned mgs (lanes 1, 3 and 5–7) and a reference E. coli strain (lanes 2 and 4). Detergent-solubilized cells were supplemented with radiolabelled (+*) and nonradiolabelled (+) substrates, and matrix lipids. Enzymatic assay, followed by lipid extraction, and TLC analysis were performed as described in ‘Materials and methods’. As, application spot; MGalDAG, migration distance of synthetic and biological α- or β-MGalDAG/MGlcDAG reference lipids (Wieslander & Rosén, 2002).

Figure 2

Galactolipid synthesis in vitro in lysate from Escherichia coli carrying cloned mgs (ec) and whole cell lysate from B. burgdorferi (bb) cells. Solubilized cells were supplemented with radiolabelled (+*) and nonradiolabelled (+) substrates and matrix lipids. Enzymatic assay, lipid extraction, and TLC analysis were performed as described in ‘Materials and methods’. Phosphatidylglycerol was included due to its stimulatory effect on the recombinant enzyme (cf. Fig. 1). Lipid X: diacylglycerol (lanes 1–3), cholesterol (lanes 4–5, 8), unknown (lanes 6–7); CG, cholesteryl galactoside; As, application spot. Lane 8, reference 14C-cholesterol.

Thus, bbMGS catalyzed synthesis of MGalDAG when expressed in E. coli, and its activity was stimulated by the anionic lipid phosphatidylglycerol. It was therefore, concluded that bbMGS is the functional ortholog of alMGS.

Enzymatic activities and transcription of mgs in B. burgdorferi

To verify the functional activity of bbMGS in B. burgdorferi, solubilized spirochetal cells were used in enzymatic assays. The biosynthesis of MGalDAG in E. coli (Fig. 2, lane 1) could also be demonstrated in Borrelia cell extracts (Fig. 2, lane 2). Radiolabelled diacylglycerol was used in this experiment; however, identical results were obtained when the UDP-sugar was labelled (data not shown). Addition of different combinations of cholesterol and UDP-Gal/UDP-Glc to the cell-detergent extracts revealed synthesis of a cholesterylgalactoside (Fig. 2, lanes 4–8). As this lipid was not synthesized by recombinant bbMGS, it is assumed that this activity must reside in another enzyme (see Discussion); however, this was not investigated further.

To determine whether mgs is transcribed in both infectious and noninfectious Borrelia strains, RT-PCR was performed on mRNA purified from the infectious B. burgdorferi N40 and the noninfectious, high-passage B. burgdorferi B31. The results showed that equal amounts of mgs mRNA were detected in both strains (data not shown), suggesting that transcription and expression are independent of factors affecting infectivity.

Lipid composition

To compare in vitro and in vivo enzyme activities of the lipid biosynthesis, the lipid composition of B. burgdorferi membranes was determined by isotope labelling (Fig. 3). Four parallel cultures of B. burgdorferi were grown in media supplemented with 14C-cholesterol, 33P-ortho-phosphate, 14C-galactose, and 14C-glucose, respectively. Lipid extraction, followed by TLC analysis, showed that B. burgdorferi membranes contain two major phospholipids, phosphatidylcholine and phosphatidylglycerol, in addition to the glycolipid MGalDAG, and cholesterol, or derivatives thereof. Cultivation with 14C-cholesterol revealed significant incorporation of cholesterol into the membranes (Fig. 3, lane A). In addition, it gave rise to a putative sterol derivative at the same migration distance as the cholesterylgalactoside synthesized in vitro (Fig. 2). However, in comparison with the amount of radiolabelled cholesterol incorporated into the membranes, the fraction of this putative sterol derivative was very small (Fig. 3, lane A). No corresponding product was observed when B. burgdorferi was grown in the presence of 14C-galactose or 14C-glucose (Fig. 3, lanes C and D). Phosphatidylcholine and phosphatidylglycerol were the two strongest radiolabelled products (constituting >90% of the extracted lipids) when the growth medium contained 33P-ortho-phosphate (Fig. 3, lane B). Small quantities of other phospho-containing lipids may be present, e.g. polyisoprenoid-phosphate carriers for peptidoglycan synthesis. A major distinction in radioactive incorporation was observed between the two cultures supplemented with 14C-galactose and 14C-glucose; radiolabelled galactolipid was only detected when the growth medium contained 14C-glucose (Fig. 3, lane D). According to the extent of radiolabelled isotope used here, in combination with the fact that MGalDAG is a major lipid in B. burgdorferi membranes, this difference cannot be explained by low detection limits but rather by how these sugars are metabolized (see Discussion).

Figure 3

Radiolabelled lipid profiles after growth of Borrelia burgdorferi with 14C-cholesterol (lane A), 33P-ortho-phosphorus (lane B), 14C-galactose (lane C), and 14C-glucose (lane D). In vivo lipids were extracted from cells and developed on a TLC plate together with MGlcDAG, phosphatidylglycerol, and phosphatidylcholine reference lipids (not shown), as described in ‘Materials and methods’. Lipid X, cholesterol or derivative thereof (lane A) and unknown lipid (lanes C and D); MGlcDAG, monoglycosyl diacylglycerol; CG, proposed cholesteryl galactoside; As, application spot.

Sequence analysis of bbMGS

The number of basic amino acids in bbMGS is high, nearly 20% (71 Lys+Arg), while acidic residues constitute about 10% (38 Glu+Asp). This leads to a theoretical isoelectric point (pI) of c. 10. Despite charged residues being evenly distributed over the entire protein, clusters of basic charges can be observed and, interestingly, are predominantly found in predicted helical structures (data not shown). Neither a signal peptide nor strong hydrophobic segments for secretion or transmembrane localization were predicted for bbMGS, only several segments with affinity for a bilayer interface according to the MPEx prediction server (data not shown).

Discussion

The present study shows that recombinant bbMGS, when expressed in E. coli, catalyzes glycosylation of 1,2-diacylglycerol with specificity for the donor substrate UDP-Gal, yielding MGalDAG (Fig. 1). This work also verified the presence of active bbMGS in B. burgdorferi cells (Fig. 2). Thereby, bbMGS is a functional and structural ortholog of similar enzymes in the mycoplasma A. laidlawii, other spirochetes, and many Gram-positive species (Table 1). All proteins shown in Table 1 belong to family 4 in the CAZy glycosyltransferase systematics (Carbohydrate-Active Enzymes server, http://afmb.cnrs-mrs.fr/CAZY/index.html; Campbell et al., 1997). Proteins of this family appear to use a retaining mechanism in the catalysis of glycosylation as indicated by their conserved EX7E catalytic sequence motif, yielding an anomeric bond in α-configuration. In agreement with previous results showing that the α-MGalDAG is a major lipid of the B. burgdorferi envelope (Hossain et al., 2001), the present data strongly suggest that bbMGS is a retaining glycosyltransferase and catalyzes the synthesis of α-MGalDAG. According to the CAZy systematics, each of the three sequenced borrelial genomes of B. burgdorferi, B. afzelii, and B. garinii contains only four ORFs classified as glycosyltransferases, of which bbMGS is the first to be characterized. Two of the other candidates, annotated by TIGR as BB0732 and BB0767 in B. burgdorferi, are strong homologs to the in prokaryotes conserved penicillin-binding protein and MurG, respectively, proteins implicated in peptidoglycan biosynthesis. Interestingly, the remaining and fourth CAZy classified glycosyltransferase, BB0572 in B. burgdorferi, belongs to the inverting family 2 and is thereby a good candidate as a UDP-Gal-dependent galactosyltransferase involved in biosynthesis of the identified cholesterylgalactoside; the sugar residue in this latter glycolipid has its anomeric bond in β-configuration (Ben-Menachem et al., 2003; Schröder et al., 2003). In agreement with this classification, the in vitro enzyme assay verified the synthesis of a cholesterylgalactoside in B. burgdorferi cell-detergent extract (Fig. 2), while no such product could be detected in the assays expressing recombinant bbMGS (Fig. 1).

As in the MGlcDAG synthesis of A. laidlawii (Karlsson et al., 1994; Berg et al., 2001), anionic phospholipids appear to stimulate the activity of bbMGS in the MGalDAG synthesis. This is in agreement with the sequence analysis of bbMGS, which indicated that this protein is predominantly positively charged due to its high pI value, and should consequently be prone to bind negatively charged membrane surfaces. Generally, a high pI is typical of proteins electrostatically attached to negatively charged lipid membranes (Schwartz et al., 2001), and a key feature of lipid glycosyltransferases related to bbMGS (Lind et al., 2007).

The in vivo experiment presented here (Fig. 3) verifies the membrane lipid composition of B. burgdorferi reported previously by Hossain (2001). It also shows that radiolabelled phosphorus can be incorporated into the polar phosphatidylcholine head group of B. burgdorferi, in agreement with what was shown for the identified phosphatidylcholine synthase cloned into E. coli (Wang et al., 2004). A significant difference in this in vivo labelling of B. burgdorferi was that the formation of radiolabelled MGalDAG only occurred in growth medium supplemented with 14C-glucose and not with 14C-galactose (Fig. 3). This can be explained by a recent publication showing that B. burgdorferi is apparently unable to take up and metabolize galactose to support growth (von Lackum & Stevenson, 2005). The same authors connect this inability with a plausible lack of galactose transporter in the genome of B. burgdorferi. However, glucose has the ability to support B. burgdorferi growth (von Lackum & Stevenson, 2005) in accordance with three probable phosphotransferase system (PTS) glucose transporters suggested in the B. burgdorferi genome (Fraser et al., 1997). UDP-Gal is essential for the galactolipid synthesis, as shown here by the present in vitro experiments, and it is likely that B. burgdorferi has the ability to metabolically process glucose to UDP-Gal. The latter is supported by the fact that a gene (TIGR designation bb0444), encoding a potential epimerase for conversion of UDP-Glc to UDP-Gal, is present near mgs in the B. burgdorferi genome. Analogous features are valid for the small human pathogen Mycoplasma pneumoniae (Wieslander & Rosén, 2002). There are probably limitations of galactose for B. burgdorferi in its natural environment and instead glucose most likely constitutes a major energy source for this parasitic organism.

Diacylglycerol, the acceptor substrate for bbMGS, has been proposed to be synthesized from phosphatidic acid (PA) by a PA phosphatase in Bacillus subtilis (Krag et al., 1974). The same enzymatic activity has been characterized and cloned from Streptococcus pneumoniae (Edman et al., 2003). These PA phosphatases contain a characteristic motif of many lipid phosphatases (Stukey & Carman, 1997). A homologous sequence carrying this motif was found in the spirochete T. denticola, but not in B. burgdorferi, suggesting that B. burgdorferi has another pathway for diacylglycerol synthesis or uptake of serum lipids that supply diacylglycerol directly to the cell. The latter, however, is less likely, given the strong labeling of MGalDAG with radioactive fatty acids [e.g. (Radolf et al., 1995)]. Likewise, a similar phosphatidylglycerol-phosphate phosphatase gene for the phosphatidylglycerol pathway has not been identified in the B. burgdorferi genome (Fraser et al., 1997). The gene bb0119 in the B. burgdorferi genome is a potential cdsA homologue, encoding an enzyme (CdsA) that produces CDP-diacylglycerol from CTP and PA as suggested elsewhere (Fraser et al., 1997; Martinez-Morales et al., 2003; Wang et al., 2004). CDP-diacylglycerol is then utilized as a substrate by two enzymes, Pcs and Pgs, in the synthesis of the phospholipids phosphatidylcholine and phosphatidylglycerol in B. burgdorferi (Martinez-Morales et al., 2003; Wang et al., 2004). One could speculate that CDP-diacylglycerol also serves as a precursor for biosynthesis of diacylglycerol; however, to the best of the authors knowledge, enzymatic hydrolysis of CDP-diacylglycerol yielding diacylglycerol has thus far not been found to occur in nature.

The glycolipid diglycosyl diacylglycerol (DGDAG) is a major lipid in most Gram-positive bacteria, and α-linked variants of this lipid are present in many species containing bbMGS analogs (Table 1). However, DGDAG has not been found in either Spirochaeta or Treponema spp. (Livermore & Johnson, 1974), and was not detected in B. burgdorferi in this study. Accordingly, no homologs were identified in a database search against the complete B. burgdorferi genome, using the DGlcDAG synthase sequence from A. laidlawii (Edman et al., 2003) as a query.

In A. laidlawii, MGlcDAG plays an important role in maintaining certain packing properties in the plasma membrane. With its small headgroup and long, unsaturated chains, MGlcDAG is prone to adopt a nonbilayer structure to keep the phase equilibrium close to a bilayer to nonbilayer transition, and to maintain a spontaneous membrane curvature. It is suggested that MGalDAG, which is also nonbilayer-prone (Mannock et al., 2001) and abundant in B. burgdorferi membranes, may have an analogous function in the maintenance of biophysical properties of the membrane bilayer. Furthermore, MGalDAG is abundant in both the inner and outer membranes of B. burgdorferi (Radolf et al., 1995), and yields a strong antigenic response, especially in later stage of Lyme borreliosis (Hossain et al., 2001). Also, the T-cell receptor in mouse natural killer cells was recently shown to recognize α-MGalDAG of B. burgdorferi (Kinjo et al., 2006). Therefore, based on these immunologically important aspects of the structural properties of MGalDAG, the present characterization of bbMGS is an interesting contribution. Further studies may thereby clarify the role of bbMGS in the biosynthesis of this lipid antigen and its importance for host–pathogen interactions.

Authors contribution

Yngve Ö. and Stefan B. contributed equally to this work.

Acknowledgements

The authors thank Eva Selstam for glycolipid identification with the spray assay procedure, and Betty Guo and May Ali for reviewing the manuscript. The Swedish Medical Research Council grant 07922, the Swedish Natural Science Research Council, and the J.C. Kempe Foundation supported this work.

Footnotes

  • Editor: Anthony George

References

View Abstract