|Home | About | Journals | Submit | Contact Us | Français|
Upon nutrient limitation cells of the swarming soil bacterium Myxococcus xanthus form a multicellular fruiting body in which a fraction of the cells develop into myxospores. Spore development includes the transition from a rod-shaped vegetative cell to a spherical myxospore and so is expected to be accompanied by changes in the bacterial cell envelope. Peptidoglycan is the shape-determining structure in the cell envelope of most bacteria, including myxobacteria. We analyzed the composition of peptidoglycan isolated from M. xanthus. While the basic structural elements of peptidoglycan in myxobacteria were identical to those in other gram-negative bacteria, the peptidoglycan of M. xanthus had unique structural features. meso- or ll-diaminopimelic acid was present in the stem peptides, and a new modification of N-acetylmuramic acid was detected in a fraction of the muropeptides. Peptidoglycan formed a continuous, bag-shaped sacculus in vegetative cells. The sacculus was degraded during the transition from vegetative cells to glycerol-induced myxospores. The spherical, bag-shaped coats isolated from glycerol-induced spores contained no detectable muropeptides, but they contained small amounts of N-acetylmuramic acid and meso-diaminopimelic acid.
Myxococcus xanthus is a rod-shaped, gram-negative soil bacterium with a complex life cycle and is a model for bacterial multicellular organization and differentiation (20, 37). Under nutrient-limiting conditions on solid media, M. xanthus vegetative cells aggregate to form fruiting bodies consisting of hundreds of thousands of cells in a highly coordinated way (23). During fruiting body formation the majority of cells (ca. 80%) lyse by programmed cell death mediated by the toxin MazF (27). About 20% of the cells develop into spherical myxospores which are better adapted to deleterious environmental conditions than vegetative cells (42, 54). Myxospore formation takes about 48 h in a fruiting body. In liquid culture addition of 0.5 M glycerol to the growth medium stimulates vegetative cells to form myxospores within a few hours (10). Glycerol-induced myxospores resemble the spores formed in fruiting bodies with respect to gross morphology and resistance properties; however, the two types of spores differ in important respects. For example, spores obtained using glycerol induction lack the major cell surface protein (protein S) present in spores that develop in fruiting bodies (18). The induction of myxospore formation in liquid culture has also permitted detection of changes in intermediary metabolism (22, 31, 32, 49), as well as variations in the synthesis of macromolecules (1, 12, 13, 19, 30, 39, 52, 55) that occur during development.
The cell shape of most bacteria is determined by the shape of the peptidoglycan (murein) sacculus, an exoskeleton surrounding the cytoplasmic membrane that is essential for osmotic stability (46, 51). The only previous study of peptidoglycan from M. xanthus was published about 40 years ago (52). This study showed that the amino sugar and amino acid composition of M. xanthus peptidoglycan is similar to the amino sugar and amino acid compositions of peptidoglycans of other gram-negative bacteria. The same study reported that, surprisingly and in contrast to peptidoglycan sacculi from other species, sacculi from M. xanthus disintegrated when they were incubated with trypsin or sodium dodecyl sulfate (SDS) (52). Based on these findings, it was suggested that the cell envelope of M. xanthus could contain patches of discontinuous peptidoglycan separated by other material rather than a continuous peptidoglycan sacculus (52). Given the essential role that peptidoglycan plays in determining cell shape, the morphological change from a rod-shaped M. xanthus vegetative cell to a spherical myxospore is expected to be paralleled by major changes in the shape and/or composition of the peptidoglycan sacculus. Indeed, previous work demonstrated that there was a decrease in peptide cross-linkage during glycerol-induced myxospore formation (52).
In this work we performed an analysis of the peptidoglycan sacculus of vegetative cells and glycerol-induced myxospores of M. xanthus. We found that vegetative cells contained peptidoglycan with unusual structural features that formed a bag-shaped sacculus. Furthermore, the peptidoglycan was degraded during the transition to glycerol-induced myxospores. These spores had a bag-shaped envelope, the spore coat.
M. xanthus DK1622 GJV1 (11) was grown at 30°C in CTT medium (1% Casitone [Gibco], 8 mM MgSO4, 1 mM potassium phosphate, 10 mM Tris-HCl; pH 7.6) in liquid cultures in shaking flasks or on CTT-0.5% agar plates. For induction of myxospore formation 0.5 M glycerol was added to a liquid culture of vegetative cells with an optical density of 0.2 to 0.6, which was followed by continuous incubation at 30°C. Inspection of samples by light microscopy showed that more than 95% of the induced cells developed into glycerol-induced myxospores within 2.5 to 3 h.
M. xanthus cells from a culture with an optical density of 0.4 to 0.8 which had been grown for 2.5 to 3 h in the presence or absence of 0.5 M glycerol were sedimented, resuspended in ice-cold water, and dropped into the same volume of boiling 5% SDS. Purification of peptidoglycan for high-performance liquid chromatography (HPLC) or HPLC-mass spectrometry (MS) analysis was performed as described previously for Escherichia coli (14, 15, 25). Some sacculus preparations were treated with 100 mM NaOH for 2 h to release alkaline-labile modifications and were then washed three times with water and resuspended in water. Sacculi were prepared for electron microscopy by using a procedure described previously by de Pedro et al. (8).
Muropeptides were released from sacculi with cellosyl, reduced by using sodium borohydride, and separated by reversed-phase HPLC as previously described (14). Some spore coats were mechanically broken with glass beads (0.17 to 0.18 mm; Sigma) in a FastPrep machine. If muropeptides were prepared for MS analysis, the digestion buffer used was 20 mM ammonium acetate (pH 4.8) and the reduction step was omitted.
Muropeptide HPLC fractions (0.5 ml) were concentrated in vacuo to 20 μl and then acidified by addition of 10% trifluoroacetic acid (2 μl). All samples were desalted using RP-C18 StageTips (Proxeon Biosytems, Odense, Denmark) and eluted with a 60% acetonitrile-0.2% formic acid solution (10 μl). The muropeptide eluate was loaded into a medium NanoES spray capillary (Proxeon) and then analyzed by nano-electrospray MS in positive ion mode using a Finnigan LTQ-FT Fourier transform (FT) mass spectrometer (ThermoElectron, Bremen, Germany). The mass spectrometer was used to perform survey MS scans over the mass range from m/z = 300 to m/z = 1,900 at a typical spray voltage of 1.1 to 1.5 kV. Data were acquired with an FT MS resolution setting of 100,000 (at m/z = 400) and a linear ion trap target value of 1,000,000. MS spectra were deconvoluted to generate uncharged muropeptide masses using the QualBrowser program (ThermoElectron, Bremen, Germany).
Online micro-HPLC-MS experiments were performed by injecting nonreduced muropeptide solutions (1 μl) onto a self-packed column (0.5 by 150 mm) of Reprosil-Pur C18-AQ 3μ medium (Maisch, Ammerbuch, Germany) with a flow rate of 12 μl/min. The HPLC gradient was generated using an Agilent 1100 system with flow splitting (100 μl/min; split ratio, 1:10). The gradient conditions were 5 min with 0.5% buffer B (0.1% formic acid in acetonitrile) in buffer A (0.1% aqueous formic acid), followed by a 30 min gradient to 20% buffer B. Nonreduced muropeptides that eluted from the gradient column were analyzed by micro-HPLC-MS in positive ion mode using a Finnigan LTQ-FT FT mass spectrometer (ThermoElectron, Bremen, Germany) equipped with a Finnigan Nanospray ion source (ThermoElectron). Eluate was sprayed using uncoated TaperTips (inside diameter, 100-μm; New Objective, Woburn, MA) at a spray voltage of 2.25 kV. The mass spectrometer was used to perform survey MS scans over the mass range from m/z = 300 to m/z = 1,500 in data-dependent mode. MS data were acquired with an FT MS resolution setting of 100,000 (at m/z = 400) and a trap injection target value of 500,000. The top six ions in the parent scan were automatically subjected to MS-MS analysis in the linear ion trap region at a target value of 10,000. Mass spectrum plots and spectral deconvolution were generated using the QualBrowser program (ThermoElectron, Bremen, Germany).
Peptidoglycan from vegetative cells of M. xanthus was hydrolyzed with 4 M HCl at 100°C for 15 h. The hydrolysate was evaporated to dryness in a gentle air stream, and the residue was dissolved in 100 μl water and dried again. The dried material was redissolved in 50 μl water and analyzed by thin-layer chromatography on a cellulose plate using a previously described solvent system (38). Amino acids were stained using the ninhydrin reagent. A standard mixture containing meso-2,6-diaminopimelic acid (meso-A2pm) and ll-diaminopimelic acid (ll-A2pm) was included on the same plate, and the acids separated with the expected Rf values (0.16 for meso-A2pm and 0.22 for ll-A2pm).
Samples were hydrolyzed in 6 M HCl at 95°C for 16 h. After evaporation of the acid, the pellets were dissolved with 67 mM trisodium citrate-HCl buffer (pH 2.2) and injected into an Hitachi L8800 analyzer equipped with a 2620MSC-PS column (ScienceTec, Les Ulis, France). Amino acids and amino sugars were detected after postcolumn reaction with ninhydrin.
Cells were fixed with 2.5% paraformaldehyde, 0.2% glutaraldehyde, 30 mM sodium phosphate (pH 7.4) for 15 min at 20°C and for 30 min on ice. The cells were then sedimented, washed twice with phosphate-buffered saline, and resuspended in phosphate-buffered saline to an optical density of 1.0. For transmission electron microscopy (TEM), glutaraldehyde-fixed cells were embedded in 2% agarose and 1-mm blocks were cut out. After postfixation treatment with 1% osmium tetroxide in 100 mM phosphate buffer (pH 7.2) for 1 h on ice, the blocks were rinsed with double-distilled water, treated with 1% aqueous uranyl acetate for 1 h at 4°C, dehydrated using a graded ethanol series, and then embedded in Epon. Ultrathin sections were stained with uranyl acetate and lead citrate. Samples were viewed using a Philips CM10 electron microscope. Sacculi were visualized by TEM using a previously described protocol (8). Briefly, the sacculi were labeled on a grid with an affinity-purified antibody against peptidoglycan from E. coli. After washing, the preparation was incubated with a protein A—6-nm gold conjugate. After another washing procedure, the sacculi were contrasted with uranyl acetate and viewed with the Philips CM10 electron microscope.
A sample of spore coats was placed on a microscope slide that was pretreated with 0.01% polylysine. After 2 min the remaining liquid was removed from the slide. Images were obtained by using the ×100 objective of a Zeiss Axiovert microscope and a Sony CoolSnap HQ cooled charge-coupled device camera (Roper Scientific Ltd.) attached to the microscope. Digital images were obtained with Metamorph (version 4.6.9) and were processed with ImageJ (version 1.38) to find the edges of the spore coats. The diameter of spore coats was measured by “straight-line selection.” The surface area was calculated from the measured diameter by assuming that the spore coat was flat on the surface, and it was 9.87 ± 1.81 μm2 (n = 100). The spore coat density in the sample was determined with a Thoma counting chamber (Weber, United Kingdom) using a Zeiss Axiovert microscope and a ×40 objective, and it was 4.37 × 1011 spore coats per ml. The surface density of amino acids and amino sugars was estimated from the concentration of the compounds in a spore coat sample (as determined by amino acid and amino sugar analysis), the mean surface area of the spore coats, and the density of the spore coats in the sample.
Muropeptides are the GlcNAc-N-acetylmuramic acid (MurNAc) disaccharide peptide building blocks of peptidoglycan released by treatment with a muramidase. They have characteristic, species-specific structural elements determined by the peptidoglycan biosynthetic pathway and the specificity of the enzymes involved. In all gram-negative bacteria studied thus far, the sequence of the pentapeptide is l-Ala-γ-d-Glu-meso-A2pm-d-Ala-d-Ala (36, 40). Cross-linking and proteolytic reactions result in a high degree of structural diversity of muropeptides, which is reflected by the complex muropeptide pattern (or profile) of a given species. In addition to pentapeptides, the peptidoglycans of gram-negative bacteria contain variable amounts of the corresponding monomeric tetra-, tri- and dipeptides, as well as different types of dimeric, trimeric, and tetrameric (cross-linked) muropeptides. There are also particular structures marking the chain ends (1,6-anhydro-MurNAc-containing muropeptides) or the attachment sites of Braun's lipoprotein (muropeptides with a Lys-Arg dipeptide) (14, 15, 36).
Peptidoglycan was isolated from vegetative M. xanthus cells and from E. coli as described previously (14, 15). The material obtained was digested with the muramidase cellosyl to release the muropeptides. Nonreduced muropeptides were first analyzed by a newly developed, rapid method for detection in peptidoglycan digestion samples (see Materials and Methods). The muropeptide mixture was applied to a micro-C18 reversed-phase HPLC column which was directly coupled to an LTQ-FT mass spectrometer. Ion peaks corresponding to the masses of expected muropeptides were identified in digests of the peptidoglycans from M. xanthus and E. coli (Table (Table1).1). With the exception of a few minor muropeptides, the masses of all previously described muropeptides of E. coli were identified by this method with high mass accuracy. We expected to obtain similar muropeptide profiles for E. coli and M. xanthus. Indeed, the M. xanthus sample contained several masses corresponding to the masses of muropeptides present in E. coli, including masses of the muropeptides Tetra, TetraTetra, TetraTetraAnh, TetraTetraTetra, and TetraTetraTetraAnh (see reference 15 for an explanation of the nomenclature of muropeptides). Comparison of MS-MS spectra of the major muropeptides of M. xanthus and E. coli revealed identical fragmentation products (not shown). In the M. xanthus sample the signals for anhydro (Anh) muropeptides which originated from glycan chain ends were generally greater. Also, the M. xanthus peptidoglycan masses included masses for additional anhydro muropeptides (TetraTetraTetradiAnh and TetraTetraTetraTetraAnh) not present in the E. coli sample, indicating that anhydro muropeptides might be more abundant in M. xanthus than in E. coli. The M. xanthus sample did not produce detectable mass signals for muropeptides with di- or tripeptides (for example, the muropeptide Di or Tri or the dimer TetraTri) or for muropeptides with the Lys-Arg modification originating from the covalently bound Braun's lipoprotein. On the other hand, the M. xanthus sample contained several molecules with masses corresponding to those of modified muropeptides. One modification (designated modification X [Table [Table1])1]) was present in the muropeptides Tetra, TetraTetraAnh, and TetraTetraTetraAnh and corresponded to a 100.02-Da increase in the molecular mass. Another modification (designated modification Y) was present in the muropeptides Tetra, TetraTetraAnh, TetraTetra, TetraTetraTetraAnh, and TetraTetraTetra and increased the molecular masses of these molecules by 372.16 Da. Overall, micro-HPLC-LTQ-FT MS analysis revealed remarkable differences in the muropeptide profiles of M. xanthus and E. coli. We also analyzed the muropeptide composition of M. xanthus by using classical HPLC separation and offline electrospray MS.
Muropeptides of M. xanthus were reduced with sodium borohydride and separated by C18 reversed-phase HPLC using a method described previously for E. coli muropeptides (14, 15). More than 20 muropeptides were separated. The peak pattern was strikingly different from the previously described pattern obtained for E. coli muropeptides (Fig. (Fig.11 and Table Table2)2) (14, 15). Twenty-three muropeptide fractions were collected, concentrated, desalted, and analyzed by LTQ-FT MS (Table (Table2).2). None of the major components in the monomeric region of the chromatogram corresponded to the muropeptide Di, Tri, or Penta, which was consistent with the results obtained by micro-HPLC-LTQ-FT MS. Surprisingly, both major monomeric muropeptide peaks (Fig. (Fig.1,1, peaks 1 and 2) had the expected molecular mass of the muropeptide Tetra, as determined by LTQ-FT MS with molecular mass accuracy in the low range (<5 ppm). The muropeptide Tetra of E. coli coeluted with muropeptide 2 of M. xanthus, but no E. coli muropeptide coeluted with muropeptide 1 of M. xanthus (not shown). Moreover, the M. xanthus sample contained four dimers (muropeptides 5, 6, 7, and 8) with the expected molecular mass of TetraTetra, and one of them (muropeptide 6) coeluted with E. coli TetraTetra (not shown). M. xanthus had also four dimers (muropeptides 14, 15, 16, and 17) with the expected molecular mass of TetraTetraAnh and several trimers with the molecular mass of TetraTetraTetra or TetraTetraTetraAnh. From these results we inferred that the basic building blocks in the peptidoglycan of M. xanthus exist in two versions with the same molecular mass; i.e., there are two isomers of monomers, four isomers of dimers, and (theoretically) eight isomers of trimers. The isomeric monomers and dimers could be separated using the HPLC conditions described here. The presence of such muropeptide isomers in a single species has not been described previously.
In order to determine the structural basis for the observed muropeptide isomerization, a sample of M. xanthus peptidoglycan was hydrolyzed with hydrochloric acid, and the resulting amino sugar-amino acid mixture was analyzed by thin-layer chromatography. This method is routinely used for peptidoglycan analysis during characterization of bacterial taxa at the German Collection of Microorganisms and Cell Cultures. Interestingly, myxococcal peptidoglycan contained both meso-A2pm and ll-A2pm at a molecular ratio of about 1 to 0.4. The presence of two A2pm isomers in M. xanthus was unexpected because only meso-A2pm has been found in the peptidoglycan of all other gram-negative bacteria studied thus far (36, 40). This would explain the pattern of muropeptide isomers observed. We inferred that muropeptide 1 (Fig. (Fig.1)1) is the muropeptide Tetra containing ll-A2pm, whereas muropeptide 2 is the “normal” Tetra muropeptide with meso-A2pm (hence, muropeptide 2 coeluted with meso-A2pm-containing Tetra of E. coli). Also, the observed peak ratio of muropeptide 2 to muropeptide 1 was approximately the same as the measured ratio of meso-A2pm to ll-A2pm in the peptidoglycan. We suggest the following structures for the four isomers of the major dimeric muropeptide (TetraTetra) based on their abundance and elution behavior. Compound 6 coeluted with TetraTetra of E. coli and is most likely the “normal” version with two meso-A2pm residues. Compound 5 eluted earlier than compounds 6, 7, and 8, and its abundance was lowest; therefore, it is most likely the isomer with two ll-A2pm residues. Compounds 7 and 8 presumably contain one meso-A2pm residue and one ll-A2pm residue (Table (Table22 and Fig. Fig.2).2). The four TetraTetraAnh isomers (muropeptides 14 to 17) gave rise to a peak pattern in the chromatogram similar to that produced by the TetraTetra isomers and are likely to have an analogous A2pm profile.
The average length of the glycan strands could be estimated from the fraction of anhydro-MurNAc residues. The glycan strands were relatively short in M. xanthus, and the average length was 8.9 disaccharide units (calculated using muropeptides 1 to 23). This value is, if anything, an overestimate because some of the unidentified, minor muropeptides with high retention times were likely to carry an anhydro group. Although the fraction of anhydro-muropeptides was high, we found that all detectable anhydro groups were present in oligomeric structures (dimers to tetramers), and we could not detect the monomeric molecule TetraAnh. Thus, the majority (if not all) of the anhydro ends of the glycan strands are connected to other strands by peptide cross-links. About 60.4% of the peptides in the myxococcal peptidoglycan were present in cross-links, and a relatively high proportion (18.7%) were present in trimeric structures.
We identified by using MS two modifications in the myxococcal peptidoglycan, both of which are low-abundance modifications. The X modification was present in about 4.5% of all monomers, and the Y modification (present in muropeptide 9) accounted for about 0.7% of the total UV-absorbable material and was also present in several minor muropeptides (muropeptides 18B, 19B, and 20B) that coeluted with the TetraTetraTetraAnh isomers. Such modifications to the muropeptide structure might occur by proteolytic cleavage or by addition of a structural element. The molecular mass of the Y modification (372.1645 Da) coincided exactly with the theoretical molecular mass of a Glu-A2pm-Ala tripeptide without water (372.16450 Da). MS-MS analysis of protonated muropeptide 9 showed that there was initial loss of the GlcNAc residues, which is generally observed with muropeptides. Both Glu and Ala could be lost from the resulting fragment (muropeptide 9 lacking GlcNAc) (data not shown). This was expected based on the proposed structure of muropeptide 9 with a terminal Glu residue and was not seen with muropeptides with Glu in a peptide chain. Thus, it is most likely that compounds with the Y modification were formed from an oligomeric muropeptide by proteolytic cleavage between l-Ala and d-Glu by an ld-endopeptidase, resulting in release of the GlcNAc-MurNAc-l-Ala moiety.
Two isomeric monomers (muropeptides 3 and 4) contained the X modification. The exact molecular mass of modification X was determined by subtracting the molecular masses of the unmodified muropeptides from the molecular masses of the modification X-containing muropeptides in LTQ-FT MS experiments performed with mixtures of nonreduced samples and was found to be 100.0173 ± 0.0011 Da. Based on the accuracy of the molecular mass determined, only one elemental composition of modification X appears to be possible: C4H4O3 (theoretical molecular mass, 100.016045 Da). Other possible combinations of C, N, O, and H atoms result in theoretical molecular masses which do not fit the observed molecular mass difference given the mass accuracy. Also, the observed isotope profile of modification X-containing muropeptides excluded the possibility of the presence of other atoms, such as S or P. Incubation of peptidoglycan with 80 mM sodium hydroxide for 1 h resulted in the absence of any detectable muropeptides with modification X, while all other muropeptides could be detected, indicating that muropeptides with modification X are alkali labile (not shown). However, several attempts to identify the H-X-OH compound in the supernatant of NaOH-treated sacculi by electron impact MS failed, possibly because of the small amounts of material obtained.
The MS-MS data allowed identification of the attachment site of modification X in the muropeptide Tetra. (i) Fragmentation of Tetra-X, fragmentation of M. xanthus Tetra, and fragmentation of E. coli Tetra all gave rise to a strong molecular mass signal at 534.2 Da, which corresponded to the protonated lactyl-tetrapeptide moiety. MS-MS-MS analysis of the 534.2-Da ions present in the fragmented Tetra-X or Tetra sample revealed identical fragmentation patterns. Further, fragmentation of all three compounds resulted in a positive ion with a molecular mass of 719.2 Da corresponding to the protonated and dehydrated MurNAc-tetrapeptide ion. Thus, Tetra-X has an unmodified peptide and peptide-MurNAc linkage, and modification X must be attached to either GlcNAc or MurNAc (or both). (ii) Both GlcNAc and modification X were readily lost in fragmentation experiments with protonated Tetra-X (1,040.4 Da), resulting in fragments corresponding to the loss of GlcNAc (a loss of 203.1 Da) or modification X (a loss of 100.0 Da) or both (a loss of 303.1 Da). (iii) MS-MS-MS analysis of the weak signal of the protonated Tetra-X(-GlcNAc) ion (837 Da) revealed loss of modification X (a loss of 101 Da). From these results we concluded that modification X must be attached to the MurNAc residue. It was possible to reduce Tetra-X with sodium borohydride, which resulted in a 2-Da increase in the molecular mass. Reduction requires the presence of a free OH group at the anomeric C-1 position of MurNAc, thereby excluding the possibility that modification X is attached there. Because there are no free OH groups at positions 2, 3, 4, and 5 of MurNAc, the only remaining possible position for the attachment of modification X is the C-6 OH. We therefore concluded that modification X is most likely attached to the C-6 OH of MurNAc by an alkali-labile ester bond.
Vegetative cells of M. xanthus were harvested, and peptidoglycan sacculi were prepared for TEM by repeated incubation in hot SDS and treatment with protease to remove contaminating proteins. The material obtained was immobilized on grids and contrasted with uranyl acetate. TEM revealed the presence of typical sacculi whose sizes were similar to those of M. xanthus cells. The sacculi from M. xanthus were incubated on a grid with an antibody raised against peptidoglycan from E. coli and then with a protein A—6-nm gold conjugate. Examination of the samples by TEM revealed that the M. xanthus sacculi were uniformly labeled with gold particles (Fig. (Fig.3A).3A). These results indicate that vegetative cells of M. xanthus have a continuous peptidoglycan sacculus with structural elements similar to the structural elements of the peptidoglycan of E. coli, which is in agreement with the results of the muropeptide analysis described here.
We attempted to visualize the sacculi from both myxospores and cells in the transition from the vegetative form to the spore form using TEM. We confirmed the previous observation that rod-shaped, vegetative cells of M. xanthus changed to spherical myxospores within 2.5 to 3 h during cultivation in the presence of 0.5 M glycerol in aerated liquid medium. Under these conditions, conversion to myxospores was almost quantitative, as judged from microscopic examination of the cells. Sacculi from sporulating cells or from myxospores were isolated in the same way as sacculi from vegetative cells, immobilized on grids, immunolabeled with anti-peptidoglycan antibody-protein A—6-nm gold conjugate, contrasted with uranyl acetate, and examined by TEM. Spore sacculi were spherical and bag-shaped objects whose electron density differed from that of sacculi isolated from vegetative cells (Fig. (Fig.3C).3C). These spore sacculi contained areas with very high electron density. They appeared to consist of the same material as the myxospore coats described previously (21) and were not labeled by the anti-peptidoglycan antiserum. Moreover, incubation of spore coats with cellosyl did not result in release of muropeptides; HPLC analysis of a sample failed to detect muropeptides or any other UV-absorbable material in the supernatant of the cellosyl digest even when the spore coats were mechanically broken using glass beads prior to cellosyl digestion (not shown). Samples from cells in transition from vegetative cells to spores contained both bag-shaped spore coats not labeled by anti-peptidoglycan antiserum and peptidoglycan patches of various sizes labeled with the antibody (Fig. (Fig.3B).3B). Frequently one or more patches of peptidoglycan were attached to a spherical spore coat, indicating that they could have originated from the same cell. These results show that development into myxospores involves both formation of a new spore coat, which does not consist of peptidoglycan, and degradation of the rod-shaped peptidoglycan sacculus.
Our failure to detect muropeptides in the spore coats could have been due to different factors. Either the spore coat did not contain a significant amount of peptidoglycan, or it contained peptidoglycan which could not be degraded by muramidase or recognized by the antiserum. To determine if the spore coat contained peptidoglycan, we hydrolyzed the material, quantified the amino acids and amino sugars, and related the amounts of these compounds to the surface area of spore coats. Isolated spore coats had an average surface area of 9.87 ± 1.81 μm2 (n = 100) as determined by light microscopy (see Materials and Methods). Table Table33 shows the compounds identified and the amount per unit of surface area. The spore coat material contained variable amounts of all of the amino acids present in proteins. Glycine, alanine, and glutamic acid were the most abundant amino acids. The spore coats also contained glucosamine and small amounts of A2pm and muramic acid. Because the latter two compounds are present only in peptidoglycan, it is likely that the spore coats contained a small amount of peptidoglycan.
The new spore coat layer could be observed in thin sections of glutaraldehyde-fixed spores prepared and visualized by TEM. Both vegetative cells and spores contained an inner cell membrane and an outer cell membrane separated by periplasmic space (Fig. (Fig.4).4). The resolution and contrast of the electron micrographs were not high enough to recognize the peptidoglycan layer. The spores had extra membrane loops and membrane vesicles on the surface, presumably because the surface area-to-volume ratio decreased during the transition from the rod shape to the spherical shape and thus the cells accumulated extra membrane. A thin spore coat layer surrounded the outer membrane and the extra membrane material of the spores (Fig. (Fig.4B4B).
The presence of a meso-A2pm residue at position 3 of the stem peptide has been the hallmark of the peptidoglycan of gram-negative bacteria and of some gram-positive species (for example, Bacillus subtilis) (46). In this work we found that in M. xanthus peptidoglycan a significant fraction (about 30%) of the stem peptides had an ll-A2pm residue in place of the “normal” meso-A2pm residue. ll-A2pm has been detected in the peptidoglycans of several gram-positive species, including Streptomyces spp. and members of the families Propionibacteriaceae and Intrasporangiaceae (40). The occurrence of ll-A2pm in combination with meso-A2pm has been found in species of the actinomycete Kitasatosporia, where ll-A2pm was found in the aerial mycelium and meso-A2pm was found in the vegetative mycelium of the same strain (43). ll-A2pm has not been described as a component of peptidoglycan in any gram-negative species. Also, ll-A2pm is not present in the peptidoglycan of an unclassified member of the Myxococcales, Myxobacter sp. strain AL-1 (16). There are three slightly different bacterial pathways for the synthesis of meso-A2pm, which is also a precursor of l-lysine (5, 50). Sequence comparisons revealed that M. xanthus strain DK1622 contains homologues of all four genes (dapC, dapD, dapE, and dapF) of the succinylase pathway, in which ll-A2pm is the direct precursor of meso-A2pm. Therefore, it is possible that M. xanthus has cytoplasmic pools of both ll-A2pm and meso-A2pm. The amino acid ligase MurE is responsible for addition of the third amino acid (most often a diamino acid) to the peptidoglycan precursor UDP-MurNAc-l-Ala-d-Glu. In most cases MurE exhibits high specificity for its amino acid. This has been shown for the meso-A2pm-adding enzyme from E. coli and the l-Lys-adding enzyme from Staphylococcus aureus (3). However, sometimes MurE appears to lose its strict specificity. The peptidoglycan of some species of Bifidobacterium contains both l-Lys and l-Orn at position 3 (40). Another example is MurE from Thermatoga maritima, which can add l-Lys and d-Lys with comparable efficiencies (4). The genome of M. xanthus DK1622 contains a single murE homologue (accession number NC_008095). Presumably, this myxococcal MurE is capable of adding both meso-A2pm and ll-A2pm to the peptidoglycan precursor.
In M. xanthus all the peptidoglycan biosynthetic steps following the MurE step appear to function with both meso-A2pm- and ll-A2pm-containing peptides, resulting in the variety of muropeptide structures demonstrated here. It is quite remarkable that both the ll-A2pm- and meso-A2pm-containing peptides can participate in cross-linking reactions, as shown by the presence of all four possible isomers of the dimeric muropeptides. Cross-linking reactions are catalyzed by a set of acyl serine transferases (the penicillin-binding proteins [PBPs]), which form a new peptide bond between the tetrapeptidyl residue of a donor peptide and the -amino group of the A2pm residue of an acceptor peptide. Either the PBPs of M. xanthus are generally nonspecific with respect to the configuration of A2pm, or there are separate PBPs specific for ll-A2pm-containing peptides and for meso-A2pm peptides. The presence of ll-A2pm could also have implications for the activity of peptidoglycan endopeptidases cleaving the peptide cross-links.
Although we have determined the likely sum formula (C4H4O3) of modification X, our attempts to establish the chemical structure of this molecule were not sucessful because of the low abundance of the modified muropeptides, which made it impossible to obtain amounts sufficient for nuclear magnetic resonance analysis. However, there may be other myxococcal strains or bacterial species containing larger amounts of muropeptides with modification X that are sufficient for nuclear magnetic resonance analysis. The structure of modification X differs from the structures of previously described modifications in glycan strands (45), but, like the frequently found O-acetyl or phosphate groups, modification X is most likely attached to the C-6 OH of MurNAc. Possible biological compounds consistent with C4H6O4, the sum formula of H-X-OH, are succinic acid and methylmalonic acid. We also identified several minor muropeptides with the Y modification. These muropeptides appear to be degradation products resulting from cleavage by an endopeptidase between l-Ala and A2pm. Although enzymes with this specificity are present in other bacteria (48), the peptidoglycan endopeptidase of M. xanthus responsible for producing such muropeptides with modification Y has yet to be identified.
The peptidoglycan of M. xanthus has relatively short glycan strands; the average length is about 9 disaccharide units, which is significantly less than the average length of glycan strands reported for E. coli (21 to 35 disaccharide units) (14, 17, 47). The presence of shorter glycan strands in M. xanthus might be compensated for by increased peptide cross-linkage, which occurs to a greater degree in M. xanthus than in E. coli. In contrast to E. coli, in M. xanthus all detectable glycan chain ends were at cross-linked peptides (and not at monomeric, non-cross-linked peptides), which might be an important structural feature of peptidoglycan containing short glycan strands. Other species with relatively short glycan strands are the gram-negative organism Helicobacter pylori (6, 7) and the gram-positive organism S. aureus (2). Other remarkable features of the peptidoglycan from M. xanthus are the absence of detectable amounts of tripeptides and the low levels of pentapeptides. In M. xanthus pentapeptides are almost quantitatively cleaved to tetrapeptides by dd-carboxypeptidase(s), whereas the tripeptide-producing ld-carboxypeptidases are either not present or poorly active. Significant (and variable) amounts of tripeptides are present in the peptidoglycans of many gram-negative bacteria (36). In addition to M. xanthus, Caulobacter crescentus is gram-negative species with very low levels of tripeptides (26). Unlike the situation in M. xanthus, in C. crescentus there are significantly increased amounts of pentapeptides in the peptidoglycan. Whether the relative amounts of tri-, tetra-, and pentapeptides are important for specific physiological features in M. xanthus and other species is not known, but in E. coli trimming of pentapeptides to tetrapeptides by dd-carboxypeptidases is required to maintain the normal rod shape under certain conditions (9, 28, 44).
A previous study reported that sacculi of vegetative M. xanthus cells disintegrate upon treatment with hot detergent (SDS) and protease, suggesting that there is a discontinuous layer (52). We could not confirm these observations. We observed intact peptidoglycan sacculi of M. xanthus using electron microscopy (Fig. (Fig.3).3). These sacculi remained intact after incubation with 4% SDS (4 h at 95°C or 18 h at 80°C) and with different proteases (chymotrypsin or pronase; incubation for 12 h at 37°C) (not shown). It is possible that the sacculi used in the previous study were contaminated with one or more endogenous peptidoglycan hydrolases. High numbers of these autolysins are known to be present in M. xanthus (41). For preparation of our sacculi we paid particular attention to rapidly and quantitatively inactivating the autolysins by dropping the ice-cold cell suspension of M. xanthus cells into a boiling SDS solution, followed by repeated boiling steps in an SDS solution to remove attached proteins.
One of our initial goals was to establish the structure of the peptidoglycan in myxospores. This could be done only with glycerol-induced myxospores obtained in liquid culture, because the amount of pure spores that could be isolated from fruiting bodies was too small for peptidoglycan preparation and analysis. We realize that glycerol-induced myxospores differ in several ways from myxospores formed naturally in fruiting bodies. An important difference is the absence in glycerol-induced myxospores of surface protein S, which is synthesized during development of the fruiting body and which is the major component of the thick spore coat (18). Glycerol-induced myxospores form a thinner polysaccharide spore coat containing 14% protein, 8% glycine, and less than 1% organic phosphorus (21). In electron micrographs the appearance of the spore coats isolated in this work (Fig. (Fig.3C)3C) and the appearance of the spore coats described previously (21) are similar. Although the isolation procedures were not identical, the different spore coat samples appeared to originate from the same layer surrounding the outer membrane of the glycerol spores (Fig. (Fig.3C3C).
It is not known from the literature whether myxospores contain peptidoglycan. To our surprise, we found that glycerol-induced myxospores do not contain a detectable amount of muropeptides. We first reasoned that the muramidase cellosyl might not be able to access the peptidoglycan because it is shielded by the spore coat. However, there was no release of muropeptides by cellosyl even after the spore coats were mechanically broken with glass beads (not shown), eliminating the possibility that there is shielding effect. Electron microscopy showed that spore coats were not recognized by antiserum against peptidoglycan, whereas patches of peptidoglycan were present on spore coats in a sample from cells which were in the process of glycerol-induced sporulation. These observations established that the peptidoglycan sacculus is degraded during glycerol-induced sporulation. Nonetheless, the spore coats contained small amounts of N-acetylmuramic acid and diaminopimelic acid, suggesting that they contained some peptidoglycan. The surface densities of meso-A2pm and muramic acid in spore coats (0.52 × 1010 and 1.40 ×1010 nmol/μm2, respectively [Table [Table3])3]) are significantly lower than the reported surface density of meso-A2pm in E. coli (7.0 × 1010 nmol/μm2; calculated from the average number of meso-A2pm molecules per cell and the average cell surface area ). Considering that the amount of peptidoglycan in E. coli can form at most two complete layers (24, 35, 53), the glycerol spore coats from M. xanthus cannot have a complete layer of peptidoglycan. There are several possible explanations for the observed small amount of peptidoglycan in glycerol spores: (i) some peptidoglycan fragments were retained in the coats after degradation of the sacculus; (ii) parts of the sacculus were not degraded during glycerol-induced sporulation; or (iii) the residual peptidoglycan originated from a small fraction of cells which did not sporulate or which started but did not complete the sporulation process.
From this work we concluded that glycerol-induced myxospore development includes synthesis of a new spore coat and degradation of most if not all of the peptidoglycan sacculus. This is in sharp contrast to the formation of endospores in gram-positive bacteria, such as B. subtilis or Clostridium sporogenes. Although these species degrade the peptidoglycan of the mother cell, they retain a structurally modified peptidoglycan sacculus around the spore (33, 34). Also, streptomycetes keep their peptidoglycan during development from aerial hyphae to spore chains (29). It remains to be established whether myxospores developing in a fruiting body retain peptidoglycan and whether glycerol-induced myxospores rebuild a peptidoglycan sacculus once they age.
The absence of a peptidoglycan sacculus implies that in glycerol-induced myxospores the spore coat is responsible for osmotic stability, a function which is consistent with the bag shape and the size of isolated spore coats shown in this and previous work (21). The molecular architecture of the spore coat is not known. It is possible that, like peptidoglycan, the spore coat forms a single netlike (or cross-linked) macromolecule able to withstand the spore's turgor. Other interesting questions arising from this work are, how is the synthesis of the spore coat coordinated with the degradation of the peptidoglycan sacculus during spore formation and how is the rod-shaped peptidoglycan sacculus resynthesized during germination.
We thank Miguel de Pedro of Universidad Autónoma de Madrid, Madrid, Spain, for providing anti-peptidoglycan antibody and technical advice; Dave Dunbar of the Chemical Analysis Services Unit, Newcastle University, Newcastle upon Tyne, United Kingdom, for performing the electron impact MS analysis; and Richard Daniel of Newcastle University for technical advice concerning determination of the surface area of the spore coats.
This work was supported by the European Commission through the EUR-INTAFAR project (grant LSHM-CT-2004-512138).
Published ahead of print on 7 November 2008.