|Home | About | Journals | Submit | Contact Us | Français|
Cell differentiation is widespread during the development of multicellular organisms, but rarely observed in prokaryotes. One example of prokaryotic differentiation is the Gramnegative bacterium Myxococcus xanthus. In response to starvation, this gliding bacterium initiates a complex developmental program that results in the formation of spore-filled fruiting bodies. How the cells metabolically support the necessary complex cellular differentiation from rod-shaped vegetative cells into spherical spores is unknown. Here, we present evidence that intra-cellular lipid bodies provide the necessary metabolic fuel for the development of spores. Formed at the onset of starvation, these lipid bodies gradually disappear until they are completely used up by the time the cells have become mature spores. Moreover, it appears that lipid body formation in M. xanthus is an important initial step indicating cell fate during differentiation. Upon starvation, two subpopulations of cells occur: cells that form lipid bodies invariably develop into spores, while cells that do not form lipid bodies end up becoming peripheral rods, which are cells that lack signs of morphological differentiation and stay in a vegetative-like state. These data indicate that lipid bodies not only fuel cellular differentiation but that their formation represents the first known morphological sign indicating cell fate during differentiation.
In multicellular organisms, cellular differentiation is widespread and enables cells to perform highly specialized functions in tissues and organs. Although, the occurrence and significance of cellular differentiation in prokaryotes is less well established certain examples such as the development of heterocysts in cyanobacteria (Wolk, 2000) or the formation of stress-resistant spores (Strauch and Hoch, 1992;Piggot, 1996) have been intensively studied. However, the most advanced form of cellular differentiation in prokaryotes occurs in myxobacteria. These Gram-negative soil bacteria are characterized by their multicellular social behavior and their remarkable ability to undergo a complex developmental cycle (Kaiser, 2003). When in contact with a solid surface, Myxococcus xanthus grows and moves as a swarm spreading biofilm-like over the surface. During movement the cells secrete lytic enzymes and antimicrobial compounds that allow them to prey on other bacteria as food source (McBride and Zusman, 1996). When the prey bacteria are used up and nutrients run low, the cells stop moving outwards but start aggregating and initiate a complex developmental program that culminates in the formation of multicellular fruiting bodies that are visible with the naked eye (Zusman et al., 2007). Inside these fruiting bodies, cells differentiate into spores that are resistant to heat, desiccation, and nutrient deprivation. While the large majority of the aggregated cells follow this developmental fate, a smaller subpopulation of cells, termed peripheral rods, remains rod-shaped and do not aggregate or sporulate (O’Connor and Zusman, 1991c). Although peripheral rods superficially resemble vegetative cells, biochemical analyses show that these cells express markers that clearly distinguish them from vegetative and sporulating cells (O’Connor and Zusman, 1991a). Therefore, it has been speculated that peripheral rods are “scout cells” that continue to search for prey while the large part of the myxobacterial population rests as spores. Thus, cellular differentiation in myxobacteria is highly orchestrated and involves multicellularity through aggregation and the development of various specialized cells that differ morphologically, functionally, and metabolically (Shimkets, 1999;Kaplan, 2003).
Although it is clear that M. xanthus cells undergo massive reprogramming of their gene expression patterns during differentiation and fruiting body formation (Kroos and Inouye, 2008), it is less clear how they metabolically support the necessary associated morphological changes. These changes include, among others, the extensive remodeling of the cell envelope due to the change of cell shape from a rod to a sphere (White, 1993), the synthesis of a complex multilayered spore coat (White, 1975), the formation of a two-chromosome complement (Tzeng and Singer, 2005), and the synthesis of spore-specific lipid components (Ring et al., 2006). What makes these feats even more remarkable is the fact that the cells have to synthesize and assembly all of these novel components and structures under condition of starvation and consequently severely reduced metabolic means. Therefore, it has been a longstanding puzzle how M. xanthus and other bacteria accomplish such a task under such adverse circumstances.
Here, we present evidence that M. xanthus uses lipid bodies as intermediary carbon and energy storage to ensure successful completion of its complex developmental cycle. Our data indicate that lipid body formation in M. xanthus is coordinated during starvation-induced fruiting body formation. Upon the onset of starvation, the cells convert excess metabolites and, potentially, biomass into lipids that then subsequently are used to fulfill the cell’s metabolic requirements during the transformation from vegetative cells into mature resting spores. This interpretation is supported by the observation that these newly formed lipid bodies gradually disappear during the maturation of the spores until they are entirely used up when the cells have completed the transformation process. Moreover, vegetative M. xanthus cells that have been pretreated with kanamycin to induce lipid body formation prior to starvation, form significantly more fruiting bodies and spores than untreated cells further demonstrating a link between the formation of lipid bodies and the cell’s ability to differentiate. Interestingly, experiments with purified peripheral rods show that these cells, while experiencing the same starvation do not form lipid bodies, are unable to form fruiting bodies, and do not differentiate into spores. Thereby supporting the idea that peripheral rods are following a very different developmental path resulting in a subpopulation of cells that is morphologically, functionally, and metabolically different. Intriguingly, the formation of lipid bodies is the first identifiable morphological feature that sets these two different subpopulations of cells apart at the beginning of the differentiation process. Therefore, we propose that M. xanthus not only uses lipid bodies as intermediary storage to complete their developmental differentiation process during sporulation but that the formation of these structures itself plays a central role in cell fate determination during this process.
Light microscopic observations of M. xanthus wild type cells growing on agar or in liquid culture containing low concentrations of kanamycin (20-30 μg/ml) revealed the formation of large intracellular inclusions (Fig. 1A, B). Similar intracellular inclusions were formed when aerobically growing cultures were shifted to oxygen-limiting conditions (Fig. 1D, E). To further explore the formation of these inclusions in M. xanthus, we tested the effect of sudden nutritional step-downs on this process. For these experiments, M. xanthus cells that grew exponentially in 1% CTT medium were harvested by centrifugation and resuspended in CTT medium containing ten or twenty fold less casitone. Again, within a few hours of growth at 32°C the cells formed intracellular inclusions (Fig. 1C), indicating that various metabolic stress situations induced the same cellular response.
Careful examination of cells under all three stress conditions, kanamycin exposure, anaerobiosis, and lower nutrients showed that the number of inclusions formed depended on two major factors: The physiological condition of the cells at the onset of the experiment and the severity of the imposed stress. Generally, the highest number of inclusions was observed in cells that were rapidly growing and subsequently severely stressed. For example, addition of low concentrations of kanamycin (20-30 μg/ml) to exponentially growing wild type cells resulted in formation of a few to a dozen inclusions per cell (Fig. 1A, B; Fig. 2), whereas cells upon exposure to kanamycin concentrations close to the amount used for the selection of kanamycin-resistant strains (40-50 μg/ml) over time became completely filled with inclusions (Fig. 1F, G; Fig. 2).
As indicated above, the number of inclusions varied in M. xanthus and reflected the physiological state of the cells and the severity and duration of the stress. In order to better quantify these effects, we studied the relative total volume and the distribution of these particles either under low stress (20 μg/ml kanamycin) or severe stress conditions (40 μg/ml kanamycin). After 6-12 h growth on kanamycin-containing agar, the relative total volume and the location of inclusions were determined as described in the experimental procedures (Fig. 2A). On average, cells under low stress conditions developed three to four inclusions that occupied ca. 4% of the cell volume, whereas under severe stress conditions the cells formed many more inclusions that filled up to 12% of the cell volume.
Since cells under low stress conditions showed an unusually high number of lipid bodies in close proximity to the cell poles (Fig. 1A, C, E), we systematically measured their cellular distribution to find out whether their arrangement was random or ordered. Fig. 2B shows that under this condition the majority of lipid bodies were polarly arranged and that, upon calculation, this arrangement was statistically significant suggesting that the inclusions are not arranged at random (χ2 test, p value = 1.07*10-36). Moreover, the careful inspection of the lipid bodies formed under these condition revealed yet another non-random aspect of their arrangement. This less obvious aspect is easily explained using the cells seen in Fig. 1A, B. The upper of the two cells in the picture contains four lipid bodies that show a highly peculiar arrangement. First, there are two “pairs” of lipid bodies, two smaller and two larger ones. Secondly, equally sized “pairs” of lipid bodies are found at the same cellular locations; the two small ones at the cell pole, while the two larger ones are equidistantly away from the poles. What could cause such a peculiar non-random arrangement? Most likely, the smaller lipid bodies were more recently formed because they have accumulated less material. The fact that these lipid bodies are at the cell poles supports the idea that these structures, at least under low stress, are probably formed at this location. But what could explain the equidistant arrangement of the larger two lipid bodies? The easiest explanation is the hypothesis that these symmetrically arranged lipid bodies were originally formed at the cell poles. But when the cells grew in length, the polarly formed lipid bodies were “left behind”. Using this criterion, more than 60% of all lipid bodies at low stress were most likely formed at the cell poles, with about 28% still being there. Moreover, we cannot exclude that the remaining lipid bodies were also originally polarly formed, but because they exist as unpaired single structures it is impossible to tell using this criterion. This interpretation also applies to the lower cell in Fig. 1A, B that contains an equidistant pair and one unpaired single lipid body. Therefore, six out seven lipid bodies in these two cells were most likely polarly formed. In contrast, under severe stress conditions the cells formed not only higher numbers of lipid bodies but they appeared to be distributed at random throughout the entire cytoplasm of the cells obscuring any particular arrangement pattern Fig. 2B (χ2 test, p value = 0.68).
In myxobacteria, starvation plays a key role in the initiation of a complex developmental program that ultimately results in the formation of spore-filled multicellular fruiting bodies (Diodati et al., 2008). Based on our observation that a sudden nutritional step-down in the growth medium induced lipid body formation in M. xanthus, we decided to study whether lipid bodies are also formed during fruiting body and spore development. Therefore, liquid grown vegetative M. xanthus cells were spotted on starvation agar and cells were examined with the light microscope. In addition, for a total of six days every 24 h developing fruiting bodies were processed for electron microscopy to evaluate the formation of inclusions during the conversion of rod-shaped vegetative cells into round myxospores at high resolution (Fig. 3).
In line with the observations described above, the severe starvation stress during the fruiting body assay initiated the accumulation of large numbers of lipid bodies in the majority of cells within the first 6-12 h of starvation. However somewhat surprisingly, a small fraction of cells reproducibly did not form lipid bodies under these conditions. Thin sections of induced cells showed that dozens of lipid bodies were formed even before any morphological change indicated the impending conversion of the cells into spores. Once inclusions were formed, they could be detected for up to 72 h and, in some cases, even 96 h after the conversion of the cells into spores had started (Fig. 3C). Analysis of several hundred differentiating spores of M. xanthus wild type cells revealed not a single cell that underwent morphological differentiation without formation of lipid bodies highlighting the importance of this cellular process for spore development and cell fate determination. Conversely, many of the vegetative cells found at the basal zone of the fruiting bodies lacked inclusions (Fig. 3E, 3F) and concomitantly did not show morphological signs of differentiation further supporting a link between these two processes. As this part of the fruiting body contains a high number of cells that eventually differentiate into peripheral rods, we tested whether the cells lacking the inclusions possibly are these specifically differentiated, albeit vegetative-like looking cells. Therefore, we isolated peripheral rods (O’Connor and Zusman, 1991c;Tzeng and Singer, 2005) and examined thin sections of these cells in the electron microscope. As can be seen in Fig 3G, differentiated peripheral rods do not form lipid bodies that are characteristic for cells that eventually differentiate into spores. Consequently, lipid body formation is the first morphologically detectable sign that indicates cell fate determination during developmental differentiation in M. xanthus.
Interestingly, the formation of lipid bodies during fruiting body and spore development showed a sequence of highly coordinated morphological changes. Although these changes were not completely synchronized in the spores of a single fruiting body, the majority of spores showed the following sequence of changes: until about 24 h into the spore development, the inclusions maintained their initial pre-spore size of 50-200 nm and nearly completely filled the interior of the spore in cross sections (Fig. 3A). During the following hours, the size and number of the inclusions started slowly to decrease (Fig. 3B), until after about 72 h only remnants of the original structures could be detected (Fig. 3C), that upon spore maturation completely disappeared (96 h or longer; Fig. 3D). Closer inspection of cross sections of the disappearing inclusions revealed a prominent circle of dark masses surrounding each individual structure (Fig. 3C). These structures may represent cytoplasmic complexes that bind to the surface of the inclusions and may play a role in the mobilization of their content during spore formation.
As our results indicated that cell fate during developmental differentiation in M. xanthus may be linked to the cell’s ability to form lipid bodies, we decided to further characterize this relationship. Our basic hypothesis thereby was that if lipid body formation plays a role in cell fate determination then cells should show defects in forming fruiting bodies and/or spores if they are impaired or unable to form lipid bodies. Vice versa, cells that had formed lipid bodies before the onset of differentiation may be able to generate more fruiting bodies and/or spores than “unprepared” control cells. Four different sets of experiments were performed to test these hypotheses.
First, we isolated highly purified peripheral rods (PRs), differentiated cells that do not contain or form lipid bodies and that, according to these considerations, should be impaired in their ability to differentiate into spores. When 1×1010 peripheral rods, a number of cells that always facilitates differentiation and spore formation in the undifferentiated wild type cells, were spotted onto starvation agar, they were unable to form fruiting bodies and spores (Fig. 4C). Although these data confirmed our hypothesis, they have to be viewed in the light that because the PRs are differentiated cells they may have lost their ability to form fruiting bodies and/or spores due to physiological and genetic changes unrelated to their inability to form lipid bodies.
To demonstrate that the lack of lipid body formation was most likely responsible for the inability to differentiate into spores, we repeated these experiments with an M. xanthus mutant strain that is impaired in the formation of the most abundant and important fatty acid (FA) family in myxobacteria. M. xanthus DK5624 carries a deletion of the bkd locus and a disruption in the mvaS gene, resulting in only residual activity in pathways leading to the formation of isovaleryl-CoA, the precursor for iso-FAs (Bode et al., 2006b). When 1×1010 undifferentiated DK5624 cells were spotted onto starvation agar they failed, like the peripheral rod cells, to form fruiting bodies and/or spores (Ring et al., 2006) indicating that lipid body formation might be a crucial event enabling cell differentiation.
We then tested the opposite scenario using cells that had already formed lipid bodies before the onset of the differentiation experiment. To induce lipid body formation in these undifferentiated wild type cells we treated them with low concentration of kanamycin (20-30 μg/ml) for 3 h and confirmed the formation of the lipid bodies using light microscopy. Initial attempts to use these “primed” cells immediately for sporulation assays failed possibly due to the toxicity of the added antibiotic. Therefore, we removed the drug by replacing the kanamycin-containing medium with fresh CTT and incubated the cells for an additional 1h at 32°C. After confirming that the majority of these cells still contained their formed lipid bodies, we used them for fruiting body assays. As can be seen in Fig. 4C these “primed” lipid body-containing cells reproducibly produced more fruiting bodies than the untreated control cells. To quantify this effect, we used the Volocity4 image analysis software (Improvision Inc., Waltham, MA) to measure the area of the spots that were covered by fruiting bodies. Comparisons between spots of control cells and “primed” cells showed that the latter covered 14% more area of the spots with fruiting bodies under these condition (Fig. S1). To examine whether the higher number of fruiting bodies also resulted in a higher total number of spores, we attempted to count the spores from spots of each of these two cell types. These experiments showed that the fruiting bodies formed by the kanamycin-pretreated cells contained viable spores. However, the number of germinating spores from these fruiting bodies was about 1000-fold lower than for the wild type ones. Upon closer inspection, it became clear that the reason for this difference was most likely not a different number of spores per fruiting body but the fact that the fruiting bodies formed by kanamycin-pretreated cells were covered with an extremely resilient layer of carbohydrates effectively preventing spore liberation. Despite using various dispersal techniques such as ultrasonication and mechanical disruption we were unable to liberate the spores without compromising their ability to germinate. Nevertheless, electron micrographs indicate that the spore content and density in these different fruiting bodies is comparable suggesting that the pretreated cells produce more spores further indicating a link between lipid body formation and cell fate determination during developmental differentiation. Moreover, the observation that pretreated cells are able to protect their fruiting bodies with a stronger carbohydrate layer may indicate that the “primed” cells had excess metabolic reserves available during fruiting body formation.
Finally, we tested the effect of “slow” starvation on lipid body formation and fruiting body development (Fig. S2). For these experiments CTT-grown cells were washed in CF medium and spotted onto CF agar with or without 0.2% citrate (Hagen et al., 1978). Every 6 h cell samples were taken from the spots, stained with nile red and examined in the fluorescent light microscope for the presence of lipid bodies. While very low amounts of nutrients in the CF medium had no effect on the timing of lipid body and fruiting body formation, the presence of citrate reproducibly delayed the formation of lipid bodies by about 24 h (Fig. S2 lower panel). However, importantly these delayed cells, again, only initiated fruiting body formation after first successfully forming lipid bodies indicating that these two processes are intimately linked and that lipid body formation is a prerequisite for developmental differentiation.
In order to characterize the observed inclusions further, we developed a protocol to isolate these structures based on their unique property to float after centrifugation (see experimental procedure). Therefore, bacterial cells were grown in liquid CTT medium and formation of inclusions was induced either by adding kanamycin to the medium (20-30 μg/ml) or by incubating the cells under anaerobic conditions. Total bacterial cell homogenates containing inclusions were obtained by means of ultrasonication and the structures were purified through repeated rounds of centrifugation. After each centrifugation the inclusions formed a bright yellow to orange-coloured layer floating on top of the centrifuge tube (Fig. 5A). The color most likely results from the incorporation of lipophilic pigments like carotenoids (Elias-Arnanz et al., 2008) or DKxanthenes (Meiser et al., 2006) into the highly lipophilic lipid bodies. These inclusions were collected and after three to four cycles of centrifugation, samples were obtained that were devoid of cellular debris and considered pure by electron microscopy using negative staining and thin sectioning (Fig. 5B, D). For some experiments, the purified inclusions were centrifuged once more using a buffer that contained up to 2 M NaCl in order to remove non-specifically attached proteins and membrane fragments. Although this treatment reduced somewhat the small amounts of surface-associated material visible in thin sections (Fig. 5E) repeated salt treatments were not possible because they destabilized and lysed the isolated inclusions resulting in a bright coloured oil film floating on top of the centrifuge tubes.
Inclusions that had been isolated and purified from kanamycin-induced M. xanthus wild type cells were used for the chemical analysis of their lipid composition. From the ultra-structural data we were expecting that their interior would be composed of neutral lipids whereas phospholipids would be the constituents of the outer membrane layer (see Figs. 1D, 5D, 5E). Indeed, analysis of intact lipids showed a high amount of neutral lipids like triacylglycerols (TAGs) or diacylglycerols (DAGs) with ether derivatives of these compounds (e. g. 1,2-di-(13-methyltetradecanoyl)-3-(13-methyltetradecyl)glycerol [TG-1] (Ring et al., 2006)) being the main components (Table 1). This is in accordance with i15:0-O-alkylglycerol (OAG) being a major compound in the analysis of FAs after methanolysis of the purified inclusions (Table S4). In order to determine the phospholipid species, which most likely are the components of the lipid monolayer surrounding the bodies, we analyzed the lipid bodies by direct injection into an electrospray ionization mass spectrometer (ESI-MS) and analyzed the fragmentation pattern of phospholipid species. The two dominant compounds are 1-i15:0- alkyl-2-i15:0-acyl phosphatidylethanolamine (AEPE) and 1-i15:0-vinyl-2-i15:0-acyl phosphatidylethanolamine (VEPE) which are 1-alkyl- and 1-vinylether derivatives of the most abundant myxobacterial phosphatidylethanolamine (PE) (Fig. 6) as described previously (Ring et al., 2006).
The composition (Table 1) and structure (Fig. 6) of the main components of the different lipid classes was determined by a detailed analysis of the GC/MS fragmentation pattern according to published procedures (Hsu and Turk, 1999). However, for the TAG species no positional information of the different acyl chains was possible. Additionally, several minor peaks of the different lipid classes could be detected by GC/MS, which could not be elucidated due to overlapping or very small peaks. Overall, up to 30 TAG species and 10-20 DAG and monoacylglycerol (MAG) species could be detected. The distribution of their major FAs is similar to the whole-cell FA profile (Table S4). The differentiation between iso- and straight-chain (sc) FAs resulted from the comparative analysis of DK1622 (wild type) and DK5614, which produces only traces of straight chain (sc)-FAs as previously published (Bode et al., 2006a). Several TAG species were not present in the DK5614 samples indicating that these molecular species are composed at least partially of sc-FAs (data not shown). This especially allows the differentiation between iso-15:0 which is the major FA in M. xanthus and 15:0, which is only present in minute amounts. Taken together these results indicated that the inclusions are indeed lipid bodies filled with neutral lipids (mostly TAGs) surrounded by a phospholipid layer.
In order to correlate the accumulation of inclusions with changes in the lipid profile we analyzed cells grown under various stress conditions and cells undergoing fruiting body formation by GC/MS. As expected an overall increase of neutral lipid could be observed comparing vegetative cells with cells grown anaerobically, with kanamycin or undergoing fruiting body formation (Table 1, Fig. 7). Moreover, detailed analysis of neutral lipid composition under anaerobic conditions showed that TAG accumulation did not require cell growth (Fig. 8).
In addition, lipid profile analysis confirmed the ultrastructural results indicating that peripheral rods do not form lipid bodies. As expected from the already known FA profile (Ring et al., 2006), no etherlipids could be detected in isolated purified peripheral rods (Tzeng and Singer, 2005) (Table 1). Moreover, the only detectable neutral lipid species in peripheral rod cells was 1,2-iso-15:0-diacylglycerol (DAG), which is a major neutral lipid species in vegetative cells. In contrast, the spores contained, as expected, large amounts of different neutral lipids (Table 1).
As formation of lipid bodies during development seemed to be important for spore formation, we also analysed the lipid profile of DK5208, which is defective in production of the csgA encoded C-signal protein (Kaiser, 2004;Sogaard-Andersen, 2008) resulting in defective aggregation, fruiting body formation, and sporulation. Although several neutral lipids have been found in DK5208, their amount was much smaller than in wild type cells. Moreover, much less of the developmental specific changes observed in the wild type (e.g. high amounts of etherlipids) could be detected in this mutant strain (Table 1). Similarly, analysis of DK5624 which is impaired in the production of iso-FAs revealed an even more severe phenotype with no TAGs but only DAGs and MAGs being produced (Table 1). The latter result is in accordance with the previously described loss of TG-1 production in DK5208 and DK5624 and several other developmental deficient mutants (Ring et al., 2006). TG-1, an etherlipid derivative of the TAG tri-iso15:0-acyl glycerol, was shown to be able to rescue sporulation of some developmental mutants which are impaired in the production of the major fatty acid iso15:0 and therefore have only trace amounts of TG-1 (Ring et al., 2006).
The isolated and purified inclusions of M. xanthus possessed a spherical shape and had diameters that ranged from 50 nm to about 0.5 μm (Fig. 5B, 5D). This large variability in size was also observed during in vivo measurements of inclusions, which could be labelled with nile red (Fig. 1B) once more confirming the lipid nature of the inclusions. Moreover, ultrasonication of large inclusions did not result in the formation of smaller structures. Therefore, the observed variability was most likely not an isolation artefact but reflected the actual size distribution of the inclusions inside the living cells. In electron micrographs of thin sections of whole cells or isolated inclusions the content of the structures was usually not preserved and the inclusions appeared empty (Fig. 1F, G; 3A, B). These results were obtained regardless whether conventional chemical- or cryo-embedding procedures with or without lipid-retaining osmium tetroxide (Zingsheim and Plattner, 1976) were used. However, the content of the inclusions could be preserved by adding lipid-retaining malachite green during the fixation process, especially when the overall fixation times were extended (Fig. 1D, E) (Teichman et al., 1974). Regardless of the preservation of their content, all lipid bodies were surrounded in thin sections by an electron-dense layer of about 4-5 nm thickness (Fig. 1D; 5D, E). Both, the appearance and the dimension of this membrane were consistent with data reported for other pro- and eukaryotic lipid bodies (Murphy, 2001) and may indicate that these structures are surrounded by monolayer membranes.
SDS-PAGE identified three proteins associated with isolated lipid bodies, two major ones, MXAN_5582, MXAN_0412, and one minor one, MXAN_2536 (Fig. 5C and Supporting Information). MXAN_5582 is a 37 kD protein that showed similarity to flotillin-like and antifreeze proteins that in eukaryotes are often lipid raft-associated. MXAN_0412, a 28 kD protein, is a member of a conserved family of proteins termed DUF124 that is found in many organisms (pfam.sanger.ac.uk/family?acc=PF01987). Finally, MXAN_2536 showed high similarity to several long-chain fatty acyl CoA ligases (FACL), which play a pivotal role in the transport and activation of exogenous fatty acids prior to their subsequent degradation or incorporation into phospholipids. To test any role of these proteins in the formation of lipid bodies, we generated the plasmid insertion mutant strains MR5582 (MXAN_5582kan), MR0412 (MXAN_0412kan), and EB2536 (MXAN_2536kan). However, no difference in lipid body formation or TAG accumulation could be observed when these strains were compared with the wild type under various stress conditions. As TAG biosynthesis has been shown to depend on diacylglycerol acyl transferases (DGAT) in prokaryotes, we then disrupted the only candidate gene encoding such an enzyme in M. xanthus, MXAN_1127, which showed 25% identity/45% similarity to the wax ester synthase/acyl-CoA:diacylglycerol acyltransferase (WS/DGAT) of Acinetobacter baylyi ADP1 that has been functionally characterized (Kalscheuer and Steinbüchel, 2003;Stöveken et al., 2005). However, again no phenotype was observed with respect to lipid profile, lipid body formation or development.
Cellular differentiation, the ability of cells to stably change the expression pattern of their genes, is an important adaptation strategy of organisms to deal with environmental changes. A particularly intensive-studied model of bacterial cellular differentiation is the formation of spores. During this differentiation process actively growing vegetative cells convert into dormant spores, cells that are resting and largely resistant to environmental stress. Although certain aspects of this process such as the regulation of gene expression and the role of transcription factors are relatively well known, other aspects such as the question how the cells metabolically support the necessary complex cellular differentiation are only poorly understood. This question is particularly interesting in the light of the fact that spore formation is initiated upon starvation and that the concomitant nutrient limitation complicates any energy-dependent cellular differentiation processes. While these circumstances are challenging enough, the bacterium M. xanthus has to cope with one further extra complication. Upon starvation, these social bacteria do not immediately form spores, but instead initiate a complex behavioral program leading to the formation of fruiting bodies, mound-shaped multicellular cell aggregates. Only when the cells have begun to aggregate, do they start to differentiate into spores making it all the more challenging for the cells to sustain the execution of both the behavioral and the differentiation program under severe nutrient restriction.
So, how does M. xanthus accomplish this feat? It appears, that lipids are a crucial part of the answer. And although lipids are, by far, the most widespread storage compounds in eukaryotes (Murphy and Vance, 1999;Murphy et al., 2000), they are only rarely found in prokaryotes (Alvarez and Steinbüchel, 2002) making M. xanthus one of the few known prokaryotes to do so. Given the duration and the complexity of the differentiation process of M. xanthus, lipids are, in fact, ideal storage compounds. In part, because their oxidation releases more energy than carbohydrates or proteins and, in part, because they are water insoluble making them highly compact and, therefore, extremely efficient storage compounds that can provide both energy and metabolites. Not surprisingly, M. xanthus uses TAGs not only during its developmental cycle (Fig. 3) but as general storage compounds under nutritional stress during vegetative growth (Fig. 1). Moreover, our data indicate that the processes controlling lipid body formation may be at the core of cell fate determination. These processes appear to use metabolic parameters predictive for the cell’s ability to successfully differentiate to initiate two very different developmental programs: cells that do not form lipid bodies end up becoming peripheral rods, while cells that form lipid bodies invariably develop into spores. It appears that during spore formation, lipid bodies have multiple functions. Some of these functions are obviously linked to their capacity to provide energy storage compounds. Consequently, they may fuel cell movements during development, the extensive remodelling of the cell shape, the synthesis of a diploid chromosome (Tzeng and Singer, 2005) and the synthesis, secretion, and assembly of an elaborate multi-layered spore coat. However, lipid bodies may play additional, less obvious, roles that are not directly linked to their energy storage function. One of these roles appears to involve a minor lipid species that is indispensable for spore germination (Ring et al., 2006). Intriguingly, by providing these signalling molecules, lipid bodies may play important roles in developmental processes that occur long after the structures themselves have disappeared. Therefore, lipid bodies appear not only to fuel the cellular differentiation, but their formation appears to represent the first known morphological sign indicating cell fate during development, while their content enables the cells achieving this goal.
Various stress conditions such as nutritional step-down, anaerobiosis, or antibiotic treatment that disrupted or even abolished cell growth induced the formation of lipid bodies in M. xanthus (see e. g. Fig. 8). Intriguingly, a more detailed analysis of this phenomenon revealed that the number and cellular location of the formed lipid bodies varied and strongly depended on the severity of the stress (Fig. 1, ,2).2). Mild stress led to the formation of only a few such structures that were mostly found close to the cell poles (Fig. 2B). In contrast, severe stress resulted in the simultaneous formation of many lipid bodies that almost completely filled the cell (Fig. 1F, G and and2).2). Therefore, lipid body formation in M. xanthus is a controlled, yet flexible, process that allows the cells to form the appropriate number of lipid storages necessary to survive. The observation that lipid body formation under more “physiological” low stress conditions occurred preferentially at the cell poles may indicate that enzymes and/or structures involved in the formation of these inclusions are polarly arranged. However, under severe stress the cells were able either to relocate these enzymes and/or structures or to synthesize new ones, because lipid body formation occurred at many locations simultaneously. In fact, the speed with which these lipid bodies occurred was so fast that it appears unlikely that the structures could be formed at the cell poles and then re-distributed. Although, we cannot, at the moment, rule out this possibility. A similar flexibility might exist for the cellular origin of the lipids. As the cells can form at the same rate a few or many lipid bodies, it is likely that the cells use different starting material and pathways for these syntheses. We postulate that during mild stress the cells appear to form lipid bodies mainly through the conversion of intermediary metabolites. However, this explanation can hardly account for the massive formation of lipid bodies during severe stress. Under these circumstances, the cells most likely convert biomass such as membranes and proteins into lipids. This process seems particularly important for fruiting body formation of M. xanthus (Shimkets, 1999;Kaiser, 2004). Not only is this developmental differentiation induced by starvation and accompanied by massive formation of lipid bodies, but upon conversion of rodshaped cells into round spores roughly two-thirds of the cellular membranes are no longer needed (Shimkets and Seale, 1975). Consequently, the cells appear to recycle these materials by converting them into lipids that later are used to fulfill the metabolic requirements of the cells during differentiation. Similarly, TAGs are produced in the postexponential growth phase of Streptomyces and their accumulation might be linked to antibiotic production, which is an important part of development in Streptomyces (Olukoshi and Packter, 1994; Arabolaza et al., 2008). The importance of lipid body formation in M. xanthus is evident in strains such as DK5208 (Sogaard-Andersen, 2008) that produce little amounts of neutral lipids (Table 1) or strains that are defective in 1,2-di-(13-methyltetradecanoyl)-3-(13-methyltetradecyl)glycerol [TG-1] production (Ring et al., 2006) and, as a consequence, are developmentally impaired. Moreover, addition of TAGs or fatty acids can at least partially complement these defects (Ring et al., 2006;Rosenbluh and Rosenberg, 1989;Rosenbluh and Rosenberg, 1990). Here we show, that the same holds true for further strains. In particular, DK5624, a strain that is impaired in its ability to synthesize iso-fatty acids shows severe developmental defects, while the lipid body-free peripheral rods cannot differentiate at all. Can the formation of lipid bodies be correlated with the complementation of DK5624 after the addition of isovaleric acid (IVA)? Although this was not explicitely mentioned previously, we now know that two lipid species (RT 7.072 and RT 16.575; see Bode et al., 2006b) are in fact iso15:0 DMA and iso15:0 OAG, respectively. The production of these two etherlipid-derived compounds is even more significant during fruiting body formation as one would expect (unpublished data). However, as the major etherlipid compounds found in M. xanthus during fruiting body formation are neutral lipids as shown here, which can only form lipid bodies, we are quite confident that complementation of DK5624 with IVA indeed restores lipid body formation. In contrast, “kanamycin-primed” cells have more lipid bodies resulting in more fruiting bodies and higher numbers of spores; while slowly starving cells delayed lipid body and fruiting body formation. Not only do these experiments corroborate the link between lipid body formation and differentiation, but they also indicate a causal relationship between the two processes. However, these results also raise a number of important questions.
What induces lipid body formation and how is this linked to differentiation? Obviously, metabolic imbalances induce lipid body formation. But it appears that the severity and, more importantly, the duration of starvation decide the cell’s fate. The importance of the severity becomes clear from the observation that there exists a rather defined nutritional threshold controlling differentiation. Only the complete absence (TPM agar) or very small amount of nutrients (CF and CF plus citrate agar) permit differentiation, while the addition of casitone suppress the process. The fact that transient starvation induces lipid body formation but does not trigger differentiation points to the importance of the duration. It takes at least six hours of starvation before the first differentiation-specific responses can be detected. Therefore, it is tempting to speculate that M. xanthus like Bacillus subtilis uses metabolic check points to decide whether to progress towards differentiation or to terminate the process and that their may even exist an as of yet un-identified point of no return upon which differentiation becomes inevitably. What makes these processes particularly interesting in M. xanthus is the fact that one subset of cells, the peripheral rods, do not form lipid bodies despite starving. So, how can cells respond so radically different to the same stimulus? The easiest answer may be that the cells in order to differentiate rely on a multitude of signals both cell-autonomous and cell-nonautonomous and that upon starvation the cells start to control each other fates through these signals. In responding to these signals, cells that eventually differentiate into peripheral rods appear to selectively suppress the genes necessary to form lipid bodies. As a consequence, peripheral rods even at high concentration are unable to form fruiting bodies or spores. Interestingly, these results differ from earlier observations that indicated that peripheral rods are only severely impaired in their ability to form fruiting bodies (O’Connor and Zusman, 1991b). Although it is currently unclear what accounts for these different results, it may be because we used highly purified isolated peripheral rods.
Ultimately of course, the question is what role does lipid body formation play during differentiation? Are lipid bodies a consequence or a cause of the process? Or to put it more simply, can cells differentiate without lipid bodies and, vice versa, do cells containing lipid bodies always differentiate? Given the fact that we have not identified lipid body formation-specific genes we cannot answer these questions with certainty. However, all of our data point to a tight correlation between lipid body formation and differentiation. All strains that are defective in fatty acid synthesis are also defective in fruiting body and spore formation. Conversely, cells that possess lipid bodies at the onset of starvation appear not only to differentiate but also to do more successfully forming more spores; while cells that delay lipid body formation also delay developmental differentiation (slowly starving cells). Moreover, electron microscopic evaluation of many hundreds of spores, have never lead to the observation of a single differentiating cell that did not contain lipid bodies. Together these data indicate that lipid body formation is an essential step during differentiation constituting the first known morphological sign indicating the fate of a cell. Therefore, understanding this process will be crucial for our understanding of cell fate determination in M. xanthus. In particular, it will be important to elucidate the metabolic parameters used by the cells to decide whether to become spores or peripheral rods. Similar complex population dynamics can be observed in cultures of B. subtilis during sporulation-related cannibalism (Gonzalez- Pastor et al., 2003). In this connection it is interesting to note that recently the sporulation specific accumulation of an unusual diacylglycerol ether in M. xanthus has been described (Ring et al., 2006). This etherlipid, named TG-1, could restore developmental and sporulation defects of mutants with a reduced level of iso-fatty acids. Although the cellular location of the etherlipid was not determined, here we could show that TG-1 is a major compound of isolated lipid bodies (Table 1). Therefore, lipid bodies in M. xanthus could possess dual functions during differentiation by providing both energy and developmentally important signal molecules. In this capacity, M. xanthus lipid bodies would resemble eukaryotic adipocytes that also provide at the same time energy (lipids) and signal molecules (lipid-soluble hormones) (Badman and Flier, 2007).
A still unresolved question is the molecular mechanism by which bacteria actually form lipid bodies. Several different models have been proposed to describe the events that lead to the initiation and maturation of these structures (Wältermann and Steinbüchel, 2005). According to one hypothesis, the process starts with the bifunctional enzyme WS/DGAT synthesizing an oleogenous layer on the surface of the cytoplasmic membrane. This layer coalesces into small droplets that over time fuse into larger structures, the so-called lipid prebodies. Upon further synthesis the prebodies are eventually released from the membrane maturing into lipid bodies (Wältermann et al., 2005). An alternative hypothesis proposes a cytoplasmic micelle-based assembly process as suggested for formation of polyhydroxyalkanoate (PHA) granules in bacteria (Jurasek and Marchessault, 2004;Stubbe and Tian, 2003). Here, a coat of lipid layer-embedded enzymes synthesizes lipids that form the growing core of the micelle. Finally, lipid body formation has been suggested as a result of membrane budding events similar to lipid body formation in plants (Murphy and Vance, 1999;Murphy et al., 2000). As in the lipid prebody model, the involved enzymes are bound to the cytoplasmic membrane, but the initial accumulation of lipids occurs between the two leaflets of the membrane. Further synthesis of lipids eventually forces the pinching-off of small vesicles that, over time, mature into large lipid bodies. Although our data are not conclusive enough to decide which of these models correctly describes lipid body formation in M. xanthus, most of our observations challenge basic tenets of the lipid prebody- and the micelle-hypotheses. For instance, we have never been able to detect lipid prebody-like structures in our embedded cells such as those described in Rhodococcus opacus PD630 (Alvarez et al., 1996;Wältermann and Steinbüchel, 2005). This is remarkable because we used cryo-embedding protocols that are generally known to be better at resolving cellular fine structure and capturing dynamic processes (Baumeister and Steven, 2000). Moreover, we found in our preparations many extremely small lipid bodies measuring only a few nanometers in diameter (Fig. 5B). As ultrasonication of larger lipid bodies did not result in the generation of such miniature structures, it is very unlikely that they are an artefact of the isolation. Given the fact that these structures have a perfect spherical shape and a complete membrane envelope, we hypothesize that they are indeed lipid bodies. This interpretation would explain why we are unable to detect the actual formation of lipid bodies inside the cells. The newly-formed structures are simply too small to be visible on thin sections that are substantially thicker than these extremely small lipid bodies. How then, if not via lipid prebodies, are these structures made in M. xanthus? The fact that we were unable to detect oleosin- or phasin-like proteins associated with the lipid bodies and the lack of obvious homologs of these proteins in the M. xanthus genome strongly argues against a micelle-based mechanism. Oleosins (Capuano et al., 2007) and phasins (Pötter and Steinbüchel, 2005) are highly abundant proteins found in membranes of plant lipid bodies and bacterial PHA granules, respectively. Moreover, the fact that the lipid bodies are predominantly located at the two cell poles under mild stress suggests that they are formed at this location. Indeed there is increasing evidence that the cell poles in bacteria are involved in several different processes and that not only proteins (Mignot et al., 2005;Leonardy et al., 2007) are located specifically at the cell poles but also distinct lipid classes (Mileykovskaya and Dowhan, 2005;Mileykovskaya and Dowhan, 2000). Therefore, we suggest that a membrane budding-based process occurring preferentially at the cell pole is involved in lipid body formation in M. xanthus.
As mentioned above, only three proteins, two major ones (MXAN_0412, MXAN_5582) and one much less abundant one (MXAN_2536) were consistently associated with the isolated lipid bodies. Moreover, no specific function in lipid body formation could be assigned to any of these proteins, because deletions of the corresponding genes had no effect on the number and distribution of lipid bodies. These observations raise two important questions. First, do these co-purified proteins play a role in lipid body formation, and, secondly, are their other lipid body-specific proteins in M. xanthus?
While the association of the flotillin-like MXAN_ 5582 and the putative long chain fatty acyl CoA ligase MXAN_2536 (FACL) with the lipid bodies was not too surprising, the presence of MXAN_0412 was somewhat puzzling. In particular, because an X-ray-based model of MXAN_0412 did not reveal prominent hydrophobic features (Fig. S3). However, detailed analyses indicated that the Cys184 of MXAN_0412 might be palmitoylated thereby explaining its lipid body-association (Figs. S3-S6). As the M. xanthus genome lacks paralogs of MXAN_0412 and MXAN_5582, the lack of a detectable lipid body-phenotype in mutant strains indicates that these proteins are most likely not essential and only co-purified. This is further supported by qRT-PCR analysis showing that the transcription of the two corresponding genes does not correlate with lipid body accumulation (Fig. S7). The situation appears to be different for the FACL MXAN_2536, which is most likely involved in the fatty acid metabolism. Here, functional redundancy may account for the lack of a mutant strain phenotype. As at least six other putative FACLs are present in the genome (MXAN_2959, MXAN_3518, MXAN_1528, MXAN_7148, MXAN_1573, MXAN_0225), which show identities and similarities ranging from 26-35% and 40-51%, respectively. In fact, such redundancies are often observed in M. xanthus as there is evidence for multiple signaling or biochemical pathways leading to the same outcome (Bonner et al., 2006;Bode et al., 2009). This phenomenon is particularly common in the fatty acid metabolism (Curtis and Shimkets, 2008) and makes the construction of loss-of-function mutants extremely difficult. If MXAN_2536 is involved in lipid body formation the question arises whether other such proteins exist. We used bioinformatics to identify MXAN_1127 as a potential bifunctional wax ester synthase/acyl-CoA:diacylglycerol acyltransferase (WS/DGAT). DGATs catalyze the final esterification of diacylglycerol to TAG and are the only enzymes uniquely involved in TAG biosynthesis (Michal, 1999). Although DGAT activity and putative WS/DGATs have been found in different bacteria (Kalscheuer and Steinbüchel, 2003;Stöveken et al., 2005;Holtzapple and Schmidt-Dannert, 2007;Kalscheuer et al., 2007;Arabolaza et al., 2008), the disruption of MXAN_1127 had no influence on the cell’s ability to form lipid bodies indicating that this protein is either not a WS/DGAT or can be functionally complemented. Probably not surprising, the transcription of MXAN_1127 did not correlate with lipid body accumulation in qRT-PCR; a result that is in contrast to the findings for the WS/GDAT and DUF124 homologue ACIAD1953 (24% identity/40% similarity to MXAN_0412) of A. baylyi ADP1 that showed a clear correlation (Fig. S8). Moreover, the heterologous expression of MXAN_1127 in Pseudomonas putida did not result in lipid body formation in this bacterium either; again, indicating that MXAN_1127 may not be a functional WS/DGAT or is at least not active in P. putida. Interestingly, the formation of TAGs by WS/DGAT enzymes might not be the only pathway used to synthesize these compounds, because in some bacteria (e.g. Rhodococcus opacus) disruption of the corresponding genes does not result in TAG-negative mutants (Wältermann et al., 2000). Consequently, the identification of proteins that are involved in the synthesis of the lipids and/or the formation of lipid bodies will be important goals of future experiments. In M. xanthus, both genome-wide genetic screens and a better enrichment of minor, lipid body-associated proteins offer feasible approaches to identify these proteins. Ultimately, the characterization of such proteins should help to understand the processes involved in the regulation and formation of lipid bodies and the roles these structures play in the physiology, metabolism, and development of this and other bacteria.
M. xanthus wild type strain DK1622 and its mutants (Table S1) were grown at 30°C or 32°C in CTT medium (1% casitone, 1 mM K2PO4, 8 mM MgSO4, 10 mM Tris/HCl, final pH 7.6) on a rotary shaker. In addition, cells were also cultivated on CTT medium that was solidified with 1.5% agar. For induction of lipid body formation, exponentially growing cells of DK1622 were transferred to CTT medium containing 20 or 30 μg/ml kanamycin and the culture was grown for an additional 2-6 h. Alternatively, exponentially growing M. xanthus cells were shifted from aerobic to anaerobic culture condition to induce lipid body formation. Therefore, 15 or 50 ml conical Falcon tubes were filled up to the top with liquid CTT medium containing the M. xanthus cells and were placed back into the rotary shaker and incubated further at 32°C and 250 rpm for 2-24 h. Starvation-induced formation of fruiting bodies and spores was studied using TPM medium, which is identical to CTT medium except that it lacks casitone. In slow starvation experiments the cells were incubated on CF agar with or without 0.2% sodium citrate (Hagen et al., 1978).
To study the development of fruiting bodies and spores, wild type cells of an exponentially grown culture were harvested by centrifugation, washed twice with TPM medium and resuspended in TPM medium to a density of 5 × 109 cells ml-1. Multiple aliquots (20 μl) of this suspension were spotted in a circular pattern on a large TPM agar plate (150 mm diameter) and incubated at 32°C for a total of six days. Developing fruiting bodies were examined every day under a stereomicroscope to monitor the progress of the formation. In addition, every 24 h an agar block with a spot of developing fruiting bodies was excised from the large agar plate and processed for electron microscopy. Finally, samples were taken every 24 h to monitor the conversion of vegetative cells into spores by observing the highly refractile spores using a light microscope.
For the analysis of TAG accumulation during fruiting body formation large scale fruiting body formation was performed as described (Ring et al., 2006). Cells were scraped off the agar and the resulting cell pellets were stored at -20°C. For the sporulation assays using peripheral rods, vegetative cells were spotted onto starvation medium and incubated for 5 days. A sterile spatula was used to scrape off the thin layer of peripheral rod cells that surrounded the initial cell spot. The rod cells were re-suspended in TPM medium and their density measured with a spectro photometer. After centrifugation the peripheral rods were resuspended at a concentration of 5 × 109 or 1 × 1010 cells ml-1, spotted onto fresh TPM agar plates and incubated at 32°C for 6 days.
To test whether kanamycin-induced lipid body formation prior to the assay influences the total number of fruiting bodies, wild type cells were grown in 2.8 l Fernbach flasks containing 450 ml CTT medium. After the culture reached a density of OD600≈0.6, 50 ml of the culture was transferred into sterile 500 ml Erlmeyer flasks (five flasks). One of these flasks was used as a control, while 20 or 30 μg/ml kanamycin was added to the other flasks. After further incubation for 3 h at 32°C, the kanamycin-treated cells were harvested by centrifugation and resuspended into 50 ml fresh CTT medium without kanamycin. Upon incubation for a further hour, all cells were harvested, washed once with sterile TPM medium, re- suspended in TPM at a concentration of 5 × 109 cells, spotted onto fresh TPM agar plates and incubated at 32°C for 6 days.
Spores were quantified according to standard procedures. Briefly, individual fruiting bodies were scraped off the TPM agar with a sterile spatula and transferred into fresh TPM buffer. The samples were sonicated in a water bath for 10 min or until all visible aggregates were dispersed. Samples were diluted serially in TPM, mixed with soft agar and spotted onto CTT agar plates. The number of germinating spores was determined by counting the number of colonies after six days at 32°C.
M. xanthus DK1622 cells were grown either in liquid or on agar-solidified CTT medium. After induction of lipid body formation through kanamycin treatment or anaerobiosis, the cells were harvested by centrifugation or by scraping them off the agar surface. After harvesting, cells were resuspended in 10 mM Tris/HCl (pH 7.5) and broken on ice by sonication using a Branson Sonifier (3/8” flat tip, output intensity 5, 50% duty cycle). Cellular debris was removed by centrifugation (15 min at 50 000 × g) and the resulting homogenate was centrifuged in an ultra-centrifuge (1 h at 400 000 × g). The yellow- to orange-coloured lipid bodies floating on top of the centrifuge tube were collected, gently resuspended by shaking or mild sonification (3/8” flat tip, output intensity 1, 50% duty cycle), and the ultracentrifugation step was repeated at least once more. After centrifugation, the floating lipid bodies were collected and either directly used for analysis or resuspended in 10 mM Tris/HCl (pH 7.5) containing 2 M NaCl and once more ultracentrifuged for 1 h at 400 000 × g to remove unspecifically associated hydrophobic proteins (Tzen and Huang, 1992). The purified lipid bodies were collected and used for biochemical and ultrastructural analysis. Light and electron microscopy, and SDS-gel electrophoresis was used to judge the purity of the lipid body preparations.
Isolated purified lipid bodies were fixed in 2% glutaraldehyde in 20 mM potassium phosphate buffer (pH 7.5) on ice for three hours. After several brief rinses with the phosphate buffer, the lipid bodies were postfixed in 1% OsO4 in the buffer on ice overnight. The lipid bodies were then dehydrated through a graded acetone series, infiltrated overnight at room temperature in an acetone-Epon 812 mixture (1:1), embedded in fresh Epon 812, and polymerized at 60°C for 36 h. To reduce the loss of the lipid content of the isolated lipid bodies during the dehydration process, some samples were first fixed with 2% glutaraldehyde in 20 mM potassium phosphate buffer (pH 7.5) containing 0.1% malachite green (Teichman et al., 1974), before they were washed with buffer, postfixed with osmium tetroxide/malachite green and embedded in Epon according to standard procedures.
For lipid body labeling, a stock solution of nile red was prepared in ethanol (0.5 mg ml-1), sterile filtered, and kept protected from light. Staining of lipid bodies was either done using fixed or unfixed, living bacteria. For the staining of fixed cells, the glass slide with the bacterial suspension was placed on a hotplate at 42°C until all the liquid was evaporated, stained for 30 min with the nile red stock solution diluted in ethanol (10 μg ml-1), washed briefly with water, and observed by epifluorescence. For the staining of living cells, the dye was added directly to the CTT medium of an exponentially growing M. xanthus culture at the onset of lipid body induction either through kanamycin addition or removal of air (anaerobiosis). The amount of stock solution added resulted in a 1:100 dilution of the dye, and the cells were incubated afterwards at 32°C and 250 rpm for 2-6 h. At different time points, cells were removed, concentrated by centrifugation and viewed using phase contrast and epifluorescence (Greenspan et al., 1985). As nile red fluorescence is quenched in water, residual dye in the medium did not interfere with the observation.
For light microscopy, 5-8 μl large droplets of lipid body-containing, nile red-stained M. xanthus cells were transferred to microscope slides. After placing cover slips on top, the cells were viewed with a 100/1.4x phase contrast objective on a Nikon Eclipse E800 microscope using a combination of visible (phase contrast) and blue light (excitation filter 465-495 nm, dichromatic mirror 505 nm, and long pass barrier filter 515-555 nm; Nikon filter set B-2E/C). Care was taken that the cells were not exposed to the fluorescent light before the camera settings were optimized and the phase contrast image was recorded. Pictures were taken with a CCD camera (SPOT RT Slider, Model 2.3.1; Diagnostic Instruments, Inc.) using the SPOT software package (SPOT Basic Version 4.0.9) and saved as *.tif files that were imported into Adobe Photoshop for presentation and measurement.
For electron microscopy, cells and whole fruiting bodies on agar were either high pressure frozen using the Leica EM PACT instrument and cryo-substituted as described (Hoiczyk and Baumeister, 1995) or conventionally processed. Epon-embedded samples were thin-sectioned and mounted on 200 mesh Formvar carbon-coated copper grids (Electron Microscopy Sciences). Thin sections were routinely poststained with 2% (wt/vol) uranyl acetate and lead citrate (Reynolds, 1963). Whole cells and isolated lipid bodies were negatively stained on glow discharge-treated 400 mesh carbon-coated copper grids (Electron Microscopy Sciences) using either 2% (wt/vol) uranyl acetate or 1.5% (wt/vol) sodium phophotungstate at pH 6.8 containing 0.015% glucose to promote homogenous staining. For specimen observation, a Philips CM12 was used at an acceleration voltage of 100 kV. Images were recorded on Kodak ISO 165 black and white film at nominal magnifications ranging from 500 to 52 000x.
All recorded images were imported into Adobe Photoshop and digitally processed for presentation. During the processing, suited areas of the original images were digitally cut out and assembled into multi-image figures. The original images were not digitally altered unless differences in brightness made it necessary to match individual images. In these cases the change of brightness was applied to the complete image before assembly of the final figure. No other form of digital manipulation was performed.
The captured phase contrast and fluorescence images were used to analyze both the relative total volume of lipid bodies per cell as well as their cellular location. For the first set of measurements, 200 individual cells were analyzed and the number of lipid bodies determined by counting the highly refractile lipid body droplets in the phase contrast images and cross correlating them with nile red-positive inclusions visible in the fluorescence images. Next, the diameter of the lipid bodies was measured and used to calculate the total volume of these structures in a given cell. Finally, the relative total volume of the lipid bodies was determined using the individual cell volume as 100%. For the second set of measurements, the cellular locations of 400 individual lipid bodies were determined with respect to the cell poles. Therefore, individual cells were optically “cut” in two identical halves and the distance of every lipid body from the cell pole measured. Then the relative cellular position of each lipid body was determined by dividing the distance of the lipid body from the cell pole through the half length of the cell. Finally, these relative cellular positions were used to calculate the total number of lipid bodies located within the first 10% of the cell (starting from the cell pole), the next 10%, the over next 10%, and so on. A chi-square test was performed to analyze whether statistically significant differences existed between these different groups of data and the p value was calculated to judge whether the observed cellular distribution of lipid bodies was random or not. All graphs were generated using the software program Prism.
For the quantification, light microscopic images of the developed fruiting bodies were captured on a Nikon TE200 inverted microscope using a 2.5x objective and Hofman modulation contrast. Pictures were taken with a CCD camera (SPOT RT Slider) using the SPOT basic software program, saved as *.tif files and imported into the software program Volocity4 (improvision). Representative equally sized areas of the fruiting body spots were cropped using a digitally created mask (see Fig. S1). The cropped images were saved and individual fruiting bodies were digitally identified using the intensity module. The threshold for the identification was selected so that all visually identified fruiting bodies within the mask were captured. The size of the fruiting bodies was automatically calculated and used to determine the relative total covered area using the cropped image as 100%.
Cell pellets were extracted with hexane – isopropanol (2:1, V/V). Spores were broken up by ultrasonication in the presence of glass beads. The extracts were dried under nitrogen and redissolved in hexane – isopropanol to give a concentration of 10 mg of total lipids per ml.
Analysis of fatty acids as their methyl esters was carried out as described (Bode et al., 2006b). For analysis of total lipids by high-temperature GC/MS, an aliquot of 100 μl of lipid extract (corresponding to 1 mg of total lipids) was mixed with 10 μg of the internal standard cetyl palmitate and evaporated to dryness. The residue was dissolved in 80 μl of hexane – MTBE (1:1) and allowed to react with 20 μl of MSTFA for 30 min at 37°C. 2 μl of this solution were injected into an Agilent 6890N gas chromatograph with a 5973 mass spectrometer using a pulsed splitless injection technique. The column was a SGE HT5 high-temperature polydimethyl-(5%-phenyl)-siloxane column (25 m × 0.32 mm × 0.1 μm, SGE, Griesheim, Germany), and the carrier gas was helium at a flow rate of 1.5 ml/min. Inlet and GC-MS transferline temperatures were set to 300°C. Column oven temperature was kept at 200°C for 3 min, then increased to 400°C at 5°C/min and held there for 10 min before cooling down to 200°C at 15°C/min. Quantification was made by peak areas with respect to the peak area of cetyl palmitate.
The lipid extract was diluted ten-fold with hexane – isopropanol containing 10 mM of lithium chloride. Using a syringe pump, this dilution was injected into a Thermo LTQ Orbitrap instrument equipped with an ESI ion source operating in positive ionization mode. Ions in the mass range between m/z 700 and 850 were subjected to further fragmentation at a collisional energy of 35 V. MS/MS spectra of [M+Li]+ ions of triacylglycerols and their ether analogues were interpreted according to literature (Hsu and Turk, 1999).
The protein profiles of lipid body membrane preparations of wild type cells were analyzed using Tris-glycine SDS-PAGE gradient gels (Invitrogen). Isolated lipid bodies were solubilized in 10 mM Tris buffer (pH 7.5) containing 1% SDS and the lipid body-associated proteins were precipitated using methanol/chloroform (Wessel and Flügge, 1984). After precipitation, the protein samples were boiled in SDS-sample buffer, separated on 4-20% gradient gels, and stained with colloidal Coomassie (Neuhoff et al., 1988) according to the manufacturer’s instructions (Invitrogen). Pre-stained protein marker (See Blue, Invitrogen) was used as a size standard.
Protein bands were cut out for “in-gel” digestion using an Ettan spot picker (GE Healthcare) and the gel plugs were transferred into a 96-well microtiter plate (MTP). First the gel pieces were washed with 100 μl of water for 10 min at room temperature. The plates were shaken at 200 rpm on a MTP shaker (Mixmate Eppendorf). The next washing step was performed with 100 μl of mixture of water/acetonitrile (1:1) for 10 min at room temperature. This step was repeated several times until the gels were completely destained. A final washing step was done with 100 μl of pure acetonitrile for 10 min and finally all liquid was removed and the gel pieces were air dried for 5 min, followed by the application of 10 μl of digestion buffer (5 ng/μl trypsin (Promega) in 40 mM NH4HCO3 buffer). More NH4HCO3 buffer without trypsin was added if necessary until all gel pieces were covered completely with liquid. The plate was sealed to avoid evaporation and incubated overnight at 37°C. The proteolysis was stopped by the addition of 5 μl of 0.1% trifluoroacetic acid (TFA) in water and all samples were stored at -20°C.
For Edman-degradation, the trypsin-digested peptide fragments were separated on an HPLC. Individual peptide fragments were N-terminally sequenced and the corresponding proteins were identified using the M. xanthus genome sequence information (Goldman et al., 2006). For MALDI measurement 0.8 μl of the digest were spotted on a 384 opti-ToF steel plate and the samples were directly mixed on the plate with 0.4 μl α–cyano cinnamic acid (2.5 mg/ml in water/acetonitrile (1/1), containing 0.1% TFA). All samples were spotted twice. For co-crystallization the samples-matrix mixtures were dried at room temperature.
The peptides resulting from the “in-gel” digestions were analyzed in positive reflector mode in a 4800 MALDI ToF/ToF analyzer (Applied Biosystems) The detector covers a mass range from 800 to 4000 Da. MS spectra were generated with 1500 shoots distributed randomly all over the sample spot. Routinely the three most intense peptide peaks were automatically used for MS-MS measurements, which resulted in the peptide sequence. For peak recognition the apex algorithm was used, peaks were only included in the analysis if the isotopic distribution fits to the typical peptide distribution and the signal-to-noise ratio was greater than thirty.
Both the MS and MS-MS data for each spot was analyzed using the MASCOT search algorithm (Papin et al., 2003), trypsin peaks were ignored. The mass tolerance was set to 70 ppm for MS and 0.1 Da for MS-MS measurements and the search was performed against all entries.
For nanoelectrospray-MS measurements tryptically digested MXAN_0412 (in acetonitrile) was diluted with an equal volume of nanoelectrospray solution (49% MeOH and 2% formic acid in H2O) and introduced into the LTQ Orbitrap (Thermo Scientific) at a flow rate of approximately 200 nl/min using a TriVersa Nanomate device (Advion Biosciences). Each full scan consisted of 10 microscans and 30 to 70 spectra were averaged for each measurement. For CID fragmentation, the relative collision energy was set to 30% and accumulation times were up to 4000 ms. Data analysis was performed using the Xcalibur 2.0 software.
Mutants of M. xanthus were made by introduction of inactivation constructs made with a TOPO cloning kit (Invitrogen, Karlsruhe, Germany) as described (Bode et al., 2006b). Fragments for inactivation were amplified from genomic DNA of M. xanthus using primers 1127-1 and 1127-2 for MXAN_1127 (potential DGAT), 0412-1 and 0412-2 for MXAN_0412 (putative lipid body associated protein with DUF124), 2536-1 and 2536-2 for MXAN_2536 (fatty acyl CoA-ligase) and 5582-1 and 5582-2 for MXAN_5582 (lipid body associated protein). The resulting internal fragments of the desired genes were cloned into plasmid pCR2.1-TOPO (Invitrogen) and the resulting plasmids were purified from E. coli TOP10 (Invitrogen) and introduced into M. xanthus DK1622 by electroporation as described previously (Jakobsen et al., 2004) resulting in strains MR1127 (MXAN_1127kan), MR0412 (MXAN_0412kan), EB2536 (MXAN_2536kan) and MR5582 (MXAN_5582kan), respectively. Because the plasmid cannot replicate in M. xanthus, kanamycin-resistant electroporants result from homologous recombination. These incorporate the plasmid into the chromosome and thereby disrupt the respective genes, which was verified as described previously using a PCR protocol based on a plasmid-specific and a gene-specific primer pair (Jakobsen et al., 2004).
For the heterologous expression of MXAN_1127, a 1424 bp-fragment containing the whole MXAN_1127 coding sequence was obtained by PCR with primers 1127exp_EcoRINdeI-1 and 1127exp_SalI-2 and cloned into the EcoRI and SalI sites of pMR06Kan. The fragment was then subcloned into the Pseudomonas expression vector pCOM10 (Smits et al., 2001), giving rise to expression construct pCOMMXAN_1127. This construct was introduced into Pseudomonas putida KT2440 by triparental mating as described previously (Hill et al., 1994). For the construction of pMR06kan a 1 kb-fragment containing a kanamycin-resistance gene (npt) was obtained from pCR2.1-TOPO (Invitrogen) with primers kan_BamHI-1 and kan_MluI-2. Both fragments were digested with the appropriate restriction enzymes. A 1.5 kb-fragment containing the origin of replication and the lacZα gene with the multiple cloning site was amplified from pUC18 with primers pUCorilac_PauI-1 and pUCorilac_BglII-2 and digested with PauI and BglII. This fragment (named orilac) was ligated to the fragment of the resistance gene yielding the desired cloning vector. All oligonucleotides used for mutant construction are listed in Table S2.
qRT-PCR analyses of the M. xanthus genes MXAN_0412 and MXAN_5582 were performed as described previously (Bode et al., 2006b). Similarly, the number of transcripts for WS/DGAT and ACIAD1953 in A. baylyi ADP1 grown in LB or MSM medium (Kalscheuer and Steinbüchel, 2003) was analyzed and all oligonucleotides used for RT-PCR are listed in Table S2.
The authors would like to thank Eva Luxenburger for expert help in lipid and protein ESI-MS analysis, Fernando Pineda for help with the statistical analysis, Rolf Müller for his excellent support, the Bioimaging Facility at the University of Delaware for the usage of their Leica EM PACT high pressure freezer, and an anonymous reviewer for valuable advice that greatly helped improving the manuscript. This work was supported by a Deutsche Forschungsgemeinschaft Emmy Noether fellowship to H.B.B. and a Faculty Innovation Fund from the Bloomberg School of Public Health as well as a Public Health Service Grant GM085024 from the National Institutes of Health to E.H.