During sporulation in Bacillus subtilis, the committed-cell undergoes substantial membrane rearrangements to generate two cells of different sizes and fates: the mother cell and the forespore. Here, we demonstrate that the master transcription factor Spo0A reactivates lipid synthesis during development. Maximal Spo0A-dependent lipid synthesis occurs during the key stages of asymmetric division and forespore engulfment. Spo0A reactivates the accDA operon that encodes the carboxylase component of the acetyl-CoA carboxylase enzyme, which catalyses the first and rate-limiting step in de novo lipid biosynthesis, malonyl-CoA formation. The disruption of the Spo0A-binding box in the promoter region of accDA impairs its transcriptional reactivation and blocks lipid synthesis. The Spo0A-insensitive accDA0A cells were proficient in planktonic growth but defective in sporulation (σE activation) and biofilm development (cell cluster formation and water repellency). Exogenous fatty acid supplementation to accDA0A cells overcomes their inability to synthesize lipids during development and restores sporulation and biofilm proficiencies. The transient exclusion of the lipid synthesis regulon from the forespore and the known compartmentalization of Spo0A and ACP in the mother cell suggest that de novo lipid synthesis is confined to the mother cell. The significance of the Spo0A-controlled de novo lipid synthesis during B. subtilis development is discussed.
Forty years have passed since the formulation of the ‘fluid mosaic model’ to interpret the structure and biochemistry of a biological membrane (Singer and Nicolson, 1972). However, how cells build a functional membrane remains one of the most captivating mysteries in cellular biology (Engelman, 2005; Lee, 2005). In organisms ranging from bacteria to humans, membranes undergo changes and remodelling as the cell cycle progresses and as differentiated cell types are established (Schneiter and Kohlwein, 1997; Emoto and Umeda, 2001; Jahn et al., 2003; Burton et al., 2007). At the microscopic scale, the two main components of a biological membrane, lipids and proteins, must be correctly packed, and they must interact with each other to create a functional membrane (Matsumoto et al., 2006; Bogdanov et al., 2008; Dowhan and Bogdanov, 2009; 2011). However, the role of lipids in membrane homeostasis and development still remains poorly understood (Schneiter and Kohlwein, 1997; Jahn et al., 2003; Engelman, 2005; Lee, 2005; Mileykovskaya and Dowhan, 2005; Choard, 2008).
In this context, spore formation in the Gram-positive bacterium Bacillus subtilis provides a useful model for investigating the formation of new membranes and their remodelling during development (Stragier and Losick, 1996; Errington, 2003; Hilbert and Piggot, 2004; Higgins and Dworkin, 2012). As illustrated in Fig. 1A, complex and remarkable membrane lipid-associated morphogenetic rearrangements (e.g. the biogenesis, migration, curvature, engulfment, fusion and fission of membranes) accompany the development of a bacterial spore (Errington, 2003; Higgins and Dworkin, 2012). Concurrently with these dynamic membrane remodelling events, different critical cell fate determinants become asymmetrically distributed on the sporulation membranes (Ramamurthi and Losick, 2005; 2009). Sporulation is triggered by starvation-related signals that activate, by phosphorylation, the master regulatory protein Spo0A. During the initial developmental stages, the formation of an asymmetrically positioned septum delimited by two membranes of opposite topology creates a small and specifically localized cell, which is known as the forespore, as well as a larger cell, which is called the mother cell (Fig. 1A, T0–TII) (Losick and Stragier, 1992; Stragier and Losick, 1996). During sporulation, the polar septum is formed prior to the incorporation of a complete chromosome into the forespore. Approximately, one-third of the forespore-destined chromosome is trapped in the forespore at the time of polar septum formation (Fig. 1A, TIIi). This atypical chromosome partitioning creates a transient genetic asymmetry where the mother cell becomes diploid for two-thirds of the chromosomal genes, which are simultaneously absent in the forespore, after asymmetric division (Fig. 1A, TIIi). The dissolution of the polar septum peptidoglycan allows for the fusion of the two membranes and the SpoIIIE-dependent DNA translocation (yellow cylinder in Fig. 1A, TIIii) of the remaining two-thirds of the forespore-destined chromosome into the smaller compartment (Burton et al., 2007). While chromosome translocation occurs, the mother cell membrane initiates a phagocytosis-like process to engulf the forespore (Fig. 1A, TIII–TIV). During engulfment, the mother cell membrane (green in Fig. 1A) migrates around the forespore membrane (red in the figure), fuses to pinch off the forespore so that it becomes a free protoplast surrounded by two lipid bilayers (green and red in the figure) within the mother cell (Fig. 1A, TIV). Engulfment eventually creates three topologically distinct membranes within the sporangium: the cytoplasmic membrane and the mother cell and forespore membranes (blue, green and red colours in Fig. 1A respectively). Following these events, the forespore matures into a dormant and highly resistant spore that is released following lysis of the mother cell (Fig. 1A, TV–TVII). In parallel with these numerous membrane phenomena, cell type-specific gene expression is triggered immediately after completion of the asymmetric division (TII) by the active and septum-localized key cell fate phosphatase SpoIIE (violet circle in Fig. 1A, TII), which releases active σF into the forespore cell (not shown in Fig. 1A for simplicity). This activation initiates the forespore-specific lineage of gene expression, which continues with the activation of σE by the σF-dependent gene-product SpoIIR in the mother cell compartment (criss-cross regulation) (Losick and Stragier, 1992; Stragier and Losick, 1996; Higgins and Dworkin, 2012). How the activity of SpoIIR, which is produced in the forespore and secreted into the septal membranes, is restricted to the mother cell site of the sporulating cell is unknown (Errington, 2003; Hilbert and Piggot, 2004; Higgins and Dworkin, 2012). The recently discovered role of lipids as topological determinants and lipochaperones suggests that these non-protein elements of the cell may also be important for the proper sorting, localization and functionality of key membrane determinants (Emoto and Umeda, 2001; Vanounou et al., 2003; Barák et al., 2008; Dowhan and Bogdanov, 2009; 2011). The numerous and dynamic membrane remodelling events that occur during the morphogenesis of the spore (Fig. 1A) also suggest that the metabolism of fatty acids (FAs) and membrane lipids (Fig. 1B) may have an active structural and/or regulatory role during the development in B. subtilis.
Active lipid synthesis in the absence of growth is linked to spore development
The treatment of B. subtilis cultures with subinhibitory doses of ethanol does not affect vegetative growth but instead produces a developmental blockage that inhibits Spo0A activation (Gottig et al., 2005). Interestingly, the same dose of ethanol that inhibits Spo0A activation also inhibits de novo lipid synthesis (Gottig et al., 2005). Therefore, we explored the possibility of a previously undetected connection between the activity of the master regulator Spo0A (Fig. 1A) and de novo lipid formation (Fig. 1B) during the morphogenesis of the spore.
To test this hypothesis, we measured the levels of de novo lipid synthesis through specific incorporation of [1-14C]-acetate into lipids in the presence of sublethal doses of alcohol (ethanol) that were added at specific developmental times before and after Spo0A activation (the onset of sporulation). Supplementation of the sporulation-supporting medium [Difco sporulation medium (DSM) broth] with alcohol 30 min before T0 (T−0.5) reduced the de novo synthesis of the three main lipid species present in B. subtilis (phospholipids, PLs; neutral glycolipids, GLs; and FAs) more than 90% before the entry into stationary phase (Fig. 2A and B, lanes 1 and 2). In contrast, when alcohol was added to the DSM broth 30 min after T0, the observed levels of de novo lipid synthesis were comparable to those observed in a parallel culture that was not treated with alcohol (Fig. 2A and B, lanes 3 and 4). This substantial difference in de novo lipid synthesis levels was dependent on whether the addition of alcohol occurred before (full sensitivity) or after T0 (full insensitivity). Importantly, the lipid synthesis effect correlated robustly with the onset of the sporulation phase as evidenced by the expression pattern of the Spo0A-controlled early developmental gene spoIIE (Fig. 2C), which suggests a requirement for Spo0A activity for active lipid synthesis after growth has ended. Supporting this conclusion, when B. subtilis was grown in Luria–Bertani (LB) broth, a medium that permits active vegetative growth but does not support active sporulation, lipid synthesis was active during the vegetative phase and significantly decreased at the onset of the stationary phase (Fig. 2D and E, T0–T3). As expected from the low levels of active Spo0A in LB broth (Hilbert and Piggot, 2004), spoIIE was not expressed during the process of bacterial growth (Fig. 2F).
Spo0A reactivates de novo lipid synthesis during spore development
The key role of Spo0A in the formation of new lipids after vegetative growth was confirmed by measuring the de novo lipid levels of Spo0A-proficient and Spo0A-deficient B. subtilis cultures grown under sporulation conditions (Fig. 3A and B). For both cultures (spo0A+ and spo0A−), the rate of de novo lipid synthesis per cell (measured as the incorporation of [1-14C]-acetate into the lipids) reached a maximum at the mid-logarithmic phase (T−2) and then decreased until it reached a minimum at the end of vegetative growth (T0) (Fig. 3C). However, after the onset of the stationary phase and during active sporulation, only the Spo0A-proficient culture was capable of re-establishing active lipid synthesis and was thus able to produce new lipid levels comparable to those observed during active vegetative growth (Fig. 3A and B, T0–T3). Culture aliquots of Spo0A-proficient and Spo0A-deficient cells were taken at different times during growth in DSM broth and labelled with the fluorescent dye FM4-64 (specific for membrane staining) to follow the percentage of cells that went through the processes of asymmetric division and engulfment during the sporulation phases. Interestingly, most Spo0A-dependent lipid synthesis occurred at developmental stages (2–4 h after the onset of the stationary phase of the Spo0A-proficient culture) that are expected to require high levels of new membrane formation for the biosynthesis of the asymmetric septum and forespore engulfment by the mother cell membrane (Fig. 3C and data not shown). In contrast, the B. subtilis culture deficient in Spo0A activity continued to demonstrate decreased levels of new lipid synthesis during the stationary phase in DSM broth (Fig. 3B and C) and its cells did not go through the process of asymmetric division as evidenced by fluorescence microscopy (data not shown). To confirm whether the regulatory impact on de novo-synthesized lipids could be attributed to the absence of Spo0A activity, we performed a complementation experiment with a spo0A mutant strain expressing an ectopically integrated copy of the sad 67-spo0A gene under IPTG control (strain RG1491, Table S1) (Arabolaza et al., 2003). Under IPTG supplementation, this strain produces a constitutively active form of Spo0A (Sad67), and it was able to reactivate the de novo lipid synthesis under sporulation conditions only when IPTG was added to the DSM broth (data not shown).
Spo0A reactivates the global synthesis of membrane lipids harbouring linear and branched fatty-acyl chains during spore development
[1-14C]-acetate is a specific precursor for the synthesis of linear FAs (n-C14 to n-C18). [1-14C]-acetate treatment results in membrane lipids that harbour straight fatty-acyl chains (Fig. 1B). However, in membranes of the Bacillus genus, branched (iso- and anteiso-) fatty-acyl moieties are also major membrane lipid components (Fig. 1B). Therefore, to confirm that the role of Spo0A in the global reactivation of lipid synthesis was not restricted to linear FA synthesis (i.e. affecting the substrate preference of the FabHA and FabHB condensing enzymes for straight-FAs instead branched-FAs precursors) (Kingston et al., 2011), we measured the incorporation of [1-14C]-isoleucine, a specific precursor for the synthesis of branched FAs, into newly formed lipids. As illustrated in Fig. 4A, in the wild-type culture, strong reactivation of de novo lipid synthesis at the onset of sporulation was observed using [1-14C]-isoleucine. In contrast, the Spo0A-deficient culture was unable to reactivate de novo branched-lipid synthesis during the stationary phase (Fig. 4B). The capacity of these two FA precursors ([1-14C]-acetate and [1-14C]-isoleucine) to account for the bulk of total fatty-acyl moieties present in B. subtilis membranes demonstrates that Spo0A induces global reactivation of the synthesis of the membrane lipids (GLs and PLs) that harbour linear and branched fatty-acyl chains during sporulation.
To further examine the regulatory role of Spo0A in lipid synthesis during sporulation, we took advantage of the fact that, unlike mammals, bacteria possess a FA synthase (FASII) complex that is composed of monofunctional proteins (Fujita et al., 2007; Zhang and Rock, 2008). Each FASII protein catalyses a step in the biosynthetic pathway of lipid formation that begins with the key metabolite malonyl-CoA (Fig. 1B). Therefore, using [2-14C]-malonyl-CoA as a specific substrate, we compared the in vitro lipid synthesis activities of protein extracts prepared from isogenic Spo0A-proficient and Spo0A-deficient B. subtilis cultures grown in sporulation-supporting broth. The in vitro FA synthesis activities of the Spo0A-proficient (Fig. 4C) and the Spo0A-deficient protein extracts (Fig. 4D) were similar during the vegetative phase (T−2–T0). However, during spore development, only protein extracts prepared from stationary-phase cultures of the Spo0A-proficient cells demonstrated high lipid synthesis activity when [2-14C]-malonyl-CoA was used as a specific precursor for in vitro FA synthesis (compare the levels of [14C]-FAs synthesized by both types of protein extracts during T1–T4 in Fig. 4C and D).
In summary, these results (Figs 2-4) demonstrate, for the first time, that new lipid formation (FA and membrane lipid synthesis) during the stationary phase remains active only under sporulation conditions. In addition, the data indicate a novel role for Spo0A in controlling the activity of the central metabolic pathway of lipid synthesis, and thus new membrane formation and remodelling (see Fig. 1A), during the morphogenesis of the spore cell.
Spo0A reactivates soluble fatty acid synthesis, the first stage of de novo lipid formation
A key checkpoint in coupling FA synthesis to the synthesis of membrane lipids (GLs and PLs) is the formation of long-chain acyl-ACP intermediates that connect stages I and II of lipid synthesis (Fig. 1B and Fujita et al., 2007; Zhang and Rock, 2008). To investigate the stage (I or II) during which Spo0A-dependent reactivation of lipid synthesis occurs, we uncoupled FA synthesis from membrane lipid synthesis by specifically inhibiting the latter process (Mindich, 1970; Grau and de Mendoza, 1993). Although glycerol starvation of B. subtilis glycerol-auxotrophs (glyc) results in the profound inhibition of membrane PL and GL synthesis, the synthesis of soluble FAs (long-chain [14C]-FAs) continues at a normal rate, and de novo long-chain [14C]-FAs accumulate as free species (Mindich, 1970; Grau and De Mendoza, 1993). We took advantage of the fact that sporulation can also be induced by resuspension of a growing culture in a poor medium (Sterlini-Mandelstam sporulation medium). Cultures of glyc strains proficient or deficient in Spo0A activity (strains B42 or RGB42 respectively, Table S1) were grown in glycerol-supplemented growth medium (Schujman et al., 1998, a medium that allows for the physiological manipulation of the glycerol auxotroph without affecting its proficiency in sporulation) at 37°C until the mid-logarithmic phase was reached. We then resuspended the cultures in Sterlini-Mandelstam poor medium in the absence or presence of glycerol to induce sporulation. Both cultures were then exposed to [1-14C]-acetate and cultured for 3 h at 37°C prior to lipid synthesis analysis. Upon glycerol starvation, the Spo0A-proficient glyc mutant strain, B42, halted new membrane lipid formation and accumulated more than 90% of the radioactivity in the soluble fraction of the de novo long-chain [14C]-FAs (compare lanes 1 and 2 in Fig. 5A). In contrast, the glyc spo0A double-mutant strain, RGB42, was unable to synthesize soluble FAs when resuspended in sporulation medium without glycerol supplementation (Fig. 5A, lane 3). Because soluble FAs did not accumulate in strain RGB42 after glycerol removal (Fig. 5A, lane 3) and because the only difference between strains B42 and RGB42 was proficiency in Spo0A activity, we conclude that Spo0A specifically reactivates soluble FA synthesis, the first stage in the formation of new lipids (Figs 1B and 5A).
Spo0A regulates the first and rate-limiting step of de novo fatty acid synthesis: malonyl-CoA formation
The lack of medium-chain [14C]-FAs in the glycerol-starved spo0A-mutant RGB42 cells (lane 3 in Fig. 5A) compared with the long-chain [14C]-FAs synthesized in the glycerol-starved BR42 cells (lane 1 in Fig. 5A) strongly suggests that Spo0A participates in the reactivation of soluble FA synthesis during a critical step upstream of the enzymatic activity of the FabHA and FabHB condensing enzymes, which are required for the initial condensation and two-carbon elongation of malonyl-ACP with acetyl-CoA or branched-CoA esters (formation of medium-chain FAs, see Fig. 1B). To test this hypothesis, we treated cell-free protein extracts prepared from wild-type (Spo0A-proficient) B. subtilis cultures with cerulenin, a specific inhibitor of FabF, the condensing enzyme that controls the formation of long-chain FAs (Fujita et al., 2007; Zhang and Rock, 2008). As expected, this treatment blocked the formation of radioactive long-chain FAs but permitted the accumulation of medium-chain [14C]-FAs (Fig. 5B). These medium-chain FAs, absent in glycerol-starved glyc-spo0A cells (see Fig. 5A lane 3), are formed via the activities of FabHA and FabHB, which are cerulenin-insensitive condensing enzymes. This result confirms that Spo0A regulates a crucial step in FA synthesis that occurs prior to the enzymatic activities of FabHA and FabHB (compare the absence of de novo-formed FAs in Fig. 5A lane 3 with the presence of de novo-synthesized FAs in Fig. 5B lane 2).
What enzymatic step of de novo FA synthesis (see Fig. 1B), preceding the activities of the three condensing enzymes (FabHA, FabHB and FabF), is under Spo0A control? Interestingly, through in silico bioinformatic analysis, in vitro chromatin immunoprecipitation and gene microarray analysis, two independent research groups have proposed that the genes comprising the bicistronic operon accDA, which encodes the β and α subunits (carboxyltransferase component) of the acetyl-CoA carboxylase (ACC) enzyme, belong to a new group of genes that are directly and positively controlled by Spo0A (Liu et al., 2003; Molle et al., 2003). ACC catalyses the first and rate-limiting step of de novo FA synthesis, which entails the ATP-dependent formation of malonyl-CoA from acetyl-CoA and bicarbonate (see Fig. 1B and Fujita et al., 2007; Zhang and Rock, 2008 for references). In addition to its central role in new lipid synthesis, malonyl-CoA negatively controls the activity of the master repressor of de novo lipid synthesis in B. subtilis, the transcription factor FapR (Fujita et al., 2007; Zhang and Rock, 2008). Taking into account the reports of Liu et al. (2003) and Molle et al. (2003) and based upon our in vivo and in vitro results, we hypothesized that the de novo formation of malonyl-CoA constitutes the key step of de novo lipid synthesis under developmental Spo0A control.
Therefore, we analysed the role of Spo0A in the in vivo expression of the genes that encode the ACC enzyme carboxyltransferase component, which is responsible for malonyl-CoA formation. As illustrated in Fig. 6A, the expression of the accDA operon was similar between the Spo0A-proficient and Spo0A-deficient cultures (strains RG1606 and RG1607 respectively, Table S1) during vegetative growth in DSM broth. In both cultures, the expression of the accDA operon, measured as β-galactosidase activity driven by the accDA promoter, was high at low cellular densities and decreased as the cultures reached the end of the vegetative phase. This behaviour suggests that accDA expression could be growth-rate regulated during the vegetative growth as previously suggested for the expression of accBC that codes for the biotin carboxyl-carrier component of ACC (Marini et al., 2001). Interestingly, at the end of the vegetative phase, a significant difference in accDA operon expression was observed between the two isogenic cultures. While accDA expression in the Spo0A-deficient culture continued to decrease until reaching an undetectable level of β-galactosidase activity, strong reactivation (starting at T0) of β-galactosidase expression was detected in the Spo0A-proficient cells. The reactivation of accDA promoter-driven expression reached a maximum at T3. This pattern correlated well with the patterns of both in vivo (Figs 3C and 4A) and in vitro (Fig. 4C) reactivation of de novo lipid biosynthesis in the wild-type cultures of B. subtilis grown in sporulation-supporting medium (DSM). As expected, the reactivation of carboxyltransferase expression at the end of vegetative growth was not observed when the Spo0A-proficient and Spo0A-deficient cultures were grown under non-sporulating conditions (growth in LB broth, Fig. 6A and data not shown). These in vivo results are in complete agreement with the in vitro results of Molle et al. (2003), who through the use of gel shift binding assays and chromatin immunoprecipitation experiments, demonstrated the specific binding of Spo0A to accDA. In toto, the present in vivo (Fig. 6A) and former in vitro results (Liu et al., 2003; Molle et al., 2003) suggest the physiological dependency of accDA expression on Spo0A activity.
How does Spo0A regulate accDA expression? As shown in Fig. 6B, the promoter region of the accDA operon contains conserved −35 and −10 sequences, separated by an ideal spacer of 17 bp, with homology to the housekeeping σA-dependent promoters. Upstream of the −35 region is a perfect Spo0A-binding box (5′-TTGTCGAA-3′) that could be necessary for maintaining active expression of accDA during the stationary phase of growth in Spo0A-proficient cultures under sporulation conditions (Fig. 6A, strain RG1606). Therefore, the expression of accDA during the vegetative phase of growth (in LB or DSM) would be driven by the housekeeping σA-containing RNA polymerase irrespective and independent of Spo0A activity. Otherwise, to maintain active expression of accDA during the stationary phase of growth (i.e. under sporulation conditions) the presence of an active Spo0A, capable of binding to the Spo0A-box present in the promoter region of the operon (as previously shown by Liu et al., 2003), would be essential to reactivate accDA expression. To demonstrate the essentiality of Spo0A for the reactivation of accDA expression during sporulation, we constructed a Spo0A-proficient mutant strain that harboured a mutation in the Spo0A-binding box present in the promoter region of the accDA operon (strain RG1130, see Experimental procedures and Table S1). In addition to the accDA0A mutation, the strain RG1130 harboured an erythromycin resistance cassette (eryr), used for the selection of the construction, upstream of the regulatory region of accDA (see Fig. 6B). During growth in sporulation-supporting medium (DSM broth, 200 r.p.m. at 37°C), there were no significant differences in the rate of growth or the cellular yield reached at the beginning of the stationary phase between eryrΩaccDA0A and wild-type cells (data not shown), indicating that the eryrΩaccDA0A construct did not affect the vegetative growth of RG1130 cells. To confirm that Spo0A governs accDA expression during sporulation, we measured accDA expression (Fig. 6C) and de novo FA synthesis (Fig. 6D) in cultures that harboured the mutation in the Spo0A-binding box of accDA (eryrΩaccDA0A mutant cells) during growth in LB and DSM broths. As illustrated in Fig. 6C, during the vegetative phase in DSM or LB broth, cultures of the Spo0A-proficient cells harbouring the reporter fusion driven by the mutated promoter (PeryrΩaccDA0A-lacZ::amyE, strain RG1158, see Table S1) exhibited a pattern of β-galactosidase expression similar to that of the cells harbouring the wild-type reporter fusion (PaccDAwt-lacZ::amyE, strain RG1606, see Fig. 6A). However, during the stationary phase (after T0), the expression of the mutated promoter PeryrΩaccDA0A was not reactivated in the Spo0A-proficient culture and remained downregulated throughout the stationary phase (Fig. 6C). The absence of transcriptional reactivation of β-galactosidase expression driven by cells that harboured the PeryrΩaccDA0A promoter during the stationary phase of growth in DSM broth was not due to the presence of the erythromycin resistance cassette (eryr), as β-galactosidase expression was normally reactivated by cells in which the accDA promoter harboured (in the same position as in strain RG1158) an erythromycin resistance cassette but a fully functional Spo0A-binding box (PeryrΩaccDAwt, strain RG1160, Table S1) (see Fig. 6C). In accord with the arrest of sporulation-linked accDA expression when the Spo0A-binding box present in the promoter region of accDA was disrupted (strain RG1158, Fig. 6C), de novo lipid synthesis in the Spo0A-insensitive eryrΩaccDA0A cells (strain RG1130) was not restored under sporulation conditions (Fig. 6D). As expected, the reactivation of de novo lipid synthesis during sporulation in DSM broth was not affected in the eryrΩaccDAwt strain RG1160 (Fig. 6D). Overall, these results confirm that Spo0A regulates the reactivation of malonyl-CoA synthesis and becomes the master regulator of de novo FA synthesis and membrane lipid homeostasis during sporulation.
Spo0A-dependent de novo fatty acid synthesis is required for spore development
The strict Spo0A dependency of the formation of new malonyl-CoA (see Fig. 6A and C) and the inability of the Spo0A-proficient accDA0A cells to synthesize new lipids under sporulation conditions (see Fig. 6D) strongly suggest that de novo FA synthesis could have a structural and/or regulatory role during development in B. subtilis. As shown in Fig. 7A, the Spo0A-proficient accDA0A cells (strain RG1130) failed to make spores in DSM broth unless they were supplemented and grown in the presence of exogenous FAs (palmitic and oleic acids, nC16:0 and nC18:1 respectively). The exogenous FAs completely restored the sporulation proficiency of the Spo0A-insensiive accDA0A cells to that of the parental wild-type strain (Fig. 7A). This result confirms that the sporulation deficiency of the Spo0A-proficient accDA0A cells was due to a deficiency in the reactivation of de novo lipid synthesis at the onset of the sporulation phase because the supplementation with exogenous FAs permitted the accDA0A cells to bypass their inability to synthesize new FAs under Spo0A control. Our data correlate the oligosporic phenotype of the Spo0A-insensitive accDA0A cells (strain RG1130, Fig. 7A) with its inability to restore accDA expression (Fig. 6C), and hence form new lipids and membranes (Fig. 6D), under Spo0A control.
To test whether the inability to make new lipids in the accDA0A cells blocked spo0A expression and/or Spo0A activity (and hence downregulated spore formation), we measured the expression pattern of key sporulation genes that are under temporal and spatial regulation during the morphogenesis of the spore. Neither spo0A expression (Fig. 7B) nor Spo0A activity (Fig. 7C) was affected by the inactivation of the Spo0A-binding box in accDA. These results confirm that the downregulation of the sporulation proficiency of accDA0A cells was not due to the inability of these cells to trigger the onset of the sporulation programme. In addition, the expression of the σF-controlled regulatory gene spoIIR was not affected in accDA0A cell cultures (Fig. 7D), but the expression of the σE-controlled spoIID gene was severely downregulated (Fig. 7E). These results are in agreement with a previous report that demonstrated that the inhibition of de novo FA synthesis in wild-type cells with cerulenin, an antibiotic that does not affect malonyl-CoA formation but inhibits the activity of FabF (Schujman et al., 1998 and Fig. 1B), blocked sporulation at the developmental stage of pro-σE processing. Together, these results suggest that de novo FA synthesis has a regulatory role in the specific activation of σE in the mother cell compartment (see Discussion).
The asymmetric nature of de novo lipid synthesis during spore development
Beyond its primary role of supplying structural PLs and GLs for new membrane formation, is it possible that the robust de novo lipid synthesis machinery that we discovered here plays an unknown role during development? Interestingly, most, if not all, of the enzymes involved in PL and GL formation were located in the membrane of the asymmetric septum during sporulation (Nishibori et al., 2005; Matsumoto et al., 2006). In addition, Spo0A functions as a cell type-specific transcription factor in the mother cell compartment for several hours after asymmetric division and does not function in the forespore chamber (Arabolaza et al., 2003; Fujita and Losick, 2003). When spo0A is ectopically integrated into the chromosome to restrict Spo0A expression and activity to the smaller compartment, the emerged sporulation phenotype is defective, indicating that the mother cell-compartmentalized expression of Spo0A is essential for the success of the sporulation programme (Fujita and Losick, 2003). More recently, a key protein of the metabolism of FAs, acyl carrier protein (ACP), which is in charge of the transport and solubilization in the cytoplasm of the fatty-acyl groups during FA and PL synthesis (see Fig. 1B), was shown to be asymmetrically compartmentalized in the mother cell compartment during sporulation (Martínez et al., 2010). Taken together, our results and previous studies (Arabolaza et al., 2003; Fujita and Losick, 2003; Martínez et al., 2010) indicate that Spo0A-activated de novo lipid synthesis occurs only in the mother cell compartment of sporulating cells to provide soluble fatty acyl-ACP substrates to the septum-localized FL synthases (Nishibori et al., 2005; Matsumoto et al., 2006) during sporulation. In addition to the mother cell compartmentalization of Spo0A and ACP (Fujita and Losick, 2003; Martínez et al., 2010), and supporting our hypothesis, we found that all the genes encoding the enzymes involved in the synthesis of FAs and most, if not all, the genes encoding enzymes involved in the synthesis of the lipid species present in B. subtilis membranes (PG, PE, CL, lysyl-PG, glucosyl-MAG and glucosyl-DAG) were located outside of the chromosomal region that is initially trapped in the forespore compartment immediately after polar septum formation (lipid gene asymmetry, Fig. 8A). It has been proposed that transient genetic asymmetry would allow the transient asymmetry of gene expression in favour of one of the two cell compartments of the sporangium (Arigori et al., 1999; Frandsen et al., 1999; King et al., 1999; Losick and Dworkin, 1999). The proposed genetic asymmetry might be responsible, at least in part, for the establishment of σF-dependent cell type-specific gene expression during sporulation (Dworkin and Losick, 2001; Dworkin, 2003; Hilbert and Piggot, 2004; Higgins and Dworkin, 2012). Among the complete set of genes (more than 20) encoding the enzymes involved in de novo lipid synthesis, which are transiently trapped in the mother cell compartment, are the ACP-coding gene, acpP, which is essential for the synthesis of soluble fatty-acyl moieties (stage I of the pathway of lipid synthesis, Fig. 1B), and three key genes (gpsA, ugtP and pgsA) involved in membrane lipid synthesis (stage II of the pathway of lipid synthesis, Fig. 1B) that are located far from the oriC and are thus among the last genes to be translocated by SpoIIIE (Burton et al., 2007) to the forespore compartment after asymmetric cell division (see Fig. 8A). Thus, our data suggest that the temporary exclusion or asymmetry of the complete set of genes involved in FA and membrane lipid synthesis (lipid gene asymmetry) from the forespore (Fig. 8A), the compartmentalization of ACP into the mother cell (Martínez et al., 2010), and the mother cell-compartmentalized activity of Spo0A (Arabolaza et al., 2003; Fujita and Losick, 2003), which is responsible for the reactivation of de novo FA synthesis during sporulation (this work), produce an imbalance in which the newly synthesized lipids are preferentially formed in the mother cell (asymmetric lipid synthesis) (Fig. 8B and Discussion).
Spo0A-dependent de novo fatty acid synthesis is required for biofilm development
In addition to sporulation, Spo0A regulates other adaptive and developmental pathways that are thought to be important for the expression of the different social behaviours of B. subtilis (Kearns and Losick, 2005; Aguilar et al., 2007; Abee et al., 2011; Shank and Kolter, 2011). One of these pathways is the capacity of B. subtilis to form floating biofilms, called pellicles, in standing cultures (Branda et al., 2001; Hamon and Lazazzera, 2001; Lombardía et al., 2006) and architecturally complex colonies on solid media (Branda et al., 2001; Lombardía et al., 2006). Although non-domesticated strains of B. subtilis, such as the Marburg-related strain NCIB3610 (Branda et al., 2001; 2006) and the natto strain (Lombardía et al., 2006; Vidyalaxme et al., 2012), make robust, highly structured and long-lasting biofilms, the domesticated strain JH642 is also capable of making floating biofilms (Hamon and Lazazzera, 2001; Hamon et al., 2004; Lazazzera, 2005) (Fig. S1). Therefore, we wanted to know if the de novo lipid synthesis that is under Spo0A control and is essential for spore development is also necessary for biofilm development. To explore this possibility, we used a modified version of LB broth (LBY medium, see Experimental procedures) that enhanced biofilm formation in several B. subtilis strains that we tested, including the domesticated strains 168 and JH642 (Fig. S1 and A. Rovetto and R. Grau, unpubl. results). As shown in Fig. 9A, the accDA0A cells (strain RG1130) were severely impaired in their capacity to make a biofilm (floating pellicle) in comparison with the biofilm proficiency of wild-type cells of the JH642 strain inoculated in the same LBY medium after 36 h of incubation at 30°C (Fig. 9A). The quantification of the rate of biofilm formation (see Experimental procedures) confirmed the inability of the accDA0A cells to construct a biofilm (Fig. 9B). To confirm that the deficiency in biofilm development of the Spo0A-insensitive accDA0A cells was due to a deficiency in de novo FA synthesis, we supplemented the LBY broth with exogenous FAs (nC16:0 and nC18:1) and followed the kinetics of biofilm formation. The supplementation with exogenous FAs completely restored the proficiency of biofilm formation in the accDA0A cells (Fig. 9A and B).
Why would de novo lipid synthesis be required for biofilm development? The proficiency of the accDA0A cells to grow planktonically but not to develop spores raises the possibility that de novo FAs synthesized under conditions that allow the formation of the biofilm would have an unnoticed function in biofilm development. We hypothesized that the requirement of active and Spo0A-dependent de novo lipid synthesis during the development of the biofilm would not be restricted to the structural role of lipids as components of the cell membranes (Flemming and Wingender, 2010). Therefore, how do B. subtilis cells develop a biofilm? The extra cellular matrix (ECM) of bacterial biofilms contains several extracellular polymeric substances mainly formed by polysaccharides, proteins, nucleic acids and lipids (Flemming and Wingender, 2010; Abee et al., 2011). In addition to the highly hydrated extracellular polymeric substances (polysaccharides and proteins), other secreted molecules (lipids) have hydrophobic properties that would enhance the repellency properties of the biofilm (Fig. S2 and Conrad et al., 2003). In several Gram-positive bacteria, including B. subtilis, membrane glycolipids have been implicated in biofilm formation, but an explanation of the mechanism by which these lipids affect biofilm formation is lacking (Lazarevic et al., 2005; Theilacker et al., 2009). In addition to the presence of lipids, the hydrophobic characteristics of the biofilm have been attributed to the covalent modification of extracellular polymeric molecules with FA-related molecules (Neu and Poralla, 1988; Neu et al., 1992) and the secretion of biosurfactants to the ECM of the biofilm (Davey et al., 2003; López and Kolter, 2010; Epstein et al., 2011). The ECM of the biofilm made by B. subtilis contains an exopolysaccharide (EPS) of an incompletely known structure (EPS) and the proteins BslA (formerly named YuaB; Kobayashi and Iwano, 2012; Kovács et al., 2012) and TasA (an amyloid protein with antimicrobial properties, Branda et al., 2001; 2006). TasA and the EPS are secreted to the ECM in the early stages of biofilm development to allow the cells in the biofilm (pellicles) to be held together and to form clusters resembling bundled chains of cells (Branda et al., 2001; 2006; Romero et al., 2010). Biofilm-deficient deletion-mutants strains unable to synthesize TasA and/or EPS do not form cell clusters (Branda et al., 2001; 2006; Kobayashi and Iwano, 2012), while bslA (yuaB)-mutant cells do not form pellicles after 24 h of incubation but form submerged cell clusters that resemble the cell clusters found in wild-type floating pellicles (Kobayashi and Iwano, 2012). The amphiphilic nature of BslA, its participation in biofilm formation after the cell clusters are allowed to be formed by EPS and TasA and its topographical distribution encasing the biofilm suggest that this protein (BslA) is the major contributor to the surface repellency and hydrophobicity of B. subtilis biofilms (Kobayashi and Iwano, 2012; Kovács et al., 2012).
To determine at which stage of biofilm development the active Spo0A-dependent de novo lipid synthesis machinery is required, we compared the phenotype of accDA0A mutant cells grown under biofilm-supporting conditions (growth in LBY broth) with the biofilm phenotype of different B. subtilis mutant strains deficient in the synthesis of the ECM components TasA (tasA), EPS (epsG) and BslA (bslA). As shown in Fig. 10, the biofilm phenotype of accDA0A cells was similar to the biofilm phenotype of tasA and epsG cells (Fig. 10A and B) in that they all failed to form bundled chains (clusters) of cells (Fig. 10C). In contrast, although, the BslA-deficient mutant cells also failed to make floating pellicles, they were capable to form cluster of cells similar to the ones formed by wild-type cells (Fig. 10C). One important property of B. subtilis biofilms, apart from the capacity to form cell clusters, is their water repellency (Epstein et al., 2011; Kobayashi and Iwano, 2012 and Fig. S2). It is proposed that the robustness of the B. subtilis ECM is reached by BslA acting synergistically with TasA and EPS in a way to increase the surface repellency and impermeability of the biofilm (Epstein et al., 2011; Kobayashi and Iwano, 2012; Kovács et al., 2012). Therefore, we investigated whether de novo-formed FAs are also required to achieve the hydrophobicity (water repellency) of the B. subtilis biofilm (Fig. S2). As shown in Fig. 10D, in contrast to wild-type JH642 cells, accDA0A cells (along with the eps, bslA and tasA cells) completely failed to repel the entrance of water into the biofilm. This result indicates a novel role of de novo FA synthesis, under Spo0A control, in the capacity of B. subtilis cells to abandon the vegetative lifestyle and develop into a biofilm.
Spo0A, lipid homeostasis and development
Biological membranes are more complex than was appreciated when the ‘fluid mosaic model’ was proposed forty years ago (Singer and Nicolson, 1972; Engelman, 2005; Lee, 2005). Biological membranes form the interface between the cell and its environment and are key players in cellular homeostasis and signal transduction (Jahn et al., 2003; Choard, 2008; Fairn and Grinstein, 2008). At the simplest level, to make a functional membrane, two major components (the membrane proteins and their lipid partners) must be correctly synthesized and packed (Dowhan and Bogdanov, 2009; 2011). But how is the biogenesis of a membrane controlled and linked to the correct execution of a genetic programme? During development in B. subtilis, Spo0A is the key transcription factor that triggers the developmental pathways of spore and biofilm formation. Here, we demonstrated for the first time that Spo0A also controls the biosynthesis of the second major component of the membrane: the lipids (Fig. 1). We demonstrate the existence of active and robust de novo FA and membrane lipid synthesis during sporulation (Figs 2-4). We also demonstrate that the reactivation of the de novo lipid synthesis during development is due to the Spo0A-dependent reactivation of the synthesis of malonyl-CoA, the central metabolite of FA synthesis and lipid homeostasis in the cell (Figs 5 and 6). Among the many genes whose products participate in the de novo lipid synthesis machinery (Fig. 1B), only one genetic unit, the accDA operon, appears to be controlled by Spo0A. Why does Spo0A only regulate the activity of the operon involved in malonyl-CoA synthesis? And, how does Spo0A control the global synthesis of FAs and membrane lipids by controlling only accDA expression? In addition to its essential role as a precursor of FA synthesis, malonyl-CoA is a direct and specific negative modulator of FapR, which is a conserved transcriptional repressor of several genes involved in FA and membrane lipid synthesis (the fap regulon) in Gram-positive bacteria (see Fujita et al., 2007; Zhang and Rock, 2008 for reviews). The binding of malonyl-CoA to FapR prevents the binding of FapR to (and/or promotes the release of FapR from) its target DNA sequences, an event that derepress de novo lipid synthesis (Fujita et al., 2007). Therefore, Spo0A functions as the key regulator of de novo lipid synthesis during development by providing appropriate levels of malonyl-CoA for new lipid formation and linking the levels of this central metabolite to the downregulation of the activity of the lipid synthesis repressor FapR (Fujita et al., 2007). Reinforcing the view of the central role of Spo0A in lipid synthesis homeostasis during development is the fact that the only genes involved in FA synthesis that are not controlled by FapR are those encoding the ACC enzyme responsible for malonyl-CoA formation (Fujita et al., 2007; Zhang and Rock, 2008), and these acc genes are the only genes involved in lipid synthesis that are controlled by Spo0A (Liu et al., 2003; Fujita and Losick, 2003) (Fig. S3).
Lipid asymmetry and development
Although the roles of lipids during development remain poorly understood, FAs have been proposed to play key regulatory roles during the establishment of protein asymmetry in a variety of model organisms, such as the nematode Caenorhabditis elegans (Rappleye et al., 2003), the plant Arabidopsis thaliana (Baud et al., 2003) and the yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe (Schneiter and Kohlwein, 1997; Iwamoto et al., 2004). It was found that depletion of FAs or loss of acetyl-CoA carboxylase activity in C. elegans through the pod-2 mutation causes a loss of asymmetry and mislocalization of the PAR-3/PAR-2 cell fate determinants during the first developmental division, supporting a regulatory rather than structural role of de novo-synthesized FAs (Rappleye et al., 2003; Golstein and Macara, 2007). Here, we demonstrated that the Spo0A-controlled de novo FA synthesis was required for sporulation proficiency (Fig. 7) and biofilm development (Figs 9 and 10). In addition, our results in combination with previous reports (Arabolaza et al., 2003; Fujita and Losick, 2003; Nishibori et al., 2005; Matsumoto et al., 2006; Martínez et al., 2010) strongly suggest the asymmetric nature of the de novo lipid synthesis machinery (confined to the mother cell compartment) during sporulation (Fig. 8). Why should de novo lipid synthesis be asymmetric during the development of the spore?
We demonstrated that maximal Spo0A-dependent de novo lipid synthesis occurred during the key developmental stages of polar septum formation and engulfment of the forespore by the mother cell membrane (see Figs 3 and 4). The formation of an asymmetrically positioned septum and the phagocytosis-like process of forespore engulfment by the mother cell membrane constitute hallmark events during which active membrane remodelling (membrane curvature, fusion, migration and fission) accompanies the correct positioning and activation of key cell fate determinants to specific membrane sites (Hartwell and Weinert, 1989; Cutting et al., 1990; Ireton and Grossman, 1992; Stragier et al., 1998; Abanes-De Mello et al., 2002; Ramamurthi and Losick, 2005; 2009; Broder and Pogliano, 2006; Campo et al., 2008; Meyer et al., 2003).
Fujita and Losick (2003) proposed that one reason for the high and prolonged activity of Spo0A in the mother cell compartment was to ensure a high expression level of the operon that governs the synthesis of the mother cell transcription factor σE (also named SpoIIGB). σE is synthesized as an inactive precursor, pro-σE, which is processed to its active form, σE, by the developmental membrane protease SpoIIGA, after the formation of the asymmetric septum (Imamura et al., 2008). Both proteins, which are encoded by the spoIIG operon, are synthesized under the direct, positive transcriptional control of Spo0A (Hilbert and Piggot, 2004). SpoIIGA initially resides at the cytoplasmic membrane, but soon after asymmetric division, it is thought to become concentrated (by a diffusion-and-capture mechanism) on both membranes of the sporulation septum (Fawcett et al., 1998; Stragier et al., 1998; Fujita and Losick, 2002; Rudner et al., 2002; Chary et al., 2010). Under physiological conditions, pro-σE, present in both chambers, is processed to active σE only in the mother cell compartment of the sporulating cell. The action of SpoIIR is essential for this cell type-specific activation (Karow et al., 1995; Londoño-Vallejo and Stragier, 1995). This regulatory protein is produced in the forespore, under the control of σF, and is secreted into the intermembrane space of the sporulation septum. Once in the space between the two septal membranes, SpoIIR exclusively activates SpoIIGA localized on the mother cell membrane but not the SpoIIGA located in the forespore membrane (Karow et al., 1995; Londoño-Vallejo and Stragier, 1995; Fawcett et al., 1998). How σE becomes activated only in the mother cell compartment is an unsolved question. SpoIIR is produced in low amounts, and it has been proposed that the SpoGA molecules inserted into both septal membranes compete for the limited amount of SpoIIR that is secreted into the intermembrane space (Chary et al., 2010). Ordinarily, the amount of SpoIIGA molecules localized at the mother cell membrane far exceeds the amount of SpoIIGA localized at the forespore membrane because of the continuous and robust mother cell-restricted spoIIG expression driven by Spo0A (Fujita and Losick, 2003), and the mother cell site would win the competition to activate σE (Chary et al., 2010). However, under physiological conditions, how does 100% of the SpoIIR secreted into the intermembrane space interact exclusively with the mother cell membrane SpoIIGA and not with SpoIIGA localized at the forespore membrane? In a previous work, we demonstrated that the early addition (before T0) of the antibiotic cerulenin to B. subtilis cells grown in sporulation broth (DSM) does not inhibit the onset of sporulation but blocks the processing of pro-σE to active σE (Schujman et al., 1998). In the present work, we also showed that the absence of de novo FA synthesis in accDA0A cells grown under sporulation conditions impaired the expression of σE-dependent genes (Fig. 7). Therefore, both sets of results (Schujman et al., 1998 and this work) suggest that de novo FA synthesis plays a regulatory role in the activation of SpoIIR and/or SpoIIGA to stimulate mother cell-compartmentalized σE activity.
Interestingly, in a recent work, Diez et al. (2012) proposed that SpoIIR-dependent σE activation is linked to de novo FA synthesis. They presented evidence for a model in which SpoIIGA (localized on either the forespore or mother cell membrane) interacts with SpoIIR, which allows for its vectorial concentration in the sporulation septum. There, on both septal membranes, SpoIIR is susceptible to covalent modification (acylation) with the fatty-acyl molecules that originated from de novo FA synthesis but not from de novo PL synthesis or previously synthesized FAs (‘aged FAs’) (Diez et al., 2012). Based on our results, we propose that, under physiological conditions, the de novo-synthesized fatty-acyl molecules needed for the processing and activation of SpoIIR (see Supporting Information) originate from the mother cell compartment, where de novo lipid acid synthesis is active (green membrane in Fig. S4). In this scenario, only the SpoIIR molecules that interact with SpoIIGA at the mother cell membrane (Fig. S4) would be acylated with de novo-synthesized FAs (Diez et al., 2012) to allow for SpoIIR processing and activation at the mother cell site of the sporulating cell. Because de novo FA synthesis is absent in the forespore compartment (once it receives its complete chromosome with the full set of genes for lipid synthesis, 15–20 min after completion of the asymmetric division, the activity of the lipid synthesis activator Spo0A is turned off, and the ACP molecules are restricted to the mother cell compartment, see Arabolaza et al., 2003; Fujita and Losick, 2003; Martínez et al., 2010), SpoIIR molecules interacting with SpoIIGA positioned on the forespore membrane of the sporulation septum (Diez et al., 2012) would not be activated, restricting σE activation exclusively to the larger compartment (Fig. S4).
Lipid synthesis and biofilm development
The production, secretion and assembly of the diverse components of the ECM are crucial for the capacity of B. subtilis (and bacteria in general) to develop into a mature biofilm (López and Kolter, 2010; Abee et al., 2011; Epstein et al., 2011; Kobayashi and Iwano, 2012). We envision that the requirement of de novo FA synthesis during the development of the biofilm in B. subtilis (Fig. 9) is not restricted to the supply of new lipids for cell membrane formation. This hypothesis is based on two experimental observations. First, we demonstrated that the Spo0A-proficient accDA0A cells were capable of healthy growth during the vegetative phase in DSM broth, with a generation time of symmetric (vegetative) division and a final cellular yield similar to those exhibited by wild-type cells, but failed to activate a specific developmental stage (activation of σE-dependent genes) during the stationary (sporulation) phase. These results suggest a regulatory, rather than structural, role for FAs during sporulation (see Figs 8 and S4). Second, the complete loss of water repellency of the accDA0A cells developed under biofilm-supporting conditions (Fig. 10D) strongly indicates that the new FAs formed under Spo0A control play an important role in the development of the floating capacity, impermeability and hydrophobic interactions of the cells within the biofilm (Figs 9 and 10). All of these properties are major determinants in biofilm formation. From these results, we infer that the deficiency in biofilm formation of accDA0A cells (Fig. 9) could have been due to a specific blockage during the development of the biofilm rather than a structural role of de novo-synthesized lipids as components of the cell membranes in the biofilm. When comparing the biofilm phenotype of accDA0A cells with the phenotype displayed by different biofilm-deficient mutant strains of B. subtilis (mutant strains tasA, epsG and bslA), we demonstrated that standing cultures of accDA0A cells developed weak pellicles similar to the ones formed by epsG and tasA cells (Fig. 10). The accDA0A cells failed to form structured cell clusters and floating pellicles unless exogenous FAs were added to the growth medium, suggesting that the biofilm-defective phenotype was due to the absence of de novo FA synthesis under conditions that supported the formation of the biofilm (Figs 9 and 10). These results indicate the requirement of de novo FA synthesis at an early stage of biofilm development, before cell cluster formation. What would be the role of de novo-synthesized lipids at an early stage of the development of the biofilm in B. subtilis? In addition to the presence of lipids (Conrad et al., 2003; Lazarevic et al., 2005; Theilacker et al., 2009; Abee et al., 2011), the hydrophobic characteristics of the biofilm have been attributed to the covalent modification of extracellular polymeric molecules with FA-related molecules (Neu and Poralla, 1988; Neu et al., 1992) and the secretion of biosurfactants to the ECM of the biofilm (Davey et al., 2003; López and Kolter, 2010; Epstein et al., 2011). Because the strain used in this study (JH642) does not form surfactin (the main biosurfactant required for the formation of robust and long-lasting biofilms in undomesticated strains of B. subtilis, Branda et al., 2001), it is tempting to speculate that ECM components of the B. subtilis biofilm (i.e. TasA, BslA and/or the exopolysaccharide) might be modified with FA-related molecules to achieve the extreme repellency and floating capacity of B. subtilis biofilms (Figs S2 and 10) (Epstein et al., 2011; Kobayashi and Iwano, 2012; Kovács et al., 2012). Although, to date, a post-translational modification of any ECM component of the biofilm of B. subtilis has not been reported (Abee et al., 2011; Kovács et al., 2012), this possibility should not be completely ruled out.
Strains, media and growth conditions
The B. subtilis strains used in this study are described in Table S1. B. subtilis was propagated in LB broth for experiments performed under non-sporulating conditions. For sporulation, B. subtilis was either grown in DSM, which triggered sporulation via nutrient exhaustion (Arabolaza et al., 2003), or was induced to sporulate via a resuspension method in Sterlini-Mandelstam medium (Schujman et al., 1998). β-Galactosidase activity was assayed as previously described, and the specific activity is expressed in Miller units (Arabolaza et al., 2003).
A procedure consisting of consecutive PCR steps (Reder et al., 2008) was used to obtain a B. subtilis strain carrying point mutations in the 0A-box (ttgtcgaa) present in the promoter region of the accDA operon. In the first step, an overlapping extension protocol was used to inactivate the 0A-box present in the promoter region of the accDA operon by introducing site-specific mutations in that 0A-box (ctgtagag); the mutated nucleotides are shown in bold. To this end, two sets of primers were used in two separate amplification reactions using 168 chromosomal DNA as template. One pair of primers was used to amplify a DNA fragment of 1059 bp that consisted of the 0A-box together with upstream sequences; the forward primer UP-F (5′-aagatattgcagcgccaaac-3′) and the reverse primer UP-R (5′-ctcaactctacagctttttattcgatg-3′). The second pair of primers was used to amplify a DNA fragment of 922 bp that consisted of the 0A-box together with the downstream sequences; the forward primer DW-F (5′-aaaaagctgtagagttgaggatgac-3′) and the reverse primer DW-R (5′-atgaataaccgcatcaagctg-3′). Once purified, these two PCR products, presenting a small overlapping sequence containing the mutated 0A-box, were mixed together and extended through the short homologous region in a third PCR using only the flanking primers (underlined). In this way, a linear DNA fragment of 1962 bp, which enclosed the complete promoter region of the accDA operon containing a mutated 0A-box, was obtained.
This 1962 bp DNA fragment was used as template for a modified two-step fusion protocol (Wach, 1996) designed to insert an erythromycin resistance marker upstream of the mutated 0A-box. The antibiotic resistance cassette was obtained through a standard PCR protocol using pMUTIN4 (Vagner et al., 1998) as template and the primers ER-F (5′-gtgccacctgacgtctaag-3′) and ER-R (5′-agtctggactgggctgtgtag-3′) as described (Reder et al., 2008). Using the 1962 bp DNA fragment as template and two sets of primers in separate amplification reactions, two DNA fragments with almost no overlapping sequence were obtained. One pair of primers was used to amplify an upstream DNA fragment of 1005 bp; the forward primer UP-F (5′-aagatattgcagcgccaaac-3′) and the reverse primer UP-RII (5′-cttagacgtcaggtggcacacaaggttgacagcgcattc-3′). The second pair of primers was used to amplify a downstream DNA fragment of 736 bp; the forward primer DW-FII (5′-ctacacagcccagtccagactcttgtcgtttcaacaccttatc-3′) and the reverse primer DW-RII (5′-aacttaatacgccttcctgcattc-3′). The erythromycin resistance marker is indicated with italics. The three DNA fragments were mixed and fused in a stepwise manner through PCR using the flanking primers UP-F and DW-RII. A similar procedure was performed using 168 chromosomal DNA as a template to obtain a wild-type linear DNA fragment containing the promoter region of the accDA operon with an erythromycin resistance marker upstream of the wild-type 0A-box. Both linear DNA fragments containing the erythromycin resistance marker upstream of the mutated and the wild-type 0A-boxes in accDA were used to transform competent wild-type B. subtilis cells to produce the strains RG1130 (eryrΩaccDA0A) and RG1340 (eryrΩaccDAwt) respectively. Transformed cells, into which we inserted the erythromycin resistance cassette by a double cross-over event, were selected on LB plates containing 1 μg ml−1 erythromycin, and chromosomal DNAs of these strains were sequenced to verify the correct resistance marker insertion, the presence of the desired mutation in the 0A-box and the absence of any other mutation in the strains.
To construct the β-galactosidase reporter strains RG1606 (accDA-lacZ cat at the amyE locus), RG1158 (eryrΩaccDA0A-lacZ cat at the amyE locus) and RG1160 (eryrΩaccDAwt-lacZ cat at the amyE locus) we PCR-amplified DNA fragments containing the promoter regions of accDA, eryrΩaccDA0A or eryrΩaccDAwt (see above) from the chromosomal DNAs of the wild-type strain 168 and strains RG1130 and RG1340 respectively. The oligonucleotides 5′-AAACAAGAAAAGGTACCATTATTGG-3′ (KpnI site underlined) and 5′-CATACCCGAAGCTTTTTATCCAACTC-3′ (HindIII site underlined) were used as primers. The amplified PCR fragments were cloned into PJM116. After linearization, the plasmids were introduced separately via transformation into competent 168 (or JH642) cells to generate strains RG1606, RG1158 and RG1160, as previously described (Arabolaza et al., 2003).
Analysis of radioactive lipids
To measure in vivo FA synthesis, B. subtilis strains were grown in their corresponding culture media with aeration for the indicated periods of time. Aliquots (1 ml) of the cultures were exposed to 10 μCi of sodium [1-14C]-acetate or [1-14C]-isoleucine at the time points indicated in the figure legends. After the incubation periods, the lipids were extracted as previously described (Grau and De Mendoza, 1993). FAs, PLs and GLs were separated by thin-layer chromatography on silica gel plates developed in petroleum ether/ether/acetic acid (70:30:10) and autoradiographed. In vitro FA synthase activity was measured as previously described (Schujman et al., 1998). Cerulenin was added at a final concentration of 10 μg ml−1, and [2-14C]-malonyl-CoA was added last to initiate the reactions at 37°C for 30 min. This was followed by lipid extraction, chromatography in petroleum ether/ether/acetic acid (70:30:10) and autoradiography. In all cases, appropriate areas of the silica gel were scraped into vials containing scintillation solution to determine their radioactive content, as previously described (Grau and de Mendoza, 1993, Schujman et al., 1998).
For biofilm (pellicle) formation, overnight B. subtilis cultures were grown in LB medium to the stationary phase. Then, 50 μl of each culture was diluted in 2 ml of fresh LB medium with (LBY) or without (LB) supplementation with 6.0 % yeast extract (biofilm growth medium). All the tubes (or microplates when indicated) were statically incubated at 30°C for 48 h.
Microtitre plate assay for B. subtilis biofilm formation rate
This assay measured the amount of cells adhered to the surface of the microtitre plate wells (Hamon and Lazazzera, 2001). B. subtilis cells were grown in 96-well polyvinylchloride (PVC) microtitre plates at 37°C in LBY biofilm medium. The bacterial inoculum for the microtitre plates was obtained by growing the cells in LB to mid-exponential phase and then diluting the cells to an OD600 of 0.08 in biofilm growth medium. Samples of 100 μl of the diluted cells were then aliquoted to each well of a 96-well PVS microtitre plate, including 12 wells with biofilm growth medium only (negative control). At various time points during the incubation of B. subtilis cells in the microtitre plate wells, the biofilm formation was monitored by staining with crystal violet (CV). Growth medium and non-adherent cells were carefully removed from the microtitre plate wells. The microtitre plate wells was incubated at 60°C for 30 min. Biofilms were stained with 300 μl of 0.3 % CV in 30% methanol at room temperature for 30 min. Excess CV was then removed, and the wells were rinsed with water. The CV that had stained the biofilms was then solubilized in 200 μl of 80% ethanol/20% acetone. Biofilms formation was quantified by measuring the Abs570nm for each well using a Beckman Coulter Multimode Detector (DTX 880). Each data point is an average of 12 wells, and error bars indicate the standard error (P < 0.05). Representative data from one of at least four independent experiments are shown.
This research was supported by grants to R. G. from CONICET (National Research Council of Argentina) and ANCyT (National Agency of Science and Technology) and the aid of the Pew Latin-American Program in Biological Sciences (Philadelphia, USA), the Fulbright Committee (Washington, DC, USA) and Fundación Antorchas (Buenos Aires, Argentina). We thank Rafael Carlucci and Betina Bonguiovani for assistance in the graphical design of figures. M. P., W. R., C. L. and A. G. are PhD. fellows from CONICET and P. de O. is a PhD. fellow from ANCyT. R. G. is a Career Investigator from CONICET and an International Fulbright Scholar. We thank to the anonymous referees that with their suggestions and observations contributed to the improvement of this work.