For all cloning procedures Q5 polymerase (NEB) was used. Constructs were assembled according to Gibson et al. (2009) from gBlocks® (IDT) or PCR products and the vector backbone. Sequence integrity was verified by DNA sequencing (GATC biotech). Bacterial strains and plasmids are listed in Table 3. Complementation of the P_II mutant was performed as described by Wolk et al. (1984) with plasmids pVZ322-P_II-Ven and pVZ322-P_IIS49E-Ven.

ACC of E. coli was extracted as described previously (Gerhardt et al., 2015). BirA was expressed in E. coli as described before (Gerhardt et al., 2015). P_II protein from Synechocystis and Synechococcus was purified as described previously (Heinrich et al., 2004). For expression of Synechocystis BCCP in E. coli BL21(DE3) was grown in 2YT medium at 37°C and expression was induced with IPTG at an OD₆₀₀ of 0.8. Induced culture was cultivated at 25°C over night. Cells were harvested at 4000 × g for 10 min., cell pellets were combined with a cell suspension overexpressing BirA in biotinylation buffer (50 mM HEPES pH8, 10 mM KCl, 5% v/v glycerol, 5 mM MgCl₂, 1 mM Biotin, 10 mM ATP, and 1 mM Benzamidine). Cells were homogenized with a Branson Sonifier S-250A and the lysate was incubated for 1 h at 37°C, followed by 4°C over night to biotinylate BCCP. BCCP was extracted from the cleared cell lysate (centrifugation for 30 min. at 25,000 × g at 4°C) through Ni-NTA affinity chromatography. Cleared lysate was loaded on a wash buffer (50 mM TrisHCl pH7.5, 100 mM KCl, 20% v/v glycerol, and 50 mM imidazol) equilibrated Histrap FF Crude column (GE healthcare). The column was washed with 10 column volumes wash buffer and bound protein was eluted with elution buffer (50 mM TrisHCl pH7.5, 100 mM KCl, 20% v/v glycerol and 500 mM imidazol). The eluted protein was dialyzed against a storage buffer (50 mM HEPES pH 7.8 100 mM KCl, 50% v/v glycerol) over night at 4°C. Biotinylation of BCCP was verified using immunoblotting and subsequent detection of biotinylated proteins using streptavidin-HRP conjugate with chemiluminscence.

ACC activity was measured by coupling ACC catalyzed ATP hydrolysis to the activities of pyruvate kinase (PK) and lactate dehydrogenase (LDH) as described (Beez et al., 2009; Broussard et al., 2013). The reaction buffer consisted of 50 mM imidazole, 50 mM KCl, 20 mM MgCl₂, 0.2 mM NADH, 1 mM phosphoenolpyruvate, 10 mM ATP, 0.5 mM DTT, 4.4 units of LDH, 6 units PK, and 10 mM NaHCO₃. The pH of the final reaction mixture was 7.5. Following concentrations of ACC subunits were used for the enzyme assay: 10 nM carboxyl transferase (tetramer), 20 nM biotin carboxylase (dimer) and 200 nM biotin carboxyl carrier protein (monomer). Different concentrations of P_II and 2-OG were used as indicated in the text. The reactions were pre-incubated for 15 min. and started by the addition of acetyl-CoA 400 μM. The oxidation of NADH to NAD⁺ was recorded at 25°C over 20 min. in a SPECORD 200 photometer (Analytik Jena) at 340 nm. From the slope of decreasing absorption, reaction velocity was calculated with an extinction coefficient for NADH of 6220 M⁻¹. For the determination of catalytic constants, the data were fitted to Michaelis-Menten equation using GraphPad prism software

Prior to protein co-precipitation experiments BCCP conformation was checked by size exclusion chromatography (20 mM potassium phosphate buffer pH 7.8 100 mM NaCl) ensuring properly folded BCCP was used for experiments. 30 μl Ni-NTA agarose coated magnetic beads (Quiagen) preequilibrated in binding buffer (50 mM TrisHCl pH 8.0, 100 mM NaCl, 0.1% w/v N,N-Dimethyldodecylamine N-oxide (LDAO), 10% v/v glycerol, and 20 mM imidazole) were used. Binding was performed in 700 μl binding buffer with magnetic beads, 30 μg BCCP, and 35 μg P_II for 20 min. at room temperature. Unbound protein was washed off, three times with 300 μl binding buffer and bound proteins were eluted in 20 μl elution buffer (50 mM TrisHCl pH 8.0, 100 mM NaCl, 0.1% w/v N,N-Dimethyldodecylamine N-oxide (LDAO), 10% v/v glycerol, and 500 mM imidazole). Various metabolites were added to the binding buffer with final concentrations as indicated. Eluted fractions were analyzed by Tricine-SDS PAGE (Schägger, 2006) and stained with InstantBlue (Expedeon). Stained gels were scanned and band intensities were analyzed densitometrically. Scanned images were gray scaled and inverted with Adobe PhotoshopCS6, mean pixel intensities were determined for the P_II protein band and used as a proxy for protein abundance for subsequent analysis.

Synechocystis sp. PCC 6803 was grown in BG11 medium (Rippka et al., 1979) at 27°C, supplemented with 5 mM NaHCO₃ on a rotary shaker at light intensities of 50–80 μmol photons s⁻¹m⁻². For imposing nitrogen-starvation conditions, cells were first grown in BG11 medium to an optical density (750 nm) of 0.4–0.6, harvested by centrifugation at 4000 × g for 10 min., then washed with BG11-N (BG11 lacking NaNO₃), pelleted again at 4000 × g for 10 min., and finally re-suspended in BG11–N (supplemented with 5 mM NaHCO₃) to an OD₇₅₀ of 0.4. For growth with ammonium, cells were grown to OD₇₅₀ of 0.6–0.8 and diluted in BG11-N medium to OD₇₅₀ 0.1. The medium was buffered with TES pH8, supplemented with NaHCO₃ and NH₄Cl to a final concentration of 5 mM.

To estimate the intracellular acetyl-CoA levels 20 ml of growing culture was pelleted at 4000 × g for 10 min. and frozen at −80°C until measured. Cell pellets were suspended in 200 μl 1 M cold perchloric acid. Suspended cells were lysed using a FastPrepR-24 (MP Biomedicals) for 30 s and 6.5 m/s five times with glass beads (0.1–0.11 mm diameter). Cell debris and glass beads were pelleted at 13,000 × g at 4°C for 10 min. The supernatant was neutralized with 3 M KHCO₃ and excess KHCO₃ was removed through centrifugation at 13,000 × g for 2 min. at 4°C. The clear supernatant was used for acetyl-CoA measurements using the Acetyl-CoenzymeA kit (Sigma-Aldrich) according to the manufacturer's instruction. Fluorescence intensities were measured using a SpectraMax M2 microplate reader with λ_ex = 535 nm and λ_em = 587 nm.

Fatty acids were quantified as described previously (Wawrik and Harriman, 2010). Cell pellets of 2 ml culture were thawed in in 200 μl saponification reagent (25% methanol in 1N NaOH) and lysed with glass beads (0.1–0.11 mm diameter) in a FastPrepR-24 (MP Biomedicals) for 30 s and 6.5 m/s five times. Cell lysates were saponified for 30 min. at 95°C and vortexed every 5 min. Cell extracts were neutralized with 200 μl neutralization reagent (1N HCl, 100 mM Tris pH 8.0) and copper reagent (9 vol. aq. 1 M triethanolamine, 1 vol. N-acetic acid, 10 vol. 6.45% (w/v) Cu (NO₃)₂·3H₂O). Samples were vortexed for 2 min. and 250 μl chloroform was added and vortexed for additional 2 min. Phase separation was achieved by centrifugation and 50 μl of the organic phase was transferred in two separate new tubes. In one tube 50 μl 2-butanol were added and used as blank. The second tube was mixed with 1% (w/v) sodium diethyldithiocarbamate in 2-butanol leading to color development in the sample. Absorption was measured at 440 nm in a SpectraMaxM2 microplate reader and the absorption of the blank was subtracted from the sample manually. Lipid concentration was estimated based on a standard curve with palmitic acid.

Two hundred milliliters exponentially growing culture were harvested at 4000 × g at 25°C, the cell pellet was washed once with water, pelleted at 20,000 × g for 3 min., frozen in liquid nitrogen and stored at −80°C until used. Cell pellets were lyophilized for 16 h. Pentadeconoic acid was added to 20 mg CDW which was used for saponification with 1 ml 3,75 M NaOH in 50% methanol (v/v) for 35 min. at 100°C. Free fatty acids were methylated by addition of 2 ml methylation reagent (3.25 M HCl in 45% methanol (v/v) for 12 min. at 80°C. Fatty acid methyl esters (FAME) were extracted with 2 ml n-hexane through vortexing and 10 min. incubation on a revolving laboratory mixer. The organic phase was transferred in a new vial to which 3 ml 0.3 M NaOH were added and incubated for 10 min. on a revolving laboratory mixer. The organic phase was transferred in a GC vial and evaporated under nitrogen gas flow at 60°C. FAME were dissolved in 50 μl dichlormethane and analyzed by gas chromatography.

GC analysis was performed with a Shimadzu GC9A equipped with a FID detector and a DBWAX-30 W (30 m × 0.319 mm) column with nitrogen as carrier gas. 5 μl of sample was injected, the injector and detector temperature was set at 250°C. The Oven temperature increased from 160° to 200°C at a rate of 4°C per minute, and from 200 to 240°C at 8°C per minute and remained constant for 10 min. at 240°C. Fatty acid methyl esters were identified based on retention times determined with commercially available fatty acid methyl esters. FAME were quantified using response factors with pentadecanoic acid as internal standard.

Two hundred milliliters exponentially growing culture was pelleted at 4000 × g for 10 min. at 25°C. The pellet was washed with deionized water, cells were pelleted at 20,000 × g for 3 min. and the pellet was frozen at −20°C until further use. Bacterial pellets were dried in a centrifugal evaporator for 16 h at 25°C. Dried cell matter (15–40 mg CDW) was used for lipid extraction as described before (Bligh and Dyer, 1959). Dried material was transferred in a glass vial with a PTFE lined screw cap lid, suspended in 3 ml Methanol:Chloroform (2:1), vortexed vigorously and incubated for 1 h on a revolving laboratory mixer. After incubation 1 ml chloroform and 1.8 ml deionized water were added, vortexed and phase separation was induced through centrifugation for 10 min. at 4000 × g. The organic phase was transferred in a fresh glass vial and the aqueous phase was extracted twice with 1 ml chloroform followed by 4 ml Isooctane:Ehtylacetate (3:1). All organic phases were combined and solvents were evaporated under nitrogen gas stream. Lipids were suspended in either 200 μl Chloroform: Methanol (1:1) or Hexane:Ether:Acetic acid (80:20:1).

To 100 μl Synechocystis cell suspension 1 μl Bodipy® 493/503 (10 mg/ml in DMSO) was added and incubated for 5 min. Cells were pelleted at 10,000 × g for 2 min. and cell pellets were suspended in PBS buffer pH 7.5. Two microliter were dropped on a poly lysine coated glass slide and examined using a Leica DM5500B microscope. Image acquisition was performed with a Leica DFC360FX black and white camera, fluorescence images were recolored using Leica application suite. Green fluorescence was detected using an excitation filter BP470/40 and an emission filter BP525/50. Fluorescence images were acquired with 100 ms exposure time. Bright field images were acquired with 6 ms exposure time. Intensity levels of images were adjusted using PhotoshopCS6.

Lipid extracts were spotted on silica gel 60 (Merck Millipore) TLC plates. Phospholipids were resolved using Chloroform:Methanol:NH₄OH (70:30:5) as mobile phase (Merritt et al., 1991). Glycolipids were visualized spraying the plates with 2.4% (w/v) α-naphtol in 10% sulfuric acid 80% (v/v) ethanol and baking the plate at 120°C until purple spots were visible (Wang and Benning, 2011). Neutral lipids were resolved using a Hexane:Ether (90:10) mobile phase and stained with iodine vapor (Ruiz-Lopez et al., 2003). Individual spots were scraped of and lipids were extracted with Hexane:Ether:Acetic acid (80:20:1).

Solvent extracted lipids from silica gel were subjected to saponification and FAME were synthesized as described above. FAMEs were detected using a Shimadzu GC17A with a QP-5000MS (GC-MS) using an optima 5MS (15 m × 0.25 mm) column with Helium as carrier gas. 5 μl of sample was injected, the injector temperature was set at 320°C. The column was heated to 90°C and the temperature was hold for 5 min., heated up at a rate of 20°C/min. to 200°C, heated at a rate of 4°C/min. to 300°C and hold for 2 min. at 300°C. The MS detector voltage was set at 1.65 keV.

The E. coli acetyl-coenzyme A carboxylase (ACC) was used in a previous study as a model system to investigate the effect of GlnB/GlnZ from Azospirillum brasilense and GlnB/GlnK from E. coli on enzyme activity (Gerhardt et al., 2015). Here, we first investigated the effect of several characterized Synechococcus elongatus PCC7942 P_II protein variants (ScP_II) on ACC activity. Initial assays were carried out at a fixed concentration of 10 mM ATP. Synechococcus GlnB (ScGlnB) was able to efficiently inhibit the E. coli ACC activity and increasing concentrations of GlnB correlated with increased inhibition of ACC (Figure 1A). The maximum inhibition was calculated to be 93% (SE: 6.8%) with an EC₅₀ for ScGlnB of 0.31 μM (SE: 0.066 μM). As interaction of BCCP and Azospirillum GlnB was shown to be affected by 2-OG, ACC activity was measured in presence of 1 μM GlnB and various 2-OG concentrations (Figure 1B). Increasing concentrations of 2-OG were able to efficiently relief ACC from GlnB-dependent inhibition and the IC₅₀ value for 2-oxoglutarate was calculated to be 4.8 μM (SE: 0.2 μM), which is almost exactly the K_d of the first 2-OG binding site (5.1 μM) of GlnB (Fokina et al., 2010b). To reveal, which positions in ScP_II are important for ACCase regulation, various variants of ScP_II were tested in their ability to inhibit ACC activity (Figure 1C). Point mutations in the T-loop of R45 and R47 to alanine and the phosphomimetic S49D/S49E variants were not as efficient in inhibiting ACC activity as wild type P_II. In contrast, mutations of S49G, Y51A, and E54A in the T-loop and E85A were not affected in inhibiting ACC activity. Two P_II variants (I86N and R103H) were, however, completely unable to inhibit ACC activity. Addition of 1 mM 2-OG to the reaction relieved ACC inhibition in all P_II variants. This confirms the previous assumption, that GlnB regulation of ACCase activity is highly conserved in bacteria.

Since the aim of this study was to characterize the physiological effect of P_II on the acetyl-CoA metabolism, but the P_II mutant of Synechococcus accumulates second site mutations in pipX (Espinosa et al., 2009) we decided to study this effect in the P_II mutant of Synechocystis, in which pipX and ntcA are not affected. Even though ScGlnB shares 95% sequence identity with GlnB of Synechocystis (SyGlnB) we wanted to verify the interaction of SyGlnB and Synechocystis BCCP proteins in vitro. To this end, recombinant proteins were expressed and purified from E. coli. His-tagged BCCP was used as bait protein using Ni-NTA coated magnetic beads. BCCP GlnB interaction was strictly dependent on the presence of Mg²⁺ ions and ATP. Like in A. brasilense and E. coli, 2-OG negatively affected the ATP-dependent P_II binding to BCCP (Figure 2A). No P_II protein could be recovered in the presence of ADP. An ATP titration experiment was performed and the amount of co-precipitated protein was plotted against the ATP concentration (Figure 2B). The apparent EC₅₀ for ATP was determined to be 68 μM (SE 13.2 μM) through non-linear fitting and is in good agreement with the K_d of the third ATP binding site of cyanobacterial GlnB (47.4 μM), which exhibits three anticooperative sites (Fokina et al., 2010b). The same type of analysis was performed for 2-OG, titrated in the presence of a fixed concentration of 0.5 mM ATP (Figure 2C). The apparent IC₅₀ value was calculated, assuming dose response dependent inhibition using a standard slope. The resulting IC₅₀ for 2-oxoglutarate was determined to be 41.3 μM (SE 1.7 μM). This value is lower than the K_d of the third GlnB 2-OG binding site (106.7 μM) but well above the Kd of the second site (11.1 μM) (Fokina et al., 2011), which suggests that occupation of the third 2-OG binding determines dissociation of the SyGlnB-SyBCCP complex. GlnB is known to be phosphorylated in vivo at position Ser49 under nitrogen-poor conditions or high CO₂-supply to nitrate-grown cells. In the case of P_II-NAGK interaction, Ser49 phosphorylation prevents complex formation (Heinrich et al., 2004) and the phosphomimetic variant S49D was unable to interact with NAGK (Llácer et al., 2007). As shown above, the phosphomimetic variants of ScGlnB (S49D and S49E) had reduced efficiency in inhibiting E. coli ACCase. To find out, how phosphomimetic variants SyGlnB are affected in binding the cognate SyBCCP protein, the affinity of SyGlnB variants S49D, S49E, S49C, and the wild type protein were tested toward SyBCCP through pull down experiments (Figure 2D). Instead of using the S49G variant we decided to use the S49C variant as mutation of S49 to glycine could have a negative impact on complex stability (Lüddecke and Forchhammer, 2013). The S49E variant was completely unable to bind BCCP. The other negatively charged variant S49D, weakly interacted with BCCP, showing only about 20% maximal binding as compared to wild-type GlnB. Likewise, the EC₅₀ for GlnB increased 4-fold compared to wild type GlnB. By contrast, substitution of Ser49 to Cys had only a minor effect on GlnB-BCCP interaction (about 90% maximal binding), indicating that the negative charge at position 49 that impairs BCCP-GlnB interaction. As ATP binding influences the T-loop conformation, a titration of ATP with the two variants S49C and S49D was performed (Figure 2E). The S49C mutation increases the calculated EC₅₀ value for ATP from 68 to 143 μM (SE 24 μM) and to 231 μM (SE 36 μM) for the S49D variant. Maximum binding of GlnB was calculated to be 88.2 mean pixel intensity (SE 3.8 mpi) for the S49C variant which was almost identical to the wild type protein (86.6 mpi; SE 3.8 mpi), but was much lower for the S49D variant with 32.2 mpi (SE 1.7 mpi) at saturating ATP concentrations. On the one hand, the doubling of the EC₅₀ for ATP implies that substitution of serine 49 to cysteine (which is bulkier) requires increased ATP concentrations to fit the T-loop into a conformation that binds to BCCP. At excess ATP concentrations, the mutation had no influence on the total amount of GlnB that can be co-precipitated with BCCP, in agreement with the GlnB titration experiment above. On the other hand, when the T-loop carries the S49D mutation, more than three times higher ATP concentrations were required to enforce the appropriate conformation for complex formation with BCCP. Moreover, the stability of the complex was reduced to one third, as compared to the complex with wild-type GlnB. Taken together, introduction of a negative charge at position 49 in the T-loop of P_II destabilizes the BCCP-GlnB complex. This, together with the fact, that the S49E SyGlnB variant was completely unable to interact with BCCP, strongly indicates that phosphorylated P_II will not be able to interact with BCCP.

The in vitro experiments showed that GlnB directly binds BCCP and affects ACC activity, and furthermore, interaction is sensitive to Ser49 modification. From these findings, we hypothesized that phosphorylation of P_II should have an impact on either fatty acid or acetyl-CoA metabolism during varying carbon-nitrogen regimes, which correspond to different degrees of P_II phosphorylation in Synechocystis (Forchhammer and Tandeau de Marsac, 1995a). This regulation should be abolished in a Synechocystis P_II mutant. To examine this prediction, wild type and P_II mutant strains were grown with different nitrogen and carbon supply (nitrate or ammonia as nitrogen source, gently shaking without aeration, corresponding to the lowest CO₂ supply; or vigorous bubbling with either ambient air (0.04%) or 2% CO₂). The expected phosphorylation status of P_II was verified (Supplementary Figure 1) and cellular acetyl-CoA levels as well as total fatty acid concentrations were determined in exponentially growing cultures under these conditions. Regardless of the carbon or nitrogen regime, the acetyl-CoA level in the P_II mutant was always much lower than in the wild type (Figure 3A). Remarkably, the acetyl-CoA levels in the wild type differed with changing carbon and nitrogen regimes. In particular in nitrate grown cells, the acetyl-CoA levels decreased significantly in presence of 2% CO₂ supply. Under these conditions, P_II displays the highest degree of phosphorylation, and acetyl-CoA levels in wild-type and mutant cells are similar. However, under any condition that leads to a low degree of P_II phosphorylation (either nitrate grown with limiting CO₂-supply or ammonia grown cells), the acetyl-CoA levels were strongly increased, whereas it stayed low in the P_II deficient mutant. Total fatty acid levels did not differ as much as the acetyl-CoA levels, but slightly higher fatty acid levels in the P_II mutant were always visible. The differences were particularly significant in ammonia grown cells with low carbon supply, where P_II is always present in the non-phosphorylated state in the wild type (Figure 3B). The carbon regime had a marked impact on the fatty acid content, in both strains. Increased CO₂ supply favored a higher intracellular lipid content. This effect is probably due to improved CO₂-fixation, that will ultimately result in increased CO₂-fixation products than can flow into various anabolic pathways.

Nitrogen starvation represents the situation of maximal P_II phosphorylation (Forchhammer and Tandeau de Marsac, 1995a). If the assumption is correct, that P_II phosphorylation abrogates its inhibitory effect on ACCase, then the differences in acetyl-CoA levels between wild-type and P_II mutant should disappear under those conditions. Therefore, we analyzed acetyl-CoA and total fatty acid levels of cells subjected to 8 h nitrogen-starvation and compared it to conditions during exponential growth with nitrate as nitrogen source. In agreement with our expectation, in 8 h nitrogen-starved cells, the acetyl-CoA levels dropped in the wild-type to the low levels observed in the P_II mutant (Table 1). As acetyl-CoA levels in E. coli decrease during late exponential (Chohnan and Takamura, 1991) and stationary phase, the growth phase dependence of acetyl-CoA levels was measured in Synechocystis strains. In the wild-type, acetyl-CoA levels were high during exponential growth and decreased with increasing optical densities. As already shown above, strongly reduced levels of acetyl-CoA in the P_II mutant were visible over all time points (Figure 4A). Complementation of the P_II mutant with the wild-type glnB gene was able to complement the low acetyl-CoA level phenotype, but introduction of the gene encoding the P_II S49E variant, which was not able to interact with BCCP retained the mutant phenotype. Total fatty acid levels generally increased during growth and the difference between wild type and P_II mutant got smaller at the later stages of growth but was significantly different in the first 48 h of growth (Figure 4B). The complemented strain had similar fatty acid levels as the wild type, but the S49E variant was not able to complement the P_II mutant phenotype. The difference between wild-type and P_II mutant in fatty acid levels during ammonia-supplemented growth was verified using GC analysis. GC results matched the values obtained with the colorimetric assay but additionally provided qualitative information, how fatty acid composition might be altered. As shown in Table 2, mutation of GlnB shifted the molar composition of fatty acids, which increased the amount of palmitic acid by about 15% at the same time decreasing the amount of linoleic acid to the same extent. The fatty acid profile of the P_II complemented strain was very similar to that of the wild type, whereas the S49E strain had a fatty acid composition reminiscent of the P_II mutant exemplifying that the S49E P_II variant is a loss-of function mutant with respect to regulation of fatty acid metabolism. Triple unsaturated fatty acids were increased in both complemented strains.

Intracellular lipids can be visualized microscopically using the hydrophobic dye Bodipy® 493/503, which gives a green fluorescence and specifically stains neutral lipids (Gocze and Freeman, 1994). Therefore, we examined wild-type and P_II deficient mutant cells by fluorescence microscopy. A strong intracellular fluorescence signal could be detected in some wild type cells taken from early exponential phase of growth, as exemplarily shown in Figure 5A. The number of lipid bodies per cell was determined and is shown in Figure 5B. Cells of the P_II mutant have at least one or two lipid bodies (mean 1.6 lipid droplets per cell), whereas only few cells have lipid bodies in wild type (mean 0.39 lipid droplets per cell). Lipid droplets formed transiently in the early phase of growth and disappeared with increasing optical densities, possibly being converted to phospholipids. To gain further insights in this phenotype, total lipids were extracted from exponentially growing cultures and the phospholipid content was analyzed using thin layer chromatography. No significant difference in phospholipid content was apparent between wild type and the P_II mutant excluding the accumulation of phospholipids in the observed vesicles. Hence the extracts were subjected to thin layer chromatography using a system, which is able to resolve more hydrophobic lipids (Figure 5C). Staining with iodine vapor revealed spots occurring in both wild type and P_II mutant and an additional spot only present in the P_II mutant. These spots migrate similar to a triacylglycerol standard (composed of C12, C14 and C16 triacylglycerols) and sesame oil (a complex mixture of C16 and various C18 fatty acids containing triacylglycerols). Stained spots were scraped off, extracted and converted to fatty acid methyl esters for GC/MS analysis. The lower spot contained primarily palmitic and stearic acid and minor traces of pentadecaonic and heptadecanoic fatty acid. The upper spot present in the P_II mutant contained primarily palmitic and stearic fatty acids (with no pentadecaonic and heptadecanoic fatty acids present). No unsaturated C16 or C18 fatty acids could be detected in both spots.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.