Introduction

Rhodospirillum rubrum has a powerful set of metabolic tools allowing photoautotrophy, chemoautotrophy, photoautolithotrophy, and heterotrophy (Gordon and McKinlay 2014; Imhoff and Trüper 1992; Munk et al. 2011; Selao et al. 2011). This bacterium is well studied due to its biotechnological potential in the production of pigments, vitamins, coenzymes (Carius et al. 2013), and hydrogen (Basak and Das 2007). R. rubrum is also known to accumulate the biopolymer polyhydroxybutyrate (PHB) with the vast majority of PHB accumulation studies performed in anaerobic conditions (Brandl et al. 1989; Do et al. 2007; Smith et al. 2008). PHB is primarily considered a carbon and energy storage polymer, accumulated as a response to a nutrient imbalance conditions (Steinbüchel and Lütke-Eversloh 2003). There is growing evidence supporting additional physiological roles of this type of biopolymer. A link between polyhydroxyalkanoate (PHA) accumulation and stress tolerance was observed in P. aeruginosa, showing that PHA has more roles than merely a carbon and energy storage molecule (Jendrossek and Pfeiffer 2014; Pham et al. 2004; Raberg et al. 2014; Zhao et al. 2007). There are suggestions that PHA metabolism also acts as a regulatory mechanism optimising the carbon and energy flow in the cell (Arias et al. 2013; Eggers and Steinbüchel 2014; Escapa et al. 2012). It has been speculated that PHB has a potential to act as an electron sink during anaerobic photoheterotrophic growth of R. rubrum (Gordon and McKinlay 2014).

In the current study, we performed a proteomic analysis of chemoheterotrophic growth of R. rubrum under PHB-accumulating conditions, which showed the expression of ribulose 1,5-bisphosphate carboxylase (RuBisCO), the ethylmalonyl-CoA (EMC) pathway and biopolymer PHB accumulation proteins. We undertook biochemical and physiological studies to examine the effect of stimulated PHB accumulation on metabolic processes in R. rubrum. We show here that RuBisCO, an enzyme previously exclusively associated with anaerobic and photosynthetic growth (Grammel et al. 2003), is active in R. rubrum under aerobic growth conditions in the dark, but the expression and activity are negatively influenced by PHB accumulation. Moreover, PHB unusually stimulates acetate assimilation into biomass and not into further PHB accumulation. Finally, the EMC pathway is used for acetate assimilation in R. rubrum, with the level and profile of crotonyl carboxylase/reductase (Ccr) activity similar in the wild type and mutant with a 6-h time delay between them for when maximum Ccr activity is observed.

Materials and methods

Bacterial strain, medium and growth conditions

All experiments were performed with R. rubrum S1 strain (ATCC 11170) or R. rubrum ΔphaC1ΔphaC2 (Klask et al. 2015). Cultures were grown in modified SMN medium (Bose et al. 1961) containing the following, per litre: 3 g KH2PO4; 4.5 g K2HPO4; 0.2 g MgSO4∙7H2O; 75 mg CaCl2∙2H2O; 20 mg EDTA; 0.1 mL vitamin solution (per litre: 150 mg biotin; 44 mg thiamine hydrochloride; 44 mg nicotinamide; 22 mg p-aminobenzoic acid; 22 mg riboflavin); 1 mL trace elements solution (per litre: 4 g ZnSO4∙7H2O; 1 g MnCl∙4H2O; 0.2 g Na2B4O7∙10H2O; 0.3 g NiCl2∙6H2O; 1 g Na2MoO4∙2H2O; 1 g CuCl∙2H2O; 7.6 g FeSO4∙7H2O) and 1.25 g for full or 0.31 g (NH4)2SO4 for nitrogen-limited growth.

Precultures of R. rubrum were incubated for 48 h in 2 mL of modified SMN medium in 13-mL sterile Sarstedt tubes wrapped in tin foil at 30 °C and 200 rpm in the dark. R. rubrum cultures were grown in 250-mL conical flasks wrapped in aluminium foil, containing 50 mL of modified SMN medium, at 30 °C and 200 rpm in the dark. For cultures grown with malate and acetate (MA) 4 g L−1 disodium malate (Sigma-Aldrich, Dublin, Ireland) and 3.7 g L−1 sodium acetate (Sigma-Aldrich, Dublin, Ireland) were added. These quantities corresponded to an equal amount of carbon (1.08 gc L−1) coming from each carbon source. Acetate was added at T24. For control experiments, R. rubrum was grown with 8 g L−1 (2.16 gC L−1) disodium malate (M) as a sole source of carbon and energy.

Cell density and photosynthetic membrane formation was monitored spectrophotometrically (OD660, OD680, OD880; Unicam Helios d UV/VIS spectrophotometer). Cell dry weight (CDW) was obtained by lyophilising the cell pellet (FreeZone Plus 6 Liter Cascade Freeze Dry Systems, Labconco, Missouri, USA) obtained after centrifugation. The photosynthetic membrane formation was assessed using OD880/OD660 ratio (Carius et al. 2011), which was in the range 0.5–0.55 in all our experiments indicating that the cultures were fully aerobic.

Cells were collected at different time points of growth for different experiments, harvested by centrifugation (5000 g for 10 min at 4 °C; Benchtop 5430R centrifuge; Eppendorf, Hamburg, Germany) and washed twice with 10 mL of phosphate buffer (50 mM, pH 7). The pellets were used immediately for further analysis or stored at −80 °C.

PHA extraction and content determination

The polymer content was assayed by subjecting the lyophilised cells to acidic methanolysis as previously described (Lageveen et al. 1988). The PHA monomers’ methylesters were assayed by GC using a Hewlett-Packard 6890N chromatograph equipped with an HP-Innowax capillary column (30 m × 0.25 mm, 0.50 μm film thickness; Agilent Technologies) and a flame ionisation detector (FID), using the temperature programme previously described (Lageveen et al. 1988). Total PHA content was determined as a percentage of CDW.

Nutrient analysis

Nitrogen concentration (absolute amount of N) was determined using the method of Scheiner (1976). Malate utilisation was determined from supernatant by HPLC using a C18 ACE column (250 mm × 4.6 mm, particle size 5 μm; Advanced Chromatographic Technologies Ltd.). The samples were isocratically eluted with 20 mM KH2PO4 (pH 2)/methanol (95:5) at a flow rate of 1 mL min−1. Acetate consumption was determined by acidifying 1 mL of the supernatant with 50 % H2SO4 (v/v), followed by addition of 100 μL saturated NaCl and extraction with 3 mL of chloroform/methanol (1:1, v/v) solution. The organic layer was analysed by a GC-FID detector fitted with an HP-Innowax column (30 m × 0.25 mm, 0.25 μm film thickness, Agilent Technologies), using the following conditions: 120 °C for 1 min, temperature ramp for 10 °C min−1 up to 250 °C and holding for 2 min.

Protein concentration

The protein concentration was measured using the Bicinchoninic Acid Kit (Sigma-Aldrich, Dublin, Ireland).

Proteomic analysis

Pellets obtained from 50-mL cultures grown with M or MA for 24, 30 or 72 h were lysed in an appropriate volume of 8 M urea solution to give a uniform protein concentration of 100 μg μL-1. The samples were reduced with 5 mM dithiothreitol and alkylated with 10 mM iodoacetamide. pH was adjusted to 7 with 200 mM NH4HCO3, and proteins were digested using proteomics grade trypsin in a convenient, single-use 1-μg package (Trypsin Singles, Sigma-Aldrich, Dublin, Ireland). The pH of the solution was brought to 2.5 with formic acid, and peptides were purified using ZipTip C18 columns (Merck Millipore, Cork, Ireland). The resulting tryptic peptide mixtures were resuspended in 0.1 % (v/v) formic acid and analysed by Q Exactive Hybrid Quadropole-Orbitrap Mass Spectrometer (MS) (Thermo Scientific, Paisley, UK) connected to a Dionex UltiMate 3000 (RSLCnano) chromatography system (buffer A: 97 % water, 2.5 % acetonitrile, 0.5 % acetic acid; buffer B: 97 % acetonitrile, 2.5 % water, 0.5 % acetic acid. All solvents are of LCMS grade). Each sample was loaded onto a fused silica emitter (75 μm ID, pulled using a laser puller (Sutter Instrument P2000)), packed with Reprocil Pur C18 (1.9 μm) reverse phase media column, and was separated by an increasing acetonitrile gradient over 60 min at a flow rate of 250 nL min−1. The mass spectrometer was operated in positive ion mode with a capillary temperature of 320 °C and a potential of 2300 V applied to the frit. All data was acquired with the MS operating in automatic data-dependent switching mode. A high-resolution (70,000) MS scan (300–1600 m/z) was performed using the Q Exactive to select the 12 most intense ions prior to MS/MS analysis using HCD.

Spectra were identified, and proteins were quantified using the Andromeda peptide identification algorithm integrated into MaxQuant (Cox and Mann 2008; Cox et al. 2011), searching against the UniProt database of R. rubrum S1 (www.uniprot.org, downloaded October 23, 2013). Proteins with a 1.5-fold change or higher and a significant change in t test (FDR 0.05; S0 0.1) were automatically accepted, while spectra with no specific change were manually checked for quality. Label-free quantification (LFQ) was used to compare the expression level of proteins present in different samples (Wang et al. 2003). David bioinformatics was used for the functional annotation and addition of KEGG pathways (KEGG 2015; Huang Da et al. 2009).

Metabolomic analysis

The pellets obtained from 50-mL cultures were resuspended in an appropriate volume of cold methanol 75 % (v/v) with respect to total protein concentration (80 μg μL-1). Cell suspensions were sonicated for 10 min on ice (Sonic Dismembrator FB120, Fisher Scientific, Dublin, Ireland; 40 % amplitude, 1 s on 1 s off) and centrifuged for 5 min (16,000 g, 4 °C), and the cell extracts were dried under a flush of N2. Samples were reconstituted with 590 μL deuterium oxide and 10 μL sodium trimethyl-[2,2,3,3-2H4]-propionate (TSP) (0.005 g mL−1). 1H nuclear magnetic resonance (NMR) spectra were acquired on a 600-MHz Varian NMR spectrometer (Varian Limited, Oxford, UK) by using the first increment of a nuclear Overhauser effect spectroscopy (NOESY) pulse sequence at 25 °C. Spectra were acquired with 16,384 data points and 512 scans. Water suppression was achieved during the relaxation delay (2.5 s) and the mixing time (100 ms). All spectra were referenced to TSP at 0.0 ppm. 1H NMR spectra were processed manually with Chenomx software (version 7.5; Chenomx Edmonton, Canada) and were phase and baseline corrected. The metabolites were quantified using Chenomx Profiler. Statistical analyses were performed using IBM SPSS Statistics 20. Differences in metabolic levels across the sample groups were examined using ANOVA and independent t tests.

Enzymatic assays

Harvested cells were resuspended in phosphate buffer (50 mM, pH 7), lysed using BugBuster (Merck Millipore) according to the manufacturer’s instructions and centrifuged (16,000 g, 30 min, 4 °C). The supernatant was used for the assays. The total protein concentration in the supernatant was determined by BCA assay. All enzymatic assays were performed in triplicate for each time point, and results were statistically analysed using ANOVA and independent t tests.

Ribulose 1,5-biphosphate carboxylase activity assay

The activity of RuBisCO was determined by following the rate of NADH (Sigma; εNADH = 6.3 mM−1 cm−1) oxidation by cell-free extracts of R. rubrum at 340 nm as previously described (Reid et al. 1997). Briefly, the assay (200 μL) contained 100 mM bicine buffer (pH 8), 25 mM KHCO3, 0.25 mM NADH, 3.5 mM ATP, 20 mM MgCl2, 5 mM phosphocreatine, 4.7 U glyceraldehyde-3-phosphate dehydrogenase from rabbit muscle (E.C. 1.2.1.12; Sigma-Aldrich, Dublin, Ireland), 4.7 U 3-phosphoglyceric phosphokinase from baker’s yeast (E.C 2.7.2.3; Sigma-Aldrich, Dublin, Ireland), 4.7 U creatine phosphokinase from bovine heart (E.C 2.7.3.2; Sigma-Aldrich, Dublin, Ireland) and 20 μL of cell-free extract containing 100 μg μL-1 of total proteins as a source of RuBisCO. The reaction was started by adding 0.5 mM ribulose 1,5-bisphosphate (Sigma-Aldrich, Dublin, Ireland). Appropriate controls were used to demonstrate the specificity of the reaction, including reactions without ribulose 1,5-bisphosphate or NADH, and a reaction with cell-free extract boiled at 100 °C for 5 min (negative controls).

Crotonyl-CoA carboxylase/reductase activity assay

The enzymatic activity of Ccr was assayed as previously described (Erb et al. 2009a), with slight modifications. Cell-free extract was used as a source of Ccr activity. The 182-μL standard reaction contained the following: 180 mM Tris–HCl (pH 7.9), 0.25 mM NADH or NADPH (Sigma-Aldrich, Dublin, Ireland), 0.5 mM crotonyl-CoA (Sigma-Aldrich, Dublin, Ireland) and 20 μL of cell-free extract (100 μg/μL total protein concentration). The reaction was started by the addition of mixture of 20 μL 50 mM KHCO3 and 6 μL 1 M acetic acid to dissolve CO2 from bicarbonate by acidification. This “CO2 solution” was immediately added to the reaction mixture, and decrease in absorbance was followed at 340 nm (εNADH = 6.3 mM−1 cm−1). Appropriate controls were used to demonstrate the specificity of the reaction, including a reaction without crotonyl-CoA (negative control), a reaction without CO2 source (crotonyl-CoA reduction reaction), a reaction with cell-free extract boiled at 100 °C for 5 min (negative control) and a reaction without NADH (negative control). The Ccr activity was normalised per total protein concentration.

ATP and NADPH/NADP quantification

Intracellular levels of ATP and NADPH/NADP were quantified using a bioluminescent enzymatic kit (ADP/ATP Ratio Assay Kit, Sigma-Aldrich, Dublin, Ireland) and NADPH/NADP quantification kit (Sigma-Aldrich, Dublin, Ireland) according to manufacturers’ instructions.

The analysis of the 13C-labelled PHB

The strain was grown in 50 mL of SMN, nitrogen limiting, supplemented with disodium malate (non-labelled) and sodium acetate-1-13C added at T24 (Sigma-Aldrich, Dublin, Ireland) for 72 h. The polymer was extracted from lyophilised cells using Soxhlet extraction method (Koller et al. 2005) and hydrolysed. Analysis was conducted using an Agilent 6890 gas chromatograph coupled to a 5973 mass selective detector. The methyl ester of the monomer dissolved in chloroform was injected into an HP5-MS column (30 m × 0.25 mm × 0.25 μm) and oven temperatures held at 60 °C for 4 min, then raised to 170 °C at 10 °C min−1 and held at that temperature for 3 min. The mass spectrometer was operated in selected ion monitoring (SIM) mode measuring ion currents m/z 103 (corresponding to a monomer molecular ion containing only 12C atoms), 104 (corresponding to a monomer molecular ion containing one 13C and three 12C atoms), 105 (corresponding to a monomer molecular ion containing two 13C and two 12C atoms) and 106 (corresponding to a monomer molecular ion containing three 13C and one 12C atoms). The mass spectrum of the unlabelled and labelled methyl ester, which had a retention time of 3.3 min, is shown in Fig. S1 indicating the key fragment ion m/z 103 and 105. Incorporation of isotope enrichment was determined after correcting for natural isotope abundances.

Results

Chemoheterotrophic growth of R. rubrum and PHB production

While acetate is a well-known precursor for PHB accumulation, R. rubrum cannot utilise acetate aerobically in the dark without the presence of CO2 or C4-dicarboxylic acids (Eisenberg 1953). C4-dicarboxylic acids found in the TCA cycle are good growth substrates for R. rubrum, but they are poor substrates for PHB accumulation (Brandl et al. 1989; Ormerod et al. 1961). Thus, we decided to use a combination of MA to allow aerobic growth and sufficient PHB accumulation in R. rubrum. We tested acetate addition at different time points (T0, T12, T24; data not shown). Since aerobic metabolism of acetate in the dark requires CO2 (provided by oxidative decarboxylation of malate to pyruvate and pyruvate to acetyl-CoA), addition of acetate at earlier time points (T0, T12) resulted in lower biomass and PHB yield (data not shown). This is probably due to the known inhibition of the activity of the enzymes of the pyruvate dehydrogenase complex (PDC) by acetyl-CoA (Lüderitz and Klemme 1977). We analysed biomass accumulation, substrate consumption, and PHB accumulation over time when acetate was added to malate grown cultures at T24 (Fig. 1). Acetate assimilation (0.015 gC L−1 h−1) represses malate metabolism, with malate consumption rate decrease from 0.036 to 0.004 gC L−1 h−1 observed in the period between T36 and T48 i.e. 12 h after the addition of acetate to cells growing on malate (Fig. 1b). In addition, malate consumption over a 72-h period was 0.009 gC L−1 h−1, while overall malate consumption in the control experiment without acetate addition was 0.017 gC L−1 h−1 (Online Resource, Fig. S3). The decrease in malate consumption corresponds to an increased acetate consumption rate and the onset of PHB accumulation (Fig. 1a, b). Besides the 5-fold increase in PHB associated with increased acetate consumption between T36 and T48, there was a 2.5-fold increase in PHB-free biomass (0.24 g L−1 at T36 and 0.6 g L−1 at T48).

Fig. 1
figure 1

Growth profile of R. rubrum S1 when malate and acetate were used as co-substrates. a Cell dry weight, CDW (filled circle), nitrogen (filled triangle) and PHB accumulation (grey bars) when 4 g L−1 (1.08 gC L−1) disodium malate and 3.7 g L−1 (1.08 gC L−1) sodium acetate were used. b Malate (dotted line, filled diamond) and acetate (full line, filled square) consumption over time. The error bars represent standard deviation between three biological replicates

Full size image

Acetate is a more reduced compound than malate, providing more reducing equivalents. The level of NADPH is an important regulator of PHB metabolism (Hauf et al. 2013); thus, we measured the presence of the oxidised and reduced forms of this molecule in R. rubrum S1. An hour after acetate addition, R. rubrum cells showed a 2.2-fold higher NADPH/NADP+ ratio in comparison with T24 (Online Resource, Fig. S2A). Acetate addition also coincided with a 1.7-fold increase in ATP levels, (Online Resource, Fig. S2B).

PHB promotes acetate assimilation into biomass

The GC-MS analysis of the PHB polymer extracted at the end of the growth period when 13C-labelled acetate (one carbon atom labelled, 13C-1-acetate) was used showed that 72 % of the PHB monomer units contained two labelled carbons (Table 1; Online Resource, Fig. S1), which means that the monomer unit was formed by condensation of two acetyl-CoA molecules originating from supplied acetate. Twenty-three percent of monomers contained one labelled carbon (Table 1; Online Resource, Fig. S1), suggesting that these monomers are a combination of acetate supplied at T24 (13C-1-acetate) and acetate formed during oxidative decarboxylation of malate to pyruvate followed by oxidative decarboxylation of pyruvate to acetyl-CoA. Thus, acetate accounts for 83.5 % of the monomers in PHB. However, only 15 % of the total acetate consumed was converted into PHB (0.03 gC L−1). While acetate is a direct precursor to PHB in R. rubrum, the conversion is inefficient. We have compared the ability of R. rubrum to grow on M and MA. Given that the residual biomass achieved by M- and MA-grown cells was similar and that MA cells showed 2-fold lower malate consumption compared to M-grown cells (Table 1), the remainder of the acetate supplied contributes to biomass generation.

Table 1 Comparison of growth and PHB accumulation in malate (M)- and malate-acetate (MA)-grown R. rubrum S1 and R. rubrum ΔphaC1ΔphaC2, a mutant incapable of PHB accumulation
Full size table

To determine if PHB accumulation affected acetate assimilation into biomass, a double knockout mutant R. rubrum ΔphaC1ΔphaC2 (Klask et al. 2015) incapable of PHB accumulation was investigated. The residual biomass achieved by the wild type and the knockout was the same, and the total carbon consumed was the same, but the acetate consumption was 1.9-fold lower in the knockout mutant (Table 1). The PHB-negative mutant consumes 0.38 gC L-1 of acetate (Table 1), while 0.7 gC L−1 of acetate is assimilated into residual (PHB-free) biomass in the wild type. Removing the ability of R. rubrum to accumulate PHB reduces its ability to assimilate acetate in biomass.

Proteomic analysis of R. rubrum S1

To further the investigation of the metabolism of acetate and PHB accumulation in R. rubrum, the proteome of R. rubrum was analysed using a “shotgun” approach (Online Resource, Table S1; Table 2). The genome of R. rubrum S1 was reported to encode 3850 proteins and 83 RNAs (Munk et al. 2011). Over 130,000 peptide tandem mass spectra were used to identify 11,495 unique peptides, in turn generating 1845 high-confidence (≥0.99 ProteinProphet score) protein identifications. This represents a substantial proportion (48 %) of the predicted R. rubrum proteome and reflects the extensive MS analysis. Moreover, the proteins exhibited a wide range of annotated biophysical (molecular mass, isoelectric point), biochemical (functional annotations) and structural (domains) properties, suggesting that the analysis was not biased in favour of, or against, any protein class. LFQ was used as a semi-quantitative measure of protein abundance, which allowed for monitoring the protein expression patterns in different samples (Cagney et al. 2005; Wang et al. 2003).

Table 2 Comparison of protein expression profiles in R. rubrum when PHB accumulation was stimulated by acetate addition
Full size table

The changes in the R. rubrum proteome associated with acetate addition and stimulation of PHB accumulation were analysed (Table 2; Online Resource, Table S1). The proteins with different expression patterns (578 proteins; Table S1) over the growth period were mapped to different metabolic pathways (Table 2).

PHB synthesis stimulated by acetate addition affects the expression and activity of RuBisCO

Surprisingly, RuBisCO was detected in the proteome of R. rubrum grown aerobically in the dark (Table 2). RuBisCO is an enzyme involved in CO2 fixation and to date is associated exclusively with anaerobic growth of R. rubrum S1 (Grammel et al. 2003). RuBisCO was detected at all time points analysed, with a 2.5-fold decrease observed between T30 and T72 (Table 2). The other enzymes of the Calvin cycle were also present, but there was no significant change in their relative expression across the cultivation period (Online Resource, Table S1). The expression of the RuBisCO large subunit did not show a statistically significant change over the growth period (Online Resource, Table S1). However, to determine whether the detected enzyme is active and whether there is a difference in the activity with regard to PHB accumulation, we measured the RuBisCO activity in cell-free extracts of R. rubrum. Cell-free extracts exhibited RuBisCO activity, which was highest at T24 and T25, and decreased over the cultivation period (Fig. 2). There was an 11-fold difference in RuBisCO activity between T24 and T72. PHB was first detected at T36 (about 3 % of CDW, Fig. 1a), which coincided with a 5-fold decrease in RuBisCO activity compared to T24 (Fig. 2). Malate-grown R. rubrum accumulated low levels of PHB (8 %, Online Resource, Fig. S3). These cells exhibited a 2.7-fold decrease in RuBisCO activity between T24 and T72 (Online Resource, Fig. S4).

Fig. 2
figure 2

RuBisCO activity was assayed by following the rate of NADH oxidation by cell-free extracts of R. rubrum (black bars) and R. rubrum ΔphaC1ΔphaC2 (grey bars) using ribulose 1,5-biphosphate as the substrate. The asterisks represent statistically significant difference between the wild-type and mutant strains and different time points. The error bars represent standard deviation between three biological replicates

Full size image

The RuBisCO activity was also detected in R. rubrum ΔphaC1ΔphaC2 (Fig. 2). The activity remained at the same level over the course of the cultivation period, suggesting that the decrease in RuBisCO activity in the wild-type strain was associated with PHB accumulation.

The expression of core proteins of the ethylmalonyl-CoA pathway is affected by acetate

Signature proteins of the EMC pathway, previously described as a new acetate assimilation strategy in a purple non-sulphur bacterium R. sphaeroides (Erb et al. 2009b), were identified in the proteome of R. rubrum grown aerobically in the dark (Table 2). The EMC pathway and PHB synthesis pathway share the first two steps i.e. condensation of two acetyl-CoA molecules into acetoacetyl-CoA which is subsequently reduced to (R)-3-hydroxybutyryl-CoA (Alber et al. 2006). (R)-3-hydroxybutyryl-CoA is further reduced to crotonyl-CoA in the EMC pathway or polymerised to PHB. Therefore, these two synthetic pathways compete for the same intermediate. The signature enzymes of the EMC pathway, Ccr, methyl-malonyl-CoA epimerase and methyl-malonyl-CoA mutase showed a 1.3- to 2.2-fold increase in relative expression levels in MA-grown cells between T24 and T30, suggesting that acetate was assimilated via the EMC pathway (Alber et al. 2006; Erb et al. 2009b). Indeed, cells grown on malate alone had up to 2-fold lower levels of EMC pathway core proteins at T30 compared to MA-grown cells (Online Resource, Table S1). The expression of the EMC pathway core proteins decreased between T30 and T72, after cells started to accumulate PHB (Table 2). An established enzymatic assay was used to confirm the presence of Ccr and to establish if accumulation of PHB affects its activity (Erb et al. 2009a). Erb and colleagues proposed that Ccr requires NADPH as a cofactor, crotonyl-CoA as a substrate and CO2 for the activity (Erb et al. 2007; Erb et al. 2009a). While no reaction occurred when NADPH was used as the cofactor, NADH oxidation by the cell-free extracts of R. rubrum was observed (Fig. 3). Appropriate controls showed that the cell extract did not oxidise NADH in the absence of crotonyl-CoA. The enzyme was able to catalyse the reaction in the absence of CO2. However, the reaction was about 3-fold higher in the presence of CO2 suggesting that the enzyme is more active at carboxylating crotonyl-CoA than simply reducing it. The Ccr activity increased 1.8-fold 1 h after acetate addition (Fig. 3), with the highest activity achieved at T36. PHB accumulation was detected at T36, and a 5-fold increase in PHB levels between T36 and T48 was accompanied by a 2-fold decrease in Ccr activity (Fig. 3). The Ccr activity was detected in R. rubrum grown with malate only (Online Resource, Fig. S5). However, MA-grown cells exhibited 1.8-fold higher activity compared to malate-grown cells at T30 and T36 (Online Resource, Fig. S5).

Fig. 3
figure 3

Comparison of Ccr activity exhibited by cell-free extracts of R. rubrum (black bars) and R. rubrum ΔphaC1ΔphaC2 (grey bars) grown aerobically in the dark with malate and acetate as cosubstrates. The Ccr activity is expressed as the rate of NADH oxidation. The asterisks represent statistically significant difference (p < 0.05) in Ccr activity between the wild-type and mutant strains and different time points. The error bars represent standard deviation between three biological replicates

Full size image

The peak of Ccr activity in R. rubrum ΔphaC1ΔphaC2 was detected at an earlier time point in comparison with that in the wild type (Fig. 3). It seems that when PHB accumulation is allowed, the common precursor (R)-3-hydroxybutyryl-CoA is first directed to PHB synthesis, which may stimulate the activity of the EMC pathway and acetate assimilation into biomass. At T48, the Ccr activity decreases in both the wild-type and the PHB-negative strains (Fig. 3). However, the activity is 2-fold higher in the wild type possibly allowing more acetate to be assimilated into biomass.

The effect of acetate addition on enzymes involved in central metabolic pathways and PHB metabolism

The enzymes of the PDC, pyruvate dehydrogenase (subunits α and β), dihydrolipoyl transacetylase and dihydrolipoyl dehydrogenase showed differences in their relative expression levels in R. rubrum after acetate was added to the growth medium (Table 2). The PDC is a large membrane-bound complex in R. rubrum and serves as the main gateway into the TCA cycle during aerobic metabolism (Berg et al. 2002; Lüderitz and Klemme 1977). The PDC oxidatively decarboxylates pyruvate to form the activated acetate i.e. acetyl-CoA, a high-energy intermediate that sits at the crossroads of central metabolism (Bentley 2000; Kennedy 2001). Six hours after acetate addition, a 1.2–1.5-fold decrease in relative expression levels of all PDC enzymes was observed, possibly as an effect of acetyl-CoA formation. According to the KEGG database (1), two enzymes can also be involved in acetate activation: acetyl-CoA synthase (Q2RNC6) and acetyl-CoA hydrolase (Q2RT18). While both enzymes were detected under both conditions, the hydrolase showed a 2-fold increase in relative expression levels 6 h after acetate addition (T30) and a 3.3-fold increase at T72 (Table 2).

Despite the fact that no PHB was detected at T24 and there was an 8-fold increase in PHB levels between T36 and T72 (Fig. 1a), the majority of proteins involved in PHB metabolism did not show a significant difference over the cultivation period (Online Resource, Table S1). However, the phasin Q2RP67, a protein covering the surface of PHB storage granules, had a 3.8-fold higher expression at T72 compared to T24 (Table 1). An increase in expression was observed for phasin Q2RQI1, with a 4.4-fold increase at T30 and a 21.4-fold increase in relative expression levels at T72 compared to T24. The expression of PhaR (Q2RXR4), another granule-bound protein, was observed at all time points tested and increased 2.3-fold between T30 and T72.

High expression levels of the synthase PhaC2 (Q2RRN2), which catalyses the polymerisation of (R)-3-hydroxybutyryl-CoA, were observed at T24 and T30, before any PHB was detected (Online Resource, Table S1; Fig. 1). Similar to the PhaC2 (Q2RRN2), the expression of PHB depolymerases Q2RNZ5 and Q2RU10 was detected at T24 and remained at the same level throughout the cultivation period (Online Resource, Table S1). Early expression (T24) of PHB metabolism-related proteins and simultaneous expression of the polymerase and depolymerases reflect a highly dynamic nature of PHB metabolism and its involvement in processes other than simple carbon and energy storage.

The differences in proteome are reflected in metabolite levels in R. rubrum S1

(R)-3-hydroxybutyrate, a monomer unit of PHB, was detected as an intracellular metabolite (Table 3). Since PHB synthesis and degradation occur simultaneously in bacterial cells, the detected (R)-3-hydroxybutyrate could be the precursor for PHB synthesis as well as the degradation product of PhaZ activity (Uchino et al. 2007). Succinate was amongst the most abundant metabolites detected in cell extracts of R. rubrum grown aerobically in the dark. Succinate levels first increased 3.1-fold, followed by a 1.7-fold decrease (Table 3). This is in keeping with the expression and activity of the EMC pathway signature enzyme (Fig. 3), with succinyl-CoA being one of the products of this pathway. The presence of intracellular acetate at T24, before acetate addition (Table 3), suggests malate is metabolised to acetate in R. rubrum. Acetate was detected in R. rubrum grown with malate as a sole source of carbon and energy as well (Online Resource, Table S2). A 2.7-fold increase of acetate levels at T36 is probably the combined effect of acetate addition to malate-grown cells at T24 and the ability of M-grown cells to produce acetate from malate (Table 3).

Table 3 Metabolite levels in R. rubrum when PHB accumulation was stimulated by acetate addition at T24 to malate grown cells
Full size table

Glutamate is a nitrogen donor in many biosynthetic reactions, and it was previously shown that this amino acid was the dominant amino acid in cell extracts of R. rubrum grown both under nitrogenase repressed and derepressed conditions (Kanemoto and Ludden 1987). Nitrogen limitation occurred at T48 of the cultivation period (Fig. 1). Unlike in the case reported by Kanemoto and Ludden, where the glutamate level decreased under nitrogen limitation, in our experiments intracellular glutamate increased at T72 compared to T36 (Table 3).

Discussion

Redox homeostasis is a fundamental requirement for sustained metabolism and growth in all biological systems. The intracellular redox potential is determined by the NADH/NAD+ and the NADPH/NADP+ ratio (Green and Paget 2004). Principally, high intracellular energy charge, characterised by an elevated pool of acetyl-CoA, NAD(P)H, ATP and others, promotes PHA formation (Haywood et al. 1988; Uchino et al. 2007). The levels of ATP and NADPH increased in R. rubrum after acetate was added (Online Resource, Fig. S2), which can explain the onset of PHB accumulation (Fig. 1). In addition, NAD(P)H and acetyl-CoA negatively regulate the activity of the enzymes of the PDC (Lüderitz and Klemme 1977). Indeed, after acetate addition to the malate-grown R. rubrum, the expression of PDC enzymes decreased (Table 2) which led to a decrease in malate consumption (Fig. 1b).

What is surprising is that acetate assimilation into biomass is stimulated by the presence of PHB since the PHB-negative mutant consumes 2-fold less acetate and that acetate assimilation into biomass is much more efficient than into PHB under PHB-accumulating conditions (Table 1). It is widely reported that PHB can function as a carbon and energy storage material (Grage et al. 2009; Steinbüchel and Lütke-Eversloh 2003; Verlinden et al. 2007), forms complexes with calcium and polyphosphate for DNA uptake (Madison and Huisman 1999) and is involved in stress response (Pham et al. 2004; Raberg et al. 2014). It was also suggested that PHA might be involved in maintenance of efficient metabolism (Escapa et al. 2012). The early and simultaneous expression of the proteins involved in the PHB metabolism (Table 2; Online Resource, Table S1) suggests that the PHB synthesis and degradation operate as a continuous cycle, potentially optimising the carbon flow in the organism. Furthermore, PHB is likely acting as an electron sink in R. rubrum as well as a carbon storage material. R. rubrum uses hydrogen as an electron sink and the Calvin cycle to distribute excess electrons in the cell (Gordon and McKinlay 2014). When grown aerobically in the dark, R. rubrum cannot produce hydrogen (Ormerod et al. 1961; Selao et al. 2011). Furthermore, RuBisCO has not been reported to be expressed or active in aerobically grown R. rubrum (Cook et al. 1988; Grammel et al. 2003). Thus, the organism when grown aerobically in the dark would lack two key means to metabolise excess electrons. However, we observed that the expression of active RuBisCO in R. rubrum grown aerobically and the presence of PHB coincide with the reduction of RuBisCO activity. The decarboxylation of malate to pyruvate and pyruvate to acetyl-CoA (Kleijn et al. 2010) could stimulate RuBisCO expression. Indeed, we observed the presence of pyruvate dehydrogenase and malate decarboxylase (malic enzyme) in the proteome of R. rubrum (Table 2; Online Resource, Table S1). Furthermore, acetate was detected as one of the metabolites in R. rubrum when only malate was supplied as a carbon and energy source (T24; Table 3), suggesting malate is metabolised to acetyl-CoA through oxidative decarboxylation.

The normal role of RuBisCO and the Calvin cycle would be to metabolise carbon dioxide and oxidise reduced cofactors such as NAD(P)H (Gordon and McKinlay 2014). When this is coupled to metabolic pathways that reduce these cofactors, such as oxidative decarboxylation of malate (Kleijn et al. 2010), the cell has a closed cycle of electron flow (Fig. 4). However, PHB synthesis also requires reduced cofactors such as NAD(P)H. Thus, it is logical that RuBisCO expression and activity decrease with PHB accumulation to prevent oxidation of the reduced cofactors by the Calvin cycle and allow their oxidation for PHB precursor synthesis (Figs. 1, 2, and 4 and Table 1). Under photoautotrophic conditions, the level of CO2 and the presence of reaction by-products regulate the level of RuBisCO expression and activity (Leustek et al. 1988; Pearce 2006). While end product repression by acetyl-CoA and related metabolites was observed in P. oxalaticus (Meijer and Dijkhuizen 1988), the addition of acetate did not repress RuBisCO activity in R. rubrum, as RuBisCO activity under aerobic, dark conditions in the PHB-negative mutant remains constant over time (Fig. 2). Thus, the decrease in expression and activity of RuBisCO is a consequence of stimulated PHB accumulation (Figs. 1 and 2). It is possible that PHB takes over the role of the main electron sink in R. rubrum, thereby maintaining the redox homeostasis.

Fig. 4
figure 4

The metabolism of malate and acetate in R. rubrum during aerobic growth in the dark. The main metabolic pathways with differential expression of enzymes involved and the intracellular metabolites that showed significant difference under tested growth conditions are depicted. Distinct pathways are represented by different colours. Intracellular metabolites with a significant change upon acetate addition are depicted in different colours to mark potential pathways where these metabolites could arise. The interchangeable forms of NAD(P)H suggest that reducing equivalents formed in a catabolic process could be recycled in an anabolic process, and vice versa. R. rubrum possesses an H+-translocating transhydrogenase (Table S1; Q2RSB2, Q2RSB4), which couples the transfer of hydride-ion equivalents between NAD(H) and NADP(H)

Full size image

R. rubrum lacks isocitrate lyase, hence the glyoxylate shunt is not operative (Kornberg and Lascelles 1960). Acetate is most likely assimilated through the EMC pathway (Alber et al. 2006), as we detected the expression of signature enzymes of this pathway (Table 1). The presence of acetate coincides with changes in PHB levels and both the activity and expression of the first enzyme of the core part of the EMC pathway namely Ccr (Fig. 3). The higher expression of the core proteins at T30 (Table 2) is likely a result of the presence of acetate. There was a difference in the peak of Ccr activity between the wild-type and the PHB-negative mutant (Fig. 3). When PHB accumulation was allowed, the Ccr activity was highest at T36, while the highest activity in the PHB-negative mutant occurred at T30. It is possible that under PHB-accumulating conditions, (R)-3-hydroxybutyric acid, the common intermediate for the PHB and EMC pathways, was firstly polymerised into PHB, as evidenced from the later increase in Ccr activity in the wild-type strain. However, the increase in Ccr activity was followed by a decrease in both the wild type and the mutant (Fig. 3). The decrease in expression levels of the core proteins of the EMC pathway at T72 is unlikely to be a response to higher levels of PHB as the mutant strain has a similar profile of Ccr activity (Fig. 4). A 2.3-fold increase in the common intermediate of the EMC pathway and PHB synthesis of (R)-3-hydroxybutyric acid was detected in R. rubrum between T36 and T72 (Table 3). This might be the consequence of the decreased Ccr activity. As there was no further increase in PHB level after 72 h of cultivation (data not shown) and PHB synthesis and degradation occur simultaneously in bacterial cells, the increase in (R)-3-hydroxybutyrate could also be the consequence of the PhaZ activity (Uchino et al. 2007).

Another finding supporting the active EMC pathway in R. rubrum is the presence of succinate as an intracellular metabolite (Table 3) (Erb et al. 2007). The difference in expression levels of enzymes belonging to the EMC pathway at T24 and T30 corresponded with the 3.1-fold difference in succinate between T24 and T36 (Tables 1 and 2). Succinate was not detected as a metabolite during aerobic growth of R. rubrum in the dark in defined mineral medium used by Rudolf and Grammel (2012), indicating increased flux through the EMC pathway.

Fig. 4 summarises results and shows the connection between MA metabolism, PHB metabolism, the EMC pathway and the Calvin cycle. During aerobic growth of R. rubrum with malate as a carbon and energy source, malate is oxidatively decarboxylated to pyruvate, and pyruvate is further decarboxylated to acetyl-CoA, generating CO2 and NAD(P)H which could fuel the Calvin cycle (Fig. 4). In addition, produced CO2 allows acetate assimilation (Eisenberg 1953), and assimilated acetate acts as a direct precursor for PHB synthesis (Fig. 4). When PHB accumulation is stimulated in R. rubrum by acetate addition, it further stimulates the conversion of acetate into biomass, thus driving carbon flow in a direction contrary to the normal perceived flow. Furthermore, PHB becomes the main electron sink in the cell, decreasing the role of the Calvin cycle but not affecting the EMC pathway (Gordon and McKinlay 2014; Laguna et al. 2011).