Introduction

Triacylglycerol (TAG) is a storage compound commonly found in eukaryotic organisms such as plants, animals, microalgae, fungi, and yeast (Leman 1997; Ratledge 1989). However, it is now known that some members of prokaryotes are also able to accumulate such a compound (Alvarez 2010; Alvarez and Steinbüchel 2010; Alvarez and Steinbüchel 2002). With a few exceptions, the ability to accumulate TAG is common in actinobacteria belonging to the Rhodococcus, Nocardia, Mycobacterium, and Streptomyces genera (Alvarez et al. 1997; 1996; Barksdale and Kim 1977; Daniel et al. 2004; Olukoshi and Packter 1994). In recent years, these lipids have attracted great interest as potential resources for biotechnological purposes (Wältermann et al. 2000; Li et al. 2008). The potential of bacterial TAG may be similar to that of vegetable sources, including their use as feed additives, cosmetics, oleochemicals, lubricants, and other manufactured products. In addition, bacterial oils have been proposed as a source for biofuel production (Alvarez 2010; Holder et al. 2011).

The biosynthesis and accumulation of TAG is a complex process that involves several catalytic enzymes that participate at different metabolic levels. In this context, Holder et al. (2011) reported at least 261 genes implicated in the Rhodococcus opacus PD630 TAG cycle based on metabolic reconstruction and gene family analysis. On the other hand, Chen et al. (2013) described at least 177 genes associated with lipid metabolism in the same strain based on transcriptome and lipid-droplet proteome analyses.

The main biosynthetic pathway for TAG biosynthesis in rhodococci, known as the Kennedy pathway (Kennedy 1961), involves the sequential esterification of glycerol-3-phosphate producing phosphatidic acid (PA) (Fig. 1). PA is a key molecule for the biosynthesis of membrane glycerophospholipids through the synthesis of the liponucleotide intermediate CDP-diacylglycerol (CDP-DAG) which is a precursor of phosphatidylinositol, phosphatidylglycerol, and phosphatidylserine in bacteria (Parsons and Rock 2013; Zhang and Rock 2008). PA can also be dephosphorylated by a phosphatidic acid phosphatase (PAP, EC 3.1.3.4) to yield diacylglycerol (DAG), which is the direct precursor for TAG synthesis through an additional step of acylation catalyzed by the wax ester/diacylglycerol acyltransferase (WS/DGAT) enzymes encoded by the atf genes in rhodococci (Fig. 1). Because PAP and WS/DGAT might catalyze the rate-limiting steps in the TAG formation in oleaginous actinobacteria, the identification of the genes encoding for both types of enzymes constitutes an important issue for a better understanding of the glycerolipid metabolism in such microorganisms. Rhodococci exhibit a high redundancy of genes encoding for WS/DGAT enzymes (Alvarez et al. 2008; Hernández et al. 2008, 2013; Holder et al. 2011; Villalba et al. 2013). Two atf genes were previously cloned and widely characterized in R. opacus PD630 (Alvarez et al. 2008; Hernández et al. 2013). This strain has been one of the best oleaginous bacterium studied to date regarding TAG metabolism and its biotechnological potential (Alvarez et al. 1996, 1997, 2000; Voss and Steinbüchel 2001; Holder et al. 2011; MacEachran et al. 2010; Wältermann et al. 2000; Hetzler and Steinbüchel 2013). Similar results have been reported in Rhodococcus jostii RHA1, an oleaginous strain able to accumulate high levels of TAG under culture conditions reported for PD630 strain (Hernández et al. 2008; Ding et al. 2012; Villalba and Alvarez 2014). According to these studies, it is clear that several proteins, including the WS/DGAT enzymes, as well as the cellular culture conditions are involved in the biosynthesis and accumulation of TAG in R. opacus PD630 and probably in other oleaginous Rhodococcus strains. In contrast, the role of PAP enzymes in bacterial TAG accumulation is barely known, and so far, no PAP has been identified in rhodococci. This enzyme has been well characterized in higher eukaryotic cells and microorganisms like yeasts, microalgae, and fungi (Kocsis and Weselake 1996). Two families of PAP enzymes, Mg2+-dependent (PAP type 1, PAP1) and Mg2+-independent (PAP type 2, PAP2), have been described in those organisms. The PAP1 family utilizes PA as a unique substrate and is localized in the soluble fraction of the cell (Carman and Han 2006). In contrast, the PAP2 enzymes, currently known as a family of lipid phosphate phosphatases (LPPs), can utilize a broad range of substrates such as PA, lysophosphatidic acid (LPA), sphingosine-1-phosphate, and diacylglycerol pyrophosphate (DGPP), among others, and are localized in the cells as integral membrane proteins. In eukaryotic cells, PAP enzymes have been associated with lipid metabolism and lipid signaling (Carman and Han 2006).

Fig. 1
figure 1

Kennedy pathway reactions for the biosynthesis and accumulation of TAG. Putative proteins for each reaction of the Kennedy pathway (left panel) and their locus numbers for selected Rhodococcus strains (right panel) are showed. GPATs glycerol-3-phosphate acyltransferases, AGPATs acyl-glycerol-3-phosphate acyltransferases, PAPs phosphatidic acid phosphatases, DGKs diacylglycerol kinases, DGATs diacylglycerol acyltransferases

Full size image

The first member of the PAP1 family of enzymes (Pah1) has been purified and characterized from membrane and cytosolic fractions in yeast cells (Carman 1997). Pah1 has been associated with DAG biosynthesis, since a pah1 mutant accumulated PA and produced reduced amounts of DAG and TAG (Han et al. 2006). Although the genes encoding for the family of PAP1 enzymes are highly conserved among eukaryotes, they are not in prokaryotes. In contrast, the PAP2 superfamily (pfam 01569) includes a high number of enzymes (including the LPP enzymes) occurring in eukaryotic and prokaryotic cells. In yeasts, for example, PAP2 enzymes DPP1 and LPP1 are integral membrane proteins with six transmembrane spanning regions and are localized in the vacuole and Golgi compartments of the cell, respectively (Toke et al. 1998a, b). Because prokaryotes do not possess any homologs of PAP1, the study of PAP2-like enzymes in lipid metabolism and, potentially, in lipid signaling becomes a crucial issue in these organisms. Only a few PAP2-like enzymes have been characterized in prokaryotes (Dillon et al. 1996; Carman 1997; Zhang et al. 2008; Comba et al. 2013). Interestingly, Nakamura et al. (2007) identified and characterized five plastidic enzymes (LPPs β, γ, δ, ε1, and ε2) with PAP activity in Arabidopsis thaliana, all of which were evolutionarily associated with a LPP of the cyanobacterium Synechocystis sp. PCC6803 (synLPP) forming, thereby, a subfamily of PAP2 enzymes with “prokaryotic” origin.

Within the members of the PAP2 superfamily, the enzyme PgpB of Escherichia coli was the only enzyme known to display PAP activity (Dillon et al. 1996). Enzymatic activities in vitro have demonstrated that PgpB has a broad substrate spectrum being able to use phosphatidylglycerol phosphate (PGP), PA, LPA, DGPP, and undecaprenyl pyrophosphate (C55-PP) as substrates (Dillon et al. 1996 and Touze et al. 2008). Recently, Rucker et al. (2013) co-expressed the native PgpB enzyme and a WS/DGAT enzyme of Acinetobacter baylyi to produce TAG in E. coli.

On the other hand, Comba et al. (2013) reported two PAP2-like enzymes named Lppα and Lppβ in the actinobacterium Streptomyces coelicolor. Both enzymes showed a significant PAP activity when expressed in E. coli. In addition, overexpression of these enzymes in S. coelicolor resulted in an enhanced proportion of TAG in total intracellular lipids.

The identification and characterization of PAP2 enzymes in rhodococci are essential for enhancing our understanding in lipid metabolism in these oleaginous bacteria. Moreover, the manipulation of such genes may provide an interesting strategy for optimizing the production of bacterial oils for the biofuel industry. In this work, we performed a thorough bioinformatic analysis to identify putative PAP2 enzyme(s) in rhodococcal strains. Based on this information, we identified and characterized the ro00075 gene of R. jostii RHA1 which encodes for a putative PAP2 protein (GenPept accession no. YP_700069). Additionally, we explored the effect of ro00075 overexpression on TAG accumulation in both oleaginous strains (RHA1 and PD630) and in non-oleaginous Rhodococcus fascians F7.

Materials and methods

Strains, culture conditions, and plasmids

The strains and plasmids used in this work are listed in Table 1. E. coli strains were grown on solid or in liquid Luria-Bertani (LB) medium at 37 °C. Rhodococcus strains were cultivated aerobically at 28 °C in LB medium or minimal salt medium (MSM) according to Schlegel et al. (1961). Glucose was used in MSM as the sole carbon source at a final concentration of 1 % (w/v). For nitrogen-limiting conditions, to allow lipid accumulation (storage conditions), the concentration of ammonium chloride was reduced to 0.1 g L−1 (MSM0.1). In MSM0 culture medium, the addition of ammonium chloride was omitted. Cells were harvested at specific time-points, washed with sterile NaCl solution (0.85 %, w/v), and dried at 37 °C for chemicals analyses. When LBS medium was used, 10 % (w/v) of sucrose was added to LB medium. Antibiotics were used at the following final concentrations: 100 μg/mL ampicillin (Ap), 50 μg/mL kanamycin (Km), 30 μg/mL nalidixic acid (Na), 5 μg/mL gentamycin (Gm), and 34 μg/mL chloramphenicol (Cm) in both E. coli and Rhodococcus strains. For overexpression analysis of genes under the acetamidase promoter (Pace) of pJAM2 and the thiostrepton promoter (PtipA) of pTip-QC2, 0.5 % (w/v) of acetamide and 1–3 μg/mL of thiostrepton were respectively added to cell cultures at time zero.

Table 1 Strains and plasmids used in this study
Full size table

DNA analysis, amplification, cloning, and sequencing

Chromosomal DNA, plasmids, and DNA fragments were isolated and analyzed by standard methods. For specific DNA amplification, the PCR assay was performed with different specific primers listed in Table 2. The general thermocycler parameters used were as follows: 5 min at 94 °C, 30 cycles of 1 min at 94 °C, 30 s at 60 °C, 1 min at 72 °C, and finally 5 min at 72 °C. The PCR products were cloned into pGEM-T-easy vector and subjected to DNA sequencing.

Table 2 Oligonucleotides used as PCR primers
Full size table

Strategy to delete the ro00075 gene in RHA1 strain

Mutagenic plasmid pK18mobsacB-Δro00075 RHA1 was constructed to delete ro00075, as follows. An upstream region of ro00075 was amplified by PCR using the MAH-F/ro00075a set of primers containing a BamHI and XbaI restriction site, respectively (Table 2). The resulting 700-bp amplicon was cloned into pGEM-T-easy vector (pGEM-T-easy/ro00075-up), digested with BamHI/XbaI, and subcloned into BamHI/XbaI-digested pK18mobsacB, resulting in pK18mobsacB-ro00075up. An 811-bp amplicon of the ro00075 downstream flanking region including the ro00075 stop codon was obtained using the ro00075b/MAH-R set of primers containing a XbaI and HindIII restriction site, respectively (Table 2). This amplicon was cloned into pGEM-T-easy vector (pGEM-T-easy/ro00075-down), digested with XbaI/HindIII, and subcloned into XbaI/HindIII linearized pK18mobsacB-ro00075up yielding in pK18mobsacB-Δro00075 RHA1 (Table 1). Then, the pK18mobsacB-Δro00075 RHA1 plasmid was transferred via conjugation into R. jostii RHA1 cells using the E. coli S17-1 as the donor. The first event of recombination was selected on LB plates supplemented with kanamycin and nalidixic acid. For the second recombination event, individual colonies were cultivated in LB during 48 h without antibiotics and then several aliquots were plated on LBS plates. Alternatively, pK18mobsacB-Δro00075 RHA1 plasmid was transferred into RHA1 pTip-QC2/RO00075 strain. The first event of recombination was selected on LB plates supplemented with kanamycin, nalidixic acid, and chloramphenicol. For the second recombination event, individual colonies were cultivated during 48 h in LB in the presence of an inducer (thiostrepton) and then several aliquots were plated on LBS plates in the presence of the inducer. The genotype of the resulting mutant strain (RHA1 Δro00075-pTip-QC2/RO00075) was verified by PCR from chromosomal DNA with primers ro00075F/ro00075R (200 bp), MAH-F/ro00075R (800 bp), ro00075F/MAH-R (~900 bp), and MAH-F/MAH-R (1,500 bp) (see Table 2 and Fig. S1).

Cloning of the ro00075 gene

The ro00075 gene was amplified from total genomic DNA of R. jostii RHA1 by PCR using the primers RO00075F and RO00075R (Table 2). The resulting PCR product was cloned in pGEMT-easy vector, replicated in E. coli JM109, and verified by DNA sequencing. To achieve overexpression of ro00075 in Rhodococcus strains, a BamHI/XbaI digest from pGEMT-easy/ro00075 was subcloned into the BamHI/XbaI site of the shuttle E. coli-Mycobacterium-Rhodococcus vector pJAM2, which contain an inducible acetamidase promoter (Pace), and six His codons downstream to the XbaI site, yielding pJAM2/ro00075.

Alternatively, a BamHI/HindIII fragment from pJAM2/ro00075 was subcloned into the BamHI/HindIII site of the expression pTip-QC2 vector yielding pTip-QC2/ro00075.

In order to heterologously express ro00075 in E. coli C41 (DE3), the BamHI/HindIII fragment from the pJAM2/ro00075 containing the ro00075 gene with the C-terminal His6-tag fusion was purified and subcloned into the BamHI/HindIII sites of pET23a expression vector (Novagen), yielding the plasmid pET23/ro00075. Both pET23 (control) and pET23/ro00075 plasmids were maintained in E. coli DH5α and transferred to E. coli C41 (DE3) to perform the membrane protein expression analysis. All the plasmids described in this section are listed in Table 1.

DNA transfer and genotype screening in Rhodococcus cells

All replicative plasmids were transferred to R. jostii RHA1, R. opacus PD630, and R. fascians F7 by electroporation. Electroporation assays were carried out as described by Kalscheuer et al. (1999) using a Model 2510 electroporator (Eppendorf-Netheler-Hinz, Hamburg, Germany). The electrotransformants containing the ro00075 gene under Pace-promoter and PtipA-promoter were checked using primers aceF/RO00075R and thioF/RO00075R, respectively (Table 2).

In F7 strain, the presence of the couples pTip-QC2/ro00075-pJAM2/atf1 or pTip-QC2/ro00075-pPR27ace/atf2 was checked by colony PCR using the primers thioF/ro00075R to detect the ro00075 gene and the primers aceF/Atf1MHR or aceF/Atf2MHR to detect the atf1 and atf2 genes, respectively (Table 2).

Membrane preparation of E. coli and PAP activity assay

E. coli C41 (DE3) strains harboring plasmids pET23 and pET23/ro00075 were grown in LB at 37 °C until OD600nm 0.6. Gene expression was induced by addition of 0.1 mM isopropyl-beta-d-thiogalactopyranoside (IPTG) followed by overnight incubation at 23 °C and 120 rpm. Cells were harvested by centrifugation at 4,000×g for 20 min at 4 °C, washed twice with buffer A (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 1 mM EDTA, 10 mM β-mercaptoethanol), and resuspended in the same buffer. The next steps were all done at 4 °C. Cell disruption was carried out by sonication (Vibra-Cell™, Sonics & Material, Inc.) in the presence of 1 mM phenylmethylsulfonyl fluoride (PMSF). The lysate was cleared by centrifugation at 15,000×g for 30 min to separate cell debris, and the supernatant was ultracentrifuged at 120,000×g for 2 h to pellet the membrane fraction. The resulting pellet was washed twice with buffer B (20 mM Tris-HCl pH 7.5, 10 mM β-mercaptoethanol, 0.5 mM PMSF) and resuspended in the same buffer. Protein concentration was quantified by the Lowry assay using BSA as the standard (Lowry et al. 1951).

To test the phosphatase activity of the putative PAP, phosphatidic acid was used as the enzyme substrate. The DAG generated in the reaction was measured by LC-MS/MS. Standard phosphatase assays were performed in a 100-μL reaction mixture containing 25 mM Tris-HCl, pH 7.5, 2.5 mM Triton X-100, and 0.25 mM dipalmitoylphosphatidic acid (DPPA; Avanti Polar Lipids, Alabama, USA) as the substrate. Aliquots of membrane fractions of the corresponding strains were added to initiate the reaction, and after incubation at 30 °C, the reactions were quenched by adding methanol/chloroform (2:1). Subsequent lipid extraction was performed by the addition of chloroform and distilled water. The organic phase was dried and solubilized in 50 μL of mobile phase (water/methanol), and 5-μL aliquots were injected for HPLC and LC-MS/MS analysis. The organic extracts were separated on a ZORBAX Eclipse XDB-C8 column (3.0 × 50 mm, particle size = 1.8 μm; Agilent, USA) using a binary solvent system of water (solvent A) and methanol (solvent B). A linear gradient from 80 % B to 100 % B was applied between 0 and 25 min. Both solvents were supplemented with 5 mM ammonium acetate. The outlet of the liquid chromatograph was connected to a micrOTOF mass spectrometer (Bruker Daltonics, Bremen, Germany) operating in the positive-ion mode. The data was acquired online in the mass range m/z 35–1,000. Dipalmitoylglycerol (DPG) was detected as the transition [M+NH4]+ → [M-R-OH]+ ion (m/z 586.5 → m/z 313.3). A calibration curve was done using DPG as the standard (Avanti Polar Lipids, Alabama, USA), under the same conditions as the phosphatase reaction. DAG concentration in the samples was calculated by the linear regression equation obtained from the calibration curve. A unit of enzymatic activity was defined as the amount of enzyme that catalyzed the formation of 1 nmol of product/min. Specific activity was defined as units per milligram of protein. PAP activity was linear-dependent to time and protein concentration within the range tested.

Bioinformatical analyses

To analyze ro00075 and its homologs, we used the available Rhodococcus genomes in the NCBI database (http://www.ncbi.nlm.nih.gov/) for R. jostii RHA1, R. opacus PD630, R. opacus B4, R. erythropolis PR4, and Rhodococcus equi 103S or in the Rapid Annotation Subsystem Technology (RAST) server (http://rast.nmpdr.org/) for R. fascians F7. Protein screening and alignments were carried out using BLAST 2.2.17 (Altschul et al. 1997) and ClustalW (Thompson et al. 1994, Bioedit software) algorithms. Reference protein sequences were retrieved from the NCBI database. Identities were determined from alignments of full-length sequences. Transmembrane domains were detected using the TMHMM server available at http://www.cbs.dtu.dk/services/TMHMM/. Predicted secondary structure and hydrophobicity profiles were obtained with Protean (from DNASTAR Lasergene) software. Phylogenetic analyses were carried out from curated sequences using the neighbor joining method with the program MEGA 5.1.

Lipid analysis

E. coli strains harboring pET23a and pET23/ro00075 were grown in LB media at 37 ° C until OD600nm 0.6. Then, protein expression was induced by addition of 0.1 mM IPTG and 3 μCi [14C]acetate was also added at the same time and the culture was kept 12 h at 23 °C. Total lipids of E. coli strains were extracted as described by Bligh and Dyer (1959) directly from 14C-labeled cells. The lipid extracts were dried and analyzed by thin-layer chromatography (TLC) on silica gel 60 F254 plates (0 ± 2 mm, Merck), using the solvent systems hexane/diethyl ether/acetic acid (70:30:1, v/v/v) (Rotering and Raetz 1983). After TLC, spots were visualized by autoradiography, which was carried out by exposure for 2 days at −80 °C in Carestream® Kodak® BioMax® MR films. Spots were quantified using the ImageJ software version 1.47v (http://rsbweb.nih.gov/ij/). Unlabeled tripalmitin (Sigma), oleic acid (Sigma), glycero-3-phosphoethanolamine (Sigma), and dipalmitin (Sigma) were used as reference substances for TAG, free fatty acid (FFA), phospholipid (PL), and DAG, respectively and visualized by Cu-phosphoric staining.

Semiquantitative analyses of total intracellular lipids in Rhodococcus cells were carried out by thin-layer chromatography (TLC). For this, 5 mg of dried cells was extracted with 300 μL chloroform/methanol (2:1, v/v) for 90–120 min. For neutral lipids analysis, 15 μL (strains RHA1 and PD630) or 30 μL (strain F7) of chloroformic phase was subjected to TLC on silica gel 60 F254 plates (Merck) using hexane/diethyl ether/acetic acid (80:20:1, v/v/v) as the solvent (Wältermann et al. 2000). Spots samples were compared with a mixture containing tripalmitin (Merck), dipalmitin (Sigma), dl-α-palmitin (Sigma), palmitic acid (Merck), and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (Sigma) used as TAG, DAG, MAG, FFA, and PL references substances, respectively. All lipid fractions were visualized using iodine vapor, and spots of samples were quantified using the ImageJ software.

A colorimetric method described in previous works (Duncombe 1963; Wawrik and Harriman 2010; Hernández et al. 2013) was performed for the quantitative determination of total fatty acids in Rhodococcus cells. For this, dry cells (5–10 mg, depending on the used strain) were hydrolyzed with alkaline reagent (1 N 25 % methanol in NaOH) at 95–100 °C for 3 h with vigorous agitation (each 30 min). The soaps of fatty acids were neutralized with concentrated acetic acid, and the resultant free fatty acids were treated with copper reagent, extracted with chloroform, and developed with revealing reagent (diethyldithiocarbamate in 2-propanol). The resultant colored samples were spectrophotometrically read at 440 nm. The standard curve was performed with oleic acid as representative fatty acid of TAG in rhodococcal cells.

Results

Occurrence of putative PAP2 proteins in Rhodococcus genus and RO00075 sequence analysis

Bioinformatic analysis of the R. jostii RHA1 genome showed the occurrence of several genes coding for proteins belonging to the PAP2 superfamily, but only ro00075 was originally annotated as a gene encoding a putative PAP enzyme. The predicted RO00075 protein possesses three conserved domains reported for the PAP2 superfamily (cl00474, pfam_01569). These domains, which comprise the consensus sequences KxxxxxxRP (domain 1), PSGH (domain 2), and SRxxxxxHxxxD (domain 3), are shared by a superfamily of lipid phosphatases that do not require Mg2+ ions for activity (Stukey and Carman 1997; Carman and Han 2006). An additional analysis of the primary sequence of RO00075 using the tool CD-Search, an interface from NCBI to search the conserved domains with protein or nucleotide query sequences, showed a high association between the RO00075 sequence and a specific region known as PAP2_like_2 single-domain (cd03392), which usually occurs in bacterial membrane lipid phosphatases. This specific region includes the amino acid residues reported as part of the active site in the domains described above (K and R in domain 1; SGH in domain 2; and R, H, and D in domain 3), and they are also present in RO00075 (Table 3). In order to identify additional putative PAP enzymes in rhodococci, several Rhodococcus genomes were analyzed based on the presence of this specific region and by homology analysis using the BLAST algorithm and the RO00075 sequence as query. The analysis revealed the occurrence of three additional hypothetical membrane proteins (RO03136, RO0497, and RO08553) in R. jostii RHA1 and a variable number of putative lipid phosphatases in the other Rhodococcus strains (Table 3). The highest redundancy was observed in R. jostii RHA1 (four putative PAPs), whereas only one putative PAP was found in R. equi 103S and R. fascians F7. In all cases, redundancy of PAP enzymes was lower in comparison with that of other enzymes of the Kennedy pathway such as the DGATs and AGPATs (Fig. 1).

Table 3 Putative PAP2 proteins containing the PAP2_like_2 specific hit (cd03392) and their three catalytic domains in several Rhodococcus species
Full size table

The sequence alignment between RO00075 of R. jostii RHA1 and all PAP2-like proteins found in the different rhodococci used in this study showed the highest identity (I, 86 %) with the protein OPAG_07226 of R. opacus PD630, an oleaginous bacterium able to accumulate significant amounts of TAG (up to 60 % by cellular dry weight) with carbon sources such as gluconate, glucose, and fructose (Alvarez et al. 1996). On the other hand, the lowest identity (I, 29 %) was observed with the protein F7_3389 of R. fascians F7, a non-oleaginous strain, which accumulates low amounts of TAG after growth with those carbon sources. Figure 2a shows the full alignment of RO00075, OPAG_07226, and F7_3389. Both RO00075 and OPAG_07226 proteins exhibited not only a similar primary structure but also a similar secondary structure with a predicted molecular weight of 24.0 kDa. Further, both proteins were predicted to be integral membrane proteins containing a six-transmembrane topology structure according to the TMHMM server (data not shown) and their hydrophobicity profiles (Fig. 2b). On the other hand, the protein found in F7 strain showed a predicted five-transmembrane topology and significant differences in its primary and secondary structures compared to RO00075 (Fig. 2). Moreover, the first amino acid residues forming the first transmembrane domain in RO00075 were absent in F7_3389 (Fig. 2). The alignment of RO00075 with the second protein identified in PD630 strain (OPAG_02574) showed a low identity (I, 23 %) between them, confirming that the OPAG_07226 is the corresponding ortholog of RO00075. Interestingly, the two proteins found in R. opacus B4 showed low identities (48 % with ROP_ pROB01-01880 and 41 % with ROP_50370) when comparing with RO00075. Similar identities were observed for proteins found in R. erythropolis PR4 (48 % for RER_10170 and 43 % for RER_46290). Finally, the second lowest identity was obtained with a PAP2-like protein found in the non-oleaginous strain R. equi 103S (REQ_09750; I, 34 %). These results show a high genetic variability of rhodococcal PAP2-like proteins, suggesting that the occurrence of these different enzymes may be not only species-dependent but also strain-dependent in rhodococci.

Fig. 2
figure 2

Analysis of RO00075 sequence. a Sequence alignment between RO00075 of R. jostii RHA1 and its homologs OPAG_ 07226 of R. opacus PD630 and F7_3389 of R. fascians F7. Identical amino acids (black), homologous residues (gray), putative signal peptide, and putative conserved PAP2-like domains (underlined) are shown. b Predicted secondary structure and hydrophobicity profiles of RO00075, OPAG_ 07226, and F7_3389. To obtain the hydrophobicity plots, a window size of 19 was used

Full size image

Phylogenetic analysis using curated sequences of LPPs from different organisms showed that RO00075 as well as the other rhodococcal proteins were evolutionary associated with the protein synLPP of the cyanobacteria Synechocystis sp. PCC6803 forming a group of “prokaryotic” LPPs. This group was closely related to the clade of “plastidic LPPs with prokaryote origin” of A. thaliana reported by Nakamura et al. (2007) but not with eukaryotic type LPPs (data not shown). Further, phylogenetic analysis using only proteins found in selected Rhodococcus strains allowed us to separate all rhodococcal PAP2-like protein sequences into two groups: the first one (small PAPs) includes proteins with short (177–185 aa) amino acid length and the second group (medium/large PAPs) includes proteins with medium (228 aa) and long (273–376 aa) amino acid length (Fig. 3). While the F7_3389 protein of R. fascians F7 forms a subgroup within the small proteins of rhodococci, RO00075 and its ortholog OPAG_07226 form a subgroup within the medium PAPs (Fig. 3). Interestingly, we observed that the occurrence of different kinds of PAP enzymes in the studied strains coincided with their genome sizes and their abilities to accumulate TAG. Rhodococci with smaller genomes such as R. fascians (5.3 Mb) and R. equi (5.0 Mb) contain only PAP2-like proteins with short amino acid length (177–185 aa), whereas rhodococcal strains with larger genomes, such as R. erythropolis (6.9 Mb), R. opacus B4 (8.83 Mb), R. opacus PD630 (9.15 Mb), and R. jostii RHA1 (9.7 Mb) contain, additionally, larger PAP2-like proteins. Further, the oleaginous rhodococcal strains possess more than one PAP2 enzyme distributed in the two groups described above (small PAPs and medium/large PAPs), while strains described as non-oleaginous bacteria only contain enzymes in the group of small PAP2 proteins. Besides, the medium PAPs were present only in Rhodococcus strains able to accumulate the highest TAG levels, such as R. jostii RHA1 and R. opacus PD630; in fact, this last strain possesses only a medium PAP2 protein within the group of medium/large PAPs (Fig. 3). The transmembrane domain analysis using all PAPs of Fig. 3 revealed that small, medium, and large PAPs contain a five-, six-, and seven-transmembrane topology, respectively (data not shown).

Fig. 3
figure 3

Phylogenetic tree of putative PAP2 enzymes from different Rhodococcus species. A neighbor-joining algorithm was used (software MEGA 5.1). Bootstrap values are shown along the branches

Full size image

Finally, when we extended alignment analyses to other actinobacteria such as Mycobacterium, Streptomyces, and Nocardia, RO00075 always grouped with proteins containing a PAP2_like_2 single-domain (cd03392) in those bacteria but with identities less than 40 %. Further, most of those proteins were represented by large or small PAPs (data not shown).

Heterologous expression of ro00075 in E. coli

In order to characterize and to assign a functional role of RO00075 as a PAP2 enzyme, we cloned the ro00075 gene into the pET23a expression vector and transferred it into E. coli C41 (DE3), an optimized strain for overproducing membrane proteins (Wagner et al. 2008). The strain producing a recombinant version of the membrane-associated lipid phosphatase (RO00075-his6) was analyzed for DAG production. Recombinant cells were grown to mid-log phase and then cultivated for 16 h at 23 °C after induction with IPTG and the addition of [14C]acetate. Then, we analyzed the total lipid profile of this recombinant strain by metabolic labeling of [14C]acetate into the different lipid fractions. As shown in Fig. 4, heterologous expression of ro00075 in E. coli C41 (DE3) resulted in a twofold increase in DAG levels in comparison with control cells, whereas the free fatty acid and phospholipids fractions showed no changes. On the other hand, the specific Mg2+-independent PAP activity from purified E. coli membrane homogenate of C41 (DE) strain expressing ro00075 was measured. As is shown in Table 4, the membrane fraction isolated from the strain expressing ro00075 exhibited a fourfold increase in PAP activity.

Fig. 4
figure 4

Heterologous expression of ro00075 in E. coli and analysis of its lipid profile. Total lipid extract from [14C]acetic acid-labeled culture of E. coli C41 (DE3) pET23/ro00075 was analyzed on a silica gel TLC plate and developed in hexane/diethyl ether/acetic acid (70:30:1, v/v/v). E. coli C41 (DE3) containing pET23 empty was used as a control strain. Both control and sample were induced with IPTG. Unlabeled diacylglycerol (DAG), triacylglycerol (TAG), phospholipid (PL), and free fatty acid (FFA) were used as reference substances

Full size image
Table 4 PAP activity in membrane of E. coli expressing ro00075
Full size table

In vivo role of RO00075 in oleaginous R. jostii RHA1 strain

The in vivo role of RO00075 in TAG metabolism was studied by means of a mutagenesis strategy. To delete ro00075, an unmarked single-gene deletion strategy was performed in R. jostii RHA1 using the sacB counterselection system (see “Materials and methods”). A large number of transconjugant colonies were observed on LB plates after the first homologous recombination event of the suicide plasmid pK18mobSacB/Δro00075 RHA1. However, several attempts to delete the ro00075 gene failed after the second recombination event on LBS plates, with the wild type genotype being restored in all cases. Only after transferring of the suicide plasmid pK18mobSacB-Δro00075 RHA1 in a RHA1 strain containing an extrachromosomal copy of ro00075 in the pTip-QC2 vector, a mutant version was obtained after the second recombination event on LBS plates supplemented with the inducer (thiostrepton). The genotype of this strain was confirmed by PCR analysis from chromosomal DNA by means of several primers pairs (Table 2, Fig. S1).

Interestingly, the mutant strain was able to grow in liquid culture media with high nitrogen levels such as LB or MSM1 (NH4 +, 1 g L−1) without the inducer and only a slight delay on cell growth was observed (data not shown). In contrast, a significant decrease in cell growth was observed after cultivation in nitrogen-limiting MSM0.1 medium (NH4 +, 0.1 g L−1) in the absence of the inducer (Fig. S2). This growth was partially recovered after the addition of the inducer to the culture medium (Fig. S2).

On the other hand, in order to analyze the effect of chromosomal deletion of ro00075 on TAG accumulation, both mutant and control strain were grown in LB medium in the presence of the inducer to generate cell biomass, and then cells were resuspended in a mineral medium free of nitrogen (MSM0) with and without the addition of the inducer. As is shown in Fig. 5a, a decrease in TAG and DAG contents was observed in the mutant strain without the inducer. When the inducer was added to the cell culture, TAG as well as DAG contents were restored at similar levels to that of the control strain (Fig. 5a). These results were in agreement with quantitative analyses; whereas the mutant strain showed lower amounts of total fatty acids (up to 20 % less by cellular dry weight) in the absence of the inducer, a similar lipid content was observed in both control and mutant strain when the inducer was added to this last (Fig. 5b).

Fig. 5
figure 5

Lipid analysis of R. jostii Δro00075 mutant strain. Cells were grown overnight in LB medium with thiostrepton as the inducer, harvested, washed twice, and then incubated for 48 h in nitrogen-free MSM with glucose as the carbon source. a TLC analysis and b total fatty acid content of R. jostii RHA1 pTip-QC2/RO00075, RHA1 Δro00075-pTip-QC2/RO00075 without inducer, and RHA1 Δro00075-pTip-QC2/RO00075 with inducer (I). In the TLC analysis, a mixture of reference lipids was used as control (TAG triacylglycerol, FFA free fatty acid, DAG diacylglycerol, MAG monoacylglycerol, PL phospholipid). Bars represent the mean ± S.D. of triplicate determinations

Full size image

Overexpression of the ro00075 gene in oleaginous Rhodococcus strains and its effect on TAG accumulation

To further evaluate the in vivo role of RO00075 in TAG metabolism of oleaginous Rhodococcus cells, ro00075 was overexpressed in R. jostii RHA1 as well as in the taxonomically related strain R. opacus PD630 using both pJAM2 and pTip-QC2 vectors as expression systems. pJAM2 and pTipQC2 and its derivatives containing the ro00075 gene were transferred into Rhodococcus cells by electroporation, and the resultant recombinant strains (Table 1) were evaluated for TAG production by TLC analysis and total fatty acid quantification from dried cells. Expression of ro00075 using the pJAM2 expression vector promoted an increase of total fatty acid content in both R. jostii RHA1 and R. opacus PD630 cells during cultivation in MSM0.1 or MSM0 with glucose as the sole carbon source (Fig. 6a). Similar results were obtained when ro00075 was expressed using the pTipQC2 expression vector under the PtipA promoter, although the absolute values of total fatty acid content were slightly higher than those obtained using pJAM2 vector with the ace promoter (Fig. 6b). However, the relative differences in total fatty acid content between controls cells and ro00075-overexpressing cells were similar after 48 and 120 h for both expression systems with an average increase of 10–15 % by cellular dry weight (Fig. 6a, b). Since the expression level of ro00075 was comparable in both systems (pJAM2 and pTipQC2) as indicated by western blot analysis (data not shown), the slightly lower increase of total fatty acid content in pJAM2-harboring cells is attributed to the increase of the nitrogen level by the addition of acetamide into the culture media with the concomitant reduction of the total lipid content. In addition, TLC analysis revealed that the increase in the total fatty acid content in recombinant strains expressing ro00075 was exclusively associated with the increase of the TAG fraction in both RHA1 and PD630 strains (Fig. 7a, b).

Fig. 6
figure 6

Total fatty acid content in overexpressing ro00075-recombinant strains. Cells were grown in LB medium for 24 h, harvested, washed, and then incubated for 48 and 120 h under storage lipid conditions (0.1 g L−1 NH4 + MSM and nitrogen-free MSM) with glucose (RHA1 and PD630 strains) or fructose (F7 strain) as the carbon source. a pJAM2-derivative strains, b pTipQC2-derivative strains, and c F7-derivative strains co-expressing both ro00075 and atf genes

Full size image
Fig. 7
figure 7

TLC-lipid analysis of whole-cell extracts of overexpressing ro00075-recombinant strains. Cells were grown in LB medium for 24 h, harvested, washed, and then incubated for 120 h under storage lipid conditions in MSM0 with glucose (RHA1 and PD630) or fructose (F7) as the carbon source. a Overexpression of ro00075 in strain RHA1, b expression of ro00075 in PD630, c expression of ro00075, and co-expression of ro00075/atf2 in F7 strain. The arrows indicate the increased lipid fractions in samples. TAG triacylglycerol, FFA free fatty acid, DAG diacylglycerol, MAG monoacylglycerol, PL phospholipid

Full size image

Effect of ro00075 expression on TAG accumulation in non-oleaginous R. fascians F7

R. fascians F7 as well as other strains of this species is able to accumulate only low amounts of TAG after cultivation under nitrogen-limiting conditions (Alvarez et al. 1997, 2013; Alvarez 2003). For this reason, in this study, we consider the strain F7 as a non-oleaginous microorganism. Recombinant plasmid derivatives of pJAM2 and pTip-QC2 containing ro00075 were introduced into R. fascians F7 cells in order to analyze the expression of this gene on TAG biosynthesis in a non-oleaginous bacterium. While F7 pJAM2/RO00075 strain did not present a significant difference in comparison with control strain (Fig. 6a), the strain F7 pTip-QC2/RO00075 showed an increase up to 7 % by cellular dry weight in total fatty acid content (Fig. 6b) under lipid accumulation conditions and fructose (cells exhibited low growth with glucose, but an optimum growth with fructose) as the sole carbon source. Interestingly, unlike what was observed in RHA1 and PD630 strains, the increase of total fatty acid content in F7 pTip-QC2/ro00075 strain (Fig. 7c) occurred at the expense of several lipid fractions (FFA, DAG, and TAG). We also analyzed the effect of the co-expression of the coupled genes ro00075/atf1 and ro00075/atf2 on the biosynthesis and accumulation of TAG in this non-oleaginous strain. Both atf1 and atf2 genes actively involved in TAG accumulation in the oleaginous strain PD630 (Alvarez et al. 2008; Hernández et al. 2013) were expressed at similar levels in F7 pTip-QC2/ro00075-pJAM2/atf1 and F7 pTip-QC2/ro00075-pPR27/atf2 strains as indicated by western blot analysis (data not shown). While co-expression of ro00075/atf1 genes produced a similar effect on lipid accumulation than the expression of the single ro00075 gene (Fig. 6c), the co-expression of ro00075/atf2 genes resulted approximately in a fourfold increase in total fatty acid content (Fig. 6c). In agreement with the quantitative analyses, TLC analysis revealed that the increase in the total fatty acids was produced mainly by a further increase of the FFA and TAG fractions in the strain F7 pTip-QC2/ro00075-pPR27/atf2 (Fig. 7c). Finally, when an analysis by BLAST using the foreign DGATs as query was performed, no ortholog of the atf1 product (OPAG_07257) was found in F7 strain, whereas the protein (F7_3568) of F7 strain exhibited a good identity (I, 53 %) with the atf2 product (OPAG_00138).

Discussion

In this study, we report the occurrence of a diverse number and types of putative PAP2-like proteins coded in the genomes of different rhodococcal species. All investigated species possess small PAP2-like proteins (177–185 aa), whereas R. jostii, R. opacus, and R. erythropolis additionally possess a set of larger PAP proteins (228–376 aa). Interestingly, these last species are able to accumulate significant amounts of TAG, in contrast to R. equi or R. fascians which usually produce lower amounts of TAG with different carbon sources, such as gluconate, glucose, fructose, or hexadecane (Alvarez et al. 1997; Alvarez 2003). In addition, medium-PAPs (228 aa) were only present in TAG-accumulating specialists as R. opacus PD630 and R. jostii RHA1, but absent in the analyzed non-oleaginous rhodococcal strains. These results suggest that medium PAPs may be specific proteins of those specialized TAG-accumulating rhodococci. In this context, RO00075 and OPAG_07226, with similar primary and predicted secondary structures, may constitute key enzymes with identical function in specialized TAG-accumulating R. jostii RHA1 and R. opacus PD630, respectively.

The high diversity of PAP2-like enzymes suggests that different proteins may have appeared during evolution of rhodococci with differing or specialized functions in lipid metabolism of cells, probably to adapt glycerolipid biosynthesis to the fluctuating physiological and environmental conditions. The appearance of additional rhodococcal PAP proteins through the evolution might permit cells to form different cellular pools of DAG for the selective production of phospholipids or TAG or potentially may also allow independent regulation of isoenzymes at the gene expression level or enzymatic activity. Interestingly, all active PAP2 enzymes described to date, including the non-related eukaryotic type LPPs, have shown a predicted six-transmembrane topology (Dillon et al. 1996; Carman 1997; Toke et al. 1998a, b; Sigal et al. 2005; Carman and Han 2006; Zhang et al. 2008; Touze et al. 2008; Comba et al. 2013). Further, the crystal structure and the predicted six-transmembrane topology of the lipid phosphatase PgpB of E. coli have been confirmed experimentally in a recent work (Fan et al. 2014). In this context, a five-transmembrane topology observed in rhodococcal small PAPs might lead to a loss of their functionality during evolution. This result fits to our finding that non-oleaginous rhodococcal strains have only small PAPs. However, additional experiments with the different types of PAPs enzymes analyzing their enzymatic activity and their participation in TAG metabolism are certainly needed.

Since RO00075 of RHA1 strain contains all the conserved domains of PAP2 proteins, and considering that it belongs to a phylogenetic subgroup only represented in TAG-accumulating specialist rhodococci, we considered this protein as the best candidate for being the PAP involved in the TAG biosynthesis pathway of R. jostii. In this context, we investigated the possible role of RO00075 in the accumulation of TAG in the strain RHA1 and its effect on other rhodococcal strains.

The functional role of RO00075 as a PAP enzyme was first analyzed by heterologous expression of the ro00075 gene in E. coli C41 (DE3). The increase of DAG levels in ro00075-expressing strain suggests that RO00075 is a membrane-associated lipid phosphatase enzyme able to catalyze the in vivo formation of DAG from intracellular phosphatidic acid or other analogous lipids from the host cell. In addition, the significant increase in the PAP activity observed in the membrane fraction of the strain expressing the ro00075 gene indicates that RO00075 is able to catalyze the in vitro formation of DAG from phosphatidic acid, confirming its role as a membrane-associated PAP enzyme type 2. Similar results were reported for two PAPs (Lppα and Lppβ) of the actinobacterium S. coelicolor (Comba et al. 2013).

In order to analyze the in vivo role of RO00075 in its native host R. jostii RHA1, we performed a mutagenesis strategy to delete the ro00075 gene. Several attempts to delete ro00075 failed, suggesting that this gene may be important for growth of parental RHA1 cells at least under the used conditions. However, when extrachromosomal copies of ro00075 were present in RHA1 cells, the chromosomal deletion of ro00075 occurred under similar conditions. This result suggested that a minimum expression level of the ro00075 gene may be necessary for growing on solid media. A similar result was reported by Nakamura et al. (2007), in which a knockout mutant for LPPγ (the primary plastidic PAP2 in Arabidopsis) could not be obtained and a lppγ homozygous mutant was only isolated under ectopic overexpression of LPPγ, suggesting that loss of LPPγ may cause lethal effect on viability of that plant.

Interestingly, no inducer was necessary for growth of RHA1 mutant strain in liquid nitrogen-rich media, discarding the essentiality of ro00075 under these conditions. However, some genes have been efficiently expressed in pTip-QC series vectors even in the absence of the inducer (Hirofumi et al. 2010). Due to this non-stringent response of the pTip-QC2 vector used in this study, a basal expression of ro00075 supplied in trans might be sufficient to ensure the cellular growth under nitrogen-rich conditions. In contrast, a basal expression of ro00075 in mutant strain seemed to be insufficient to ensure a normal cell growth under nitrogen-limiting conditions and being resumed only after inducer addition. Altogether, these results suggested that ro00075 may play a possible role in cellular growth, especially under lipid accumulation conditions. Further studies are necessary to verify the potential importance of ro00075 for initial cell growth in R. jostii RHA1.

Further, RO00075 seems to play a relevant role in the TAG metabolism of RHA1 strain. The decrease in total fatty acid, TAG, and DAG contents in mutant strain during cultivation of cells in the absence of the inducer demonstrates that this enzyme contributes actively to the DAG pool for TAG biosynthesis in RHA1 strain. Since the mutant without inducer was still able to accumulate TAG to some extent, the remaining PAP enzymes might be involved in providing DAG for TAG biosynthesis.

Overexpression of ro00075 in RHA1 and PD630 resulted in an increase in TAG content under lipid accumulation conditions (MSM0.1 and MSM0). These results also confirm that RO00075 is involved in the biosynthesis and accumulation of TAG in the native strain R. jostii RHA1 as well as in the taxonomically related strain R. opacus PD630, which possesses an orthologous protein (OPAG_07226, also named PD630_LPD04081) very similar to RO00075 (I, 86 %). Interestingly, according to recent studies by Chen et al. (2013) in R. opacus PD630, expression of the gene encoding for the protein PD630_LPD04081 was higher in the low-nitrogen MSM than in the nitrogen-rich media. Since TAG accumulates to high levels under nitrogen-limiting conditions, the increased expression of this gene suggests that it may be involved in TAG synthesis. The results reported in this work are consistent with those studies.

On the other hand, expression of ro00075 in the non-oleaginous strain R. fascians F7 resulted in a slight increase in total fatty acids at the expense of FFA, DAG, and TAG. This result also supports the involvement of RO00075, which possesses no ortholog in F7 strain, in DAG levels, and in concomitant TAG production. Interestingly, several works have been reported that both lipid fractions, TAG and FFA, which increased in F7 cells can be used for biofuel production (Kosa and Ragauskas 2011; Liang and Jiang 2013; Wu et al. 2014; Janßen and Steinbüchel 2014). However, single expression of ro00075 from an oleaginous bacterium seems not to be sufficient to significantly increase TAG biosynthesis in the non-oleaginous strain R. fascians F7, at least under the conditions used in this study. A possible reason for this result might be due to the flux of precursors in central and lipid metabolism being not sufficient to support a significant increase in TAG accumulation in a non-oleaginous strain. Alternatively, since the predicted structure of RO00075 was different compared to the native PAP found in F7 cells, RO00075 may exhibit a limited activity in this non-oleaginous strain (Fig. 2b). In this context, the different structure of the RO00075 protein might be not well recognized by the anchoring machinery of F7 cells or, alternatively, the activity of this protein may also be modulated by a posttranslational allosteric process in native cells, which may not occur in F7 cells. On the other hand, the two native WS/DGAT enzymes found in F7 strain, which were required to convert DAG to TAG, might have low activity. In order to analyze this last possibility, we analyzed the effect of the co-expression of the coupled genes ro00075/atf1 and ro00075/atf2 in TAG biosynthesis in this non-oleaginous strain. Since the effect on lipid accumulation by co-expressing of ro00075/atf1 genes was similar to that observed by expressing of the single ro00075 gene, this result suggests that the atf1 product from PD630 is not active enough to support a significant TAG accumulation by F7 cells, at least under the conditions used in this study. However, co-expression of ro00075/atf2 genes in the strain F7 resulted in a higher total fatty acid content, at the expense of TAG and FFA fractions, mainly demonstrating that both RO00075 and Atf2 were sufficient to improve the de novo TAG biosynthesis using precursors from the glycerophospholipid metabolism of non-oleaginous rhodococci. Interestingly, these results are in agreement with the fact that only an homolog for the atf2 product could be found in F7 cells. On the other hand, co-expression of both PAP and WS/DGAT enzymes has been previously reported in E. coli as an attempt to establish TAG formation in this bacterium. For example, Rucker et al. (2013) co-expressed the native PgpB enzyme (with PAP activity) of E. coli and the AtfA enzyme of A. baylyi to produce TAG in E. coli. Although E. coli cells were able to produce TAG to some extent, a loss of its cellular viability after 8 h following induction was also observed. A similar study was performed by Comba et al. (2013) using the same biosynthetic route, but expressing the genes sco0958 (DGAT) and lppα or lppβ (PAPs) from S. coelicolor in E. coli. Interestingly, the growth of F7 cells was no affected after co-expression of two genes from oleaginous rhodococcal strains (ro00075 RHA1/atf2PD630), thus suggesting that this genetic procedure may be useful as an alternative form to significantly increase TAG production in a non-oleaginous strain. Further, the results observed in this study and those previously reported are in agreement with the hypothesis that both kinds of enzymes, DGAT and PAP, might catalyze the rate-limiting steps in TAG formation.

Understanding rhodococcal lipid metabolism is of great interest regarding the development of new technologies through the design and refinement of these oleaginous cell factories. In this work, we report the functional elucidation of an individual gene coding for a PAP2 enzyme in R. jostii RHA1 involved in the production of DAG as a precursor for TAG biosynthesis. The manipulation of this gene alone or in combination with other identified rhodococcal genes also involved in TAG accumulation may contribute to establish diverse strategies to increase lipid production for industrial applications. Thus, our results provide new elements and tools for further cell engineering to achieve sustainable and cost-effective single-cell oil production in oleaginous bacteria.