Introduction

Triacylglycerols (TAG) and wax esters (WE) are besides the predominant polyhydroxyalkanoic acids widespread hydrophobic intracellular storage compounds for carbon and energy in bacteria (Steinbüchel 1996). TAG and WE inclusions (oxoesters of long-chain fatty acids with glycerol or primary fatty alcohols, respectively) are formed by the wax ester synthase/acyl-CoA:diacylglycerol acyltransferase (WS/DGAT; the atfA gene product) at the cytoplasmic surface of the plasma membrane (Wältermann and Steinbüchel 2005). WS/DGAT was recently characterized as a novel class of acyl-CoA-dependent acyltransferase that exhibits no sequence homology to any known acyltransferase involved in storage lipid synthesis in eukaryotes. The promiscuous WS/DGAT (AtfA) is the key enzyme for biosynthesis of TAG and WE in Acinetobacter baylyi strain ADP1 (ATCC 33305) and probably also in most other lipid accumulating bacteria (Kalscheuer and Steinbüchel 2003; Wältermann et al. 2007). AtfA is an unspecific acyltransferase which simultaneously synthesizes WE and TAG by utilizing long-chain fatty alcohols and diacylglycerols as acceptor substrates and fatty acid CoA thioesters (acyl-CoA) as acyl donor substrates (Kalscheuer and Steinbüchel 2003). The biochemical analyses showed that AtfA of A. baylyi strain ADP1 exhibits a very broad substrate range, accepting primary, secondary, cyclic, and phenolic acyl alcohols, diols and dithiols as well as mono- and diacylglycerols (Kalscheuer et al. 2003; Stöveken et al. 2005; Uthoff et al. 2005). Depending on the metabolism of the microorganism, the media composition, or genetic engineering, diverse types of fatty acid esters can be synthesized in vivo by expressing WS/DGAT in different recombinant hosts leading to new biological, technological, or therapeutic applications (Stöveken and Steinbüchel 2008).

Numerous homologous proteins especially in actinomycetes were identified by an alignment of the amino acid sequence of WS/DGAT with the publicly accessible sequence databases. A general overview on occurrence and characteristics of putative WS/DGAT proteins in bacteria is given by Wältermann et al. (2007). It is remarkable that a large group of up to 14 homologous WS/DGAT proteins is present in most actinomycetes. In Mycobacterium tuberculosis, 15 homologous proteins were discovered suggesting an important function of these proteins (Daniel et al. 2004). The capability of TAG accumulation is also widely distributed among members of the genus Streptomyces. Streptomycetes are filamentous Gram-positive bacteria with unusual complex cell morphology. The complete genome sequences of Streptomyces coelicolor A3(2) (DSM40783) and Streptomyces avermitilis MA-4680 (DSM 46492) were published recently by Bentley et al. (2002) and Ikeda et al. (2003). In contrast to the large number of homologs found in mycobacteria, only three WS/DGAT homologous proteins in S. coelicolor A3(2) and a single WS/DGAT homologous protein in S. avermitilis MA-4680 were identified (Kalscheuer et al. 2003). The accumulation of TAG as insoluble cytoplasmic granules in the cell up to 5% of the cellular dry weight (CDW) has already been described for S. coelicolor A3(2) and Streptomyces lividans 66. It has been proposed that the accumulated TAGs might act as a carbon source for the biosynthesis of antibiotics (polyketide compounds) derived from acetyl-CoA/malonyl-CoA precursors (Olukoshi and Packter 1994; Packter and Olukoshi 1995). However, this hypothesis was disproven by the recently published report on the TAG biosynthesis pathway of S. coelicolor M145 (Arabolaza et al. 2008). The opposite behavior was observed in a disruption mutant, which stored about 70% less TAG but produced almost 20% more actinorhodin compared to the wild type, and although all three WS/DGAT homologous genes of S. coelicolor M145 were deleted, a decreased but still residual storage lipid accumulation remained in the triple mutant, indicating the presence of alternative neutral lipid biosynthesis pathways. Similar observations of residual lipid production were also made for WS/DGAT deletion mutants of A. baylyi strain ADP1 (Kalscheuer and Steinbüchel 2003), Alcanivorax borkumensis SK2 (Kalscheuer et al. 2007), and Rhodococcus opacus strain PD630 (Alvarez et al. 2008).

In this study, we investigated the role of WS/DGAT homologous proteins in the neutral lipid metabolism of actinomycetes with regard to the question whether TAG biosynthesis in Streptomyces sp. occurs exclusively by this new type of acyltransferases, by so far undiscovered DGAT isoenzymes, or even by alternative acyl-CoA-independent biosynthesis pathways.

Materials and methods

Bacterial strains and plasmids

Bacterial strains and plasmids used in this study are listed in Table 1.

Table 1 Bacterial strains and plasmids used in this study
Full size table

Media and cultivation conditions

S. avermitilis MA-4680 was grown in modified mineral salts medium (MSM; Schlegel et al. 1961) containing 0.05% (w/v) yeast extract, 1.0% (w/v) glucose as carbon source, and 0.1% (w/v) NH4Cl. Cells were cultivated for 72 h at 30°C. For induction of storage lipid biosynthesis, a reduced concentration of 0.01% (w/v) NH4Cl was used (“storage conditions”). For preparation of protoplasts, cells were cultivated in yeast extract–malt extract medium (Hopwood et al. 2000); after protoplast transformation, R5 medium was used for regeneration (Okanishi et al. 1974) and CASO (Merck) medium for selection. After conjugational transfer, cells were regenerated on mannitol soy flour medium (Hobbs et al. 1989) and selected on modified MSM. Recombinant strains of Escherichia coli were cultivated at 37°C on Luria–Bertani medium (Sambrook et al. 1989) containing 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) to induce expression of the protein encoded by pET23a(+). For preparation of electrocompetent cells of E. coli DY378, cells were cultivated at 30°C and induced for 15 min at 42°C (Yu et al. 2000). All liquid cultures were incubated aerobically in baffled Erlenmeyer flasks on an orbital shaker and were inoculated (1%, v/v) from a well-grown preculture. Solid media contained 1.8% (w/v) agar–agar. If required, antibiotics were added at the following concentrations: 75 μg ampicillin (Ap) ml−1, 50 μg apramycin (Apra) ml−1, 25 μg chloramphenicol (Cm) ml−1, 50 μg kanamycin (Km) ml−1, 12 μg thiostreptone (Tsr) ml−1, and 20 or 10 μg nalidixic acid (Ndx) ml−1 for E. coli or Streptomyces, respectively.

Electron microscopy studies

For transmission electron microscopy (TEM), cells of S. avermitilis MA-4680 were cultivated in modified MSM under storage conditions, harvested, and washed twice with 125 mM sodium phosphate buffer (pH 7.4). Cells were fixed with 4.6% (w/v) paraformaldehyde and postfixed with 1% (w/v) osmium tetroxide in 0.1 M phosphate buffered saline (pH 7.3). For thin sectioning, samples were embedded in Lowicryl K4M (Polyscience); sections were made with an Ultracut S apparatus (LEICA Mikroskopie und Systeme) using a diamond knife and were then positioned on a 200-mesh copper grid. Imaging was performed with an H-500 TEM (Hitachi) in the bright-field mode at 75 kV acceleration voltage at room temperature. Micrographs were taken with an Agfa-Gevaert 23 D 56 film.

Lipid analysis

Thin layer chromatography (TLC) of lipid extracts from whole cells was done as described previously (Kalscheuer and Steinbüchel 2003). Cetylpalmitate or oleyl oleate and triolein were used as reference substances for WEs and TAGs, respectively. Fatty acid analysis of whole cells and purified TAGs was done by gas chromatography (GC) and gas chromatography/mass spectrometry analysis (MS) according to Kalscheuer et al. (2004). Fatty acid methyl esters were analyzed on an Agilent 6850 GC (Agilent Technologies) equipped with a BP21 capillary column (50 m × 0.22 mm, film thickness 250 nm; SGE) and a flame ionization detector (Agilent Technologies).

Preparation of crude cell extracts

Cultivated cells of Streptomyces sp. were centrifugated (20 min, 6,238×g, 4°C) and washed with 125 mM sodium phosphate buffer (pH 7.4). Cells were resuspended in the same buffer, incubated with 1 mg lysozyme ml−1 for 45 min on ice, and passaged through a French pressure cell (1,000 MPa) to obtain crude extracts. Cells of E. coli were grown to an OD600 nm of 0.5 before IPTG was added to a final concentration of 1 mM. Induced cultures were incubated for further 3 h, harvested by centrifugation (3,345×g, 20 min, 4°C), washed twice, and resuspended in sodium phosphate buffer. Cells were disrupted by ultrasonification and protein concentrations were determined by the Bradford (1976) method.

Determination of enzyme activities

WS/DGAT activity was determined in a total volume of 250 μl containing 12.5 μg bovine serum albumin ml−1, 4.72 μM [1-14C]palmitoyl-CoA (specific activity 1.961 Bq pmol−1; Hartmann Radiochemicals), 125 mM sodium phosphate buffer (pH 7.4), and different acceptor molecules at a concentration of 3.75 mM. 1-Hexadecanol and 1,2-dipalmitoylglycerol were used as standard substrates for assaying WS and DGAT activity, respectively. Alternatively, a range of various-chain-length linear alcohols (1-butanol, 1-decanol, 1-hexadecanol, 1-tetracosanol), various acyl-glycerols (1,2- and 1,3-dipalmitoylglycerol, 1-, 2-, and 3-monopalmitoylglycerol) as well as cyclic and aromatic alcohols (cyclohexanol, 2-cyclohexylethanol, phenylethanol) were tested as acyl acceptors. The enzyme reaction was performed as described previously by Stöveken et al. (2005). The reaction products were separated by TLC using different solvent systems for the separation of linear, cyclic, and aromatic WEs, monoacylglycerols, DAGs, or TAGs. After separation of lipids and staining of the TLC plates with iodine vapor, spots corresponding to the reaction products were scraped from the plates, and radioactivity was measured by scintillation counting.

Cloning of AtfA homologous gene from S. avermitilis MA-4680

The gene encoding the AtfA homologous protein SAV7256 of S. avermitilis MA-4680 was amplified from total genomic DNA by polymerase chain reaction (PCR). The resulting product was cloned as HindIII-XhoI fragment into the expression vector pET23a(+) colinear to the T7 promoter to obtain pET23a::SAV7256. This construct was then transformed into E. coli BL21(DE3).

SDS-PAGE and immunoblot analysis

Protein content was analyzed according to Bradford (1976). Protein samples were resuspended in gel loading buffer and separated in 12.5% (w/v) sodium dodecyl sulfate (SDS) polyacrylamide gels as described by Laemmli (1970). Proteins in the polyacrylamide gels were visualized using Coomassie brilliant blue R250 according to Weber and Osborn (1969). For immunological detection of WS/DGAT homologous proteins, the proteins in the polyacrylamide gels were transferred onto a nitrocellulose membrane (Towbin et al. 1979). Proteins on the membrane were analyzed immunologically employing 2 μg polyclonal rabbit anti-WS/DGAT IgGs ml−1 (Stöveken et al. 2005). IgGs were visualized on immunoblots using goat antirabbit IgG–alkaline phosphatase conjugates (diluted 1:20,000 in 10 mM potassium phosphate buffer, pH 7.2; 0.5%, w/v, NaCl; 0.1%, v/v, Tween 20).

PCR and RT-PCR analysis

PCR was performed as described by Sambrook et al. (1989) using Plantinum Pfx polymerase (GIBCO BRL Life Technologies). For RT-PCR, the OneStep RT-PCR kit (Qiagen) was used according to the manufacturer’s instructions employing oligonucleotide primers listed in Table 2.

Table 2 Oligonucleotides used in this study as primers for PCR, diagnostic PCR, RT-PCR, and sequencing
Full size table

DNA sequencing, sequence analysis, and alignments

DNA sequencing was performed by employing IRD800-labeled primers, a model 4000L DNA sequencer (Li-COR), and a SequiTherm EXCEL II Long-Read L-C kit-LC (Epicentre) according to the manufacturer’s instructions. Obtained sequences were analyzed with Genamics Expression software (http://genamics.com/expression/index.htm). Sequence comparisons and alignments were performed using the BLAST online service available on National Center for Biotechnology Information (NCBI), and alignments were done with BioEdit (Hall 1998) and ClustalW (Thompson et al. 1994).

Transcription analysis of SAV7256 in S. avermitilis MA-4680

Cells were cultivated separately on modified MSM under growth as well as under storage conditions. After harvesting, the cells were resuspended in Tris–ethylenediaminetetraacetic acid buffer (pH 8.0) containing 1 mg lysozyme ml−1 and incubated for 20 min at 37°C. The isolation of RNA was performed by using the RNeasy RNA purification kit (Qiagen) according to the manufacturer’s protocol. Additionally, 0.1 mm glass beads (Carl Roth) and a Silamat® S5 (Ivoclar Vivadent) were used for extraction of RNA. RT-PCR was performed using 1 μg RNA as template. A negative control for DNA contamination was done by adding the RNA template after the reverse transcriptase step.

Isolation, manipulation, and transfer of DNA

Isolation of total DNA of Streptomyces cells was performed according to Marmur (1961). The isolation of plasmid and cosmid DNA was conducted according to the method of Birnboim and Doly (1979). DNA manipulations and other standard molecular biology techniques were performed according to Sambrook et al. (1989). DNA was introduced into Streptomyces cells either by protoplast transformation according to Hopwood et al. (2000) or by conjugation using E. coli ET12567 with the helper plasmid pUZ8002. Competent cells of E. coli were prepared and transformed by the CaCl2 procedure as described by Hanahan (1983). Electrocompetent cells of E. coli DY378 were prepared according to the method of Yu et al. (2000).

Gene disruption and knock out constructs

For inactivation of the SAV7256 gene, an internal 0.4-kbp DraII fragment of SAV7256 was deleted and replaced by a ΩKm cassette yielding pGEM-T Easy::SAV7256ΩKm A 44.1-kbp fragment containing the gene SAV7256 from cosmid CL_222_G04 (obtained from the genomic library of S. avermitilis MA-4680; The Kitasato Institute, Japan, http://avermitilis.ls.kitasato-u.ac.jp/) was cloned into the singular BamHI site of vector pKU402. For gene disruption mutagenesis, the intact gene SAV7256 in CL_222_G04 was replaced by the disrupted gene SAV7256ΩKm yielding pKU402::CL_222_G04::SAV7256ΩKm using the E. coli strain DY378 which is suitable for integration of short linear DNA (at least 35–50 bp) by homologous recombination (Yu et al. 2000).

Amino acid sequence data

The primary amino acid sequences of proteins investigated in this study are deposited under following accession numbers: AtfA (AF529086) from A. baylyi strain ADP1, SAV7256 (NP_828432) from S. avermitilis MA-4680, SCO0958 (NP_625255), SCO0123 (NP_624462), and SCO1280 (NP_625567) from S. coelicolor A3(2).

Results

Identification of a WS/DGAT homologous protein in S. avermitilis MA-4680 and primary structure comparison

Since it is known that streptomycetes are capable of accumulating lipids in the cytoplasm (Olukoshi and Packter 1994; Packter and Olukoshi 1995), the genome sequence of S. avermitilis strain MA-4680 was searched for genes coding for homologs of the recently detected promiscuous acyltransferase of A. baylyi strain ADP1 (Kalscheuer and Steinbüchel 2003). AtfA is present in A. baylyi and supposably in most other bacteria that are capable of synthesizing TAG and WE (Wältermann et al. 2007).

Based on the amino acid sequence of AtfA, one homologous protein in S. avermitilis MA-4680 referred to as SAV7256 was identified. Up to now, no biological function could be assigned to this hypothetical protein. SAV7256 showed an amino acid homology of 25.3% to AtfA. The three WS/DGAT homologs SCO0958, SCO0123, and SCO1280 of strain S. coelicolor A3(2), which were investigated recently (Arabolaza et al. 2008), showed amino acid homologies ranging from 17.7% to 26.6% to AtfA (Table 3). SAV7256 from S. avermitilis MA-4680 and SCO0958 from S. coelicolor A3(2) exhibited the highest homology (81%) to each other, as well as to AtfA (25.3% and 26.6%, respectively). The other two homologs SCO0123 and SCO1280 from S. coelicolor A3(2) exhibited only weak homologies to each other (29%), as well as to SAV7256 and SCO0958 (Table 4).

Table 3 Genes for WS/DGAT homologs detected in the genomes of S. avermitilis MA-4680 and S. coelicolor A3(2)
Full size table
Table 4 Homologies between different WS/DGAT homologous proteins
Full size table

The HHAxxDG active site motif found in AtfA of A. baylyi strain ADP1, which has been shown to be essential for catalytic activity (Stöveken et al. 2009), was strictly conserved in the acyltransferase homolog of S. avermitilis MA-4680 (SAV7256) and in one AtfA homolog of S. coelicolor A3(2) (SCO0958; Fig. 1). In the two other acyltransferase homologs of S. coelicolor A3(2), this motif was less conserved (“HAxxDG” in SCO0123 and “HxxxDG” in SCO1280).

Fig. 1
figure 1

Amino acid sequence alignment of the WS/DGAT enzymes AtfA from A. baylyi strain ADP1, SAV7256 from S. avermitilis MA-4680, and SCO0958 from S. coelicolor A3(2). Analysis was done using the ClustalW program (Thompson et al. 1994). Residues identical in all sequences are shaded in dark gray; similar residues are shaded in light gray. A putative active site motif is boxed

Full size image

Although the amino acid sequences of SCO0958 and SAV7256 showed a relative weak overall homology to AtfA, several highly conserved amino acid residues were identified (Fig. 1). Furthermore, the proteins exhibited highly similar hydrophobicity profiles (Fig. 2), which might be indicative for a similar structure and function of SAV7256, SCO0958, and AtfA.

Fig. 2
figure 2

Hydrophobicity plot of the WS/DGAT enzymes SAV7256 and SCO0958 compared to AtfA. Calculation was done as described by Kyte and Doolittle (1982). Hydrophobic regions show values >0

Full size image

Occurrence and composition of intracellular storage lipid inclusions

S. avermitilis MA-4680 was cultivated in modified MSM medium containing 0.01% (w/v) NH4Cl and 1% (w/v) glucose, which are conditions known to induce TAG and WE biosynthesis in other bacteria. Transmission electron micrographs of thin sections revealed a large number of inclusions in the cells occupying the major part of the cytoplasm (about 70–80%; Fig. 3a, b). Each cell contained between ten and 30 well-separated granules with diameters between 100 and 500 nm exhibiting no coalescence. All granules stained positive with the lipophilic fluorescent dye Nile red (Spiekermann et al. 1999), which indicates the accumulation of abundant amounts of hydrophobic compounds in the granules (Fig. 3a).

Fig. 3
figure 3

Structure and accumulation of neutral lipid inclusions in S. avermitilis MA-4680 grown under growth (upper line) and storage (lower lines) conditions for 72 h at 30°C. Fluorescence images after staining with Nile red (a) as described in “Materials and methods” section, corresponding phase-contrast (b), and differential interference contrast images (c). Ultrathin sections were analyzed by TEM (d, e) as described in “Materials and methods” section. Scale bars: a–c 2, d 1, and e 0.5 μm

Full size image

Thin layer chromatography of whole cell extracts revealed the presence of TAG, while WE were not visible. Combined GC/MS analyses of the methyl esters obtained from whole cells, cultivated under storage conditions, as well as from purified TAG, confirmed a similar fatty acid composition with a high amount of iso- and anteiso-methyl-branched fatty acids. The cellular fatty acid composition is listed in Table 5, showing that anteiso-pentadecanoic acid was the predominant constituent. Only when S. avermitilis was cultivated on a mixture of 1% (w/v) glucose and 0.3% (v/v) hexadecanol, very small amounts of WE (probably less than 5%) could be detected in total lipid extracts by TLC (data not shown). These WE represented a mixture of hexadecanoic acid and various alcohols. GC/MS analysis of TAG, accumulated under the same conditions, revealed only small differences in the fatty acid spectrum compared to those TAG obtained under storage conditions. Cultivation on hexadecanol or hexadecane alone did not result in the accumulation of any WE or TAG (data not shown). Evidence for the presence of poly(3-hydroxybutyrate) or other polyhydroxyalkanoates in the cells was not obtained. Therefore, the inclusions visible in Fig. 3 consist most probably of TAG.

Table 5 Fatty acid composition of whole cells of S. avermitilis MA-4680 after cultivation of cells in modified MSM containing 0.01% (w/v) NH4Cl and 1% (w/v) glucose for 72 h
Full size table

Enzymatic detection of DGAT and WS activities and transcription analysis in S. avermitilis MA-4680

In crude extracts obtained from cells of S. avermitilis MA-4680 grown under storage conditions, high DGAT activity was detected (specific activity 143 pmol (mg min)−1). In contrast, very low WS activity was measured (specific activity 1.3 pmol (mg min)−1). DGAT activity increased in the cells during cultivation under storage conditions, which correlated with the accumulation of TAG (Fig. 4). These data are consistent with the high TAG and nondetectable WE content of the cells. The increase of DGAT activity under lipid storage conditions indicates the induction of gene expression of one or more DGAT enzymes. To examine whether SAV7256 from S. avermitilis MA-4680 is expressed under different cultivation conditions, RT-PCR was performed. For this, cells were grown in MSM containing 1% (w/v) glucose and 0.1% (w/v) NH4Cl or 0.01% (w/v) NH4Cl, allowing accumulation of TAGs. Transcripts of the SAV7256 gene were detected under both tested cultivation conditions. Thus, SAV7256 is potentially responsible for the basal TAG accumulation in strain S. avermitilis MA-4680 (Fig. 5).

Fig. 4
figure 4

Time courses of WS and DGAT activities related to cell density (CDW; a) and TAG production (b) in S. avermitilis MA-4680. Cells were cultivated under storage conditions for 72 h at 30°C. TAGs were extracted each from 1.5-mg lyophilized cell material with 200 μl chloroform/methanol (1:1, v/v) for 30 min. The supernatant was subjected to silica TLC using the solvent system hexane/diethylether/acetic acid (80:20:1, v/v/v) for TAG analysis

Full size image
Fig. 5
figure 5

Agarose gel electrophoresis (1%, w/v) of RT-PCR products of SAV7256 in S. avermitilis MA-4680. Expression of SAV7256 was analyzed in samples derived from cells cultivated under growth as well as under lipid storage conditions for 72 h at 30°C. The left side shows the 1,344 bp transcript of SAV7256 amplified from RNA from S. avermitilis under growth conditions, the right side under lipid storage conditions accordingly. M PstI-digested λ DNA, + RT-PCR assay, − control for DNA contamination

Full size image

Next, it was examined whether the existing polyclonal antibodies produced against AtfA from A. baylyi strain ADP1 (Stöveken et al. 2005) can be used to detect SAV7256 heterologously expressed in E. coli. The Western blot analysis on SAV7256 from recombinant E. coli as well as on crude extracts of S. avermitilis grown under normal growth as well as under lipid storage conditions resulted in the immunological detection of SAV7256 based on a cross reaction with the AtfA antibodies (data not shown). The expression level of SAV7256 isolated from crude extracts was similar under both cultivation conditions and in agreement with the results of transcription analysis.

To provide a final proof that SAV7256 was the only or at least major enzyme responsible for the storage lipid accumulation in the investigated S. avermitilis strain, we tried to generate a knock out mutant defective in the gene. However, even after several attempts with different variations of the constructs and transfer methods, a SAV7256 knock out mutant could not be obtained.

Heterologous expression in E. coli and biochemical characterization of the acyltransferase

The identified WS/DGAT homologous gene SAV7256 was heterologously expressed in E. coli BL21(DE3) as described in the “Materials and methods” section. Protein expression was confirmed by SDS-polyacrylamide gel electrophoresis, and enzyme activity was determined by a radiometric assay using linearly saturated long-chain C16:0-CoA ([1-14C]palmitoyl-CoA) as acyl donor. If not stated otherwise below, 1-hexadecanol and 1,2-dipalmitoylglycerol were used as standard substrates for assaying WS and DGAT activity, respectively. SAV7256 surprisingly showed a high WS but low DGAT activity of 71.3 pmol (mg min)−1 and 16.8 pmol (mg min)−1, respectively. These values corresponded to 46% and 31%, respectively, of the activities of AtfA from A. baylyi ADP1 (positive control) and therefore, SAV7256 could be designated as an acyltranferase. As expected, the negative vector control E. coli BL21(DE3) (pET23a) proved low WS (0.27 pmol (mg min)−1) as well as low DGAT (0.21 pmol (mg min)−1) activity. It is noteworthy that the ratio of WS to DGAT activity is in stark contrast to enzyme activities detected in crude extract of S. avermitilis MA-4680 grown under storage conditions with high DGAT and low WS activity.

The substrate specificity of the S. avermitilis acyltransferase SAV7256 was measured in crude extracts of E. coli BL21(DE3) expressing pET23a::SAV7256. Since E. coli does not possess endogenous WS and DGAT activity, the results of such measurements allowed a general view on the substrate range of the enzyme from S. avermitilis, regardless of its purity. Using [1-14C]palmitoyl-CoA as acyl donor and 1-hexadecanol as reference acyl acceptor (100%), the highest activity of SAV7256 was measured with the medium-chain-length alcohol 1-decanol (110.9%) as substrate. The short-chain-length C4 alcohol as well as the long-chain-length C24 alcohol was utilized by this enzyme with lower activity. Lower activities were also measured for the acyl-glycerols 1,2- and 1,3-dipalmitoylglycerol (6.6 and 1.5%, respectively) as well as with 1-, 2-, or 3-monopalmitoylglycerol (5.8%, 14.3%, and 12.7%, respectively). The substrate range of SAV7256 was further tested for cyclic and aromatic alcohols. 2-Cyclohexylethanol was acylated with relative high specific activity (24.9%), whereas the activity with cyclohexanol was rather low (1.2%). Phenylethanol, the only aromatic compound tested in this study, could also serve as substrate for SAV7256. In summary, enzyme SAV7256 from S. avermitilis showed a preference for medium-chain-length linear fatty alcohols and exhibited generally a narrower substrate range than that of AtfA of A. baylyi strain ADP1 (Stöveken et al. 2005; Table 6).

Table 6 Specific activities of SAV7256 from S. avermitilis MA-4680 and AtfA from A. baylyi strain ADP1 in use with different acyl acceptors
Full size table

Discussion

Detailed investigations on the TAG accumulation in S. coelicolor were already done by Olukoshi and Packter (1994), Shim et al. (1997), and Arabolaza et al. (2008), whereas investigations in S. avermitilis MA-4680 were only carried out on whole cell lipid extracts so far (Cropp et al. 2000). In this study, we have identified one protein in S. avermitilis MA-4680, which is homologous to the promiscuous AtfA, the key enzyme for neutral lipid biosynthesis in A. baylyi ADP1. The high degree of homology between SAV7256 from S. avermitilis and SCO0958 from S. coelicolor A3(2), their verified acyltransferase activities, as well as the same organization of the gene loci, strongly suggest a similar function and orthologous origin. Moreover, the low homologies between the S. coelicolor genes SCO0958, SCO0123, and SCO1280 (Arabolaza et al. 2008) indicate that they may have been developed through gene duplication in the region of the chromosome arm leading to paralogous genes.

In this study, the accumulated storage lipids of S. avermitilis were biochemically analyzed in detail. The results shown in Fig. 3 are in accordance with the amount, structure, and size of accumulated TAG grana (76–87% of the CDW, 100–600 nm) described for R. opacus strain PD630 (Alvarez et al. 1996). TAG and WE inclusions in lipid accumulating bacteria are usually organized in form of spherical lipophilic grana (Wältermann and Steinbüchel 2005); it was recently that also rectangular, disk-shape, needle-like, or irregularly shaped forms were observed in the n-hexadecane-utilizing bacterium Acinetobacter sp. strain M-1 and in the hydrocarbonoclastic marine bacterium A. borkumensis SK2 (Ishige et al. 2002; Kalscheuer et al. 2007). Such untypical forms are probably caused by a biased fatty acid composition of the lipids with a high content of saturated fatty acids. A pecularity of streptomycetes is the presence of iso- and anteiso-methyl-branched fatty acids, representing more than 60% of total cellular fatty acids which are rarely found in other bacteria. The fatty acid profile of TAG purified from S. avermitilis MA-4680 was identical to that of total cells, which is in accordance with observations in S. coelicolor M145 (Arabolaza et al. 2008). Furthermore, TAG from S. avermitilis contained a substantial portion of odd chain fatty acids. This was similar to TAG from R. opacus, whereas odd chain fatty acids normally occur only in small quantities in bacteria (Fulco 1983; Alvarez et al. 1996; Kalscheuer et al. 2001).

Lack of de novo biosynthesis of WE despite the presence of endogenous WS activity and the intracellular availability of acyl-CoA for TAG biosynthesis indicate that, unlike A. baylyi, S. avermitilis is unable to reduce acyl-CoA to fatty alcohols and therefore could not form WEs. This is in agreement with reports on Mycobacterium smegmatis (Kalscheuer and Steinbüchel 2003). The presence of an acyl-CoA-reductase (acr1) required for fatty alcohol formation was described for A. baylyi ADP1 (Reiser and Somerville 1997); however, no homologous gene to acr1 was identified in the genome of S. avermitilis MA-4680.

In crude extracts of S. avermitilis, the increase in DGAT activity correlated with TAG accumulation. Thus, it can be assumed that TAG accumulation in this strain occurred via the classical acyl-CoA-dependent DGAT pathway. Due to the contrary ratio of DGAT and WS activities found in crude extracts of S. avermitilis and recombinant E. coli with high DGAT activity in cells of S. avermitilis and only relatively low DGAT activity of SAV7256, it is likely that SAV7256 is only in part responsible for TAG synthesis in this strain. However, neither further WS/DGAT homologous genes nor genes homologous to the eukaryotic acyltransferases DGAT1 and DGAT2 (Cases et al. 1998, 2001) could be identified in the genome of S. avermitilis MA-4680. Therefore, this suggests the presence of yet unidentified DGAT isoforms, as it has been proposed for some other TAG accumulating bacteria (Kalscheuer et al. 2007). The definite role of SAV7256 for neutral lipid production in S. avermitilis can only be determined by a knockout mutant which became unfortunately not available.

Due to the inability of S. avermitilis to synthesize WE de novo, synthesis of linear WE may not be the main function of SAV7256. The involvement of this acyltransferase in synthesis of yet unidentified fatty acid ester substances, which could not be detected by the applied methods, is also conceivable.

Arabolaza et al. (2008) recently generated disruption mutants of S. coelicolor M145 defective in the three AtfA homologs SCO0958, SCO0123, and SCO1280. Deletion of SCO0958 led to a reduced de novo TAG biosynthesis of approximately 70%. Mutations in the genes for the other two proteins showed no influence on neutral lipid accumulation. In addition, overexpression of SCO0958 resulted in increased TAG levels showing that this enzyme is the major acyltransferase for TAG accumulation in S. coelicolor M145. Remarkably, even in the triple mutant of strain M145, a reduced but substantial level of TAG accumulation remained, indicating the existence of an alternative non-WS/DGAT-dependent TAG biosynthesis pathway. It seems obvious that not WS/DGAT implemented production of storage lipids is mediated, at least in part, by an acyl-CoA-dependent pathway. Moreover, a substantial phospholipid:diacylglycerol acyltransferase (PDAT) activity was described for S. coelicolor (Arabolaza et al. 2008). In this acyl-CoA-independent reaction phospholipids act as acyl donors and DAG as acyl acceptor. However, it is unclear to what extend PDAT, which mediates TAG synthesis in plants and yeast (Dahlqvist et al. 2000), contributes to TAG production in S. coelicolor. PDAT activity was not determined in S. avermitilis in the current work. Trace amounts of TAGs in an atfA deletion mutant of A. baylyi strain ADP1 (Kalscheuer and Steinbüchel 2003) and in an atfA1 and atfA2 double deletion mutant of A. borkumensis SK2 (Kalscheuer et al. 2007), as well as in the atf1 disruption mutant of R. opacus strain PD630 (Alvarez et al. 2008), support the speculation about alternative pathways.

S. avermitilis MA-4680 is a bacterium with high industrial potential due to the production of the secondary metabolite avermectin, which is used in agriculture, veterinary, and human medicine as antiparasitic agent (Yoon et al. 2004). During fatty acid de novo synthesis, the central intermediates acetyl-CoA and propionyl-CoA are used for the synthesis of linear and branched fatty acids. Avermectin biosynthesis additionally requires the precursors isobutyryl-CoA and 2-methylbutyryl-CoA, which are synthesized by the degradation of l-isoleucine and l-valine (Ikeda et al. 1999). Based on these pathways, TAG and avermectin biosynthesis share the same metabolic intermediates (Fig. 6). A quantitative correlation between avermectin and lipid production was observed (Novák et al. 1990). In contrast to avermectins, the fatty acid moieties of TAG can be mobilized by intracellular lipases and catabolized through β-oxidation into the precursor substrates, which in turn can serve for biosynthesis of avermectins (Fig. 6). In literature, it is discussed that TAG accumulated in the early stationary phase of Streptomyces represented a depot for the synthesis of secondary metabolites during the late stationary phase (Olukoshi and Packter 1994; Shim et al. 1997). Thus, the accumulated TAG in S. avermitilis may fulfill the function of a mobilizable depot for precursors of the avermectin biosynthesis, or TAG and avermectin synthesis are in competition to the limited supply of the same precursors. The hypothesis of a competition between TAG and antibiotic production was corroborated by a TAG-deficient mutant of S. coelicolor. It was found that the SCO0958 gene disruption mutant stored approximately 70% less TAGs but produced 20% more actinorhodin in comparison to the wild-type strain (Arabolaza et al. 2008). These facts support the assumption of competition for the same precursors. In the future, the generation of a SAV7256 mutant of S. avermitilis MA-4680 will be necessary to understand and explain the competition behavior between TAGs and avermectins, particularly with regard to industrial applications. A better understanding of the lipid metabolism in S. avermitilis and its engineering may lead to improved production strains.

Fig. 6
figure 6

Correlation between biosynthesis of TAGs and avermectins in S. avermitilis MA-4680. TAG and avermectin production share the same precursor substrates. In contrast to avermectins, TAGs can be mobilized and the released fatty acids can be degraded by β-oxidation

Full size image