Active Multienzyme Assemblies for Long-Chain Olefinic Hydrocarbon Biosynthesis

ABSTRACT

Bacteria from different phyla produce long-chain olefinic hydrocarbons derived from an OleA-catalyzed Claisen condensation of two fatty acyl coenzyme A (acyl-CoA) substrates, followed by reduction and oxygen elimination reactions catalyzed by the proteins OleB, OleC, and OleD. In this report, OleA, OleB, OleC, and OleD were individually purified as soluble proteins, and all were found to be essential for reconstituting hydrocarbon biosynthesis. Recombinant coexpression of tagged OleABCD proteins from Xanthomonas campestris in Escherichia coli and purification over His₆ and FLAG columns resulted in OleA separating, while OleBCD purified together, irrespective of which of the four Ole proteins were tagged. Hydrocarbon biosynthetic activity of copurified OleBCD assemblies could be reconstituted by adding separately purified OleA. Immunoblots of nondenaturing gels using anti-OleC reacted with X. campestris crude protein lysate indicated the presence of a large protein assembly containing OleC in the native host. Negative-stain electron microscopy of recombinant OleBCD revealed distinct large structures with diameters primarily between 24 and 40 nm. Assembling OleB, OleC, and OleD into a complex may be important to maintain stereochemical integrity of intermediates, facilitate the movement of hydrophobic metabolites between enzyme active sites, and protect the cell against the highly reactive β-lactone intermediate produced by the OleC-catalyzed reaction.

IMPORTANCE Bacteria biosynthesize hydrophobic molecules to maintain a membrane, store carbon, and for antibiotics that help them survive in their niche. The hydrophobic compounds are often synthesized by a multidomain protein or by large multienzyme assemblies. The present study reports on the discovery that long-chain olefinic hydrocarbons made by bacteria from different phyla are produced by multienzyme assemblies in X. campestris. The OleBCD multienzyme assemblies are thought to compartmentalize and sequester olefin biosynthesis from the rest of the cell. This system provides additional insights into how bacteria control specific biosynthetic pathways.

INTRODUCTION

Long-chain olefinic hydrocarbons are abundant in nature. First described in 1929 in cabbage leaf extracts, plant C₂₉ ketones and hydrocarbons were proposed to arise from a head-to-head condensation of fatty acids, with a loss of one of the “head” carboxylic acid groups as carbon dioxide (1). Bacterial long-chain olefinic hydrocarbons were characterized structurally in 1969 during studies on the lipid components of Micrococcus luteus (2).

The genes and enzymes responsible for the head-to-head mechanism of hydrocarbon biosynthesis in bacteria were elucidated some 40 years later by the company LS9, Inc. (now Renewable Energy Group), and patented (3). A subsequent study in 2010 reported the presence of the four hydrocarbon biosynthetic genes, oleABCD, in 69 prokaryotes from many deeply rooted branches of the prokaryotic tree of life (Fig. 1) (4). The cellular function(s) of the ole gene cluster has been studied only in Shewanella oneidensis MR-1. The presence of the oleABCD genes was found to promote more rapid cell growth during a shift to colder temperatures, consistent with those genes being commonly found in cold-temperature bacteria from polar and marine environments (4–6). However, bacteria containing ole gene clusters inhabit a diversity of ecological niches, suggesting there may be a range of functions for long-chain olefinic hydrocarbons.

FIG 1.

Olefin biosynthesis enzymatic pathway and gene clusters. Gene clusters 1 and 2 are the two most common gene arrangements found in oleABCD-containing organisms. The two fatty acyl-CoAs are typically linear alkyl chains between 10 and 16 carbons in length.

Only recently have the activities of each Ole enzyme been assigned (Fig. 1), as initial work was hampered by the instability and hydrophobicity of the pathway intermediates. OleA is a soluble protein that condenses C₁₀-C₁₆ fatty acyl groups via a nondecarboxylative Claisen condensation reaction to form a β-keto acid (7). If the β-keto acid is not quickly reduced by OleD, it spontaneously decarboxylates in aqueous solution to a ketone, as observed in earlier studies (7, 8). Physiologically, OleD catalyzes an NADPH-dependent reduction of the β-keto acid intermediate to produce a β-hydroxy acid that is significantly more stable than the OleD substrate (9). Until recently, OleC was thought to catalyze the final reaction in olefin biosynthesis. However, OleC is now known to react with the product of the OleD reaction to generate a thermally labile β-lactone (10). The β-lactone undergoes spontaneous and nonbiological decarboxylation to an olefin when monitored by gas chromatography (GC), which led to the initial incorrect assignment of OleC function (10, 11). Recent data suggest that OleB acts as an unprecedented β-lactone decarboxylase to yield the final olefin product (10). Homologous gene clusters lacking an oleB gene have been identified in Streptomyces, and these produce β-lactone natural products rather than olefins (10). Microbial secondary metabolites containing β-lactones often serve as antibiotics and are known as general esterase and protease inhibitors (12, 13).

It was considered here that Ole enzymes might form a multienzyme complex to effectively process the hydrophobic, unstable, and potentially toxic intermediates of olefin biosynthesis. The presence of an oleBC gene fusion in actinobacterial ole gene clusters further suggested a physical interaction among the enzymes of olefin biosynthesis. In the present study, we report the purification of active protein assemblies consisting of OleB, OleC, and OleD proteins. Assemblies of OleBCD were obtained by recombinantly coexpressing all four ole genes from Xanthomonas campestris pv. campestris ATCC 33913 (X. campestris) in Escherichia coli with various affinity tag combinations. Nondenaturing gels immunoblotted with anti-OleC showed OleC from an X. campestris cell lysate migrated as a high-molecular-weight band, similarly to purified OleBCD assemblies. The results from size exclusion chromatography and electron microscopy were consistent with ordered assemblies of OleBCD, with an average molecular mass of ∼1.9 MDa.

RESULTS

Individual purification and physical properties of the four OleABCD enzymes.

The purification of individually expressed OleA, OleC, and OleD proteins derived from different bacteria had been reported previously (5, 7–11), while the purification of an OleB protein is reported here for the first time, to our knowledge. In this study, all four X. campestris Ole proteins were purified in a recombinant form from separate E. coli expression host cell lines. In contrast to conclusions reached in a previous study (11), we found that OleB is required for the reconstitution of hydrocarbon biosynthesis, consistent with the ubiquitous occurrence of oleB in the olefin gene cluster. The OleA and OleD reactions were previously demonstrated to produce a β-keto acid and β-hydroxy acid, respectively (Fig. 1) (7, 9). OleC, purified from Stenotrophomonas maltophilia, was reported to produce an olefin (11), but we recently demonstrated that OleC from that organism produces a β-lactone and that OleB is required to transform that to the final olefinic hydrocarbon (10).

Relevant physical properties of the Ole proteins are reported in Table 1. The oligomeric state of each protein was determined by gel filtration chromatography. However, detergents were required in the purifications of OleB and OleD, which could affect the observed oligomeric states. Moreover, all attempts to concentrate either protein resulted in immediate precipitation, even in the presence of detergents. Glycerol increased the purification yields of OleB, OleC, and OleD to the levels shown in Table 1 but did not boost the yield of OleA. Since the proteins all appeared to be in the soluble fraction of the cell, the insoluble nature of OleB and OleD in the absence of detergents suggested to us that Ole proteins might associate with each other in vivo.

TABLE 1.

Physical properties of the X. campestris Ole proteins

Property	Value or characteristic for:
	OleA	OleB	OleC	OleD
Subunit molecular mass (kDa)	36.6	34.1	58.5	36.1
Oligomeric state	Dimer	Tetramer	Monomer	Aggregatea
Required detergent		Triton X-100		Tween 20
Yield (mg/liter of culture)	15	1.9	5.8	1.1
Precipitated on concn?b	No	Yes	No	Yes

^aX. campestris OleD runs at >600 kDa on SEC but remains active and soluble. Stenotrophomonas maltophilia OleD is a dimer (9).

^bPrecipitation of the sample after spin concentration in 30,000-MWCO Amicon spin filters.

OleBCD form active multienzyme assemblies without OleA.

To examine our hypothesis that Ole enzymes form a multienzyme complex, we simultaneously expressed oleABCD in E. coli with different affinity tag arrangements (1 to 3) shown in Fig. 2A. Tag arrangement 1 contained an N-terminal His₆ tag on OleB and a C-terminal FLAG tag on OleC (His₆-OleBADC-FLAG). Even when no detergent was used, three clear SDS-PAGE bands corresponding to the size of OleC, OleA, and/or OleD, as well as OleB eluted from the Ni²⁺ column when the imidazole concentration reached ∼140 mM. OleA and OleD are nearly indistinguishable by SDS-PAGE, as they differ by only 508 Da. Subsequent purification over an anti-FLAG column further purified these three bands. However, the addition of myristoyl coenzyme A (myristoyl-CoA) substrate and cofactors generated no olefin product unless separately purified OleA was added back to the reaction mix. The reaction products of OleA with copurified OleBCD were found to be identical to our synthetic cis-olefin standard (see Fig. S1 in the supplemental material). These data suggested that the association between OleBCD survives the purification, whereas OleA is lost during chromatography. For comparison, the individually expressed OleA, OleB, OleC, and OleD were reconstituted and also shown to produce olefin from CoA-charged fatty acids. In both samples, the absolute configuration of the major product was cis-olefin. No appreciable difference in the rate of olefin production was observed, possibly because the rate-limiting step for in vitro assays is release of the hydrophobic olefin from OleB.

FIG 2.

Plasmid expression, purification, and SDS-PAGE analysis of Ole protein coexpression. (A) The table represents all tag combinations constructed. The placement of the His₆ and FLAG tags on the left and right of the protein corresponds to N- and C-terminal tags, respectively. Colors in the table represent the grouping of genes on a single vector for expression. (B) Purification of OleBCD (tag arrangement 4) over His affinity column with increasing amounts of imidazole. Peak fractions indicated by the bar were collected, concentrated, and loaded onto the anti-FLAG column (lane labeled “Load” on gel in panel C). (C) SDS-PAGE showing the post-His affinity column concentrate and the anti-FLAG purification of OleBCD tag arrangement 4. Bound samples were washed twice (W1 and W2) with buffer before elution with FLAG peptide. Note that OleC, the weakest-intensity band, contained the FLAG tag. Abs, absorbance.

To examine if the tag placement altered OleA's binding to the putative complex of OleBCD, expression of tag arrangements 2 and 3 was undertaken. Tag arrangement 2 (His₆-OleABDC-FLAG) resulted in the coelution of three bands at ∼100 mM imidazole from the Ni²⁺ column that were identified by mass spectrometry (MS) as OleB, OleC-FLAG, and OleD, while His₆-OleA (MS identified) eluted separately at 200 mM imidazole. The observation that some OleBCD was retained on the Ni²⁺ column until imidazole was added suggested an interaction with His₆-OleA, which was disrupted prior to His₆-OleA being competitively eluted off the column. Tag arrangement 3 (His₆-BACD) resulted in the coelution of OleBCD with no OleA activity. Taken together, these data suggest that OleBCD form active stable multienzyme assemblies with which OleA weakly associates.

Optimizing the copurification of OleBCD.

To obtain pure assemblies of OleBCD routinely, OleA was no longer coexpressed with OleBCD; only oleBCD genes were expressed in the E. coli host. Of the three additional tag arrangements tested (4 to 6, Fig. 2A), only the purification of tag arrangement 4, His₆-BDC-FLAG, yielded active assemblies after Ni²⁺ and anti-FLAG columns. In contrast to observations made with OleB and OleD proteins purified independently, the copurified OleBCD did not require detergents and could be concentrated without precipitation. A His₆ tag on OleD (either N- or C-terminal, tag arrangements 5 and 6) appeared to prevent formation of the enzyme assemblies. The purification of tag arrangement 4 is shown in Fig. 2B and C. Glycerol was found to increase purification yields but was not required to obtain active assemblies. The individual components of this OleBCD complex were confirmed by MS, and this sample was used in further characterization studies.

OleBCD complex identification in native X. campestris.

The level of expression of olefinic hydrocarbons in wild-type bacteria is modest (4, 5), and so we used sensitive immunoblotting methods and nondenaturing gels to identify the anticipated multienzyme assembly within the native X. campestris host. A polyclonal antibody against X. campestris OleC was raised against a specific peptide predicted to be on the protein surface, as described in Materials and Methods. The polyclonal antibody was then validated by showing binding activity in immunoblots against purified OleC, copurified OleBCD, and lysates of E. coli expressing OleBCD. Figure 3 shows a nondenaturing gel that had been immunoblotted in which a lysate fraction derived from X. campestris cells is compared with copurified OleBCD and OleC alone. The band from the supernatant of X. campestris lysate migrates similarly to the OleBCD assemblies and more slowly than OleC alone. These data are consistent with the idea that the assemblies of OleBCD that form in E. coli likely assemble in a manner similar to that in the native host, X. campestris. It was also found in the soluble fraction, similar to recombinant expression in E. coli.

FIG 3.

Anti-OleC immunoblot on a nondenaturing protein gel. Native OleC from the supernatant of lysed X. campestris cells migrates closely with recombinantly expressed and purified X. campestris OleBCD rather than individually purified OleC. Xc, X. campestris.

Size estimation of OleBCD assemblies by gel filtration chromatography.

Gel filtration chromatography was used to obtain insight into the molecular size of copurified OleBCD. Ni²⁺ column-purified OleBCD (tag arrangement 4) eluted in the void of a GE 16/600 Superdex 200 gel filtration column, indicating a size larger than 600 kDa. The void fraction was collected and found to be active upon addition of OleA, cofactors, and myristoyl-CoA. We subsequently attempted to purify OleBCD assemblies with detergents to see if a minimal functional complex could be isolated. However, OleBCD still ran in the void with the addition of Tween 20 or Triton X-100, while 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), a zwitterionic detergent, led to complete dissociation of OleBCD at all concentrations.

Subsequently, OleBCD was analyzed on a Sephacryl gel filtration column that is able to separate globular proteins up to 8 MDa. OleBCD (tag arrangement 4) eluted significantly after the void (45 ml) at a volume of 65.0 ml, consistent with a high molecular weight (Fig. 4). The observed peak lies outside the range of commercially available protein standards (up to 600 kDa), but a molecular mass centered at ∼1.9 MDa was determined by extrapolation. The activity of each fraction was assayed in the presence of excess OleA and activity tracked precisely with the A₂₈₀, consistent with stable active assemblies of OleBCD (Fig. 4). SDS-PAGE analysis of fractions showed peak OleBCD protein staining intensity consistent with the UV absorbance and activity data.

FIG 4.

Size exclusion chromatography of Ni²⁺ column-purified OleBCD (tag arrangement 4). The peak absorbance at 65.0 ml corresponded to an estimated size of 1.9 MDa when extrapolated from protein standards. Two hundred microliters of each 4-ml fraction was tested for OleBCD activity by the addition of OleA, myristoyl-CoA, and cofactors.

Electron microscopy imaging of OleBCD assemblies.

The suggested size based on gel filtration prompted analysis of OleBCD by transmission electron microscopy (TEM) using negative staining. This required purification of OleBCD (tag arrangement 4) without glycerol using Ni²⁺, anti-FLAG, and gel filtration columns to obtain a highly purified sample. Glycerol increases the purification yield of OleBCD assemblies but prevents sample dehydration for effective TEM imaging. Multiple purifications of OleBCD produced very similar electron micrographs, one of which is shown in Fig. 5.

FIG 5.

Electron micrograph of OleBCD assemblies with particle size analysis. (A) OleBCD assemblies by TEM. (B) Histogram analysis of >150 particles from four different frames, with representative micrographs shown below. Measurement was of the greatest particle diameter. Average assembly size is 30.3 nm, and the median is 29.6 nm.

TEM analysis of the negatively stained OleBCD revealed distinct assemblies for OleBCD, primarily ranging in size from 24 to 40 nm in diameter. The annular hexagonal assemblies with an average diameter of ∼27 nm are particularly striking for their organized structure (Fig. 5 and S2). Some of the sample heterogeneity may be different orientations of the same structure on the TEM grid, different size complexes, assembly intermediates, or the result of the dehydration and negative-staining protocol required for TEM. We note that size exclusion chromatography (SEC) does show that OleBCD activity is associated with a broad peak (Fig. 4). The relationship between the assemblies pictured in Fig. 5 remains unclear, but further work is under way to obtain more high-resolution structural information by cryo-electron microscopic (cryo-EM) methods.

Estimated stoichiometry of the OleBCD assemblies.

To estimate the stoichiometry of each of the OleB, OleC, and OleD proteins within the combined assemblies, gel band intensities of the purified OleBCD, shown in Fig. 2C, were compared to band intensities of a gel standard curve made with each purified protein. Fig. S3 shows equimolar amounts of separately purified and quantified OleB, OleC, and OleD by SDS-PAGE, as determined by Bradford assay and UV₂₈₀. Using ImageJ, the intensities of the bands were plotted against protein concentration to generate standard curves for each of the Ole proteins (Fig. S4). Comparing the band intensities of the copurified OleBCD assemblies (Fig. 2C) with these standard curves revealed a stoichiometry of nearly 4:1:4 of OleB:OleC:OleD within the combined assemblies. Analysis of two other gels gave similar results. Taking this result with the ∼1.9-MDa size as assigned by SEC, this would be most consistent with a single complex composed of 24, 6, and 24 subunits of OleB, OleC, and OleD, respectively. However, the broadness of the SEC peak and the various sizes of the particles observed by cryo-EM might also be consistent with other multiples of the 4:1:4 stoichiometry. It is also possible that a mixture of different component combinations (perhaps assembly intermediates) led to the overall observed 4:1:4 stoichiometry. However, olefin production upon the addition of OleA, cofactors, and substrate occurs in all fractions following SEC, and activity tracks with the amount of protein in the fraction (Fig. 4). This indicates that OleB, OleC, and OleD are present in all fractions, although they may not be in the same proportions. We note that the observed correlation between activity and total protein indicates only that the rate-determining step of olefin biosynthesis correlates with total protein.

DISCUSSION

Many biosynthetic pathways are assembled as a single large protein with multiple domains, for example, polyketide synthases and type I fatty acid synthases (14, 15). In other cases, individually expressed proteins self-assemble in the cytosol, for example, the lumazine synthase-riboflavin synthase protein complex (16, 17). The structure and function of a number of polyketide synthase and fatty acid synthase systems have been studied in significant detail (18–20). In contrast, the higher-order structure of hydrocarbon biosynthetic machinery has been far less characterized.

In the present study, the OleABCD enzymes from X. campestris were expressed and purified individually and shown to reconstitute long-chain olefin biosynthesis. Coexpression studies demonstrated that OleBCD forms active stable assemblies. While hundreds of bacteria are thought to produce long-chain head-to-head hydrocarbons, X. campestris has been used as a model system because the enzymes are relatively stable and have been shown to possess broad substrate specificity, making them suitable for bioengineering (4). OleA purifies as a stable soluble dimer, and the structure and mechanism of this component have been the subject of several studies (7, 21, 22). OleA catalyzes the Claisen condensation of fatty acyl-CoA substrates, producing a β-keto acid that feeds into the OleBCD enzymes (7).

There are several reasons that multienzyme OleBCD assemblies might be functionally superior to individual OleB, OleC, and OleD proteins operating independently within a bacterial cell. First, a multienzyme assembly of Ole proteins could protect the cell by sequestering the reactive Ole pathway β-lactone intermediate and preventing a nonspecific reaction with cytosolic proteins. The β-lactone moiety is common in microbial natural products and can function as an antibiotic by reacting with active-site residues in essential esterase and protease enzymes (12, 13). The β-lactone tetrahydrolipstatin from Streptomyces toxytricini is very similar in structure to the β-lactone produced by OleC, and its biosynthesis is encoded by an ole-like gene cluster that lacks the oleB decarboxylase gene (10, 12). Tetrahydrolipstatin is a potent human pancreatic lipase inhibitor and is the only FDA-approved over-the-counter antiobesity drug. Since the β-lactone of olefin biosynthesis is only an intermediate, the sequestration of a potentially harmful compound by an enzyme complex likely affords sufficient protection to the bacterium. It is also relevant that oleB and oleC are a gene fusion in all actinobacteria, which comprise approximately 30% of known ole-gene-containing organisms. It is unknown if assemblies of OleBCD form in these bacteria, as the enzyme fusion should allow the direct transfer of the β-lactone from the OleC to OleB domains for decarboxylation. Fusion creates a 1:1 stoichiometry between OleB and OleC, but our protein quantitation for the X. campestris multienzyme assemblies suggests a 4:1 relationship, respectively, suggesting there may be different configurations between species.

An additional benefit of an OleBCD complex would be to promote efficient trafficking of the highly hydrophobic intermediates that comprise the Ole pathway from one active site to another. It is common for the multistep biosynthesis or degradation of hydrophobic compounds to be accomplished by either membrane-bound enzymes or enzyme complexes (14, 15). However, sequence analysis of Ole proteins did not predict any transmembrane domains or lipid anchor sequences. Moreover, the enzymes have been observed in the soluble protein fraction of the native host X. campestris and in the recombinant host E. coli. An example of a well-characterized bacterial complex producing and transferring hydrophobic intermediates is the fatty acid synthesis (fatty acid synthase type I [FAS I]) machinery from Mycobacterium tuberculosis (23). Cryo-EM studies of FAS I from M. tuberculosis revealed a 2.0-MDa homohexameric enzyme complex with a diameter of 25 nm and the same hexagonal symmetry observed in our annular assemblies by TEM of the OleBCD assemblies. However, the FAS I complex uses an acyl carrier protein within a central cavity to ferry enzyme-tethered intermediates between active sites for iterative fatty acid elongation reactions. While a central cavity or porous channeled structure is possible based on the size and molecular weight of the OleBCD assemblies, there is no evidence for any enzyme-tethered intermediate shuttle in the Ole biosynthetic pathway at this time.

In summary, we propose the following model for long-chain olefin biosynthesis in X. campestris (Fig. 6). OleA receives fatty acyl-CoA products from the fatty acid biosynthetic machinery, acting as a shuttle between that complex and an OleBCD complex. It is likely that OleA has some binding affinity for an OleBCD complex, as suggested by assemblies of OleBCD showing weaker affinity for His₆-OleA on the Ni²⁺ column. However, this interaction is not sufficiently strong to survive passage through several chromatographic steps. The current data support the idea that the consecutively catalyzed OleD-OleC-OleB reactions occur sequestered within multienzyme structures.

FIG 6.

Model for the biosynthesis of long-chain head-to-head olefins in X. campestris showing that OleA acts predominantly as a soluble cytosolic dimer, condensing acyl-CoAs, docking, and providing its reaction product to the OleBCD complex, which ultimately releases a long-chain olefin.

The product olefin is ultimately found in the membrane fraction, and the mechanism of transfer from the soluble OleBCD assemblies to the membrane remains to be elucidated. Further studies by cryo-EM will help reveal more details of the OleBCD assembly architecture and the relationship between the component enzymes. Long-chain hydrocarbon production has been confirmed experimentally in at least 20 different bacteria (2, 4, 6–9), and sequence analysis reveals oleABCD gene clusters in hundreds of other bacteria, suggesting that multienzyme OleBCD structures may be widespread among microorganisms.

MATERIALS AND METHODS

Chemicals.

All reagents were purchased from Sigma-Aldrich, St. Louis, MO, with the following exceptions. NADPH was purchased from EMD Millipore, Billerica, MA. Glycerol was purchased from IBI Scientific, Peosta, IA. Synthetic cis- and trans-olefins were prepared as described previously (10).

Cloning, expression, and purification of Ole proteins.

All protein sequences for ole genes used in this study were from X. campestris pv. campestris ATCC 33913 (OleA, accession no. NP_635607.1; OleB, accession no. NP_635611.1; OleC, accession no. NP_635613.2; and OleD, accession no. NP_635614.1). E. coli codon-optimized genes were ordered from DNA2.0, Inc. Polymerase, restriction enzymes, ligase, and supercompetent cells were purchased from New England BioLabs and used according to the manufacturer's instructions. The four vector backbones, pET28b⁺, pET30b⁺, pCOLA, and pCDF, were obtained from Novogen. A complete list of constructs, gene insertion sites, and antibiotic selectable markers from this study can be found in Table S1 in the supplemental material.

The following general expression and purification protocols were developed for all Ole enzyme configurations. Plasmids were transformed into E. coli BL21(DE3) cells (Invitrogen) and plated on Luria-Bertani (LB) agar with appropriate antibiotic selection (50 μg/μl). Single colonies were selected and grown at 37°C in 5 ml of LB overnight as starter cultures. LB medium (1.0 liter) containing antibiotic was inoculated with a 5-ml starter culture and grown at 37°C until an optical density at 600 nm (OD₆₀₀) of 0.50 was reached. Protein expression was then induced with 100 μM isopropyl β-d-1-thiogalactopyranoside (IPTG), and growth was shifted to 16°C overnight. Cells were harvested by centrifugation and resuspended in 5 ml of buffer containing 200 mM NaCl, 20 mM HNa₂PO₄ (pH 7.4), and 10% glycerol (buffer) per liter of LB medium. Sonication or French pressure cell lysis was used to disrupt the cells, followed by centrifugation at 35,000 × g. The supernatant liquid was filtered through 0.45-μm- and 0.22-μm-pore-size filters (Millipore) and loaded onto a GE HisTrap HP 5-ml column (His column). Gradient elution in buffer with 500 mM imidazole (pH 7.4) was achieved using an Äkta fast protein liquid chromatography instrument (General Electric Company). Ole proteins typically eluted around 200 mM imidazole. Coexpressed protein mixtures containing a DYKDDDDK (FLAG) tag were then manually loaded onto a column containing 2.5 ml of anti-FLAG M2 resin (Sigma-Aldrich) (anti-FLAG column). The anti-FLAG column was washed with 8 column volumes of buffer before eluting with 10 ml of 100 μg/ml FLAG peptide. The protein in the solution was concentrated using 30,000-molecular-weight-cutoff (30,000-MWCO) spin filters (Amicon) and frozen at −80°C for later use.

The following protein-specific modifications were made to the above-mentioned protocol. Whenever buffer additives were required, as described below, they were included for all purification steps. For OleBCD assemblies, vector combinations, shown in Fig. 2A, were transformed into E. coli under both kanamycin and streptomycin selection. No detergents were needed during purification of Ole assemblies, but 10% (vol/vol) glycerol was found to increase protein yield. For electron microscopy experiments, glycerol was omitted from the purification buffer. For OleA, after IPTG induction, cells containing pET28b⁺ oleA with an N-terminal His₆ tag were grown at 37°C for 4 h before purification in buffer. For OleB, buffer required 0.05% (wt/vol) Triton X-100 (critical micelle concentration [cmc], ∼0.02% [wt/vol]) for purification of pET28b⁺ expressed OleB with a His₆ N-terminal tag. Triton X-100 interfered with A₂₈₀ measurements, necessitating that protein elution be followed by the Bradford assay and protein gels (24). Protein could not be concentrated without precipitation. For OleC, an N-terminal His₆ tag on OleC does not purify, necessitating the C-terminal His₆ tag of pET30b⁺ for expression and purification. For OleD, buffer containing 0.02% (wt/vol) Tween 20 (cmc, ∼0.007% [wt/vol]) was required to purify OleD with an N-terminal His₆ tag expressed from a pET28b⁺ vector. Protein could not be concentrated without precipitation.

Olefin activity assays.

Ole enzyme activity assays were conducted in glass vials at room temperature with 500 μl of buffer (200 mM NaCl, 20 mM NaPO₄ [pH 7.4]). Reactions were conducted in duplicate and contained 10 μg of OleA and 30 μg of copurified OleBCD, as measured by the Bradford assay. Assays involving separately purified OleA, OleB, OleC, and OleD contained 10 μg of OleA, 12.1 μg of OleB, 5.1 μg of OleC, and 12.8 μg of OleD to mimic the putative 4:1:4 B:C:D stoichiometry of 30 μg of copurified OleBCD. ATP, NADPH, and MgCl₂ were added to final concentrations of 1 mM, 500 μM, and 1 mM, respectively, in all assays. The slight difference in buffer additives to Ole proteins was ignored, as they were found not to significantly affect reactions. Reactions were initiated by the addition of myristoyl-CoA to a concentration of 50 μM and quenched with 500 μl of ethyl acetate at appropriate time points. Five microliters of organic phase was removed and analyzed by GC, using both a flame ionization detector (FID) and a mass spectrometry (MS) detector, as previously described (10). The program conditions were as follows: start temperature, 100°C; ramp rate, 10°C/min; final temperature, 320°C; hold time, 5 min; and total time, 27 min. A synthetic olefin standard, cis-13-heptacosene, was synthesized as described previously (10). The retention time of cis-13-heptacosene was 18.26 min.

Mass spectrometry protein identification.

Ole proteins were separated using 12.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Bands were excised, trypsin digested, and submitted to the University of Minnesota Center for Mass Spectrometry and Proteomics for peptide identification using liquid chromatography-mass spectrometry (25). Peptides were initially separated by a Paradigm Platinum Peptide Nanotrap (Michrom Bioresources, Inc.) precolumn, followed by an analytical capillary column (100 μm by 12 cm) packed with C₁₈ resin (5 μm, 200 Å MagicC18AG; Michrom Bioresources, Inc.). Mass spectrometry was performed on an LTQ (Thermo Electron Corp., San Jose, CA).

Immunoblots for OleC in native X. campestris lysate.

Polyclonal antibodies against an X. campestris OleC peptide (AIDDAAIPEWSGVR) were raised by GenScript in rabbit. A secondary horseradish peroxidase-conjugated donkey anti-rabbit IgG antibody was purchased from Jackson ImmunoResearch Laboratories. X. campestris (ATCC 33913) for immunoblots was grown in 250 ml of LB medium at 30°C before harvesting and sonication generated the X. campestris lysate. OleC and OleBCD (Fig. 2A, tag arrangement 4) were purified as described above. Native gels were made as described by Nadano et al. (26) and contained 1.9% polyacrylamide and 0.75% agarose in 40 mM Tris-acetate buffer at pH 8.4. Protein was mixed with loading buffer (15% sucrose in Tris-acetate) and run at 100 V. Gels were transferred to a nitrocellulose membrane and washed thoroughly with 200 mM NaCl, 20 mM NaPO₄ (pH 7.4), and 0.1% Tween 20 as a blocking agent. Membranes were incubated with the primary antibody overnight, washed, and incubated with the secondary antibody for 1 h. ECL Plus Western blotting substrate (Pierce) and Classic Blue autoradiography film BX (MidSci) were used to visualize the membrane.

Gel filtration chromatography.

A HiLoad 16/600 Superdex 200 (GE; normal column) and a HiPrep 16/60 Sephacryl S-400 HR (GE; ultrahigh-molecular-weight [ultrahigh-MW] column) were used to separate the proteins and determine molecular weight. The normal small-size column can separate globular proteins up to 600 kDa, while the ultrahigh-MW column can separate globular proteins up to 8.0 MDa. A gel filtration high-MW (HMW) calibration kit was purchased from GE to calibrate the gel filtration columns (up to a molecular mass of 669 kDa), with higher molecular masses being extrapolated. Typically, 2.0 mg of protein was loaded onto columns and run according to the manufacturer's specifications.

Electron microscopy of OleBCD assemblies.

Following His affinity column, anti-FLAG, and gel filtration purification, OleBCD assemblies were characterized using transmission electron microscopy (TEM). Grids (400-mesh Cu-carbon grids) were glow discharged at 15 mA and 39 Pa for 1 min using a Pelco easiGlow glow discharge cleaning system (Ted Pella, Inc., Redding, CA). Samples containing 0.08 mg/ml OleBCD were drop-cast onto these grids and blotted with filter paper to remove excess sample. Grids were washed twice with deionized water and blotted dry. Uranyl formate (0.05%) stain was added to each grid and blotted a final time. Negative-stain micrographs were collected using a field emission gun (FEG) Tecnai G2 F30 TEM (FEI, Hillsboro, OR). Images were recorded at room temperature on a Gatan 4k by 4k charge-coupled-device (CCD) camera (Gatan, Inc., Pleasanton, CA). ImageJ was used to analyze the sizes of >150 particles in four TEM images. The size was recorded as the greatest diameter of the particle.

SDS-PAGE standard curve.

Quantification of staining intensities of the Ole proteins on SDS-PAGE was carried out using individually purified Ole proteins, as described above. Concentrations of OleB, OleC, and OleD were determined by the Bradford assay and supported by UV absorbance at 280 nm (A₂₈₀). OleB required Triton X-100 to prevent protein precipitation and so could not be verified by A₂₈₀. A 4.0 μM master mix of the three proteins was created and loaded in different volumes into wells of a 12.5% SDS-PAGE gel. Gels were stained with SimplyBlue SafeStain (Life Technologies), and images were analyzed using ImageJ.