Production of 10-methyl branched fatty acids in yeast

Background Despite the environmental value of biobased lubricants, they account for less than 2% of global lubricant use due to poor thermo-oxidative stability arising from the presence of unsaturated double bonds. Methyl branched fatty acids (BFAs), particularly those with branching near the acyl-chain mid-point, are a high-performance alternative to existing vegetable oils because of their low melting temperature and full saturation. Results We cloned and characterized two pathways to produce 10-methyl BFAs isolated from actinomycetes and γ-proteobacteria. In the two-step bfa pathway of actinomycetes, BfaB methylates Δ9 unsaturated fatty acids to form 10-methylene BFAs, and subsequently, BfaA reduces the double bond to produce a fully saturated 10-methyl branched fatty acid. A BfaA-B fusion enzyme increased the conversion efficiency of 10-methyl BFAs. The ten-methyl palmitate production (tmp) pathway of γ-proteobacteria produces a 10-methylene intermediate, but the TmpA putative reductase was not active in E. coli or yeast. Comparison of BfaB and TmpB activities revealed a range of substrate specificities from C14-C20 fatty acids unsaturated at the Δ9, Δ10 or Δ11 position. We demonstrated efficient production of 10-methylene and 10-methyl BFAs in S. cerevisiae by secretion of free fatty acids and in Y. lipolytica as triacylglycerides, which accumulated to levels more than 35% of total cellular fatty acids. Conclusions We report here the characterization of a set of enzymes that can produce position-specific methylene and methyl branched fatty acids. Yeast expression of bfa enzymes can provide a platform for the large-scale production of branched fatty acids suitable for industrial and consumer applications.


Background
Biobased lubricants account for only a small percentage of the total global lubricant supply [1,2] despite environmental and sustainability benefits for their use. A major barrier to increased biobased lubricant utilization is that they lack the proper combination of viscosity, low pour point and high oxidation resistance to match petroleum-based and synthetic oils. Most natural fatty acids achieve lower melting temperatures through desaturation, but this leaves them prone to oxidative degradation. Linear saturated long-chain fatty acids with appropriate viscosities have high melting temperatures unsuitable for most lubricant applications. To date, the biobased fatty acid most closely matching the chemical properties of an ideal lubricant is a product termed "isostearic" acid, which is a mixture of branched fatty acids (BFAs) methylated at various positions on the stearate chain [3,4]. This non-specificity of branch point leads to mixed chemical properties. Isostearic acid is produced by chemical transformation of unsaturated fatty acids, a process that also yields significant amounts of polymer byproducts [4], limiting the production volume and profitability of isostearic acid.
A group of naturally occurring fatty acids achieves lower melting temperatures while maintaining oxidative stability via carbon chain branching on saturated fatty acids. For example, many bacteria including Escherichia coli produce cyclopropane fatty acids (CFAs), which have a low melting temperature [5], but the cyclopropane ring is unstable. Iso and anteiso methyl branched fatty acids are produced by Bacillus subtilis [6], but the methyl branches located close to the end of the carbon chain do not substantially lower the melting temperature [7]. Fatty acids with random branched sites suitable for biodiesel have been produced from branched-CoA intermediates in E. coli and P. pastoris by combining the expression of genes from fatty acid and valine and isoleucine biosynthetic pathways [8], however, through this strategy the branch position cannot be controlled. A good candidate for a well-defined, biobased lubricant is 10-methyl stearate. This fully saturated, long-chain fatty acid achieves a low melting temperature (13.2 °C) due to the methyl branch that occurs close to the middle of the fatty acid chain [9]. 10-Methyl fatty acids occur naturally in small amounts in the membranes of a few bacterial species. 10-Methyl stearate, or tuberculostearic acid, was first observed in mycobacterium, and later confirmed to exist in a variety of actinobacteria [10,11]. 10-Methyl palmitate has also been detected in a subset of γ-proteobacteria [12][13][14]. Large-scale production of 10-methyl BFAs has been hampered by the fact that these organisms do not produce large amounts of BFAs and are not amenable to commercial production.
The two-step enzymatic production of 10-methylstearic acid was defined through detailed biochemical analysis of Mycobacterium phlei [15]. First, a methylase transfers a methyl group from S-adenosylmethione (SAM) to the C10 of a phospholipid-bound oleic acid molecule [16]. The reaction is similar to that catalyzed by cyclopropane fatty acid synthase (cfa) enzymes, except the product is a C10-methylene rather than a C9,10 cyclopropane. In a second step, unique to the bfa pathway, a reductase saturates the methylene double bond, resulting in the production of 10-methylstearic acid. Although the chemical mechanism of this reaction was proposed 50 years ago [15], the genes responsible for BFA production had not been cloned. Two reports claimed to have identified genes responsible for the production of 10-methylstearic acid [17,18], but neither gene contained any redox activity to carry out the second step in the production pathway. A recent report identified a two-gene pathway in mycobacteria for the production of 10-methylstearic acid [19] and these genes were functional in cyanobacteria [20].
For large-scale production of branched fatty acids, the oleaginous yeast Yarrowia lipolytica and the industrial workhorse Saccharomyces cerevisiae are promising host organisms. Y. lipolytica naturally accumulates fatty acids in triacyglycerols (TAGs) and has been previously engineered to reach very high lipid contents in bioreactors [21,22]. Methyl branched fatty acids are produced from an oleic acid precursor, which is highly abundant in engineered Y. lipolytica strains [23]. S. cerevisiae is a well-known industrial yeast with an extensive literature base and genetic toolkit for advanced engineering. Both Y. lipolytica and S. cerevisiae have been engineered for enhanced free fatty acid production and secretion [24][25][26], which could aid the recovery of BFAs.
To enable the production of 10-methyl BFAs, we identified and characterized enzymes that produce site-specific BFAs. Although the major products in yeast are 10-methyl C16 and C18 BFAs, the enzymes methylated substrates with a range chain-lengths and unsaturation sites, expanding the potential product range. Using the bfa genes from Thermomonospora curvata, we found that fusion of BfaA and BfaB into a single protein increased the production of 10-methyl BFAs. We demonstrated secretion of BFAs in S. cerevisiae and accumulation of high levels of BFAs in Y. lipolytica.

Gene identification
Bfa and Tmp gene identification began with a review of biochemical literature describing bacterial species with 10-methyl branched fatty acids. A list of representative strains was generated whose genomes were sequenced (Additional file 1: Tables S1 and S2), and the genomes were queried for genes sharing homology to the E. coli cyclopropane fatty acid synthase via the GenBank and Metacyc [27] databases. Gene candidates were selected based on their restricted presence in 10-methyl fatty acid producing species and adjacency to a likely redox-active gene. A list of protein sequences and alignments is available in Additional file 3. Candidate gene operons were then cloned in E. coli and whole cell biomass assayed for the presence of non-native branched fatty acids.

Escherichia coli expression vector construction
Bacterial genomic DNA was obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ), Germany. Plasmids were constructed with standard molecular biology techniques using the "yeast gap repair" method [28]. For E. coli expression vectors, the empty expression vector pNC53 was restriction digested with enzyme PmeI (New England Biolabs, MA), creating a double strand break between the tac promoter and trpT' terminator sequences on this vector. bfa gene operons were PCR amplified from genomic DNA with primer flanking sequence such that the bfaB ATG start site integrated into the end of the tac promoter via homologous recombination. The stop codon of the bfaA gene similarly integrated into the beginning of the trpT' terminator region. E. coli translation of the operon-embedded bfaA gene relied on native translation signals from the donor organism DNA. Where necessary, the first codon of bfaB was altered from GTG or TTG to ATG; otherwise the native codon sequence was kept in the E. coli expression vectors. Vectors were checked by DNA sequencing and restriction digest for correct construction. All plasmids used in this study are listed in Additional file 1: Table S4.

Exogenous fatty acid supplementation in E. coli
Unsaturated fatty acids were purchased from Nu-Check Prep, Inc., Elysian MN. Fatty acids were dissolved in DMSO at a concentration of 100 mg/mL, with the exceptions of palmitoleic acid, oleic acid, and vaccenic acid, which were dissolved in ethanol at the same concentration. A 10-methylstearic acid reference standard was obtained from Larodan AB, Sweden. Preliminary tests were performed with E. coli Top10 (Invitrogen) for evaluation of bfa operon expression vectors. Initial screening for active bfa and tmp operons was performed in 50 mL LB medium supplemented with 100 mg/L ampicillin and 100 mg/L oleic acid at 37 °C and 200 rpm in baffled shake flasks for 41 h. For evaluating conversion of other unsaturated fatty acids, E. coli strains NS1161 (empty vector), NS1162 (T. curvata bfaAB), NS1237 (M. hydrocarbonoclasticus tmpBA), and NS1238 (T. halophila tmpBA) were used. NS1161 was constructed by transforming the control plasmid pNC53 into E. coli CGSC 9407 (aka JW1653-1 Keio collection) which has a kanR disruption of the native E. coli cyclopropane fatty acid synthase (cfa) gene. Strains NS1162, NS1237, and NS1238 were constructed in a similar manner by transforming plasmids pNC906, pNC1074, and pNC1076 containing the bfaAB operon from T. curvata and tmpBA operons from M. hydrocarbonoclasticus, and T. halophila, respectively. E. coli strains were grown in LB media supplemented with 100 mg/L ampicillin and 100 mg/L of unsaturated fatty acid with a 5-mL working volume at 37 °C in a rotary drum roller for 24 h. All strains used in this study are described in Additional file 1: Table S5.

Fatty acid compositional analysis
Escherichia coli and yeast cells were harvested by centrifugation, targeting 2-20 mg of lipid in the cell pellet (Tables 1, 2, Additional file 1: Table S3, Additional file 2: Figure S1B, Fig. 5) or the entire volume from the 96-well plate assay (Figs. 2, 3 and 5). Cells were washed twice with 1 mL of deionized water, resuspended in 100 μL deionized water, and frozen at − 80 °C. For Tables 1, 2 and Fig. 2a, cells were lyophilized to dryness and subjected to acid-catalyzed transesterification using 1.25 M hydrochloric acid in methanol (Sigma) at 85 °C for 90 min with mixing by vortexing at 30 min and 60 min to produce fatty acid methyl esters (FAMEs). During the course of this study, we identified acid-catalyzed degradation products arising from 10-methylene fatty acids, similar to that reported earlier for cyclopropane fatty acids [29,30]. We later determined that base (sodium methoxide) catalyzed transesterification did not cause degradation, and this method was used for all other analyses. In the base-catalyzed transesterification 250 μL of 0.5 M sodium methoxide in methanol (Acros Organics) was added to the cell pellets and incubated at 50 °C for 30 min with mixing by vortexing at 15 min. 250 μL of 0.6 N hydrochloric acid in water and 1 mL isooctane were then added to each sample and mixed by pipetting. The tubes or plate were centrifuged at 2500×g at 22 °C to separate the organic and aqueous layers. A sample of the FAME-containing isooctane layer (top layer) was analyzed by gas chromatography equipped with a flame ionization detector (GC-FID) (Agilent Technologies 7890B GC) and VF-23 ms capillary column (20 m × 0.15 mm × 0.15 µm, Agilent Technologies). The FAME-containing sample was injected in the liner at 335 °C with a split ratio of 20:1. The column flow was constant at 0.35 mL/min He. The initial oven temperature was 120 °C, which was then raised to 130 °C at 5 °C/min, and then to 220 °C at 6 °C/min, finally to 240 °C at 40 °C/min and held for 1.5 min. The compounds of interest were identified by comparison of retention times with authentic standards.

Methylene fatty acid identification
Yarrowia lipolytica strain NS1165, expressing T. curvata bfaB under control of the Y. lipolytica TEF1 promoter, and the parental strain NS1009 were cultured in bioreactors before harvesting, washing, freezing, and lyophilization to dryness followed by lipid extraction by bead milling in chloroform-methanol. Extracted material was analyzed by 13 C nuclear magnetic resonance (NMR), and by GC-FID after sodium methoxide catalyzed transesterification.

BfaA co-factor assay
Escherichia coli strains NS1161, NS1163, and NS1164 were used in this experiment. Strain NS1163 was constructed by transforming plasmid pNC963, containing the T. curvata bfaB gene under control of the constitutive tac promoter, into E. coli CGSC 9407. Strain NS1164 was constructed by transforming plasmid pNC964, containing the T. curvata bfaA gene under control of the constitutive tac promoter, into E. coli CGSC 9407.
Strain NS1163 was grown in 2 × 500 mL LB media in 2 L baffled flasks supplemented with 100 mg/L ampicillin for 24 h at 37 °C. After cultivation, cells were harvested by centrifugation at 3000×g for 15 min in an Eppendorf 5810 R clinical centrifuge and washed twice in 100 mL PBS buffer. After concentration to 40 mL PBS buffer, cells were heat inactivated at 85 °C for 30 min. Inactivated cells were then dispensed into 1 mL aliquots and disrupted with 0.3 g of 0.1 mm glass beads using a MP fastprep-24 on "E. coli" setting (MP biomedicals, LLC). Whole cell lysed suspension was collected by micro-centrifugation at 2000×g for 30 s to remove beads and then 0.7 mL of suspension per tube was transferred to new tubes and frozen at − 80 °C until further use.
On the day of assay, strains NS1161 and NS1164 were grown via inoculation from overnight cultures (1:1000 dilution) in 50 mL LB medium supplemented with 100 mg/L ampicillin in 37 °C and 200 rpm in baffled shake flasks. After 4 h of cultivation, cells were harvested at 5 °C, washed 1× in ice cold PBS and then resuspended in 750 μL PBS in 1 mL plastic screw tubes. 0.3 g of 0.1 mm glass beads were added, and cells were lysed with a MP fastprep-24 on the "E. coli" setting. The cell suspension was then micro-centrifuged for 5 min at 12,000×g, and the supernatant transferred to a fresh tube and held on ice until assay. The 1 mL reaction contained 700 μL of NS1163 whole lysate with 10 mM NADPH, 10 mM NADH solution, and 100 μL of cell free extract as dictated by the assay conditions. Omitted components were replaced by PBS buffer. Assay tubes were sealed and rotated on a drum roller at 37 °C for 16 h. To end the assay, tubes were frozen at − 80 °C, then lyophilized to dryness followed by in situ extraction and transesterification with methanolic HCl. Fatty acid profiles were determined by GC with flame ionization detection, and the 10-methyl fatty acid peak area was compared to the total fatty acid peak area to determine assay activity.

Yarrowia lipolytica strains and cultivation
All Y. lipolytica strains are modifications of the wild-type strain YB-392 from the ARS collection (https ://nrrl.ncaur .usda.gov) and are listed in Additional file 1: Table S5. Gene deletions (tgl3 and fad2) were carried out similar to our previous work [31]. Targeting to the designated locus was increased by treating the cells with hydroxyurea to Table 1 BFA production in bfaAB and tmpBA expressing E. coli E. coli Top10 cells containing the indicated plasmid were cultivated in LB medium supplemented with 100 mg/L oleic acid (for bfaAB operons) or 100 mg/L palmitoleic acid (for tmpBA operons) and antibiotic. Percent conversion was measured by dividing the 10-methyl/10-methylene fatty acid content by the sum of the remaining precursor fatty acid and the 10-methyl/10-methylene fatty acids present in E. coli cell mass synchronize the cells in S phase, using the local promoter (TGL3 or FAD2) to drive the NAT marker, which reduces the selection frequency of random integration events, and adding 2 kb of homology downstream of the targeted gene. To allow for marker removal, this cassette also contained the negatively selectable hsvTDK gene, followed by a 450-bp direct repeat of homology sequences from the target gene promoter that are upstream of the sequence used for integrations, which enables the full removal of all non-native DNA from the deletion locus. Integration of the disruption construct was selected with 500 mg/L nourseothricin on YPD plates, and gene deletions were verified by PCR. Selective marker removal was carried out by growth on YPD plates (10 g/L yeast extract, 20 g/L Bacto Peptone, 20 g/L glucose, and 20 g/L agar), followed by selection on YPD with 30 μM 5-fluoro, 2-deoxyuridine (FUDR). Marker removal was verified by failure to grow on YPD with nourseothrycin and PCR of the locus. Gene overexpression of the native sequences of all transgenes was carried out by random integration of an expression construct containing the gene of interest and either the HYG or NAT gene and the hsvTDK gene.
The positive and negative marker genes were flanked by identical 406 bp sequences of the Y. lipolytica TEF1 promoter, which allowed for recombination-driven marker removal. These TEF1 promoter regions drove expression of the positive marker and overexpression gene of interest, and successful recombination left a single TEF1 promoter driving the gene of interest on the chromosome. Marker removal was verified by failure to grow on the positive antibiotic marker and overexpression phenotypes were confirmed after marker removal to ensure the transgene remained in the genome. All plasmids from this study are listed in Additional file 1: Table S4. Yarrowia lipolytica transformations were carried out by a standard lithium acetate heat shock transformation [22]. Cells were pre-grown on solid or liquid YPD overnight. For gene deletions, cells were further incubated in liquid YPD containing 50 mM hydroxyurea for 4 h at 30 °C. For transformation, cells were incubated with transforming DNA, 80 μL 60% polyethylene glycol 4000, 5 μL 2 M dithiothreitol, 5 μL 2 M lithium acetate pH 6, and 2 μL 10 mg/mL single-stranded salmon sperm DNA for 1 h at 39 °C. Cells were outgrown in YPD overnight before plating on selective media (300 mg/L hygromycin or 500 mg/L nourseothricin).

BFA plate growth and fatty acid analysis
Individual transformant colonies were inoculated into 2-mL deep-well plates containing 300 μL per well of shake-flask media (0.5 g/L urea, 1.5 g/L yeast extract, 0.85 g/L casamino acids, 1.7 g/L yeast nitrogen base without amino acids and ammonium sulfate, 100 g/L glucose and 5.1 g/L potassium hydrogen phthalate to adjust the pH to 5.5). Cells were grown for 96 h at 30 °C at 900 rpm at 70-90% humidity. Cell pellets were washed once with water, freeze-dried and whole cells were subjected to transesterification and fatty acids were analyzed by gas chromatography (GC-FID).

Saccharomyces cerevisiae plasmid and strain construction
Strain CEN.PK113-5D strain was obtained from EURO-SCARF (Frankfurt, Germany), strain MLM1.0 was obtained from the authors of Ferreira et al. [32], and strain Y&Z001 was obtained from the authors of Zhou et al. [24]. The bfaA-B gene codon-optimized for S. cerevisiae (see Additional file 3 Sequences and alignments) was cloned into p416 and p426 [33] using the Gibson Assembly ® Master Mix (New England Biolabs) with a synthetic gene fragment (GenScript, NJ, US) containing overlapping sequences to the plasmids linearized by digestion with BamHI and XhoI. S. cerevisiae strains were transformed using a PEG/LiAc technique [34].

curvata bfa, M. hydrocarbonoclasticus tmp, and T. halophila tmp gene operons expressed in E. coli
Exogenous fatty acids were separately fed to Δcfa E. coli expressing the T. curvata bfa, M. hydrocarbonoclasticus tmp, and T. halophila tmp gene operons and a reference strain with a control vector. FAME profiles were compared across strains and to a control medium with no fatty acid supplementation. Conversion of the exogenously fed fatty acid was calculated as the combined percentage of 10-methyl and 10-methylene fatty acid peak areas relative to the sum of the unsaturated and branched fatty acid peak areas.

Growth medium
Saccharomyces cerevisiae strains with auxotrophies were grown on YPD plates. URA3 plasmid carrying strains were grown on selective growth medium containing 6.9 g/L yeast nitrogen base without amino acids (Formedium, Hunstanton, UK), 0.77 g/L complete supplement mixture without uracil (Formedium), 20 g/L glucose and 20 g/L agar. Shake flask cultivations were performed in minimal medium containing 20 g/L glucose, 5 g/L (NH 4 ) 2 SO 4 , 14.4 g/L KH 2 PO 4 , 0.5 g/L MgSO 4 •7H 2 O adjusted to pH 6. After sterilization, 2 mL/L trace element solution [35] and 1 mL/L of vitamin solution were added.

Saccharomyces cerevisiae shake-flask cultivations
Biological triplicates were pre-cultivated in 5 mL minimal medium at 200 rpm and 30 °C for 18 h. Subsequently, the pre-culture was diluted into 15 mL minimal medium in a 100-mL shake flask to an OD 600 of 0.1. Shake flasks were incubated at 200 rpm and 30 °C for 72 h. A spectrophotometer (Genesis 20, Thermo Fisher Scientific, Waltham, MA, USA) was used to measure cell density at the end of the shake-flask cultivations.

Quantification of free fatty acids
Free fatty acids (FFA) were simultaneously extracted and methylated by dichloromethane containing methyl iodide as methyl donor [36]. Briefly, 200 μL aliquots of whole cell culture were taken into glass vials, then 10 μL 40% tetrabutylammonium hydroxide (base catalyst) was added immediately followed by addition of 200 μL dichloromethane containing 200 mM methyl iodide as methyl donor and 100 mg/L pentadecanoic acid as an internal standard. The mixtures were shaken for 30 min at 1400 rpm with a vortex mixer, and then centrifuged at 5000×g to promote phase separation. A 160 μL dichloromethane layer was transferred into a GC vial with glass insert and evaporated for 4 h to dryness. The extracted methyl esters were resuspended in 160 μL hexane and then analyzed by gas chromatography (Focus GC, ThermoFisher Scientific) equipped with a Zebron ZB-5MS GUARDIAN capillary column (30 m × 0.25 mm × 0.25 μm, Phenomenex) with a flame ionization detector (ThermoFisher Scientific). The GC program was as follows: 50 °C hold for 2 min; ramp to 140 °C at a rate of 30 °C per minute, then ramp to 280 °C at a rate of 10 °C per min, and hold for 3 min. The inlet temperature was kept at 280 °C. The injection volume was 1 μL. The flow rate of the helium carrier gas was set to 1.0 mL/min. Final quantification was performed using Xcalibur software.

SAM requirement calculations
The minimum estimated requirements for SAM usage during BFA production were calculated assuming only two cellular pathways would use this metabolite: phosphatidylethanolamine (PE) methylation to phosphatidylcholine (PC) and the BFA pathway. This assumption was based on the study of Ye et al. [37], demonstrating that knockout of PE methylation reaction led to up to 30-fold accumulation of SAM in the cell. SAM requirement was calculated then according to the following formula: n SAM = n PC * 3 + n BFA, in which n SAM represents moles of SAM required per g of cell biomass, n PC is the calculated moles of PC per g of cell biomass calculated from the molecular weight of 1-stearoyl-2-linoleoyl-snglycero-3-phosphatidylcholine and the lipid quantification analysis from Ferreira et al. [25] and Zhou et al. [24], and n BFA represents the moles of 10-methyl BFA produced per g of cell biomass.

BfaA-B localization
Cells were grown for 48 h in minimal medium with 2% glucose and then photographed for green fluorescent protein (GFP) fluorescence using standard methods. Nile Red staining was performed as reported by Ciamponi et al. [38].

1-L Y. lipolytica batch fermentations
Frozen stocks of strains NS1009, NS1227, and NS1647 were patched onto each of two YPD plates and grown overnight at 30 °C. A 10 µL loopful of cells from each patch was used to inoculate separate 250-mL baffled Erlenmeyer flasks with 50 mL of medium consisting of 100 g/L glucose, 0.5 g/L urea, 1.5 g/L yeast extract, 0.85 g/L casamino acids, 1.7 g/L YNB without AA or ammonium sulfate, and 5.11 g/L potassium hydrogen phthalate. The pH of the flask medium started at 5.5 and declined during growth. Inoculum flasks were cultured overnight at 30 °C with constant agitation of 200 rpm in a New Brunswick I26 incubator shaker. A volume of each flask culture required to initiate a 1 L bioreactor (Dasgip, 1.2-L vessels) at a cell density of 0.5 OD 600 was transferred to separate sterile conical tubes. Each conical tube was then brought to 50 mL with sterile deionized water and centrifuged at 3100×g for 3 min in an Eppendorf 5810 R centrifuge. The supernatant was decanted, and the cells were resuspended in 50 mL sterile deionized water. Inoculum was added to designated 1-L working-volume bioreactors with medium consisting of: 150 g/L glucose, 0. Batch process parameters were pH 3.5 automatically adjusted with 10 N sodium hydroxide, 30 °C, aeration with 0.3 vvm air, and agitation at 1000 rpm. Conditions were held constant and were such that dissolved oxygen was never limiting and always above 50%. A 10-mL sample was taken from each culture once per day. For all time-points, broth analysis was conducted via HPLC. Total dry cell weight (DCW) and total fatty acid content were measured gravimetrically by a two-phase solvent extraction. Fatty acid composition was measured by GC analysis. All results for each biological duplicate bioreactor run were averaged.

Two-phase solvent extraction for DCW and total fatty acid content
Broth volume from each harvested culture sample was added to a separate pre-weighed 2-mL screw-cap microfuge tube (USA Scientific, 1420-8799) to achieve a cell mass between 15 and 20 mg. Samples were washed twice with deionized water and centrifuged at 21,130×g for 2 min. Pelleted cells were then resuspended in 200 µL of deionized water, frozen at − 80 °C for 30 min, and freeze-dried overnight. After complete drying, each tube was weighed to obtain the DCW. To each freezedried sample and three empty (control) microfuge tubes, 400 mg of glass beads (Sigma, G8772) and 400 µL of a 3:2, cyclopentyl methyl ether (CPME):methanol solution was added. Samples were bead-beaten with maximum agitation (BioSpec Mini-Beadbeater 8) for 2 min and cooled for 10 min to allow accumulated pressure inside the tubes to normalize before opening. After cooling, 640 µL of CPME followed by 640 µL of 10% (w/v) CaCl 2 ·6H 2 O were added to each sample and vortexed. Samples were then centrifuged for 2 min at 21,130×g, creating two distinct layers. 660 µL of the top CPME layer (75% of total CPME volume) was removed and dispensed into glass vials. Samples were evaporated under compressed air until no visual solvent remained and then lyophilized overnight for total solvent removal. The remaining lipid was weighed, with any residual mass in the averaged blank samples subtracted from the mass of the experimental samples.

Identification of genes to produce 10-methyl fatty acids
To identify enzymes that produce 10-methyl branched fatty acids, we examined the genomes of several actinobacteria in the actinomycetales order that were reported to produce 10-methyl branched fatty acids. We searched for a gene or gene set with predicted protein domains for the two activities required for 10-methyl BFAs production: transferring a methyl group from SAM to a monounsaturated phospholipid acyl-chain resulting in a methylene branched fatty acid, and reduction to a methyl group via electrons donated by NADPH (Fig. 1a). We identified genes with over 25% amino acid identity to the E. coli Cfa enzyme that showed amino acid conservation in the Cfa bicarbonate ion binding site [39,40]. Although multiple cfa homologs were present in the genomes of some organisms known to produce BFAs (Additional file 1: Table S1), in every instance we identified a single operon containing two genes, which were annotated in database collections as a cyclopropanefatty-acyl-phospholipid synthase and a FAD oxidase protein (Fig. 1b, Additional file 1: Table S1, Additional file 3). The genes we identified were distinct from the M. tuberculosis ufaA1 or uma1 genes that have previously been ascribed to 10-methylstearic acid production. We originally termed these novel operons tms for their natural product, ten methyl stearate [41], but updated the name to bfa when we found they are homologous to the bfaAB gene operon from M. chlorophenolicum and others reported by Machida et al. [19]. To simplify expression of the two-gene operon in eukaryotic cells, we also created single-gene fusion enzymes of BfaA and BfaB connected by a flexible linker from the Y. lipolytica FAS2 enzyme (Fig. 1c).
Bacteria in the Desulfobacteraceae order of γ-proteobacteria also produce 10-methyl fatty acids, although the fatty acid methylation biochemistry in these bacteria is less extensively studied. Rather than 10-methylstearic acid, the primary mid-chain branched fatty acid in these bacteria is 10-methylpalmitic acid [12]. We performed a similar genome-based search in Desulfobacteraceae (Additional file 1: Table S2) and identified a two-gene operon with methyl transferase and reduction activities, which we refer to here as the tmp (ten methyl palmitate) gene operon. The tmp operon is distinct from the bfa operon, as the gene order differs and the BfaA protein is not homologous to the TmpA putative methylene reductase (Fig. 1b, and below).

bfaA and bfaB expression in E. coli
To test whether heterologous expression of the bfaA and bfaB genes enables 10-methyl fatty acid production, we cloned the bfaAB operons from several donor organisms behind the constitutive tac promoter and expressed them in E. coli. In four of the eleven tested vectors, a new fatty acid was observed upon addition of exogenously fed oleic acid (Table 1). This novel fatty acid co-eluted with a 10-methylstearic acid analytical standard in GC-FID analysis, had the identical molecular weight as 10-methylstearic acid in GC-MS analysis, and was demonstrated to be fully saturated based on GC-FID analysis before and after chemical hydrogenation (data not shown). The highest conversion of oleic acid occurred with the bfaAB operons from T. fusca and T. curvata (Table 1). We did not further pursue the seven heterologous operons that did not demonstrate activity in our assay, possibly due to protein misfolding or other expression problems. A third bfa ORF of unknown function present only in some bacterial species (Additional file 1: Table S1, bfaC) was not necessary for branched fatty acid production (data not shown), similar to previous results [19].

Methylene intermediate
An exo-methylene intermediate fatty acid species had previously been identified during 10-methyl fatty acid biosynthesis from whole cell extracts of Mycobacterium phlei [42]. To test whether recombinant expression of the putative first enzyme in the biosynthesis pathway, BfaB, would result in exo-methylene fatty acid synthesis, we cloned the T. curvata bfaB gene in a Y. lipolytica expression vector and transformed it into NS1009, a strain enriched for oleic acid production through removal of the genes for the delta-12 desaturase Fad2 and the lipase regulator Tgl3. Extracted lipid from the bfaB expression strain had characteristic 13 C NMR signals for a methylene and quaternary carbon that were not present in a control extract (Additional file 2: Figure S1A). The resulting fatty acid methyl ester chromatograph contained three new peaks we assigned as 10-methylenestearic acid, 10-methylenepalmitic acid, and 8-methylenepalmitic acid (Additional file 2: Figure S1B), resulting from the methylation of three monounsaturated fatty acids present in our Y. lipolytica strains; C18:1 Δ9, C16:1 Δ9 and C16:1 Δ7 (the product of a single round of β-oxidation of oleic acid we and others have previously observed [23,43,44]).

bfaA co-factor usage
Mycobacterium phlei whole cell extracts utilize the cofactor NADPH to reduce the methylene branch to a methyl branch in 10-methylstearic acid synthesis [15]. To evaluate whether the recombinant T. curvata bfaA enzyme used the same co-factor, we assayed crude cell extract from E. coli expressing T. curvata bfaA for activity on disrupted, heat inactivated recombinant T. curvata bfaB E. coli cells that contained 10-methylene acyl chains. We detected 10-methylstearic acid production in the presence of NADPH, but not NADH or in the absence of a reducing co-factor (Additional file 1: Table S3).

Acyl chain substrate range
To test the acyl-chain substrate specificity of heterologously expressed T. curvata BfaB and BfaA enzymes, we fed exogenous unsaturated free fatty acids to a Δcfa E. coli strain expressing the bfa pathway and an empty control vector. By comparing characteristic GC-FID retention times, we found that methylation occurred on monounsaturated acyl-chains from 14 to 20 carbons at the Δ9, Δ10, and Δ11 double bond positions ( Table 2). The highest percent conversion to methylated fatty acids occurred with 16 and 18 carbon fatty acids at the Δ9 and Δ11 positions.

tmpBA expression in E. coli
Similar to our evaluation of the bfa gene operon, we expressed several tmpBA gene operons in E. coli and measured branched fatty acid production ( Table 1). Four of the five operons produced the branched intermediate, 10-methylenepalmitic acid. However, none of them produced 10-methylpalmitic acid, indicating the tmpA gene was not active in this assay. Intriguingly, TmpA does not resemble BfaA by domain homology (Fig. 1), but has similarity to anaerobic archaeal geranylgeranyl reductases that reduce the double bonds in isoprenoid-derived membrane lipids [45]. It is possible that TmpA is oxygen sensitive or requires an electron carrier not present in our E. coli cells. We did not pursue further tmpA characterization because our a b c  Table 2). The highest percent conversion to methylated fatty acids occurred with 14 and 16 carbon fatty acids unsaturated at the Δ9 position.

Production of 10-methylene BFAs in yeast
To produce 10-methylene fatty acids in yeast, eleven bfaB candidate genes were expressed in Y. lipolytica. The bfaB genes were subcloned into Y. lipolytica expression vectors and transformed into NS1009. Multiple individual transformants for each bfaB gene were selected to undergo a 4-day lipid accumulation assay in 96-well plates, after which 10-methylene fatty acids were measured. Both 10-methylenepalmitic acid and 10-methylenestearic acid were detected at varying levels in individually isolated strains, and for simplicity we report the total levels of all C16 and C18 10-methylene fatty acids (Fig. 2a). Low levels of 10-methylene BFAs were produced by BfaB from M. smegmatis, H. subflava and R. opacus, while higher production was measured for BfaB from T. fusca and T. curvata (Fig. 2a). The isolate producing the most 10-methylene fatty acids in this initial screen was strain NS1117 expressing T. curvata bfaB on a replicating plasmid. The activity of six TmpB enzymes were similarly assayed in Y. lipolytica, and the genes from M. hydrocarbonoclasticus and T. halophila produced low levels of 10-methylene fatty acids (Fig. 2b). These two genes, plus the tmpB gene of D. balticum showed higher activity in S. cerevisiae when expressed from a high-copy plasmid (Additional file 2: Figure S2), suggesting that the enzymes are active in yeast, but activity is limited in Y. lipolytica. We selected the T. curvata bfaB gene for further BFA production in Y. lipolytica. To construct stable strains, we chromosomally integrated the T. curvata bfaB gene in subsequent constructions. We utilized a random integration approach, which allowed screening of individual isolates with different total expression levels, to assess the range of phenotypes associated with overexpression of each gene. When we screened 96 individual integration colonies for the highest 10-methylene BFA content, we isolated strain NS1195, which demonstrated 10-methylene fatty acid levels up to 28% of total fatty acids in our plate-based assay (Fig. 2c). Consistent with the idea that gene expression was limiting, integrating a second copy of T. curvata bfaB with selection from 96 colonies yielded strain NS1647, which had a 10-methylene BFA content of 39% of total fatty acids (Fig. 2c). With each integration we saw a concomitant decrease in monounsaturated fatty acids, the substrate for the BfaB enzyme. Strains producing high levels of 10-methylene BFAs exhibited decreased growth in Y. lipolytica (data not shown, discussed further below), similar to observations from expression of Bfa enzymes in cyanobacteria [20].

Production of 10-methyl BFAs in yeast
To produce 10-methyl BFAs, the 10-methylene BFA-producing strain NS1117 was transformed with various bfaA and tmpA genes. No activity was detected for any tmpA gene (data not shown), similar to our results in E. coli. In contrast, Y. lipolytica strains expressing the C. glutamicum, T. fusca and T. curvata bfaA enzymes all exhibited some production of 10-methyl fatty acids (Fig. 3a). The strain expressing both T. curvata bfaA and T. curvata bfaB had the highest conversion rate, suggesting either that T. curvata BfaA has the most activity in yeast, or there is a direct interaction between the BfaA and BfaB proteins from the same organism that serves to increases conversion of BFAs. However, all strain isolates expressing both T. curvata bfaA and T. curvata bfaB showed inefficient production of 10-methyl BFAs, with the majority of accumulated BFAs remaining in the methylene form (Fig. 3a).
To increase the conversion of 10-methylene BFAs to 10-methyl BFAs, we linked the T. curvata bfaA and bfaB genes in frame to produce the single fusion proteins, BfaA-B and BfaB-A (Fig. 1c). When expressed in Y. lipolytica, the BfaA-B fusion protein significantly increased 10-methyl BFA production so that ~ 90% of BFAs are in the fully saturated form (Fig. 3b). The BfaB-A fusion did not work as efficiently in Y. lipolytica, where it produced more 10-methyl BFAs than the unjoined genes, but high levels of methylene BFAs remained (Fig. 3b). Both BfaA-B and BfaB-A efficiently produced 10-methyl BFAs when expressed from the Y. lipolytica TEF1 promoter on a high-copy (2μ) plasmid in S. cerevisiae (Additional file 2: Figure S3), indicating the lack of BfaA reduction activity for bfaB-A expressing Y. lipolytica strains is likely due to inefficient read-through in that strain, not defective enzyme activity in the fusion protein.
To further enhance 10-methyl BFA production, we added a second copy of bfaA-B to our best Y. lipolytica BFA production strain. Among the individual isolates assayed, two had very high levels of 10-methyl BFA production, exceeding 50% of total fatty acids (data not shown). However, these two isolates also exhibited significant growth and lipid accumulation defects (data not shown), precluding their use in large-scale production. Throughout our studies, we repeatedly observed an association between very high BFA content and poor growth. We hypothesize that these growth defects are not accidental mutations arising from random integration events, but rather the specific effect of high levels of BfaB or BfaA-B activity. Therefore, we chose to proceed with the analysis of an isolate, NS1227, that demonstrated lower BFA composition (36%), but better growth in our lipid accumulation conditions (Fig. 3c).

Secretion of BFAs in S. cerevisiae
We tested whether secreting free fatty acids (FFAs) could increase the production and turnover of BFAs. We initially tested FFA secretion in Y. lipolytica similar to previous studies [26], however the secretion strains constructed in the YB-392 background grew poorly and were not appropriate for industrial fermentation. Therefore, we utilized S. cerevisiae strains engineered for enhanced free fatty acid production based on preventing fatty acid activation and degradation through deletion of FAA1, FAA4 and POX1 [24,25]. In addition, Y&Z001 carries a synthetic acetyl-CoA and malonyl-CoA overproduction pathway [24], whereas MLM1.0 has upregulated phospholipid synthesis and turnover [25]. Although these strains are engineered to prevent fatty acid activation and degradation, we have previously observed that secreted FFAs seem to originate from phospholipid acyl chains because they persist even when other pools of cellular fatty acids have been depleted and their production is correlated with increasing phospholipid levels [25]. Using S. cerevisiae constructs expressing codon-optimized bfaA-B with the S. cerevisiae TEF1 promoter, we found that the wild-type CEN.PK strain grew well with singlecopy or high-copy bfaA-B plasmids (Fig. 4a), although this low lipid strain is not appropriate for large-scale production. In both FFA production strains, high-copy bfaA-B expression caused highly variable growth, total free fatty acid (FFA) and BFA production. However, low copy expression was tolerated well (Fig. 4a-c). Y&Z001 and MLM1.0 strains expressing bfaA-B produced 56 and 53 mg BFA/L, which corresponded to 7.6 and 7.2 mg BFA/gDCW, respectively, in shake-flask assays (Fig. 4c).
We next tested whether BFA production strains have limited levels of the methyl donor, SAM, which is also required for a variety of cellular methylation reactions [46]. The most prevalent essential pathway utilizing SAM is the production of phosphatidylcholine (PC) from phosphatidylethanolamine (PE) [37]. We estimate that the wild-type CEN.PK strain requires 23-40 μmol SAM/g dry cell weight (DCW), and the MLM1.0 increased lipid strain [25] would require 73-119 μmol SAM/g DCW. We tested whether increasing intracellular SAM levels could improve growth and BFA production. SAM is not stable in growth media, however, previous studies have established that methionine supplementation increases intracellular SAM levels in S. cerevisiae [47], but with toxic effects [48,49]. We chose to supplement with 1.3 g/L methionine to maximize SAM levels while minimizing growth effects. We compared total FFA and BFA production in the absence or presence of methionine in shakeflask fermentation experiments. The MLM1.0 strain that over-produces phospholipids showed slightly increased total FFA production (Students t-test p = 0.103, standard deviations do not overlap) with methionine supplementation (Fig. 4d), consistent with the role of SAM in PC production. All strains showed some increase in BFA production with methionine (Fig. 4e), however, the results were not highly significant (students t test p values of 0.122 for wild type, 0.096 for Y&Z001 and 0.042 for MLM1.0). Methionine supplementation decreased the growth of all strains (Fig. 4f ). approximately 8 mg total BFA/gDCW, a significant improvement over the wild-type strain that produced 2 mg BFA/gDCW in this assay. Methionine supplementation of Y. lipolytica strains did not improve BFA production (data not shown), although it has been reported that methionine addition does not increase SAM levels in this organism [50], due to unique sulfur metabolism pathways [51]. These results suggest that SAM levels may become limiting when BFA production occurs at the same time as increased phospholipid synthesis, such as during cell-growth.
One hypothesis for the poor growth observed in strains producing very high levels of BFAs is that BFAs themselves are toxic to cells, possibly due to changes in membrane structure and fluidity similar to those seen with cyclopropanated phospholipids [52]. To understand where BFAs are produced in the cells, we examined BfaA-B localization in live S. cerevisiae by linking the protein to GFP. In bacteria, the Bfa enzymes are thought to act at the plasma membrane although they are not integral membrane proteins (Fig. 1a), similar to cfa synthase [40]. In the Y&Z001 strain both GFP-BfaA-B and BfaA-B-GFP fusions localized to the plasma membrane and punctate structures resembling lipid bodies (Additional file 2: Figure S4A). The localization of the GFP fusion proteins to lipid bodies was confirmed using the neutral lipid dye Nile Red in a strain engineered for increased lipid droplets [32] (Additional file 2: Figure S4B). This staining pattern suggests that BFAs are likely produced in the yeast plasma membrane and/or the lipid body membrane, and we have confirmed that they accumulate to high levels in TAG oil in Y. lipolytica (data not shown).

Production of 10-methylene and 10-methyl BFAs in bioreactors
To better characterize BFA production, we performed 1 L batch fermentations for the strains that had the best combination of growth, total lipid production and BFA production: Y. lipolytica strains NS1227 (2 copies of T. curvata bfaA-B) and NS1647 (2 copies of T. curvata bfaB), using the parental strain NS1009 as a control. Two independent fermentations were carried out for each strain and the average results were reported. Of total cellular fatty acids produced by 190 h, NS1647 accumulated 40% 10-methylene fatty acids and NS1227 accumulated 37% 10-methyl fatty acids (Fig. 5a). However, as noted in our plate assays, cells producing 10-methylene and 10-methyl BFAs showed a decreased rate of cell mass accumulation when compared to NS1009 (Fig. 5b) and consumed proportionately less glucose (Additional file 2: Figure S5). The decreased cell weight can be explained both by the slower cell division observed at our earliest sample time (Fig. 5b first time point), as well as lower total lipid levels accumulated in stationary phase in the BFA-producing strains (Fig. 5c). NS1647 accumulated 10-methylene BFAs faster than NS1227 accumulated 10-methyl fatty acids (Fig. 5a), likely due to different levels of expression of the randomly integrated gene copies. Additionally, the BfaA-B protein is significantly larger (887 amino acids) than BfaB alone (420 amino acids), which might decrease the protein accumulation rate through slower translation or folding. To compare the two BFA-producing strains, we calculated cell-specific productivity to normalize the BFA levels to the lipid-free dry cell weight (LFDCW) for each strain. NS1227 and NS1647 showed no significant difference in 10-methyl or 10-methylene BFA production, with respective average rates of 8.8e−4 and 7.7e−4 g BFA per g of LFDCW per hour (Fig. 5d, Additional file 2: Figure S5). The total BFA content of both strains was also similar; NS1227 produced BFAs at 11.3% of the total cell mass with a titer of 1.3 g/L 10-methyl fatty acids, and NS1647 produced BFAs at 10% of the total cell mass with a titer of 1.2 g/L 10-methylene fatty acids (Fig. 5e).

Discussion
We demonstrate here production of 10-methylene and 10-methyl branched fatty acids. 10-Methylene fatty acids are produced in yeast from Δ9-10 monounsaturated C16 and C18 fatty acids using methyltransferases from multiple species of both mycobacteria (bfaB) and γ-proteobacteria (tmpB). 10-Methyl fatty acids were observed when both bfaA and bfaB genes were expressed, and production was most efficient when the two genes were fused together as bfaA-B. Overexpression of the T. curvata BfaB or the BfaA-B fusion protein in Y. lipolytica enabled the production 10-methylene and 10-methyl BFAs, respectively, with a titer of at least 1.2 g/L in a 1 L batch fermentation process. Although 10-methyl BFAs were isolated from mycobacterium in 1943 [53], the enzymes responsible for the two-step synthesis of 10-methylstearic acid from oleic acid were not reported until recently [19]. The BfaA and BfaB enzymes identified here are homologous to those identified by Machida et al. [19]. In addition, we isolated a second set of genes from γ-proteobacteria that likely carry out a similar methylation reaction on palmitic acid in vivo, to produce 10-methylpalmitatic acid (tmp). The tmpB genes showed robust methylation activity in S. cerevisiae, but we were unable to observe activity for the tmpA genes in aerobically growing yeasts or E. coli. We found that the T. curvata BfaB and the M. hydrocarbonoclasticus and T. halophila TmpB can methylate a variety of C14-C20 fatty acids that are monounsaturated at the Δ9, Δ10, and Δ11 positions, with the BfaB enzyme preferring C16-C18 fatty acids and TmpB preferring C14-C16 fatty acids. Consistently, we found TmpB activity was robust in S. cerevisiae, where C16 fatty acids are more prevalent compared to our Y. lipolytica strain, which produces predominantly C18 fatty acids [23]. Both bfa and tmp two-gene operons were identified by homology searches for an ORF homologous to the E. coli cfa gene with an adjacent gene containing reductase activity. We believe that the bfa and tmp gene sets were independently evolved as evidenced by the lack of homology between the BfaA and TmpA proteins, the opposite gene order in each family, and the distant evolutionary relationship of the bacterial species from which they were isolated.
We consistently observed that production of large amounts of BFAs in yeast resulted in decreased growth and lipid accumulation. Overexpression of bfaB alone was sufficient to affect growth and lipid, indicating that neither bfaA nor the 10-methyl product is required for the effect. This finding is in agreement with the effects of M. chlorophenolicum bfaA and bfaB overexpression in cyanobacteria [20]. The methylation reaction by BfaB utilizes SAM as the methyl donor, and one possibility is that the co-factor may be limited under bfaB overexpression conditions, competing with essential cellular functions that also require SAM. Our results in S. cerevisiae suggested that SAM can be limiting for phospholipid and BFA production, and it is possible that modifications in culture conditions and genes in the SAM pathway could improve BFA production in yeast. Alternatively, and nonexclusively, the production of high levels of methylene or methyl BFAs in the cell or organelle membranes may be harmful to the cells, as the methyl addition is expected to alter membrane fluidity and structure [54]. Increasing the turnover of BFAs from cell membranes may be important for decreasing the effects on cell proliferation.
In this study, we have explored two routes to accumulate BFAs, through secretion of FFAs or accumulation in TAGs. Although we do not know how BFAs are mobilized after synthesis in the plasma membrane, we have demonstrated that BFAs are secreted by S. cerevisiae cells lacking FFA activation and degradation pathways. The titers of total fatty acids, including BFAs, were low in this system, indicating that improvements are necessary to render this production method commercially feasible. The advantages of the secretion system would be to simplify FFA recovery and purification, and to enable cost-saving continuous fermentation. As an alternative, we have also shown that BFAs accumulate efficiently in TAGs in Y. lipolytica, where we demonstrated the highest composition yet reported for 10-methyl BFA production, at 37% of total fatty acids. Previous studies reported BFA compositions of 4.1% in Synechocystis and 14% in the native producer M. chlorophenolicum [20]. Total DCW contents were not calculated in the previous studies, but natural lipid contents relative to DCW reported for Synechocystis are 10-20% [55,56], indicating the BFA content is approximately 0.4-0.8% of DCW. In the current study, the strains expressing BFA genes in Y. lipolytica had vastly increased BFA contents of 10-11.3% DCW. It is possible that BFA content and productivity can be further improved by combining T. curvata bfaA-B expression with genetic modifications previously demonstrated to increase lipid yield in Y. lipolytica [22]. Additionally, Imatoukene et al. demonstrated improved cyclopropane fatty acid production in Y. lipolytica through a combination of genetic and process improvements [57], and a similar approach could be applied to 10-methyl BFA production.

Conclusions
BFAs are of interest due to their potential applications in a wide variety of commercial products. In Y. lipolytica we produced primarily 10-methylstearic acid, a fully saturated long-chain fatty acid with high viscosity, low pour point and high oxidative stability. We also produced 10-methylenestearic acid, an unusual fatty acid with potential for chemical derivatization to novel oleochemicals. We identified genes capable of producing mono-methyl derivatives of a variety of other fatty acid chain-lengths and positions, which could be tailored to specific applications. The yeasts S. cerevisiae and Y. lipolytica are promising hosts for the further development of BFA production processes for these unique and valuable fatty acids.