Castor Stearoyl-ACP Desaturase Can Synthesize a Vicinal Diol by Dioxygenase Chemistry1[OPEN]

The Ricinus communis stearoyl-ACP desaturase is capable of dioxygenase chemistry, converting oleoly-ACP to the natural product erythro-9,10-dihydroxystearoyl-ACP. In previous work, we identified a triple mutant of the castor (Ricinus communis) stearoyl-Acyl Carrier Protein desaturase (T117R/G188L/D280K) that, in addition to introducing a double bond into stearate to produce oleate, performed an additional round of oxidation to convert oleate to a trans allylic alcohol acid. To determine the contributions of each mutation, in this work we generated individual castor desaturase mutants carrying residue changes corresponding to those in the triple mutant and investigated their catalytic activities. We observed that T117R, and to a lesser extent D280K, accumulated a novel product, namely erythro-9,10-dihydroxystearate, that we identified via its methyl ester through gas chromatography–mass spectrometry and comparison with authentic standards. The use of 18O2 labeling showed that the oxygens of both hydroxyl moieties originate from molecular oxygen rather than water. Incubation with an equimolar mixture of 18O2 and 16O2 demonstrated that both hydroxyl oxygens originate from a single molecule of O2, proving the product is the result of dioxygenase catalysis. Using prolonged incubation, we discovered that wild-type castor desaturase is also capable of forming erythro-9,10-dihydroxystearate, which presents a likely explanation for its accumulation to ∼0.7% in castor oil, the biosynthetic origin of which had remained enigmatic for decades. In summary, the findings presented here expand the documented constellation of di-iron enzyme catalysis to include a dioxygenase reactivity in which an unactivated alkene is converted to a vicinal diol.

Di-iron clusters within the active sites of enzymes facilitate the binding of molecular oxygen and its derivatives and are able to perform redox chemistry, which results in a range of chemical outcomes (Edmondson and Juynh, 1996). All di-iron enzymes characterized to date belong to one of two separate classes, one soluble and the other membrane-bound (Shanklin and Somerville, 1991). Both classes have the ability to catalyze the oxidation of unactivated C-H bonds to give a range of chemical outcomes (Shanklin and Cahoon, 1998;Fox et al., 2004). For instance, both soluble and membrane di-iron enzyme classes contain desaturase enzymes that perform the stereo-and regioselective introduction of Z-(cis) double bonds into unactivated lipid acyl chains. The reactions are thought to proceed via a radical mechanism initiated by abstraction of a specific hydrogen from substrate (Buist, 2004). Double bond formation ensues via the abstraction of a second neighboring hydrogen. As predicted by Bloch (1969) and subsequently confirmed by x-ray crystallography (Lindqvist et al., 1996;Bai et al., 2015), the boomerang shape of the substrate binding channel within the desaturase drives the formation of the (Z)-olefinic fatty acids.
There is a diverse constellation of chemical outcomes performed by variant enzymes that are structurally related to the prototypical desaturase. The membranebound di-iron-containing plant fatty acid desaturase (FAD) family of FAD2 variant enzymes perform a variety of chemical transformations. Using oleate as substrate, either desaturated or hydroxylated products are obtained; using linoleate as a substrate, the corresponding epoxide, a conjugated double bond, or an acetylenic bond can be produced. Changes in chemoselectivity are based on a relatively small number of amino acid sequence differences that presumably alter the relative orientation of the substrate with respect to the active site oxidant (Bhar et al., 2012). For instance, changes to only four amino acid side chains was sufficient to predominantly convert a FAD2 into a hydroxylase and vice versa (Broun et al., 1998;Broadwater et al., 2002). Despite our increasing understanding of specificity determining residues within the FAD2-related di-iron enzymes, further interpretation has been hindered by the lack of structural information for these enzymes. Recently published structures of several mammalian membrane-bound desaturases suggest it will be possible to solve one of the plant FAD2 class at some point and we will be able to correlate changes to the enzyme structure with distinct functional outcomes (Bai et al., 2015;Wang et al., 2015).
This work was initially designed to evaluate the individual contributions of T117R, G188L, and D280K in castor desaturase to allylic alcohol formation. During these experiments, we discovered a novel dioxygenase reactivity of the soluble desaturase that results in the conversion of oleoyl-ACP to erythro-9,10-dihydroxystearate. The same product was found in TMS-derivatized methyl esters from castor seed where it constitutes ;0.7% of the total fatty acids

RESULTS
As part of our continuing structure-function analysis of di-iron enzymes, we analyzed the contributions of Figure 1. GC-MS elution profiles of TMS derivatives. GC images of TMS derivatives of 18:1-ACP substrate (A) and product distributions for the castor desaturase triple mutant T117R/G188L/ D280K (B), and each of the single mutants T117R (C), G188L (D), and D280K (E) reveals a novel fatty acid species labeled as peak "5." Product profile of wild-type castor desaturase is included (F) as a control. Peak identities: Z18:1D9 (1), Z18:1D11 (2), Z18:1D10 9OH (3), and E18:1D10 9OH (4). Figure 2. The novel fatty acid product is 9,10-dihydroxystearate. Comparison of mass spectra of TMS derivatives of the novel enzymatic product produced by the castor desaturase T117R mutant (A) and an authentic erythro 9,10 dihydroxy stearate standard (C), and the fragmentation pattern giving rise to the major ions at 215 and 259 AMU (B). each of the mutations within the castor desaturase T117R/G188L/D280K triple mutant that converts oleoyl-ACP into (E)-10-18:1-9-OH (Whittle et al., 2008). Each of the individual mutants was constructed and tested for its activity using oleoyl-ACP as substrate. In each case, the product profiles were determined by gas chromatography-mass spectrometry (GC-MS) analysis. The results are shown in Figure 1. The GC elution profile of substrate is shown in Figure 1A and features a peak corresponding to 18:1D9 methyl ester (peak 1). A minor shoulder peak can be attributed to 18:1D11 (peak 2) and is a well-known artifact of the expression system. As shown in Figure 1B, the triple mutant T117R/G188L/ D280K converted most of the oleoyl-ACP substrate into a mixture of the Z(cis)18:1D10 9OH (peak 3) and E(trans) 18:1D10 9OH allylic alcohol (peak 4) isomers, with the E form predominating by ;3-fold over the Z form.
Reactivity of the Castor Desaturase Single Mutants T117R, G188L, and D280K Each of the single mutants was active with respect to the oleoyl-ACP substrate ( Fig. 1, C-E). The T117R mutant produced ;15-fold more of the E 18:1D10 9OH isomer than the corresponding Z isomer. However, a new peak (labeled "5" in Fig. 1C) became apparent at an elution time that was not characteristic of the silylated derivatives of commonly occurring fatty acid methyl esters. The G188L mutant produced an ;1:1 mixture of E and Z isomers of 18:1D10 9OH (Fig. 1D), but no detectable trace of the novel fatty acid species (peak 5). The D280K mutant was less active than T117R and G188L, producing only a small amount of the E isomer of 18:1D10 9OH (Fig. 1E), along with a small amount of the novel fatty acid (peak 5). As expected, the wild-type desaturase showed very little activity with its natural product oleoyl-ACP, but close inspection revealed the production of a trace of novel species (peak 5) based on its elution time and mass spectra (Fig. 1F).
The Novel Fatty Acid Product (Peak 5) Is 9,10-Dihydroxystearate Mass-spectral analysis of the peak-5 product produced by the T117R mutant (Fig. 1C) revealed a molecular ion of 474 AMU, consistent with an 18C fatty acid methyl ester containing two silylated hydroxyl groups ( Fig. 2A). Fragmentation of the product between the two silyl groups produced fragments of 259 AMU for the carboxyl-containing fragment and 215 AMU for the methyl-containing fragment (diagrammed in Fig. 2B), consistent with the presence of vicinal hydroxyl groups at C9 and C10. The identity of the peak-5 product was confirmed by comparison of its fragmentation pattern with that of a silylated authentic commercial standard of erythro-methyl 9,10-dihydroxy stearate (Fig. 2C). Analysis of the peak-5 product from the D280K mutant also showed the same fragmentation pattern.

9,10-Dihydroxystearate Produced by the T117R Mutant Is Solely in the erythro Configuration
Fatty acids containing vicinal midchain hydroxy groups may exist as threo or erythro diastereoisomers (Fig. 3). To distinguish between these possibilities, we compared the GC elution times of the novel product from T117R with those of authentic threoand erythro-9,10-dihydroxystearate standards (Fig. 4, A-C, respectively). The T117R product eluted as a single defined peak without any detectable shoulders (Fig. 3A) and coeluted with authentic erythro standard (Fig. 4C). The authentic threo standard (Fig. 4B) eluted ahead of that of the T117R product (Fig. 4A). When a small amount of the T117R product was mixed with either the threo standard ( Fig. 4D) or the erythro standard ( Fig. 4E), two peaks were seen for the sample spiked with threo standard whereas a single coeluting peak was seen for the sample spiked with erythro standard. These results . The structural relationships of compounds discussed in this work. Shown are 1 Stearoyl ACP, showing two hydrogens at C-9,10 that are removed by desaturase; 2 Oleoyl ACP, the product of a stearoyl 9,10 desaturation; 3 Erythro-9(R), 10 (R)-dihydroxystearoyl ACP, the predicted product of a one-step direct oleate dihydroxylation; and 4 Threo-9(S), 10 (R)-dihydroxystearoyl ACP, a possible product of an enzymatic two-step oleate epoxidation/hydrolysis sequence.
The Hydroxyl Oxygens at Both C9 and C10 Are Derived from Molecular Oxygen The oxygen atoms in either of the two hydroxyl groups could in principle arise from water or molecular oxygen (Fig. 5). To distinguish between these possibilities, T117R, oleoyl-ACP, and all assay components were first degassed by multiple gas exchange cycles employing vacuum and O 2 -free argon with the use of a Schlenk line (Arnold and Bohle, 1996) to remove residual atmospheric 16 O 2 from the sealed reaction vials. Assay reactions were subsequently incubated in the presence of 16 O 2 or 18 O 2 . We used mass-labeled 18:1 d 2 -11,11 oleoyl-ACP for these assays to ensure the product we observed was derived from the enzymatic reaction rather than from endogenous oleate contaminant. Analysis of the methylated silylated products from reaction under air yielded the expected 217 and 259 AMU products (the methyl fragment increased by 2 AMU relative to unlabeled product results from the substitution for the two hydrogens at C11 for deuterons; Fig. 6). The same experiment performed under 18 O 2 resulted in the production of fragments of 219 and 261 AMU, consistent with the incorporation of one 18 O at each of the hydroxyl positions.
The Formation of 9,10-Dihyroxystearate from Oleate Is the Result of a Dioxygenase Reaction The incorporation of molecular oxygen at the 9 and 10 positions of oleate could in principle result from a single dioxygenase reaction, or from two sequential monooxygenase reactions. To distinguish between these possibilities, we degassed the samples as described above and performed a reaction under an atmosphere containing an equimolar fraction of 16 O 2 and 18 O 2 (Fig. 7B) and performed MS on methylated acetonide derivatives of the product (Fig. 7E). Acetonide derivatives were used because they protect vicinal hydroxy groups while maximizing the detectable mass ion of the product. If the reaction operates via a dioxygenase mechanism, then the oxygen atoms at both hydroxyl positions should derive exclusively from either 16 O 2 or 18 O 2 , resulting in either M or M14 species. Alternatively, if the mechanism employs two sequential monooxygenase reactions, a 1:2:1 pattern of M:M12:M14 would be expected by random incorporation of either 16 O or 18 O at each hydroxyl position. Consistent with a dioxygenase mechanism, reactions performed under an equimolar mix of 16 O 2 and 18 O 2 yielded only M and M14 peaks (355 and 359), with no detectable 357 species (Fig. 7B). Individual control 16 O 2 and 18 O 2 reactions only showed the expected 355 and 359 major species accompanied by minor peaks at M11 and M12 that approximate the natural abundance of 13 C (Fig. 7, A and C, respectively). That M11 and M12 peaks originate from natural 13 C was confirmed by the fragmentation of equivalent derivatives of an authentic Figure 4. The 9,10-dihydroxystearate produced by the castor T117R mutant is solely in the erythro configuration. GC images of 9,10dihydroxy-stearates are compared for the reaction product of T117R (A) to those of standards: threo configuration (B), the erythro configuration (C), a mixture of the T117R product and the threo standard (D), and the T117R product and the erythro standard (E). erythro-9,10-dihydroxystearate, which showed the same proportions of M, M11, and M12 species (Fig. 7D).
The Native Castor Desaturase Can Convert Oleoyl-ACP to 9,10-Dihydroxystearate The formation of dihydroxystearate with selected mutated desaturases prompted us to probe for the formation of this compound by the wild-type enzyme. Interestingly, using a prolonged time of incubation (240 min) with oleoyl-ACP as substrate, we were able to identify production of 9,10-dihydroxystearate (peak 5) at low levels (Fig. 8). This compound was accompanied by lesser amounts of E 18:1D10 9 OH (peak 4).
Castor Oil Contains Erythro-9,10-Dihydroxystearate The observation that the native castor desaturase can produce small amounts of 9,10-dihydroxystearate correlates well with an early report by King (1942), who isolated a small amount of 9,10-dihydroxystearate from castor oil. We sought to confirm this observation and analyzed a fatty acid extract of castor seeds by GC-MS after methylation and silylation. Chromatograms of castor seed fatty acid derivatives (Fig. 9A) showed the expected common C16 and C18 fatty acids, along with a major peak of ricinoleic acid that is followed by a small discrete peak (labeled "8" in Fig. 9A, inset) of ;0.7% (of total fatty acids), which corresponds to the elution time of disilylated methyl 9,10-dihydroxystearate. Mass spectral analysis of this peak revealed fragments of 215 and 259 AMU confirming its assignment as 9,10-dihydroxystearate (compare Fig. 9B with Fig. 2, A and C). Based on the in vitro assays using purified enzyme reported above, we hypothesize that this 9,10 dihydroxystearate arises from the dioxygenation of oleoyl-ACP product of the stearoyl-ACP desaturase. If this were the case, the 9,10-dihydroxystearate would be in Figure 5. Two potential schemes for the conversion of oleate to erythro 9,10 dihydroxystearate by a di-iron-containing desaturase-dioxygenase. The initial bridged hydroperoxo species in both mechanisms is inspired by large-scale multireference ab initio calculations on a related enzyme (Chalupský et al., 2014). the erythro form as originally proposed (Morris and Crouchman, 1972). We therefore conducted coelution studies with authentic threo or erythro standards (Fig. 9, C-E). The 9,10 dihydroxystearate isolated from castor eluted as a single peak (Fig. 9C) with the same mobility as that of the authentic erythro standard (Fig. 9E). By contrast, two peaks were seen in the spiking experiment using the threo standard (Fig. 9D).

DISCUSSION
Stereoselective dihydroxylation reactions are important to the chemical industry (Borrell and Costas, 2017) because diols serve as valuable synthons. The osmiumbased asymmetric dihydroxylation reaction (Crispino and Sharpless, 1993) is a prominent example of controlled olefin oxidation and was (in part) recognized by the award of the 2001 Nobel Prize in Chemistry to its inventor, Karl B. Sharpless. In addition, biocatalytic diol formation from aromatics by whole-cell mutant Pseudomonas cultures has furnished the synthetic chemist with a variety of enantiomerically pure cyclohexadienecis-diols. (Hudlicky and Thorpe, 1996). Much effort has also been expended to develop iron-based biomimetic catalytic methodology for this reaction (Oloo and Que, 2015). Herein, we report the details of our investigation into a "green chemical approach": the castor D 9 18:0-ACP desaturase-mediated syn-dihydroxylation of an unactivated alkene in the form of oleoyl-ACP to erythro-9,10-dihydroxystearoyl-ACP.
Stearoyl-ACP desaturase belongs to the nonheme diiron subclass of oxidative enzymes that have been shown to mediate a variety of chemical transformations including dehydrogenation and mono-oxygenation. Typical products include primary, secondary, and allylic alcohols in addition to the conversion of double bonds to epoxides (Wallar and Lipscomb, 1996). However, a diiron center performing dioxygen chemistry to convert a double bond to a vicinal diol as reported here is without precedent. The closest comparable example we are aware of is arylamine oxygenase (CmlI) from the chloramphenicol biosynthesis pathway, which incorporates two oxygens from O 2 into the aryl-nitro product; however, this occurs in two consecutive mono-oxygenations Figure 7. The 9,10-dihydroxy stearate formation is the result of a single dioxygenase reaction. GC images and corresponding mass spectra of acetonide derivatives of 9,10 dihydroxy stearate from reactions carried out under 16 O 2 (A), equimolar 16 O 2 and 18 O 2 (B), and 18 O 2 (C). Also depicted is an authentic erythro 9,10 dihydroxy stearate standard (D) along with a diagram of its fragmentation (E). Figure 8. Upon prolonged incubation, the castor wild-type desaturase can convert 18:1 substrate to erythro-9,10-dihydroxystearate. Peak identities: Z18:1D9 (1), Z18:1D11 (2), E18:1D10 9OH (4), and 9,10dihydroxystearate (5). (Komor et al., 2017). We envision the conversion of alkene to vicinal erythro-diol in this work to be mechanistically related (Fig. 4) to that described for Rieske cisdiol-forming dioxygenases (Ensley et al., 1982;Karlsson et al., 2003). More specifically, we envision involvement of a bridged hydroperoxo-di-iron species similar to that proposed by Solomon and Srnec (Chalupský et al., 2014) for the conversion of stearate to oleate by two consecutive hydrogen atom abstractions: "-CH 2 -CH 2 -" to "-CH5CH-." When presented with an alkene moiety, the vinyl hydrogens are unavailable for abstraction for steric reasons and this same species is forced to transfer two oxygen atoms to substrate as shown in Fig. 4 (Pathway 1). Our oxygen-labelling experiments rule out an epoxidation/hydrolysis route (Pathway 2). It is possible that our T117R mutant may change the molecular architecture of the substrate binding cavity, altering the relative orientation of the substrate with respect to the hydroperoxo-di-iron group and facilitating deoxygenation relative to the wild-type enzyme. That the diol is produced as the erythro diastereoisomer, in which both hydroxy groups occur on one face (Fig. 3), is consistent with the geometry of the active site substratebinding cavity with respect to the di-iron active site oxidant (Lindqvist et al., 1996), in which stearate binds in a quasi-eclipsed conformation at C9 and C10, projecting the pro-(R) hydrogens toward the active site oxidant (Behrouzian et al., 2002). Future availability of a crystal structure of the T117R mutant in complex with bound oleoyl-ACP, or of the T117R mutant alone or with substrate bound as previously modeled (Whittle et al., 2008), would be useful starting points for probing mechanistic models using computational methods such as density functional theory. Indeed, homology modeling was recently shown to be a useful approach for elucidating selectivity mechanisms of desaturase enzymes such as FAD2 and FAD3 (Cai et al., 2018).
The low levels of 9,10-dihydroxystearate in castor suggests that this system is not optimized to produce this particular product. Higher levels of the diol may accumulate via enzymes with active site geometries that permit more efficient dioxygenation. Cardamine impatiens is an example of a plant that accumulates ;25% of 9,10-dihydroxystearate (and its chain-elongation products) in its seed oil (Mikolajczak et al., 1964). It is tempting to speculate that it contains a desaturase that has undergone mutation/selection to optimize the production of the diol from the initial alkene product. Examples of desaturases with multiple sequential oxidation activity include English ivy (Hedera helix) that can perform D9followed by D4 desaturation on stearoyl-ACP (Guy et al., 2007); FM1, a fungal membrane desaturases that sequentially inserts a D12 followed by a D15 double bond into oleoyl-phosphatidyl ethanolamine (Cai et al., 2018); and an insect multifunctional enzyme that functions as a D11 desaturase, D11 acetylenase, and D13 desaturase (Serra et al., 2007).
C. impatiens, there exists a variant acyl-ACP thioesterase that cleaves the vicinal diol fatty acid from its ACP adduct in addition to specialized acyltransferases and other components that facilitate its transfer from the plastid to triglyceride storage lipids.
More than 70 years ago, 9,10-dihydroxystearate was reported as a component of castor oil (King, 1942) at ;1% of the total fatty acids (Sreenivasan et al., 1956). The stereochemistry of the diol was later determined to be the erythro configuration (Morris and Crouchman, 1972). Consistent with these earlier reports, castor oil samples evaluated in this work contained ;0.7% of erythro-9,10-dihydroxystearate. That the wild-type castor desaturase can produce this compound was an entirely unanticipated result and resolves a longstanding mystery. In addition, our results underscore the remarkable plasticity of the nonheme di-iron catalytic center found in the desaturase family of enzymes. It appears that subtle changes in the active site architecture found in these versatile oxidants can allow new reaction pathways to emerge. Further detailed mechanistic work is needed to understand the relationship between reaction outcome and details of the active site architecture.

Mutant Construction
Synthesis of the castor (Ricinus communis) desaturase triple mutant T117R/ G188L/D280K and D280K single mutants were as described in Whittle et al. (2008) and Guy et al. (2011). The single mutants T117R and G188L were identified by mutagenesis-selection experiments (Whittle and Shanklin, 2001). The open reading frames were introduced into pET9d using XbaI and EcoRI restriction sites and the resulting clones were validated by sequencing.

Mutant Analysis
Desaturases, and variants thereof, were overexpressed in Escherichia coli BL21(DE3) with the use of pET9d. Recombinant desaturase was enriched to .90% purity by 20CM cation exchange chromatography (Applied Biosystems). Desaturation reactions (600 mL; Cahoon and Shanklin, 2000) were performed by incubation of the desaturase with 18:0-and 18:1-ACP substrates in the presence of recombinant spinach ACP-I (Beremand et al., 1987). Uniformly deuterated stearate was obtained from Cambridge Isotope Laboratories and 9,10 d2 oleate and 11,11 d2 oleate was obtained from the collection of Tulloch (1983). Experiments reported herein were replicated three or more times and representative results are presented.

Fatty Acid Analysis
Fatty acid methyl esters (FAMEs) were prepared by addition of 2 mL of 1% (v/v) NaOCH 3 in methanol and incubated for 60 min at 50°C. Fatty acid methyl esters were extracted twice into 2 mL of hexane after acidification with 100 mL of glacial acetic acid. Hexane was evaporated to dryness under a stream of N 2 , and samples were resuspended in hexane for GC analysis. FAMEs were dried and resuspended in 100 mL of N,O-bis(trimethylsilyl) trifluoroacetamide (BSTFA) and trimethylchlorosilane (TMCS; Supelco) for 45 min at 60°C to create trimethyl silyl derivatives. Samples were analyzed with an HP5890 gas chromatograph (Agilent) fitted with a 60-m 3 250-mm SP-2340 capillary column (Supelco). The oven temperature was raised from 100°C to 160°C at a rate of 25°C min 21 and from 160°C to 240°C at a rate of 10°C min 21 with a flow rate of 1.1 mL min 21 . Mass spectra were analyzed using an HP5973 mass selective detector (Agilent). For 18 O experiments, oxygen was removed from the sample cell by repeated evacuation and purging of the cell with O 2 -free argon using a Schlenk line. Two mixtures were prepared-one containing desaturase enzyme, buffer, ferredoxin NADPH 1 reductase, and substrate, and the other containing ferredoxin and NADPH. The two anaerobic mixtures were transferred to sealed reaction vials containing an atmosphere composed of either 16 O 2 , 18 O 2 (Cambridge Isotope Laboratories), or an equimolar mixture of 16 O 2 and 18 O 2 . Reactions were terminated by the addition of toluene, and fatty acids were esterified and silylated as described above for experiments designed to fragment the fatty acid to reveal the position of the vicinal hydroxyl groups. Alternatively, for the labeled oxygen experiments designed to determine the reaction mechanism, fatty acids were converted to methyl esters after which vicinal hydroxy groups were converted to their acetonide derivatives (Singh et al., 2008). To achieve this, methyl ester samples were dried under nitrogen and resuspended in 40 mL of 4 mM of ZrCl 4 catalyst in diethyl ether, 200 mL of dichloromethane (CH 2 Cl 2 ), and 5 mL of dimethoxypropane. The mixture was incubated with shaking at 22°C for 2 h The mixture was extracted with 3 mL of chloroform (CHCl 3 ) and 1 mL of water, separated by centrifugation (at 1,500g for 5 min) and the lower phase was collected and dried under nitrogen before resuspension in hexane for GC-MS analysis. Samples were analyzed on an HP6890/5973 GC-MS equipped with a 30-m 3 250-mm HP 5MS capillary column (Supelco). Oven temperature was held at 100°C for 2 min, raised to 300°C at the rate of 20°C min 21 , and held for 2 min.

Accession Numbers
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number M59857.