The tightly coupled nature of the electrophilic alkylation reaction sequence catalysed by 24-SMT (sterol C-24-methyltransferase) of land plants and algae can be distinguished by the formation of cationic intermediates that yield phyla-specific product profiles. C-24-methylation of the cycloartenol substrate by the recombinant Glycine max (soybean) 24-SMT proceeds to a single product 24(28)-methylenecycloartanol, whereas the 24-SMT from green algae converts cycloartenol into two products cyclolaudenol [∆25(27)-olefin] and 24(28)-methylenecycloartanol [∆24(28)-olefin]. Substrate analogues that differed in the steric-electronic features at either end of the molecule, 26-homocycloartenol or 3β-fluorolanostadiene, were converted by G. max SMT into a single 24(28)-methylene product. Alternatively, incubation of the allylic 26-fluoro cyclosteroid with G. max SMT afforded a bound intermediate that converted in favour of the ∆25(27)-olefin product via the cyclolaudenol cation formed initially during the C-24-methylation reaction. A portion of the 26-fluorocycloartenol substrate was also intercepted by the enzyme and the corresponding hydrolysis product identified by GC-MS as 26-fluoro-25-hydroxy-24-methylcycloartanol. Finally, the 26-fluorocycloartenols are competitive inhibitors for the methylation of cycloartenol and 26-monofluorocycloartenol generated timedependent inactivation kinetics exhibiting a kinact value of 0.12 min−1. The ability of soybean 24-SMT to generate a 25-hydroxy alkylated sterol and fluorinated ∆25(27)-olefins is consistent with our hypothesis that (i) achieving the cyclolaudenyl cation intermediate by electrophilic alkylation of cycloartenol is significant to the overall reaction rate, and (ii) the evolution of variant sterol C-24-methylation patterns is driven by competing reaction channels that have switched in algae from formation of primarily ∆25(27) products that convert into ergosterol to, in land plants, formation of ∆24(28) products that convert into sitosterol.
- enzyme evolution
- isotopically sensitive branching
- sterol C-24-methyltransferase (24-SMT)
Phytosterols are a diverse group of 24-alkyl sterols which in eukaryotic cells can provide architectural support in membranes and ‘metabolic’ roles to signal changes in plant growth and development [1–4]. The multiple functions of phytosterols, together with their structural complexity, suggest that novel biosynthetic pathways and functions remain to be identified for this isoprenoid class. Indeed, novel and recently discovered pathways to ergosterol in the kinetoplastid protozoan Trypanosoma brucei and green algae Chlamydomonas reinhardtii are distinct from each other [5,6] and both of them are partly (T. brucei) or completely (C. reinhardtii) different from the fungal ergosterol biosynthesis pathway [7,8]. These new observations, which take into account isotopically labelling experiments and co-metabolite analysis, suggest that the ergosterol biosynthesis pathways in the green algae and fungi resulted from convergent evolution, with the formation of variant sterol C-methylation products of 24-SMT (sterol C-24-methyltransferse) catalysis [∆25(27)-olefins compared with ∆24(28)-olefins] pivotal to the evolutionary change that led from the ergosterol synthesized in green algae to sitosterol formed in land plants  (Figure 1). 24-SMT enzymes have also attracted considerable interest in drug design for treatment of disease since this class of catalyst is not synthesized by humans, yet they are important to many opportunistic pathogens and parasites responsible for human suffering [9–12].
Significant complexity for phytosterols arises from the 24-SMTs that initiate the biosynthesis of different classes of C28-, C29- and C30-24-alkyl sterols. 24-SMTs catalyse highly regio- and stereo-specific C-24-methylations, hydride and methyl transfers as well as deprotonation reactions while at the same time excluding solvent from the active site to prevent premature quenching by water of the reactive cationic reaction intermediates. These membrane-associated 160–172-kDa proteins show 49–77% sequence identity across kingdoms and contain four conserved substrate-binding segments [13,14]. Cloned enzymes from fungi, plants and protozoa show a high degree of substrate specificity; the 24-SMT (EC 220.127.116.11) and 28-SMT (EC 18.104.22.168) from plants prefer cycloartenol and 24(28)-methylenelophenol respectively, whereas the 24-SMT from protozoa and fungi prefer zymosterol (EC 22.214.171.124). The majority of the characterized 24-SMTs form only one or a few products from a single substrate. It is interesting that these enzymes have evolved to handle highly reactive electrophilic species that typically convert into ∆24(28)- or ∆25(27)-olefinic products rather than for the cation to be intercepted affording enzyme inactivation that can terminate phytosterol production.
Soybean (soya bean; Glycine max) 24-SMT provides a model phytosterol catalyst to investigate phyla-specific C-24-methylation reactions, as this cloned and purified land plant 24-SMT produces the main C-24-alkylidene product(s) via a regiospecific ∆24(28) reaction pathway , although a vestigial ∆25(27) reaction pathway can be induced to form algal-like second C1-transfer products by incubation of 24(28)-methylenecycloartanol. Additional evidence of regiospecificity for the soybean 24-SMT reaction is shown by the conversion of 24-methylcycloartenol (added steric bulk at C-24) to the uncommon 24(28)-methylene-25-methylcycloartanol . Fluorinated substrate analogues have also proven to be extremely useful in uncovering relevant information about intermediates of C-24-methylation reactions. Isosteric substitution of hydrogen by electron-withdrawing fluorines exerts a dramatic decrease in reactivity of ∆24-substrates by increasing the energy of the transition state for formation of the C-25-carbocation. Thus 24-fluorocycloartenol tested with soybean 24-SMT failed to convert into a 24(28)-methylene product detectable by GC-MS analysis, but was recognized electronically and exhibited competitive-type kinetics (Ki=32 μM) against cycloartenol (kcat=0.02 min−1) and a partition ratio (kcat/kinact) of 0.07 . These results suggested the vinylic fluoride intermediate proceeded to partition in the direction of enzyme alkylation rather than undergo methyl-methylene elimination to form the exocyclic double bond characteristic of 24(28)-methylene sterol products.
In the present paper, we address the question of whether channel switching in the C-24-methylation reaction of soybean 24-SMT catalysis can be manipulated to become more primitive in product specificity by incubation of substrate mimics that contain added bulk or electronegativity at either end of the sterol frame, and we draw mechanistic and functional inferences of these findings for the evolutionary origins of sitosterol in land plants.
MATERIALS AND METHODS
NMR spectra were recorded at room temperature (23°C) on a Varian Unity Inova 500 MHz NMR spectrometer, JEOL 400 MHz NMR spectrometer or Varian Mercury Plus 300 MHz spectrometer. 1H-NMR: 500, 400 or 300 MHz [CDCl3 (deuterated chloroform)/chloroform at 7.260 p.p.m.]; 13C-NMR: 75 MHz (CDCl3/chloroform at 77.00 p.p.m.); 19F-NMR: 282.35 MHz (CDCl3/CFCl319F resonating at 0.000 p.p.m.). Chemical shifts are reported relative to internal standards listed. Synthetic compounds and enzymatic reaction products were analysed by GC-MS using a Hewlett-Packard 6890 GC-MSD [70 eV EI (electron impact), scan range 50–550 a.m.u. (atomic mass units)] equipped with 30 mm length×0.25 mm i.d. (inner diameter) fused silica column coated with Zebron ZB-5, film thickness 0.25 μm (Phenomenex) with He as carrier at a flow rate of 1.2 ml/min. Cool On-Column injection was used. The oven was programmed to be isothermal at the first minute of 170°C, and then ramped to 280°C at 20°C/min. The source temperature was 230°C. Cholesterol under these conditions eluted at 13.8 min (old column) or 14.5 min (new column).
Fluorinated analogues and other compounds were chromatographed by gravity-flow chromatography on high-purity grade 60–200 μm high-purity 60 Å (1 Å=0.1 nm) silica gel (BDH Chemicals) with diethyl ether graded into a mixture of hexanes (Fisher) and select fractions further purified by semi-preparative reverse-phase C18 chromatography performed on TSKgel (120 Å, 300 mm length×7.8 mm i.d., 10 μm particle size; Tosoh Bioscience) or Whatman Partisil (10-ODS-3, 250 mm length×9.4 mm i.d., 10 μm particle size; GE Healthcare). As necessary, sterol mixtures in these HPLC fractions were purified to homogeneity by analytical reversed-phase chromatography performed on Zorbax (ODS, 250 mm length×4.6 mm i.d.; Agilent). The solvent employed for HPLC was methanol. Detection of sterol eluting from the HPLC was monitored at 210, 240 or 282 nm using a diode array multiple wavelength detector. UV maximum absorbance (λmax) was from the diode array detector of the analytical HPLC system. GC of sterols are reported at RRTc (retention times relative to cholesterol).
Chemicals and reagents were purchased from Sigma and used without further purification or treatment unless otherwise noted. SAM (S-adenosylmethionine; also known as AdoMet) was the iodide salt, [methyl-2H3]SAM was the tetra p-toluenesulfonate salt purchased from C/D/N Isotopes Medical Isotopes. [methyl-3H3]SAM was purchased at 5–15 Ci/mmol from PerkinElmer and diluted with non-radioactive SAM to specific activity of 10 μCi/μmol. DAST (diethylaminosulfur trifluoride) was purchased from Alfa Aesar. Cycloartenol, lanosterol, 3-fluorolanosta-8,24-diene and 3-acetoxycycloart-24(25)-en-26-al were obtained or synthesized by methods reported previously [18,19]. All substrates, as shown in Figure 2, were of 95% or greater purity by capillary GC analysis.
Synthesis of 26-homocycloartenol and 26-fluorinated analogues of cycloartenol
A general outline for the SC (side chain) extension and fluoroalkylation in position C-26 of cycloartenol of corresponding enal derivatives is shown in Figure 3 with the SC modifications reported at SC-1 to SC-7. The preparation and characterization of these compounds are as follows. 26-Homocycloartenol 4: to a solution of 10 ml of THF (tetrahydrofuran), dried with sodium-benzophenone ketyl, 71 mg of 3-acetoxycycloart-24(25)-en-26-al SC-2 in 10 ml of THF was added and reacted with methylene Wittig ylide under nitrogen (prepared in situ from triphenylphosphonium methyl bromide in THF with 1.0 ml of 1.6 M n-butyllithium in hexanes) to form the conjugated C-24–C-26 diene construction SC-6. Deprotection of 3-O under basic conditions afforded 60 mg of product of the corresponding 3β-ol SC-6 (125 μmol, 85% yield). SC-63-alcohol (38 mg) was reduced by 0.5 cm3 of Raney nickel/dihydrogen in 7 ml of benzene to produce 19 mg of mixed isomers 26-homocycloarten-25(26)E-enol, 25(26)Z-26-homocycloarten-25(26)Z-enol, (25R)-26-methylenecycloartenol and (25S)-26-methylenecycloartenol, and 26-homocycloartanol (approximately 11 μmol, 25% yield 4). Preparative HPLC using a TSKgel analytical column eluted with methanol at 1 ml/min at ambient temperature afforded 4 (SC-7) and small amounts of the Δ25(26) isomer (detected by relevant olefin chemical shifts in the NMR). Isolated yield: 5%. Chemical data for 4: RRTc: 1.68. EI-MS m/z [rel. int. (relative intensity)]: 440 (10) [M+ (molecular ion)], 425 (17) [M+−methyl], 422 (22) [M+−H2O], 407 (36) [M+−methyl−H2O], 379 (18) [M+−propyl−H2O], 353 (11) [M+−part of A ring−H], 315 (8) [M+−SC], 300 (35), 95 (100), 69 (93) [C-23–C-27 allylic cation+], 55 (79). δH: 4.905 (H1–24, m), 3.263 (H1–3, m), 1.872 (H2–26), 1.561 (H3–27, s), 1.247 (H2–23, m), 0.962 (H3–33), 0.944 (H6–18, 30, s), 0.861 (H3–32, s), 0.841 (H3–21, m), 0.787 (H3–31, s), 0.529 (H1–19 endo, d, J 4.2 Hz), 0.308 (H1–19 exo, d, J 4.2 Hz); λmax end absorption.
26-Fluorocycloartenol 2: to a solution of methanolic sodium borohydride (20 mg in 5 ml) was added to the previously described 3-acetoxycycloart-24(25)-en-26-al SC-2 (140 μmol)  and the reaction mixture stirred at 0°C for 2 h. The reaction was quenched by saturated aqueous ammonium chloride and the precipitate filtered off to give 3-acetoxycycloart-24(25)-en-26-ol SC-3 (120 μmol, 86% yield). DAST (0.2 ml) was reacted with SC-3 in 20 ml of dichloromethane at reflux for 5 min, quenched with 50 ml deionized water, and extracted three times with 50 ml of dichloromethane to produce 46 mg of fluorinated product that contained a mixture of 3-acetoxy-26-fluorocycloart-24(25)-ene SC-4 and its isomer 3-acetoxy-24-fluorocycloart-25(26)-ene in a 5:1 ratio (38 mg, 79 μmol, 66% yield). The protecting group at C-3 was removed by hydrolysis in 10 ml of 10% methanolic potassium hydroxide at room temperature overnight. The resulting free-alcohol mixture was added to 50 ml of deionized water and extracted three times with 50 ml of dichloromethane to yield 20 mg (45 μmol) of 2 in 90% yield and 52% yield from 3-acetoxycycloart-24(25)-en-26-al SC-2; 2 was purified by Whatman HPLC. Isolated yield: 17%. RRTc: 1.53; EI-MS m/z (rel. int.): 444 (22) [M+], 429 (48) [M+−methyl], 426 (55) [M+−H2O], 424 (1) [M+−HF], 411 (100) [M+−methyl−H2O], 383 (48) [M+−propyl−H2O], 357 (35) [M+−allylic radical], 315 (13) [M+−SC], 304 (76), 95 (79). δH: 4.976 (H1–26, m), 4.921 (H1–26, m), 4.766 (H1–24, dq, J 48.5 Hz, 7.5 Hz), 3.284 (H1–3, m), 1.735 (H3–27, s), 0.964 (H6–18, 30, s), 0.888 (H3–32, s), 0.876 (H3–21, d, J 6.0 Hz), 0.806 (H3–31, s), 0.550 (H1–19 endo, d, J 4.0 Hz), 0.334 (H1–19 exo, d, J 4.0 Hz). δF: −177.0 (m); λmax end absorption.
26,26-Difluorocycloartenol 3: 3-acetoxycycloart-24(25)-en-26-al SC-2 (93 μmol) was added to a solution of 0.5 ml of DAST in 5 ml dichloromethane and refluxed for 1 h under nitrogen, followed by quench with 60 ml deionized water and extraction three times with 60 ml dichloromethane, to yield 37.6 mg of 3-acetoxy-26,26-difluorocycloart-24(25)-ene SC-5 and the isomer 3-acetoxy-24,26-difluorocycloart-25(26)-ene in a 3:1 mixture (evident by the 1H-NMR spectrum). O-Deprotection was performed by 5 ml of 10% methanolic potassium hydroxide at room temperature to yield the corresponding C-3-alcohol 26,26-difluorocycloartenol SC-5 3 and its 24,26-difluoro-25(26)-ene isomer. To the reaction mixture, 60 ml of deionized water was added, and this was extracted three times with 60 ml of dichloromethane to give 34.5 mg of mixed isomers. The desired SC-5 product was purified by semi-preparatory HPLC (Whatman Partisil 10-ODS-3 column eluted with methanol at 20°C at 2.5 ml/min), followed by a second round of purification of select HPLC fractions using analytical HPLC (Zorbax ODS, eluted with methanol at 20°C at a flow rate of 1.0 ml/min). Isolated yield: 45%. RRTc: 1.56. EI-MS m/z (rel. int.): 462 (6) [M+], 447 (19) [M+−methyl], 444 (21) [M+−H2O], 442 (1) [M+−HF], 429 (53) [M+−methyl−H2O], 401 (29) [M+−propyl−H2O], 375 (21) [M+−ethyl−H2O−2HF], 322 (40), 315 (5) [M+−SC], 95 (100). δH: 5.902 (H1–26, t, J 65 Hz), 5.696 (H1–24, m), 3.282 (H1–3, m), 1.576 (H3–27, s), 0.964 (H6–18, 30 s), 0.904 (H3–21, d, J 4.4 Hz), 0.888 (H3–32, s), 0.806 (H3–31, s), 0.550 (H1–19, endo, d, J 4.0 Hz), 0.334 (H1–19 exo, d, J 4.0 Hz). δF: −114.2 (d, J 113.0 Hz); λmax end absorption.
Enzyme assay and product analysis
Procedures for the heterologous overexpression in Escherichia coli and isolation and incubation of the soluble fraction from a 13300 g preparation of recombinant soybean 24-SMT (G. max SMT1) has been described previously . The standard linear range assay procedure for G. max SMT1 involved incubation in triplicate in 10 ml test tubes containing 20 mM phosphate buffer (pH 7.5), 5% glycerol, soluble fraction protein (2 mg) and the sterol substrate emulsified in Tween 80 (final concentration of 1.2 g/l); sterol substrate was varied from 5 to 100 μM and SAM cofactor fixed at 100 μM (diluted with 0.6 μCi of [methyl-3H3]SAM from the stock solution). The reaction mixture was incubated at 35°C for 45 min in a total volume of 600 μl. The incubation mixture was terminated by brief vortex-mixing and the addition of a solution (600 μl) of 10% methanolic KOH. The C-methylated sterol products were extracted in hexane and dried: an aliquot of the hexane extract was analysed by GC-MS or in the case of radioactive samples analysed by scintillation counting to determine the conversion rate. Control experiments of [methyl-3H3]SAM without sterol were conducted with each enzyme preparation to determine the radioactive background which generally afforded less than 500 d.p.m.; for optimal sterol methylation, the amount of radioactivity recovered from the organic extract was approximately 1×106 d.p.m. per assay. Protein concentration was measured by the dye-binding method .
The standard initial velocity data were evaluated by SigmaPlot 2001 with the enzyme kinetics module software package. The data were fitted to the equation v=VmaxS/Km+S using a non-linear least squares approach. Data for kinetic constants were initially determined by varying the substrate concentration in the presence or absence of inhibitor at fixed concentrations and SAM fixed at saturation (100 μM), followed by another series of assays in which the concentration of inhibitor varied while substrate concentration was held constant . Km and Vmax values typically had R values of >98 with an error in work-up of triplicate samples generally of 5–10%. Steady-state inhibition was determined with the software package in analogous fashion to the initial velocity data analysis. To investigate whether the inhibition kinetics were of the competitive, non-competitive or uncompetitive type, the data were fitted in a double reciprocal plot to the respective equation using non-linear squares analysis . The product distribution of organic extracts generated by 24-SMT incubated with substrate analogues was determined at saturating levels of co-substrates and with sufficiently large preparations (3 mg/ml of protein) to ensure accuracy by GC-MS peak integration from the total ion current chromatogram. Product yields determined by capillary GC were accurate to approximately 1% (50 ng). Modification experiments to establish time-dependence of SMT inactivation were carried out using 100000 g soluble fractions by established procedures [16–18]. Co-localization experiments were performed using 100000 g supernatant enzyme preparations incubated with no substrates (sterol or SAM; control), cycloartenol or 26-fluorocycloartenol paired with SAM or [methyl-3H3]SAM respectively and the resulting samples loaded on to a Q-Sepharose column (2.5 cm×5 cm) eluted with step-wise gradient from 100 to 250 mM NaCl in buffer A [20 mM phosphate, 2 mM MgCl2, 2 mM 2-mercaptoethanol, 1 mM EDTA and 5% glycerol (v/v), pH 8.0] and emphulogen detergent (to a final concentration of 0.4%) as described previously . The eluted fractions were monitored for radioactivity directly or assayed for SMT activity according to the experimental design.
RESULTS AND DISCUSSION
Incubation of cycloartenol, 26-homocycloartenol, lanosterol and 3β-fluorolanostadiene with soybean 24-SMT
Cycloartenol and lanosterol have been previously incubated with the recombinant G. max 24-SMT and, from steady-state determinations (Vmax/Km) of different substrates added to a supernatant preparation, lanosterol was reported to be 65% as competent as the optimal substrate cycloartenol (680 pmol/min per mg), whereas the yeast substrate zymosterol was only 9% as competent as the cycloartenol substrate . Cycloartenol converted into a single product, 24(28)-methylenecycloartanol; the identity of the product was confirmed by MS and 1H-NMR .
Using fresh enzyme preparations of soybean 24-SMT assayed with lanosterol or cycloartenol, we made a renewed search for ∆25(27)-olefin products or by-products that could be originally bound covalently to the enzyme and then released from the adduct on saponification affording a ‘diol’ sterol that contains hydroxy groups at C3 and C25 (Figure 1). Possible enzyme-generated products of cycloartenol methylation that possess a ∆25(27)- or ∆24(28)-alkylated sterol SC, such as cyclolaudenol and 24(28)-methylenecycloartanol, elute differently in GC at RRTc of 1.54 and 1.57 respectively. The difference in the elution times of these sterols provide a separation factor of 1.02 that can be used to monitor for ∆25(27)-product in the product mixture. Polar diol-sterols, on the other hand, that contain a C-25-hydroxy group, are expected to elute several minutes later than the corresponding monol sterol with the 3β-OH group . Notably, we demonstrated the ability of 24-SMT to form diol sterols, including the alkylated products at C-24 and C-26 characterized by GC-MS and NMR, from incubations of a ∆22,24-diene or fused 26,27-sterol SC containing substrates that yield 3,22-dihydroxy and 3,24-dihydroxy sterols respectively [16,22]. In the present study, the identity of total sterols in the non-saponifiable lipids from methylation of soybean 24-SMT on different substrates was elucidated through GC-MS analysis. Thus the GC chromatogram of cycloartenol incubation showed, within our limits of detection, two major GC peaks at RRTc 1.43 (substrate) and 1.57 [∆24(28)-product]; the substrate was converted into 24(28)-methylenecycloartanol product of 67% yield, similar to that reported previously [15,21]. Incubation of lanosterol also showed two distinct peaks at RRTc 1.33 (substrate) and 1.45 [∆24(28)-product] with conversion to product at 44% yield, consistent with our previous work . As further noted from the GC chromatogram of the lanosterol incubation (Figure 4A), there are no detectable polar metabolites in the region 19–32 min that can signal formation of ‘diol’ sterols. As expected, the corresponding mass spectra of the products from cycloartenol and lanosterol incubations with 24-SMT were indistinguishable from the authentic specimens of 24(28)-methylenecycloartanol and eburicol previously prepared in this laboratory (Supplementary Figure S1 at http://www.biochemj.org/bj/456/bj4560253add.htm).
To investigate further whether modified structural features introduced at the distal and proximal ends of the sterol frame, likely to be key for enzyme recognition, show changes in behaviour, we chemically altered the size and polarity of the ∆24-substrate at C-3 and C-26 to form the analogues 3-fluorolanostadiene and 26-homocycloartenol. Both analogues, as judged by GC-MS analysis, were converted into 24(28)-methylene product at approximately the same 12% yield, reinforcing the fidelity in reaction mechanism against variant stereoelectronic changes in the substrate. The capillary GC retention time of the products relative to the retention time of cholesterol of 1.83 for 24(28)-26-homocycloartenol and 1.11 for 3-fluoro-lanosta-8,24(28)-diene and the mass spectra of these compounds (Supplementary Figure S1) that showed M+ 454 and M+ 428 respectively, are consistent with the structures assigned. Again, as shown in the capillary GC chromatogram of the enzyme-generated product profile from 3-fluorolanostadiene (Figure 4B), there is no evidence for the biosynthesis of ∆25(27)-olefin at RRTc 1.09 or ‘diol’ sterols at or approximately RRTc 1.42. Although the formation of a single product is typical of activity assays with ∆24(25)-substrates, even prolonged incubation with alternative substrates of up to 16 h and higher amounts of total protein (3–5 mg) failed to generate SMT products other than the conventional 24(28)-methylene sterol SC as judged by GC-MS analysis.
Catalytic competence and inhibition with 26-fluorocycloartenols
A major determinant of the structure and stereochemistry of the ultimately formed C-24-methylated sterol is believed to be the precise conformation of the ∆24-substrate SC at the enzyme active site. Therefore it was of some surprise that the added steric bulk from the extended SC of C-26-homocycloartenol, a feature that might disrupt the binding mode of the substrate, failed to alter the product outcome. We tested another possibility to affect catalysis, that is, to alter the electronics of the intermediate by replacing C-26H with fluorine. The heteroatom of the modified substrate should only slightly perturb the size and shape of the sterol molecule while at the same time the fluoro substituent exerts a strong influence on the electron-withdrawing chemistry at the site of replacement. This structural modification is anticipated to decrease the nucleophilicity of the substrate 24(25)-double bond and hence slow the timing of C-24-methylation in such a way as to affect the stability of the bound intermediate in the activated complex. We surmise charge–charge interactions from the allylic fluorine adjacent to the cation-stabilizing residue(s) (possibly Tyr83 in region 1 of the primary structure of soybean 24-SMT ) can effectively push the catalytic nucleophile away from C-28 and closer to C-27 of the intermediate that, thereby, favours the ∆25(27) reaction pathway in the direction of cyclolaudenol formation or the repositioned base can move to quench the cyclolaudenyl cation and terminate the reaction.
The derived steady-state kinetic parameters and catalytic efficiency of the 24-SMT reaction for the fluorinated olefins paired with SAM, along with comparable data for the unmodified substrate cycloartenol paired with SAM under the same conditions are reported in Table 1. The Km and Vmax for the physiological substrate were determined to be 31 μM and 793 pmol/min (mg of total protein). Substitution of fluorine for hydrogen at C-26 of the cycloartenol SC had little effect on the binding constants (Km) of the substrate analogues, but suppressed the rate of formation of C-24-methylation metabolites by a factor of approximately 10, yielding catalytic competency (Vmax/Km) for 26-fluorocycloartenol of 2.6 and 26-difluorocycloartenol of 1.3. This kinetic behaviour, coupled with the resemblance of fluoro-analogues to the true substrate and intermediate, would predict these to be good competitive inhibitors. Indeed, inhibition studies showed that 26-fluorocycloartenol and 26-difluorocycloartenol are competitive inhibitors against cycloartenol (Figure 5) and the substrate analogues exhibited Ki values close to the Km of cycloartenol and hence bound to the enzyme's active site in a way similar to that of the natural substrate. The value for Ki of the fluorinated olefins was obtained from the appropriate slope and intercept replots, and is listed in Table 1.
Co-localization and mechanism-based inactivation studies with 26-fluorocycloartenol
To test whether 26-fluorocycloartenol can also undergo specific covalent binding of SMT, we first performed an experiment to track the bound product co-localized with the protein through chromatographic steps that otherwise led to release of the substrate and turnover product from the catalyst. Previously, we showed that the soluble 100000 g bacteria fraction of soybean SMT can be purified by applying this supernatant to a Q-Sepharose column (2×) followed by FPLC using a Mono Q column ; the most active fractions from the Q-Sepharose chromatography step at approximately 165 mM in the salt gradient are enriched to approximately 85% by SDS/PAGE (12% gel)  (Supplementary Figure S2 at http://www.biochemj.org/bj/456/bj4560253add.htm). In the case of a microsomal preparation of Candida albicans SMT, purification through the same protein purification steps outlined for the soybean SMT led to a complete loss of endogenous sterol in the fractions eluted from the Q-Sepharose column . With these observations in mind, the soybean SMT was incubated with (A) 100 μM 26-fluorocycloartenol paired with 100 μM [methyl-3H3]SAM (total 3H of 8×105 d.p.m.) or (B) cycloartenol paired with 100 μM [methyl-3H3]SAM (total 3H of 5×106 d.p.m.) in the usual way. The pooled preparations were partially purified by Q-Sepharose column chromatography using the methods adapted from previous reports [15,24] and most of the radioactivity was retained on the column. For Q-Sepharose chromatography, elution was performed by a step-wise gradient of increasing salt concentrations (100–230 mM) and the radioactivity determined in each of the six 30 ml fractions. In the first fraction of 100 mM salt, low radioactivity was detected by scintillation counting. However, this fraction neither showed significant SMT activity after incubation with cycloartenol and SAM nor was sterol detected by GC-MS analysis, suggesting the residual 3H is associated with SAM contaminants. In the next five fractions, various amounts of SMT were detected by SDS/PAGE analysis  (Supplementary Figure S2). Methyltransferase from (B) above eluting in several of these fractions was devoid of radioactivity. Alternatively, the same enzyme preparation of 26-fluorocycloartenol paired with [methyl-3H3]SAM of (A) yielded approximately 2×104 d.p.m. in the Q-Sepharose column fractions that typically contain SMT, consistent with enzyme alkylation (Supplementary Figure S3A at http://www.biochemj.org/bj/456/bj4560253add.htm). In a separate experiment, SMT processed directly without prior assay with [methyl-3H3]SAM, and each Q-Sepharose fraction assayed with cycloartenol and [methyl-3H3]SAM, yielded a high activity fraction of approximately 1×105 d.p.m. that corresponded to the same fractions containing 24-methyl 26-fluorocycloartenol complexed with SMT. Alternatively, methyltransferase inactivated by 26-fluorocycloartenol and the soluble enzyme–analogue complex loaded on to a Q-Sepharose column was assayed without measurable recovery of catalytic activity, indicative of an irreversible reaction (Supplementary Figure S3B).
To further test for enzyme inactivation by the fluorinated substrate, increasing concentrations of 26-fluorocycloartenol incubated with soybean SMT (1 mg) at 35°C in phosphate buffer at pH 8.0 afforded pseudo-first-order time-dependent inactivation kinetics, as shown by the linear dependence of the log of residual activity against time (Figure 6). The rate of inactivation of SMT was saturable, with a maximum rate of inactivation, kinact, of 0.12±0.01 min−1 (t1/2=5.4 min) and a Ki for 26-fluorocycloartenol of 77 μM (similar to that reported in Table 1). These values compare very favourably with the steady-state kinetic parameters for the normal substrate cycloartenol. As expected, no inactivation occurred when the incubations were carried out with sterol only, consistent with the absolute dependency of the co-substrates for the reaction to generate a covalently bound product.
Enzymatic C-24-methylation of 26-fluorocycloartenols
Initial overnight incubations monitored by GC-MS showed that the soybean 24-SMT catalysed the C-24-methylation of 26-fluorocycloartenols to multiple products at RRTc of 1.65, 1.66 and 2.13 at a ratio of 19:50:31 of approximately 9% yield. These products are considered to be the pair of ∆25(27)-olefin and ∆24(28)-olefin monol- and diol-fluorinated derivatives of cyclolaudenol, 24(28)-methylenecycloartanol and 25-hydroxycyclolaudenol, on the basis of their GC retention times and mass spectral characteristics relative to that of the corresponding parent compound lacking a fluorinated group [25,26] (Figure 7A, and Supplementary Figure S4 at http://www.biochemj.org/bj/456/bj4560253add.htm). Inspection of the molecular ion clusters for M+, M+−CH3, M+−H2O and M+−CH3−H2O, characteristic fragmentation ions in the mass spectrum of sterols, of 24(28)-methylenecycloartanol and 26-fluorocyclolaudenol indicated a quantitative shift in the parent ion from m/z 440 to 458 and a shift of 14 a.m.u. in the high mass end of the other three ions. Relevant fragmentation ions for SC cleavage and the A and B ring in 24(28)-methylenecycloartanol, or its isomer cyclolaudenol, appearing at m/z 353 and m/z 300 respectively, were moved to 355 and 318 m/z in 26-fluorocyclolaudenol. Entire cleavage of the 26-fluoro SC (C9H16F) is seen at m/z 316 for M+−SC and 297 for M+−SC−H2O−2H. Notably, an ion is present at M+−20 for HF elimination and, as expected, this ion is absent from the 24(28)-methylenecycloartanol spectrum. Other diagnostic ions in the fragmentation pattern typical of 9β,19-cyclosteroids at 371 (8%) due to loss of C2, C3 and C4 and ions at m/z 55, 69 and 95 a.m.u. are in agreement with the steroidal character of the metabolite [25,26]. The observed M+ is consistent with the expected molecular mass of the ∆25(27)-methylation product of 26-fluorocycloartenol turnover. The fragmentation pattern of the other fluorinated monol, 26-fluoro-24(28)-methylenecycloartanol, lacked M+ and M+−CH3, but the mass spectrum continued to be dominated by ions found in its isomer, including the relevant peak at 438 a.m.u. (M+−HF, 10%) consistent with retention of the SC fluoro group. The diol metabolite eluting at RRTc 2.13 is predicted to possess a molecular mass of M+ 476 a.m.u. However, as was the case for the one of the fluorinated monols lacking M+, the MS of this material shows peaks that correspond to m/z 456 (3%) for M+−HF, followed by a strong peak at m/z of 438 (21%) for M+−HF−H2O, m/z at 423 (29%) for M+−HF−H2O−CH3 and a strong m/z 316 for M+−SC cleavage. To further support the generation of the diol metabolite from subsequent saponification of a putative fluorosterol-SMT ester linkage, a separate preparative incubation of 26-fluorocycloartenol with recombinant enzyme was performed followed by saponification-free extraction. In this case, the enzyme assay was terminated by the addition of chloroform/methanol [2:1 (v/v)] and the resulting organic extract analysed by GC-MS; only two GC peaks were evident and they corresponded to the monol products. This extraction method using chloroform/methanol should not and did not release any sterol covalently bound to the enzyme. The trapping and tentative identification of 26-fluoro-25-hydroxy-24-methylcycloartanol represents the first example of trapping a high-energy intermediate considered to be part of the normal sterol C-24-methylation reaction that generates ∆25(27)- and ∆24(28)-alkylated products.
GC analysis of the total sterol mixture by GC-MS from the 26-difluorocycloartenol incubation showed that the product set was similar to that of the 26-fluorocycloartenol incubation, except 26-difluorocyclolaudenol (monol), 26-difluoro-24(28)-methylenecycloartanol (monol) and 26-difluoro-25-hydroxy-24-methylcycloartanol (diol) distributed in a ratio of 6:4:90 at approximately 1% yield (Supplementary Figure S5 at http://www.biochemj.org/bj/456/bj4560253add.htm). Owing to the increased destabilizing inductive effect from one to two fluorine substituents on the cyclolaudenyl cation that thereby greatly enhanced the entropic barriers to catalysis posing limitations on the reaction pathway, the enzyme was alkylated to greater extent from 26-difluorocycloartenol evidenced in formation of the 26-difuorinated diol. Alternatively, the increases in formation of adduct leaves little intermediate for turnover in the direction of deprotonation to monol formation. MS analysis of the enzyme-generated products detected by GC show the first two GC peaks at RRTc 1.67 and 1.70 (slightly longer than their mono-fluoro-counterparts by a retention factor of 1.02) possess the same molecular mass of M+ 476 (Supplementary Figures S6A and S6B at http://www.biochemj.org/bj/456/bj4560253add.htm), consistent with a C-24-methyl group in the fluorinated sterol products. The late eluting GC peak at 2.15 possessed an M+ 442 for M+−CH2F2 (predicted M+ 494) and other diagnostic ions compatible with the 26-difluoro diol structure (Supplementary Figure S6C).
Isotopically sensitive branching experiments
To explore the mechanism of this unexpected transformation and to confirm product identities, we carried out incubations of recombinant soybean 24-SMT with 26-fluorocycloartenol paired with [methyl-2H3]SAM and analysed the resulting enzyme-generated products by GC-MS . Mass spectral analysis of total sterols from the non-saponifiable lipids clearly demonstrated high incorporation of three deuterium atoms in the first product eluting in GC (M+ 461), two deuterium atoms into the second product eluting in GC (M+−HF, 440) and three deuterium atoms in the third product eluting in GC (M+−HF, 459), corresponding to the two monol and diol products respectively (Supplementary Figure S5). The incorporation of three deuterium atoms (>99%) from [methyl-2H3]SAM into the diol SC established the compound contains a C-24-methyl structure and that it is a true biosynthetic metabolite.
C-24-methylation of 26-fluorocycloartenol paired with [methyl-2H3]SAM gave three products analogous to a control of the fluorinated sterol paired with [methyl-3H3]SAM, but in a ratio of 23:38:39 (Figure 7B). GC analysis of the product profile showed the proportion of C-24-methyl Δ24(28)-olefin to C-24-methyl Δ25(27)-olefin in the control incubation was markedly decreased in assays in the presence of [methyl-2H3]SAM and the proportion of the monol to diol changed from 69:31 to 61:39 respectively. Thus the presence of deuterium at C-28 in the initial product formed slows down the rate of the elimination producing ∆24(28)-olefin and the proportion of ∆25(27)-olefin increases. These results reveal isotopically sensitive branching in C-24-methylation can lead to multiple products synthesized by the 24-SMT by partitioning along different reaction channels at the active site. The increase in overall production of deuteriated diol coupled with the change in product ratios towards ∆25(27)-olefin formation suggests a pronounced deuterium isotope effect during the stabilization of the cyclolaudenyl cation or rearrangement of the cyclolaudenyl cation to the cyclobranyl cation and indicates that proton loss associated with methyl→methylene eliminations at C-28 can be a rate-limiting step in the biosynthesis of phytosterols. Furthermore, the combination of kinetic and chemical results using fluorinated sterol acceptor paired with SAM or [methyl-2H3]SAM presented above are consistent with mechanism-based inactivation of soybean SMT in which a derailment intermediate is intercepted by a nucleophilic residue within the active site, leading to covalent binding to the protein.
Mechanistic and phylogenetic implications
The fine-tuning of 24-SMTs under the pressure of natural selection has resulted in efficient catalysis that impart regio-and stereo-chemical precision to phytosterol diversity . A major contribution to product specificity in the first C1-transfer reaction catalysed by soybean 24-SMT appears to come from a tight active site composed of two separate binding pockets for ∆24-sterol and SAM. The pockets are in close proximity with numerous aliphatic and polar contacts to the acceptor molecule that subtly influence the substrate-binding orientation. It is now understood that 24-SMTs studied to date catalyse the methyl transfer from SAM only on the S-face of the substrate double-bond in an SN2 reaction and the ensuing reversible 1,2-hydride shift from C-24 to C-25 on the R-face of the original double-bond undergoing methylation lead to regio-specific deprotonations at C-27 or C-28 [29–31]. These findings provide support for the suggestion that 24-SMTs have evolved divergently from a common ancestral protein, because the identity in the stereospecificity reflects the similarity in the active-site structure, in particular, in the geometrical relationship between the bound intermediate and the contact amino acids that participate in reaction channelling.
However, the soybean 24-SMT can also catalyse the second C1-transfer reaction from 24(28)-methylenecycloartanol to three products of the 24E- and Z-ethylidene and 24β-ethyl ∆25(27)-SCs. The paradoxical ability for 24(28)-methylenecycloartanol to convert into multiple products in an otherwise sterically constrained active site can be rationalized due to the improved access to the deprotonating base by the exocyclic double-bond for a 24H, 24CH3 or 26CH3 group. The active site amino acids directly involved in these channelling events are expected to be highly conserved and invariant in their spatial position. Thus the deprotonating base that promotes the formation of the isofucosteryl cation (24-ethyl C-24 cation intermediate) from the ∆24(28)-substrate is likely to be the same conserved residue used to deprotonate the C-24-methyl substituent of the cyclobranyl cation affording 24(28)-methylenecycloartanol. There is also prevalence for substrate-specific kinetic competition between slow deprotonation and fast alkylation of the bound cyclolaudenyl intermediate that can affect the product profile for C-24-methylation pathways. For example, as our results show the resulting ‘sessile’ C25 carbocation produced on methylation of 26-fluorocycloartenols can be quenched by the general catalytic base through alkylation, or the positive charge can be eliminated from the 1,2-hydride shift of H24 to C25 to produce the 24(28)-methylene product or from methyl-methylene elimination to generate the ∆25(27)-alkyl product.
What possible relevance does the cryptic ∆25(27)-alkylation pathway of 24-SMT have in soybean plants? This 24β-methyl sterol biosynthesis route is down-regulated in the soybean plant where the products of the ∆25(27)-alkylation pathway appears to serve no useful function to the plant physiology outside of possibly acting as membrane inserts. Alternatively, there are cogent reasons to believe that ∆25(27)-alkylated sterols may be harmful to plant growth because ∆25(27)-24-alkyl sterols bear the wrong C-24 orientation (β rather than α) for plant hormone (brassinosteroid) production [32,33]. In crop plant-insect ecology, these alkylated compounds fail to undergo de-alkylation by phytophagous insects , such that when ∆25(27)-alkyl sterols are fed to corn pests they disrupt normal larval growth and adult development . Therefore it would appear that, in the evolution of the C-24-methylation reaction pathway in plants, there was selection for the ∆24(28)-alkylation route that led to sitosterol.
Accordingly, how 24-SMTs specify product outcome has drawn a great deal of attention. In previous work, we have shown that single mutations in substrate-binding segments as well as in spatially distinct conserved amino acids in the primary structure can result in more promiscuous activity, which can lead to specificity switches [14,23,26]. Thus it appears the evolvability of the C-24-methylation pathway is correlated to residues that either line the 24-SMT-binding pocket directly or modulate the shape and chemical reactivity of the active-site surface through second-tier interactions. Although the structure–function analyses of the 24-SMTs are informative and clearly indicate that the structural framework for reaction channelling is highly conserved, more detailed information on the spatial determinants of mechanism and stereochemistry of these enzymes must await evaluation of the corresponding crystal structures.
Presheet Patkar and T. Thuy Minh Nguyen performed the kinetic and product analysis, Crista Thomas isolated and purified the substrates from commercial sources, and Brad Haubrich and Ming Qi prepared and characterized the substrate analogues. W. David Nes conceived the work, analysed the data and wrote the paper.
This investigation was supported by a National Science Foundation Grant [grant number MCB 0929212 (to W.D.N.)]. The support of the National Science Foundation for NMR spectroscopy instrumentation [grant number CHE-1048553] is gratefully acknowledged.
Abbreviations: a.m.u., atomic mass unit; CDCl3, deuterated chloroform; DAST, diethylaminosulfur trifluoride; EI, electron impact; i.d., inner diameter; M+, molecular ion; rel. int., relative intensity; RRTc, retention times relative to cholesterol; SAM, S-adenosylmethionine; SC, side chain; 24-SMT, sterol C-24-methyltransferase; THF, tetrahydrofuran
- © The Authors Journal compilation © 2013 Biochemical Society