Oxidative Functionalization of Trinor-18α-olean-17(22)-ene Derivatives. Annulation of the E-Ring by an Intramolecular Aldol Reaction

cis-Dihydroxylation of trinor-18α-olean-17(22)-ene 2 with osmium tetroxide led to diol 9. Its cleavage with lead tetraacetate gave tetracyclic ketoaldehyde 10. By comparison, the ozonation of trinor-18α-olean-17(22)-ene 2 in the presence of p-toluenesulfonic acid gave the corresponding ketoacetal 12. Both products were subjected to an intramolecular aldol reaction under the acidic conditions and yielded unusual triterpenes bearing a bicyclo[4.3.1]decane fragment (22). Further manipulation of the protective groups afforded compounds useful in triterpene synthesis, especially in the preparation of potentially biologically active saponins based on a tetracyclic terpene core.


■ INTRODUCTION
Birch is a hardwood tree of the genus Betula widespread in the Northern Hemisphere, particularly in areas of temperate and boreal climates. Its wood has had important historical and cultural significance since ancient times. 1,2 Nowadays, birch wood is widely used in the paper industry. The outer part of birch bark (the waste generated during paper production) is an extremely rich source of lupane-type triterpenes, mostly betulin (1, Scheme 1), which are isolated in appreciable amounts (up to 30% of dry mass). 3,4 Betulin has interesting biological properties, including antiviral and antitumor activities. 5−8 Lupane triterpenoids have also been used as intermediates in the synthesis of triterpenoids employed for biological studies. 9−18 In the pursuit of new, flexible starting materials for the functionalization of the lupane core, 19−22 we guessed that the easily available 3β-O-acetyl-trinor-18α-olean-17(22)-ene (2) 23 is a candidate for further transformation (Scheme 1). Special interests for us were the tetracyclic compounds with the general formula 3, which can be obtained by oxidative cleavage of the double bond in 2. They are structural analogues of radermasinin (4), a cytotoxic triterpene lactone isolated from Radermachia sinica. 24 In the present paper, we report on a simple and high yield conversion of the olean-17 (22)-ene scaffold by an oxidative functionalization of the double bond. The obtained derivatives may be used as starting materials for the synthesis of saponins and more complex triterpene derivatives.
In the first attempt, the dihydroxylation of 2 with an excess of OsO 4 in pyridine afforded cis-diol 9 in high yield (80%) as a single diastereoisomer; the complete oxidation of the double bond took 7 days (Scheme 3). Catalytic dihydroxylation of 2 in catalytic version (OsO 4 /NMO) gave no product. 26,27 According to the well-known mechanism of dihydroxylation with OsO 4 , only the cis-diol was formed. We expected that the steric interaction of D-and E-ring protons precluded formation of 17β,22β-diol (Figure 1). By a comparison, no steric interaction influenced the OsO 4 attack from the α-side. The 17α,22α configuration of the new stereogenic centers in 9 was determined by the NMR experiments (see Supporting Information).
Then, we studied the oxidative cleavage of the 1,2-diol group in 9, which, as we expected, should afford ketoaldehyde 10 as a main product. Treatment of 9 with NaIO 4 on silica gel 28 for 7 days, gave a single product (Scheme 3). Despite a prolonged time of the reaction, most of the starting material (53%) was recovered. In the 1 H NMR spectra of the product, a doublet of the −CHO proton was detected at δ = 9.81 ppm. Unexpectedly, only one signal of the carbonyl group at δ = 206.5 ppm, belonging to the aldehyde moiety, was observed in the 13 C NMR spectra. Analysis of HMBC and NOE correlations suggested that a pentacyclic product 11 was obtained (in 27% yield) instead of the expected tetracyclic 10. Therefore, in the next attempt, we used lead tetraacetate in pyridine as an oxidizer. At room temperature, two products were obtained. In the 1 H NMR spectrum of the main product, a typical multiplet of the −CHO proton was observed at δ = 9.75 ppm, whereas both expected signals of the carbonyl groups belonging to the ketone (δ = 202.4 ppm) and aldehyde groups (δ = 213.7 ppm) were present in the 13 C NMR spectra. Further analysis of HMBC and NOE correlations confirmed that, in this case, compound 10 was isolated as the main product (78%), whereas 11 (7%) was identified as the minor Scheme 1. Proposed Transformation of Betulin Scheme 2 a a Reagents and conditions: i. 10% Pd/C, H 2 (7 bar, quantitatively); ii. i-PrOH, Al(i-PrO) 3 , reflux, 75%; iii. POCl 3 , pyridine (98%).

Scheme 3 a a
Reagents and conditions: i. OsO 4 , pyridine (80%); ii. NaIO 4 supported on SiO 2 (27%); iii. Pb(OAc) 4 , pyridine (78% of 10, 7% of 11); iv. Pb(OAc) 4 , benzene (90% of 10); v. O 3 , methanol, CH 2 Cl 2 (45−59% of 12, 15−21% of 11, 3−7% of 13, and 11−15% of 14); vi. O 3 , p-TsOH, methanol, CH 2 Cl 2 (71% of 12); vii. NaClO 2 , amylene; viii. Jones reagent, acetone; ix. MeOH, HCl (75% of 17 after two steps; 98% of 20); x. MeI, K 2 CO 3 , DMF (75% of 13 after two steps; 80% of 19 after two steps); xi. Ac 2 O, Py (95%); xii. (a) LiAlH 4 , (b) Ac 2 O, Py (79% after two steps). The Journal of Organic Chemistry pubs.acs.org/joc Article product. Presumably, the pentacyclic compound 11 was formed in an intramolecular pyridine catalyzed aldol reaction as the basic component of this reaction. To confirm this assumption, we kept aldehyde 10 in pyridine at 50°C for 12 h. However, no cyclization was observed. Apparently, the mechanism of this reaction is more complex and requires the presence of a metal ion. To avoid formation of 11, we have tested benzene as a neutral solvent. As a result, tetracyclic aldehyde 10 was obtained as a sole product in 90% yield. No traces of 11 were detected in the reaction mixture. Both products (10 and 11) have limited shelf life. Further optimization was focused on excluding the use of toxic OsO 4 . With alkene 2 in hand, we tried to run ozonation of the C17(22) double bond to prepare 10. Because of the low solubility of 2 at −78°C, we used a dichloromethane− methanol mixture as a solvent system to provide sufficient solubility of the starting material. Under these conditions, the ozonation of 2 was fast; however, a mixture of products was obtained (Scheme 3). 29,30 Careful separation of the reaction mixture and analysis of the isolated products revealed that four compounds were obtained in this reaction. It must be noted that the composition of the mixture and the yields of the products changed significantly from batch to batch. Usually, acetal 12 (45−59%) and pentacyclic aldehyde 11 (15−21%) were isolated as the main products. In some cases, 13 (3−7%) and 14 (11−15%) were also isolated as byproducts. Surprisingly, the expected aldehyde 10 was not formed during the ozonation of 2.
The formation of acetals and esters is rather unexpected during the ozonolysis under neutral conditions. However, in his seminal publication, Schreiber has shown 31 that the incorporation of an acid or base during the ozonolysis of cycloalkenes, followed by a reductive workup, promoted a formation of differently substituted products, including acetals and esters, in high yields. 32,33 As suggested, such products were formed by the addition of methanol, a participating solvent, to the carbonyl oxide 15, reactive intermediate in the Criegee mechanism, 34 and subsequent transformation of α-alkoxy hydroperoxide 16 during workup (Scheme 4). It is interesting that, in the case of ozonation of 2, acetal 12 and ester 13 were formed in the absence of a catalyst. We supposed that the unpredictability of the ozonation of 2 and fluctuation in the product proportions were caused by an uncontrolled decomposition of α-alkoxy hydroperoxide 16 under neutral conditions. To confirm the above assumption, we repeated this reaction under the Schreiber's conditions. The ozonation of 2 in the presence of NaHCO 3 caused decomposition of the starting material. By comparison, the reaction performed in the presence of p-TsOH (approximately 10% w/w) afforded acetal 12 as a sole product in high yield (71%, Scheme 3).
All compounds prepared above (10−12) are valuable starting materials in the synthesis of differently substituted analogues. Therefore, in the next step, we have tested their reactivity and scope of possible transformations. We were especially interested in the synthesis of compounds bearing the free −OH groups, starting materials for the preparation of saponins (triterpene glycosides). 35 Then, we initially converted aldehyde 10 into acid 14 by its oxidation using a procedure developed by Clive. 36 The same reaction can be performed using the Jones reagent. Acid 14 was then transformed into The Journal of Organic Chemistry pubs.acs.org/joc Article methyl ester by treatment with acidic methanol which afforded 17, bearing a free 3β-OH group (75% yield after two steps). Moreover, treatment of 14 with methyl iodide in the presence of potassium carbonate afforded fully protected ester 13 in 75% yield. The same compound 13 was obtained by acetylation of 17 under the standard conditions in 95% yield. Similar oxidation of 11 to free acid 18 with the Clive's method, followed by esterification with methyl iodide, gave ester 19 (80% yield after two steps). Treatment of 19 with acidic methanol afforded ester 20 with the free 3-OH group (98%). The reduction of 11 with LiAlH 4 , followed by acetylation of the crude reaction mixture, yielded diacetate 21 (79% after two steps). Its structure was confirmed by a single-crystal X-ray analysis ( Figure 2).
Finally, we have attempted the hydrolysis of acetal 12 into aldehyde 10 by a treatment with p-TsOH in acetone. This reaction required high loading of acid (at least 40% w/w) for a complete transformation. Column chromatography of the reaction mixture gave two fractions. The first contained a compound which was identified as 11 (18%, Scheme 5). The second fraction was composed of an inseparable mixture of two epimeric compounds (22, 56%). They were chromatographically separated after acetylation under standard conditions. On the basis of the analysis of the NMR spectra, we proposed their structures as diacetates 23 (46%, after two steps) and 24 (28%, after two steps). Both structures were confirmed by single-crystal X-ray analysis ( Figure 2). Similarly, treatment of aldehyde 10 with p-TsOH in acetone afforded the same products in an identical ratio. Notably, the subjecting of 12 to acidic conditions led to the hydrolysis of the acetal function and the formation of aldehyde 10, which immediately cyclized by intramolecular aldol reaction, 37 affording cyclic products 11 and 22. When acetone was replaced by toluene, the cyclization of 10 was highly selective toward 22 (74%). Similarly, cyclization of acetal 12 in toluene solution gave the epimers 22, but in lower yield (50%). In both cases, only traces of 11 were detected. Neither equilibration nor retro-aldol reactions were observed when compounds 11 and 22 were subjected to acidic conditions. Compounds 22−24 belong to modified triterpenes with an unusual bicyclo[4.3.1]decane framework. 38−43 To the best of our knowledge, compounds bearing the bicyclo[4.3.1]decane motif in their structure have never been synthesized by an intramolecular aldol reaction. This methodology was, however, used for the preparation of derivatives with the bicyclo[3.3.1]nonane fragment, 44−46 found in some natural compounds. 47−50 The possible reaction mechanism is presented in Scheme 6. When an aldehyde's carbonyl group is protonated (structures A and B), cyclization leads to product 22 having a sevenmembered ring. Protonation of the ketone's carbonyl group (structures C and D) should result in the formation of fivemembered ring products. In this case, however, compound 11α was formed selectively. Probably, the formation of 11β is precluded by steric interactions of the aldehyde group with protons of the ring D and the angular methyl group at the C8. Stereochemistry of the newly generated stereogenic center at C16 (in 22) and C17 (in 11) was determined by the structure of aldehyde 10; formation of the new carbon−carbon bond is possible only from the β-side of the D-ring.

■ DFT CALCULATIONS
Initially, 18 structures of 10 differing in the arrangement of the side chains were considered, while the remaining part of the molecules were unchanged. The analysis of energy revealed that only three structures have a significant population (0.22, 0.13, and 0.48 molar fractions, Figure 2S, rotamers I, II, and III), having an aldehyde group far from the ketone fragment. In the next step, the structures of four tautomers derived from III (i.e., from the low-energy structure) were optimized, and their molecular energies were estimated ( Figure 3S). An analysis of the molecular energies of III and its tautomers together with the assumption of equilibrium between species indicated a negligible molar fraction of tautomeric forms.
In the third step, calculations were performed for the protonated compounds. All input structures (cations) were constructed starting from the most stable rotamer III and related tautomers. A proton was located at either the ketone or aldehyde groups ( Figure 4S). The structure with H + at the ketone group appeared to be the most stable. However, the most interesting results were obtained starting from formally unfavorable rotamers. Optimization resulted in structures stabilized by hydrogen bonds (HB) and/or other weak interactions, such as C···O, CH···O, and OH···C ( Figure 3).

Scheme 6. Tentative Reaction Mechanism
The Journal of Organic Chemistry pubs.acs.org/joc Article In one case, we observed the transformation of aldehyde to enol, forced by the transfer of H ( Figure 4). Twice, the ring closure occurred during optimization ( Figure 5).
The last series of calculations concerned the analysis of the hydrogen bonds expected for 11. Three initial structures differing in the arrangements of the OH bonds were   The Journal of Organic Chemistry pubs.acs.org/joc Article considered ( Figure 6), the first with the aldehyde group CH(OH + ) involved in HB formation, the second structure without HBs, and the third one HB formed by the ketone group (COH + ). Surprisingly, an open chain product has been obtained during optimization in the last case, with molecular energy lower than cyclic XI without HB. This "linear" form was stabilized by the C···C, O···O, and OH···C interactions. The optimization process appeared to be reversible; cyclic 11 with HB has been obtained during optimization when a "linear" reagent with the right OH bonds pattern was used as the input structure. Concluding, the first calculations revealed that rotamers of 10 having the CHO and CO functionalities far from one another seem to be preferential ones. However, the optimization and examination of seemingly disadvantageous rotamers of protonated 10 resulted in some structures having both CO groups close to each other. Analysis by Atom-In-Molecule (AIM) methodology shows that these rotamers are stabilized by numerous weak interactions, such as OH···O, CH···O, and C···HO hydrogen bonds, and/or interactions between C and O atoms. Weak interactions preorganize 10, enabling the reacting fragments to come closer, and facilitating certain reactions. In particular, the ring closure leads to 11 or 22, depending on the conformation of the reagent. The results of the calculations are in agreement with the proposed reaction mechanism.

■ CONCLUSIONS
The presented results clearly show that the oxidative functionalization of the double bond of trinor-18α-olean-17(22)-ene derivatives led to synthetically useful products. The proper choice of oxidants and reaction conditions gave highly functionalized tetracyclic triterpenes and unusual products with the bicyclo[4.3.1]decane fragment. In all cases, the key compounds were obtained as the major products. Prepared derivatives are convenient starting materials for further synthesis of tetracyclic terpene analogues, especially the synthesis of saponins. The most interesting achievement is the synthesis of triterpenes bearing a bicyclo[4.3.1]decane fragment. To our best knowledge, this is the first observation of an intramolecular aldol reaction leading to this unusual structure. 51  ■ EXPERIMENTAL SECTION General Procedures. Silica gel HF 254 and Silica gel 230−400 mesh (E. Merck) were used for TLC and column chromatography, respectively. 1 H and 13 C{ 1 H} NMR spectra were recorded at 298 K with Varian NMR-vnmrs600 or vnmrs500 spectrometers, using standard experimental conditions and Varian software (ChemPack 4.1). Configurational assignments were based on the NMR measurements generated using two-dimensional techniques like COSY and 1 H− 13 C gradient selected HSQC (g-HSQC), as well as 1 H− 13 C gradient selected HMBC (g-HMBC). Internal TMS was used as the 1 H and 13 C NMR chemical shift standard. J values are given in Hertz. High-resolution mass spectra (HRMS ESI) were acquired with Mariner and MaldiSYNAPT G2-S HDMS (Waters) mass spectrometers. Optical rotations were measured with a Jasco P-2000 automatic polarimeter. Single crystal X-ray diffraction measurements were carried out on an Agilent Supernova diffractometer at 100 K with monochromated Cu Kα radiation (1.54184 Å). The structures of compounds 21, 23, and 24 were determined on crystals prepared in a chloroform/methanol solvent system by slow evaporation at room temperature.
All DFT calculations have been performed using the Gaussian program suite. 53 Molecular energies (a.u.) have been estimated at the B3LYP/6-311++G(2d,p) level of theory, using B3LYP/6-31G(2d,p) optimized structures. All calculations were performed assuming isolated molecules. The protonated compounds were calculated as cations. Topological analysis: detection of the weak interactions, hydrogen bonds, and bond critical points was performed by the AIMAll program package. 54 Some preliminary calculations were The Journal of Organic Chemistry pubs.acs.org/joc Article performed using simplified structures ( Figure 1S) to save computational time; then the results were used to plan the calculations for full molecules 10, 11, and 22. The population of a series of rotamers was estimated on the basis of molecular energy, using the Boltzmann distribution. The details of the calculations, some key figures as well as atomic coordinates are enclosed in the Supporting Information. 20,29-Dihydrobetulin 3β,28-di-O-acetate (7). Betulin diacetate (5) was converted to dihydrobetulin diacetate (7) by a modified procedure of Lehn. 55 To a solution of betulin diacetate (5, 5.27 g, 10.00 mmol) in THF (70 mL) and methanol (70 mL) was added 10% Pd/C (300 mg), and the mixture was hydrogenated under 7 bar of hydrogen for 48 h. Then, the whole mixture was filtered through a short silica pad (hexane−ethyl acetate−methanol, 5:3:1 as an eluent) to afford dihydrobetulin diacetate (7, 5.25 g, quant) as a white solid. 56 No further purification was necessary. 1  To a solution of betulin 3β-O-acetate (6, 3.15 g, 6.50 mmol) in THF (35 mL) and methanol (55 mL) was added 10% Pd/C (230 mg), and the mixture was hydrogenated under 7 bar of hydrogen for 3 days. Then, the whole mixture was filtered through a syringe filter to remove the catalyst and evaporated to dryness under reduced pressure. No further purification was necessary. Dihydrobetulin 3β-acetate (8) was obtained quantitatively as a white solid. 56 Method B. Dihydrobetulin 3β-O-acetate (8) was prepared from dihydrobetulin 3β,28-di-O-acetate (7) by selective deacetylation according to a modified procedure of Thibeault. 25 A mixture of dihydrobetulin 3β,28-di-O-acetate (7, 10.00 g, 18.91 mmol), Al(i-OPr) 3 (11.65 g, 57.04 mmol), and i-PrOH (300 mL) was stirred under reflux for 24 h. The crude mixture was concentrated under reduced pressure and water (300 mL) was added. The suspension was slightly acidified with 2 M HCl and extracted with chloroform (3 × 100 mL). The combined organic layers were washed with saturated NaHCO 3 (20 mL) and concentrated under reduced pressure. Column chromatography of the residue (hexane−ethyl acetate, 15:1 to 10:1) afforded 6.86 g (75%) of 8 as a white solid. 1  3β-O-Acetyl-19α-isopropyl-28,29,30-trinor-18α-olean-17(22)-ene (2). Compound 2 was prepared from dihydrobetulin 3-O-acetate (8) according to the procedure published for betulin 3-Oacetate. 57 POCl 3 (19.5 mL, 210 mmol) was added to a solution of dihydrobetulin 3-acetate (8, 6.33 g, 13.00 mmol) in anhydrous pyridine (50 mL) and heated at 60°C in an oil bath for 24 h. Then, the mixture was carefully poured onto ice (500 g). The product was extracted with chloroform (3 × 100 mL), and the organic extracts were concentrated under reduced pressure and filtered through a short silica path. Organic solvents were evaporated; column chromatography of the residue (hexane−ethyl acetate, 40:1 to 20:1) gave the title compound (2, 5.98 g, 98%) as a foam, sufficiently pure for further transformation. [ 3-O-Acetyl-19α-isopropyl-28,29,30-trinor-17α,18α-oleanan-3β,17α,22α-triol (9). 58 To a solution of 2 (1.000 g, 2.133 mmol) in pyridine (30 mL) was added OsO 4 (600 mg, 2.36 mmol), and the mixture was stirred in the dark for 7 days. Then, pyridine was co-evaporated with toluene under reduced pressure, and the residue was dissolved in ethyl acetate. Water (40 mL), Na 2 S 2 O 5 (2.0 g), and Na 2 S 2 O 3 ·5H 2 O (3.0 g) were added, and the mixture was stirred until decomposition of osmate ester was detected on TLC (2−3 days). Then, water (100 mL) was added, and the product was extracted with chloroform (3 × 30 mL). Combined organic extracts were concentrated and the residue was purified by column chromatography Compounds 10 and 11. Method C. To a vigorously stirred suspension of silica gel (230−400 mesh, 400 mg) in CH 2 Cl 2 (4 mL) was added a solution of NaIO 4 (56 mg, 0.26 mmol) in water (1 mL), and the heterogeneous mixture was stirred for 15 min to form a flaky suspension. Diol 9 (100 mg, 0.200 mmol) in CH 2 Cl 2 (5 mL) was then added and stirred for 7 days. The mixture was filtered through sintered glass, silica gel was washed with CH 2 Cl 2 , and the solvents were evaporated under reduced pressure. The residue was purified by column chromatography (hexane−ethyl acetate, 9:1 to 5:1) to afford 11 (27 mg, 27%) as a foam and recovered diol 9 (53 mg, 53%).
Method D. A mixture of diol 9 (201 mg, 0.400 mmol) and lead tetraacetate (320 mg, 0.65 mmol) in pyridine (10 mL) was stirred at room temperature for 30 min. Then, two drops of glycerin were added to decompose an excess of lead tetraacetate and the solvents were co-evaporated with toluene under diminished pressure. The residue was purified by column chromatography (hexane−ethyl acetate, 9:1 to 5:1) to afford 10 (156 mg, 78%) and 11 (14 mg, 7%), both as a foams.
Method E. A mixture of diol 9 (402 mg, 0.800 mmol) and lead tetraacetate (600 mg, 1.20 mmol) in benzene (15 mL) was stirred at room temperature for 30 min. Then, two drops of glycerin were added to decompose an excess of lead tetraacetate and the solvents were evaporated under diminished pressure. The residue was purified by column chromatography (hexane−ethyl acetate, 9:1 to 5:1) to afford 10 (360 mg, 90%) as a foam.  Ozonolysis of 2. Method F. Ozone was bubbled through a solution of 2 (4.69 g, 10.00 mmol) in MeOH (100 mL) and CH 2 Cl 2 (100 mL) at −78°C until the disappearance of the starting material on TLC (30 min). Oxygen was passed through the solution for an additional 15 min to remove an excess of ozone, and Me 2 S 2 (10 mL) was added. The mixture was then left to warm to room temperature and stirred overnight. Solvents were evaporated under reduced pressure and the residue was purified by column chromatography (hexane−ethyl acetate, 9:1 to 5:1) to afford 13 (320 mg, 6%), 58 12 (3.06 g, 56%), 11 (902 mg, 18%), and crude 14 (565 mg, 11%) in order of appearance, all as foams.
Method G. Ozone was bubbled through a solution of 2 (910 mg, 1.94 mmol) and p-TsOH (100 mg) in MeOH (25 mL) and CH 2 Cl 2 (25 mL) at −78°C until the disappearance of the starting material on TLC (30 min). Oxygen was passed through the solution for an additional 15 min to remove an excess of ozone. Then, the mixture was stirred for 1 h at room temperature to ensure the complete acetal formation and Me 2 S 2 (1 mL) was added. The reaction was worked up following Method F. Acetal 12 (754 mg, 71%) was obtained as the sole product.
Data  Compound 14. Method H. Aldehyde 10 (201 mg, 0.400 mmol) was dissolved in a mixture of THF (5 mL), tert-BuOH (15 mL), and 2-methyl-2-butene (5 mL). The solution was cooled in an ice bath, and a solution of NaH 2 PO 4 ·2H 2 O (600 mg) and NaClO 2 (720 mg) in water (10 mL) was added. The solution was stirred at 0°C for 10 min; then the temperature was raised to room temperature and stirring was continued for 30 min. A saturated solution of NH 4 Cl (0.5 mL) and 15 mL of water were added. Product was extracted with dichloromethane (3 × 100 mL), and the combined organic extracts were evaporated to dryness. Short column chromatography of the residue (hexane−ethyl acetate, 9:1 to 5:1, and hexane−ethyl acetate− methanol, 5:3:1) gave crude acid 14 (202 mg) as an amorphous powder. 58 Method I. To a cooled in an ice bath solution of 10 (100 mg, 0.200 mmol) in acetone (10 mL) was added Jones reagent (0. 8 mL) dropwise, and the mixture was stirred at room temperature for 1 h. Then, isopropanol (1 mL) was added and stirring was continued for an additional 20 min. The solution was decanted, and the precipitated solid mass was washed with acetone (4 × 10 mL). Combined organic extracts were evaporated under reduced pressure and the residue was purified by column chromatography (hexane−ethyl acetate, 5:1, to hexane−ethyl acetate−methanol, 5:3:1) to afford crude acid 14 (100 mg) as amorphous powder.