Synthesis of bodinieric acids A and B, both C-18 and C-19-functionalized abietane diterpenoids: DFT study of the key aldol reaction

The first synthesis of C-18- and C-19-bifunctionalized abietane diterpenoids, bodinieric (or callicapoic) acids, via an aldol reaction has been developed. This key aldol reaction was very sensitive to steric hindrance. This fact has been studied by deuterium exchange experiments and DFT methods. Optimization of this reaction led to the synthesis of anti-inflammatory bodinieric acids A and B, starting from abietic acid.


Introduction
Diterpenoids of the abietane class have been known for almost 200 years since the discovery of abietic acid (1, Fig. 1) but new members are still being discovered and useful pharmacological applications are under study. 1 From the organic synthesis point of view, the preparation of abietanes has long been a matter of interesting research. 2 The abietane carbon framework characterized by a tricyclic ring system ( Fig. 1) with several stereocenters and a wide grade of oxygenation pattern on the skeleton invites synthetic chemists to be creative and efficient. The recent report, in 2018, on isolation of related abietane congeners of dehydroabietic acid (2, DHA) with both C-18-and C-19oxygenated carbons, such as bodinieric acid A (3, deacetylcallicapoic acid M5), and B (4, also known as callicapoic acid M4) attracted our attention. 3 Their important biological activities as a selective spleen tyrosine kinase (SYK) inhibitors have led to a patent. 4 These natural products reminded us an old synthetic work on another diterpenoid, possessing exactly the same stereochemistry at C4 and rare constitution, that is, both C18 and C19 oxidized and with the same oxidation pattern at this carbons in ring A, (À)-scopadulcic acid A (5, Fig. 1), as we had worked in the eld 20 years ago. 5 We became interested in developing a scalable synthetic route towards those acids for further study, including the control of the stereoselectivity at the quaternary stereocenter at C-4. Herein, we describe the rst synthesis of bodinieric acids A (3) and B (4) through a key aldol reaction of which we give further insight based on deuterium exchange experiments and DFT computational studies.

Results and discussion
The original retrosynthetic analysis is outlined in Scheme 1. In a similar approach to the work of Ziegler and Wallace to complete the synthesis of (AE)-scopadulcic acid A in 1995, 6 we envisaged that bodinieric acid B (4) could be readily obtained from the aldol 6 (Scheme 1) by Pinnick oxidation. Conversion of the isopropyl moiety into a methyl ketone moiety would then afford bodinieric acid A (3). Aldol 6 would be prepared by treatment with formalin and base of a mixture of known aldehydes 7a,b used in the synthesis of callitrisic acid (4-epidehydroabietic acid) from dehydroabietic acid (2) by Pelletier and Herald. 7 The necessary aldehydes for the key aldol reaction could be obtained aer opening of the epoxide 8, which in turn is synthesized from a mixture of olens 9 coming from an oxidative decarboxylation of DHA (2), obtainable from (À)-abietic acid (1).
With this straightforward synthetic plan in mind, we prepared DHA (2) from commercial (À)-abietic acid (1) following our method by methylation, aromatization with Pd/C catalyst and hydrolysis (Scheme 2). 1c, 8 We thought that the preparation of the exo-olen 4(18), though as a mixture of regioisomers, would be easy and fast by treatment of DHA (2) with lead tetraacetate. But like other researchers found, the reaction was capricious and low yielding in the desired exo-olen. 7,9 We sought an alternative and decided to use the methodology of Barrero and co-workers based on the elimination of a formate moiety (Scheme 2). 10 Thus, methyl dehydroabietate 1c (10) obtained from (À)-abietic acid (1) was converted into dehydroabietinal (11) in almost quantitative yield, which was oxidized with m-CPBA in the presence of disodium phosphate to furnish formate 12 in 73% yield. Heating in collidine at reux of 12 led to a mixture of olen regioisomers 9, ca. 5 : 1 : 1 (4,18-; 3,4-; 4,5-), in 87% yield where the major component contained the exocyclic 4,18-double bond. Continuing with our synthetic plan, epoxidation of that mixture of olens 9 (ca. 69% purity in exo-olen) with an excess of m-CPBA afforded epoxide 8, which was treated with BF 3 etherate in toluene to give a 1 : 1 mixture of epimeric aldehydes 7a and 7b (88%, two steps). The introduction of the remaining carbon at C-4 and manipulation of the functionality at this site were the remaining steps to be accomplished. To our surprise, our rst attempt of the planned aldol reaction 6 (aq. HCHO/ Na 2 CO 3 ) with aldehydes 7a,b (ca. 1 : 1 mixture) did not work as expected since only b-aldehyde 7a reacted, recovering unaltered a-aldehyde 7b. At this point, we hypothesized that using a stronger base such as NaOH might help but little progress in the aldol reaction was observed when using recovered a-aldehyde 7b and NaOH as a base, even with slow addition of aq. HCHO.
Initially, it was attributed this high stereoselectivity to the presence of the axial C-20 methyl moiety, which probably hinders the approach of the base to the most hindered b-face.
This fact was investigated by computational methods and deuterium exchange experiments (see below) in order to determine the causes of the different behavior in the enolization step and hydroxymethylation of both aldehyde epimers at C-4, 7a and 7b.
The completion of the syntheses of both bodinieric acids A and B is shown in Scheme 3. At this stage, the key aldol reaction of aldehyde 7a with aq. HCHO/Na 2 CO 3 rapidly gave the desired a stereoselectivity as a single isomer, hydroxy-aldehyde 6 in high yield (70%). Selective oxidation of the aldehyde group of 6 with NaClO 2 provided bodinieric acid B (4, callicapoic acid M4) in 82% yield, whose 1 H and 13 C NMR spectra (in CDCl 3 and acetone-d 6 ) and HRMS (calcd for C 20 H 29 O 3 [M + H] + : 317.2117; found: 317.2111) were in complete agreement with the reported data for the natural product 4. 3,12 Interestingly, Zhang and coworkers and Wang and co-workers dened wrongly this natural product as 18-hydroxydehydroabietic acid instead of 18hydroxycallitrisic acid which contains the C-19 carboxylic acid, and 18-hydroxy-8,11,13-abietatetraen-19-oic acid instead of 18hydroxy-8,11,13-abietatrien-19-oic acid, respectively. 3 Our next synthetic target was bodinieric acid A (Scheme 3). Starting from bodinieric acid B (4), selective dehydrogenation at C-15 with dichloro dicyano quinone (DDQ) provided the isopropenyl moiety in 14 (bodinieric acid C, as a ca. 3 : 2 mixture with unreacted starting material). Subsequent oxidative cleavage of 14 with catalytic OsO 4 and oxone as co-oxidant 13 furnished bodinieric acid A (3) in moderate yield (53% brsm, two steps) which was not further optimized (10.2% overall yield from abietic acid, 13 steps, see Scheme S1 ‡). The data for synthetic 3 were in excellent agreement with the isolation data ( 1 H NMR, 13 C NMR, [a] D ). 3a A conformational study of 3 was also carried out to estimate the 13 C data with the GIAO (gauge including atomic orbital) method, implemented in the Gaussian package (see Fig. S1 and S2 and Tables, S1 and S2 ‡). 14 Sixteen different conformations were obtained in acetone (solvent used in NMR experiments), the most stable being that indicated in Fig. 2. This conformation contains a hydrogen bond between the carbonyl of the acid group and the hydroxyl proton. A good correlation is observed between the experimental and theoretical 13 C NMR data. In particular, our experimental values for 3 gave a mean deviation of 1.82 (Table S2 ‡). Only some sensitive carbons to the rotation of the carboxylic group such as C5 give a high deviation.

Deuterium exchange experiments and computational study
To further get insight on the mechanism of the enolization, an experiment of deuterium exchange of hydrogens H-4 was designed and executed successfully. Thus, a ca. 1 : 1 mixture of aldehydes 7a/7b was subjected to equivalent conditions of the aldol reaction, without adding formaldehyde, with the corresponding solvents capable of releasing protons as deuterated versions (MeOD and D 2 O). The results (see ESI, Fig. S9-S12 ‡) indicate that aer 60 minutes of reaction the proton H-4 of isomer 7a (b-aldehyde) was exchanged by deuterium, while that of isomer 7b (a-aldehyde) apparently remained unaltered. It is believed that with longer reaction times the isomer 7b should also react but in the real system with formaldehyde this reaction will be disfavored since formaldehyde will decompose in basic media.
The enolate formation of both aldehyde epimers 7 using as base HO À and CO 3 2À was studied by DFT methods. In Scheme 4, there is a representation of the mechanism of enolate generation, starting from 7a and 7b. Initially, both epimers form a molecular complex (MC1) with the base, which evolves through the corresponding transition state (TS) to another molecular complex (MC2). MC2 is composed by the enolate and the conjugated acid of the used base. We have considered that ion HO À is solvated by three H 2 O molecules surrounding the oxygen with negative charge (see Fig. S7 ‡). 15 In a similar way, the anion CO 3 2À is also surrounded by three H 2 O molecules (see  Table S4 ‡ and in Fig. 3, the enolate formation of the epimer 7a (CHO-18b) with HO À occurs through the TSa with a Gibbs free energy difference DG ¼ 6.2 kcal mol À1 from MC1a. The enolate generation of the epimer 7b (CHO-18a) is clearly disfavored from the kinetic point of view (TSb, DG ¼ 12.5 kcal mol À1 from MC1b). Nevertheless, both enolate formations are very endergonic (MC2a, DG ¼ 3.5 kcal mol À1 and Scheme 3 Synthesis of bodinieric acids A and B. MC2b, DG ¼ 7.5 kcal mol À1 ). As a result, a very low concentration of complex MC2 is obtained and therefore, low concentration of enolate.
It must be considered that formaldehyde under the conditions of reaction is an aqueous solution stabilized by methanol. Those conditions provide a low concentration of formaldehyde, since there are several equilibria on reacting formaldehyde with water to give methylene glycol (HOCH 2 OH) and poly(oxymethylene)glycols and with methanol to form hemiformal (CH 3 OCH 2 OH) and poly(oxymethylene)hemiformals. 16 Thus, the low concentration of enolate and free formaldehyde makes the alkylation a slow reaction. This fact could be also accompanied by a side reaction, the Cannizzaro reaction, 17 that normally occurs under basic conditions. As a nal result formaldehyde decomposes without reacting with the enolate.
On the other side, the use of CO 3 2À as base changes the scenario notably. As it is shown in Table S4, ‡ Fig. 3 and 4, the reaction of 7b (CHO-18a) with CO 3 2À leads to MC2d through TSd. The Gibbs free energy difference is 8.1 kcal mol À1 but the process is still very endergonic (MC2d, DG ¼ 3.7 kcal mol À1 ) and therefore, disfavored. However, the enolate formation of 7a (CHO-18b) is fast (TSc, DG ¼ 6.9 kcal mol À1 ), leading to MC2c with only 0.1 kcal mol À1 above the energy of MC1c. This allows a high concentration of the complex MC2c and therefore a high concentration of the corresponding enolate. In the so basic media of CO 3 2À vs. OH À , the Cannizzaro reaction would be less important and, therefore, less decomposition of formaldehyde will occur. These two circumstances explain the experimental nding that only the epimer 7a reacts in a carbonate media (CO 3

2À
). As it can be seen in Fig. 4, both the transition state and the molecular complex resulting from the enolate formation of aldehyde 7b with CO 3 2À (TSd and MC2d, respectively), have a high steric hindrance with H-2b, H-6b and methyl 20. This is the main cause of the increase in energy of these species. This circumstance is even worse in the real situation since in the simulation only three water molecules have been considered and, in fact, there are more water molecules solvating the base. The use of a more real model would have several consequences: (a) on considering a higher number of explicit water molecules would imply the stabilization of the reactant base and the MC1 complex due to H-bonding as compared to the transition state, which raises the barrier height and hinders reaction. 18 (b) It is known that steric hindrance, i.e. S N 2 reactions, increases the energy of the transition state. 19 In our case, this effect would be similar especially for the steric characteristics of TSd, which would hinder the removal of the proton by the base.
(c) If we consider the effect of the solvated cation the steric effects would be even higher. 20 As an overall result, it can be deduced that in a more real computational model the reactants will be stabilized in comparison with the TS. Therefore, including more explicit water molecules would probably lead to slower reactions, 18a which is in agreement with the deuterium exchange experiment.

Conclusions
In summary, the rst stereoselective synthesis of both C-18 and C-19 oxygenated abietane diterpenoids, in particular, bodinieric acids A and B, has been completed using an aldol reaction as key step enabling the synthesis of additional congeners of this new group of natural products. Those metabolites were isolated very recently and had not been previously prepared synthetically. Our synthesized materials established and conrmed the originally proposed absolute stereochemistry of each natural product since the structure of the starting material (À)-abietic acid is already known by single-crystal X-ray determination. 21 The key aldol reaction was very sensitive to steric hindrance. This fact has been studied by deuterium exchange experiments and DFT methods. Further application of the intermediates obtained and biological evaluation of these synthetic natural products and related analogues are underway and will be reported in due course.

General experimental procedures
The melting points were measured with a Büchi M-560 apparatus. Optical rotations were measured using a 10 cm cell in  a Jasco P-2000 polarimeter in DCM unless otherwise stated. NMR spectra were recorded on a 300 MHz spectrometer ( 1 H: 300 MHz, 13 C: 75 MHz) and referenced to the solvent peak at 7.26 ppm ( 1 H) and 77.00 ppm ( 13 C) for CDCl 3 and 2.05 ppm ( 1 H) and 29.84 ppm ( 13 C) for acetone-d 6 . All spectra were recorded in CDCl 3 as solvent unless otherwise stated. Complete assignments of 13 C NMR multiplicities were made on the basis of DEPT experiments and further assignment of signals with the help of COSY and HSQC experiments. J values are given in Hz. MS data were acquired on a QTOF spectrometer. Reactions were monitored by TLC using Merck silica gel 60 F 254 (0.25 mm-thick) plates. Compounds on TLC plates were detected under UV light at 254 nm and visualized by immersion in a 10% sulfuric acid solution and heating with a heat gun. Purications were performed by ash chromatography on Merck silica gel (230-400 mesh). Commercial reagent grade solvents and chemicals were used as purchased unless otherwise noted. Combined organic extracts were washed with brine, dried over anhydrous MgSO 4 , ltered, and concentrated under reduced pressure. The starting material, methyl dehydroabietate (10), was obtained following our reported protocol 8 with the following modications: abietic acid (1, >70%, TCI) was methylated under standard conditions with dimethyl sulfate (1.2 equiv.) and K 2 CO 3 (1.3 equiv.) in acetone (0.33 M in abietic acid, 30 g scale). The resulting methyl abietate was aromatized with 10% Pd/C catalyst (1% of ester mass), heating at 240 C for 3 h. Next, compound 10 was reduced with LiAlH 4 (5 equiv.) in THF (0.23 M in methyl dehydroabietate, 8 g scale) and subsequently the crude product was oxidized with PCC (1.5 equiv.) in DCM (0.24 M in dehydroabietinol, 5 g scale) to give dehydroabietinal 11 (quantitative) which was used in the next step without further purication. The carbon numbering of all synthetic compounds corresponds to that of natural products.

Computational methods
All calculations were carried out with the Gaussian 09 suite of programs. 14 Initially, density functional theory 22 calculations (DFT) have carried out using the B3LYP 23 exchange-correlation functionals, together with the standard 6-31G** basis set, in gas phase. 24 Subsequently, the inclusion of solvent effects have been considered by using a relatively simple self-consistent reaction eld (SCRF) method 25 based on the polarizable continuum model (PCM). 26 Geometries have been fully optimized with PCM. As solvents we have used H 2 O. In the calculations, three water molecules have been explicitly included. The values of enthalpies, entropies and free energies in H 2 O were calculated with the standard statistical thermodynamics at 298.15 K. 24 The stationary points were characterized by frequency computations in order to verify that TSs have one and only one imaginary frequency. The intrinsic reaction coordinate (IRC) paths 27 were traced in order to check the energy proles connecting each TS to the two associated minima of the proposed mechanisms using the second order González-Schlegel integration method. 28 18-Norabieta-8,11,13-trien-4-yl formate (12). A suspension of m-CPBA (75%, 10.0 g, 0.043 mol, 2.5 equiv.) and Na 2 HPO 4 (6.7 g, 0.047 mol, 2.7 equiv.) in DCM (60 mL) was added to a solution of crude aldehyde 11 (5.0 g, 0.017 mol) in DCM (120 mL) and heated at reux for 4 h. Then, the mixture was cooled to rt and diluted with Et 2 O (150 mL), washed with saturated aqueous NaHCO 3 (3 Â 80 mL), brine (2 Â 50 mL), dried, and concentrated. The resulting oily yellow residue was chromatographed on silica eluting with n-hexane-EtOAc (9 : 1) to give 3.7 g (73%) of formate 12 as a yellowish oil: 19-Norabieta-4(18),8,11,13-tetraene (9). A yellowish solution of formate 12 (3.7 g, 0.012 mmol) in 2,4,6-collidine (16 mL) was heated at reux for 7 h. During this time, the solution became dark orange then cooled to rt and diluted with Et 2 O (110 mL). The mixture was washed with 10% HCl (3 Â 30 mL) and brine (3 Â 30 mL), dried and concentrated. The resulting dark orangered oil residue was dissolved in n-hexane and few drops of DCM. This solution was chromatographed on a short pad of silica eluting with n-hexane to give 2.75 g of olens 9 (87%, exocyclic D 4(18) ca. 69% purity estimated by 1 H NMR), containing the three corresponding regioisomers in an inseparable mixture by standard ash chromatography ca. 5 : 1 : 1 (4,18-; 3,4-; 4,5-). The 1 H and 13 C NMR data for the major exocyclic isomer are: 4a,19-Epoxide-18-norabieta-8,11,13-triene (8). A solution of the olens 9 (760 mg, 3.0 mmol) in DCM (70 mL) was treated with an excess of m-CPBA (75%, 1.2 g, 5.2 mmol, 1.7 equiv.) and stirred at rt for 3 h 30 min. The reaction mixture was quenched with 5% aqueous Na 2 S 2 O 3 (20 mL) and washed with a solution at 50% made with saturated aqueous NaHCO 3 (3 Â 20 mL), brine, dried and concentrated. The resulting crude epoxide (810 mg) was isolated as a yellowish oil and was used in the next step without further purication (it contained about 10% of 19norabieta-4(18),8,11,13-tetraen-3a-ol). Rearrangement of 4a,19-epoxide-18-norabieta-8,11,13-triene (8) (mixture aldehydes 7a,b). Boron triuoride etherate (400 mL, ca. 2 equiv.) was added at 15 C to a solution of the previous crude 4a-19-epoxide 8 (400 mg, 1.48 mmol) in dry toluene (10 mL) under Ar atmosphere. The reaction mixture was stirred for 3 min and saturated aqueous NaHCO 3 (1 mL) was added. Then, the mixture was diluted with Et 2 O (15 mL), washed with brine, dried and concentrated to give a dark pale oil (410 mg). The oily residue was carefully chromatographed on silica eluting with nhexane-EtOAc (9 : 1) to give 245 mg (88% based on the D 4(18)isomer in the alkenes 9 mixture) of epimeric aldehydes 7a,b (NMR data vide infra) as a colorless oil which were used in the next step directly (aldol reaction).