Regio- and Stereocontrolled Synthesis of Oligostilbenoids: Theoretical

26 Feb 2013 - ... evaluation of their radical scavenger activities. Elisa Beneventi , Silvia Conte , Maria Rita Cramarossa , Sergio Riva , Luca Forti...
0 downloads 0 Views 563KB Size
Article pubs.acs.org/jnp

Regio- and Stereocontrolled Synthesis of Oligostilbenoids: Theoretical Highlights at the Supramolecular Level Saraswati S. Velu,†,‡,# Florent Di Meo,§,# Patrick Trouillas,*,§,⊥,∥ Juan-Carlos Sancho-Garcia,▽ and Jean-Frédéric F. Weber† †

Atta-ur-Rahman Research Institute for Natural Product Discovery (RiND), Faculty of Pharmacy, Universiti Teknologi MARA, Campus Puncak Alam, 42300 Bandar Puncak Alam, Selangor D. E., Malaysia ‡ Jeffrey Cheah School of Medicine and Health Sciences, Monash University Sunway Campus, Building 3, Jalan Lagoon Selatan, Bandar Sunway, 46150, Selangor D. E., Malaysia § Laboratoire de Chimie des Substances Naturelles, Faculté de Pharmacie, Université de Limoges, 2 Rue du Docteur Marcland, F-87025 Limoges, France ⊥ Laboratoire de Chimie des Matériaux Nouveaux, Université de Mons, Place du Parc 20, B-7000 Mons, Belgium ∥ Regional Centre of Advanced Technologies and Materials, Department of Physical Chemistry, Faculty of Science, Palacký University of Olomouc, tr. 17 listopadu, 771 46 Olomouc, Czech Republic ▽ Departamento Química Física, Universidad de Alicante, Apartado de Correos 99, E-03080, Alicante, Spain S Supporting Information *

ABSTRACT: Oligostilbenoids (e.g., ampelopsin F, viniferin, pallidol) result from homogeneous or heterogeneous coupling of monomeric stilbenoid units, leading to various chemical structures. Oligostilbenoid synthesis is regio- and stereocontrolled. To tackle this regio- and stereocontrol, a supramolecular chemistry approach is required that can be achieved by quantum chemistry. The stability of noncovalent π-stacks, formed between two stilbenoid units prior to oxidation, is accurately evaluated with density functional theory (DFT) including dispersive effects (within the DFT-D formalism). These noncovalent arrangements drive the regiocontrol. The rest of the chemical pathway is a succession of dearomatization and rearomatization stages. The thermodynamics and kinetics of the processes are calculated with classical hybrid functionals. This study allows discrimination between the two main possible chemical pathways, namely, radical−neutral and radical−radical reactions. The former appears more likely, thermodynamics and kinetics being in perfect agreement with the experimental 1:2 ratio obtained for ampelopsin F:pallidol analogues, respectively.

N

antifungal,5−7 antioxidant,8,9 anti-HIV,10 cytotoxic,10−12 and anti-inflammatory.13,14 Over the past decade, the synthesis of oligostilbenoids has received particular attention; a seminal contribution concerning the “programmable synthesis design” of resveratrol oligomers was recently proposed.15−17 The literature usually describes stilbenoid oligomerization as a classical phenolic oxidative coupling. This approach is globally correct but not sufficient to understand the apparent inconsistencies concerning regio- and stereoselectivity, leading

aturally occurring oligostilbenoids form a specific group of polyphenolic compounds. In spite of their relatively small number, they have a significant economic impact, as they are constituents of widely used plant species, e.g., dipterocarp timber trees from Southeast Asia and grapevine.1,2 Their attractiveness originates from their structural diversity, which includes rings of unusual sizes, various types of fused rings with or without oxygen atoms, and numerous stereogenic centers. Biogenetically, oligostilbenoids result from homogeneous or heterogeneous coupling of monomeric stilbenoid units (i.e., resveratrol, isorhapontigenin, pterostilbene) into dimeric to octameric species (Figure 1). The compounds from this group exhibit diverse biological activities including antibacterial,3,4 © 2013 American Chemical Society and American Society of Pharmacognosy

Received: October 9, 2012 Published: February 26, 2013 538

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

refers to the dimerization process initiated by a single, noncontrolled, experimental condition (presence of hard Lewis acid).7 In this case, such supramolecular arrangements would provide the regio- and stereocontrol. In an attempt to validate this hypothesis, the present work provides a quantum-calculation-based description of π-stacking interactions involved in pterostilbene dimerization into analogues of ampelopsin F and pallidol. Section 2 justifies the choice of the method of calculation to properly describe π−π interaction (Section 2.2), density functional theory (DFT) including dispersive correction being recommended in this case. Section 3.1 describes the oxidative initiation and the formation of the subsequent phenoxy radical. Section 3.2 proposes a molecular description of the π-stacking complexes, showing the different stabilizing contributions. The coupling may occur after oxidative initiation of at least one monomer and the presence of the subsequent phenoxy radical (see Scheme 1).18 Then, two types of reaction pathway are suggested in solution (Section 3.3): the oxidative coupling of one phenoxy radical monomer with (i) a native monomer (i.e., radical−neutral reaction) or (ii) another phenoxy radical monomer (i.e., radical−radical reaction). Section 3.3 also provides the thermodynamics of the entire mechanism of action, and the initial bond formation is kinetically rationalized.



METHODS OF CALCULATION

Reaction Mechanism. Over the past decades, DFT methods have been widely used to study the electronic structure of natural compounds, allowing high accuracy at a relatively reasonable computational time with respect to post-Hartree−Fock (HF) methods. Oxidative reactions of polyphenols (i.e., oxidation by Hatom transfer and radical coupling) were accurately evaluated within the classical hybrid DFT framework, the B3P86 functional being particularly well adapted.28−30 The 6-31+G(d,p) basis set is used since it provides similar results to the larger and more computationally demanding 6-311+G(2d,3pd) basis set.28 Geometries and energies including the zero-point correction (V), enthalpies (H), and Gibbs energies (G) at 298 K of reactants, intermediates, and products were calculated at the B3P86/631+G(d,p) level of theory. The ground-state geometries are confirmed by vibrational analysis that indicated the absence of imaginary frequencies. Transition states (TSs) are confirmed by the presence of one imaginary frequency assigned to the normal mode of the corresponding reaction coordinate. The capacities of H-atom transfer (HAT) from a given compound R-H and inducing radical formation from pterostilbene, RNA3, and RNP3 (Scheme 1) are evaluated by the bond dissociation enthalpy (BDE) calculated as follows:

Figure 1. Chemical structures of stilbenoid monomers and dimers.

to the formation of the various specific skeletons under different reaction conditions (e.g., solvents and/or metal oxidants). Trying to understand the biosynthesis of these compounds, we came up with a conceptual framework clearly showing regio- and stereoselectivity. This regio- and stereoselectivity was also observed in vitro, in which reaction conditions were able to control the types of dimers that were synthetized.7 An in-depth analysis of the outcome of stilbene oxidative coupling18,19 led us to derive a key hypothesis:18 πstacking interaction is the driving force to achieve specific regioand stereoselectivity, allowing self-association of two stilbenoid partners in solution, thus orienting further reactions. π−π-Interactions are involved in many natural processes including (i) copigmentation in plants,20 (ii) ligand−protein and ligand−nucleic acid interactions,21 (iii) solid-state arrangements,22 and (iv) charge transfer.23,24 π-Stacking has already been suggested as an important contribution in chemical reactions contributing to regio- and stereospecificity.18,25 πStacking interactions in the solid state allow control of the photodimerization of olefins.26 Similar to how benzene and hexafluorobenzene produce face-to-face stacks, (E)-pentafluorostilbene crystallizes with long stacks of alternating phenyl and pentafluorophenyl rings.26 These stacks produce a single isomer of the cyclobutane photodimer. Stilbenoids are fully π-conjugated systems and may in principle interact noncovalently through π−π interactions in solution. An NMR-based dynamical and structural study of resveratrol in DMSO-d6/D2O showed that the molecule can engage in strong autostacking interactions.27 In stilbenoid oligomerization, π−π interactions would play a crucial role in the self-assembly of stilbenoids prior to the oxidation and phenolic radical coupling, which then would determine the type of skeletons that are produced. The present theoretical work

BDE(R − H) = H298K(R − H) − H298K(R•) − H298K(H•) (1) where H298K refers to the electronic plus the corrections to enthalpy obtained at 298 K. The hard and soft (Lewis) acid and base (HSAB) principle can be quantified by the chemical hardness (η), which is calculated as follows:

η=

I−A 2

(2)

where I and A are the adiabatic ionization potential and the adiabatic electron affinity, respectively. This is a global parameter obtained to quantify the hardness or softness of a given molecular system in terms of its acid and base behavior. It allows rationalizing the global capacity of a compound to react according to the HSAB principle. The Fukui function f k(r) was developed to provide the atomic picture of this reactivity.31 For a given atom k, it is calculated as follows: 539

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

Scheme 1. Radical−Neutral Chemical Pathway of the Oxidative-Coupling Dimerization Process of Pterostilbene

fk = f k+ (r ) + f k− (r )

where qk(N), qk(N − 1), and qk(N + 1) are the electronic populations of atom k in its neutral, radical-cation, and radical-anion forms, respectively. In the present study, the Fukui nucleophilic contribution is used to rationalize the reactivity of stilbenoids with FeCl3·6H2O; the higher the f k‑(r) value, the higher the atomic nucleophilicity. All these calculations were performed with Gaussian09.32 Description of π-Stacking Interactions. Classical hybrid functionals are known to poorly describe noncovalent weak interactions such as π-stacking interactions of conjugated systems.

(3)

where f k+(r) and f k‑(r) are the electrophilic and nucleophilic contributions of the Fukui function:

f k+ (r ) = qk(N + 1) − qk(N )

(4a)

f k− (r ) = qk(N ) − qk(N − 1)

(4b) 540

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products



Among new DFT refinements, DFT-D, a successful approach to circumvent the use of sophisticated and expensive post-HF methods, has been developed by Grimme.33 It consists of the addition of dispersion correction based on the well-known dependence of the interactions between weakly overlapping systems as a function of R−6.34,35 It appears particularly relevant to calculate noncovalent interactions within an acceptable accuracy/computational time ratio. The pairwise-like dispersion energy (Edisp) is thus calculated in a post self-consistent field fashion:

E DFT ‐ D = E DFT + E D

RESULTS AND DISCUSSION Initiation Process: Pterostilbene Oxidation. Oxidation of polyphenols (ArOH) has extensively been rationalized, showing the importance of HAT from the OH groups of ArOH to the oxidative agent. Four different chemical pathways exist: (i) HAT-PCET (proton coupled electron transfer) (ArOH + R• → ArO• + RH);43,44 (ii) electron transfer proton transfer (ET-PT) (ArOH + R• → ArOH+• + R− → ArO• + RH);45,46 (iii) sequential proton loss electron transfer (SPLET) mechanism (ArOH → ArO− + H+; ArO− + R• → ArO• + R−; R− + H+ → RH);47−49 and (iv) adduct formation (ArOH + R• → ArOH−R → other metabolites or ArO• + RH in the presence of H2O).42 The ArO-H BDE appeared as a relevant thermodynamic descriptor to evaluate the capacity of HAT, disregarding the mechanism of action. BDEs of resveratrol and pterostilbene were recalculated with the same methodology for the sake of comparison and were found in agreement with previous determination performed for resveratrol.50 The 12-OH BDE is similar for both compounds, being 81.7 and 81.4 kcal mol−1, respectively (Table 1). The 3-

(5)

where ED is the dispersive energy having the RAB−6-dependent decay:34 N−1

Edisp = − s6

N

∑∑ A

B>A

C6AB 6 RAB

fdmp (RAB)

(6) CAB 6

is the dispersion where s6 is a functional-dependent scaling factor, coefficient for the atomic pair AB, RAB is the interatomic distance for atoms A and B, and fdmp(RAB) is a damping function that avoids nearsingularities for small interatomic distances.34 The refined version DFT-D2 has been widely used over the past years.36 We recently reparametrized a new s6 value (0.78) for the B3P86-D2 functional, providing accuracy to evaluate π-stacking complexation in flavonoid derivatives.20 Geometries of π-stacking complexes were thus obtained at the B3P86-D2(s6=0.78)/6-31+G(d,p) level. The robustness of this methodology was tested on flavonol and anthocyanidin self-association and flavonol:anthocyanidin complexation, with respect to high-level calculations and experimental evidence of these chemical arrangements.20 The intermolecular interaction energies are calculated as follows:

Table 1. Bond Dissociation Enthalpies (BDE, kcal mol−1), Nucleophilic Function ( f k−(r), |e|), and Chemical Hardness (η, eV) of Pterostilbene, Resveratrol, RNA2, and RNP2



Ei

i

(7)

position

pterostilbene

1- or 3-OMe 12-OH 1- or 3-OMe 12-OH C-6 C-6

RNA2 RNP2

where EComplex denotes the energy of the complex, and the summation runs over both free partners. EComplex includes a basis set superposition error (BSSE), estimated using the traditional counterpoise method: AB AB BSSE = [EAB (A) − EAAB(A)] + [EAB (B) − E BAB(B)]

EAB AB(A)

compound

resveratrol

Free Partner

ΔE int = EComplex −

Article

BDE 81.4 88.1 81.7 28.9 30.8

f k−(r)

η

0.017 0.030

4.3

OH BDE of resveratrol is much higher (88.1 kcal mol−1), and no OH group exists at this position in pterostilbene. This indicates that the HAT mechanism would preferentially take place at the 12-OH. However, these BDE values are relatively high compared to those of strong H-atom donors such as quercetin or catechin.51 Thus, the HAT requires the presence of a strong oxidant such as FeCl3·6H2O. Moreover, FeCl3 is a strong Lewis acid, while pterostilbene is a relatively strong base (η = 4.3 eV, Table 1), showing that according to the HSBA principle, the reaction between these compounds is very likely. The complexation with FeCl3·6H2O mainly occurs at O-12. Indeed, the highest Fukui nucleophilic contribution is obtained for this group, f k−(r) being 0.030, 0.017, and 0.017 at the O atom of the 12-OH, 1-OCH3, and 3-OCH3 groups, respectively (Table 1). The spin density distribution of the phenoxy radical generated after HAT from the native stilbene is mainly delocalized over the p-vinylphenol moiety (Figure 2).52 This distribution highlights different reactive sites (mainly at O-12, C-9, C-11, C-13, and C-7) for further reactions such as radical coupling. In principle this allows several different C−C or C−O combinations, leading to several dimers. However all the possible dimers are not observed experimentally, and the dimerization appears strongly regiocontrolled, as only compounds obtained from the C-7−C-7′ and C-7−C-8′ bond formation are observed, namely, pallidol and ampelopsin F analogues (Scheme 1). π-Stacking Complexation. On the basis of our previous hypothesis involving the formation of π-stacking self-association between two pterostilbene partners,18 here we explore the

(8)

EAB AB(B)

and are the energies of two given free partners A where and B, respectively, as obtained in the AB complex geometry with the AB AB basis set; EAB A (A) and EB (B) are the energies of A and B, respectively, as obtained in the AB complex geometry with the A and B basis sets, respectively. All the calculations were performed with ORCA.37 Solvent Description. Solvent effects were taken into account implicitly for all molecular systems including reactants, π-stacking complexes, intermediates, and products. The integral equation formalism−polarizable continuum model (IEF-PCM)38,39 and conductor-like screening (COSMO)40,41 models were used for thermodynamics of oxidative coupling and π-stacking complexation, respectively. In these types of models the solute is embedded in a shapeadapted cavity surrounded by a dielectric continuum, which is characterized by its dielectric constant ε. Calculations were performed in benzene (ε = 2.27, nD = 1.50), MeOH (ε = 32.63, nD = 1.33), CH2Cl2 (ε = 9.08, nD = 1.42), and H2O (ε = 80.40, nD = 1.33). Implicit PCM models are known to provide reasonable descriptions of solvent effects. The general trend is globally sufficiently accurate with polyphenols.42 The only general weakness occurs for solvents having high H-bonding capacities (mainly H2O). To improve accuracy, empirical corrections should be added, e.g., for HAT reactions, because adding explicitly H2O molecules would require too much computational resources. In the present work, the discussion is mainly based on calculations performed in CH2Cl2 to be compared with experimental data. Calculations with other solvents were mostly performed to evaluate the global influence of the solvent polarity.18 IEF-PCM and COSMO models were performed with Gaussian09 and ORCA, respectively. 541

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

Table 2. Stabilizing Energies (ΔEint, kcal mol−1) of the SelfAssociation Complexes of Stilbene, Pterostilbene, and the Mono-oxidized Pterostilbene stilbene

Figure 2. Spatial spin density distribution of the phenoxy radical obtained after HAT from the 12-OH group of pterostilbene. Only atomic values higher than 0.1 are reported.

pterostilbene

mono-oxidized pterostilbene

orientation

re/re

re/si

re/re

re/si

re/re

re/si

head-to-tail head-to-head T-shape

−3.2 −3.2 −7.3

−4.8 −3.2

−7.9 −7.8

−6.2 −8.1

−7.4

−7.7

exactly faces a C atom of the other partner). In these complexes intermolecular H-bonding slightly distorts the structures (the parallel alignment) but reinforces the stabilization. As previously described53 and confirmed here (at our level of calculation), the self-association is more likely of the T-shaped rather than cofacial π-stacked type in the absence of any substituent (i.e., stilbene, Figures 1 and 3b). The former arrangement is stabilized by around 7 kcal mol−1 while the stabilizing energy of the latter rearrangement is around 4 kcal mol−1 (Figure 3b, Table 2). The presence of OH and OMe groups, as in pterostilbene, induces (i) an increase of πconjugation due to their electron donor capacity, thus increasing the capacity for dispersive interactions between both partners, and (ii) the capacity of intermolecular H-bond formation between both partners. As a consequence, the cofacial π-stacking between two pterostilbene units is dramatically favored (ΔEint ranging from −6.2 to −8.1 kcal mol−1), no T-shaped arrangement being stabilized (Table 2).

different possibilities for π-stacking arrangements, namely, head-to-head or head-to-tail with both possible approaches re/re and re/si (Figure 3a). For all four possibilities the complexes were found to be stabilized by around 8 kcal mol−1: ΔEint is −7.8, −8.1, −7.9, and −6.2 kcal mol−1 for head-to-head (re/re), head-to-head (re/si), head-to-tail (re/re), and head-totail (re/si), respectively (Table 2). The differences in stabilizing energies between the different orientations appeared not significant, indicating that all possibilities may occur in solution with similar Boltzmann ratio, i.e., 25% for each conformation. The minimum distance in these complexes is around 3.5 Å (Figure 3a), as usually observed in π-stacking complexes. However, the π-stacking alignments are not ideal (Figure 3a), as rings are displaced but not in a classical manner (so-called parallel-displaced stacking, in which the center of one partner

Figure 3. Optimized geometries for self-association complexes for (a) pterostilbene and (b) stilbene exhibiting the distances of H-bond (blue dashed lines) and π-stacking interactions (black dashed lines). Side (up) and top (bottom) views are proposed here for all π-stacking complexes. 542

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

respectively. On the basis of an accurate molecular picture of these complexes our results confirm that favorable arrangements are possible prior to oxidation in order to rationalize the regiocontrolled formation of pallidol and ampelopsin F. π-Stacking complexes between a neutral stilbene moiety and a 12-O radical species, namely, neutral−radical complexes, can also exist in solution. The stabilizing energies of these complexes are similar to those obtained for the neutral−neutral complexes (Table 2). This confirms that π-stacking complexation not only is a reactant complex that would be formed just before the bond formation and after oxidation of one partner but truly exists in solution prior to oxidation. Complete Dimerization Mechanism: A Matter of Dearomatization and Rearomatization. In principle, the bond formation that initiates the formation of pallidol and ampelopsin F types of compounds occurs either between one phenoxy radical and the other native unit engaged in the πstacking complex (radical−neutral reaction) or between two phenoxy radicals (radical−radical reaction). Radical−Neutral Reaction. It is of major importance to note that, contrary to our previous hypothesis,18 the re/si and re/re (or si/si) approaches do not favor C-7−C-8′ and C-7−C7′ bond formations, respectively. The bond formation is more likely driven by the head-to-tail or head-to-head orientations, C7−C-7′ and C-7−C-8′ being slightly favored by the head-tohead and head-to-tail, respectively (Table 3). However, when the bond is formed, the subsequent intermediates RNA1 and RNP1 (Scheme 1) can flip-flop around the torsion angle defined by this bond. The Gibbs energy of activation for such a torsion rearrangement is around 8 kcal mol−1, meaning that this step is not limiting in solution. In both cases, the unpaired electron, mainly located on C-7 (Figure 5), attacks C-8′ or C-

This surprisingly high stability of the substituted-stilbenoid selfassociation is attributed to the combination of both intermolecular H-bonding and π-stacking. The latter contribution remains crucial since no energetic potential well (i.e., no stabilized complex) was found when using classical hybrid functionals that do not include dispersion (e.g., B3P86). The stabilizing energies are slightly higher than those we recently computed (at the same level of calculation) for quercetin self-association (ranging from −9.1 to −11.6 kcal mol−1) and significantly higher than those obtained for cyanidin:quercetin complexes (−13.9 kcal mol−1).20 Quercetin self-association involves π-type interaction with three conjugated aromatic rings versus two for pterostilbene. This is, however, compensated by intermolecular H-bonding that is stronger in pterostilbene than in quercetin self-association complexes (average H-bond lengths of 2.1 and 2.7 Å, respectively).54 The solvent effect slightly influences stabilization when increasing the dielectric permittivity from 2.27 to 80.4.55 A usual asymptotic behavior was observed (Figure 4), showing

Figure 4. Impact of the solvent polarity on the binding energies of the head-to-tail π−π pterostilbene self-association complexes.

that for ε higher than 10 the stabilizing energy remained unchanged. Specific interactions such as intermolecular Hbonding with solvent cannot be taken into account here. This would require using explicit solvent (at least for the first solvation shell), which is unfeasible for these molecular systems at this level of theory. However this would be critical in the case of H2O, while most of the experiments on which our previous study was based were performed in CH2Cl2 or CH2Cl2/MeOH (7:3) mixtures.18 Such π-stacking arrangements place the atoms of the 7,8double bonds of both partners in close proximity (C-7−C-7′ distance is ranging from 3.4 to 4.6 Å, while that of C-7−C-8′ from 3.6 to 4.0 Å; see Table 3). According to the spin density distribution occurring in the phenoxy monomeric radical, it appears that the most likely bond formations are C-7−C-7′ or C-7−C-8′, giving rise to pallidol and ampelopsin F analogues,

Figure 5. Spin density distributions of the radical intermediates involved in the different chemical pathways. Only atomic values higher than 0.1 are quoted here.

Table 3. Bond Distances (in Å) in re/re and re/si Alignments for Both Head-to-Tail and Head-to-Head Orientations in the Pterostilbene Self-Association Complexes head-to-tail

7′, respectively. This bond formation is endergonic, with a Gibbs energy of 28.9 and 27.8 kcal mol−1, respectively (Table 4). This indicates that the energy consumed to break the πconjugated system is higher than the energy released to establish the C-7−C-8′ or C-7−C-7′ covalent bonds. The spin distributions of RNA1 and RNP1 are poorly delocalized, less than in the phenoxy radical partner, thus confirming the destabilization of the products (Figure 5). Moreover, this

head-to-head

alignment

C-7−C-7′

C-7−C-8′

C-7−C-7′

C-7−C-8′

re/re re/si

4.4 4.6

3.6 3.7

3.4 3.7

4.0 3.5 543

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

Table 4. Gibbs Energy (ΔG°, kcal mol−1) of the Radical−Neutral and Radical−Radical Mechanisms Following Schemes 1 and 2, Gibbs Energy of Activation (ΔG#, kcal mol−1) of the Limiting Steps of the Radical−Neutral Processes, and Relative Gibbs Energy of Stabilization of Ampelopsin and Pallidol (ΔΔG°, kcal mol−1) for the Radical−Neutral Mechanism synthesis (a)

(b)

ampelopsin

pallidol

radical−neutral

radical−neutral

radical−radical

(c)

tricuspidatol

radical−neutral radical−radical

a

ΔG°

ΔG#

RNA1 formation RNA1 → RNA2 RNA3 → ampelopsin F

28.9 −1.3 −16.7

37.7

RNP1 formationa RNP1 → RNP2 RNP3 → pallidol RRP1 formationb RRP1 → RNP2

27.8 5.5 −29.5 8.3 −24.1

35.8

RNT1 formationa RNP1 → RNT1 RRT1 formationb

28.0 0.0 9.5

mechanism

reaction a

ΔΔG° 5.5a

0.0a

8.1

Refers to Scheme 1. bRefers to Scheme 2.

(from two radicals) does not compensate for the loss of aromaticity of both phenolic rings. This means that the reaction can occur only if it continues to evolve to a more stable molecular system (Scheme 2). This step of the radical−radical process is, however, much more favorable than that of the radical−neutral reaction. The subsequent nucleophilic attack of C-6′ on C-8, leading to RNP3 (Scheme 2), is highly favorable (Gibbs energy of −24.1 kcal mol−1), leading to the pallidol analogue. The C-7−C-8′ bond formation is highly improbable compared to C-7−C-7′ bond formation, mainly due to the low spin density at C-8′, making this site less reactive than the neighboring C-7′. Thus the radical−radical pathway intrinsically avoids the formation of the ampelopsin F analogue, providing only the pallidol analogue.

reaction exhibits a similar Gibbs energy of activation for both pathways (37.7 and 35.8 kcal mol−1, respectively). This shows that the bond formation is the limiting step, also confirming that the torsion rearrangement is not limiting. After the establishment of the bond between C-7 and C-8′ (RNA1) or C-7 and C-7′ (RNP1), the C-7′ radical attacks C-6 (−1.3 kcal mol−1) or C-8′ (5.5 kcal mol−1) to produce semifused ring systems (RNA2 and RNP2, respectively; see Scheme 1). The Gibbs energies of these reactions are close to 0 (slightly negative and positive, respectively) due to the loss of aromaticity in the quinone methide group, only partially compensated by spin delocalization (Figure 5). The formation of RNA2 appears easier than that of RNP2, as in the former product the spin is better delocalized (spin densities at C-6′ of 0.45 and 0.54 for both intermediates, respectively; see Figure 5). Further dissociation of the H atom from RNA2 and RNP2 (Scheme 1), leading to RNA3 and RNP3, respectively, is highly favored, as the corresponding C−H BDEs are very low, i.e., 34.2 and 35.2 kcal mol−1, respectively (Table 1). The HAT from these intermediates (RNA2 and RNP2) is sufficiently facile (highly labile H atom) to permit these species to be reduced by any surrounding molecules, including FeCl3 or other intermediates present in the solution. Both intermediates RNA3 and RNP3 complete the semifused ring systems by nucleophilic attack of C-6′ onto C-8 followed by rearomatization leading to ampelopsin F and pallidol analogues, respectively. This step is crucial to rationalize the entire process, exhibiting Gibbs energies of −16.7 and −29.5 kcal mol−1, respectively. Following the radical−neutral process, the overall formation of the pallidol analogue thus appears thermodynamically favored with respect to that of ampelopsin F (ΔΔG° = 5.5 kcal mol−1; see Table 2). Radical−Radical Reaction. In principle this reaction may occur between C-7 and C-7′ or C-8′ depending on the complex-of-approach. However, when the second unit is oxidized, two phenoxy radicals face each other, both having a high spin density at C-7 and C-7′. This leads to a reaction that occurs “spontaneously”,56 that is, following a potential energy surface as described for other polyphenol dimerizations.5730 In this case, the reaction is exothermic (ΔH of −6.3 kcal mol−1) with a low energetic barrier. Taking entropy into account, this reaction appears endergonic (Gibbs energy of 8.3 kcal mol−1), suggesting that the formation of the closed-shell system RRP1



CONCLUDING REMARKS On the basis of accurate quantum methodology, in particular taking dispersive effects into account, the present work has addressed many of the concerns hypothesized from experimental data on silbenoid oligomerization.18 First, the occurrence of π-stacking complexes comprising substituted stilbenoid derivatives is confirmed; they may exist in solution, prior to oxidation. These complexes are stabilized by intermolecular H-bonding and π-stacking interactions. Solvent effects, substitution pattern, and metal oxidant coordination are expected to profoundly influence these interactions, as all these factors tend to alter the electronic distribution in the system. In the presence of AgOAc, pallidol and ampelopsin F analogues are not formed; only δ-vinferifin derivatives are formed.18 The formation of this type of compound involves neither C-7−C-7′ nor C-7−C-8′ bond formation, but C-7−C-11′. Ag(I) ions efficiently bind to the 7,8-double bond (probably giving 2:1 stilbenoid:Ag coordination complexes), disrupting planarity and rendering π-stacking complexation unlikely. The presence of noncovalent stacks prior to oxidation fully explains the regiocontrol that was observed in pterostilbene dimerization.18 Supramolecular chemistry appears mandatory to address all concerns of stilbenoid oligomerization. Using the adequate methods, quantum calculations appear to be a powerful tool to elucidate these supramolecular arrangements. We believe that these 3D models may allow rationalizing and even predicting other issues in the field of regio- and 544

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

The radical−radical reaction may provide only pallidol-like derivatives and cannot operate as the major mechanism. If that were the case, the pallidol analogue concentration would be much higher than that of ampelopsin F when considering the thermodynamics and kinetics of this process. The present work clearly shows that π-stacking complexes are likely to occur in solution and are crucial to rationalize stilbenoid oligomerization. Their existence in the active sites of enzymes involved in biosynthesis has not yet been demonstrated. Nonetheless, the metabolon responsible for oligomerization of stilbenoids has not yet been elucidated. The hydrophobic character classically described for most of active sites is in favor of such stacking forces; however many other parameters should be considered, mainly size and bonding constraints in the active sites.

Scheme 2. Radical−Radical Chemical Pathway of the Oxidative-Coupling Dimerization Process of Pterostilbene



ASSOCIATED CONTENT

S Supporting Information *

Detailed results of the solvent impact on geometries and residual charge transfer in noncovalent complexes, XYZ coordinates of the structures performed. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Tel: +33 555 435 927. E-mail: [email protected]. Author Contributions #

S.V. and F.D.M. contributed equally.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank the “Conseil Régional du Limousin” for financial support and CALI (CAlcul en LImousin) for computing facilities. Research in Limoges is also supported by COST actions (FA1003 “East-West Collaboration for Grapevine Diversity Exploration and Mobilization of Adaptive Traits for Breeding” and 0804 “Chemical Biology with Natural Compounds”). Financial support by the French Embassy in Malaysia and the MICINN of Spain (project CTQ2011-27253) is gratefully acknowledged. The authors gratefully acknowledge the support by the Operational Program Research and Development for Innovations−European Regional Development Fund (project CZ.1.05/2.1.00/03.0058 of the Ministry of Education, Youth and Sports of the Czech Republic).



stereoselective synthesis of polyphenols; for example, the synthesis of stilbenoid tetramers is a direct perspective of this work. Second, the distinction between radical−neutral and radical− radical reactions is rationalized. It is clear that an ampelopsin F analogue originates from C-7−C-8′ bond formation, while a pallidol structure derives from C-7−C-7′ bond formation. The formation of π-stacking complexes fully explains the regiocontrol at the 7,8-double bond but is not sufficient to decide between C-7−C-7′ and C-7−C-8′ bond formation. However, the thermodynamics of the entire process is clearly in favor of the formation of pallidol derivatives, which is in perfect agreement with the 1:2 ratio obtained for ampelopsin F:pallidol analogues.18 It must be emphasized that in the presence of H2O the C-7−C-7′ bond formation can also provide the wellstabilized tricuspidatol-like product according to the mechanism reported in Schemes 1 and 2.

REFERENCES

(1) Gorham, J. The Biochemistry of the Stilbenoids; Chapman & Hall, 1995. (2) Weber, J. F. F.; Wahab, I. A.; Marzuki, A.; Thomas, N. F.; Kadir, A. A.; Hadi, A. H. A.; Awang, K.; Latiff, A. A.; Richomme, P.; Delaunay, J. Tetrahedron Lett. 2001, 42, 4895−4897. (3) Sotheeswaran, S.; Sultanbawa, M. U. S.; Surendrakumar, S.; Balasubramaniam, S.; Bladon, P. J. Chem. Soc., Perkin Trans. 1 1985, 159−162. (4) Zgoda-Pols, J. R.; Freyer, A. J.; Killmer, L. B.; Porter, J. R. J. Nat. Prod. 2002, 65, 1554−1559. (5) Bokel, M.; Diyasena, M. N. C.; Gunatilaka, A. A. L.; Kraus, W.; Sotheeswaran, S. Phytochemistry 1988, 27, 377−380. (6) Ducrot, P.-H.; Kollmann, A.; Bala, A. E.; Majira, A.; Kerhoas, L.; Delorme, R.; Einhorn, J. Tetrahedron Lett. 1998, 39, 9655−9658. (7) Ge, H. M.; Huang, B.; Tan, S. H.; Shi, D. H.; Song, Y. C.; Tan, R. X. J. Nat. Prod. 2006, 69, 1800−1802. 545

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546

Journal of Natural Products

Article

(8) Tanaka, T.; Ito, T.; Nakaya, K.-i.; Iinuma, M.; Takahashi, Y.; Naganawa, H.; Matsuura, N.; Ubukata, M. Tetrahedron Lett. 2000, 41, 7929−7932. (9) Morikawa, T.; Xu, F.; Matsuda, H.; Yoshikawa, M. Chem. Pharm. Bull. 2010, 58, 1379−1385. (10) Dai, J.-R.; Hallock, Y. F.; Cardellina, J. H.; Boyd, M. R. J. Nat. Prod. 1998, 61, 351−353. (11) Ito, T.; Tanaka, T.; Nakaya, K.-i.; Iinuma, M.; Takahashi, Y.; Naganawa, H.; Ohyama, M.; Nakanishi, Y.; Bastow, K. F.; Lee, K.-H. Tetrahedron 2001, 57, 7309−7321. (12) Ito, T.; Akao, Y.; Yi, H.; Ohguchi, K.; Matsumoto, K.; Tanaka, T.; Iinuma, M.; Nozawa, Y. Carcinogenesis 2003, 24, 1489−1497. (13) Huang, K.-S.; Lin, M.; Yu, L.-N.; Kong, M. Tetrahedron 2000, 56, 1321−1329. (14) Waffo-Teguo, P.; Lee, D.; Cuendet, M.; Mérillon, J.-M.; Pezzuto, J. M.; Kinghorn, A. D. J. Nat. Prod. 2000, 64, 136−138. (15) Snyder, S. A.; Gollner, A.; Chiriac, M. I. Nature 2011, 474, 461− 6. (16) Snyder, S. A.; Breazzano, S. P.; Ross, A. G.; Lin, Y.; Zografos, A. L. J. Am. Chem. Soc. 2009, 131, 1753−65. (17) Snyder, S. A.; Zografos, A. L.; Lin, Y. Angew. Chem., Int. Ed. 2007, 46, 8186−91. (18) Velu, S. S.; Buniyamin, I.; Ching, L. K.; Feroz, F.; Noorbatcha, I.; Gee, L. C.; Awang, K.; Wahab, I. A.; Weber, J.-F. F. Chem.Eur. J. 2008, 14, 11376−11384. (19) Velu, S. S.; Thomas, N. F.; Weber, J.-F. F. Curr. Org. Chem. 2012, 16, 605−662. (20) Di Meo, F.; Sancho-Garcia, J. C.; Dangles, O.; Trouillas, P. J. Chem. Theory Comput. 2012, 8, 2034−2043. (21) Zondlo, N. J. Nat. Chem. Biol. 2010, 6, 567−568. (22) Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990, 112, 5525−5534. (23) Aragó, J.; Sancho-García, J. C.; Ortí, E.; Beljonne, D. J. Chem. Theory Comput. 2011, 7, 2068−2077. (24) Gierschner, J.; Cornil, J.; Egelhaaf, H.-J. Adv. Mater. 2007, 19, 173−191. (25) Rezazgui, O.; Boëns, B.; Teste, K.; Vergnaud, J.; Trouillas, P.; Zerrouki, R. Tetrahedron Lett. 2011, 52, 6796−6799. (26) Coates, G. W.; Dunn, A. R.; Henling, L. M.; Dougherty, D. A.; Grubbs, R. H. Angew. Chem., Int. Ed. Engl. 1997, 36, 248−251. (27) Bonechi, C.; Martini, S.; Magnani, A.; Rossi, C. Magn. Reson. Chem. 2008, 46, 625−629. (28) Trouillas, P.; Marsal, P.; Siri, D.; Lazzaroni, R.; Duroux, J. L. Food Chem. 2006, 97, 10. (29) Anouar, E.; Calliste, C. A.; Košinová, P.; Di Meo, F.; Duroux, J. L.; Champavier, Y.; Marakchi, K.; Trouillas, P. J. Phys. Chem. A 2009, 113, 13881−13891. (30) Kosinova, P.; Gazak, R.; Duroux, J.-L.; Lazzaroni, R.; Kren, V.; Assfeld, X.; Trouillas, P. ChemPhysChem 2011, 12, 1135−1142. (31) Yang, W.; Mortier, W. J. J. Am. Chem. Soc. 1986, 108, 5708−11. (32) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; et al. Gaussian 09, Revision A.02; Wallingford, CT, 2009. (33) Grimme, S. J. Comput. Chem. 2004, 25, 1463−1473. (34) Grimme, S. J. Comput. Chem. 2006, 27, 1787−1799. (35) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104−154119. (36) We are aware that the new generation DFT-D3 has been developed including atom pairwise-specific dispersion coefficients and a new set of cutoff radii as defined in the damping function. The reparametrized s6 = 0.78 provided better accuracy for interaction energies in p-stacking complexes than DFT-D3. (37) Neese, F. WIREs 2012, 2, 73−78. (38) Cossi, M.; Scalmani, G.; Rega, N.; Barone, V. J. Chem. Phys. 2002, 117, 43−54. (39) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999−3094.

(40) Sinnecker, S.; Rajendran, A.; Klamt, A.; Diedenhofen, M.; Neese, F. J. Phys. Chem. A 2006, 110, 2235−2245. (41) The default atom-sphere radii (i.e., van der Waals radii increased by 20%) as implemented in the ORCA package are used. (42) Anouar, E.; Kosinova, P.; Kozlowski, D.; Mokrini, R.; Duroux, J. L.; Trouillas, P. Phys. Chem. Chem. Phys. 2009, 11, 7659−7668. (43) Burton, G. W.; Doba, T.; Gabe, E.; Hughes, L.; Lee, F. L.; Prasad, L.; Ingold, K. U. J. Am. Chem. Soc. 1985, 107, 7053−65. (44) De Heer, M. I.; Mulder, P.; Korth, H.-G.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc. 2000, 122, 2355−2360. (45) Jovanovic, S. V.; Steenken, S.; Tosic, M.; Marjanovic, B.; Simic, M. G. J. Am. Chem. Soc. 1994, 116, 4846−51. (46) Jovanovic, S. V.; Steenken, S.; Hara, Y.; Simic, M. G. J. Chem. Soc., Perkin Trans. 2 1996, 2497−2504. (47) Zhang, H.-Y.; Ji, H.-F. New J. Chem. 2006, 30, 503−504. (48) Litwinienko, G.; Ingold, K. U. J. Org. Chem. 2003, 68, 3433−8. (49) Foti, M. C.; Daquino, C.; Geraci, C. J. Org. Chem. 2004, 69, 2309−2314. (50) Cao, H.; Pan, X.; Li, C.; Zhou, C.; Deng, F.; Li, T. Biorg. Med. Chem. Lett. 2003, 13, 1869−1871. (51) Trouillas, P.; Marsal, P.; Siri, D.; Lazzaroni, R.; Duroux, J.-L. Food. Chem. 2006, 97, 679−688. (52) The spin density delocalization of the resveratrol radical obtained after HAT from O-3 is less extended than from O-12, the radical being nonplanar. This explains the higher BDE obtained for 3OH compare to 12-OH. It must also be stressed that in this case the spin density at C-7 is very low. (53) Gierschner, J.; Oelkrug, D. In Optical Properties of Oligophenylenevinylenes; American Scientific Publishers, 2004; pp 219−238. (54) The head-to-tail re/si orientation presents both the highest pstacking alignment and the lowest H-bond distance (2.3 Å) but also exhibits the lowest interaction energy (DEint = −6.2 kcal mol−1) with respect to the other geometries. (55) The intermolecular distance is only slightly decreased from 3.98 to 3.92 Å from the gas phase to H2O (e = 80.4), respectively. In MeOH (e = 32.6) this distance is the same as in water. The intermolecular H-bonding is also slightly modified but in a highly complex-dependent way, i.e., strengthening or weakening. However, the global stabilization is not modified, nor even the related conclusions. (56) When optimizing a radical−radical complex, the C-7−C-7′ bond is automatically formed, meaning that (i) no radical−radical complex may exist in solution and (ii) the double oxidation systematically produces the C-7−C-7′ bond formation. (57) This is a typical radical−radical reaction in which the system is (i) a triplet state when both units are far from each other, (ii) a singlet when the bond is formed, and (iii) a complex mixture of both triplet and singlet states.

546

dx.doi.org/10.1021/np300705p | J. Nat. Prod. 2013, 76, 538−546