Theoretical Study on the Biosynthesis of the Mandapamates: Mechanistic Insights Using Density Functional Theory

Density functional theory (B3LYP-D3(BJ) and ωB97XD) calculations have been used to assess the stereochemical outcomes of the proposed transannular [4 + 2] cycloaddition pathway for the biosynthesis of mandapamate and isomandapamate from macrocyclic intermediates. Calculations reveal that the topological shift between macrocyclic conformers is vital in controlling the stereoselectivity of the downstream steps toward the isomeric mandapamates. A stepwise 4 + 2 type process is energetically favored over a concerted [4 + 2] pathway at room temperature, and is consistent with the stereochemistries found in the natural products.


■ INTRODUCTION
Mandapamate 1 and its stereoisomer isomandapamate 2 are novel polycyclic cembranoid diterpenes found in soft corals. 1 Other polycyclic cembranoids found in soft corals include plumarellide 3, intricarene 4 and bielschowskysin 5, and the closely related C19 norditerpenes ineleganolide 6 and sinulochmodin C (7). 2 Each of these polycyclic structures is purported to be derived in nature from macrocyclic furanocembranoid structures, e.g., 8 3 and 9, 4 via sequences involving oxidation/hydration followed by transannular cyclizations (Figure 1). 5 Thus, in the case of the mandapamates 1 and 2, they are thought to be biosynthesised from the hypothetical precursor 10 via 11/12, leading to the key exo-enol ether intermediate 15 which then undergoes transannular [4 + 2] cycloaddition (Scheme 1a).Stable exoenol ether structures similar to 15 were first reported in corals by Fenical et al. 6 Others were described later, e.g., 16 and 17, 7 and exo enol ether structures are also thought to be implicated in biosynthetic pathways which lead to plumarellide 3 and to bielschowskysin 5. 5 Mandapamate 1 and isomandapamate 2 share a common angular [5,6,7]-carbocyclic ring system with (3R, 6S)-and (3S, 6R)-stereochemistry respectively at their C3 and C6 centers (Figure 1).This stereogenicity originates from the relative orientation of the oxygen atoms on the bridged cycloheptenes in their structures.We postulated that during the proposed biosynthesis of the mandapamates there is likely to be restraints imposed on the conformations of their macrocyclic enol ether/furanoxonium ion intermediates, viz.13−15 (Scheme 1a), which would predetermine the stereochemical outcomes of the subsequent transannular cyclizations, thereby leading to the different stereochemistries present in the two natural products.The origins of these proposed conformational preferences are not known, which has motivated us to establish some theoretical support for their existence, and thereby importance, using density functional theory (DFT).
The complexities of the ring systems and the oxidation patterns present in the naturally occurring polycycles 1−7, combined with the interesting biological properties they display, have made them challenging targets for the synthetic chemist.Pertinent to our study, there has been a recent novel synthesis of the core carbocyclic ring system in bielschowskysin 5 which was based on an intramolecular [2 + 2] cyclization involving an exo enol ether intermediate akin to that shown in compound 15. 8,9 Other synthetic studies implicating enol ether intermediates have been carried out toward plumarellide 3, 10 and total syntheses of both ineleganolide 6 and sinulochmodin C (7) have recently been accomplished. 11Earlier Pattenden et al. 12 and Trauner et al. 13 independently achieved a biomimetic synthesis of (+)-intricarene 4 from bipinnatin J (8) implicating an intramolecular dipolar [5 + 2] cycloaddition reaction as the key step, and in 2011, Li and Pattenden presented biomimetic syntheses of ineleganolide 6 and sinulochmodin C (7) from the macrocyclic 5-episinuleptolide 18 (believed to be derived from deoxypukalide 9 in vivo) via sequences of transannular Michael reactions. 14Some studies have been made toward the synthesis of the mandapamates 1 and 2 exploring an intramolecular [4 + 2] cyclization strategy to produce their [5,6,7] tricyclic ring systems, viz.19 to 20 and 21 to 22 (Scheme 1b, c), 15,16 but with limited success.
The challenges presented by the synthesis of the naturally occurring polycycles 1−7, implicating their likely biosynthesis precursors, have also prompted theoretical chemists to examine features of the mechanisms of these likely transformations.Quantum chemical calculations are now widely used to uncover new mechanisms 17 as well as seek theoretical support for biosynthetic proposals. 18Related computational studies have recently been reviewed.18a Tang and Paton, for example, recently used DFT to probe the pathway to the cyclobutane rings in the furanocembranoid providencin 19 and in bielschowskysin 5. Thus, based on quantum chemical calculations of the thermal and photochemical pathways toward the formation of the cyclobutane ring in bielschowskysin 5, they concluded that while a photo [2 + 2] cycloaddition pathway is computed to be highly efficient, a thermal ringclosure process in a single step in water is also energetically feasible at room temperature. 20More recently, alongside their synthetic work toward bielschowskysin 5, Scesa et al. also used DFT calculations to analyze the conformation of the likely furanocembranoid precursors, e.g., deoxypukalide 9 to its formation.8b Earlier, Lygo et al. used DFT calculations to study the steps leading to the polycyclic ring system found in plumarellide 3 and compared a stepwise cyclization pathway with a [4 + 2] cycloaddition of a furanoxonium ion intermediate. 21Hence, transannular cycloaddition reactions implicating furanoxonium ions as key intermediates have been verified both experimentally 5b,e,10,16 and computationally with DFT. 22

The Journal of Organic Chemistry
In the proposed biosynthesis of the mandapamates 1 and 2 from the furanocembranoid precursor 10 (Scheme 1a) the key exo enol ether intermediate 15, can exhibit both E-and Zgeometry, each of which can undergo either a concerted or stepwise transannular 4 + 2 cycloaddition process leading to the isomeric mandapamates.Known furancembranes akin to the hypothetical precursor 10 have been isolated and/or synthesized, including bipinnatin J (8), 3,12,13 Z-/E-deoxypukalide 9, 4 leptolide, 23 pukalide aldehyde, 23 molestin E, 24 in addition to two model precursors designed by Mehta 15 and Pattenden. 16The existence of these furancembranes gives us confidence that our proposed precursors could be derived from naturally existing furancembranes, if they themselves are not natural products.
To gain insight into this biosynthesis proposal we have carried out detailed quantum chemical calculations of the key steps and the likely stereochemical and conformational requirements of any macrocyclic furanoxonium ion intermediate implicated in competitive transannular 4 + 2 cycloaddition processes.The key point for this calculation is The Journal of Organic Chemistry to understand the feasibility of the proposed transannular reactions leading to mandapamate and isomandapamate, despite the possibility that some side reactions might occur experimentally.For example, transannular cycloaddition between the furan ring and the exo C(15)�C is a possibility.However, our calculations revealed that the associated transition state (TS) has a very high activation energy barrier of over 59.0 kcal mol −1 , which rules out this pathway (Figure S7).Similarly, selective dihydroxylation or epoxidation of the C7−C8 double bond with the existence of C15−C16 double bond could be problematic.Therefore, a good number of synthetic attempts/trails would be required to find the right reaction conditions to achieve these transformations.

■ COMPUTATIONAL METHODS
All the calculations reported were carried out in Gaussian 16. 25 Unless stated otherwise, B3LYP-D3(BJ) 26 /def2-TZVP 27 //B3LYP/6-31G-(d) 28 and ωB97XD 29 /def2-TZVP//B3LYP/6-31G (d) relative free energies at 298.15 K, obtained from optimized species including an implicit SMD description of water, 30 are quoted throughout, the latter in parentheses.The vibrational frequencies were computed at the same level of theory as for the geometry optimizations to confirm whether each optimized structure is an energy minimum (possessing zero imaginary frequencies) or a TS (possessing one single imaginary frequency), and to obtain the zero-point vibrational energy and thermal corrections under 298.15K and 1 atm pressure.All transition states were confirmed to connect reactants and products by intrinsic reaction coordinate calculations. 31All Gibbs free energies discussed in the manuscript were corrected using the quasi-rigid rotor-harmonic

The Journal of Organic Chemistry
oscillator approach proposed by Grimme 32 and implemented in the GoodVibes code 33 (Figures S5, S6 and Table S5).The 3D graphics of molecules were generated using CYLview. 34Electrostatic potential surface (EPS) analysis based on the noncovalent interactions was performed using Multiwfn (v3.8_dev) 35 and visualized using VMD. 36he level of the theory that we have adopted was based on functional/basis set benchmark studies carried out in this work (Tables S1, S2 and S3).The B3LYP-D3(BJ) and ωB97XD functionals provided similar accuracy and qualitative conclusions.We selected B3LYP/6-31G(d) level for the geometry optimization in order to achieve the best balance between accuracy and computational cost.The 0.0 kcal mol −1 reference includes the single point energies of 11b, H 2 O, H 3 O + and CH 3 OH.The structures of H 2 O, H 3 O + and CH 3 OH were optimized and computed at the same level of 11b.Conformational analyses of 11 and 12 were carried out via random searching in the GMMX 3.1 module of Gaussian 16 using the MMFF94 37 force field.The computational details are given in the Supporting Information.

■ RESULTS AND DISCUSSION
Conformational Analysis of the Macrocyclic Furanocembranes 11/12 to 13. Referring to the biosynthesis proposal (Scheme 1a) we first analyzed the expected topological shift in the macrocyclic ring of the intermediates 11 to 13 resulting from conformational changes in the orientation of their furan rings.Using relaxed coordinate scans, the potential energy surfaces for these conformational changes were obtained at the B3LYP/6-31G(d) level through the incremental rotation of the O−C6−C7−C8 dihedral angles (10°increments each frame, see Supporting Information).As shown in Figure 2a, the energy minimum for conformer 11a is almost equi-energetic with conformer 11b (+1.1 kcal mol −1 ).These two conformers differ in the relative location of the furan-ring plane.The conformer 11a is parallel while the conformer 11b is orthogonal to the macrocycle.The activation Gibbs free energy toward the TS TS 11a,11b is 3.4 kcal mol −1 with a reverse barrier of 2.3 kcal mol −1 (11b → TS 11a,11b ), so that the interconversion between conformers 11a and 11b occurs readily.Following a pivotal exothermic process (furan dearomatisation), the conformers 11a and 11b are converted into the more stable furanoxonium ion intermediates Int-13a and Int-13b respectively.Interestingly, the configurations (E-vs Z-) of the Δ 6,7 double bonds in 13 are divergent at this stage and they will be retained in the downstream steps, thereby controlling the C3 and C6 stereocenters in the final mandapamates.The pathways for the interconversion of the epoxy compounds (cis-12a⇄cis-12b, trans-12a⇄trans-12b) presented similar transition states TS cis-12a,cis-12b and TS trans-12a,trans-12b (Figure 2b).Compounds cis-12 and trans-12 differ in the orientations of their epoxy moieties.For the cis epoxide, the activation Gibbs free energy toward the TS TS cis-12a,cis-12b is 6.6 kcal mol −1 with a reverse barrier of 3.3 kcal mol −1 (cis-12b → TS cis-12a,cis-12b ).In the corresponding trans epoxide, the activation Gibbs free energy toward the TS TS trans-12a,trans-12b is 4.6 kcal mol −1 with a reverse barrier of 4.0 kcal mol −1 (trans-12b → TS trans-12a,trans-12b ).These two transition states (TS c i s -1 2 a , c i s -1 2 b and TS trans-12a,trans-12b ) also connect to the corresponding parallel conformers (cis-12a and trans-12a) and to the orthogonal conformers (cis-12b and trans-12b) which are in equilibrium and lead to the formation of the furanoxonium ion intermediates Int-13a and Int-13b with E-and Z-configured (Δ 6,7 )-double bonds, respectively.regioselectivity.To investigate further the origin of the conformational preference in the transition states TS 11a,11b , TS cis-12a,cis-12b and TS trans-12a,trans-12b , the EPS analysis was performed.As displayed in Figure 2c, d, in the transition states TS 11a,11b and TS cis-12a,cis-12b , the moiety containing a hydroxy or epoxy group at C7 shows a negative surface potential (blue, rO•••O 2.70 Å), while the oxygen atom on furan ring also shows a negative surface potential (blue, rO•••O 2.87 Å), indicating that there is electrostatic repulsion between them.However, in the TS TS trans-12a,trans-12b , the oxygen atom on the furan ring shows a neutral surface potential (white, rO•••O 3.70 Å), indicating that there is less electrostatic interaction.This also explains why the activation energy barrier of TS cis-12a,cis-12b is higher than that of TS trans-12a,trans-12b .Therefore, the EPS analysis suggests that the electrostatic interaction would be the driving force for the rotation about the furan ring.
Pathway for the Formation of Mandapamate 1.As already discussed, the proposed biosynthesis of the mandapamates 1 and 2 (Scheme 1a) proceeds via the central stable furanoxonium ion intermediate Int-13b derived from 11b and 12b.The conformations of 11b and 12b are orthogonal and hence the intermediate Int-13b will have a Z-configured (Δ 6,7 )-double bond.This Z-configured (Δ 6,7 )-Int-13b is a direct optimization result of the pivotal furan dearomatisation process from 11b and 12b.Because extensive searches could not locate transition states connecting both sides of 11b/12b and int-13, we consider that this furan dearomatisation process is exothermic and is likely to be a barrierless process on the PES.The furanoxonium ion intermediate Int-13b is next quenched by methanol leading to the intermediate Int-14b (Figure 3).This process proceeds via TS1b with an activation Gibbs free energy of 9.2 kcal mol −1 .The resulting protonated intermediate Int-23b then undergoes deprotonation to afford the stable exo enol ether-cyclic ketal 15b.Lastly, 15b undergoes a transannular [4 + 2] cycloaddition via TS2b leading to mandapamate 1 with an activation Gibbs free energy of 16.0 kcal mol −1 .However, the overall energy barrier of this pathway (Int-13b → TS2b, 36.3 kcal mol −1 ) seems to be overly high for a biosynthesis route.We therefore investigated the alternative, stepwise, intramolecular cyclization pathway from 13b to mandapamate 1.Indeed, our calculations demonstrated that the Z-configured (Δ 6,7 ) intermediate Int-13b could undergo transannular C−C bond formation via TS3b (Figure 3) with an activation Gibbs free energy of 6.6 kcal mol −1 to generate the thermodynamically stable cyclopentane-substituted intermediate Int-24b.A second transannular C−C bond formation through TS4b then leads to the intermediate Int-25b containing the tricyclic ring system found in mandapamate 1 with an activation Gibbs free energy of 25.5 kcal mol −1 .Methanolysis of the intermediate Int-25b, to Int-26b, followed by deprotonation finally produces mandapamate 1. Attempts to locate the transition states connecting both sides of int-25 and int-26 (MeOH nucleophilic attack) were unsuccessful, indicating this is a barrierless process.Compared with the aforementioned concerted [4 + 2] cycloaddition pathway, the overall energy barrier in this stepwise route (Int-13b → TS4b, 24.9 kcal mol −1 , without enzymatic intervention) is more reasonable.Our calculations revealed that B3LYP-D3(BJ) and ωB97XD functionals predicted the same rate-determining step, with M06-2X predicted a much lower Gibbs free activation energy for TS4b, see Table S3 in the Supporting Information, ΔG ⧧ (TS4b) is ca. 25 kcal mol −1 predicted by B3LYP-D3(BJ) and ωB97XD, but 21.4 kcal mol −1 predicted by M06-2X.One might anticipate that with The Journal of Organic Chemistry enzymatic intervention, the energy barriers would typically decrease by at least a few kcal mol −1 .The TS TS3b is favored over the TS1b by 12.6 kcal mol −1 and the TS TS4b is favored over the TS2b by 11.4 kcal mol −1 respectively.
Relative Stereochemistries of the Mandapamates 1 and 2. Our calculations reveal that in the proposed biosynthesis of the mandapamates, the orientations of the oxygen atoms in the furan ring of the furanoxonium ion intermediate Int-13 will be maintained once the latter is  The distortion/interaction model originally conceived for bimolecular reactive systems, can also be applied to intramolecular reactions with a fragmentation scheme. 38As displayed in Figure 5a, we computed the strain-related energies of distorting various fragments in the intramolecular transition structures TS2, TS3 and TS4.The fragments (bold moieties) are obtained after being separated and filled valences with hydrogen atoms where the covalent bonds have been broken.The distortion/interaction analysis revealed that the distortion energies of TS3 (TS3a: ΔE ⧧ dist = 29.9kcal mol −1 , TS3b: ΔE ⧧ dist = 34.6 kcal mol −1 ) representing first C7−C11 bond formation in the stepwise pathway are lower than that of TS2 representing concerted pathway (i.e., the concerted formation of the bonds C6−C14 and C7−C11).Compared with TS2, the bond distance of the second C6−C14 bond formation in TS4 are shorter by 0.85 and 0.63 Å respectively, indicating the stronger interaction in the transition states TS4 (TS4a: ΔE ⧧ int = −35.6 kcal mol −1 , TS4b: ΔE ⧧ int = −24.4kcal mol −1 ) of the stepwise pathway.Through constrained coordinate scans, we obtained the PES describing the mechanism of formation of the transannular C−C bonds C6−C14 and C7−C11 at the B3LYP/6-31G(d) level.The most striking feature of the [4 + 2] cycloaddition pathway toward mandapamate 1 is that there is only one single saddle point connecting the intermediate 15b and 1 (Figure 3), and clearly this corresponds to the concerted TS TS2b shown in Figure 5b (left).By contrast, the stepwise 4 + 2 type pathway to mandapamate 1 presents two saddle points connecting three species with much lower energy barriers in Figure 5b (right).Thus, the first transannular C−C bond formation occurs between C7 and C11 via TS TS3b (Figure 5b, right), and then the second transannular C−C bond forms immediately via TS TS4b, leading to formation of the key intermediate Int-25b (Figure 3) containing the tricyclic ring system in the natural product.Likewise, the above is also applicable to the pathway toward isomandapamate 2, due to the presence of the same carbocyclic skeleton, albeit with different orientations of the oxygen-bridged moiety.

■ CONCLUSION
We have explored the concerted and the stepwise pathways leading to the formation of the tricyclic ring systems in the mandapamates 1 and 2 using quantum chemical calculations (Scheme 2).Our calculations reveal that the topological shift between conformers in the macrocyclic precursors to 1 and 2 is significant in controlling which one leads to the (3R, 6S)mandapamate 1 and which to the (3S, 6R)-isomandapamate 2 via Z-configured (Δ 6,7 ) Int-13b and E-configured (Δ 6,7 ) Int-13a respectively.Our study therefore highlights the significance and importance of analyzing the conformations of macrocyclic molecules in order to appreciate their inherent ring-strained reactivity. 39The energy barriers between the conformers Int-13a and Int-13b (ΔΔG 11a⇄11b = 1.1 kcal mol −1 , ΔΔG cis-12a⇄cis-12b = 3.3 kcal mol −1 , ΔΔG trans-12a⇄trans-12b = 0.6 kcal mol −1 ) would allow facile conformational changes in conditions in vivo.Comparing the proposed cyclization routes to the ring systems in the mandapamates, calculations demonstrate that a stepwise sequence is energetically favored over an alternative concerted [4 + 2] cycloaddition pathway at room temperature.Our calculations also demonstrate that the stepwise pathway explains the formation of the different stereochemistries found in the naturally occurring isomeric mandapamates 1 and 2. Our study has provided some new mechanistic insight.However, the actual biosynthetic pathway may differ from that suggested by our calculations.For example, enzymes can alter reaction pathways beyond simply lowering a TS barrier.We can only rely on synthetic strategies in the lab to explore the chemistry of synthesizing macrocyclic precursors as well as biomimetically synthesizing mandapamates 1 and 2. Synthetic work toward the synthesis of macrocyclic furanocembranoid 11/12 for the biomimetic syntheses of mandapamates 1 and 2 is currently being carried out in our laboratory.Progress will be reported in due course.

Data Availability Statement
The data underlying this study are available in the published article and its Supporting Information The Journal of Organic Chemistry

Scheme 1 .
Scheme 1.(a) Biosynthetic Proposal for the Formation of Mandapamate 1 and Isomandapamate 2 a