A Relay Strategy Actuates Pre-Existing Trisubstituted Olefins in Monoterpenoids for Cross-Metathesis with Trisubstituted Alkenes

A retrosynthetic disconnection–reconnection analysis of epoxypolyenes—substrates that can undergo cyclization to podocarpane-type tricycles—reveals relay-actuated Δ6,7-functionalized monoterpenoid alcohols for ruthenium benzylidene catalyzed olefin cross-metathesis with homoprenyl benzenes. Successful implementation of this approach provided several epoxypolyenes as expected (E/Z, ca. 2–3:1). The method is further generalized for the cross-metathesis of pre-existing trisubstituted olefins in other relay-actuated Δ6,7-functionalized monoterpenoid alcohols with various other trisubstituted alkenes to form new trisubstituted olefins. Epoxypolyene cyclization of an enantiomerically pure, but geometrically impure, epoxypolyene substrate provides an enantiomerically pure, trans-fused, podocarpane-type tricycle (from the E-geometrical isomer).


■ INTRODUCTION
Biomimetically inspired polyene cyclizations have emerged as a powerful synthetic strategy for the stereocontrolled construction of complex polycarbocyclic scaffolds of biological significance, 1 where epoxypolyene cyclizations of terminally functionalized geranyl units with nucleophilic aromatic headgroups have provided synthetic access to podocarpane-type tricyclic diterpene skeleta ( Figure 1a). 2 Such cyclization substrates are typically constructed in two steps via metalcatalyzed cross-coupling methodology of an electrophilic geranyl species in conjunction with a benzylic organometallic, andeither before or after C−C bond construction regioselective functionalization of the geranyl alkene at the terminus of the chain (Figure 1a). 3 Each of these steps is subject to a potential disadvantage: the former is subject to competing allylic S N 2′ substitution, and the latter to nonperfect regioselective oxidation, regardless of the order of implementation. 4 During the course of our studies, we had reason to consider an alternative disconnection of such functionalized linear monoterpenoid derivatives by olefin cross-metathesis, but of the two terminal olefin species that are revealed, the epoxide-containing component is synthetically nonsimplified (Figure 1b). Nonetheless, a "reconnection" operation 5 reveals a geraniol derivative with a pre-existing trisubstituted olefin that we expected could be actuated for cross-metathesis by the application of Hoye's relay strategy. 6 For reasons outlined below, we also elected to "reconnect" the terminal alkene component from the initial disconnection as a trisubstituted alkene.
The catalyst(s) of choice for the above proposition would be the commercially available well-defined ruthenium benzyli-denes as developed by Grubbs. 7 Such catalysts are widely used to accomplish the ring-closing metathesis of disubstituted, trisubstituted, and even tetrasubstituted olefins. 8 In contrast, and quite surprisingly, there are only limited reports on the formation of unfunctionalized trisubstituted olefins (as required here) by cross-metathesis using ruthenium benzylidene pre-catalysts. 9 Grubbs and co-workers initially showed that ruthenium pre-catalyst 1 was competent for the crossmetathesis of geminally disubstituted olefins with terminal olefins (Figure 1c). 10,11 Subsequently, Robinson and coworkers showed that the cross-metathesis of sterically challenging allyl branched 1,1-disubstituted olefins performed considerably better using a (terpenoid) prenyl rather than an allyl partner using precatalyst 2 (Figure 1d). 12 With this latter literature precedent in mind, we therefore selected trisubstituted olefins as the cross-metathesis partners (Figure 1b, reconnection). 13 As envisioned, this overall stratagem not only opens up the possibility of an alternative, modular, synthetic route to such cyclization precursors, but perhaps more significantly could provide a general approach to the functionalization of pre-existing trisubstituted olefins in acyclic monoterpenoid alcohols by cross-metathesis (Figure 1e). 14 Herein, we report the success of this unprecedented olefin− olefin combination to form new unfunctionalized trisubstituted olefins by cross-metathesis (Figure 1e), where the overall transformation can be classified as a relay cross-metathesis. This relay cross-metathesis reaction ("ReXM") distinguishes itself from the very limited literature precedent for such reactions by being the first such example to form isolated, unconjugated, trisubstituted alkenes where all previous reports have formed conjugated alkenes. 15,16 ■ RESULTS AND DISCUSSION We commenced our investigations with two main objectives in mind: (i) demonstration of proof-of-principle ReXM of monoterpenoid alcohol derivatives with homoprenylbenzenes to prepare representative epoxypolyene cyclization substrates; (ii) exemplification of the method as a general approach for the functionalization of pre-existing trisubstituted olefins in acyclic monoterpenoid alcohols. Accordingly, we assembled relaymodified Δ 6,7 -functionalized monoterpenes (E)-5a and (Z)-5a from geraniol [(E)-3] and nerol [(Z)-3] via allylation 17 and epoxidation with mCPBA (Scheme 1). We also prepared diols (S)-and (R)-5b via Sharpless dihydroxylation 18 of triene (E)-4 in excellent enantiomeric purityconfirmed by conversion to their respective benzoates 5c and chiral stationary phase highperformance liquid chromatography (HPLC) analysis (see Supporting Information)and thence acetonides (S)-and (R)-5d (Scheme 2) by ketalization. Relay-free acetonide (S)-5e was prepared from geranyl acetate as a control substrate by the use of Scafato's methods. 19 Control substrate (S)-5f was prepared by the action of Grubbs catalyst 1 on (S)-5d, thereby inherently confirming the ability of the allyl group to function as a relay in this situation. Boronates (S)-and (R)-5g were also prepared from diols (S)-and (R)-5b by direct condensation with phenyl boronic acid in ethyl acetate. These latter substrates, now incorporating UV-active chromophores, could be analyzed directly by HPLC for enantiomeric purity and were found to have identical enantiomeric excesses to benzoates (S)-and (R)-5c (see the Supporting Information).
Attention now turned to the assembling of a collection of suitable trisubstituted alkenes for this study. Trisubstituted alkenes 7a and 7d−7k were prepared by the Wittig reaction of isopropyl phosphonium iodide (6) with aldehydes (Scheme 3). 20 Alternatively, trisubstituted alkene 7c could be prepared by the reaction of the corresponding benzylic Grignard reagent with prenyl bromide under Pd(0) catalysis. 3b The former method is preferred, as nonperfect regioselectivity from competing S N 2′ attack is possible in the latter. Prenyl acetone 7l was commercially available, as was terminal alkene 7b, which was used for control experiments (vide infra).
With these substrates in hand, we selected relay (E)-5a and trisubstituted alkene 7a as the partner olefin to test in the proposed ReXM reaction. It is well established that trisubstituted olefinsclassified as type III olefins 21 do not homodimerize, and this prompted us to use trisubstituted alkene 7a in excess with the expectation that this would thereby help facilitate the desired cross-metathesis. Although various attempts to mediate the proposed ReXM in toluene or dichloromethane solution failed, neat epoxide (E)-5a underwent smooth ReXM using 10 mol % 1 with trisubstituted alkene 7a (5 equiv) at 50°C to provide functionalized epoxypolyene 8a in excellent isolated yield (Table 1, entry 1). Surprisingly, the use of Hoveyda−Grubbs precatalyst 2 under identical conditions gave only a trace of the product 8a in a complex product mixture (entry 2), and all further metatheses were conducted with catalyst 1. In further stark contrast, the use of terminal olefin 7b under the same conditions (entry 3) with epoxide (E)-5a gave instead direct cross-metathesis product 9a and isomerized vinyl ether 10a as the major epoxide-containing products, demonstrating that the use of a trisubstituted alkene is critical for these reactions. Control experiments with acetonides (S)-5d−f (entries 5−7) 22 verify also the vital role of the relay in this ReXM process, and a comparison with the reaction with Z-epoxide (Z)-5a (entry 4) Scheme 2. Synthesis of Various Δ 6,7 -Functionalized Monoterpenes 5b−g Scheme 3. Synthesis of Various Trisubstituted Alkenes Table 1. ReXM of Relay-Actuated Δ 6,7 -Functionalized Monoterpenoids with Homoprenyl Benzenes Using 10 mol % GII Catalyst (1) establishes the olefin geometry in the relay substrate as unimportant. Further examples of epoxides (E)-and (Z)-5a with various homoprenyl benzenes 7c-g establish the generality of the method (entries 8−13). 23 In all successful cases, the ReXM products 8a-g were obtained with moderate E-olefin selectivity (ca. 2−3:1), as inseparable isomers, which is a limitation of the method. 24 However, these selectivities are directly comparable to those previously reported for the formation of trisubstituted olefins by cross-metathesis with ruthenium benzylidene precatalysts (cf Figure 1c,d). 10−12 A possible catalytic cycle for this ReXM process using representative epoxide (E)-5a with homoprenyl benzene 7a invokes Diver 15 for the conversion of A to B with loss of dihydrofuran ( Figure 2). The regioselective reactions of ruthenium species of type B with trisubstituted olefins have been proposed by Robinson, 12 which would produce the ReXM product 8a and ruthenium isopropylidene C. In this scenario, the catalytic cycle would be closed by re-initiation of ruthenium isopropylidene C 11a on the terminal olefin of relay epoxide (E)-5a with concomitant loss of isobutylene. 25 This mechanism is consistent also with the results obtained using nerol versus geraniol-derived substrates (cf, Table 1, entries 1 vs 4 & entries 8 vs 9) as the same ruthenium alkylidene of the type B should be formed after initial relay metathesis.
With the ReXM method established for the reaction with homoprenyl benzenes, we explored further reactions with a variety of relay substrates and different trisubstituted alkenes as a general method for the functionalization of pre-existing trisubstituted olefins in acyclic monoterpenoid alcohols (Table  2). Thus, epoxide (E)-5a underwent smooth ReXM with aliphatic trisubstituted alkene 7h to give ReXM product 8h in excellent yield ( Table 2, entry 1). α-Branching of the alkyl chain as in olefin 7i (entry 2) proved to be detrimental to the process, where β,β-dimethylstyrene (7j) and prenylbenzene (7k) (entries 3−4) as partner olefins also failedproducing only truncated alkene 5hpresumably on the basis of increased steric demand in each of these partner olefins. Readily available prenyl acetone 7l gave the ReXM product 8i (entry 5), but diol (S)-5b unexpectedly failed to undergo ReXM (entry 6), resulting in truncated compound 5i and    In order to overcome the inherent E/Z mixture limitation of this cross-metathesis method, we elected to demonstrate an epoxypolyene cyclization with the expectation that any resulting products would have more marked polarity differences. Accordingly, we prepared ReXM product (R)-8c from enantiomerically pure epoxide (R,E)-5a and homoprenyl methoxybenzene (7c) in good yield (81%) as an inseparable 2:1 E/Z mixture (Scheme 4). Boron trifluoride-promoted epoxypolyene cyclization of this E/Z mixture provided single enantiomer podocarpane-type tricycle 11 (56% yield based on E-8c) as a single diastereoisomer, which was readily separated away from the other components in the reaction mixture. 27 To the best of our knowledge, tricycle 11 has not previously been prepared in a single enantiomer form, 28 thereby validating the utility of this two-step metathesis-cyclization sequence. 29

■ CONCLUSIONS
In conclusion, we have designed and demonstrated a novel ruthenium benzylidene-catalyzed relay cross-metathesis ("ReXM") reaction for the preparation of podocarpane-type epoxypolyene cyclization substrates from relay-actuated Δ 6,7functionalized monoterpenoid alcohols with homoprenyl benzenes. It constitutes also a general method for the crossmetathesis of pre-existing trisubstituted olefins in other relayactuated Δ 6,7 -functionalized monoterpenoid alcohols with various other trisubstituted alkenes to form new trisubstituted olefins, thereby facilitating the ability to valorize terpene biomass. The limitation inherent in the method regarding E/Z selectivity requires further advances in catalyst development to provide E-and Z-selective ruthenium benzylidene catalysts for trisubstituted olefins. However, in this situation, this can be overcome by cyclization of a E/Z-epoxypolyene substrate to give a separable, enantiomerically pure, podocarpane-type tricycle (from the E-geometrical isomer) in comparable yield to such cyclizations already reported in the literature. 2 ■ EXPERIMENTAL SECTION Experimental Techniques. All reactions were carried out in oven-dried glassware. Air-sensitive reactions were performed under a positive pressure of nitrogen unless stated otherwise. Reaction temperatures other than room temperature were achieved using an oil bath, ice/water bath, or dry ice/acetone. "Concentrated" refers to concentrating of the solution in vacuo. "Chromatographed" refers to flash column chromatography on silica gel, particle size 33−70 or 40− 63 μm, unless otherwise stated. "DCVC" refers to dry column vacuum chromatography on silica gel, particle size 33−70 μm. 30 Analytical thin-layer chromatography was performed on silica gel 60 F254 precoated aluminum-backed plates and visualized with either irradiation with UV light (254 nm) or potassium permanganate, vanillin, or phosphomolybdic acid staining. Brine refers to a saturated aqueous NaCl solution.
Characterization. Fourier transform infrared (IR) spectra were recorded neat using an attenuated total reflection (ATR)-IR spectrometer and absorptions are reported to the nearest wavenumber. The (expected) very weak CC and sp 2 C−H bond stretches for trisubstituted alkenes 7 and 8 failed to be automatically pick peaked because they fell under the peak picking threshold, although they can be observed (in most cases) by careful inspection of the spectra. 29 1 H and 13 C NMR spectra were recorded on either a Bruker DRX-400 or Bruker AV-400. Chemical shifts (δ) are expressed in parts per million (ppm) relative to the residual solvent peak. 1 H NMR spectra were recorded at 400 MHz. 13 C NMR spectra were recorded at 101 MHz. NMR acquisitions were performed at 298 K unless stated otherwise. Abbreviations are: s, singlet; d, doublet; t, triplet; q, quartet; qu., quintet; m, multiplet. High-resolution mass spectrometry (HRMS) was conducted by the Imperial College Department of Chemistry Mass Spectrometry Service.
Reagents. Allyl bromide was distilled freshly before use; otherwise all reagents were obtained from commercial suppliers and used as received.