Total Synthesis of the Schisandraceae Nortriterpenoid Rubriflordilactone A

Abstract Full details of the total synthesis of the Schisandraceae nortriterpenoid natural product rubriflordilactone A are reported. Palladium‐ and cobalt‐catalyzed polycyclizations were employed as key strategies to construct the central pentasubstituted arene from bromoendiyne and triyne precursors. This required the independent assembly of two AB ring aldehydes for combination with a common diyne component. A number of model systems were explored to investigate these two methodologies, and also to establish routes for the installation of the challenging benzopyran and butenolide rings.


Introduction
The Schisandraceae family of climbing plants are widely distributed throughout east Asia. Extracts of thesep lants have been employedi nt raditional medicine for thousands of years,w ith uses ranging from antihepatitis and anticancer properties, to antioxidant and immune regulatory activity. [1] Due in part to this ethnopharmacological history,m uch effort has been dedicated to the characterization of their bioactive constituents, resulting in the isolation of more than 420 triterpenoids from Schisandraceae speciess ince1 973. [2] Around at hird of these have been termed "schinortriterpenoids", whichs pecifically refers to nortriterpenoidsi solated from the Schisandra genus; as ubset of representative structures is depicted in Figure 1. The schinortriterpenoids are thought to derive from the rearrangemento facycloartane-type carbon skeleton, [2] and invariably feature complexp olyoxygenated ring systemsw ith numerous stereocenters. Aside from their inherent skeletal complexity,t hey have attracted more recent attention due to their moderate anti-HIVactivity,c oupledw ith low cytotoxicity. [2] Following the first isolation of as chinortriterpenoid (micrandilactone A) in 2003, [3] the first total synthesis of am ember of this natural product family was achieved by  with the synthesis of (AE)-schindilactone A( 1). [4] Sincet hen, total syntheses of severalo ther members of this family have been reported, including rubriflordilactone A [5] (2,L i, 2014), [6] schilancitrilactones Ba nd C( 3, 4,T ang, 2015), [7] propindilactone G( 5,Y ang, 2015), [8] rubriflordilactone B( 6,L i, 2016), [9] and lancifodilactone G(7,Y ang, 2017). [10] Despite pronouncedstructural diversity,s everalf eatures remain commont om ost of these naturalp roducts, includingafused AB ring systemc omprisingag-lactone and gem-dimethyl substituted tetrahydrofuran, an eighbourings even-membered Cr ing, and an a-me-  -g-lactone,w hich is unsaturated in most family members. Amongt his myriado fs ynthetically attractive targets, [11] our group has been particularly interested in the rubriflordilactones, [12] for which we reported at otal synthesis rubriflordilactone Ai n2015. [13] Here we disclose full details of the evolution of our strategy towards this naturalp roduct, including the exploration of an umber of model systems that provided valuable information in the synthetic campaign, and insighti nto the scalability of the ultimate synthetic route.

Results and Discussion
From ar etrosynthetic perspective (Scheme 1), we planned that the butenolideGr ing could be introduced by al ate-stage Mukaiyamaa ldol-type addition of silyloxyfuran 9 onto an oxocarbenium ion, which could be formed from an acetal derivative 8.F ollowing functional group manipulations including Fr ing formation, pentacycle 10 was furtherd econstructed via two metal-catalyzed polycyclization strategies, [14] involving either palladium-catalyzed cyclization of bromoendiyne 11, [15] or cobalt-catalyzed cyclotrimerization of triyne 12, [16] both of which offered ap owerful means to construct the core CDE ring system and its central pentasubstituted arene Dr ing in a single step. These substrates would be accessed by addition of the diyne 13 to appropriate AB ring aldehydes 14 or 15.N otably,t his strategy neatly segregates the AB ring system, which is commont om osts chinortriterpenoid natural products, from the rest of the framework, where most variation is found. Investigations commencedw ith the constructiono ft he AB ring system, which would need to bear either ab romoalkene (14)o rt erminal alkyne (15)m oiety.T he former route began with hydrostannylationo fb utyne-1,4-diol, [17] followed by ar egioselectivem onosilylation (16,S cheme2,y ields are given for the largest scales these reactions were performed on). Stille coupling with 2,3-dibromopropene, followed by ap rotecting group switch,g ave allylic alcohol 18 bearing the requisite bromoalkenes ide chain. AS harpless asymmetrice poxidation [18] allowed the smooth installation of the key AB ring stereocentres, and the resulting epoxide underwent regioselective ring opening [19] with allylmagnesium chloride to give primary alcohol 19. Formation of the b-lactone 20 in three steps enabled the introductiono ft he Br ing gem-dimethyl motif by doublea ddition of methylmagnesium bromide, to generate diol 21.T he formation of significant amounts of byproductk etone 22 was also observed (see below).
As imilar strategy was followed for the synthesis of the analogousa lkyne derivative 23.T he route this time started with propargyl alcohol, whichwas efficientlyc onvertedt oa lkyne 24 in two steps, before undergoing a syn-carbocupration [20] with a propargylic cuprater eagent [21] generated in situ from the corresponding Grignard, to give enyne 25 after reduction of the ethyl ester.C ompound 25 underwent an equivalent sequence of transformationsa sd escribed above to give diol 23,t his time with aT MS-protected alkyne sidechain. In comparing the efficiency of these two approaches, the first striking differencea ppearsd uring the synthesis of Sharpless epoxidation substrates 18 and 25,w here the route to the latter proved significantly higher yielding( 87 %t o25 from propargyl alcohol versus 17 %t o18 from butyne-1,4-diol). The root of this problem was the capricious Stille coupling of 16 with 2,3-dibromopropene, which is not only inconvenient on large scale due to toxicityi ssues,b ut was also consistently plaguedb yt he formation of side-product 27,w hich resulted from as econd Stille couplingo fd esired product 17 with stannane 16.S everalc onditions weres creened in attempts to improve this step, as election of whicha re depicted in Figure 2a.
Neitheri ncreasing the equivalents of dibromopropene (entry 2), nor slow additionofthe stannane to amixture of catalyst and electrophile (entry 3), improved the yield, albeit the latter conditions did suppress the formation of 27.S witching to other sources of palladium did not improvem atters (entries 4, 5), including the use of aP d II -succinimide complex reported by Taylor and Fairlamb to give superiory ields in p-allyl Stille couplings to common Pd 0 sources. [22] Surprisingly,u se of an allylic carbonatec oupling partner (X = OCO 2 Me, Entry 6) delivered none of the expected p-allyl Stille coupling product, [23] yieldingi nsteada1:0.86 mixture of alcohols 28 and 29,a gain highlighting the unexpected reactivity of the vinylic CÀBr bond over the allylic electrophile. The reluctance to form a pallylpalladium intermediate from dibromopropene may be due to the inductive electron-withdrawing effect of the bromine atom at the 2-position disfavoring p-complexation of the metal prior to oxidative addition. In af urther attemptt oo vercome this problem, we examined bromoalkene installation by alkyne bromoboration (using BBr 3 or B-Br-9-BBN); an umber of desilylated alkynes were screened from the sequence towards alkyne 15,b ut all led only to decomposition.
As econd major difference between the two routes arose during the addition of methylmagnesium bromide to b-lactone 20,a nd its equivalent in the alkyne route (30,F igure 2b). When using 20,t ertiary alcohol 21 was isolated in 66 %y ield along with epimerizedk etone 22 (24 %), which due to being a poorly-separable epimeric mixture provedi mpossible to recycle in satisfactory yield. On performing the same reactiono n alkyne-lactone 30,n oe pimerization of byproduct ketone 26 was observed, andi ts subsequent reaction with methylmagnesium bromide smoothly delivered an additional 34 %o fd iol 23 (an overall yield of 75 %). Ap ossible explanation for this difference in reactivity involves formation of complex 31 following addition of MeMgBrt ol actone 20,v ia chelation of the tertiary alkoxide, OPMB group,and carbonyl oxygen to the magnesium ion. The stereoelectronically-favored attack of as econd methyl nucleophile necessitates pseudo-axial approach from the hindered concavef ace of this chelate, which instead undergoes epimerization duet oa lignment of the s (CÀH) and p* (C=O) orbitals in this conformation.O nt he other hand, when performing this reactiono n30,t he presenceo fanon-coordinating OTBS group affords al ess constrained intermediate, whichi sa ble to accessc onformationss uch as 32 in whicht he methyl nucleophile can now approachw ith less steric hindrance, and epimerization is disfavored due to poor alignment of the s (CÀH) and p* (C=O) orbitals. Despite these complications, we were nonetheless able to access appreciable quantities of the AB ring carbon frameworks for each of the two derivatives, with the synthesis of 21 performed on more than gram scale with good yields.W ith these intermediates in hand, we now addressed completion of the AB rings, and the unveiling of the aldehyde sidechainfrom the pendent allyl group.
The second part of the optimized route towards the AB ring aldehyde 14 commencedw ith an oxidative cleavage of the alkenei n21,w hich resulted in the regioselective formation of lactol 33 (Scheme 3). This ring was destined to act as aprotecting group for the neighbouringt ertiarya lcohol, in order to avoid unreactive substrates later in the sequence. [12a] Oxidative PMB ether cleavagew as found to proceed optimally by as hort exposure to trifluoroacetic acid; subsequentm ethyl acetal formation delivered diol 34.T his was oxidized under Parikh-Doering conditions, andc onverted to lactol 36 via a( Z)-selective Ando olefination/lactonization, [24] followed by hydrolysis of the methyl acetal. Finally,c yclization of the Br ing wasr eadily achieved by ah igh-yielding oxa-Michael reactionu nder mildly basic conditions, delivering the AB ring aldehyde 14.
When the same sequence was performed on the alkyne derivative 23,w en ow observed the formation of am ixture of four isomersa fter the initial oxidative cleavage step, correspondingt ot wo regioisomeric lactols, each as an epimeric mixture. The differenceb etween this and the bromoalkene route can presumably be explained by the increased steric bulk of the bromoalkene sidechain disfavouring lactols equiva- T hese isomers could not be separated, and were carried through the next two steps as am ixture, after which we were able to separate the isomeric aldehydea cetals 39 and 40 in good overall yield (77 %o ver two steps). Pleasingly,t hese successfully converged to lactol 41 after olefination and acetal hydrolysis, presumablyd ue to acid-catalyzed isomerization during methyla cetal deprotection. AB ring aldehyde 15 was isolated as as ingle compound after oxa-Michaelc yclization.
We now hadi nh and the aldehydes required for the different cyclization strategies, which were both due to be coupled with the CDE ring diyne 13.T he synthesis of this crucial component started from carboxylic acid 42 (Scheme 4, obtained in two steps from 1,5-pentane diol), [25] whichu nderwent esterification with enantiopure alcohol 43 (prepared by enzymatic kinetic resolution, absolute configuration confirmed by Mosher ester analysis). [26] The resulting ester 44 underwent ad iastereo-selectiveI reland-Claisen rearrangement, setting up the two vicinal stereogenic centreso fc arboxylic acid 45.The stereoselectivity of this rearrangement arisesf rom the high (Z)-selectivity of the enolization in the presence of excesst riethylamine, as observed by Collum et al. [27] It is worth noting that the use of other conditions, such as LiHMDS/TMSCl, [25] led to inferiord iastereoselectivity (9:1) compared to the direct rearrangement of the lithium enolate. Carboxylica cid was converted to its methyl ester (TMSCHN 2 )t oi mprove the subsequent reduction to primary alcohol 46,w hichw as oxidized to aldehyde 47 (95 %y ield over three steps);d irect reduction of acid 45 to alcohol 46,o ro ft he intermediate methyl ester to aldehyde 47, delivered poor yields/ decomposition. Stork-Zhao olefination [28] of 47 and in situ elimination of the intermediate (Z)-vinyl iodide yieldeda lkyne 48,w hichw as converted to its benzyldimethylsilane derivative 49 in good yield. Installation of the second alkyne was achieved by PMB ether deprotection,oxidation, Ramirez olefination, and Fritsch-Buttenberg-Wiechell rearrangement to give diyne 13 in excellent yield. Overall, 13 could readily be prepared on multigram scale ( % 3.5 g) in a total of 12 steps from alcohol 43 (43 %o verall yield). With diyne 13 in hand, we turned our attentiont oe xploration of its conversion to the full CDEFG ring system via metal-catalyzed cyclization and FG ring construction,a sap reludet oa na ssault on the naturalproduct itself.
The palladium-catalyzed polycyclization of bromoendiynes has been known for more than 25 years, since the pioneering work from the groups of de Meijere [15c, 29] and Negishi. [15a,b] Althought his field has expanded to the synthesis of many different molecules, including natural product-like polycyclic systems [30] and axially chiralb iaryls, [31] no applicationso ft his chemistry in natural product total synthesis had been reported at the outset of our work. Ap articular challenge in the present synthetic context is the need for the palladium catalyst to mediate a7 -membered ring formation, af eat not reported in the bromoenediynem ethodological context, albeit precedented in other carbopalladative cyclizations. [32] Importantly,t he bromoenediynec yclization benefits from the presenceo fasilane substituent at the terminus of the diyne component, which avoids side reactions based on furtherc arbopalladation processes. This is particularly convenientf or our synthesis, where we envisaged that this alkynylsilane, which is converted to an arylsilane in the cyclization, would serve as am asked phenol through eventual aromatic Ta mao oxidation.
[33] 7-membered ring formation is similarly challengingfor the cyclotrimerization approach, [16a-c] but for both strategies we had already demonstrated that simple bromoendiynesa nd triynes could access the abridged CDE cores.B efore embarking upon the synthesis of the full rubriflordilactone Af ramework, we now decided to explore advanced model systems to assess the feasibility of each route with diyne 13,a nd also to establishc onditions for Fa nd Gr ing installation, and therebyt he synthesis of at runcated CDEFG rubriflordilactone Aa nalogue.
The palladium-catalyzed cyclization of bromoenediyne 51 (generated in two steps from 13 and aldehyde 52,7 3%)w as first studied (Scheme 5). We werepleased to find that the addi-tional pendenta lkene functionalityi nt his diyne was tolerated, althoughasmall optimization was necessary to reach as atisfying 69 %y ield of tricycle 53.W hen employing triyne 54 (obtained from 13 and aldehyde 55,7 7%), we were delighted to obtain an 80 %y ield of 56 for the cobalt-catalyzed cyclotrimerization under microwaveh eatingc onditions, whichw ere essentialf or success in 7-membered ring formation. [12b, 34] The cyclization also provede fficient with the TBS-protected triyne variant 57,a lbeit in slightly lower yield (73 %), which allowed the two parallela pproaches to converge at ac ommon intermediate 53 in readiness for Ta mao oxidation and cationic reductiono ft he benzylic alcohol/ ether.I nt his oxidation, the conditions optimized previously required slight adjustment [35] to achieve good yields of phenol 58,d ue to the problematic formation of disiloxane 60 as observed in crude 1 HNMR spectra. Cationic benzylic reduction of 59 proceeded smoothly,d elivering 61 in 77 %y ield. This two-step sequence could similarly be performed on the free alcohol 56.
The introduction of the Gr ing could conceivably proceed through either of these intermediates. To explorethesealternatives, we first studied two simpler model systemsl acking the 7-membered Cr ing and the Fr ing methyl group. The first of these was aldehyde 65,a na nalogue of aldehyde 63.T his was prepared from known indanone 66 [38] (Scheme 6) by Wittig olefination, alkene and ester reduction, and protection of the phenol as at riethylsilyl ether (to prevent issues with equilibrium of the intermediate phenol-aldehyde and the corresponding lactol). Promoted by BF 3 ·OEt 2 , 65 was reacted with silyloxyfuran 67 (prepared in four steps from citraconic anhydride) [39] to afford the DEG ring alcohol 68 in quantitative yield.
We first envisaged completion of the DEFG rings by intramolecular 1,6-oxa-Michael addition from diene 69.W hilst formation of this diene from 68 was unproblematic, all attempts to effect the conjugate addition failed, including aw ider ange of basic (e.g. tBuOK, NaH, KHMDS, Cs 2 CO 3 )o ra cidic (e.g. InCl 3 , Sc(OTf) 3 ,F eCl 3 ,T iCl 4 ,L a(NO 3 ) 3 ·5 H 2 O, p-TsOH) conditions. These reactions mostly resulted in no conversion (which we attributed to the poor nucleophilicity of the phenoxide ion and/ or the reversibility of addition) or decomposition.
Other pathways weret hen explored using phenol 71,w hich was accessed by acidic deprotection of the TESg roup in 68. [40] Scheme5.Reagents and conditions: a) 13,LiHMDS, THF, À78 8C; then add aldehyde; b) TBSCl, imid.,DMAP,CH 2 Cl 2 ;c)Pd(PPh 3 ) 4 (5 mol %), NEt 3  However,a ttempted intramolecular palladium-catalyzedC ÀO bond formation [41] after regioselective conversion of the phenol group to triflate 72 provedu nsuccessful. Af inal attempt to construct the Fr ing was made via an intramolecular Mitsunobu reaction, [42] which resulted only in the elimination of the alcohol to diene 69 even at temperatures as low as À78 8C. We surmised that the stability deriving from conjugation of the diene in 69 was an insurmountable barriert oc yclization, ahypothesis that was confirmed by the successful isolation of dihydroDEFGr ing product 73 when the equivalent Mitsunobu cyclization was carriedo ut on the saturated derivative 74.
With these unfruitful results in hands, we decided to adjust the order of FG ring formation,a nd examine the addition of a butenolide nucleophile to an oxocarbenium ion derived from an Fr ing acetal (or similard erivative). Based on some preliminary experiments using pyranyl acetals or acetates, it soon becamea pparent that ag ood leavingg roup would be required in the generation of the oxocarbenium ion, and we were attracted to the report of Vercellotti et al. on the use of thionyl chloride and zinc chloride to effect the formation of pyranosyl chlorides. [43] Initial attempts to form an Fr ing chloropyran from lactol 75 [44] (Scheme7)u nexpectedly resultedi n the formation of mixtures of the desired chloropyran 76,a nd dimer 77;h owever,w ef ound that by extendingt he reaction time, 76 could be isolated in excellent yield ( % 93 %, crude) and as as ingle diastereomer,w hich was assigneda st he aanomer on the basis of coupling constantsi nt he 1 HNMR spectrum. [44] Monitoring of the reaction by 1 HNMR spectroscopy revealed ar apid initial formation of the dimer (< 10 min), followed by slower conversion to 76.W eh ypothesize that acti-vation of lactol 75 by thionyl chloride/zinc chloride indeed leads to the formation of the corresponding oxocarbenium ion, which is rapidly trapped by unreactedl actol to give dimer 77,aprocess that is favoureda tt he beginning of the reaction when lactol concentration is high. However,d imer formation appearst ob eareversible process, where the action of zinc chlorider eforms an oxocarbenium that is trapped by chloride as the reaction progresses.
With chloropyran 76 in hand, we next addressed the addition of siloxyfuran 67.O ptimal results were found using 1.5 equivalents of 67 with 40 mol %o fz inc chloride in CH 2 Cl 2 , allowing the reaction mixture to slowly warm up from À40 8C to room temperature overnight.T his led to the isolation of 47 %o ft he DEFG ring system 78,a sa1:1m ixture of diastereomers. Pleasingly,f acial selectivity for addition to the oxocarbenium ion was high, presumably directed by the proximal stereocentre and stereoelectronic effects;t he low stereoselectivity at the newly formed butenolide stereocentre presumably reflects poor facial selectivity in an open transition state. In an effort to improvet his ratio, we examined epimerizationo ft his stereocentre, reasoning thatt he conversion of the newly appended Gr ing back to as ilyloxyfuranc ould improve the dr (dr = diastereomeric ratio) upon reformation of the butenolide by hydrolysis. Siloxyfuran 79 was generated by treatment of 78 with TIPSOTf and triethylamine at 0 8C, and av ariety of conditions were then screened to reform the Gr ing (e.g.,T BAF, THF;( AE)-camphorsulfonic acid (CSA), MeOH;a q. citric acid/ CH 2 Cl 2 ;A cOH/THF/H 2 O). Unfortunately,t hese all yieldeda lmost exclusively the elimination product 69,l ikely again due to the good leaving group ability of the phenol,w ith only trace amountsof78 observed as am ixture of diastereomers.
Despite the poor dr observed with this model system,t he successful incorporation of the Gr ing was nonetheless encouraging.W ew ere delighted to observe similar reactivity when switchingt ot he more elaborate CDEF ring system 64 (Scheme 8), with formation of the chloropyran derivative 80 from the corresponding dimer 81 (4 equiv SOCl 2 ,5 equiv ZnCl 2 ). Monitoring of the reactionb y 1 HNMR spectroscopy (Scheme 8a-c) showed almost complete conversion of lactol 64 to am ixture of dimer 81 and chloropyran 80 over 45 minutes, although almostt wo hours reactiont ime was necessary to reach complete conversion (Scheme 8a). The reaction pathway was confirmed by subjecting dimer 81 to similarr eaction conditions (8 equiv SOCl 2 ,1 0equiv ZnCl 2 ,S cheme 8b). The obtained chloropyran was used without further purification in the ZnCl 2 -promoted butenolide addition step. Pleasingly,t his delivered the CDEFG ring system as as ingle diastereomer at the F ring stereocentre,a nd a1 :1 diastereomeric mixture at the butenolides tereocentre (82 and 83) [45] in 45 %y ield over two steps. The truncated natural product 82 is of interest itself in the study of structure-activity relationshipsi nt he rubriflordilactones.
These collected models ystemsh ad provided valuable information on suitable method been developed to optimize incorporation of the Fa nd Gr ings. Despite poor diastereoselectivity at the butenolide stereocentre, the furan addition reactione xhibited high stereoselectivity at the Fr ing, ande xcellent regioselectivity on the butenolide itself, which gave us much confidence for our assault on the total synthesis of rubriflordilactone A. This commenced with the separatea ddition of the CDE diyne 13 to AB ring aldehydes 14 and 15 (Scheme 9). To our delight, the TBS-protected secondary alcohol 84 (deriving from aldehyde 14)u nderwent palladium-catalyzed cascade cyclization to give 85 in an excellent 91 %y ield. As observed previously,t he triynes arising from addition of 13 to aldehyde 15 could be cyclotrimerized to the ABCDE ring system without (triyne 86)o rwith (triyne 87)p rotection of the secondary alcohol as aT BS ether.H owever,adecrease in yield was noted when using the protected triyne 87 (54 %), due to an unexpected isomerization of the alkene sidechain to terminal alkene 88 (22 %). As this behaviour was not observed in the model systems, or indeed for triyne 86 (67 %o fA BCDE alcohol 89), we assume that the presence of this bulky silyl ether substituenta djacentt ot he site of cyclotrimerization must retard the initial oxidative coupling, which allows isomerization to compete. [46] Ta mao oxidation and benzylic reduction were performed on both the free alcohol 89 and its silyl ether analogue 85,d elivering phenol 90 in 65 %y ield (2 steps) and 51 %y ield (3 steps) respectively;astepwise alkene dihydroxylation/ diol cleavage then cleanly installed the Fr ing (91,8 5% yield). As observed before,t reatment of lactol 91 with thionyl chloride and zinc chlorideg enerated chloropyran 92 via dimer 93,w ith ar eaction time of 3h required for complete chloropyran formation. Introduction of the Gr ing was carriedout under the same conditions as used for the model systems, which led to the isolation of rubriflordilactone A 2 in 38 %y ield, along with 33 %o f its C23-epimer 94 (over 2s teps).
Although all spectroscopic data for synthetic (+ +)-2 werei n agreement with that of the naturalp roduct, and the Li group's synthetic sample, the specific rotation was found to be equal in value but of opposite sign to the isolations ample ([a] D 25 + 58.3 (c = 0.114 g/100 mL MeOH);l it. [a] D 25 À58.1 (c = 0.114 g/100 mL MeOH)). [5,26] However, communications with the Li group revealed that both synthetic samples were in agreement. Althought his at first suggestst hat both we (and Li et al.) had Scheme8.Reagents and conditions:a)SOCl 2 ,ZnCl 2 ,PhMe, 0 8C!RT;b)ZnCl 2 ,CH 2 Cl 2 , À40 8C!RT.Graph a:Reaction profile (monitored by 1 HNMR for conversion of lactol 64 to chloropyran 80 via dimer 81;G raph b:R eactionprofile for conversiono f81 to 80;G raph c:Example 1 HNMR timecourse experiment. synthesized the unnatural enantiomer of the naturalp roduct, the stereoselective nature of our synthetic routes, combined with the conservation of the stereochemistry of the AB rings in other schinortriterpenoid compounds (including those calculated and measured using CD spectroscopy, [47] and prepared by synthesis) indicateso therwise.A ne xplanation for this discrepancy remainsu nclear.

Conclusion
As eries of model studies provided af irm synthetic footing for the formation of the CDE rings of (+ +)-rubriflordilactone A using either ap alladium-catalyzed cascade cyclization or a cobalt-catalyzed cyclotrimerization, and for the installation of the Fa nd Gr ings thought he intermediacy of ac hloropyran. NMR studies revealed that this latter compound is formed via ap yran dimer,w hich over time is converted to the chloropyran. The two transition metal-catalyzedr outes were appliedt o the total synthesis, with late-stage convergencyb etween the approaches. Comparison of spectroscopicd ata revealed an inconsistency in the specific rotationb etween synthetica nd isolation samples, which remains unresolved. The modularn ature of the synthesis renders it suitable for application to other members of the Schisandraceae family;e fforts towards the total synthesis of such compounds are currently underway in our laboratory.

Experimental Section
Experimental details are given in the Supporting Information. These include details of the synthetic procedures, spectroscopic data, and copies of the 1 Ha nd 13 CNMR spectra for novel compounds.