Transition-metal and organocatalysis in natural product synthesis

The total synthesis of natural products, a field of organic chemistry that is both historical and contemporary, has undoubtedly entered a new paradigm over the past decade with the advent of revolutionary new synthetic methods and innovative synthetic concepts. Putting aside the downstream applications of natural-product synthesis in the interrogation of biological processes, elucidation of biogenetic origins, structural assignments and many others, its fundamental and indispensable value as a vehicle for the discovery of new synthetic transformations is well-testified and unparalleled by any other research discipline over the history of chemical science. These transformations, largely concerning carbon–carbon/carbon–heteroatom bond formations, asymmetric induction, and catalysis, constantly expand the repertoire of powerful tools available at the organic chemist’s disposal, and enable more challenging synthetic problems to be investigated. As such, this catalytic cycle of discovery fueled by natural-product synthesis continues to capture and captivate the imagination of both practitioners and students of organic chemistry around the world and will do so far beyond the foreseeable future. In particular, the recent discovery of novel transitional-metal complexes and their associated chemical transformations, and rediscovery of the unprecedented reactivity of previously documented transition-metal complexes with subtle changes in the reaction conditions and the reacting substrate have been extremely fruitful since the turn of the millennium, most notably in promoting reactions of unfunctionalized and unactivated chemical bonds. Furthermore, the application of organic compounds as promoters of chemical transformations has also witnessed increasing sophistication, substrate scope, and efficiency together with new modes of activation, which rival or at times surpass those exhibited by transition metals. Last but not least, the judiciary combination of transition-metal and organic mediators, together with tandem processes encompassing multiple reaction cycles in a programmed sequence, represents a new horizon with vast potentials that have yet to be fully understood and exploited.

The total synthesis of natural products, a field of organic chemistry that is both historical and contemporary, has undoubtedly entered a new paradigm over the past decade with the advent of revolutionary new synthetic methods and innovative synthetic concepts. Putting aside the downstream applications of naturalproduct synthesis in the interrogation of biological processes, elucidation of biogenetic origins, structural assignments and many others, its fundamental and indispensable value as a vehicle for the discovery of new synthetic transformations is well-testified and unparalleled by any other research discipline over the history of chemical science. These transformations, largely concerning carbon-carbon/carbon-heteroatom bond formations, asymmetric induction, and catalysis, constantly expand the repertoire of powerful tools available at the organic chemist's disposal, and enable more challenging synthetic problems to be investigated. As such, this catalytic cycle of discovery fueled by natural-product synthesis continues to capture and captivate the imagination of both practitioners and students of organic chemistry around the world and will do so far beyond the foreseeable future. In particular, the recent discovery of novel transitional-metal complexes and their asso-ciated chemical transformations, and rediscovery of the unprecedented reactivity of previously documented transitionmetal complexes with subtle changes in the reaction conditions and the reacting substrate have been extremely fruitful since the turn of the millennium, most notably in promoting reactions of unfunctionalized and unactivated chemical bonds. Furthermore, the application of organic compounds as promoters of chemical transformations has also witnessed increasing sophistication, substrate scope, and efficiency together with new modes of activation, which rival or at times surpass those exhibited by transition metals. Last but not least, the judiciary combination of transition-metal and organic mediators, together with tandem processes encompassing multiple reaction cycles in a programmed sequence, represents a new horizon with vast potentials that have yet to be fully understood and exploited.
In this Thematic Series, selected examples of metal-and organic-compound-promoted chemical processes that render the preparation of architecturally complex natural products, naturalproduct subdomains, or natural-product-like scaffolds, are presented. These illustrative synthetic studies are intended to showcase the most recent developments, at the same time highlight the state-of-the-art and current limitations, and in doing so set the path for the future. It is our great anticipation that this Thematic Series will instigate and inspire further investigations in this field, and challenge the existing technologies and our current mindset in target-oriented synthetic design. Ultimately, we wish that the newly acquired knowledge will translate to further advances in synthetic organic chemistry and provide more enabling tools for synthetic-chemistry-dependent research fields and beyond.

Introduction
Ryanodine (Scheme 1) [1][2][3] is a potent modulator of the intracellular calcium release channels, known as ryanodine receptors [4,5]. Its complex architecture, including eight contiguous tetrasubstituted carbons on the pentacyclic ABCDE-ring system, has posed a formidable synthetic challenge. To date, the only total synthesis of a compound in this class of natural products was reported by Deslongchamps, who constructed ryanodol in 1979 [6][7][8][9]. Most recently, we reported the synthesis of 9-demethyl-10,15-dideoxyryanodol [10] by taking advantage of the intrinsic C 2 -symmetry of the target molecule. In this synthesis, C 2 -symmetric compounds such as bicyclo[3.3.2]decene 1 were strategically designed, and application of pairwise func-tionalizations of these molecules minimized the total number of steps.
Bicyclo[3.3.2]decene 1 was prepared from C 2 -symmetric bicyclo[2.2.2]octene 2 through a ring-expansion reaction (Scheme 1) [11]. We reported the synthetic routes to racemic 2 and enantiomerically pure 2 from 3 and 5, respectively. Specifically, the dearomatizing Diels-Alder reaction between 2,5dimethylbenzene-1,4-diol (3) and maleic anhydride lead to the construction of bicyclo[2.2.2]octene 4, which was then transformed into racemic 2 through electrolysis [11]. Alternatively, the Diels-Alder reaction between 3,6-dimethyl-o-quinone Scheme 1: Structure of ryanodine and the Diels-Alder reactions for construction of the potential intermediates of ryanodine. monoacetal 5 and 1,1-diethoxyethylene provided bicyclo[2.2.2]octene 6, which was then converted to enantiopure 2 via an enzymatic kinetic resolution [12]. The Diels-Alder reaction was effectively employed in both of these syntheses for construction of the two quaternary carbons at the C1 and C5 positions of ryanodine (highlighted in gray). However, the current route to (+)-2 from racemic 6 generated the unnecessary antipode. Therefore, development of an alternative asymmetric route to 1 was planned to further improve the overall practicality. Here we report an asymmetric Diels-Alder reaction for simultaneous installation of the C1-and C5-stereocenters using the optically active cycloheptadiene derivative 7, and its derivatization into bicyclo[3.3.2]decene 1.
We assumed that the C4-stereocenter of optically active sevenmembered diene 7 would permit the requisite stereoselective Diels-Alder reaction (Scheme 1). Namely, the reaction between 7 and acrolein was expected to stereoselectively introduce the C1, C5 and C12 stereocenters to afford bicyclo[3.2.2]nonene 8. The C11-aldehyde of 8 was then to be utilized as a handle for the ring expansion to access 1. To the best of our knowledge, construction of the two quaternary carbons by the intermolecular Diels-Alder reaction of 1,4-disubstituted cycloheptadiene derivatives has not been reported [13,14].

Results and Discussion
The synthesis of optically active 7 began from crotyl chloride (Scheme 2). The carbon chain extension of crotyl chloride by treatment with acetylacetone and K 2 CO 3 [15], followed by the addition of vinylmagnesium bromide [16], provided 9. The bromoetherification of tertiary alcohol 9 by using NBS led to tetrahydrofuran 10 as a diastereomeric mixture. Next, the baseinduced elimination of HBr converted 10 to diene 11, which underwent the Claisen rearrangement at 170 °C to give rise to heptenone 12 [17][18][19][20][21]. The more thermodynamically stable silyl enol ether 13 was regioselectively formed from 12 under Holton's conditions [22], and DDQ-mediated oxidation of 13 resulted in the formation of α,β-unsaturated ketone 14. Asymmetric reduction of ketone 14 was in turn realized by applying a stoichiometric amount of (R)-CBS and BH 3 ·SMe 2 to produce 15 (82% ee) [23]. The absolute configuration of C4 was determined as S by the modified Mosher method after acylation of Scheme 2: Asymmetric synthesis of 7 and determination of the absolute configuration at C4 of 15 by the modified Mosher method. The numbers are differences in 1 H chemical shifts between 16a and 16b (Δδ = δ16a − δ16b).

Scheme 3:
Generation of 17 through the 6π-electrocyclic reaction and the Diels-Alder reaction. [24]. Finally, the hydroxy group of 15 was protected as its TBS ether to afford 7.

using (R)-and (S)-MTPACl
We then explored the Diels-Alder reaction between 7 and acrolein to construct the bicyclo[3.2.2]nonene structure (Scheme 3). The Diels-Alder reaction under thermal conditions (100 °C) induced the decomposition of diene 7, and only the starting material was recovered after 10 h (26%). Because of the low reactivity of 7, we applied a Lewis acid to facilitate the reaction. However, the reaction of 7 and acrolein in the presence of BF 3 ·OEt 2 (50 mol %) led to formation of the unex-pected bicyclo [2.2.2]octene skeletons 17a and 17b as a 2.9:1 mixture. Under these conditions, BF 3 ·OEt 2 -promoted elimination of the allylic siloxy group in 7 generated triene 18, which then isomerized into 19 via a 6π-electrocyclic reaction. Diene 19, which appeared to be more reactive than the original diene 7, then underwent the Diels-Alder reaction to provide 17a and 17b.
Formation of 17a and 17b in Scheme 3 confirmed that selective activation of acrolein in the presence of the Lewis-basic allylic oxygen was a prerequisite to the successful formation  (Table 1, entry 5) increased the yield of the adducts. It is particularly worthy of note that the use of 2,6-di-tert-butylpyridine in combination with 200 mol % of TBSOTf effectively inhibited the Lewis-acid-promoted elimination of the C4-oxy group, and that the ratio of 8 to 20 was improved from 1.6:1 to 3.4:1 by replacement of the solvent (Table 1, entries 4 and 5). Thus, we developed an effective method for synthesis of the requisite stereoisomer 8 by applying a TBSOTf-promoted Diels-Alder reaction [25,26]. Most importantly, the C4-stereocenter behaved as the control element to introduce the two quaternary carbons (C1 and 5) and the C12-stereocenter.
The selective formation of 8 out of eight possible isomers is rationalized in Scheme 4. The endo-type transition states would be favored over their exo-type counterparts, and acrolein would approach from the bottom face of 7 to avoid steric interactions with the axially oriented C2-and C4-hydrogen atoms on the top face [27]. These considerations eliminate six out of the eight stereoisomeric transition states, and leave only TS-A and TS-B, which in fact correspond to the generated adducts 8 and 20, respectively. TS-A would be preferred over TS-B due to the unfavorable interaction of the two proximal TBS groups in TS-B, allowing formation of 8 as the major compound.

Scheme 4:
Rationale of the stereoselectivity of the Diels-Alder reaction.
Having synthesized the optically active 8, the next task was the preparation of C 2 -symmetric bicyclo afforded 22 as a single stereoisomer, and the obtained 22 was oxidized with DMDO to provide α-hydroxy aldehyde 23 as a diastereomeric mixture (dr = 2.8:1). Compound 23 then reacted with benzyl hydroxylamine to produce oxime 24, LiAlH 4 -treatment of which led to 25. The regioselective ring expansion of seven-membered 25 was induced by treatment with NaNO 2 in acetic acid [28,29], resulting in the formation of eightmembered 27 through the intermediary of 26. Finally, the desilylation of 27 with TBAF and the subsequent oxidation of the resultant hydroxy group delivered the symmetric diketone 1 in optically active form.

Conclusion
In summary, we developed a synthetic route to the optically active seven-membered 7 and established the TBSOTfpromoted stereoselective Diels-Alder reaction between 7 and acrolein to construct highly functionalized bicyclo [

Experimental
General: All reactions sensitive to air or moisture were carried out under argon or nitrogen atmosphere in dry, freshly distilled solvents under anhydrous conditions, unless otherwise noted. All other reagents were used as supplied unless otherwise stated. Analytical thin-layer chromatography (TLC) was performed by using E. Merck Silica gel 60 F254 precoated plates. Flash column chromatography was performed by using 40-50 µm Silica Gel 60N (Kanto Chemical Co., Inc.), 40-63 µm Silicagel 60 (Merck) or 32-53 µm Silica-gel BW-300 (Fuji Silysia Chemical Ltd.). Melting points were measured on a Yanaco MP-J3 micro melting-point apparatus and are uncorrected. Optical rotations were recorded on a JASCO DIP-1000 Digital Polarimeter. Infrared (IR) spectra were recorded on a JASCO FT/IR-4100 spectrometer. 1 H and 13 C NMR spectra were recorded on JEOL JNM-ECX-500, JNM-ECA-500, or JNM-ECS-400 spectrometers. Chemical shifts were reported in parts per million (ppm) on the δ scale relative to CHCl 3 (δ 7.26 for 1 H NMR), CDCl 3 (δ 77.0 for 13 C NMR), C 6 D 5 H (δ 7.16 for 1 H NMR), C 6 D 6 (δ 128.0 for 13 C NMR), CO(CD 3 )(CD 2 H) (δ 2.05 for 1 H NMR) as internal references. Signal patterns are indicated as s, singlet; d, doublet; t, triplet; q, quartet, m, multiplet. The numbering of the compounds corresponds to that of the natural products. High-resolution mass spectra were measured on Bruker microTOFII.
(PA)-directed C-H functionalization reactions beginning from easily accessible PA-protected benzylamine and aryl iodide precursors.

Results and Discussion
New synthetic strategy for phenanthridine compounds. The picolinamide (PA) group has been shown to be an excellent directing group for a range of Pd-catalyzed C-H functionalization reactions [26][27][28][29][30][31][32][33][34][35]. In 2005, the Daugulis laboratory first reported that the ortho-C(sp 2 )−H bond of benzylpicolinamides could be arylated with aryl iodides under Ag-promoted Pd-catalyzed conditions [26]. In 2012, our laboratory [28] as well as that of Daugulis [27] independently reported that picolinamide substrates can undergo intramolecular dehydrogenative C-H amination reactions to afford medium-sized N-heterocycles under the catalysis of Pd(OAc) 2 with PhI(OAc) 2 oxidant. These discoveries led us to explore whether we could develop a new strategy for synthesizing phenanthridines. As outlined in Scheme 1B, we envisioned that ortho-arylated benzylamine picolinamides could undergo an intramolecular amination at the ortho ε-C-H position of the newly installed arene group to form cyclized dihydrophenanthridines, which could be further converted to phenanthridine products under oxidative conditions. Ideally, we hoped to perform both the intramolecular C-H amination and subsequent oxidation in a single step [36].
Arylation of 2-methoxybenzyl picolinamide 1 with 4-iodoanisole (2) under various conditions. We commenced the study by investigating the arylation of 2-methoxybenzyl picolinamide 1 with 4-iodoanisole (2) under various conditions (Table 1) to form our desired arylated product 3. Our initial attempt under the original Pd(OAc) 2 -catalyzed AgOAcpromoted solvent-free condition afforded the desired arylated product 3 in good yield (Table 1, entry 1). This method, however, required the use of expensive silver salt as an additive and high reaction temperature (150 °C). We next sought to replace the silver salts with cheaper reagents and lower the reaction temperature [12]. Not surprisingly, the arylation yield dropped significantly when the reaction was performed in toluene solvent at 120 °C (Table 1, entry 2). Addition of PivOH (0.3 equiv) gave little improvement (Table 1, entries 3 and 4). To our delight, the desired arylation reaction was largely restored with the application of 2 equiv of K 2 CO 3 at 120 °C for 24 h (Table 1, entry 5). Furthermore, an excellent yield was obtained when K 2 CO 3 was replaced with KHCO 3 and 0.3 equiv of PivOH was applied ( Table 1, entry 7). The most effective carboxylate ligand and solvent was found to be PivOH and toluene, respectively.
The determination of the scope of this reaction with benzylpicolinamide and aryl iodide substrates. With the optimized conditions in hand, we next explored the scope of benzylpicolinamide and aryl iodide substrates ( Figure 1). The electronic properties of benzylpicolinamide and aryl iodides had little influence on the reactivity, as benzylpicolinamide and aryl iodide substrates bearing electron-donating and withdrawing substituents react in good yields (3, 8, and 12). Significantly decreased arylation yield was observed for ortho-substituted aryl iodides (e.g., 9). The sterics of the benzylpicolinamides is also important for the regioselectivity of the arylation reaction. For instance, the less hindered ortho position is preferentially arylated (e.g., 14) when a meta substitutent is present on the benzylpicolinamide. Aryl bromides are much less reactive compared with aryl iodide substrates 4. This is in accordance with results on the Pd-catalyzed PA-directed arylation of more inert C(sp 3 )−H bonds [29].
Cyclization of biaryl compounds to form dihydrophenanthridines. Next, we investigated the cyclization of biaryl compounds to form dihydrophenanthridines via Pd-catalyzed intramolecular dehydrogenative amination of ε-C(sp 2 )-H bonds [37][38][39][40][41][42][43][44][45]. To our delight, treatment of 3 in the presence of 5 mol % of Pd(OAc) 2 and 2 equiv of PhI(OAc) 2 in toluene at 120 °C for 24 h gave the desired dihydrophenanthridine 16 in good yield ( Table 2, entry 1). In addition, a further oxidized phenanthridine 17 was obtained as a side product. Compound 17 is presumably generated through the PhI(OAc) 2 -mediated oxidation of the benzylic C-H bond to form a phenanthridinium intermediate 18, which then undergoes a removal of the PA group. Encouraged by these observations, we proceeded to explore whether the cyclization and oxidation steps can be performed in one step to give the phenanthridines in a shorter procedure. A variety of oxidants, such as 1,4-benzoquinone (BQ), KMnO 4 , ceric ammonium nitrate (CAN), and copper salts were examined [46]. The combination of PhI(OAc) 2 (2 equiv) and Cu(OAc) 2 (2 equiv) afforded the phenanthridine product 17 in highest yield ( Extension of the cyclization-oxidation step to other arylated picolinamide substrates. The coupled cyclization-oxidation step detailed above was then used to synthesize phenanthridines from other arylated picolinamide substrates ( Figure 2). In general, electron-rich arene motifs, installed by C-H arylation, gave a higher yield of phenanthridine products; electron-deficient substrates provide a lower yield. For instance, substrate 8 with a para-nitro group failed to give any cyclized product under the standard conditions. Substrates with moderately electron-withdrawing groups, such as 20 bearing a paraester group, reacted in moderate yield. The electronic properties of the benzylpicolinamide scaffold had much less influence on the reaction. For example, product 22 bearing an ortho-CF 3 substituent was obtained in 51% yield. Finally, it is noteworthy that all of the above phenanthridine products show intense blue fluorescence. We expect our synthetic strategy will afford access to phenanthridines bearing varied substitution patterns, enabling applications in biology and materials science.

Conclusion
In summary, we have developed a readily applicable two-step method for the synthesis of phenanthridines from easily accessible benzylamine picolinamides and aryl iodides. In the first step, an improved protocol allows us to carry out the Pd-catalyzed PA-directed C-H arylation reaction without the use of expensive silver additives. In the second step, application of PhI(OAc) 2 and Cu(OAc) 2 oxidant under the catalysis of  Pd(OAc) 2 affords phenanthridines in moderate to good yields. Applications of this method to the synthesis of more complex phenanthridines with novel photophysical properties are currently underway.

Experimental
General conditions: All commercial materials were used as received unless otherwise noted. All solvents were obtained from a JC Meyer solvent dispensing system and used without further purification. Flash chromatography was performed using 230-400 mesh SiliaFlash 60 ® silica gel (Silicycle Inc.  [2][3][4][5][6][7][8][9][10][11]. Isolated from the pantropical plant Catharanthus roseus, vinblastine (3) and vincristine (4) have potent antitumor activity and are clinically utilized in the treatment of Hodgkin's disease and leukemia [12][13][14]. Fragmentation of 3 under acidic conditions delivers two structural units, desacetylvindoline and velbanamine (2) [15]. The reassembling of catharanthine (1) and vindoline (5) into the parent alkaloids 3 and 4 by using FeCl 3 -promoted oxidative coupling supports the biogenesis of heterodimeric indole alkaloids [11]. Interestingly, velbanamine (2) was later identified in leaves and twigs of Tabernaemontana eglandulosa in 1984 [16]. Therefore, the syntheses of velbanamine (2) and structurally closely related alkaloids may be important for the syntheses of their dimeric alkaloids. The modification of these alkaloids in the context of clinical drug development still poses a challenge to synthetic and medicinal chemists. Since the first racemic synthesis of velbanamine (2) was disclosed by Büchi and co-workers in 1968, four racemic syntheses and two enantioselective syntheses have been reported in spite of several synthetic efforts toward the core structure [17][18][19][20][21][22][23][24]. The practical synthesis of velbanamine (2) and a general approach toward structurally related alkaloids remain an intriguing task in the synthetic community. Here, we would like to disclose our recent efforts on method development toward the efficient construction of velbanamine-type indole alkaloids. As shown in Scheme 2, an intramolecular Heck reaction (via 9-exo manner) would finalize the 9-membered ring, which was biogenetically derived from a retro-Mannich reaction from catharanthine (Scheme 1) [25]. The terminal alkene 6 can be disconnected to give amide 7, which may be derived from 2-bromotryptamine 9 and secodiene 10. Clearly, the chemoselective dioxygenation of 10 is the main theme in our synthetic endeavor. From here, we expect that current synthetic methods may provide a general basis for similar important structures, such as isovelbanamine and cleavamine.
The two terminal alkenes in compound 10 pose a challenge for chemoselective dioxygenation or iodolactonization. To address this problem, we turn our attention to differentiating the two types of terminal alkenes. Although iodolactonization as exemplified in Kita's elegant synthesis of rubrenolide resulted in a practical approach on desymmetrization [45], the subsequent hydrolysis still required an extra step to reveal the proper hydroxy group.

Results and Discussion
The pathway of cyclization Recently, Tellitu, Dominguez, and co-workers reported an intramolecular oxyamidation of alkene 11 with phenyliodine(III)-bis(trifluoroacetate) (PIFA) (Scheme 3) [46]. The lactam 12 was originally assigned as an unstable intermediate, which should be subsequently reduced to pyrrolidine 13. It was particularly striking to us that the arene unprecedentedly stabilizes the primary carbon cation through a neighboring participation in the mechanism proposed by Tellitu et al. To resolve the confusion, we synthesized the corresponding pyrrolidine through an alternative approach (Scheme 4). The amination of para-methoxyaryliodide with 2-hydroxymethylpyrrolidine in the presence of Cs 2 CO 3 and a catalytic amount of copper iodide in DMF afforded 13a in 60% isolated yield [47]. To our surprise, the proton NMR of 13a was distinct from that of the originally proposed 13 by Tellitu et al. [46]. Based on the Baldwin rule, both 5-exo-trig and 6-endo-trig are favorable in the cyclization. Meanwhile, the ring expansion may occur to constitute the functionalized piperidine ring, resembling Cossy's endeavor to synthesize velbanamine [48]. In this regard, the same coupling condition was applied for the reaction of 3-hydroxypiperidine with para-methoxyaryliodide. Unexpectedly, the spectrum of the resulting amine 13b was not consistent with that of 13.
Based on the general iodolactonization principle, the C-O dipole in an amide is aligned for a favorable nucleophilic addition due to the "double-bond character" in a planar amide [41].
In the PIFA-promoted cyclization, lactonization was more likely to be a prominent process than lactamization. Following Tellitu's procedure [46], we re-synthesized 13 and subjected it to the proton NMR in the presence of D 2 O. Three proton signals were exchanged with deuterium. X-ray analysis further contributed to the elucidation of the linear structure of 13 as shown in Scheme 5 [49]. The revised structure 13 implies that the amide 11 did not undergo the oxyamidation under the previously defined reaction conditions. Alternatively, it appears that iodolactonization of the amide dominated when compound 11 was subjected to the oxidation conditions of PIFA in CF 3 CH 2 OH. Mechanistically, the terminal double bond is activated by hyperiodine or via a halogenium-like intermediate [50][51][52]. The subsequent intramolecular ring-closing reaction in a 5-exo or 6-endo manner delivers iminolactones A. The reduction with borane gives hydroxyamine 13. In this new proposal, the iminolactone A could be unstable and difficult to purify according to Tellitu's findings [46]. Similar stability of iminolactone is documented in the literature [53][54][55].
Although we have established the amide-assisted dioxygenation of an alkene, the structure of the iminolatone A is still a mystery. In principle, both 5-exo-trig and 6-endo-trig are favorable during the cyclization (Scheme 6). However, the instability of the iminolactone A impedes further characterization, and consequently, the possible cyclization mode is difficult to clarify. We envisioned that the steric hindrance of the amide would prevent the hydrolysis of the iminolactone during the work-up stage. After several experiments, ortho-biphenylamide 14 was chosen as a starting point. To our delight, two isomers of iminolactone were separated on a silica-gel column. After the treatment of benzyl bromide in the presence of NaH, the acidic work-up delivered two diastereoisomers of lactone 17 (Route A). This was further confirmed by an alternative synthesis based on a known procedure [56], in which 3-substituted-γlactone 17 was derived from the stereoselective alkylation of Bn-protected 5-hydroxymethyl-γ-lactone 16 (Route B). Two diastereoisomers of the trans-isomer of 17 (ratio 4/1) were identical with compounds from the PIFA-promoted cyclization (Scheme 6, Route A). Based on these conclusive experiments, we believe that the 5-exo-trig cyclization is the favorable pathway during the PIFA-mediated amide-cyclization of alkene. Two isomers after cyclization were then attributed to the stereoisomers of the C=N double bond in the iminolactone C. This notion is also consistent with numerous reports on the iodocyclization with amide [41].

Chemoselective dioxygenation
The structural assignment of 13 and the verification of the reaction pathway encouraged us to explore the synthetic potential of the dioxygenation of alkene with the assistance of an amide. The dioxygenation can be "stopped" after the first PIFApromoted cyclization to form an iminolactone (such as the intermediate C in Scheme 6). Thus, if the original amide group could be recovered, it may "flow" into the second round of alkene dioxygenation. This "stop-and-flow" approach easily differentiates two alkene groups during the synthetic endeavor. In our experiment, after cyclization of amide 18a, the corresponding iminolactone was hydrolyzed under Mukaiyama's conditions (sat. Na 2 B 4 O 7 buffer in CH 3 CN) [53] to give amide 19a in 72% yield [57]. The different substituents on aniline ( Figure 1) slightly deteriorated the isolated yield. The geminal bis-substituted alkene gave a moderate yield of the corresponding dihydroxyamide 19d. The α-methylated substrate was also dioxygenated albeit in a low diastereoselectivity (dr 1:1).
When two alkene groups co-existed as in compound 20, desymmetrization delivered the mono-dioxygenated product 21 in a dr ratio of 1:1 (Scheme 7). The less polar 22a was confirmed by X-ray diffraction [58]. The second alkene was further dioxygenated by repeating the previous protocol to deliver 23 with a 3/1 diastereoselectivity from 22b. The encountered poor chemoselectivity under conventional metal-mediated conditions for the dioxygenation of seco-dienes such as 20 implies that this iterative method may find usage in organic synthesis.
In this mode, we can install four hydroxy groups on a secodiene with different protecting groups. Although several research groups have investigated the desymmetrization of seco-diene by using iodolactonization [59][60][61][62][63][64], our strategy here proves the concept of "stop-and-flow" to functionalize alkenes step-by-step with a simple hydrolysis in between.

Synthetic applications
After a simple acidic work-up, the established desymmetrization process can also be applied to prepare 3-alkyl-5-hydroxymethyl-γ-lactone, which has been widely found in natural products and compounds of pharmaceutical interest [65][66][67][68]. The usual methods including the alkylation at C-3 or the iodolactonization of amides or esters comprises multiple production steps, such as the hydrolysis of the halogen compounds and the protecting groups. However, the direct dioxygenation can easily construct γ-lactones by a simple acidic work-up as shown in Scheme 8. For example, when 24a (R = Bn) was subjected to the PIFA-mediated cyclization, the desired γ-lactone 25a was isolated after hydrolysis in 94% yield with a ratio of trans/cis of 1.9:1. Compound 25a can be easily converted into the key intermediate in the synthesis of indinavir, a protease inhibitor used as a component of a highly active antiretroviral therapy to treat HIV infection and AIDS [69]. When two alkene groups exist in 24b-d, the terminal alkene preferentially underwent cyclization to deliver a variety of 3-susbtituted-γ-lactones (25b-d). For comparison, the mCPBA-epoxidation approach in literature always demonstrated that electron-rich alkenes are more reac- tive leading to the reversal of chemoselectivity [70,71]. A cyclopropane group was also tolerated during this operation and lactone 25e was obtained in 71% yield with a dr ratio of 2:1.
Product 25f was particularly interesting since the "stop-andflow" strategy can differentiate two terminal alkenes. The more reactive geminally substituted alkene was dioxygenized to deliver all functional groups required in 8, which was designed as a key intermediate in our synthetic plan toward velbanamine (Scheme 2). The stereoselectivity still awaits further improvement. Nevertheless, as a proof of concept, the current approach allows us to develop an efficient synthetic route to access challenging synthetic targets.

Conclusion
In summary, with a comprehensive validation of PIFApromoted cyclization of alkenes, a synthetically useful desymmetrization approach via the dioxygenation of alkenes was developed. The "stop-and-flow" strategy allows us to easily functionalize seco-dienes step-by-step. Moreover, this approach also chemoselectively functionalizes terminal alkenes instead of internal ones. Substituted 5-hydroxymethyl-γlactones have been constructed in a protecting-group-free manner. The synthetic application in the efficient synthesis of velbanamine-type indole alkaloids as well as the enantioselective desymmetrization are currently pursued in our laboratory and will be reported in due course.

Supporting Information
Supporting Information File 1 Experimental descriptions, analytical and X-ray data.

License and Terms
This is an Open Access article under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Introduction
Neurotrophins are a family of endogenous proteins that are vital for neuron function, survival, and regeneration [1][2][3]. As such, they have prompted intense studies toward the treatment of various neurodegenerative diseases including Alzheimer's disease [4] and Parkinson's disease [5]. Despite their unambiguous importance, approaches to neurotrophin-based drug development have encountered problems associated with their limited oral availability, insufficient delivery to the central neural system and considerable manufacturing cost [6,7]. These limitations have stimulated the search for small molecules that can enhance or mimic neurotrophin activity as potential drug leads [8][9][10][11][12].
We have recently reported a unified synthetic strategy of 2, 3 and designed analogues using scaffold 7 as the key intermediate ( Figure 2) [24][25][26]. A potential drawback of this strategy is the late-stage modification of the A ring motif of 7 that requires additional steps for the synthesis of the target molecules. In an effort to overcome this issue, we describe here a second-generation strategy of framework 9 in which the C-1 center has been methylated early in the synthesis. As such, it represents an efficient route toward a diversity-oriented synthesis of several Illicium sesquiterpenes. The enantioselective entry to these molecules is based on an organocatalyzed asymmetric Robinson Scheme 1: Organocatalyzed asymmetric Robinson annulation.

Results and Discussion
During the past 20 years, organocatalysis has emerged as an important field in asymmetric stereoselective synthesis due to its advantages, which include high enantioselectivity, environmental friendliness and ease of handling . Organocatalyzed asymmetric Robinson annulation has long been proven to be one of the most powerful strategies to construct bicyclic systems with a chiral quaternary center [51][52][53][54][55][56][57][58]. Among them, the Hajos-Wiechert and Wieland-Miescher ketones represent two of the most famous examples [59][60][61][62][63][64][65]. With this background information in mind, we devised an enantioselective synthesis of 8 starting from commercially available dione 12, and the synthesis of 8 was previously published [25,26].
Tsuji-Trost allylation [66-68] of 12 produced compound 11, which was readily converted to 13 by an acid-catalyzed Michael addition with methyl vinyl ketone (MVK) (two steps, 63% overall yield) [69][70][71]. The organocatalyzed cyclization of 13 was achieved by optimizing the previously reported Tu/Zhang conditions [71] using D-prolinamide as the organocatalyst (Scheme 1). Performing this reaction at 80 °C gave rise to bicyclic motif 8 in about 70% ee (70 % yield after 12 h), while decreasing the temperature to 25 °C increased the enantioselectivity to over 99% (70% yield after 60 days). To compromise between high enantioselectivity and short reaction time, we decided to pursue this conversion at 40 °C where we obtained an enantiomeric excess of 90% (70% yield after 14 days).
Wittig olefination of the C-1 ketone with methoxymethylenetriphenylphosphine [75] yielded the corresponding enol methyl ether, which was hydrolyzed to the aldehyde under acidic conditions and reduced with NaBH 4 to form alcohol 15 with desired diastereoselectivity at the C-1 center (dr = 9:1) in 81% yield (over three steps) [76]. The stereochemistry of 15 was unambiguously confirmed by single-crystal X-ray analysis of the related tosylate derivative 16 [77]. Deoxygenation of the C-15 primary alcohol was performed by: (a) mesylation of the alcohol with MsCl; and (b) reductive deoxygenation with LiEt 3 BH (super hydride). The thioketal protecting group was then removed under oxidative conditions with [bis(trifluoroacetoxy)iodo]benzene (PIFA) to yield ketone 10 in good yield (66% over three steps, Scheme 2) [78]. This approach allowed us to produce a sufficient amount of enone 10 (>10 grams) for further functionalization.
Notably, this scalable approach rendered us several hundred milligrams of compound 9, paving the way for a diversityoriented synthesis. For example, a Mn(III) promoted C-2 allylic oxidation [24,101,102] would provide a C-2 oxygenated functionality. Similarly, C-10 α-substitution would provide a large diversity of neurotrophic analogues based on our recent findings [26].

Supporting Information
Supporting Information File 1 Experimental procedures for the syntheses of all new compounds.

Introduction
Throughout the past decade, we have been developing an aza-[3 + 3] annulation reaction as a general and unified strategy in alkaloid synthesis . Our aza- [3 + 3] annulation, which has been classified as Type-II [1], with Type-I aza-[3 + 3] annulation being reserved for Robinson's double Mannich-type process [33], utilizes readily accessible and easily handled vinylogous amides and vinyl iminium salts. It provides a significant complementary, if not superior, approach to aza- [4 + 2] cycloadditions in constructing piperidines, because the azadienes and imines required are not always the most accessible and/or easily handled substrates given the problems of isomerization and hydrolysis ( Figure 1) [34][35][36][37].

Results and Discussion
To examine the feasibility of an enone intramolecular aza- [3 + 3] annulation, a seven-step synthesis of the annulation substrate 10 commencing from 3-butyn-2-ol (5) was carried out (Scheme 2). Protection of secondary propargyl alcohol 5 as the THP-ether followed by alkylation of a lithium acetylide onto 1,4-dibromobutane afforded bromide 6 in 81% overall yield. Acid-mediated removal of THP followed by azide formation using NaN 3 afforded alkyl azide 7 in 94% overall yield.
Preparation of vinylogous amide 9 was then achieved by a twostep sequence: (a) LAH-mediated azide reduction to give a primary amine with concomitant reduction of the alkyne revealing the olefin functionality; and (b) dehydrative condensation of this primary amine with 1,3-cyclohexanedione. Oxidation of the secondary allylic alcohol in 9 with MnO 2 at ambient temperature provided enone 10 in 85% yield. We were poised to investigate the viability of the enone aza-[3 + 3] annulation.
Our initial efforts to employ enone substrate 10 met with failure when using up to 500 mol % of piperidinium chloride [65] (Table 1, entry 1) or 100 mol % piperidinium acetate salts (Table 1, entry 2); even at higher temperatures of 120 °C, these chloride and acetate salts proved to be ineffective (Table 1, entry 3). We then explored salts prepared from stronger acids such as (+)camphorsulfonic acid (CSA) and trifluoroacetic acid, which should tend to dissociate easier in the reaction medium and lead to a more reactive vinyl iminium salt intermediate 1b. We were The observed reactivity difference of the trifluoroacetate and camphorsulfonic acid salts, as compared to the chloride and acetate salts, can be attributed to the rate at which they promote vinyl iminium ion formation through a "balanced act" [12]. The difference in reactivity is related to the dissociation capacity of the respective amine salts. Since both "free amine" and "free acid" are needed in a synergistic manner to generate the vinyl iminium ion from the corresponding enone, the ability of the amine salt to dissociate to its "free amine" and "free acid" can exert an impact on the rate of the iminium formation. The low reactivity of the chloride salt can be attributed to its higher resistance toward dissociation, or is simply a tighter ion-pair compared to the CSA and the trifluoroacetate salt. At the same time, the formation of vinyl iminium ion from carbonyl systems and the "free amine" is also promoted by protonation of the carbonyl group via the "free acid". Consequently, the increased reactivity of the CSA and trifluoroacetate salt from the acetate salt can be attributed to a higher acidity of the acids.
After establishing the feasibility of the enone version of the intramolecular aza-[3 + 3] annulation we turned our attention to propyleine (12) (see Figure 2 and Scheme 3) as a possible target to showcase the new enone aza- [3 + 3] annulation. Propyleine (12) was isolated in 1972 from Propylaea quatuordecimpunctata in a continued effort by Tursch and co-workers [46,47] in their isolation of the azaphenalene family of defensive alkaloids from various ladybug beetles [38][39][40][41][42][43][44][45]. Mueller and Thompson [48] in 1980 found it interesting that the isomeric enamine named isopropyleine (14) was not reported in the original paper [46,47] taking into account that the isolation conditions involved acid-base extractions. It is conceivable that propyleine (12) and isopropyleine (14)  To investigate this matter, Mueller and Thompson carried out the first and the only synthesis of propyleine known to date Scheme 4: Retrosynthesis of propyleine (12). [48]. They were able to take a mixture of propyleine and isopropyleine with 1:3 ratio as determined by 1 H NMR and watch two sets of proton resonances collapse into one that corresponds to the iminium salt 13 after addition of TFA. This experiment strongly suggested that propyleine (12) and isopropyleine (14) could equilibrate under acidic conditions, thereby implying that the observed ratio of 12 and 14 represents a thermodynamic one in favor of the more stable isopropyleine [48]. With these experimental findings, Mueller and Thompson concluded that the alkaloid isolated by Tursch and co-workers [46,47] was in fact a mixture of readily interconvertable 12 and 14.
We found this controversy in the isolation and total synthesis papers an interesting one that deserves further investigations. Consequently, we performed ab initio calculations on 12 and 14 to gain insight into their relative thermodynamic stability. Models of the most stable conformers and their corresponding energies are shown in Figure 3. To our surprise, propyleine (12) is 2.59 kcal mol −1 more stable than isopropyleine (14), which is the opposite of the Mueller-Thompson postulation [48]. Our calculations suggest that the ratio obtained from the Mueller-Thompson study was likely determined by the kinetics (k B versus k R ) in the deprotonation step or the tautormerization process from the iminium salt 13, and not by thermodynamics as originally proposed. Resolving this interesting literature controversy added extra incentive for us to pursue propyleine. Retrosynthetically, we envisioned propyleine (12) to come from the decarboxylation reaction of vinylogous carbamic acid 15 [10,28], which could be derived from stereoselective hydrogenation of the endocyclic olefin in tricycle 16 (Scheme 4  (Table 2). After extensive screening of various oxidation protocols, we succeeded with the Doering-Parikh conditions [71], and the cis geometry of the enone olefin was preserved under these conditions.
With enone 17-cis in hand, we studied its intramolecular aza- [3 + 3] annulation reaction utilizing piperidinium salts (Table 3). When compound 17-cis was treated with the acetate salt in EtOAc at rt, no reaction was observed after 18 h  ( Table 3, entry 1). Heating this reaction mixture at 85 °C for 12 h only led to slow isomerization of cis-enone 17 to the thermodynamically more stable trans-isomer again with no forma-tion of the desired annulation product 16 (Table 3, entry 2). The inefficiency of the piperidinium acetate salt in the intramolecular aza- [3 + 3] annulation of enones was consistent with our previous findings (see Table 1). We were also not successful in converting enone 17-cis to the desired tricycle 16 using the more reactive trifluoroacetate salt and EtOAc as the solvent. In this case also only isomerization to trans-enone was detected (Table 3, entry 3). Heating this reaction to 130 °C led to eventual decomposition of the starting material (Table 3, entry 4).
However, when enone 17-cis was heated in toluene at 100 °C, complete consumption of starting material was observed after 5 h (Table 3, entry 5). Even though the reaction was relatively messy, the formation of the desired cycloadduct 16 was confirmed by the presence of characteristic signals in the 1 H NMR spectra of the crude mixture. Unfortunately, attempts to isolate the annulation product in pure form were not successful, probably due to the high instability of the electron-rich dihydropyridine moiety in 16. Based on our previous experience with precarious annulation products, in situ hydrogenation was then carried out, and we managed to track down ~30% of the desired reduced aza-annulation product 26, although stereoselectivity for the reduction was modest. We ultimately elected not to force our way toward propyleine using the enone aza- [3 + 3] annulation, as we succeeded in total syntheses of other members of the azaphenalene alkaloid family through annulations with enals [30,31]. While it is disappointing that this particular system may have lacked sufficient stability for this to be a suitable synthetic approach, success in an enone version of intramolecular aza-[3 + 3] annulation will allow us to find future applications.

Conclusion
Herein, we have described a successful enone version of intramolecular aza-[3 + 3] annulation reaction. Use of piperidinium trifluoroacetate salt as the catalyst and toluene as the solvent appears to be critical for a successful annulation. We also demonstrated for the first time that microwave irradiation can accelerate aza-[3 + 3] annulation reactions. An attempt to expand the scope of enone aza- [3 + 3] annulation was made in the form of propyleine synthesis as a proof of concept. While the synthesis of an enone annulation precursor was successfully accomplished, the annulation itself proved to be challenging and was only modestly successful. Future investigations are underway to pursue alkaloid synthesis via enone aza- [3 + 3] annulation.

Supporting Information
Supporting Information File 1 Experimental section.

Introduction
Lyconadin A (1, Figure 1) was isolated from the club moss Lycopodium complanatum in 2001 by Kobayashi and co-workers [1]. Subsequent to this discovery, lyconadins B-F were isolated and characterized [2][3][4]. Biological assays revealed that 1 exhibits cytotoxicity against murine lymphoma L1210 and human epidermoid carcinoma KB cells (IC 50 = 0.46 μg/mL and 1.7 μg/mL, respectively) [1]. Moreover, 1 has been shown to promote nerve growth factor biosynthesis in 1321N1 human astrocytoma cells [2]. In addition to its interesting bioactivity, lyconadin A presents a significant synthetic challenge by virtue of its unique pentacyclic skeleton, which contains six stereocenters and a pyridone ring. It is therefore not surprising that 1 has attracted the attention of the organic synthesis community. The first total synthesis of lyconadin A was reported in 2007 by Smith and Beshore [5,6], and efforts from the Sarpong [7,8] and Fukuyama [9,10] groups have also culminated in the construction of 1.
Our initial interest in lyconadin A was sparked by recognition that a 7-exo-6-exo cyclization cascade would efficiently furnish its bicyclo [5.4.0]undecane system, which is shown in bold in Figure 1. Subsequent to this observation, we performed model studies that demonstrated the viability of highly stereoselective 7-exo-trig acyl radical-6-exo-trig alkyl radical cyclizations as a means of preparing bicyclo [5.4.0]undecanes fused to aromatic rings [11]. Then, we devised an annulation protocol inspired by the work of Donohoe and co-workers [12,13] that provided access to substituted pyridones of the type found in 1 from thioester precursors [14]. Based on these encouraging results, we decided to target lyconadin A for synthesis. Herein, we provide an account of our studies directed toward the construction of this alkaloid. Specifically, we describe our efforts to prepare advanced intermediates that could be employed in the aforementioned pyridone annulation and tandem radical cyclization processes. In the course of this work, we discovered an unusual Payne-like rearrangement process that occurred in preference to the ring-opening of a hindered epoxide.

Results and Discussion
Our retrosynthetic analysis of lyconadin A is shown in Scheme 1. We reasoned that 1 could be formed by an alkylation cascade triggered by exposure of trimesylate 2 or a related electrophile to ammonia. A sequential alkylation process would serve as a viable alternative in the event of problems with this approach. In turn, cis-fused trimesylate 2 could be derived from trans-fused tricyclic ketone 3 by epimerization and standard functional-group manipulations. Based on the aforementioned model study [11], 7-exo-6-exo tandem radical cyclization of phenyl selenoester 4 was expected to produce ketone 3. Disassembly of the pyridone moiety of 4 according to our annulation protocol [14] revealed thioester 5 as a suitable precursor. We believed that this compound could be prepared from alkene 6 in two consecutive epoxidation-ring-opening sequences involving vinyl nucleophiles. We anticipated that a chiral catalyst such as one of the ketones developed by Shi and co-workers [15][16][17][18] would control the stereochemistry of the epoxidation of 6.
Presumably, the identity of the protecting groups on this substrate (i.e., R 2 and R 3 ) would be critical to the success of the reaction. After formation of the epoxide, the bulky trityl ether was envisioned to direct the subsequent ring-opening to the distal carbon [19][20][21][22]. Alkene 6 would ultimately be formed by a Myers alkylation [23] of (+)-pseudoephedrine derived amide 7 with allylic iodide 8.
The initial epoxidation substrate of type 6 that we targeted possessed benzyl and TBDPS ethers as the protecting groups. First, allylic iodide 8 was synthesized by iodination of the mesylate derived from known alcohol 9 [24] (Scheme 2). Then, coupling of methyl γ-hydroxybutyrate (10) [25] with lithiated (+)-pseudoephedrine afforded amide 11 in excellent yield. Selective silylation of the primary alcohol of 11 delivered substrate 12. Alkylation of the enolate derived from 12 with 8 according to the Myers protocol [23] furnished adduct 13 in very high yield. Although not measured directly, the dr of this compound was assumed to be very high (i.e., ≥95:5) based on the results of an alkylation conducted on a very similar substrate (see below). The configuration of the newly formed stereocenter of 13 was assigned based on the established stereochemical course of the Myers alkylation [23]. Finally, reductive removal of the chiral auxiliary with lithium amidotrihydroborate [26] produced alcohol 14, and benzylation yielded triether 15.
Asymmetric epoxidation of alkene 15 was somewhat sluggish and required superstoichiometric amounts of Shi's fructosederived ketone 16 [27]. The resulting epoxide 17 was produced in moderate yield but excellent (<95:5) diastereomeric ratio (Scheme 3). The epoxide stereochemistry was assigned based on the reported outcomes of epoxidations mediated by 16 [27]. Epoxide 17 was then subjected to ring-opening reactions with vinyl Grignard reagents in the presence of various copper salts. Surprisingly, only trace amounts of the desired product were detected, with recovered starting material and multiple byproducts typically comprising the majority of the mass balance.
Although not investigated in detail, analysis of these reactions by 1 H NMR and mass spectrometry indicated that partial debenzylation was occurring. Accordingly, we decided to replace the benzyl ether with a 2-naphthylmethyl (NAP) ether [28]. experimentation, we discovered that CuBr•Me 2 S [29] in conjunction with vinylmagnesium bromide was uniquely effective at facilitating the ring-opening of 25. However, careful inspection of the 1 H NMR spectrum revealed the presence of one less hydrogen atom than expected in the 3-4 ppm region and one more hydrogen atom than expected in the 1-2 ppm region. Clearly, neither the anticipated product 27 nor the regioisomer derived from attack at the more hindered epoxide carbon had been generated. Instead, the NMR data were consistent with the formation of a different regioisomer, tentatively identified as alcohol 26, which had been produced in good yield.
Presumably, the extremely hindered nature of internal epoxide 25 precluded its direct ring-opening, allowing alcohol 26 to form by means of a Payne rearrangement [30]. A possible mechanistic pathway for this transformation is given in Scheme 6. Coordination of a Lewis acid (likely a copper or magnesium species) to the trityl ether moiety of 25 could promote migration of the trityl group [31,32] to the epoxide, generating intermediate A. Payne rearrangement of A would then furnish epoxide B. Finally, attack of the vinylcopper complex [29] at the less-hindered carbon of the epoxide would provide 26. Acid-and Lewis acid promoted Payne rearrangements of epoxy alcohols [33,34] and epoxy methyl ethers [35] have been described, but we are unaware of any prior reports of Payne rearrangements of the bulkier epoxy trityl ethers. However, previous observations of trityl migration [31,32], although rare, do lend support to our mechanistic proposal. The NMR data for 26, while strongly supportive of the carbon backbone as drawn, do not permit an unambiguous assignment of the trityl ether to the C4 or C5 oxygen atom. An alternative Scheme 7: Synthesis of epoxide 29 from alcohol 26 (asterisks indicate relative but not absolute stereochemistry).
pathway to this carbon skeleton involving a Payne rearrangement without trityl migration can also be envisioned, and under this scenario, the trityl ether would be located at C4 rather than C5. This possibility cannot be ruled out, but it would require opening of an activated epoxonium species at the less-substituted carbon instead of the more-substituted carbon as is typically observed. Thus, we favor the mechanism shown in Scheme 6.
To provide additional evidence for the structure of 26, this compound was converted into epoxide 29 as outlined in Scheme 7. Selective detritylation was accomplished by exposure to BCl 3 at low temperature [36]. Camphorsulfonic acid was also effective for this transformation, although lengthy reaction times were required. Treatment of the resulting diol 28 with 2,4,6-triisopropylbenzenesulfonyl imidazole (TrisIm) effected regioselective sulfonylation (presumably of the less-hindered homoallylic alcohol, although this cannot be known for sure) followed by cyclization [37], delivering a single trans-disubstituted epoxide 29 of uncertain absolute stereochemistry in good yield. Examination of the 1 H NMR spectrum of 29 clearly demonstrated that a disubstituted epoxide had been generated. Alcohol 27, or the aforementioned regioisomer that would have resulted from ringopening of epoxide 25 at the more hindered carbon, would have afforded terminal epoxide 30 or oxetane 31, respectively, when subjected to this two-step sequence. While these observations do not shed light on the location of the trityl ether in 26, they do provide compelling evidence that the carbon backbone of this compound is correct as drawn and is produced by a Payne rearrangement of some type.

Conclusion
In the context of synthetic efforts targeting the polycyclic alkaloid lyconadin A, we prepared scalemic epoxide 25. A Myers alkylation and a reagent-controlled Shi epoxidation were used to construct this compound in a highly stereoselective fashion.
The bulky trityl group of 25 was intended to serve as a means of directing a ring-opening reaction to the distal carbon of the epoxide [19][20][21][22]. However, an unanticipated Lewis acid promoted Payne rearrangement intervened, producing alcohol 26 instead of the expected regioisomer 27. We believe that the extremely hindered nature of epoxide 25 prevented the desired ring-opening process, thereby enabling the unusual rearrangement to proceed. Conceivably, future studies of the scope and limitations of Lewis acid promoted Payne rearrangement-ringopening cascades could establish their utility in organic synthesis.

Supporting Information
Supporting Information File 1 Name: Experimental procedures and characterization data for all new compounds.

Introduction
Quinazoline alkaloids are a class of naturally occurring compounds with a range of medicinal properties and have been indicated for use as bronchodilators, vasodilators, anti-inflammatory agents and acetylcholinesterase inhibitors [1][2][3][4][5]. Many of the plants these products have been isolated from, such as Adhatoda vasica, Peganum harmala and Evodia rutaecarpa, have been used in folk medicine for centuries [6][7][8][9]. Since the original isolation of vasicine (1, Figure 1) in 1888 [10], the biological properties of this class of alkaloids have been extensively studied.
A number of synthetic strategies have been employed to gain access to quinazoline alkaloids [5,[11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]. Perhaps the most common method involves the condensation of an orthoaminobenzoic ester with a lactam promoted by phosphoryl chloride, known as the Niementowski reaction [3,[27][28][29][30] ( Figure 2). The availability, or lack thereof, of the corresponding lactam can determine the length and efficiency of the route. Access to the sometimes more biologically active dihydroquinazolines, such as deoxyvasicine (2), from quinazolinones requires a subsequent reduction of the amide. In 2008, our group reported the syntheses of deoxyvasicinone (4) and rutaecarpine (6) by the potassium permanganate promoted oxidation of aminals, which in turn were obtained from the condensation of ortho-aminobenzaldehydes and simple secondary amines [31,32]. A number of these aminal precursors were  prepared in generally good to excellent yields with the scope encompassing various cyclic amines and substituents on the aminobenzaldehyde aryl ring. Since then, we have demonstrated that the reaction can be run on a multigram scale [33] and have shown that dihydroquinazolines vasicine (1) and deoxyvasicine (2) can be synthesized from their corresponding aminals by using an iodine-promoted oxidation [34]. While resulting in good yields, these oxidations have the drawback of requiring large amounts of a strong oxidant for the permanganate oxidation and the necessity of stoichiometric n-butyllithium for the iodine reaction.
The conversion of the aminals formed from the condensation of aminobenzaldehydes and secondary amines to the corresponding dihydroquinazoline and quinazolinone structures under mild and catalytic conditions would be preferable to using harsh oxidants and strong bases. Han et al. have recently shown the ability of copper salts, in conjunction with oxygen, to catalyze oxidations of 2-substituted tetrahydroquinazoline aminals to quinazolines [35] (Figure 2). In addition, Reddy and co-workers have developed a catalytic system in which 2,3-substituted tetrahydroquinazoline aminals are converted to quinazolinones using tert-butylhydroperoxide (TBHP) and catalytic potassium iodide [36,37]. While these examples deal with the oxidation of bicyclic aminals, we were interested in developing methods to create dihydroquinazoline and quinazolinone alkaloids from ring-fused aminals. Here we present catalytic methods for the synthesis of both these compound classes from aminals using Cu(OAc) 2 /O 2 /AcOH and KI/TBHP systems, respectively.

Results and Discussion
Copper-catalyzed oxidations of aminals to dihydroquinazolines Copper-catalyzed oxidation reactions have received a great deal of interest in recent years [38][39][40][41][42][43][44]. Han's copper-catalyzed method for the synthesis of aminals to quinazolines results in high yields [35], but the process is not applicable to monooxidation as dihydroquinazolines are not isolated as products in these reactions. We set out to develop a method for the syn- thesis of dihydroquinazolines that would prevent further oxidation at the benzylic position. A factor complicating this effort was that dihydroquinazolines like deoxyvasicine (2) are known to auto-oxidize to their quinazolinone counterparts by exposure to air [3,[45][46][47]. We initiated our efforts by exposing aminal 7 to stoichiometric amounts of CuCl 2 in acetonitrile under a nitrogen atmosphere, which led to the formation of 2 in 81% yield (Table 1, entry 1). To improve the efficiency of the process, catalytic conditions were subsequently evaluated. When aminal 7 was heated under reflux in an oxygen atmosphere and in the presence of 20 mol % of CuCl 2 , 2 was only observed in trace amounts; deoxyvasicinone (4) and peroxide 8 were also formed as products. Switching the catalyst to Cu(OAc) 2 led to a 15% yield of the desired product 2, but the process was still unselective.
It appears that the first oxidation occurs exclusively at the aminal site to form deoxyvasicine (2). The presence of the amidine moiety apparently activates the molecule for oxidation at the benzylic position; we have observed that samples of aminal 7 can remain stable in the freezer for years, whereas 2 begins to convert to 4 within a day when exposed to atmospheric oxygen. Considering this, we reasoned that addition of a weak acid to protonate the relatively basic amidine moiety of 2 might deactivate the benzylic position toward oxidation while not interfering with the initial aminal oxidation. Indeed, using 1.1 equivalents of acetic acid as an additive with catalytic Cu(OAc) 2 in acetonitrile led to the formation of 2 in 53% yield without formation of 4 and 8 ( Table 1, entry 4). A simple change of the solvent from acetonitrile to methanol drastically improved the yield of 2 to 81% ( Table 1, entry 5). A number of different copper salts, solvents and acids were then evaluated, but none of the changes led to a further improvement in yield. It appears that under certain conditions catalyst deactivation via copper oxide formation decreased the catalyst turnover and consequently product yields.  Using the optimized reaction conditions, a range of different aminals were selectively oxidized to the corresponding dihydroquinazolines (Table 2). In general, these products were obtained in moderate to good yields. Product 10, containing a piperidine ring, required a higher reaction temperature and resulted in a lower yield than the corresponding pyrrolidine and azepane products (2 and 12, respectively). While differences in conformation may in part account for the observed differences in reactivity (X-ray crystal structures of aminals containing pyrrol-idine and piperidine revealed that the pyrrolidine-containing aminal adopts a bent structure, whereas the piperidine aminal appears relatively strain-free [34]), this finding likely relates to the reduced propensity of six-membered rings to engage in reactions that form exocyclic double bonds. The isolation of azepinoquinazoline 12 in 73% yield was gratifying but somewhat unexpected since Decker reported that samples of the compound completely oxidized to quinazolinone 23 when exposed to air for 24 h [3]. This demonstrates the need for acetic acid to protonate the amidine, preventing further oxidation. While product 16 was obtained in good yields from tetrahydroisoquinoline-aminal 15, rutaecarpane-derived product 18 was formed in only 47% yield, apparently due to unidentified side-reactions. The reaction leading to the synthesis of the dibromo-analogue of deoxyvasicine 20, even under elevated temperature and extended reaction time, still did not reach completion after 3 days. The attenuated reactivity of aminal 19 is most likely the result of the decreased electron density on the anilinic nitrogen.

KI-catalyzed oxidations of aminals to quinazolinones
Different conditions for the direct catalytic oxidation of aminals to quinazolinones were also explored. The use of Cu(OAc) 2 and methanol, while appropriate for furnishing deoxyvasicine (2) from aminal 7, did not result in satisfactory yields of deoxyvasicinone (4, Table 1). Attempts to use other copper(I) or copper(II) salts and solvents under oxygen without the addition of acid to promote the full oxidation of aminal 21 to deoxyvasicinone (4) were met with disappointment, with yields of 4 for these conditions reaching a maximum of around 40% (Table 3). In most cases, peroxide 8 was observed as a major side product.
The Cu/TEMPO/DABCO catalyst system employed by Han et al. [35] for the oxidation of aminals to quinazolines provided an increased yield of 50% (Table 3, entry 9). The best yields were obtained by using the conditions developed by Reddy and co-workers [36], namely the combined use of catalytic amounts of potassium iodide (20 mol %) and excess TBHP (5 equiv), followed by the addition of piperidine. In this instance, deoxyvasicinone was isolated in 80% yield ( Using the optimized conditions, a range of different quinazolinones were synthesized (Table 4). In general, yields were moderate to good for substrates with varying ring sizes. In this manner the natural products deoxyvasicine (4), mackinazolinone (5) and rutaecarpine (6) were prepared, in addition to the azepinoquinazolone 23, which has been demonstrated to be a more effective antitussive agent than codeine [48]. Dibromo-  deoxyvasicinone analogue 25 was obtained in relatively high yield (88%) whereas the corresponding analogue of mackinazolinone (27) was obtained in only 50% yield.
Interestingly, when quaternary aminal 28 was subjected to oxidative conditions in an attempt to prepare compound 29, deoxyvasicinone (4) was obtained as the major product in a process that involved demethylation (Scheme 1, reaction 1). The demethylation of aminals has been previously reported in cases where the product achieves aromaticity [49][50][51], which is presumably the driving force for this transformation. Aminal 30, which contains two tertiary amines and is readily obtainable by an acid-promoted hydride shift process [52][53][54], was also exposed to oxidative conditions (Scheme 1, reaction 2). We had hypothesized that quinazolinone 32 might be formed in this reaction by the debenzylation of an intermediate iminium ion. However, the major product from this reaction was identified to be 31, the apparent product of iminium hydrolysis.

Conclusion
We have demonstrated that quinazoline alkaloids and their analogues can be synthesized from aminals by using Cu(OAc) 2 / O 2 /AcOH and KI/TBHP catalyst systems. The use of acetic acid in addition to oxygen and catalytic copper(II) salts was determined to prevent overoxidation of dihydroquinazolines, allowing access to these structures under mild conditions. A number of natural products and their analogues were obtainable by these methods, which should facilitate the preparation of novel materials for biological studies.

Supporting Information
Supporting Information File 1 Experimental details, characterization data and 1 H and 13 C NMR spectra for all new compounds.

Introduction
Due to their presence in some natural products [1] and pharmaceuticals [2][3][4], the preparation of N-arylpyrroles is an active field of investigation [5]. Depending on their substituents, N-arylpyrroles could also be electron donor/acceptor molecules with a dual fluorescence ability suggesting attractive optoelectronic applications [6,7]. If the N-arylation of pyrroles is possible by Ullmann-type condensation [8][9][10], the regioselective functionalization of pyrroles is less trivial when asymmetric substrates are targeted. An indirect solution, based on the construction of substituted pyrrolidines that oxidize into elaborated pyrroles, can be employed fruitfully [11,12]. We recently described a one-pot organo-catalyzed synthesis of N-heteroarylmethylene pyrrolidines 4 [13] from readily available aldehydes 1 and imine 2 by a sequence of Mannich coupling [14][15][16][17][18][19][20][21][22][23][24], Wittig olefination with phosphonium 3, and proton-mediated hydroamination (Scheme 1). In the course of our investigations, we observed that pyrrolidine 4 could be converted into the corresponding pyrrole 5 by a simple isomerization, avoiding the use of oxidants. We describe herein the details of these observations and the scope of this methodology for the concise preparation of substituted 2-heteroaromatic decorated N-arylpyrroles.

Results and Discussion
We first observed the unexpected formation of pyrrole 5a in 50% yield after treatment of pyrrolidine 4a with KCN in DMF (Scheme 2, conditions a). Although obtained in modest yield, we found the original and unique structure of the substituted pyrrole 5a interesting, especially with the 2-pyridylmethylene decoration. In an attempt to rationalize the formation of 5a, we hypothesized that KCN acted as a nucleophilic and weak base since the level of oxidation of 4a and 5a was the same. To improve the efficiency of the transformation, a stronger nucleophilic base such as DBU (1,8-diazabicyclo [5.4.0]undec-7-ene) was tested [25]. Pleasingly, when pyrrolidine 4a was exposed to DBU in CH 2 Cl 2 , 5a was obtained in excellent yield (98%, 1 h, conditions b; Scheme 2). The reaction can also be promoted by a catalytic amount of DBU (0.2 equiv) delivering 5a (96%) after prolonged reaction time (22 h, conditions c, Scheme 2). Interestingly and despite its strong nucleophilic character, DABCO (1,4-diazabicyclo[2.2.2]octane) was unable to promote the isomerization (conditions d, Scheme 2) and the starting material was recovered.
As presented in Scheme 3, the methodology was next attempted in a one-pot process. Hence, the transformation of aldehyde 1a and imine 2a into pyrrolidine 4a was followed by the introduction of DBU leading to pyrrole 5a in 26% yield. However, proceeding stepwise and isolating the pyrrolidine 4a by a simple filtration on silica gel before isomerization is more rewarding: following this route, the global yield for the whole process reaches 59% yield. Applying this procedure, various 2-heteroarylmethylenepyrrolidines 4b-h prepared from aldehydes 1b-h and imine 2a were exposed to DBU (1.1 equiv). Pleasingly, pyrrolidines 4b-h were transformed into the corresponding pyrroles 5b-i with homogeneous efficiency. Hence, the chemistry proved to be compatible with substrates containing meta-, para-pyridyl and quinolinyl substituents, allowing the preparation of 5b (81%), 5c (78%) and 5d (98%). Pyrrolidine 4e containing an electronically deficient pyridyl residue was also converted into 5e (80%) while pyrrolidine 4f bearing a pyrazine core underwent aromatization with high efficiency to give 5f (97%). The C 2 -symmetric scaffold 4g was efficiently converted into 5g (86%) and similar treatment of pyrimidine 4h provided pyrrole 5h in high yield (91%).
The Mannich coupling was next attempted with different imines 2b-e in order to modulate the nature of the aryl moiety (Scheme 4). The electronic nature of the aniline being crucial Scheme 3: Preparation of N-arylpyrroles 5a-h (unless otherwise specified, yields in brackets refer to the isomerization step while yields in square brackets refer to the two-step procedure from the corresponding aldehyde).

Scheme 5: Possible mechanism.
for the stability of the imine and the hydroamination step, electronically rich anilines were selected to form imines 2b,c. Hence, when imines 2b,c were exposed to aldehyde 1a in the presence of catalyst 6 (available in racemic form), the Mannich adducts 7i,j were obtained and directly reacted with phosphonium salt 3. In line with our procedure, the resulting acyclic anilines 8i,j were then exposed to TFA to promote the cyclization into pyrrolidines 4i,j which upon treatment with DBU were converted into pyrroles 5i,j in 41% and 52% overall yields. While p-alkoxy substituted (R = OAllyl, OBn) anilines are compatible, the methodology proved troublesome with o-alkoxy substituted anilines, the main limitation being the formation of the corresponding imines. Similarly, imine 2d prepared from para-bromoaniline was found to be unstable and only degradation was observed during the Mannich reaction. When imine 2e, derived from the para-iodoaniline, was engaged in the process, the hydroamination step turned out to be problematic, which prevented the isolation of the corresponding pyrrolidine.
Even if not completely elucidated, a mechanism of the isomerization can be suggested in which the acrylate moiety is crucial. Indeed, without this unsaturation, it was not possible to observe the isomerization of the exo-enamine into the endo compound under basic treatment [26]. These observations suggest that DBU or KCN behave as base to promote the deconjugation of the acrylate moiety of 4a [27]. The resulting product 4a' would lead under basic treatment to pyrroline 4a" from which aromatization to 5a would be expected to follow (Scheme 5).
Having established a practical methodology for the preparation of substituted N-arylpyrroles, we next undertook synthetic transformations to extend the molecular diversity of the substrates. While attempts to perform an oxidation of the bis(heteroaryl)methylene position with elemental sulfur [28] or SeO 2 failed, the oxidation of this methylene position was regioselectively carried out by treatment of 5a-e with (NH 4 ) 2 Ce(NO 3 ) 6 (CAN), delivering the alcohols 9a-e (Scheme 6). The methylene oxidation was especially efficient with substrates containing mononitrogenated heteroaryl substituents, with yields ranging from 64-87%. Oxidation under the same conditions was found to be more troublesome with pyrazine 5f since alcohol 9f was isolated in only 20% yield. Similar treatment of pyrazine 5g and pyrimidine 5h gave a complex mixture of products. While the oxidation of the bis(aryl)methylene position with CAN has been reported [29], this is the first example of bis(heteroaryl)methylene oxidation employing this reagent [30]. In order to increase the local electron deficiency of the scaffold, 9a was oxidized with 2-iodoxybenzoic acid (IBX) into ketone 10a (98%), which presents an ideal push-pull configuration tunable with the pH by protonation of the pyridine ring. This is likely to lead to applications of 10a such as for new water-soluble molecular probes.

Conclusion
A new catalytic and regioselective preparation of substituted N-arylpyrroles decorated with various 2-heteroaromatic scaffolds is reported. Based on the isomerization of pyrrolidines prepared by a simple and efficient sequence of Mannich/Wittig olefination/hydroamination reactions, no oxidant or metallic salts were employed [31]. This study also led us to investigate the feasibility of this process with different anilines and enlarge the molecular diversity of the scaffold. So far the methodology is limited to electron-rich anilines due to the formation and reactivity of the corresponding imines and the stability of the Mannich adduct for the hydroamination step. However, this electronic configuration is ideal for the preparation of electron donor/acceptor N-arylpyrroles as demonstrated in this study. In addition, we documented an efficient C-H oxidation of the bis(heteroaryl)methylene position promoted by CAN. Technical grade N,N-dimethylformamide and dichloromethane were used for this work. Following our procedure [13], catalyst 6 was prepared from (±)-1-benzyl-3-aminopyrrolidine [18471- . N-Arylimino ethyl glyoxylates 2a-c were prepared by a condensation of ethyl glyoxylate and arylamines in toluene (c = 1 M) with MgSO 4 at room temperature. IBX (2-iodoxybenzoic acid) was prepared according to standard procedures.

Introduction
The ripostatins (A, B, and C) are a family of antibiotic natural products, isolated in 1995 by Höfle and colleagues from cultures of the myxobacterium Sorangium cellulosum (Figure 1) [1,2]. Ripostatins A and B are active against Gram-positive bacteria due to their inhibition of bacterial ribonucleic acid polymerase. These compounds inhibit chain initiation of RNA synthesis in Staphylococcus aureus, a particularly infectious bacterial strain with reported drug resistance to the antibiotics vancomycin and methicillin [3]. Ripostatin A exists as an equilibrium mixture of ketone and hemiketal forms, the ratio of which is reported to be 55:45 in methanolic solution. In the hemiketal form, the bicyclic framework of ripostatin A features an unusual in/out connectivity. The 14-membered macrocyclic core of ripostatin A contains three double bonds, arranged in a rare 1,4,7-skipped triene. The alkene geometry was determined to be (2E,5E,8E) by measurement of NOE enhancements [4].  Ripostatin B and its C15 epimer can be obtained from ripostatin A by reduction with sodium borohydride, while ripostatin C can be formed from ripostatin A by a mild base-mediated elimination. Consequently, ripostatin A was selected as the primary target for synthesis.
At the outset of our efforts, the only published synthetic study on the ripostatins was Kirschning's synthesis of C1−C5 and C6−C24 fragments of ripostatin B [5]. However, these fragments could not be connected by esterification due to steric hindrance from bulky protecting groups, as well as susceptibility of the skipped dienes therein to double bond isomerization and migration under basic conditions. However, the ripostatins intrigued others, and three total syntheses of ripostatin B were published in succession in 2012 [6][7][8]. Tang and Prusov extended their synthetic method to syntheses of 15-deoxyripostatin A, and later ripostatin A itself [9]. All of these approaches to the ripostatins share several key features: use of ring-closing metathesis to form the 14-membered macrocycle, preceded by one or more Stille couplings to generate the double 1,4-diene ( Figure 2).
Notwithstanding the successful syntheses of ripostatins, preparation of configurationally defined skipped polyenes (1,4-dienes and higher homologues) remains a significant challenge in organic chemistry. The doubly allylic protons found in these structures may be sensitive to strong base as well as hydrogen abstraction [10,11]. While classical methods for the preparation of 1,4-dienes include partial reduction of alkynes and carbonyl olefination, a variety of transition-metal-mediated processes have been developed for the synthesis of skipped dienes of varied substitution patterns [12][13][14][15]. Most recently, Sigman and colleagues have reported a palladium-catalyzed 1,4-difunctionalization of 1,4-butadiene with vinylboronic acid and vinyl triflate that can be used to rapidly access the skipped triene of ripostatin A [16].  We recognized that the C11 stereocenter and the C8-C9 trisubstituted olefin of ripostatin A mapped onto a nickel-catalyzed coupling reaction of alkynes and epoxides developed in our laboratory (Scheme 1) [17]. In the intermolecular reaction, stereospecific cis addition across the alkyne is observed, and the stereochemistry at the epoxide is preserved in the transformation. Aliphatic epoxides are opened selectively (>95:5) at the terminal position. Although very high regioselectivity with respect to the alkyne is observed when R 1 = Ph and R 2 = Me, attempts to differentiate between aliphatic alkyne substituents lead to mixture of regioisomers. When a 1,3-enyne is coupled with simple epoxides, however, >95:5 regioselectivity is observed for C-C bond formation distal to the pendant alkene [18].
In conjunction with the nickel-catalyzed fragment coupling, we wished to investigate whether it would be possible to delay introduction of the potentially sensitive 1,4,7-triene by masking it as a cyclopropyldiene, then unveiling the skipped triene portion via a 1,5-hydrogen rearrangement (Figure 3). This strategy would allow us to take advantage of the high regioselectivity in enyne-epoxide reductive coupling reactions. Furthermore, the proposed rearrangement would serve to differentiate the ester groups, as hydrogen would migrate from adjacent to the ester cis to the dienyl chain only.
While offering a unique approach to the skipped triene portion of ripostatin, this route was not without significant uncertainty. First, substrates containing a vinyl cyclopropane unit had not been previously tested under nickel-catalyzed reductive coupling conditions. Nickel(0) is known to catalyze the rearrangement of vinylcyclopropanes to cyclopentenes; however, activating substituents are commonly required [19][20][21]. Furthermore, application of a proposed 1,5-hydrogen rearrangement to ripostatin A would require that the reaction proceed to give the triene with E,E,E configuration selectively out of four possible configurational outcomes (Scheme 2).

Scheme 2:
Potential transition states and stereochemical outcomes for a concerted 1,5-hydrogen rearrangement.
Rearrangement of vinylcyclopropanes has been used to prepare 1,4-skipped dienes of varying geometry (Scheme 3). It has long been known that cis-disubstituted vinylcyclopropanes can undergo 1,5-hydrogen migration under thermal conditions to deliver acyclic 1,4-dienes [22]. In this reaction manifold, the new "acceptor-derived" double bond is formed via an endo transition state, leading to a cis olefinic configuration in the product (Scheme 3, reaction 1). Berson has quantified the energetic preference in this transformation, while Turos has shown that the presence of a silicon substituent on the "donor carbon" facilitates the hydrogen migration [23][24][25][26].
In contrast, Wilson has reported formation of the trans double bond in the opening of a cyclopropane with an adjacent hydroxy or mesylate leaving group, regardless of the initial configuration of the cyclopropane (Scheme 3, reaction 2) [27]. Braddock has demonstrated that the internal 3,4-E olefin is obtained exclusively in Prins reactions terminated by cyclopropylmethylsilane (Scheme 3, reaction 3), which may be explained by the participation of a carbocation that is stabilized by the adjacent cyclopropane ring in the bisected conformation where (CHOR)R′ is oriented anti to the cyclopropane [28][29][30]. Finally, Micalizio has described a titanium-mediated, alkoxidedirected fragment coupling reaction between vinylcylopropanes and vinyldimethylchlorosilane (Scheme 3, reaction 4) in which the stereochemical outcome of the rearrangement is orches-trated by the adjacent alkoxide, which is believed to direct formation of a tricyclic titanacyclopentane that subsequently fragments in a stereospecific manner [31].
We were intrigued by the apparent difference in selectivity observed in cyclopropane rearrangements proceeding via a neutral pathway versus those proceeding by more polar or directed mechanisms. Although the E geometry of the central olefin in the ripostatin A triene is more consistent with a polar mode of reactivity than a neutral 1,5-hydrogen migration, we still wished to investigate the outcome of rearrangement under thermal conditions. To the best of our knowledge, such a rearrangement has not been explored for structures containing an additional alkenyl substituent in conjugation, or with electron-withdrawing groups adjacent to the site of hydrogen migration. In particular, the latter's ability to facilitate the buildup of negative charge at an adjacent carbon might play an important role in the stereoelectronic course of the reaction.

Results and Discussion
Synthesis of cyclopropylenyne and reductive coupling with model epoxide Diethyl 1,3-acetonedicarboxylate (Scheme 4, 16) was rapidly identified as an inexpensive five-carbon fragment possessing the appropriate oxygenation pattern for preparation of the C1-C9 enyne fragment. However, due to keto-enol tautomer- ization, carbonyl olefination methods are of limited utility for this substrate. Instead, the ketone was protected as the mixed S,O-ketal and reduced to the diol 17. Protection of the hydroxy groups and removal of the ketal afforded ketone 19. A number of alternative promoters were investigated to avoid the use of mercury(II) salts in the ketal deprotection (including MeI, H 2 O 2 , AgClO 4 /I 2 ); however, these generally led to concomitant removal of the TBS groups.
Ketone 19 was converted to the α,β-unsaturated ester 20 using the Peterson olefination [32]. Treatment of 20 with the sulfur ylide derived from trimethylsulfoxonium iodide [33,34] led to recovery of starting material at room temperature, but decomposition at elevated temperatures. Instead, the enone was smoothly reduced to the allylic alcohol, and a Furukawa-modified Simmons-Smith reaction [35] afforded the cyclopropyl alcohol 22 in high yield.
Oxidation to the cyclopropyl aldehyde 23 offered a branching point from which either the Eor Z-substituted enyne could be synthesized, should we wish to study the rearrangement of both diene geometries. To access our initial target, the E-enyne 6, a Takai olefination [36] was used to generate the E-vinyl iodide. The vinyl iodide was sufficiently stable for purification by silica gel chromatography but, following purification, was immediately carried forward to a Sonogashira reaction with propyne [37]. The enyne could be obtained in 15:1 E/Z selectivity, in 10 steps and 35% overall yield. It was found that the use of freshly distilled THF in the Takai olefination and careful temperature control in the subsequent cross coupling were critical to the preservation of high E/Z selectivity over the course of these transformations. Once isolated, however, enyne 6 proved to be quite stable and could be stored for extended periods at 0-5 °C without appreciable isomerization or decomposition.
With a suitable cyclopropylenyne in hand, a model epoxide substrate containing the 1,3-oxygenation pattern found in ripostatin A (Scheme 5) was prepared by using a route analo-gous to one reported by Smith [38]. Allylation of 24 with (+)-Ballyldiisopinocampheylborane generated the alcohol 25 in high yield and enantioselectivity. Directed epoxidation using VO(acac) 2 and tert-butyl hydroperoxide was initially performed in order to furnish 27 directly; however, this proceeded in only modest yield and diastereoselectivity (53%, 2.8:1 syn/anti).
Although the ratio of syn/anti epoxide diastereomers could be enhanced by subjecting the mixture to hydrolytic kinetic resolution [39], greater throughput could be obtained by converting 25 to the tert-butyl carbonate, performing an iodocyclization, then cleaving the iodocarbonate and closing the epoxide under basic conditions. Silyl protection of the secondary alcohol afforded the desired model compound 28.
Enyne 6 and epoxide 28 were subjected to standard nickelcatalyzed reductive coupling conditions, and reductive coupling proceeded in good yield, leaving the cyclopropane ring intact (Scheme 6). However, the desired diene 29 was isolated along with the regioisomeric product 30 in approximately a 3:1 ratio. The desired product could be partially separated from the regioisomer by careful silica gel chromatography.
It may be instructive at this stage to consider the proposed mechanism of the nickel-catalyzed alkyne or enyne-epoxide reductive coupling reaction (Scheme 7). In contrast to the mechanistic framework proposed [40,41] for reductive coupling reactions of alkynes and aldehydes developed in our laboratory [42,43], it is believed that epoxide oxidative addition precedes alkyne addition, as opposed to concerted oxidative coupling. At least when dimethylphenylphosphine is used as ligand, this may proceed via the intermediacy of a betaine species. In the reductive coupling reaction of enynes and epoxides, the olefin coordinates to nickel and directs alkyne insertion.
Because of this directing effect, formation of the regioisomeric diene product is atypical for reductive coupling reactions of enynes and epoxides. However, in reactions of 1-phenyl-1propyne and epoxides with oxygenation in the 3-position Scheme 6: Nickel-catalyzed enyne-epoxide reductive coupling reaction.

Scheme 7:
Proposed mechanism for the nickel-catalyzed coupling reaction of alkynes or enynes with epoxides.
(Scheme 8), it was found that while epoxides containing adjacent silyl ethers afforded mainly the expected regioisomer (7:1 37a/37b), epoxides with sulfonate esters (e.g., tosyl, 32) and esters (e.g., acetyl, 33) afforded a regioisomeric mixture of opposite (albeit poor) selectivity relative to that normally observed for unfunctionalized epoxides [44]. This is proposed to be an effect of the coordinating ability of Lewis basic oxygen atoms in tosylates and esters, which may disrupt the binding and directing effect of phenyl or alkenyl groups.
In the case of the ripostatin A model system, the most likely candidate for chelation is the oxygen protected as the PMB ether. Although an eight-membered chelate might seem too large to play an important role in directing regioselectivity, the 3:1 regioselectivity observed in the "normal" direction is consistent with chelation playing a diminished role relative to the seven-membered chelates invoked for coupling of 3-oxygenated epoxides. We attempted to discern whether this interaction was the reason for the observed regioselectivity by performing the nickel-catalyzed coupling reaction with 1,2epoxyoctane, which lacks potentially chelating functional groups (Scheme 9). Although the coupling product 39 appears to be formed in the reaction with either Bu 3 P or PhMe 2 P as ligand, the non-polar nature of this molecule complicates chromatographic purification, and mixtures of what appears to be 39 along with one or more other products were obtained. Based on these results, regioisomer formation cannot be excluded.  Several aspects of the enyne synthesis and the nickel-catalyzed coupling reaction require further investigation. As 6 does not itself appear to undergo thermal rearrangement, it seems advantageous to convert this compound to the corresponding diester. Preliminary investigation indicates that oxidation of the diol derived from 6 is complicated by the 1,5-relationship of the alcohols. Despite the greater complexity inherent to this alter-native, differentiation of the alcohols allowing for sequential oxidation may be necessary.

Scheme 11:
Deuterium labeling reveals that the allylic/benzylic site is most acidic.
could be expediently accessed from the corresponding aldehyde by making use of a Claisen rearrangement to set the geometry of the γ,δ-unsaturated double bond.
In the forward direction, the allylic alcohol 44 was obtained from reaction of the alkenyllithium reagent derived from 2-bromopropene with phenylacetaldehyde (Scheme 10). In our hands, the organolithium afforded significantly higher and more reproducible yields than either the commercially available Grignard reagent or the organocerium. Johnson-Claisen rearrangement [50] of 44 proceeded smoothly to give the γ,δunsaturated ester 45. Conducting the reaction without added solvent in a microwave reactor at 170 °C allowed the reaction to proceed in just 30 minutes; a significant improvement over heating the reaction mixture in toluene under reflux, which typically required 48 hours to obtain a comparable yield. Reduction and oxidation afforded the aldehyde 47, which could then be converted to dithiane 41.
Unfortunately, attempts to couple 41 with an oxygenated epoxide fragment under a variety of conditions reported by Smith for lithiation and electrophilic trapping [51] were unsuccessful. We suspected that the lithiated dithiane was not being generated and decided to investigate this step of the reaction independently of reaction with the epoxide electrophile. To this end, 41 was treated with tert-butyllithium at -78 °C in a 90:10 THF/HMPA mixture, referred to as the method of first choice for lithiation of complex dithianes (Scheme 11). Following warming to -42 °C, the reaction was quenched with deuterated methanol. Analysis of the product by 1 H NMR revealed that no deuterium incorporation (to the sensitivity of integration) had occurred at the desired dithiane site, while approximately 80% deuterium incorporation had occurred at the allylic/benzylic site.
These results indicated that the presence of the trisubstituted C18-C19 olefin would interfere with dithiane coupling. However, given the suitability of the Claisen rearrangement for formation of this bond, we wished to preserve that transformation. Accordingly, an alternate route that would capitalize on the electrophilic nature of aldehyde 47 to form the bond corresponding to C14-C15 of ripostatin A instead was sought.

Oxy-Michael approach to epoxide
We were intrigued by a recent report by Falck describing an organocatalytic oxy-Michael addition to achiral δ-hydroxy-α,β-Scheme 12: Oxy-Michael addition to δ-hydroxy-α,β-enones. enones (Scheme 12) [52]. The hydroxy group is delivered in a directed fashion from the boronic acid hemiester generated in situ from the substrate and phenylboronic acid. It is proposed that complexation of the tertiary nitrogen to boron and coordination of the carbonyl act in a push/pull fashion, simultaneously enhancing the nucleophilicity of the boronate oxygen as well as imposing a chiral environment around the enone. Aliphatic enones react more sluggishly in this transformation; however, 3,4,5-trimethoxyphenylboronic acid may be used as a more efficient nucleophilic partner to circumvent this limitation.
Application of this transformation to the ripostatin A epoxide fragment 5 allows installation of the C13 hydroxy group via conjugate addition to the δ-hydroxy-α,β-enone 49 ( Figure 5). Although the diastereoselectivity of this reaction using substrates with a chiral center at the hydroxy group had not been investigated in the published study, substitution at the carbinol position was reportedly well tolerated for the reaction using γ-hydroxy-α,β-enones. Although the presence of chirality at the δ-position allows for diastereoselective intramolecular oxy-Michael addition of hemiacetal-derived alkoxides into α,βunsaturated esters, the extension of this reaction to ketones was not successful [53]. In turn, we intended to prepare 49 by hydrometalation of 50 and addition into aldehyde 47.
To this end, (R)-glycidol was protected as the para-methoxybenzoate ether to give the PMB glycidol 52 (Scheme 13), in which the configuration is now assigned as (S). Opening the epoxide of 52 with the ethylenediamine complex of lithium acetylide in a 1:1 THF/DMSO solvent mixture at 0 °C allowed the terminal alkyne to be accessed in 84% yield without rearrangement to the internal alkyne. Silyl protection afforded alkyne 50, which was prone to decomposition upon extended storage, even at −20 °C.
Hydrozirconation of alkyne 50 with Schwartz's reagent [54,55] was followed by transmetallation to zinc and nonselective addition into aldehyde 47. Oxidation of the resulting allylic alcohol mixture afforded the enone 53. Prior to the key oxy-Michael addition, it was necessary to remove the tert-butyldimethylsilyl protecting group. Use of TBAF under buffered (acetic acid) or unbuffered conditions proved sluggish and somewhat lowyielding, typically ~50%. Various other promoters led to decomposition (BF 3 ·OEt 2 ) or low conversion (CsF); however, a modest improvement in yield was noted with the use of 1% HCl in MeOH (with a small amount of THF to dissolve the starting material).
The oxy-Michael addition with chiral thiourea and phenylboronic acid proceeded to give a single diastereomer; however, after 48 h at 50 °C, only 17% of the syn diol was formed, and 56% of the starting material was re-isolated. Under the modified conditions for less reactive aliphatic aldehydes using 3,4,5trimethoxyphenylboronic acid, we were unable to isolate the desired diol from the reaction mixture. Since the oxy-Michael substrate derived from 53 contains an existing stereocenter at the directing hydroxy group, we also attempted to carry out the  reaction with diisopropylamine as a substitute for the thiourea catalyst. This modification afforded both the syn and anti diols in roughly a 1:1 ratio and a combined yield of 40-45%, albeit without recovery of starting material. The syn and anti diastereomers could be separated by repeated silica gel chromatography, and the desired syn diol was converted to the bis-silylated compound 54.
However, attempted oxidative deprotection of the PMB ether of 54 with DDQ led to destruction of the material. It seems likely that this is again due to the presence of the allylic/benzylic site in the molecule, although no individual decomposition products could be identified. Although it is possible that further screening of deprotection conditions might have allowed us to move forward with this route, the modest yields and long reaction time of the oxy-Michael addition severely limited material throughput. In order to proceed with the synthesis, we needed a more robust route, with the following criteria: (1) no protecting groups requiring oxidative cleavage, and (2) introduction of the C13 hydroxy group at an early stage.

Iodocyclization approaches to epoxide
In the preparation of model epoxide 28, iodocyclization was used to introduce oxygenation in a stereoselective fashion from a chiral homoallylic alcohol. Applying this disconnection, we hypothesized that we might be able to introduce the epoxide functional group of 5 by iodocyclization of the tert-butyl carbonate 55 ( Figure 6). Although the additional double bond in this substrate presents a potential site for competing reaction pathways, we were encouraged by a report by Bartlett in which the iodocyclization of the tert-butyl carbonate derivative of 1,7octadien-4-ol afforded exclusively the 6-membered carbonate derivative, with product arising from cyclization onto the more distant double bond not detected [56]. We set out to access 55 from deprotonation of methyl ketone 57 on the less hindered side and alkylation with bromide 56.
The E-allylic bromide was prepared rapidly, albeit in modest yield, with an Appel reaction [57] of the known allylic alcohol [58]. To synthesize ketone 57, we opted to utilize an asymmetric aldol reaction to set the stereochemistry of the β-hydroxy  group. Since the report of Evans's diastereoselective asymmetric aldol reaction using the boron enolates of N-acyloxazolidinones [59], numerous chiral-auxiliary-based methods have been developed for the synthesis of synor anti-propionate aldol units. However, many of these auxiliaries, including the Evans oxazolidinones, fail to give high stereoselectivities when employed in acetate aldol reactions [60]. Of the methods available, we selected Sammakia's boron enolate-based strategy using the N-acetylthiazolidinethione 58 (Scheme 14) for its high reported diastereoselectivity with aliphatic aldehydes and its avoidance of toxic tin reagents [61].
The reaction of the brightly colored 58 with but-3-enal proceeded in moderate yield, with an initial diastereoselectivity around 10:1, with further enhancement following silyl protection and purification. Although the silylated compound 60 proved to be reluctant to form the Weinreb amide, microwave irradiation allowed this process to proceed on a reasonable time scale. Grignard addition to the Weinreb amide afforded ketone 57.
Unfortunately, attempts to unite ketone 57 and bromide 56 via alkylation were unsuccessful. Although deprotection at the lesssubstituted site of the methyl ketone using LDA was verified in a deuterium quench experiment, the alkylation did not proceed at temperatures from −78 °C to 0 °C. While the ketone was re-isolated cleanly following the reaction, the bromide was converted to a mixture of olefinic compounds. Faced with the difficulty of forming the C16−C17 bond by alkylation, we considered potential routes arising from retrosynthetic disconnection of the C15−C16 bond (Figure 7). It was recognized that reaction of the enolate of ester 45, a compound previously synthesized in just two steps, and subsequent oxidation could give the β-ketoester 61. Decarboxylation of this compound would provide rapid access to the key iodocyclization substrate 55.
Aldehyde 62 was prepared by reduction of the thiazolidinethione 60 with DIBAL-H (Scheme 15). Treatment of the ester with LDA, followed by trapping with the aldehyde, afforded the aldol adduct as a mixture of up to four possible diastereomers. This was then oxidized under Ley's conditions [62] to the β-ketoester 61, itself a mixture of two diastereomers.
Initially, we attempted to induce decarboxylation of 61 by treatment with LiOH in a 1:1 water/THF mixture. No reaction was observed at room temperature, but heating to 70 °C resulted in elimination of the β-siloxy group. The Krapcho reaction offers an essentially neutral method for the decarboxylation of baseand acid-sensitive substrates [63]. Under these conditions (sodium chloride in wet DMSO at elevated temperatures) the desired decarboxylation reaction proceeded, although only in modest yield (Scheme 16). Reduction of the temperature from 180 °C to 120 °C led to much lower conversion, as expected, but did not improve mass recovery in the reaction. Somewhat more surprisingly, silyl deprotection with TBAF was also low yielding. Interestingly, when the order of these operations was reversed, the Krapcho conditions led to a complex product mix- ture, the major component of which appeared to be the dienone, formed by elimination and isomerization of the terminal olefin into conjugation.
We reasoned that the decarboxylation and silyl deprotection steps could be coupled into one operation by switching from the methyl ester to the 2-trimethylsilylethyl (TMSE) ester.
Although there was concern that if the TBS group were removed first, the free hydroxy group would undergo elimination, it seemed likely that both deprotections would proceed at ambient temperature, which might circumvent this issue. Transesterification of methyl ester 45 to TMSE ester 64 proceeded in good yield, and following an analogous procedure for aldol reaction and oxidation the TMSE β-ketoester was obtained (Scheme 17). Treatment with an excess of TBAF in THF at room temperature overnight resulted in formation of the β-hydroxyketone 63. Although the yield for this transformation remained moderate, it was higher than that obtained for either of the individual steps from the methyl series that it replaced.
The cleavage of TMSE β-ketoesters with TBAF·3H 2 O has been described in the literature as a chemoselective method for decarboxylation in the presence of other types of β-ketoesters [64].
Comparable yields for the decarboxylation to form 63 were obtained with this reagent as with the anhydrous solution, or when the reaction was run in DMF instead of THF. The use of TAS-F (tris(dimethylamino)sulfonium trifluoromethylsilicate) was clearly inferior, leading to incomplete conversion and elimination. With TBAF, partial elimination could sometimes be observed; however, this typically occurred in less than 10%. Given these results, it seems that the fluoride-mediated deprotection of TMSE β-ketoesters is deserving of further exploration and utilization as a method for the decarboxylation of sensitive synthetic intermediates.
Alcohol 63 was derivatized as the Boc carbonate, a reaction plagued by the formation of the carbonate arising from two molecules of 63. The ratio of Boc derivative to symmetrical carbonates is dependent on the acidity of the alcohol and not necessarily improved by increasing the stoichiometry of Boc 2 O [65]. Disappointingly, treatment of 55 with IBr or I 2 led to exhaustive decomposition of the material. Similarly, attempts to convert the homoallylic olefin of 55 into the epoxide via directed oxidation with VO(acac) 2 and TBHP again resulted in an intractable mixture of products.
While investigations into installing the epoxide via iodocyclization were ultimately not fruitful, in the course of this route an expeditious and perhaps underappreciated disconnection was identified in the construction of the C15−C16 bond via a simple aldol reaction, followed by TBAF-promoted decarboxylation to remove the ester. We concluded that selective reaction of the terminal olefin in the presence of the trisubstituted olefin was not a feasible proposition. Therefore, a substrate with oxygenation present at C10 and C11 from an early stage was needed as well.

Acetonide approach to epoxide
To obtain the key C10−C11 epoxide in 5 in stereoselective fashion from displacement of a leaving group at C10, a means for the selective formation of a syn-1,3-diol at C11 and C13 is required. Rychnovsky has demonstrated that alkylation of 4-cyano-1,3-dioxanes (cyanohydrin acetonides) constitutes a practical and valuable approach to syn-1,3-diol synthesis [66].
The lithiated cyanohydrin acetonides react as nucleophiles with alkyl, allyl, and propargyl halides, as well as with epoxides.
Although the alkylation itself is highly stereoselective in favor of the axial nitrile, the syn-1,3-diol stereochemistry is ultimately set in a subsequent reductive decyanation step. We planned to synthesize 5 by reaction of the cyanohydrin acetonide 67 with the epoxide electrophile 66 (Figure 8) [67].
The dimethyl derivative of L-malic acid was chemoselectively reduced with borane-dimethylsulfide and sodium borohydride to afford diol 69 (Scheme 18) [68]. The primary alcohol was protected as the TIPS ether, and the secondary alcohol subsequently converted to the TMS ether. Reduction with DIBAL-H afforded the aldehyde 71 without over-reduction to the alcohol. Acetonide formation proceeded smoothly to give 67 as an inconsequential mixture of diastereomers. However, attemps to alkylate the lithiated anion of 67 with epoxide 66 led only to recovery of starting material. Interestingly, although alkylations of the acetonide are known to be stereoselective, protonation does not appear to be, as a cis and trans mixture was obtained from the attempted reaction of a single acetonide diastereomer.
In the course of investigating why 66 and 67 failed to react, attempts were made to trap the anion of 67 with a more reactive electrophile. Allyl bromide reacted rapidly, affording the product 73 as a single diastereomer (Scheme 19). The configuration of this compound, as well as subsequent compounds along this route, could be assigned as the 1,3-syn acetonide by analysis of the 13 C chemical shifts of the acetonide methyl groups [69]. It was recognized that conversion of this olefin to the aldehyde would provide an ideal electrophile for a revised β-ketoester decarboxylation strategy. To this end, reduction of the nitrile proceeded with the expected selectivity; this arises Following the prior procedure, the lithium enolate of TMS ester 64 was reacted with aldehyde 75, and the mixture of diastereomeric alcohols was oxidized to the β-ketoester 76 (Scheme 20). This substrate did not appear to be prone to elimination, and treatment of the β-ketoester with TBAF in THF provided the decarboxylated and deprotected alcohol 77. The primary alcohol could be converted to the tosylate 78 in good yield with tosyl chloride, triethylamine, and trimethylamine hydrochloride as the catalyst.
The alcohol 77 and tosylate 78 contain all of the carbon atoms of epoxide fragment 5 in the correct oxidation state. The remaining steps required to access the epoxide consist of acetonide deprotection, displacement of the tosylate or another appropriate leaving group to obtain the terminal epoxide, and silyl protection.

Conclusion
Nickel-catalyzed reductive coupling methodologies are an attractive fragment coupling strategy for the synthesis of complex natural products. The formation of sensitive skipped diene units in this context remains largely an unsolved problem for organic chemistry, but reactions for the rearrangement of vinylcyclopropanes present an intriguing avenue for exploration. To facilitate future studies in this vein, a cyclopropylenyne corres-

Introduction
Selenofunctionalization of carbon-carbon double bonds provides practicable opportunities for rapid construction of molecule complexity [1][2][3][4][5][6][7][8], because the versatile carbon-selenium bond could either stabilize carbanions [9,10], serve as a radical precursor [11][12][13], or undergo a syn-selective oxidative elimination via the selenoxide [14,15]. A widely accepted mechanism suggests that a key discrete seleniranium ion intermediate is initially formed, and then trapped by internal amine through nucleophilic attack to furnish the product. So far, in addition to the chiral-substrate-induced strategy [16], chiral selenylating agents [17][18][19][20][21][22][23][24] are commonly designed for asymmetric selenofunctionalization of carbon-carbon double bonds. In 2010, Denmark and co-workers reported a Lewis base catalyzed asymmetric selenoetherification of olefins, whereas the enantioselectivity was not quite synthetically attractive (up to 70% ee) [25]. As chiral 3-substituted hexahydropyrroloindoline is a key structural moiety prevalent in a large number of bioactive indole alkaloids [26,27], direct access to which by selenofunctionalization has been considered to be promising but challenging. Danishefsky found that the treatment of bis(Cbz)tryptamine with N-phenylselenophthalimide (N-PSP) in the presence of a catalytic amount of p-toluenesulfonic acid (PTSA) was able to afford a racemic selenofunctionalization product in 84% yield [28]. These leading findings indicate that either Lewis base or Brønsted acid shows catalytic activity for the selenofunctionalization reaction. Since chiral phosphoric acids have been shown to be Brønsted acid/Lewis base bifunctional organocatalysts [29][30][31][32][33], we ask whether the chiral BINOL-based phosphoric acids are able to catalyze the selenofunctionalization of tryptamine derivatives.

Results and Discussion
Initially, we investigated a reaction of bis(Cbz)tryptamine reagent 1a with N-phenylselenophthalimide (N-PSP) (2a) by using phosphoric acids 3 (Table 1, Figure 1) as catalysts for the validation of our hypothesis. Encouragingly, the reaction proceeded smoothly in the presence of 10 mol % of the phosphoric acids evaluated under the assistance of 5 Å molecular sieves. Apparently, the stereoselectivity depended on the N-protecting group of tryptamine 1. When nitrogen atoms of the tryptamine were both protected with Cbz, a very poor enantioselectivity was observed regardless of the catalysts used ( Table 1, entries 1-3). Notably, only the substrate bearing an electron-withdrawing N-protecting group at the indole nitrogen (R 2 ) underwent a smooth reaction to afford the desired product. When the R 2 was replaced with a methyl group, the N-PSP directly underwent a coupling reaction with tryptamine derivatives 1 at the 2-position in 76% yield [34], indicating that the electronically rich substitution inhibited the desired selenofunctionalization reaction. Other selenofunctionalization reagents, such as 2b and 2c could also participate in the reaction under similar conditions, but both showed lower reactivity than 2a.
Interestingly, the selenofunctionalization reagent 2d, which was the best substrate in the reaction developed by Denmark [25], however, was completely unreactive in our case. After the optimal R 2 protecting group and the phenylseleno reagent were determined, we focused on the evaluation of the N-protecting group of the tryptamine (R 1 ) to improve the enantioselectivity. The Fmoc group was found to be better than any other substituents that were screened (Table 1, entries 10 and 11 versus 12). The optimization of reaction parameters including solvents and temperature found dichloroethane (DCE) to be the best solvent in terms of enantioselectivity, and the best results could be accessed by conducting at 0 °C (Table 1, entries 13  and 14).
After optimizing the reaction conditions, a variety of tryptamine analogues were synthesized for this chiral phosphoric acid-catalyzed asymmetric selenofunctionalization. As shown in Figure 2, no matter what the chemical and electronic feature of the substituents on the benzene moiety of either substrates or N-PSP, various tryptamine analogues could be smoothly transformed into the corresponding products in satisfactory yields (65-85%) and with good enantioselectivities (71-89%; Figure 2, 4b-4i). In addition, the products were solid and easy to recrystallize to enhance the optical purity. After a single recrystallization from methanol, the optical purity of some  products, such as 4e, was enhanced to >99% ee. Importantly, the configuration could be assigned by X-ray crystallography. The crystal structure of 4a (>99%) indicated that the configuration of the stereogenic centers was assigned to be (3aR,8aS) ( Figure 3) [35].
On the basis of experimental observations, we proposed a reaction mechanism (Scheme 1). The phosphoric acid acts as a bifunctional catalyst and simultaneously activates both the tryptamine derivative and N-PSP by hydrogen-bonding inter-  action. Then, the asymmetric selenofunctionalization occurred at the 3-substituted tryptamine and subsequently the proton of the phosphoric acid protonates the phthalimide anion to release phthalimide. Finally, the amide on the side chain of the tryptamine derivatives attacks the resultant iminium cation leading to the formation of the product in an enantioselective manner.
Finally, we demonstrate the synthetic application of this reaction in the construction of the 3a-(phenylselenyl)bispyrrolidino[2,3-b]indoline core structure (Scheme 2). Under the optimized reaction conditions, the enantioselective substitution reaction gave 4a in 78% yield and 86% ee. However, when the reaction was scaled up to 20 mmol, both the yield and the stereoselectivity were significantly sacrificed. To our delight, the reaction running on a similar scale could give 4a in 80% yield and 82% ee by tuning the stoichiometry of the phenylseleno reagent (2a) to 1.5 equiv by using 5 mol % of catalyst (R)-3b. After a single recrystallization from methanol, the product 4a was obtained in 50% yield and with 97% ee. The oxidative deselenation of 4a with MCPBA afforded the corresponding alcohol 5 in 95% yield. The stereochemistry of the alcohol was found to be identical to that of the parent selenide, as demonstrated previously [36].

Conclusion
In summary, we have developed a reaction for the enantioselective selenofunctionalization of tryptamine derivatives with N-phenylselenophthalimide (N-PSP) catalyzed by chiral phosphoric acids (up to 89% ee). In this context, we used this protocol to prepare the key chiral precursor of the (+)-alline. Introduction 2-Pyranone is a privilege structure that is often present in natural products and pharmaceuticals, many of which exhibit diverse molecular architectures and biological profiles [1,2]. For example, katsumadain A (1) and B (2), which were isolated from Alpinia katsumadai Hayata (Zingiberaceae), a chinese herbal drug used as an anti-emetic and stomachic agent, are two natural products bearing a diarylheptanoid scaffold that is incorporated into the styryl-2-pyranone moiety [3]. Preliminary biological evaluations showed that 1 and 2 feature anti-emetic activities on copper sulfate-induced emesis in young chicks. More recently, Rollinger et al. disclosed that katsumadain A (1) exhibited prominent in vitro inhibitory activity against the human influenza virus A/PR/8/34 of the subtype H1N1 (IC 50 1.05-0.42 μM) by targeting the enzyme neuraminidase (NA) [4,5]. Moreover, it also inhibited the NA of four H1N1 swine influenza viruses with IC 50 values between 0.59 and 1.64 μM. Therefore, katsumadain A represents an attractive lead structure for the anti-flu drug discovery [6].

Supporting Information
We recently reported the biomimetic total synthesis of katsumadain C [7], a natural product isolated from the same resource as katsumadain A and B [8]. As part of our continuous interest in the synthesis of bioactive 2-pyranone-derived natural products, we launched a project aiming to develop a highly effi-Scheme 1: Proposed biosynthetic pathway and strategic analysis for synthesis of katsumadain A. cient route for the synthesis of katasumadain A as well as its analogues, which would pave the way for their application in further biomedical investigations.
Biosynthetically, katsumadain A is assumed to be derived from styryl-2-pyranone 3 and alnustone (4) [9] through a 1,6-conjugate addition/oxa-Michael addition cascade reaction (path a, Scheme 1). Indeed, both 3 and 4 are known natural substances. Apparently, the biosynthetic pathway represents the most straightforward and convergent approach to synthesize katsumadain A. However, its efficiency might be limited to some extent, given that α,β,γ,δ-unsaturated ketone 4 could undergo a competitive 1,4-conjugate addition to provide the other natural product katsumadain B (path b, Scheme 1). Actually, the regioselectivity of a conjugate addition with α,β,γ,δunsaturated Michael acceptors remains a considerable challenge, as it is heavily dependent on the steric and electronic nature of the substrates [10,11]. Moreover, the enantioselective 1,6-conjugate addition to acyclic dienones or dienoates monosubstituted at the β-and δ-position has rarely been investigated [12][13][14][15], thus leaving open the question of whether or not a biomimetic approach towards katsumadain A might succeed. Keeping these concerns in mind, an alternative strategy was designed as a fallback, in which katsumadain A could be accessed from the lactol 5a and phosphonate 6 via a tandem Horner-Wadsworth-Emmons (HWE)/oxa-Michael addition reaction [16]. In turn, 5a could be derived from 3 and cinnamaldehyde (7) by an organocatalytic enantioselective 1,4conjugate addition followed by the hemiketal formation.

Results and Discussion
Our investigation was initiated by an investigation of the conditions that could effect the proposed biomimetic approach towards katsumadain A and katsumadain B. Both styryl-2-pyranone 3 [17] and alnustone (4) [18] were synthesized according to the literature methods. First of all, the 1,6-conjugate addition of 3 towards 4 was attempted by employing various conditions, including different basic conditions (NaH, DBU, KHMDS) by activation of the nucleophile 3 or acidic conditions (AcOH, TMSOTf, Sc(OTf) 3 and In(OTf) 3 ) by activation of the electrophile 4. However, all of these reactions failed to provide satisfactory results and only lead to the recovery or the substantial decomposition of the starting material. We then turned our attention to the organocatalytic conjugated addition reaction. Among the various documented conditions [19][20][21][22][23], the 9-amino-9-deoxyepicinchona alkaloid-promoted Michael addi- tion is particularly attractive, mainly due to the availability of the catalyst and its superior reactivity towards the activation of the unsaturated ketone substrates through formation of the corresponding iminium intermediate [22]. To our delight, when we tried the standard conditions (30% catalyst A, 60% TFA, DCM, 96 h) in our case, we isolated a product in 25% isolated yield, which was proved to be the 1,4-adduct katsumadain B (2). Encouraged by this result, we further optimized the reaction by screening different solvents (CH 3 CN, THF, DMSO and MeOH) and additives (HCl, TFA and DMAP), aiming to improve the efficiency and selectivity (1,4-or 1,6-adduct) of the reaction. In most of the cases the 1,4-conjugate addition proceeded dominantly, while no or only trace amounts of the 1,6-adduct katsumadain A (1) was observed. The best result was obtained when the reaction was performed with a substoichiometric amount of catalyst A with MeOH as a solvent, in which katsumadain A and B were isolated in a 5:6 ratio with a combined yield of 33% (Scheme 2).
With limited success regarding the the biomimetic synthesis of katsumadain A, we then moved towards the alternative approach as described in Scheme 1. We envisioned that in this scenario an organocatalytic 1,4-conjugate addition [24][25][26][27] between 3 and 7 would circumvent both the reactivity and selectivity issues, which we have struggled with in the aforementioned studies. To validate this hypothesis, we performed a systematic investigation of the organocatalytic 1,4-conjugate addition by examining various reaction parameters, including organocatalyst, acid additive, solvent temperature, and reaction temperature ( Table 1). The first reaction was performed by stirring a mixture of 3 and 7 in DCM at room temperature for 12 h a Each reaction was run with 3 (0.5 mmol) and 7 (0.6 mmol) in 2.0 mL solvent as shown above. b Each of 5a-k was obtained as a mixture of C-5 diastereoisomers (the ratio of α-isomer:β-isomer varied from 1:5 to 1:7. c The ee value of 5a-k was measured with the corresponding lactone product using chiral HPLC. in the presence of Hayashi catalyst B [28]. It was found that the desired product 5a was obtained, albeit in moderate yield and enantioselectivity (Table 1, entry 1). To our delight, the usage of benzoic acid (BA) as an additive could dramatically improve the reaction by affording 5a in a good yield (78%) and a good ee value (91%, Table 1, entry 2). Besides the catalyst B, both Jørgensen catalyst C [29] and MacMillan catalyst D [30] were also tested in this reaction, but gave inferior results (Table 1, entries 3 and 4). As to the acid additive, p-nitrobenzoic acid (PNBA) was found to afford 5a in an excellent yield, but with a decreased ee value (80%). Furthermore, the solvent effect was also examined. Among the several solvents examined, both MeOH and DMSO proved to be suitable solvent systems ( ). Finally, we found that the reaction temperature has some influence on the outcomes, with a slightly improved enantioselectivity (92% ee) obtained at 0 °C (Table 1, entry 10 and 11). Although the best ee value (93%) was achieved at −20 °C, the reaction became sluggish and the yield dropped to 45% (Table 1, entry 12). It is noteworthy that 5a was isolated as a mixture of C-5 diastereoisomers (β-isomer/α-isomer = 5:1 to 7:1) in all of the above cases. For convenience, the ee value of 5a was determined with the corresponding lactone derivative. Furthermore, the absolute stereochemistry of 5a (β-isomer) was assigned as (7S,5R) by using the Mosher ester method (see Supporting Information File 1 for details).
To evaluate the substrate scope of the reaction, we then examined different substituted styryl-2-pyranone and cinnamaldehyde derivatives as Michael addtion donors and acceptors ( Table 2). When styryl-2-pyranone 3a remained unchanged, a variety of cinnamaldehyde derivatives (7a-f) bearing either electron-withdrawing groups (4-Cl, 4-CF 3 and 4-NO 2 , Table 2, entries 2-4) or electron-donating groups (4-MeO or 3,5-MeO, Table 2, entries 5 and 6) on the phenyl ring proved to be suitable substrates, affording the corresponding products (5b-f) in good yields and enantioselectivities. Besides 3a, the Michael addition donors could also be extended to other substituted styryl-2-pyranone derivatives (e.g., 3b and 3c, Table 2, entries 7-11), all of which gave acceptable results. As proof-of-concept cases, the above outcomes indicate that the developed organocatalytic enantioselective 1,4-conjugate addition could be potentially applied to the synthesis of various bicyclic compounds bearing different aromatic moieties (Ar 1 and Ar 2 ), which paves the way to access katsumadain A and its analogues for further biomedical studies.
Having achieved the bicyclic core of katsumadain A in an efficient and enantioselective manner, we then moved towards its total synthesis through the proposed tandem Horner-Wadsworth-Emmons/oxa-Michael addition. As expected, deprotonation of 6 [31] with KHMDS at −40 °C for 0.5 h followed by the addition of the lactol 5a led to the formation of katsumadain A as the only diastereoisomer in 52% yield, apparently via the in situ generated intermediate 8.
The spectroscopic data of the synthetic katsumadain A were in accordance with those of the natural one [32]. However, we found that its optical rotation (

Conclusion
We accomplished the first enantioselective total synthesis of katsumadain A, a naturally occurring influenza virus neuraminidase (NA) inhibitor. The key elements of the synthesis featured a bioinspired, organocatalytic enantioselective 1,4-conjuate addition and a tandem HWE/oxa-Michael addition.
Due to the high efficiency and flexibility of the synthetic route it is applicable to the syntheses of both enantiomers of katsumadain A as well as their analogues. Applications of these compounds in relevant biomedical studies are ongoing in this laboratory, and the progress will be reported in due course.

Experimental
Representative procedure for the organocatalytic 1,4-conjugate addition: To a mixture of 3a (214 mg, 1.0 mmol) and 7a (163 mg, 1.2 mmol) in dry CH 2 Cl 2 (5 mL) at 0 °C was added PhCOOH (24 mg, 0.2 mmol) and catatalyst B (50 mg, 0.2 equiv). The mixture was stirred at 0 °C for 10 h before being quenched by saturated aqueous NH 4 Cl. The mixture was extracted with DCM (3 × 10 mL), and the organic layers were washed with brine and dried over MgSO 4 . The organic solvent was removed under vacuum, and the residue was purified by column chromatography (CH 2 Cl 2 :ethyl acetate = 20:1) to give 5a (284 mg, 82% yield) as a light yellow solid.

Supporting Information
Supporting Information File 1 Experimental procedures and characterization data for synthetic 1, 3a-c, 5a-k and 9a-k.

License and Terms
This is an Open Access article under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The license is subject to the Beilstein Journal of Organic Chemistry terms and conditions: (http://www.beilstein-journals.org/bjoc) The definitive version of this article is the electronic one which can be found at: doi:10.3762/bjoc.9.182

Introduction
Fluorine is the most electronegative element in the periodic table, resulting in a highly polar C-F bond. This gives fluoroorganic compounds unique properties, compared with their parent compounds [1]. Due to the rareness of organofluorine compounds in nature, synthetic fluorinated compounds have been widely applied in numerous areas, including materials, agrochemicals, pharmaceuticals and fine chemicals [2][3][4]. In this context, the stereoselective introduction of fluorine atoms in molecules has become one of the most exciting and intense research areas in the recent years.
Lewis base-catalyzed asymmetric allylic alkylations (AAA) of Morita-Baylis-Hillman (MBH) adducts [5,6], such as acetates and carbonates, have become an attractive option to access Using the established conditions, allylic alkylations of α-fluoroβ-ketoesters 1b-g with MBH carbonate 2a were found to afford the products 3ba-ga in 67-79% yield with 88-96% ee and 3:1 to 4:1 dr ( Table 2, entries 1-6). The results showed that the introduction of various aryl substituents in α-fluoro-β-ketoesters did not affect the reactivity and stereoselectivity. Subsequently, the scope of the allylic alkylation with respect to various MBH carbonates 2 and α-fluoro-β-ketoester 1a was investigated (Table 2, entries 7-20). The desired allylic alkylation adducts 3ab-o were achieved in moderate to good yields with good to excellent enantioselectivities and moderate diastereoselectivities. MBH carbonates ( Table 2, 2b-k) with electron-withdrawing groups appended on the aromatic rings were more active than those ( Table 2, 2l-m) with electron-neutral and donating groups. Excellent ee values with moderate dr values were obtained when the phenyl groups of MBH carbonates were replaced with hetereoaromatic groups, such as thiophene and furan ( Table 2, 2n-o).

Conclusion
We have developed an asymmetric allylic alkylation of MBH carbonates with α-fluoro-β-ketoesters, catalyzed by a commercially available Cinchona alkaloid. Several fluorinated adducts, with chiral quaternary carbon centres containing a fluorine atom, were successfully prepared in 50-93% yields with 84-96% ee and a dr of 3:1 to 4:1. The absolute configurations of adducts still have to be determined and will be reported in due course.

Introduction
The resolution of inflammation is a tightly governed active process effectively mediated by a range of bioactive polyunsaturated fatty acids, peptides and proteins. In 2002, a new family of endogenously generated lipid mediators involved in the resolution of inflammation named the resolvins (resolution phase interaction products) were identified by Serhan and co-workers in the inflammatory exudates of aspirin treated mice [1][2][3]. The resolvins are divided into 2 groups, the D-series resolvins D2 (1) and D1 (2) [3], which are derivatives of docosahexaenoic acid (DHA) (Figure 1) and the E-series [4] derived from eicosapentaenoic acid. Structural analysis by mass spectrometry (MS) showed that resolvin D2 (RvD2, 1) was a 17-hydroxy deriva-tive of DHA (17HDHA) [1]. However, no NMR experiments were performed due to nanogram quantities isolated and the stereochemistry was tentatively assigned based on the proposed biosynthesis via lipoxygenase modification of DHA.
RvD2 (1) prevents the adherence of polymorphonuclear leukocytes (PMN) to the blood vessel wall by promoting the shedding of L-selectin from PMNs thus preventing binding to E selectin on the endothelial cell lining of the blood vessel [5]. Furthermore, RvD2 (1) promotes the influx and phagocytic activity of macrophages, facilitating clearance of dead cells and microbial pathogens, allowing resolution of inflammation and  infection [5]. This successful evaluation of the resolvin series in preclinical models of bacterial sepsis has stimulated strong interest in their therapeutic potential, as RvD2 appears to express the unusual combination of anti-inflammatory and antimicrobial activity. Further interest in inflammationresolving lipids is stimulated by their inhibitory effects on inflammatory pain, which are mediated via inhibition of the activity of TRPV1 and TRPA1 calcium channels on sensory nerves [6]. Resolvin D1 (2) [7] has been shown to act directly on human PMNs and also regulates actin polymerization [8]. Whilst RvD1 has been shown to act on the FPR2 and GPR32 types of G-Protein-coupled receptors, the receptor(s) for RvD2 remain to be identified. Identification of the receptors mediating the combined anti-inflammatory and antimicrobial actions would facilitate efforts to identify ligands that have better druglike properties than RvD2 or its analogues. However, such efforts have been limited by the lack of availability of suitable amounts of RvD2.
The first total synthesis of (7S,16R,17S)-RvD2 (1) was communicated by Spur in 2004 [9] but this report did not include an experimental section although physical data for some compounds was provided. A similar synthesis of RvD2 was utilized by others for the production of 1 for a biological study [5] but again, there was no experimental provided. The total synthesis of resolvin D1 has also been reported [10] along with resolvins D3 [11], D5 [12], D6 [13] and resolvins E1 [4,14,15], E2 [16,17] and E3 [18] with full experimental details included for resolvins D3 [11], E2 [16] and E3 [18]. An improved synthesis of the C16-C20 fragment of resolvin E1 has also been reported [19]. We were interested in accessing amounts of RvD2 (1) for biological evaluation but without detailed synthetic sequence to follow and given the very high cost [20] of commercial 1 we elected to develop an alternative route to provide this important compound and analogues for further biological evaluation.
Herein we describe a synthesis of RvD2 (1) which includes full experimental details so that other researchers can produce useful amounts of this important compound as well as novel isomers.

Retrosynthetic analysis
A retrosynthetic analysis of RvD2 (1) is shown in Scheme 1. It was envisaged that the target compound 1 could be secured via a Sonogashira coupling to form the C11-C12 bond followed by partial reduction. This is similar to the endgame of the reported syntheses of 1 [5,9] but both of these approaches involved formation of the C9-C10 bond as the convergent step. In our approach, 1 could arise from enyne 3 and vinyl iodide 4 which could both be obtained by Wittig extension using the common linchpin phosphorus ylide derived from phosphonium salt 6 [21,22] and each of the homochiral aldehydes 5 and 7 [9].

Synthesis of vinyl iodide 4
The synthesis of fragment 4 began with the production of the aldehyde 7 as shown in Scheme 2. A Wittig reaction between hemiacetal 8 [23] and the ylide derived from 9 provided the alkene 10 [9] with excellent stereoselectivity. Oxidation of 10 with Dess-Martin periodinane then afforded aldehyde 7. The phosphonium salt 6 [21,22] was produced from propargyl bromide via silylation of the derived sodium salt with TIPSCl followed by reaction with triphenylphosphine.
Treatment of the salt 6 with n-BuLi gave the ylide and condensation with the aldehyde 7 afforded the desired E-enyne 11 along with the Z-isomer in a ratio of 2.2:1 which were easily separated by flash chromatography (Scheme 3). The minor Z-isomer could also provide novel stereoisomer analogues of RvD2 (1). Removal of the TIPS group with TBAF gave terminal alkyne 12. Alkyne 12 then underwent smooth hydrozirconation utilizing the procedure reported by Negishi [24] were ZrCp 2 HCl is generated in situ by reduction of ZrCp 2 Cl 2 with DIBALH in THF. Iodinolysis of the zirconium species then gave the diene iodide 4 in good yield. Selectivity for this process was excellent with only a trace of the regioisomer formed.

Synthesis of dienyne 3
Our approach to the aldehyde 5 began with the production of the known bis-TES ether [9] produced by an alternative procedure (Scheme 4) in which the C7 stereochemistry was introduced via asymmetric dihydroxylation [25,26]. Thus, ester 13 [27] was treated with AD-mix-α in t-BuOH/H 2 O to give diol 14 in reasonable yield. The enantioselectivity and absolute configuration of the secondary alcohol was determined by conversion of diol into the bis-(S)-Mosher ester [28,29]. Integration of the 1 H MMR spectrum indicated the e.r. was 93.7:6.3 and Mosher analysis (See Supporting Information File 2 for details) confirmed the stereochemistry of the new asymmetric center of the major enantiomer as S in accord with the predicted outcome [25]. Silylation gave bis-TES ether 15 and partial reduction of the alkyne using P2-Ni as catalyst [30] afforded the alkene 16.
Desilylation of the primary TES group in 16 and concomitant oxidation to aldehyde 5 was achieved under Swern conditions as reported by Spur [31]. Deprotonation of salt 6 with LiHMDS followed by condensation with aldehyde 5 gave the E-enyne and the corresponding Z-isomer in a 4:1 ratio. The use of LiHMDS as base was critical for reasonable yields and stereoselectivity in this case. Global deprotection of 17 with TBAF gave the enyne 3 in good yield.

Total synthesis of resolvin D2 (1)
The completion of the synthesis of RvD2 (1) is shown in Scheme 5. Sonogashira coupling [32,33] between 3 and 4 was very efficient giving the alkyne 18 in good yield. Removal of the acetonide was effected by treatment with HCl in MeOH to give the known triol 19 [9]. The final steps to 1 were similar with those previously reported [5,9]. Thus, partial reduction of the triple bond using Zn(Cu/Ag) [34] to afforded RvD2 methyl ester 20 in 76% yield. A large excess of the Zn reagent was required to obtain a good conversion of 19 into 20. The 1 H NMR spectrum (CDCl 3 solvent) of RvD2 methyl ester (20) compared well to that reported [5]. Final ester hydrolysis and mild acid work-up then gave RvD2 (1).
The synthetic RvD2 (1) had physical data identical to that reported [9,35] and we measured the specific rotation of this material for the first time ([α] D −17.5° (c 0.075, CH 2 Cl 2 )). In our hands, both RvD2 methyl ester (20) and RvD2 (1) itself were highly unstable, especially to acid. Prolonged standing in CDCl 3 or CD 3 CN solution or exposure to light caused rapid decomposition and so NMR spectra were obtained quickly. We found that RvD2 methyl ester (20) was not very soluble in CD 3 CN so spectra were best run in CDCl 3 that was filtered through basic alumina immediately prior to use. Spectra for RvD2 (1) were always measured for CD 3 CN. Even with short exposure to the solvent, we still observed degradation to unidentified compounds. Samples of RvD2 (1) can be stored in EtOH or frozen in DMSO solution but should be used immediately upon thawing. Alternatively, the triol 19 proved more stable than both RvD2 methyl ester (20) and RvD2 (1) and can be stored for longer periods prior to conversion to 1 which should be used rapidly for biological assessment to avoid degradation.

Conclusion
The total synthesis of RvD2 (1) has been completed using a common linchpin Wittig reaction. Using this approach, we were able to prepare sufficient quantities of this important inflammation resolving compound for further biological evaluation.