Synthetic studies toward eleganine A

Eleganine A is a cytotoxic indole alkaloid recently isolated from the leaves of Tabernaemontana elegans . Its unique structure arises from rearrangement of a canonical corynanthe skeleton, resulting in the presence of a 4-ethylidene-3-alkylproline core. Employing a chiron approach, we describe an efficient and scalable synthesis of the proline subunit of eleganine A, as well as efforts toward its proposed structure


Introduction
Alkaloids belonging to the corynanthe family have long attracted the interest of synthetic chemists owing to their complex architectures and intriguing biological properties.These natural products feature the monoterpene connectivity shown in Figure 1, typically resulting in tetrahydo-β-carboline core structures that can further rearrange to aspidosperma-or iboga-type alkaloids. 1,2In 2009, Ferreira and co-workers reported the isolation of eleganine A (1) from the leaves of Tabernaemontana elegans in Mozambique. 3 Eleganine A exhibited cytotoxicity and induction of apoptosis in human hepatoma cells as measured in a trypan blue cell viability assay.In addition, compound 1 features an unprecedented azabicyclo[6.2.1]undecane core structure presumed to derive from rearrangement of the corynanthe monoterpene unit.Excelsinidine (2) 4 and 17-norexcelsinidine (3) 5,6 are naturally-occurring reduced congeners of eleganine A featuring similar terpenoid connectivity.A recent biomimetic synthesis of 3 demonstrated its relationship to the geissoschizine skeleton using an elegant oxidative rearrangement approach. 7To date, there are no reported synthetic studies toward 1.
In considering a modular synthetic approach to 1, we were intrigued by the presence of a 4-ethylidene-3-alkylproline embedded within a new alkaloid framework.We previously reported the synthesis of several 4alkylideneprolines, [8][9][10] which are found in a select number of natural products including the lucentamycins, 11,12 isodomoic acids, 13 and tomaymycin. 14,15Our strategy relied on an ester enolate-Claisen rearrangement to give the linear proline precursors, setting both stereocenters and the pendant alkene geometry in a single step. 9he E configuration of the alkene in 1 precluded application of this approach, as the chair-like transition state of the rearrangement provides the opposite alkene geometry.This prompted us to consider a conceptually novel approach toward the 4-ethylidene-3-alkylprolines subunits of 1-3.Here, we report the stereospecific synthesis of the fully elaborated core of eleganine A, as well as efforts toward its proposed structure.

Results and Discussion
Our retrosynthetic plan for eleganine A is depicted in Figure 2. We envisioned that nucleophilic attack of the pyrrolidine nitrogen onto an oxocarbenium ion would provide the cyclic hemiaminal ether in the final step of the synthesis.Acylation of an appropriately substituted C2-lithiated indole with pyrrolidine derivative 5 could in turn provide 4. A protected 4-alkylidene-3-alkylprolinol (5) serves as the key intermediate in our synthetic plan, and its stereochemistry at C2 was traced back to D-serine as a chiral progenitor.We anticipated that the configuration of C3 could be set through a diastereoselective reductive Heck-type cyclization involving the vinyl halide and enoate groups in 6.The early formation of the C2-C3 bond in our approach enables the synthesis of large quantities of trans-substituted pyrrolidine 5, which we also viewed as a potential precursor to related indole alkaloids such as 2 and 3.The synthesis commenced with alkylation of readily available D-serine derivative 7 16 with (Z)-1-bromo-2iodobut-2-ene, 17 followed by Boc protection and ester reduction with lithium borohydride (Scheme 1).Compound 8 was then converted to enoate 9 in 70% yield using standard conditions.Reductive-Heck type cyclization of 9 was initially accomplished in the presence of excess Ni(COD) 2 and triethylamine to give moderate yields of 10, along with various byproducts.Given the sensitive nature of the metal complex and required use of stoichiometric reagents, we explored catalytic systems as an alternative.Gratifyingly, we found that 20 mol% of Ni(PPh 3 )Cl 2 in the presence of freshly activated zinc afforded the desired pyrrolidine 10 in 68% isolated yield.Analysis of the crude reaction mixture showed this transformation to be highly diastereoselective, with none of the minor cis isomer detected by 1 H NMR. This selectivity may be attributed to the higher activation barrier associated with a C2,C3-cis transition state, as well as conformational preorganization governed by 1,3-allylic strain, as shown in Scheme 1.In anticipation of nucleophilic attack onto the carbonyl in 10, we then converted the ester into Weinreb amide 11 to provide another potential substrate for indole acylation.Initial attempts to join the indole and pyrrolidine fragments relied on C2 lithiation of dimethyl acetal derivative 13 (Scheme 2).After screening several conditions, we found lithium tetramethylpiperidine to be the optimal base for C2 metalation. 18This was verified by trapping the lithiated intermediate with ethyl chloroformate to give 14 in good yield.Despite this encouraging result, we were unable to promote the condensation of 13 with either methyl ester 10 or Weinreb amide 11 to give 15.Attempts to employ other protected indoles (replacement of the N-Boc group) or active ester derivatives of 10 also failed to provide appreciable amounts of acyl indole products.As an alternative to direct indole acylation, we turned to the alkyne heteroannulation strategy depicted in Scheme 3. Thus, alkynylation of 11 with 4,4-dimethoxybutyne 19 afforded ynone 16, which would serve as a substrate for Larock indole synthesis.The regioselectivity of Larock heterannulation has been extensively studied and is thought to be governed primarily by steric factors in the case of asymmetric alkynes. 20This selectivity arises during the alkyne carbopalladation step wherein the larger substituent prefers to orient itself away from the forming carbon-(aryl)carbon bond.2][23] A relevant study by Chuawong and coworkers employed para-substituted diphenylacetylenes to establish a positive Hammett correlation between electronwithdrawing substituents and regioselectivity. 24In these cases, polarization of the alkyne results in preferred migration of the electrophilic Pd center to the more electronegative carbon.We observed regioselectivity consistent with this model in the heteroannulation of 16 with 2-iodoaniline.In the presence of Pd(PPh 3 ) 4 and K 2 CO 3 in THF, indole 15 was obtained in 55% yield along with 18% of the undesired regioisomer, which was readily separated by flash chromatography.With intermediate 15 in hand, the final stages of the synthesis required oxidation of the protected primary hydroxyl group and acid-promoted cyclization to form the hemiaminal ether.Thus, silyl ether cleavage with TBAF was followed by oxidation using (bisacetoxyiodo)benzene and catalytic TEMPO.Esterification of the crude carboxylic acid with trimethylsilyldiazomethane then afforded 17 in 64% combined yield over 3 steps.Oxidation immediately following silyl ether cleavage and flash chromatography was critical, as the resulting alcohol was found to decompose over several hours at room temperature.We screened various conditions to promote Boc deprotection and concomitant cyclization of 17.While one-pot procedures failed to afford desired product, we found that treatment with acidic methanol cleaved the Boc group, leaving the dimethyl acetal intact.The crude amine was then treated with dilute aq.HCl in trifluoethanol/DCM, resulting in the formation of a compound with 1 H NMR signals characteristic of a trans-enamine.This solid was recrystallized from chloroform/hexanes and X-ray diffraction confirmed its structure to be that of dimeric species 18.
Despite several efforts to obtain monomeric cyclization products, we found that acidic conditions consistently favored the formation of 18 over the heminaminal ether 1.
In observing the apparently strong propensity for formation of dimeric enamine 18, we considered the influence that the relative stereochemistry at C2 may have on the final cyclization.The C2-C3 trans relationship in eleganine A was originally assigned on the basis of anisotropic shielding of the methyl ester NMR signal by the indole  electron cloud. 25This phenomenon has been observed in a variety of related nonrearranged monoterpene alkaloids including vobasine and tabernaemontanine. 26However, in most cases the bridgehead carbon in question bears the opposite configuration, placing the ester group in closer spatial proximity to the indole ring.Subsequent to the isolation of 1, Girardot et al. identified a reduced congener of 1, dihydroeleganine A , in which the C2-C3 relationship was characterized as cis. 27This raises the possibility that the pyrrolidine ring in eleganine A may also harbor a cis substitution pattern and that this configuration may influence the stability of the product, or its propensity for dimerization.

Conclusions
In summary, we have described a concise approach to the synthesis of the eleganine A core structure using Dserine as a chiral synthon.Our strategy relies on a highly diastereoselective reductive Heck-type coupling to form the pyrrolidine ring and a Larock heteroannulation to install the indole.Attempts at late-stage hemiaminal ether cyclization resulted in the unexpected formation of a dimeric enamine whose structure was confirmed by X-ray diffraction.We are currently exploring the factors that influence the final hemiaminal ether formation step, including the possibility of configurational missassignment at the bridgehead methine in 1.A select number of monterpene alkaloids derived from corynanthe precursors feature a rearranged skeleton harboring the 4-alkylidene-3-alkylproline core of eleganine A. The described approach should find utility in the syntheses of these and related natural products.

Experimental Section
General.Unless stated otherwise, reactions were performed in flame-dried glassware under a positive pressure of argon or nitrogen gas using dry solvents.Commercial grade reagents and solvents were used without further purification except where noted.Toluene, Et 2 O, DCM DMF, and MeCN were used following passage through a Pure Process Technologies solvent purification system.Other anhydrous solvents were purchased directly from chemical suppliers.Thin-layer chromatography (TLC) was performed using Merck 60 F254 silica gel pre-coated glass-backed plates (0.25 mm).Flash chromatography was performed using silica gel cartridges (40-65 μm particle size).Reaction progress was judged by TLC analysis (single spot/two solvent systems) using a UV lamp, CAM (ceric ammonium molybdate), ninhydrin, or basic KMnO 4 stain(s) for detection purposes.NMR spectra were recorded on a 400 or 500 MHz spectrometer.Proton chemical shifts are reported as δ values relative to residual signals from deuterated solvents (D 2 O, CDCl 3 , CD 3 OD, or DMSO-d 6 ).