Unified Total Synthesis of Pyrroloazocine Indole Alkaloids Sheds Light on Their Biosynthetic Relationship

The total synthesis of seven members of the lapidilectine and grandilodine family of alkaloids has been accomplished in racemic and enantiopure form without protection/deprotection of functional groups. The two key steps, an 8-endo-dig hydroarylation and a 6-exo-trig photoredox cyclization, were catalyzed using gold. A rationale for the formation of the cyclopropane ring of the lundurines is also provided.


General information
All solvents and other chemicals were used as received, unless otherwise stated.
Methyl (±)-1-cyclohexylprop-2-en-1-ol, 1  HPLC-grade solvents were used for UV, CD, optical rotation and electrochemical measurements. CHCl 3 was additionally filtered through basic Al 2 O 3 prior to its use for optical rotation measurements of tertiary amine substrates.

S6
All compounds were obtained in pure form (≥ 98% wt purity by 1 H NMR), unless otherwise stated, and characterized with 1 H, 13 C NMR and mass spectra. The structures of the following racemic compounds were confirmed by single crystal Xray diffraction studies: Chiral HPLC analyses were carried out on an Agilent Technologies instrument HPLC 1100 series with VWD detector or HPLC 1200 series with DAD detector. The ee values were obtained by chiral HPLC for the following compounds: For other chiral compounds the ee values were assumed to be the same, as for their direct precursors or derivatives, and for every batch the measurement was repeated separately to ensure their enantiomeric purity.
The synthesized enantiopure natural products had the same sign of optical rotation as previously reported ones isolated from natural sources, with the exception of lapidilectam (4), for which our data strongly suggests to reconsider the sign of the optical rotation from previously reported (+) to (-), see below. The absolute configuration was established by single crystal X-ray diffraction studies for the following compounds: Details of single crystal X-ray diffraction studies are presented in the section "Crystallographic data", page S117.
It is important to note that Kopsia grandifolia species was previously called Kopsia lapidilecta. 6 , 7 , 8 In the modern classification, Kopsia lapidilecta describes another plant, which is endemic to Natuna island, Indonesia. 9 The previous proposal of biosynthetic relationship between pyrroloazocine indole alkaloids is depicted below. 10 The nomenclature of the compounds follows the names as given by ChemDraw however for practical reasons, we decided to use the numbering as defined in the initial isolation report to compare the NMR data of our synthetic grandilodines/lapidilectines.
The biphasic mixture was diluted with CH 2 Cl 2 (50 mL) and the organic layer was separated. The aqueous layer was re-extracted with CH 2 Cl 2 (70 mL) and the combined organic layers washed with a saturated solution of NaHCO 3 (2 × 100 mL), dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. Purification by flash column chromatography on silica gel eluting with cyclohexane/EtOAc 2:1 gave the title product as a colorless solid (484 mg, 1.66 mmol, 82%). In a similar experiment on 16 mmol scale the cyclized product 16a was isolated in 79% yield (3.70 g).
Note 1: Unlike its 5-OMe derivative, 2 15a was a more difficult substrate for the Aucatalyzed cyclization. Additive of acetic acid was crucial to avoid the deactivation of the catalytic system and to reach full conversion. The origin of acid additive effect in Au-catalyzed alkyne hydroarylation with indole has previously been discussed. 16 Note 2: Alternatively, in the presence of ligand-stabilized cationic gold complex [IPrAu(NCCH 3 )]SbF 6 (2 mol %), 15a undergoes the gold-catalyzed cyclization with a satisfactory ca. 20  Note: In a similar experiment the yield of the crude 16a was ca. 87%, and it contained ca. 5% of exo-cyclization isomer, that was separated in protection/crystallization sequence
The flask was sealed and the resulting mixture stirred at 25 ºC for 30 min, then another portion of AuCl (64.5 mg, 0.28 mmol, 2 mol %) was added and stirring was continued for 1 h, whereupon TLC indicated full consumption of the starting material.
The flask was removed from the glovebox and the mixture slowly added to a saturated aqueous solution of NaHCO 3 (300 mL) in a separating funnel. The flask was washed with CH 2 Cl 2 (2 × 50 mL) and the biphasic mixture shaken vigorously, evacuating the CO 2 formed. This operation was continued until vigorous effervescence had ceased. The biphasic mixture was diluted with CH 2 Cl 2 (150 mL) and the organic layer was separated. The aqueous layer was re-extracted with CH 2 Cl 2 (100 mL) and the combined organic layers washed with a saturated solution of NaHCO 3 (2 × 100 mL), dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. Purification by flash column chromatography on silica gel eluting with cyclohexane/EtOAc 2:1 gave the title product 16b as a colorless solid (4.03 g, 10 yield of (±)-18 based on recovered starting material = 92%).
Note 1: The reaction should be carefully monitored by TLC to minimize over hydrogenation and the formation of indoline byproducts. Note 1: Pyridine N-oxide was dried and stored in the glovebox as a crystalline solid.

Note 2:
The density of t-BuNC (obtained from Aldrich) was found to be 0.70 g/mL.   Figure   S1). In the abovementioned experiments, the vials were placed in two-diode-sockets.
The irradiance at 365 nm in the two-diode-socket, measured at the vial distance of reaction (ca. 2 mm) was found to be 2.6 mW/cm 2 (voltage = 26V, current = 0.07 A in our setup). A high intensity UV irradiation was found to be important to achieve high yield in this transformation.
Note 2: Product 10 was found to be air-sensitive, presumably because of a possible autoxidation at the allylic position. Thus, the isolation was performed within two to three hours with particular precautions (the compound was kept under Ar atmosphere, the preparative TLC chamber was purged with argon). After isolation, 10 was placed under inert atmosphere and stored in the freezer.

Block of PTFE
Two parallel lines of three diodes in each S33 (±)-10, Procedure B: In a glovebox, a solution of (±)-11 (117 mg, 0.245 mmol, 1 equiv), 4-MeOC 6 H 4 NPh 2 (135 mg, 0.49 mmol, 2 equiv) and [Ru(bpy) 3 ]Cl 2 6H 2 O (9.4 mg, 0.012 mmol, 5 mol %) in DMF (anhydrous, oxygen-free, 3.7 mL) was placed in a vessel designed for photochemical reactions ( Figure S2). The vessel was sealed, removed from the glovebox and the resulting orange solution was irradiated with blue light (LEDs) for 12 hours (an air flow was applied for external cooling). After this time, catalyst decomposition was clearly observed (deep purple color of the reaction mixture). The reaction mixture was poured onto CH 2 Cl 2 -H 2 O (50 mL:50 mL). The organic phase was separated and the aqueous layer was extracted with CH 2 Cl 2 (25 mL × 2). The combined organic layers were washed with brine (25 mL), dried over Na 2 SO 4 , filtered, concentrated, and the obtained residue was purified on a 2000 µm preparative silica gel TLC using EtOAc as eluent. The title product (±)-10 was isolated as a pale yellow solid in 40% yield (39.0 mg, dr >50:1).
Note 1: Ledxon LED stripe (1.0 m, 60 blue LEDs, λ = 470 nm) coiled outside of a crystallizing dish (Ø = 11 cm) was used for blue-light irradiation ( Figure S2). The irradiance at 470 nm of our setup, at the distance of the reaction vessel, was found to be 6.8 mW/cm 2 (calibrated photodiode).

Note 2:
The starting material was recovered as an inseparable mixture with, presumably, the chloride derivative of (±)-11 arising from Br-Cl exchange (46 mg, ca.

40%).
Note 3: Deep purple block crystals deposited on the vessel walls in the course of the photochemical reaction. These crystals were analyzed by single-crystal X-ray diffraction, and the structure [(bpy) 2 RuBr 2 ]Br was obtained. This deep-colored compound presumably interferes with the blue-light mediated photoredox process, and is responsible for the deactivation of the catalytic system.

S35
General note on the dr estimation: The diastereoselectivity of the photoredox transformation was estimated after careful examination of the 1 H NMR spectra of crude reaction mixtures and isolated products.
It was noticed that all samples of isolated 10 still contained a minute amount of an inseparable impurity that could be observed in the 1 H NMR as signals at 3.80 (s) and 3.60 (s) ppm. This impurity was assumed to be the exo-isomer of 10, due to the similarity of its chemical shifts with those observed in related compounds, containing the   hours, the reaction mixture was cooled to 0 ºC, diluted with CH 2 Cl 2 (10 mL), and deionized water (15 mL) was slowly added at 0 ºC. The resulting mixture was allowed to warm to 25 ºC and stirred for 30 minutes. Then the layers were separated, the aqueous layer was extracted with CH 2 Cl 2 (3 × 10 mL) and the combined organic layers were washed with a saturated aqueous solution of NaHCO 3 (20 mL), dried over      Note 1: Crystals suitable for single-crystal X-ray diffraction studies were obtained from CH 2 Cl 2 /cyclohexane solvent system by slow evaporation method under argon atmosphere.

Note 2:
The title product undergoes rapid decomposition when exposed to air, silica,  (+)-1: Procedure similar to the previously reported one. 21 To a suspension of (+)-grandilodine C (17.0 mg, 0.0447 mmol, 1 equiv) and activated molecular sieves (4Å, 50 mg) in anhydrous CH 2 Cl 2 (4. Note 3: Nishida et al. 21 previously reported the reduction of (+)-grandilodine C into (+)-lapidilectine B with Me 3 OBF 4 /NaBH 4 . We performed the reduction of (±)grandilodine C according to their procedure, and isolated a product mixture with a 1 H NMR spectrum nearly identical to the one reported 21 for lapidilectine B (see below).
Careful examination of our 1 H NMR spectral data revealed that the integration of S45 aromatic protons in comparison to olefinic ones was off (higher) by ca. 20%.

Moreover, HRMS(ESI+) analysis showed the presence of two molecular ions with 2H
difference. All these data suggest that a 5:1 mixture of lapidilectine B and its dihydroanalogue 26 was obtained, and it is inseparable by standard silica gel chromatography. After extensive studies, we found that this mixture can be separated on alumina TLC plates after multiple (10-12) runs eluting with cyclohexane/EtOAc 4:1. Comparison of our and previously reported 21

Preliminary studies on the construction of benzylic C-C bond.
Initial experiments on benzylic C-C bond construction were performed on

S50
We also attempted to perform a one-pot radical cyclization / benzylic C-C bond construction on substrate 11 or D 18,19 -11, but without success. 24 The and washed with acetone/CH 2 Cl 2 mixture (150 mL). The volatiles were removed, and the residue was purified on 2000 µm silica gel preparative TLC using EtOAc/CH 2 Cl 2 10:1 as eluent and acetone/CH 2 Cl 2 1:1 to collect the compound from the silica.
Evaporation of solvents and drying under high vacuum gave the title product (±)-9 as a white foamy solid in 73% yield over two steps (76.3 mg, 84 wt% purity: 3% of alkene (±)-10 and 12% of residual solvents). The obtained material was used in the next step without further purification.
Note 1: Evaporation from cyclohexane/CH 2 Cl 2 (10:1) and drying under high vacuum gave the title product as a white solid, that was more convenient to handle.  Formation of (±)-22: A solution of (±)-21 (13.2 mg, 0.03 mmol) in CD 2 Cl 2 (0.6 mL) was placed in an NMR tube and was treated with pyrrolidine (13 µL, 0.16 mmol, ca. Scheme S1. Proposed main and side pathways for the two-carbon cleavage in 9.       were obtained from CH 2 Cl 2 /cyclohexane/pentane solvent system by diffusion method.

Note 2:
The use of THF as a co-solvent was essential to achieve the desired reactivity.
In the absence of THF no conversion of lapidilectam or only traces of 1,4-reduction were observed. We think that the lack of reactivity is related to the low solubility of        Note: Crystals suitable for single-crystal X-ray diffraction studies (colorless plates)

S73
were obtained from CH 2 Cl 2 /EtOAc solvent system by slow evaporation method.         and MeO-derivative (20Me) in ca. 30% and 32% NMR yields respectively (comparison of NMR spectra see below).
Step 2: The NMR sample was transferred into a 3 mL vial, concentrated, dried under high vacuum, redissolved in CH 2 Cl 2 (0.10 mL) and treated with 50% (v/v) aqueous H 2 SO 4 (0.10 mL). The purple reaction mixture was stirred for 3 hours, diluted with

CV curves for purified carboxylic acid (±)-24 (4 mg) were obtained
showing oxidation process at ca. 1.8 V (see the plot below). Na 2 CO 3 (25 mg) was added, the voltammetry cell was sonicated for 2 minutes, and the CV curves were measured again, showing an oxidation peak at ca. 0.9 V (see the plot below).

Experimental Section
General Methode'H NMR spectra were measured in CCl, with a 60-MHz instrument (Varian 360 EM); chemical shifte (6) are reported in ppm values, referenced to tetramethyleilane as internal standard. Mass spectra were obtained with a Hewlett-Packard 5988 A spectrometer; the ratios m/z and the relative intensities (%) are reported. GLC (gas-liquid chromatographic) analyses were performed with a Hewlett-Packard 5890 A instrument connected to a Hewlett-Packard 3390 A integrator.
The electrochemical studies were conducted in methanol with tetrabutylammonium hexafhorophmphate (0.1 M) as supporting electrolyte or in distilled water with sodium sulfate as supporting electrolyte. Cyclic voltammograms were obtained with a programming function generator (Tacussel IMT-1) connected to a potentiostat (Tacuwl PJT 120-1). The working electrode was graphite with a saturated calomel reference electrode, separated from the test solution by a salt bridge containing the solvent and supporting electrolyte. The auxiliary electrode was a platinum wire. The electrochemical cell used for potential-controlled electrolysis was a conventional H-type design with anodic and cathodic compartments separated by a porous glass frit.
Isolation and purification were done by flash column chromatography on silica gel (Merck 60,76230 mesh) with hexane as eluent or by high-performance liquid chromatography (HPLC) (Waters isocratic HPLC equipment) with a semipreparative silica gel (Microporasil column and hexane-ethyl acetate as eluent. Materials-Naproxen, obtained from Naproval (Lab. Vallen Heutre), was purified by chromatography on a short silica gel column with CH2Cl, as eluent, followed by recrystallization from CH,Cl,n-hexane.
The sodium salt of naproxen was obtained by addition of an equimolar amount of CH,ONa to a solution of naproxen in CH30H. The solid remaining after evaporation of the resulting solution was dried under vacuum and washed with dichloromethane and ether.
Oxiahtion of Naproxen Salt ( I -) in MethanobA mixture of naproxen (30 mg), a n equimolar amount of KOH, and the indicated amount of ceric ammonium nitrate (Table I)  Electrochemical Oxidation-A typical electrolysis was performed as follows: 750 mg of naproxen (1H) or its salt (l-) waa dissolved in 35 mL of water (pH 8-9) or methanol, and the solution was electrolyzed for several hours at low current intensity (initially about 20 mA). Electrolysis was stopped when current intensity was about 1% of the initial value. The total charge was obtained by integration of the curve of intensity versus time. After electrolysis in methanol, the solvent was removed with a rotary evaporator, and the residue was partitioned between aqueous d i u m hydroxide and ether. The organic phaee was dried (NaaO,), filtered, and evaporated. After electrolysis in water, ether was added for extraction and then the organic layer was dried (Na,SO,), filtered, and evaporated. The product mixtures were analyzed by GLC and mass spectrometry.

Conclusions
The data show that naproxen (1H) and its conjugate salt (1 -1 undergo oxidative decarboxylation chemically, by treatment with known one-electron oxidizing reagents, or electrochemically. The product patterns can be related to the fate of intermediates I, 11, and 111, when they are generated by ejection of one electron in the photolysis of 1-and 1H.
Presumably, this behavior is involved in the photobiological reactions of naproxen and reflects the low oxidation potential of this drug.

Irradiation of lactone 8 with UVc light (254 nm)
. Lactone (±)-8 (4.0 mg, 0.01 mmol) was placed in a quartz NMR tube, dissolved in CD 3 OD (0.6 mL) and the tube was sealed with a rubber septum. The initial 1 H NMR spectrum was recorded. The NMR tube was placed inside a quartz vessel (ca. 3 cm OD) filled with water (to prevent overheating of the reaction mixture), the vessel was placed at the center of a Rayonet RPR-200 photochemical chamber reactor equipped with RPR-2537Å lamps (see Figure S4 for specifications), and it was irradiated for 20 minutes, whereupon a 1 H NMR spectrum was recorded. The NMR tube containing the reaction mixture was irradiated for additional 25 minutes, and a final 1 H NMR spectrum was recorded. Our results are summarized in the Table S8.

Irradiation of lactone 8 with UVb light (300 nm)
. Lactone (±)-8 (4.0 mg, 0.01 mmol) was placed in a quartz NMR tube, dissolved in CD 3 OD (0.6 mL) and the tube was sealed with a rubber septum. The NMR tube was then placed inside a quartz vessel (ca. 3 cm OD) filled with water (to prevent overheating of the reaction mixture), the vessel was placed at the center of a Rayonet RPR-200 photochemical chamber reactor equipped with RPR-3000Å lamps (see Figure S5 for specifications) and it was irradiated for 20 minutes, whereupon a 1 H NMR spectrum was recorded.
The NMR tube containing the reaction mixture was irradiated for additional 45 minutes and a second 1 H NMR spectrum was recorded. Irradiation was continued for an additional hour and a final 1 H NMR spectrum was recorded. Our results are summarized in Table S8. Figure S5. Specifications of our Rayonet reactor equipped with RPR-3000Å lamps.

Comparison of NMR data (CD 3 OD, 298K):
Lactone 8: and it was irradiated for 20 minutes, whereupon a 1 H NMR spectrum was recorded.
The NMR tube containing the reaction mixture was irradiated for additional 2.5 hours, and a final 1 H NMR spectrum was recorded. Our results are summarized in Table S8.

Autoxidation of (-)-Lundurine B into (-)-Lundurine A.
An HPLC-sample of (-)-lundurine B 2 (ca. 2 mg) was concentrated and kept under air for 16 months. The sample was then dissolved in CDCl 3 and its 1 H NMR spectrum was recorded showing ca. 50% conversion of lundurine B into lundurine A.

General comment
All DFT calculations were carried out using the Gaussian09 suite of programs. 32 The geometries were fully optimized without any constraints with B3LYP functional 33 and ultrafine integration grid. Solvent effects were taken into account by means of the implicit polarizable continuum model (PCM) 34

Radical cyclization
DFT calculations for radical cyclization were performed with acetonitrile as a solvent.
First, the cyclized benzylic radicals were calculated for both C16-CO 2 Me epimers: endo (R2a) and exo (R2b). For each epimer four orientations of the CO 2 Me groups (two for the carbamate and two for the ester) were considered. For both R2a and R2b, the rotamers featuring the carbamate OMe-groups and the C16-CO 2 Me pointing toward the aromatic ring were found to be slightly lower in energy than others (up to 2 kcal/mol). The same rotameric configuration was favored in the corresponding transition states of the radical cyclization leading to R2a and R2b (up to 1.5 kcal/mol), so it was selected to compute the full reaction profile. The most stable form of open a-CO 2 Me radical R1 undergoes rotation around C20-C17 bond with the formation of its rotamer R1a, that in turn bends into conformer Inta via TS R1a (14.7 S100 kcal/mol). The radical cyclization toward endo-isomer goes from Inta through TS R2a (20.2 kcal/mol). Alternatively, the low-barrier rotation around C16-C17 bond in Inta provides Intb (13.0 kcal/mol), the direct precursor to an exo-isomer R2b. Subsequent radical attack proceeds with a 23.2 kcal/mol barrier, which is 3.0 kcal/mol higher than that of the endo-pathway. This value correlates well with the observed >50:1 endo/exo ratio. Obtained barriers of 20.2 kcal/mol and 23.2 kcal/mol can be converted, through Eyring equation, into rate constants k of ca. 1·10 -2 and 6·10 -5 s -1 (25 ºC), which are, respectively, 5·10 5 and 9·10 7 times slower than the experimental k value for the standard 6-exo-trig cyclization of 6-hepten-1-yl radical (5·10 3 s -1 , 36 ca. 12.4 kcal/mol barrier). A 13.9 kcal/mol barrier (k = 4·10 2 s -1 ) was calculated for the 6hepten-1-yl radical cyclization by DFT employing our standard computational method, which is similar to the above-mentioned experimental value.

Open and cyclized forms of carbocations A-C
DFT calculations of cations A-C and A´-C´ were performed with water as a solvent.
For all cations two different orientations of the carbamate NCO 2 Me were considered.
For cations A-C two opposite orientations of ester CO 2 Me were calculated, and for cations A´-C´ two orientations of the OMe on the charged furan ring were taken into account. The most stable rotamers are presented below.

Crystallographic data for [(bpy) 2 RuBr 2 ]Br, CCDC 1567828
The asymmetric unit contains a half molecule of a metal complex and a half of a bromide anion. Both show Ci-symmetry. The measured sample is formed by at least four crystals with a ratio of 43:21:20:16. The collected data were processed with TWINABS 40 taking into account overlapping reflections. The structure has no A-level alerts in the checkcif test, and B-level alerts were commented.

Crystallographic data for (+)-12, CCDC 1567836
The sample crystallized in the orthorhombic chiral space group P2 1 2 1 2 1 and the absolute structure could be determined reliably with a Flack value based on Parsons' quotients of 0.06(7) (the Flack parameter value for the correct absolute structure determination should be 0; the inverted structure would give 1). The determination was performed on high resolution data collected with Molybdenumradiation using the previously described methodology. 46 The absolute configuration based on the absolute structure of this compound was determined to be S (C13).

Crystallographic data for (+)-grandilodine C (2), CCDC 1567838
This compound crystallized as a monohydrate. The methoxy group (O2, C20) is disordered in two orientations (ratio 90:10). The sample crystallized in the orthorhombic chiral space group P2 1 2 1 2 1 and the absolute structure could be determined reliably with a Flack value based on Parsons' quotients of 0.04(10) (the Flack parameter value for the correct absolute structure determination should be 0; the inverted structure would give 1). The determination was performed on high resolution data collected with Molybdenum-radiation using the previously described methodology. 46 The absolute configuration based on the absolute structure of this compound was determined to be R (C7), R (C13), R (C16), R (C17).

Crystallographic data for (-)-(25S2), CCDC 1567839
The asymmetric unit contains two molecules of 25S2, one of them disordered over two positions with an occupation ratio of 75:25 while the other one has a disordered fragment with a 90:10 occupation ratio, and a half molecule of acetone, which is disordered over three positions with a 25:15:10 occupation ratio. The absolute structure could be determined reliably with a Flack value based on Parsons' quotients of -0.013(8) (the Flack parameter value for the correct absolute structure determination should be 0; the inverted structure would give 1). The absolute configuration of both molecules of 25S2 in the asymmetric unit is the same in all disordered positions and was determined to be R (C4), R (C7), R (C8), S (C12).           -10  0  10  20  30  40  50  60  70  80  90  100  110  120  130  140  150  160  170  180  190  200  210 f1 (ppm) 25