Titanocene(III)-catalyzed conversion of N-(epoxyalkyl)anilines into indolines

Densely substituted indolines and azaindolines can be obtained by the titanocene(III) chloride catalyzed reductive opening of N -(oxiran-2-ylmethyl)anilines. The reaction optimization, substrate scope, and limitations are discussed, and a mechanistic pathway for the epoxide-opening rearrangement is proposed.


Introduction
The indole scaffold is the third most common aromatic building block present in bioactive molecules. 1Frequently, indoles are constructed from partially reduced precursors, such as indolines or oxidized derivatives, such as indolinones, and the latter are also often by themselves biologically active. 2 Furthermore, in the pharmaceutical industry, nitrogen replacements of carbon atoms in the indole scaffold are routinely applied for the investigation of structureactivity relationships (SAR), and indazole and azaindole (pyrrolopyridine) subunits are often found in the molecular structure of therapeutic candidates. 3Therefore, both the discovery of new methods as well as the optimization of the scope of well established protocols for the construction of these heterocycles continues to be a focal point of heterocyclic chemistry.We recently developed a synthesis of 3,3-disubstituted indolines that uses in situ generated titanocene(III) chloride 4,5 to promote a new epoxide-opening rearrangement, transforming N-(epoxyalkyl)anilines into indolines. 6Although most one-step protocols to prepare indolines through carbon-carbon bond formation involve an activation of an aryl halide using either a transition metal 7 or a radical initiator, 8 we envisioned exploring a complementary approach toward indoline preparation that uses the reductive opening of an epoxide derived from an allyl When lowering the reaction concentration to 0.03 M and using 10 mol% of precatalyst in the presence of 80 mol% of manganese powder under sonication, a 3:1 ratio of indoline 2a to tetrahydroquinoline 3a was observed by GC analysis of the crude reaction mixture (Table 1, entry 1).Next, we tested if a methyl substituted alkene substrate would react more selectively and avoid the formation of the undesired mixtures of 2a and 3a.Indeed, when epoxide 1b was treated with in situ generated titanocene (III) chloride, indoline 2b was obtained in 82% yield as a single product (entry 2).It was evident that the additional methyl group on the epoxide improved the selectivity in favor of the desired indoline product.We also observed that sonication provided no additional benefits toward promoting the annulation reaction since 2b was isolated in nearly identical yield when using conventional magnetic stirring (entry 3).To broaden the scope of the methodology, we screened for a suitable replacement for the N-phenyl group.When subjecting the secondary amine (R 1 =H) to the cyclization reaction, only a minor amount of reduced amino alcohol was isolated.The benzyloxycarbonyl (Cbz) group was the preferred choice of protecting group among the candidates screened, which included benzyl, tbutyloxycarbonyl, p-toluenesulfonyl, and trifluoroacetyl.The Cbz protecting group was optimal since it neither functioned as a radical acceptor nor participated in undesired side reactions with the tethered epoxide during the reaction.
Subjecting carbamate 1c to catalytic titanocene(III) chloride at room temperature produced a 2:1 mixture of indoline 2c to alcohol 4c (Table 1, entry 4).When the precatalyst loading was lowered to 5 mol% and then further to 3 mol%, the product ratio (determined by 1 H NMR analysis of the crude reaction mixtures) of 2c:4c increased from 3:1 to 7:1, respectively.However, when using 3 mol% of the precatalyst, 1c was recovered in 18% yield.In order to drive the reaction further to completion, the mixture was heated at reflux in THF using 3 mol% of the precatalyst.Under these conditions and subsequent hydrogenolysis of the Cbz group, indoline 7c was isolated in 63% yield over 2 steps (entry 5).Subjecting the ethoxycarbonylprotected substrate 1d to the optimized conditions afforded the desired indoline in 65% yield (entry 6).When applied to our model system 1b, these conditions produced indoline 2b in 89% yield (entry 7).High yields were also obtained when subjecting epoxide 1a to 3 mol% of titanocene dichloride and stoichiometric Mn powder in THF at reflux, affording a 3.8:1 mixture of 2a and 3a in 87% yield (entry 8). a For epoxide preparation, see experimental section.b All reactions were performed in degassed THF at reflux unless otherwise noted.c Yield was not determined; product ratio was determined by GC analysis of crude reaction mixtures.d Reaction was performed in degassed THF at room temperature using sonication.e Reaction was performed in degassed THF at room temperature using magnetic stirring.f Reaction was performed at room temperature; yield was not determined; product ratios were obtained by 1 H NMR analysis of crude reaction mixtures.
g Yield was determined over 2 steps.We also explored the sensitivity of the reaction to both ambient oxygen and water.When subjecting 1c to titanocene(III) catalysis in a reaction flask open to ambient air, we observed only a 14% yield of 2c and recovered 43% of 1c (Table 2, entry 1).This observation suggested that the desired reaction was inhibited by ambient oxygen.Subjecting 1d to the titanocene(III) catalyzed epoxide opening using a degassed mixture of distilled THF:H2O in an equivolume ratio led to the recovery of 89% of the starting material (entry 2).Although some titanocene(III) chloride-mediated processes utilize water as a co-solvent, 9 the desired catalytic process appears to be inhibited by its presence.Additional control experiments using either only 3 mol% of the precatalyst (entry 3) or only 1.5 equiv. of manganese metal (entry 4) under otherwise identical reaction conditions failed to afford indolines.These results indicated that the in situ generated reagent was responsible for promoting the annulation.In the presence of a catalytic amount of manganese metal, 2c was isolated in 35% yield, in addition to 38% of recovered 1c (entry 5).
For further substrate variation, a series of substituted N-(epoxyalkyl) anilines were prepared and subjected to titanocene(III) chloride catalysis to afford the respective indolines (Table 3).The para-and ortho-methyl substituted substrates 8a and 8b (entries 1 and 2, respectively) afforded indolines in good to modest yields.The 5-methoxyindoline 9c was isolated in low yield (21% yield over 2 steps).In contrast, electron-deficient substrates underwent the epoxideopening rearrangement to afford indolines in good yields (entries 4 and 5).For the study of the regioselectivity of the annulation, epoxide 12 (prepared in 4 steps from aniline 10) was subjected to the titanocene(III) catalyzed epoxide opening, affording a 1:1 mixture of regioisomers 13 and 14 in a combined yield of 63% (Scheme 2).Subsequent reduction of the ethyl carbamates using lithium aluminum hydride provided N-methyl indolines 15 and 16, respectively.

Scheme 2. Regioselectivity of the radical annulation reaction.
We also attempted to prepare heterocycles of higher complexity (Scheme 3).Epoxide 18 was obtained in 61% yield over 3 steps by alkylation of tetrahydroquinoline 17 with 3-bromo-2methyl propene, followed by OsO4 oxidation and epoxide formation.Subjecting the epoxide to the titanocene(III) catalyzed annulation and protection of the resulting alcohol as the acetate provided 19 in 69% yield over 2 steps.In contrast, the epoxide 21 (prepared in 4 steps from indole 20) failed to provide the pyrroloindoline 22, possibly due to the ring strain in the fused five-membered ring system.The linearly fused tricyclic hydrocarbazole 25 was obtained, albeit in low yield, from cyclohexene oxide 24.The alcohol function in 25 was converted to the carbonyl group using Dess-Martin periodinane (DMP) in 63% yield.The relative configuration of ketone 26 was assigned to be syn based on the strong nOe between the tertiary methyl group and the methine hydrogen observed in the 2D-NOESY NMR.Scheme 3. Conversions of more highly substituted substrates to tricyclic indolines.
As mentioned previously, azaindolines have found broad applications as bioisosteres of indolines in pharmaceutical research. 10The use of aminopyridine substrates offered the possibility of applying our methodology toward the preparation of azaindolines. 6,11Although the chemoselective epoxidation of alkenes in the presence of the pyridine moiety was precedented, attempts to efficiently prepare the epoxide on an unsubstituted aminopyridine substrate were unsuccessful.This negative result was primarily due to the reactivity of the pyridine nitrogen toward either m-CPBA or DMDO, i.e. the formation of N-oxides. 12In contrast, the orthochlorine substitution 13 proved to be sufficient to attenuate the reactivity of the pyridine nitrogen lone pair and prevent this undesired oxidation.A Curtius rearrangement of the known carboxylic acid 27 14 followed by subsequent trapping of the intermediate isocyanate with benzyl alcohol afforded the Cbz-protected aminopyridine 28 in 33% yield over 3 steps (Scheme 4).Subsequent methallylation and epoxidation using m-CPBA led to epoxide 29, which, when treated with catalytic titanocene(III) chloride in the presence of stoichiometric manganese powder, provided an intermediate 4,6-dichloro-5-azaindoline.Further conversion of this intermediate with Pd/C under an atmosphere of H2 gave azaindoline 30 in 52% yield over 2 steps.
Our studies of the mechanism of the titanocene(III)-catalyzed conversion of N-(epoxyalkyl)anilines into indolines and azaindolines are still evolving.Based on our current experimental data, we envision that after the in situ generation of titanocene(III) chloride, the epoxide 31 is reductively opened to form the -titanoxy radical 32.This radical may then undergo an addition onto the aromatic ring forming the cyclohexadienyl radical intermediate 33 (Scheme 5). 15Oxidation or disproportionation of the cyclohexadienyl radical followed by proton loss affords indoline 34. 16It is unlikely that the oxidation of the cyclohexadienyl radical could be occurring through the precatalyst since using 15 mol% of manganese metal in the epoxideopening rearrangement affords only 35% of the desired indoline (Table 2, entry 5).To complete the catalytic cycle, protodemetallation by collidinium hydrochloride (Coll•HCl) leads to product 35, and regenerates the titanocene precatalyst.Alternatively, the titanocene (III) chloride catalyst or the manganese chloride byproducts may serve as Lewis acids to promote an epoxide-opening Friedel-Crafts alkylation.However, 5-membered benzene annulations by epoxide opening under typical Friedel-Crafts conditions are quite inefficient. 17Furthermore, it has been shown that the manganese chloride produced in the reaction is incapable of promoting Lewis acid mediated epoxide-opening rearrangements.4c

Experimental Section
General.All moisture-sensitive reactions were performed under an atmosphere of dry nitrogen unless otherwise noted.All glassware was dried in an oven at >140 o C or flame-dried under an atmosphere of dry nitrogen unless otherwise noted.Diethyl ether was dried by distillation over sodium/benzophenone under an argon atmosphere.Tetrahydrofuran was dried either by distillation over sodium/benzophenone under an argon atmosphere or by distillation over LiAlH4 under a nitrogen atmosphere.Dry CH2Cl2 and toluene were purified by filtration through an activated alumina column.Unless otherwise stated, solvents and reagents were used as purchased without further purification.Benzyl alcohol was distilled prior to use.Collidine hydrochloride was recrystallized from absolute ethanol.Manganese and zinc metals were activated through washing with 1 M HCl, followed by rinsing with acetone and drying under vacuum.Titanocene dichloride was recrystallized from chloroform before use.Analytical thin-layer chromatography (TLC) was preformed on pre-coated silica gel 60 F254 plates (250 m layer thickness).Visualization was accomplished by using either a 254 nm UV lamp or by staining with a PMA solution (5 g of phosphomolybdic acid in 100 mL of 95% EtOH), p-anisaldehyde solution (2.5 mL of p-anisaldehyde, 2 mL of AcOH, and 3.5 mL of conc.H2SO4 in 100 mL of 95% EtOH), Vaughn's reagent (4.8 g of (NH4)6Mo7O24•4 H2O and 0.2 g of Ce(SO4)2 in 100 mL of a 3.5 N H2SO4) or a potassium permanganate solution (1.5 g of KMnO4 and 1.5 g of K2CO3 in 100 mL of a 0.1% NaOH solution).NMR spectra were recorded at room temperature in CDCl3 at 300 MHz/75MHz ( 1 H/ 13 C NMR) using a Bruker Avance 300 MHz spectrometer unless stated otherwise.Chemical shifts () are reported in parts per million and referenced from the residual solvent peak or tetramethylsilane.Data are reported as follows: chemical shift, multiplicity (app s = apparent singlet, s = singlet, bs = broad singlet, d = doublet, bd = broad doublet, dd = doublet of doublets, ddd = doublet of doublet of doublets, app t = apparent triplet, bt = broad triplet, t = triplet, dt = doublet of triplets, tt = triplet of triplets, q = quartet, app quint = apparent quintet, m = multiplet), integration, and coupling constant(s).IR spectra were recorded on either a Nicolet Avatar 360 FT-IR E.S.P. spectrometer (KBr or neat) or a Smiths Detection IdentifyIR FT-IR spectrometer (ATR).Melting points were uncorrected and determined using a Laboratory Devices Mel-Temp II.Mass spectrometry data were recorded on a VG-70-70 HF instrument.Titanocene-catalyzed radical cyclization reactions were performed under rigorous exclusion of dioxygen under a positive pressure of dry argon.Tetrahydrofuran, in addition to being distilled, was deoxygenated (freeze-pump-thaw) three times and then stored under a positive pressure of dry argon prior to all titanocene(III) catalyzed annulation reactions.Sonication was accomplished with the Sonics Vibracell device (model VCX 130) equipped with a 2 mm microtip.

Benzyl (2-methyloxiran-2-yl)methyl(phenyl)carbamate (1c). General protocol E
To a solution of 2.91 g (12.8 mmol) of benzyl phenylcarbamate in 60 mL of THF at 0 °C was added 1.02 g (25.6 mmol) of NaH (60% dispersion in mineral oil).The reaction mixture was warmed to rt over 15 min and 2.6 mL (25.6 mmol) of 3-bromo-2-methylpropene were added.The solution was stirred overnight, quenched with H2O and extracted with 3 x 20 mL of Et2O.The combined organic layers were dried (MgSO4) and concentrated in vacuo.The residue was dissolved in 80 mL of dichloromethane, cooled to 0 o C and 5.67 g (23.0 mmol) of m-CPBA (70%) was added portionwise.The reaction mixture was quenched with aq.Na2S2O3 solution and extracted with 3 x 10 mL of Et2O.The combined organic layers were dried (MgSO4), concentrated in vacuo and purified by chromatography on SiO2 (hexanes:EtOAc; 4:1 with 1% NEt3) to afford 3.01 g (

2,6-Dichloroisonicotinic acid (27).
According to a literature procedure, 24 a solution of 8.91 g (43.24 mmol) of methyl 2,6-dichloroisonicotinate in 20 mL of THF was treated with 1.24 g (51.89 mmol) of LiOH in 60 mL of H2O.The reaction mixture was stirred for 20 min at rt and concentrated in vacuo to remove the THF.The resulting solution was cooled to 0 o C and treated with 25 mL of 2 M HCl solution.After 2 h, the solid was filtered and dried to afford 6.05 g (31.Benzyl 2,6-dichloropyridin-4-ylcarbamate (28). 24To a solution of 1.00 g (5.20 mmol) of 27 in 20 mL of THF was added 540 μL (6.35 mmol) of oxalyl chloride at rt.The reaction mixture was refluxed for 2 h, cooled to rt and concentrated in vacuo.The resulting oil was dissolved in 40 mL of freshly distilled acetone, cooled to 0 o C and treated dropwise with a solution of 1.01 g (15.62 mmol) of NaN3 in 20 mL of H2O.The mixture was stirred for 90 min, and the temperature was allowed to increase from 0 o C to 10 o C. The mixture was diluted with 15 mL of distilled Et2O, and the aqueous layer was extracted with 2 x 10 mL of distilled Et2O.The combined organic layers were dried (MgSO4) and concentrated to ~10% volume.After addition of 10 mL of toluene the remaining Et2O and acetone were removed under reduced pressure.The residue was dissolved in 15 mL of toluene, stirred with MgSO4 and treated with 1.1 mL (10.41 mmol) of benzyl alcohol.The mixture was heated at reflux for 15 h behind a blast shield, cooled to rt, diluted with water and extracted with 2 x 10 mL of EtOAc.The combined organic layers were dried (MgSO4), concentrated and purified on SiO2 (hexanes:EtOAc; 20:1 to 10:1 gradient) to afford 580 mg (1.95  of 28 in 10 mL of THF at 0 o C was treated with 34 mg (0.09 mmol) of TBAI and 93 mg (3.68 mmol) of NaH (95%).The reaction mixture was stirred for 5 min, treated with 370 L (3.68 mmol) of 3-bromo-2-methylpropene, warmed to rt and stirred for 16 h.After addition of 160 mg of NaH and 400 μL of 3-bromo-2-methylpropene, the mixture was stirred until starting material was consumed according to TLC (hexanes:EtOAc; 4:1).The solution was then cooled to 0 o C, quenched with water and extracted with 3 x 10 mL of Et2O.The combined organic layers were dried (MgSO4), concentrated in vacuo and purified on SiO2 (hexanes:EtOAc; 15:1 to 10:1 gradient) to afford 497 mg (1.41   29).To a solution of 429 mg (1.22 mmol) of benzyl 2,6-dichloropyridin-4-yl(2-methylallyl)carbamate in 10 mL of dichloromethane at 0 o C was added 451 mg (1.83 mmol) of m-CPBA (70%).The reaction mixture was warmed to rt and after 6 h an additional 1 equiv (300 mg) of m-CPBA was added.The mixture was stirred for a total of 11 h, cooled to 0 o C, quenched with aq.Na2S2O3 solution and extracted with 3 x 10 mL of dichloromethane.The combined organic layers were dried (MgSO4), concentrated and purified on SiO2 (hexanes:EtOAc; 8:1) to afford 371 mg (1.01 mmol, 83%) of 29 as a colorless oil: IR (ATR) 3089, 1714, 1576, 1535, 1216, 1149, 1088 cm -1 ; 1 H

Table 1 .
Conditions screened during the optimization of the epoxide-opening rearrangement

Table 2 .
Control experiments to probe reaction limitations aReaction was performed in a flask open to the atmosphere in distilled, non-degassed THF.b Reaction was performed in a degassed mixture of distilled THF:H2O (1:1).

Table 3 .
Substrates prepared using titanocene(III) chloride catalysis a See experimental section for epoxide preparations.b Yield was determined over 2 steps.c Cbz group was not removed; i.e. step 2 was not performed.