The stereochemistry of the addition of chlorotitanium enolates of N-acyl oxazolidin-2-ones to 5- and 6- membered N-acyliminium ions

The stereoselective addition of chiral and achiral titanium enolates derived from the corresponding N-acyl oxazolidin-2-ones to 5- and 6- membered N-acyliminium ions afforded 2-substituted pyrrolidines in moderate to good diastereoisomeric ratio (5:1 to 14:1) while lower diastereoselection was generally observed in the formation of the corresponding 2-substituted piperidines. The stereochemical outcome was found to be modulated by the nature of the cyclic N-acyliminium ion (5- or 6-membered) and of its carbamate and by the N-acyl group in the enolate precursor. The preferential lk approach seems to be dictated mainly by the minimization of non-bonding interactions between the N-acyl group in the chlorotitanium (IV) enolate and the carbamate and methylene groups in the cyclic N-acyliminium ion.


Introduction
The condensation of a carbonyl compound with an amine followed by the addition of a carbon nucleophile to the intermediate iminium species, known as the Mannich reaction, is one of the classical methods for the synthesis of b-aminocarbonyl compounds and nitrogen-containing heterocycles. The use of preformed iminium salts and carbon nucleophiles such as metal enolates, silyl enol ethers, silyl keteneacetals and enamines has greatly expanded the versatility of this reaction allowing the use of milder reaction conditions and the introduction of elements of regioselective control 1 .
Despite the similarity with the aldol reaction much less is known about the structural features controlling the stereochemical outcome of the addition of prochiral carbon nucleophiles to imines or iminium ions when compared to the corresponding addition to aldehydes. Evans and co-workers put forth a topological analysis for the addition of metallic enolates to imines based on the coordination of the nitrogen lone pair to the metallic species giving rise to chelated transition states which may adopt either chairlike or boat-like geometries depending on the interplay of steric and electronic interactions 2 .
The utilization of N-acyliminium ions has attracted much attention particularly in the intramolecular version of the reaction due to their enhanced electrophilic character and the reduced bias towards Grob fragmentation displayed by the corresponding b-aminocarbonyl derivatives 3 . An open transition state with an antiperiplanar approach of carbon nucleophiles to the electrophilic center of the iminium ion has been proposed 4 .
Inspired by the early work of Fuentes and co-workers who first revealed the feasibility of the addition of boron enolates of chiral oxazolidin-2-ones to chiral 4-acetoxy-2-azetidinones 5 and by the work of Nagao and co-workers who described the addition of tin (II) enolates of chiral 3acyl-1,3-thiazolidine-2-thiones to 4-acetoxy-2-azetidinone and 5-acetoxy-2-pyrrolidinone 6 , we were attracted to study the effect of the ring size and the nature of the carbamoyl group of the N-acyliminium ion in the reactivity and stereochemical outcome of the addition of boron and chlorotitanium (IV) enolates of oxazolidin-2-ones to prochiral cyclic N-acyliminium ions expecting that diastereoselection would be attained due to the known facial discrimination displayed by the enolates of oxazolidin-2ones 7 . Additionally, the results were expected to be of potential interest in the asymmetric synthesis of nitrogen heterocycles such as pyrrolizidine, indolizidine and quinolizidine ring systems 8 .

Results and Discussion
Initially, we investigated the boron enolate addition of achiral oxazolidin-2-one 1 to 2-ethoxypyrrolidine 5, prepared by sodium borohydride reduction of an ethanolic solution of the corresponding lactam 9 . The boron enolate of achiral oxazolidin-2-one 1 was generated in CH 2 Cl 2 at 0 °C upon treatment with n-Bu 2 BOTf, according to the procedure by Evans and co-workers 10 , followed by the addition of 5 (1.0 equiv.) and an additional equivalent of n-Bu 2 BOTf at 0 °C. The diastereoisomeric ratio for (+/-)-7:(+/-)-8 was shown to be 13:1 after inspection of the 1 H NMR spectrum (300 MHz, 55 °C) of the crude product which displayed two doublets at d 1.13 and d 1.18 ppm for the methyl group at C-1' (Scheme1) 11 . N-Boc-2substituted pyrrolidine (+/-)-7 was isolated in 50% yield after column chromatography on silica gel. In comparison, the coupling of the N,O-silylketeneacetal derived from Npropionyl oxazolidin-2-one 1, prepared in situ through the addition of 1.2 equiv. of TMSOTf and 1.15 equiv. of Et 3 N in CH 2 Cl 2 at 0 °C, with 2-ethoxypyrrolidine 5 was also carried out but afforded a 2:1 mixture of (+/-)-7 and (+/-)-8, in 45% yield (Table 1, entries 1 and 2).
Much to our surprise, the experimental protocol described above for the addition of the boron enolate derived from 1 to 2-ethoxypyrrolidine 5 did not provide the corresponding N-Boc-2-substituted piperidines (+/-)-9:(+/-)-10 when 2-ethoxypiperidine 6 was added to a CH 2 Cl 2 soln. of the boron enolate of oxazolidin-2-one 1, which was recovered in almost quantitative yield, even when the reaction was carried out at room temperature and for longer reaction period. However, the reaction of 2ethoxypiperidine 6 with the N,O-trimethylsilylketeneacetal of N-propionyl oxazolidin-2-one (1) did proceed to afford a 2:1 mixture of the 2-substituted piperidines corresponding to (+/-)-9 and (+/-)-10, in 36% non-optimized yield after in situ N-Boc deprotection ( Table 1, entries 4 and 5).
The same reactivity pattern emerged when we employed the boron enolate derived from chiral oxazolidin-2-one ent-2: 1 H NMR analysis of the crude mixture revealed that a single 2-substituted pyrrolidine was formed from 5 but no reaction was observed with 2-ethoxypiperidine 6 (Scheme 1, Table 1, entries 7 and 8). N-Boc pyrrolidine (+)-11 was isolated in 55% yield after column chromatography on silica gel and had its structure established by X-ray diffraction analysis 7b .
As we were facing some difficulties reproducing the yields in the reactions with boron enolates 12 , we decided to examine the behavior of the chlorotitanium(IV) enolates of oxazolidin-2-one 1 and 2 in the presence of cyclic Nacyliminium ions. Upon addition of a CH 2 Cl 2 soln. of 2ethoxypyrrolidine 5 to a previously formed solution of the chlorotitanium (IV) enolate corresponding to 1 in CH 2 Cl 2 at -23 °C 7b , a gradual fading of the deep burgundy color of the enolate soln. was observed. 1 H NMR analysis (in CDCl 3 at 55 °C) of the crude product revealed that 2-substituted pyrrolidines (+/-)-7 and (+/-)-8 were produced in a 14:1 ratio (Scheme 1, Table 1, entry 3). Purification by column chromatography on silica gel gave (+/-)-7 in 72% yield. As before, no reaction took place when 2-ethoxypiperidine 6 was employed at -23 °C or at 0 °C (Table 1, entry 6).
At this point, it was evident that the ring size and the nature of nucleofile were modulating the reactivity and the stereochemical outcome of the reaction and an evaluation of the impact of the carbamate group of the N-acyliminium ion on the reaction course seemed in order. The reaction with the chlorotitanium (IV) enolate of achiral oxazolidin-2-one 1 proceeded with moderate to good yields with N-Cbz and N-CO 2 Me pyrrolidines 21 and 23 and piperidines 22 and 24, but with good diastereoselection only in the pyrrolidine series (Scheme 2, Table 2, entries 1-4). The same trend was observed when the enolate of chiral oxazolidin-2-one 2 (Scheme 2, Table 2, entries 5-8) was employed and the best diastereoselection was achieved with N-Boc-2-ethoxypyrrolidine 5 (table 1, entries 1 and 4).
Inspection of the 13 C NMR spectra in the 2-substituted piperidines series revealed a downfield shift of C-2 in the major isomers when compared with the minor ones (Dd 1.4-1.7 ppm) which may be diagnostic of their relative stereochemistry. The absolute configuration of 39:40 was confirmed after conversion of a 1.8:1 mixture of (-) -35 and (-)-36 of known configuration to the corresponding mixture of 39:40 (i. H 2 , Pd/C, MeOH, rt; ii. MeOCOCl, K 2 CO 3 , acetone, rt) and comparison by HPLC analysis.
In order to extend the study of the influence of the nature of the nucleophile in the stereochemical outcome of the reaction, chlorotitanium (IV) enolates derived from oxazolidin-2-ones 52 and 53 were employed (Scheme 9). Our choice was additionally guided by the potential usefulness of the corresponding adducts in the asymmetric synthesis of (2R, 2'R)-methylphenidate hydrochloride 7c and in the total synthesis of pyrrolizidine and indolizidine alkaloids 6 .
As depicted in Table 3 the formation of 2-substituted pyrrolidine derivatives occurred with excelent diastereoselection (>95:5, entries 1, 4 and 5) while moderate to good selectivity was observed in the piperidines series (entries 2, 3 and 6), superior to the diastereoisomeric ratio observed with chlorotitanium enolates from oxazolidin-2-ones 1-4 ( Table 1 and 2). The absolute configuration for 54, 60 and 62 was tentatively assigned as (2R, 1'R) based on our previous results for the addition of chlorotitanium enolates derived from 2 to a-alkoxy pyrrolidines 5, 21 and 23.
Inspection of the 13 C NMR spectra of the mixture 64:65 revealed that the major isomer 64 displayed its C-2 shift downfield (1.4 ppm) in comparison with the minor isomer 65, as observed earlier for 2-substituted piperidines depicted in Table 2. Based on that evidence (2R, 1'R) and (2R, 1'S) configuration were assigned to 64 and 65, respectively.
The absolute configuration of the majors isomers 56 and 58 was established after their conversion to (2R, 2'R) methylphenidate hydrochloride (66). Methylphenidate is a mild psychostimulant and is widely prescribed for the treatment of attention deficit hyperactivity disorder (ADHD) in children, a condition that is manifested by impulsivity, hyperactivity, and inattention. It is marketed as its racemic form despite the fact that the 2R, 2'R-isomer is several times more active than the corresponding enantiomer [14][15][16] .
Alternatively, basic hydrolysis of 58 allowed recovery of the chiral auxiliary (90% yield), and afforded carboxylic acid 69 in 90% yield. Carbamate deprotection was accomplished with in situ prepared trimethylsilyl iodide and it was followed by esterification and treatment with ethanolic HCl to provide (2R, 2'R)-methylphenidate hydrochloride 66 in 63% yield from 69.
The absolute configuration (2S, 1'R) for the minors isomers 57 and 59 was assigned based on our previous studies (Scheme 2, Table 2) in agreement with the results by Matsumura et al 7c .
From the body of information described above it emerges that the oxazolidin-2-one moiety is efficiently discriminating its approach to the N-acyliminium ion: in every case, both major and minor isomers displayed 1'S stereochemistry when (R)-4-benzyl oxazolidin-2-one was employed as chiral auxiliary revealing the preferential approach of the Re face of the corresponding chlorotitanium (IV) enolate. A Z-configured internally coordinated chlorotitanium (IV) enolate, as depicted in Scheme 11, is proposed to be the nucleophilic species in the reactions above and either an antiperiplanar or a synclinal approach of the nucleophile to the N-acyliminium ion seems to be available.
For bulky carbamates such as 5-6 (Scheme 1) the stereochemical outcome will be dictated by the relief of steric interactions involving the Boc group: while this seems to be the case for the antiperiplanar approach to Re face of the N-acyliminium ion (see A, Scheme 11), the synclinal approach to the Re face of the electrophilic species brings Boc and oxazolidin-2-one groups into close proximity (see B, Scheme 11). Despite keeping Boc and oxazolidin-2one groups apart, the antiperiplanar approach of chlorotitanium (IV) enolate to the Si face of the Nacyliminium ion develops non-bonding interactions between Boc and R 2 groups (see C, Scheme 11). A sterically  congested synclinal approach such as depicted in D (Scheme 11) seems to be of only marginal relevance in the reaction manifold due inter alia to steric hindrance involving the titanium (IV) ligands and Boc group. The steric interactions involving R 2 group and cyclic N-acyliminium ion are not, however, to be overlooked (see A and D, Scheme 11) particularly when cyclic 6-membered N-acyliminium ions are involved. The preferred half-chair conformation in these cases will bring about significant non-bonding interactions during the approach of the nucleophile which are partially relieved when more flattened cyclic 5-membered N-acyliminium ions are involved. Not surprisingly, no reaction was observed between the N-acyliminium ion derived from 6 and boron enolates prepared from 1 and (+)-2 or chlorotitanium (IV) enolates derived from 1 and 2, even at room temperature and longer reaction periods.
The reactivity was restored when N-acyl chlorotitanium (IV) enolates derived from 3 and 4 were employed in the reaction with 6 affording 16 (40% yield) and 19:20 (90% yield, 3.5:1 ratio). The higher diastereoselectivity observed in the reaction of 4 and 6 (3.5:1 ratio) when compared to the corresponding reaction with 5 (1:1 ratio) may be due to steric hindrance between the half-chair conformation of 6 and the oxazolidinone ring thus disfavoring approach C (Scheme 11).
According to the transition state model depicted in Scheme 11, utilization of less sterically demanding carbamates such as 21-24 resulted in lower levels of diastereoselection when compared to the ones observed for N-Boc carbamate 5 while keeping the preference for lk topology 17 . The diastereoisomeric ratios observed for the N-CO 2 Me and N-CO 2 Bn 2-alkoxycarbamates did not vary significantly.
The higher levels of diastereoselection observed in the reaction of the chlorotitanium enolate derived from (S)-52 with 22 and 24 contrasts with those found for (R)-2 and point out either to a relief of steric interaction upon changing a methyl group in the chlorotitanium enolate derived from 2 to a planar, conjugated phenyl ring in the enolate derived from 52 (A, Scheme 11) or to a synclinal approach as depicted in B (Scheme 11). In both cases, the lk topology 17 would be greatly favoured resulting in the preferential formation of 56 and 58. Additionally the s-cis/ s-trans orientation of the carbamate group and the stereoelectronics associated with 6-membered Nacyliminium ions may be relevant to rationalize these results.
The stereochemical outcome observed in the reaction of the di-n-butylboron enolate from ent-2 with 2ethoxypyrrolidine 5 (Scheme 1), which occurred with the same lk topology as observed for the chlorotitanium (IV) enolate from 2, do not support the coordination of the Nacyliminium carbamate group to the Z-configured internally coordinated chlorotitanium (IV) enolate, as proposed by Matsumura and co-workers 7c .

Conclusions
The stereochemical outcome of the addition of chlorotitanium (IV) enolates derived from N-acyloxazolidin-2-ones to in situ generated cyclic N-acyliminium ions was shown to be dependent on the nature of the ring size and carbamate group in the N-acyliminium ion and on the Nacyl group in the chlorotitanium (IV) enolate: 5-membered N-acyliminium ions derived from 5, 21 and 23 reacted with the chlorotitanium (IV) enolate derived from N-propionyl oxazolidinones 1 and 2 with moderated to good diastereoisomeric ratio (5:1-14:1) but poor diastereoselection was observed when the N-acyliminium ions generated from 6membered piperidines 22 and 24 were employed. Better yield and seletivities were observed for the BOC protected Nacyliminium ion derived from 5 when compared to the corresponding N-carbobenzyloxy and N-carbomethoxy analogues but N-Boc-2-ethoxy piperidine 6 failed to react with the chlorotitanium (IV) enolates derived from 1 and 2.
The preferential lk approach observed was rationalized based on the minimization of steric interactions involving the N-acyl group in the chlorotitanium (IV) enolate and carbamate and methylene groups in the cyclic Nacyliminium ions.

Experimental Section
Infrared spectra were recorded on a Nicolet Impact 410 spectrometer and 1 H-and 13 C NMR spectra were measured on a Bruker AC-300P (7.0T), VARIAN GEMINI (7.0T) or VARIAN INOVA (11.7T) spectrometers. Chemical shifts are reported in parts per million downfield from tetramethylsilane internal standard. Data are reported as follows: chemical shift, integration, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, qt = quintet, m = multiplet, br = broad) and coupling constants. Chemical shifts of 13 C NMR spectra are reported in ppm from tetramethylsilane using the solvent resonance as internal standard (deuterochloroform: 77.0 ppm, deuteromethanol: 49.0 ppm). NMR data for mixture of diastereoisomers are reported with the chemical shift of the minor diastereoisomer following the one of the major isomer. The optical rotation was measured on a Polamat A (Carl Zeiss) or Perkin Elmer 241 polarimeter. Mass spectra were obtained in the electron impact (EI) mode using an HP-5890-serie II chromatograph equipped with HP-1 column (25m x 0.20mm x 0.33mm) or Ultra 2 column (25m x 0.20mm x 0.33mm) coupled to a HP-5988 mass detector and high resolution mass spectra on a VG Autospec/ Fission Instrument. HPLC analyses were performed with HP1050 chromatograph equipped with Hypersil column (5mm, 200 x 4,6mm). Melting points were measured in open capillary tubes using an Electrothermal 9100 apparatus and are not corrected. Elemental analyses were carried on a Perkin Elmer-2400 series II CHNS/O analyser. Column chromatography was performed with Aldrich silica gel (70-230 mesh) and flash chromatograph with Merck silica gel 60 (230-400 Mesh). Thin-layer chromatography (TLC) was carried out on Alugram â SIL G/UV 254 precoated plates and were developed with potassium permanganate in all cases. Unless otherwise noted, all reactions were conducted in flame-dried glassware with magnetic stirring under an inert atmosphere of dry nitrogen or argon. Solvents and reagents were purified and dried when necessary using standard procedures 18 .

General procedure for the addition of the N,O-silylketeneacetal from 1 to cyclic N-acyliminum ions
To a solution of 1 (0.090 g, 0.62 mmol) in CH 2 Cl 2 (3.0 cm 3 ) at 0 °C was added Et 3 N (0.12 cm 3 , 0.87 mmol), and TMSOTf (0.14 cm 3 , 0.74 mmol) dropwise. The resulting homogeneous solution was stirred for 45 min at 0 °C. The solution was cooled to -78 °C, and the a-alkoxycarbamate 5 or 6 (0.57 mmol) was added, followed by catalytic amount of TMSOTf. After 2 h the reaction was quenched with satd. aq. NH 4 Cl (5.0 cm 3 ) and extracted with CH 2 Cl 2 (3 x 5 cm 3 ). The combined organic layer was dried over MgSO 4 , filtered and the solvent was evaporated under reduced pressure.

General procedure for the addition of the titanium enolates of N-acyloxazolidinones to cyclic N-acyliminium ions
To a solution of TiCl 4 (0.12 cm 3 , 1.1 mmol) in CH 2 Cl 2 (2.0 cm 3 ) at temperature T1 was added dropwise a solution of N-acyloxazolidinone (1.0 mmol) in CH 2 Cl 2 (2.8 cm 3 ). After 5 min., diisopropylethylamine (0.20 cm 3 , 1.1 mmol) was added and the mixture was stirred 1h at temperature T1. The reaction mixture was cooled to temperature T2 and a solution of a-alkoxycarbamate (1.1 mmol) in CH 2 Cl 2 (4.8 cm 3 ) was added dropwise. The mixture was then stirred 1 h at temperature T2 and the reaction was quenched with satd. aq. NH 4 Cl (5.0 cm 3 ) and extracted with CH 2 Cl 2 (3 x 5 cm 3 ). The combined organic layer was dried over MgSO 4 , filtered and the solvent was evaporated under reduced pressure.
The reaction temperatures (T1, T2), purification, yield, spectral and analytical data of the adducts prepared are as follows.