Aldol derivatives of 5-phenyl-1,4-benzodiazepin-2-on-N 4 -oxide; intriguing inertness of N-oxides in aldol reactions

In an attempt to prepare 3-substituted 7-chloro-1,3-dihydro-1-methyl-5-phenyl-2H -1,4- benzodiazepin-2-on-N 4 -oxides 7-10 , C(3) carbanion of 5-phenyl-1,4-benzodiazepin-2-on-N 4 - oxide ( 2 ) proved completely inert in aldol reaction: Detour to the target compounds 7-10 via aldols 3-6 was required. Unexpected inertness of C(3) carbanion of 2 was attributed to the high charge delocalization.


Introduction
In the course of the study of stereoselective aldol reaction of 5-phenyl-1,4-benzodiazepin-2-one 1 and its 5-pyrido-analogue 2 with aromatic and aliphatic aldehydes, we have entered the preparation of the N 4 -oxides of diastereomerically pure aldol products 7-10 as potential ligands for catalytic organometallic complexes.A number of reports appeared on successful application of N-oxides as ligands in catalytic C-C; C-O, C-S and C-H bond forming reactions, such as allylation of aldehydes, 3,4 aldol reactions of ketones, 5 cyclopropanation of styrene, 6 epoxidations, 7 oxidation of alkenes to diols, 8 desymmetrization of epoxides, 9 addition of thiols to enones, 10.11 rearrangement of thiones to thiols. 12Chiral β−hydroxy-N-oxides catalyse the enatioselective borane reduction of ketones. 13

Results and Discussion
Two synthetic alternatives to the target compounds 7-10 were considered, depending whether the N-oxidation is performed before or after aldol reaction, Scheme 1, paths A and B. We have selected the first approach in view of a. well known, technical-scale N-oxidation of N-demethyl analogue of 1,4-benzodiazepine 1 in the production of therapeutically important 3hydroxy derivative 14 (Praxiten®, generic name Oxazepam), b. the observation that in the 1 H-NMR spectra of 1 and 2 AB system of C(3)H 2 protons is centered at 4.31 ppm and 4.64 ppm, respectively, and in their 13 C-NMR spectra C(3) carbon appears at 56.59 ppm and at 67.68, respectively.This reveals strong electron-deshielding effect of the N-O group and consequently higher C(3)-H acidity in 2. The carbanion of N 4 -oxide 2 is expected more convenient for the aldol reaction then carbanion of benzodiazepine 1, similarly as the S-oxide function in I is reported by Cadoni et al. 15 to promote generation of vicinal carbanion that affords in high yield the aldol products II.To our surprise, all attempts to complete an aldol reaction with the carbanion of 2 failed.Formation of carbanion on addition of a strong base to the THF solution of 2 can be followed by the appearance of an intense orange-red color.On addition of aldehyde this color persists and no formation of aldol product can be traced by HPLC.To the contrary, coloration disappears in few minutes at -70 o C when aldehydes are added to the solution of carbanion of 1. 1¸2 At the temperatures around 40 o C decomposition of 2 is observed.To achieve our synthetic target, the pathway B in the Scheme 1 was then followed; preparation of diastereomerically pure syn and anti 7 and 8 was completed by N-oxidation of 3 and 4, without any loss of stereochemical integrity, as controlled by HPLC and 1 H-NMR.In order to test separability of the aldol products of N 4 -oxides by crystallization, their deoxo-analogs can be separated only by chromatography, 1,2 diastereomeric N 4 -oxides 9/10 were prepared from 4.0:6.0mixture of 5/6.Diastereomeric ratio remained in the product mixture but separation by crystallization failed; the products can be completely separated by chromatography.In order to explain the failed aldol reaction of N 4 -oxide 2, and to trace eventual side-products, additional experiments were performed.First, stability of the aldol products was checked by attempting a retro-aldol reaction of diastereomeric mixture 9/10.HPLC monitoring has revealed their complete stability at -70 o C; on gradual elevation of temperature only syn diastereomer 9 has returned to 2. In the separate experiments with syn-7 and anti-8, slow splitting of 7 to 2 and benzaldehyde at temperatures between ambient and 50 o C ca be traced, whereas 8 proved stable under the same conditions.

V
This experiment eliminated retro-aldol reaction of the aldol products of 2 as the origin of the synthetic failure.Large difference in the stability of syn and anti diastereomers under basic conditions can be explained by different 6-membered chelate rings they form with a lithium cation.In the syn diastereomers chelation involves carbonyl oxygen and places the large aryl group in pseudoequatorial position, 1,2 whereas in the anti-diastereomers aryl group adopts pseudoequatorial conformation on chelation to the N-oxide oxygen atom.The former enolates only can undergo retro-aldol reaction, which is inhibited for the latter ones, however.
Aldol reaction of 2 was then monitored by HPLC under conditions that allow identification of benzoin, as the product of dimerization of benzaldehyde catalyzed by the carbanion of 2, and eventual other side-products.No traces of benzoine have been identified, what excluded catalytic activity of the betaine-like carbanion of 2. 16 Assuming high delocalization of the negative charge to the carbonyl and the N 4 -oxide oxygen atoms, we envisaged formation of hemiacetale-like product III or acetal-like polycyclic product IV; its carbon analog V was reported as the addition product of N-oxide 2 to acrylates. 17No such side products were identified in the reaction solution.Besides, an attempt to trap the carbanion of 2 by benzylbromide has also failed.

2A 2B 2C
To explain difference in the reactivity of the carbanions of the compounds 1 and 2 we have assumed strong charge delocalization of the, lowering the electron density on the C(3) carbon.Charge-separated canonic structures have been invoked as the origin of lower inversion barrier for the 7-membered ring in N 4 -oxide 2. 18 Charge separation in the canonic structures 2A-2C of carbanion favors delocalization of the negative charge to the oxygen atom (2B, 2C); "soft" electronic nature of delocalized carbanion makes it inert in the attempted aldol reaction.

Experimental Section
General Procedures.IR spectra were run on Perkin Elmer 297 spectrometer for KBr pallets.C-NMR spectra were obtained with Varian Gemini XL 300 spectrometer in CDCl 3 , δ in ppm is relative to TMS as internal reference, and J in Hz.HPLC chromatography was performed on HP 1050 chromatograph with Nucleosil C18 RP column, separation was monitored by HP 1050 UV detector set up at 254 nm and connected to HP 3396A integrator.M.p.'s were determined on Electrothermal Apparatus, and are not corrected.