PREPARATION OF BETA-SUBSTITUTED TRYPTOPHAN DERIVATIVES - COMPARISON OF THE REACTIVITY OF N-METHYLINDOLE TOWARD AZIRIDINE-2-LACTONES AND AZIRIDINE-2-CARBOXYLIC ESTERS AND INTERPRETATION OF RESULTS USING MNDO CALCULATIONS

With the aim of preparing novel beta-functionalized tryptophan derivat ives, the reaction of (1S,4S,5R)- ethoxymethyl)-3-oxa-6-azabicyclo[3.1 .0]hexan-2-one (4), a newly developed rigid analogue of the synthetica lly useful aziridine-2-carboxylic esters of type 1, with N-methylindol e (6) was studied under acidic (Lewis acid) conditions. N-Methylindole reacted exclusively at C-2 of 4 to give C-[3-(1-methylindolyl)]-5-O-m ethyl-D-xylonolactone (7) in contrast to this nucleophile's known reac tivity with aziridine-2-carboxylic esters 1 at C-3 under the same cond itions. The desired beta-substituted tryptophan derivative 12 was inst ead obtained by reacting 6 with the tert-butyldimethylsilyl furanoside precursor of 4 (i.e., 9) followed by desilylation and oxidation of th e anomeric hydroxyl function with TPAP. The regioselectivity of azirid ine ring opening by 6 was rationalized by comparison of the LUMO coeff icients and atomic charge distributions for the reactive centers of th e aziridine-2-lactone 4, the aziridine-2-carboxylic ester 16, and the aziridine furanoside 9 in both their ground states and protonated stat es as determined using MNDO calculations. It was found that (1) proton ation of both 4 and 16 causes a large increase in the LUMO coefficient s at C-2 and C-3, thereby directing attack by N-methylindole (6), a so ft nucleophile, toward these centers via orbital control, as has been experimentally observed; (2) of C-2 and C-3, the higher LUMO coefficie nt was found at the former position for the N-protonated forms of both 4 and 16, suggesting that C-2 is the preferred site of attack by 6 in both cases. Though this was verified experimentally in the case of la ctone 4, the fact that aziridine-2-carboxylic esters (e.g. 16) always react with indoles at C-3 under acidic conditions indicates that in th ese compounds, steric and/or electrostatic effects rather than orbital considerations determine the course of the reaction; (3) in the case of the N-protonated aziridino furanoside 9, C-3 was calculated to have a higher LUMO coefficient than C-2, in accord with the exclusive atta ck of 6 at the former position. MNDO calculations thus appear to be a useful tool for the prediction of the reactivity patterns of rigid azi ridine structures such as 4 and 9, but are less satisfactory in the ca se of flexible aziridine-2-carboxylates in which other factors may pre dominate.