Table 1. Enantioselective synthesis of dihydroindoles 2 from azomethines 1 [a]. Cation: Li”.

Entry Azo- Catalyst [b] Prod. Yield ee Sign of methine [%] [%I rotation

[cl [dl (CHCls)

1 l a 4b L=CH1 2a 53 < I 0

3 [el l b L = H 2b 57 <10 4 l b 4b L=CHs 2b 50 >95 (-) 5 l b 5b L=CH3 2b 50 >95 (+)

7 l b 4~ L=C2Hs 2b 64[c] 73 (-)

2 l b 48 L = H 2b 51 12 (+)

6 [el l b L=CH, 2b 30 20 (+)

8 l b 46 L=CHzCH20CH, Zb 60 38 ( -1 9 l b 4e L=CH2Ph 2b 48 26 (-)

11 l c 4b L=CH, 2c 70 37 (-) 12 ld 4b L=CH3 2d 57 50 (-)

10 l b 4f L= COPh Zb 44 <10

[a] Experimental procedure: [7]. [b] 4a and 50 are commercially available; 4b-f and 5b were synthesized from 4a and 5a by either reductive methyla- tion (4b, 5b) [9] or by acylation (4f), followed by LiAlH4 reduction (4c-e). [c] Isolated yields except for entry 7, for which the ‘H-NMR yield is given. [dl Determined by ’H-NMR spectroscopy using Eu(hfc), [5, 7). [el Cation: Na”.

Mechanistically, this ring-closure reaction must be viewed as a disrotatory 1,5-electrocyclization[3~’01. To ac- count for the formation of the cis product we assume that ring closure occurs from the Z-azomethine geometry and that the imide enolate has the enolate oxygen directed to- ward the azomethine, both moieties being arranged in a helix-type conformation. The favorable influence of the hydroxy group is most probably the result of hydrogen bonding with the azomethine nitrogen (0-H . . . N) in the transition state. Furthermore, in the highly structured tran- sition state the tertiary amine function of the chiral auxil- iary is bound via Li’ to the enolate oxygen as shown in 6. This model accounts for the failure of Na’ (entry 6) and of the N-benzoyl group (entry 10) to induce high ee values. Due to the weak Lewis acidity of Na’ and the weak Lewis basicity of an amide nitrogen the interaction between eno- late and catalyst is loosened in both cases.

Since the conversion of rac-2d into rac-vindorosine has previously been reported[’], the present result can be for- mally regarded as an enantioselective synthesis of this al- kaloid.

Received: September 26, 1983; revised: November 7, 1983 [Z 572 IE]

German version: Angew. Chem. 96 (1984) 165

[I] J. W. ApSimon, R. P. Seguin, Tetrahedron 35 (1979) 2797; H. Wijnberg, CHEMTECH 1982, 116.

[2] For an early and a recent example see: Z. G. Hajos, D. R. Parrish, J. Org. Chem. 39 (1974) 1615; T. Katsuki, K. B. Sharpless, J. Am. Chem. SOC. 102 (1980) 5976.

[3] W. N. Speckamp, S. J. Veenstra, J. Dijkink, R. Fortgens, J. Am. Chem. SOC. 103 (1981) 4643. Review on 1,5-electrocyclizations: R. Huisgen, An- gew. Chem. 92 (1980) 979; Angew. Chem. Inl. Ed. Engl. 19 (1980) 947.

[4] S. J. Veenstra, W. N. Speckamp, J . Am. Chem. SOC. 103 (1981) 4645; J. Dijkink, J. N. Zonjee, B. S. de Jong, W. N. Speckamp, Heterocycles 20 (1983) 1255.

[5] S. J. Veenstra, W. N. Speckamp, J. Chem. SOC. Chem. Commun. 1982, 369.

[6] H. Hiemstra, H. Wynberg, J . Am. Chem. Soe. 103 (1981) 417.

[7] Procedure: To a well stirred solution of (IS,2R)-N-methylephedrine 4b (270 mg, 1.63 mmol) ([a]!&= +25 (c=6, EtOH)) in 5 mL THF under nitrogen at 0°C was added 0.40 mL of a 1 . 6 ~ solution of nBuLi (0.64 mmol) in hexane. Two minutes later a solution of azomethine l b (210 mg, 0.63 mmol) in 1 mL of toluene was added. The resultant blue mix- ture slowly turned red and was stirred for 45 min at 0°C. 5 mL of 2~ HCI was then added and the mixture extracted with ether (3 x 10 mL). The ether solution thus obtained was dried (Na2S04) and evaporated to leave a light yellow oil (110 mg). Purification via column chromatogra- phy [SO2 (70-150 mesh), EtOAc, hexane 1 :2] afforded 104 mg (0.31 mmol, 50%) of Zb as light yellow crystals, m.p. 118-119°C (EtOAc/ hexane); m.p. (rac-2b) 125--126°C [3, 81. [a]?$, -71 (c=5.25, CHCI3); IR(KBr): 3320 (NH), 1760 and 1685 (C=O), 1600,1385,1340,1165,935, 695 cm-I. ’H-NMR (250 MHz, CDCI,): 6=7.45-6.65 (m, 9H), 4.70 (d,

(m, 2H, NH and >CH-nPr), 3.17 (d, 1 H, J = 19 Hz, -CH.H,CONa, 2.70 (d, l H , J=19 Hz, -CH,H,CON=), 1.7-1.1 (m, 4H, CHICHI), 0.79 (t. 3H, CH,). Addition of Eu(hfc), did not change the ‘H-NMR spectrum (in the case of the racemic compound the signal originally at S=3.17 was split into two doublets of equal intensity).

[8] S. J. Veenstra, Dissertation, University of Amsterdam 1982. [9] H. T. Clark, H. B. Gillespie, S. Z. Weisshaus, J. Am. Chem. SOC. 55

[lo] D. N. Reinhoudt, G. W. Visser, W. Verboom, P. H. Benders, M. L. M.

1 H, J = 14 Hz, CH.H,Ph), 4.61 (d, 1 H, J = 14 Hz, CH,H,Ph), 3.85-3.70

(1933) 4571.

Pennings, J . Am. Chem. SOC. 10s (1983) 4775.

1,2-Dioxetane: Synthesis, Characterization, Stability, and Chemiluminescence** By Waldemar Adam* and Wilhelm J. Baader

To the best of our knowledge“’, the parent 1,2-dioxetane 1 has been observed only as a transient species in the gas phase via the formaldehyde fluorescence arising from the cycloaddition of singlet oxygen to ethenel’l. The statement . . . “isolation and characterization of 1,2-dioxetane still re- main a challenge” . . .131 encouraged us to try to prepare the parent 1,2-dioxetane. Here, we report its synthesis, charac- terization, stability, and chemilurninescen~e~~~”.

F H 2

[8 - CH2=O* --+ hu - CHzO

CHz +k Br-FH, I 4 1

CHZ-OOH

2

Synthesis: After photosensitized singlet-oxygenation of ethene in trichlorofluoromethane at - 40°C and in the gas phase both failed to produce even traces (monitored by chemiluminescence) of dioxetane 1, we returned to the now classical Kopecky methodL5). A solution of 2.82g (20.0 mmol) I-bromo-2-hydroperoxyethane (2)161 in 20 mL dichloromethane and a solution of 8.00 g (143 mmol) KOH in 20 mL water was vigorously mechanically stirred for ca. 15 min: the mixture was then warmed up from 0°C to 20°C. The organic phase was washed with 10 mL cold wa- ter, dried over MgS04, and the 1,2-dioxetane 1 distilled as a CHzClz solution at O”C/lOO torr to 2OoC/10 torr. Thin

[*I Prof. Dr. W. Adam, Dr. W. J. Baader Institut fur Organische Chemie der UniversitPt ~-8700 Wtirzburg (FRG)

[**I We thank the Deutsche Forschungsgemeinschaft, the Fonds der Chemi- schen Industrie, and the Stiftung Volkswagenwerk for generous finan- cial support.

166 0 Verlag Chemie GmbH, 0-6940 Weinheim, 1984 0570-0833/84/0202-0166 $02.50/0 Angew. Chem. Int. Ed. Engl. 23 (1984) No. 2

layer chromatography (silica gel/CHZCl2) gave a peroxide spot at RF=0.58 (positive KI/HOAc test). Attempts to re- move the CHzCIZ by distillative fractionation failed be- cause of the volatility and thermal lability of 1. In all sub- sequent operations the dioxetane was therefore handled in CH2C12 solution. The yield of the dioxetane was estimated .as ca. 0.1% by iodometric titration.

Characterization: Solutions of 1 in CH2CI2 exhibited di- rect chemiluminescence, which was significantly enhanced by the presence of 9,lO-dibromoanthracene (DBA). Con- trol experiments with hydroperoxide 2 confirmed that the latter displayed negligible chemiluminescence under com- parable conditions. The 400 MHz 'H-NMR and 100 MHz l3C-NMR spectra of a CDzCIz solution of 1 at - 40°C ex- hibited the expected signals at 6=5.38 and 6=76.14, re- spectively, which both disappeared after 1 h at 40°C. The characteristic singlet of formaldehyde (expected decompo- sition product) was observed at 6=9.60 in the 'H-NMR spectrum. These data corroborate the proposed dioxetane structure 1.

Stability: The activation parameters of dioxetane 1 were determined by DBA-enhanced chemilumine~cence[~~. Un- fortunately, 1 is very prone to catalytic dark decomposi- tion; hence, for reliable and reproducible kinetic data the activation data had to be assessed from the temperature dependence of the initial chemiluminescence intensities[71 (Table 1). Clearly, the normal chemiluminescence method (values in parentheses) results in considerable catalytic dark decomposition, as witnessed particularly by the very low activation entropy. However, the free energy of activa- tion appears to be a more reliable measure of thermal sta- bility (Table 1). At 343 K a half-life t1/2= 1.1 min is extra- polated from these rate data. Compared to tetramethyl-1,2- dioxetane 3, for which tIl2=45 min at 343 K"], the parent 1,2-dioxetane 1 is, as predicted['], considerably less sta- ble.

Table 1. Activation parameters and excitation yields for the thermal decom- position of 1,2-dioxetane 1 in toluene and, for comparison, those for tetra- methyl-l,2-dioxetane 3 in benzene. For the values in parenthesis see text.

AHC A s + AG'(343K) @'[a] OT [bl [kcal/mol] [calmol-lK-' [kcal/mol] x lo4 x lo2 OS

0.03 +0.01 0.2+0.05700 22.1k0.3[c] -3.9 23.3 (18.9+0.8)[d] (-12.8f1.6) (23.4)

3 26.9kO.l 3.7f0.3 25.8 25k14 35rt15 140

[a] Determined from DPA fluorescence via Stern-Volmer kinetics [9]. [b] De- termined from DBA fluorescence via Stem-Volmer kinetics [9]. [c] Deter- mined from the temperature dependence of the initial chemiluminescence in- tensities monitored by DBA-fluorescence 171. [d] Determined from the time profile of the DBA-fluorescence intensity at various temperatures [7]. [el Taken from [7] and 191.

Chemilurninescence: Direct chemiluminescence could be observed, but was too weak for quantitative measurements. For this reason energy transfer chemiluminescence with 9,lO-diphenylanthracene (DPA) was used to determine sin- glet excitation yields (@LpA) and DBA to determine triplet excitation yields (@ZBA)['] (Table 1). Unquestionably, the excitation yields are dramatically lower than those of the tetramethyl-1,2-dioxetane 3; but, like 3, the spin state se- lectivity (QT/@') for triplet excitation is highly favored. The thermal stability and excitation yields increase with methyl substitution.

Received: October 21, 1983 [Z 598 IE] German version: Angew. Chem. 96 (1984) 156

CAS Registry numbers: 1, 6788-84-7; 2, 88510-96-7; ethene, 74-85-1; hydrogen peroxide, 7722-84-1

111 W. Adam, G. Cilento: Chemical and Biological Generation of Excited States, Academic Press, New York 1982; Angew. Chem. 95 (1983) 525; Angew. Cfiem. Int. Ed. Engl. 22 (1983) 529.

[2] D. J. Bogan, J. L. Durant, Jr., R. S. Sheinson, F. W. Williams, Pfiotocfiern. Photobiol. 30 (1979) 3.

131 D. J. Bogan in 111, chapt. 2. [4] W. J. Baader, Dissertation, Universitat Wiirzburg, July 1983. [5] K. R. Kopecky in [I], chapt. 3. [6] 2 was prepared in 35% yield by allowing ethene to react with 1,3-dibro-

mo-5,5-dimethylhydantoin and hydrogen peroxide; colorless liquid, h.p.=3O0C/O.1 torr, 97% peroxide titer by iodometry; IR (CC14): v=3420 cm-' (s, -0OH); 'H-NMR (CDC13, 90 MHz): 6=3.62 (t, J=6.0 Hz, 2H), 4.30 (t,J=6.0 Hz, 2H), 8.90 (br. s, 1 H, -0OH); "C-NMR (CDC13, 22.6 MHz): 6=28.46 (t, C-Br), 76.21 (t, C-OOH).

[7] W. Adam, K. Zinner in [I], chapt. 5. [8] H. E. ONeal, W. H. Richardson, J. Am. Chem. SOC. 92 (1970) 6553. 191 W. Adam in [I], chapt. 4.

Total Synthesis of a Fortimicin Aglycone"" By Jurgen Schubert, Reinhard Schwesinger, and Horst Prinzbach *

Efficient methods have been developed for the synthesis of a broad spectrum of cis-1,3- and cis-l,4-(deoxy)inosa- diamines using (deoxy)anhydroinositols such as 1 (R= H, OH), which are readily accessible from benzene"'. Such cis-l,4-(deoxy)inosadiamines constitute the aglycone com- ponents of several novel, highly effective and structurally simple aminoglycosidic antibiotics (fortimicins 212], ista- mycin~[~], ~annamycins'~]).

/ Y' HO HN-CH3

1 2 3

Herein we report on a convenient entry to enantiomeri- cally pure 3-de-0-methylfortamine 1la starting from 1,2 :4,5-dianhydro-epi-inositol 1 (R= OH), which is also useful for the total synthesis of chemically modified for- timicins['I. (In the formula of fortamine 3, both the "num- bering for fortimicin" as well as the "numbering for inosi- tol" is given.) The procedure outlined in Scheme 1 excels in the high selectivity of the epoxide-opening (in 4/5d) and in the efficient resolution (chromatographic separation of the diastereomeric (R)-( +)-1 -phenylethylamine adducts 9a/10a); it makes allowance for the fact that two different N-functions could not be selectively introduced via 1 ow- ing to insufficient kinetic differentiation in the separate steps['*61. The bisurethane 4 prepared by reaction of 1 with methyl isocyanate (dioxane, lOO"C, 3 h, 95%, m.p.=203- 205 OCf7]), is cyclized regiospecifically by strong, weakly nucleophilic bases, preferably tris(dimethy1amino)methy- liminophosphorane[81 (acetonitrile, 60°C, 3 h), with attack at C-1 to give the oxazolidone rac-5a. The, a priori, possi-

[*] Prof. Dr. H. Prinzbach, J. Schubert, Dr. R Schwesinger Institut fur Organische Chemie und Biochemie der UniversitBt Albertstrasse 21, D-7800 Freiburg (FRG)

Fonds der Chemischen Industrie. and BASF AG. [**I This work was supported by the Deutsche Forschungsgemeinschaft, the

Angew. Chem. Int. Ed. Engl. 23 (1984) No. 2 0 Verlag Cfiemie GmbH. 0-6940 Weinheim, 1984 0570-0833/84/0202-0167 $02.50/0 167