analysis of 4a and 5a using the chiral shift reagent Eu(hfc)3
showed that both were formed with significant stereochem-
ical retention (84% for 4a and 74% for 5a).17 This suggests
that recombination of the initially formed biradical C is fast
in comparison to bond-rotation to achiral biradical D
(Scheme 4). To our knowledge, this represents the first
after 48% conversion.19 Treatment of 3b under the conditions
used for the rearrangement of 3a to 4a + 5a furnished
quinolizidine 4b as a single diastereomer in 55% yield.20
A
slight reduction in reaction temperature led to a small increase
in yield, but at temperatures lower than 85 °C the reaction
became sluggish. Other soluble copper catalysts also gave
inferior results, as did Rh2(OAc)4.
It is interesting to note that 4b was isolated as a single
diastereomer, given the 2:1 ratio of 4a to 5a seen in the
previous series. A possible explanation for this difference is
a higher conformational rigidity of the pyrrolidine ring prior
to carbenoid addition due to the increased steric bulk of the
phenyldimethylsilyl group, leading to a greater preference
for ylide diastereomer A (Scheme 3).
Scheme 4
Quinolizidine 4b was obtained in 77% ee (1H NMR with
Eu(hfc)3), indicating 91% stereochemical retention17 during
the [1,2]-shift. This case displayed a higher degree of
retention than either the trimethylsilyl example 3a or the
earlier proline-derived substrate.4 It is possible that rate of
randomization of the relatively more stable ester-substituted
biradical in the proline example is faster than it is for the
silyl-substituted biradicals. However, the improved degree
of retention by 3b over that by 3a, although welcome, is
not easily explained.
Finally, the suitability of the silyl-directing group as a
hydroxyl surrogate needed to be investigated. Quinolizidine
4b could be diastereoselectively reduced in high yield with
Dibal-H21 to furnish alcohol 6b (Scheme 6).22 Sodium
borohydride furnished diastereomers 6b and 7b with a
modest selectivity for axial attack product 7b. Fleming-
Tamao oxidation of 6b under Denmark’s conditions23
proceeded smoothly to give quinolizidinediol 8b in 91%
yield.24 Support for the stereochemical assignments of 6b
and 7b was obtained by the conversion of the 1:4 mixture
of 6b/7b to 8b and known symmetrical diol 9b.25
example of a silyl-directed Stevens rearrangement. Though
this novel [1,2]-shift is gratifying, this chemistry was not
applicable to the synthesis of hydroxylated quinolizidines
since the trimethylsilyl group is not suitable for subsequent
Fleming-Tamao oxidation. At this point, we next turned
our attention to the corresponding phenyldimethylsilyl
derivative.
Asymmetric lithiation and silylation under the standard
conditions8 furnished (S)-N-Boc-2-phenyldimethylsilylpyr-
rolidine 1b in 92% yield and 85% ee (Scheme 5).18 The Boc
Scheme 5
(19) Alkylation product 3b was found to decompose slowly under the
conditions of its formation, requiring that the reaction be stopped prior to
complete consumption of starting material. The same problem, though less
severe, was encountered in the alkylation of 3a.
(20) Representative Procedure for Conversion of 3b to 4b. To a
stirring solution of Cu(acac)2 (0.011 g, 0.040 mmol) in a solution of degassed
toluene (36 mL) at 85 °C was added a solution of diazoketone 3b (0.123
g, 0.404 mmol) in degassed toluene (4.0 mL) via syringe pump over 1 h.
Upon complete addition, the syringe was washed with 2 mL of toluene
and the contents were added directly to the reaction. The resulting solution
was allowed to cool to room temperature. The reaction mixture was
concentrated and immediately loaded on a 1.5 × 10 cm column and eluted
with a gradient of 20 mL each of 10, 20, and 40% EtOAc/hexanes collecting
2 mL fractions. Product-containing fractions were concentrated under
reduced pressure to give 4b (0.067 g, 58%) as a colorless oil. The product
was a single isomer as determined by proton and carbon NMR: Rf 0.66
(10% MeOH/CH2Cl2); IR (neat) 2937, 2855, 2799, 2752, 1720 cm-1; [R]20
D
+30.8 (c 0.34, CHCl3); 1H NMR (500 MHz, CDCl13) δ 7.48-7.46 (m,
2H), 7.32-7.31 (m, 3H), 2.95-2.85 (br m, 2H), 2.56 (d, J ) 11.0 Hz,
1H), 2.40 (ddd, J ) 11.0, 11.0, 4.0 Hz, 1H), 2.32 (ddd, J ) 13.0, 3.5, 3.5
Hz, 1H), 2.16 (ddd, J ) 13.5, 2.0, 2.0 Hz, 1H), 2.07 (ddd, J ) 12.0, 12.0,
8.0 Hz, 1H), 1.97-1.88 (m, 2H), 1.75-1.70 (br m, 1H), 1.59-1.44 (m,
2H), 1.39 (ddd, J ) 14.5, 11.0, 3.5 Hz, 1H), 1.14 (dddd, J ) 13.5, 12.8,
12.8, 4.0 Hz, 1H), 0.32 (s, 3H), 0.26 (s, 3H); 13C NMR (125 MHz, CDCl3)
δ 207.9, 139.2, 134.4, 128.8, 127.6, 72.7, 56.4, 55.4, 40.3, 26.8, 26.2, 26.2,
22.6, -2.1, -2.7. Anal. Calcd for C17H25NOSi: C, 71.03; H, 8.77; N, 4.87.
Found: C, 71.0; H, 8.93; N, 4.52.
group was removed with anhydrous HCl, and the free amine
was alkylated as before with 2 to provide 3b in 47% yield
(16) (a) Clark, J. S.; Hodgson, P. B.; Goldsmith, M. D.; Blake, A. J.;
Cooke, P. A.; Street, L. J. J. Chem. Soc., Perkin Trans. 1 2001, 3325-
3337. (b) Clark, J. S.; Hodgson, P. B. Tetrahedron Lett. 1995, 36, 2519-
2522.
(17) Retention was calculated as % ee of quinolizidines 4a (71%), 5a
(63%), or 5b (77%) divided by % ee of 3a or 3b (both 85%).
(18) Optical purity of 1b was determined by HPLC, using a Daicel
(21) Reduction with L-selectride or under Meerwein-Pondorff-Verley
conditions also gave 6b exclusively, but in slightly lower yield.
(22) 2D NOESY NMR data clearly indicated the expected axial alcohol
resulting from equatorial attack of the bulky hydride reagent.
i
CHIRACEL OD-H column with a mobile phase of 0.25% PrOH/hex and
a flow rate of 0.5 mL/min.
(23) Denmark, S. E.; Hurd, A. R. J. Org. Chem. 2000, 65, 2875-2886.
Org. Lett., Vol. 4, No. 17, 2002
2815