1078
J. Thiel et al.
H3 is not indicative of a chair form of ring A, the more so that the lactam fragment
3
is a factor ¯attening this ring. However, taking into account the value of JHH for
the C6-H (10.2 Hz, denoting 1,2-diaxial interaction with C5-H) and also inspecting
the model, one may conclude that the C5, C6, N1, C2, and C3 atoms occupy
approximately a planar position, whereas C4 rises above this plane towards the C7-
C8-C9 bridge. Such an envelope conformer [20] seems to be the dominant form of
ring A similar to the analogous fragment of angustifoline [21] for which the
corresponding coupling constant amounts to 10.7 Hz. The latter alkaloid, however,
features a small contribution of another conformer in which the C4 methylene
3
group is oriented in the opposite direction to that of the C7-C8-C9 bridge. J for
the perchlorate under discussion also deviates from the analogous value for
3
sparteine with a rigid chair ring A featuring JHH 13.2 Hz [22]. A contribution
from another conformer of ring A, similar to that of angustifoline, seems quite
probable.
Formation of the double bond between C17 and N16 also induces ¯attening
into ring C, comprising, in particular, C7, C17, N16, and C11 [19]. C9 may also
participate in this ¯attening, since no vicinal coupling of H9 and H11 is observed,
whereas an NOE occurs between them (see Table 2). This coupling disappears
when the H-C9-C11-H dihedral angle adopts a value of ꢀ90ꢂ, and this corresponds
to the molecular model. Besides the NOE mentioned, H11 shows the usual 1,3-
diaxial Overhauser enhancement with H15ax. Unfortunately, due to severe overlap-
ping of the NOESY signals, it was not possible to locate the 1,3-diaxial interactions
of H13 proton with H15ax and H11. In other words, the chair form of ring D could
not be con®rmed by 1,3-diaxial NOEs.
The existence of a chair fragment would have been indicated by signal of H15ax
1
provided that this resonance had appeared as a dtd in the H NMR spectrum.
Instead, one observes averaged multiplicities of not only H15ax but also of H15eq
and H11. The protons appearing in such a form belong to the planar H-
C17N16 < C15,C11 fragment which also participates in ring C together with the
chair ring B (vide supra), generating a rather rigid 3,7-diazabicyclo [3.3.1]nonane
(bispidine) system within the molecule of 4. Its rigidity is evidenced by
multiplicities of H6, H8eq, H10eq and H10ax and, in particular, by W couplings
between H10eq and H8eq as well as between H7 and H9 (see Table 2).
Geometrical stabilization of rings B and C, as deduced from the above data, is
additionally con®rmed by NOEs of H17 with H5ax and H15eq as well as by
Overhauser interactions between the pairs H15ax H11 and H10eq H11 (see Table
2). As a consequence, the sp2 nitrogen N16 from ring C introduces a strain onto the
C11-N16-C15 bond angle which amounts to 113.9ꢂ in the solid state [19]. Hence,
relative to 3, the geometry of ring D is destablized in 4. The protons bound to C11
1
and C15 appear, as mentioned above, in the H NMR spectrum in pseudo-dd or
pseudo-dt forms. Such shapes of the resonances should not result from couplings of
the allylic type between the H17 and H11, H15ax as detected by means of the DQF-
COSY spectrum, since H15eq, which does not show this kind of coupling, also
appears in a diffuse form. Moreover, a decoupling experiment (irradiating H17
8.94 ppm with different decoupler powers) did not cause any change in the shapes
of the resonances of H11, H15ax, and H15eq. Therefore, since the geometrical
stability of the hydrogens at C17, C11, and C15 cannot be the reason for the signal