Also,
as already pointed out above, due to its molecular size the NMR
relaxation characteristics of bisindoles often complicate the measurement
of structurally informative NOEs, especially at higher
field strengths where spectral congestion tends to be less of a problem
(for a given structure and at a given field strength, the NOEs can
be either positive or negative, or can fall in between these regions,
depending on the solvent and on whether the bisindole is available
in base or salt form). Moreover, the NMR signal widths of bisindoles
often prove to be inconveniently broad due to the conformational
dynamics of the system, and/or possibly because of weak intermolecular
associations, and/or as a consequence of an acid-base
equilibrium being involved (velbanamine/vindoline-type bisindoles
contain two basic nitrogens, N(4) and N(4), with pKa1 ∼
7.5
and pKa2 ∼
5.5). This signal broadening can only be alleviated by
using carefully selected solvent mixtures or pH (Figs. 9 and 10).
This, however, may modify the chemical shifts, making structurally
indicative comparisons with the literature data of relevant analogues
more difficult. With VCR analogues one faces an additional
difficulty due to the slow N-formyl rotamerism which yields two
sets of NMR signals in all conventional NMR solvents at room temperature.
An important implication of these complications is that,
as opposed to the use of LC–MS techniques as discussed above, in
reality the structure-resolving powers of state-of-the-art NMR can
most efficiently be utilized for bisindoles as the off-line investigation
of pure samples, which offers the best scenario to run 2D
experiments by avoiding the inherent limitations of LC–NMR [49].
By the same token, the most efficient way of using NMR and MS in
a complementary fashion is to run the NMR and MS experiments
from the same isolated sample.