Angewandte
Communications
Chemie
to furnish a globular molecular architecture was manifested
could be controlled by introducing a minute stereogenic
by an almost flat triple DDA–AAD hydrogen-bond array
(Figure 1b, conformer B). No evidence was observed for
either the b-hairpin-like conformer A, or bridged confor-
mer C (Figure 1). Unfolded trans-1a eluted faster than folded
cis-1a based on size-exclusion chromatography in CHCl3,
thus establishing a clear difference in size of the two isomers
to support the globular architecture (Figure 2c). A reduction
in structural flexibility by shortening the linker length [1b:
L1 = L2 =-(CH2)4-] gave identical NOESY crosspeaks, thus
suggesting a similar folded structure was formed in solution.
Unexpectedly poor solubility of the shorter analogues
prevented further investigation. Comparative kinetic analysis
of the thermal reversion of cis-1 and control azobenzene cis-2
(which lacks the ability to form a hydrogen bonding array) in
the absence of light at room temperature revealed that the
hydrogen bonding retarded the reversion rate to retain the
folded structure, albeit with a small magnitude (Figure 2d).
This observation implies that the choice of hydrogen bonding
array can be a determinant to modify reversion kinetics. As
with the o-fluoroazobenzene derivatives reported by Hecht
et al.,[17] reversion appeared to follow a first-order exponen-
tial decay, and almost complete suppression of isomerization
was observed at 38C in the dark.
element to the foldamer 1. The installation of a nonpolar Me
group next to the amidonaphthyridine unit in L2, where an
NOESY crosspeak was observed with Hpyr2 in the folded cis
isomer (Figure 2b), was envisaged to give pseudo-diastereo-
mers as well as pseudo-enantiomers, with an unfavorable
steric constraint in one pseudo-diastereomer promoting
selectivity towards the other (Figure 3b). An initial attempt
of folding was conducted with racemic trans-1c bearing
a branching methyl substituent. In striking contrast to the
achiral cis-1a and cis-1b, cis-1c displayed clear differentiation
between individual geminal methylene protons of the linker
in the 1H NMR spectrum. 1H-13C HSQC spectroscopy
revealed that five out of seven of the diastereotopic methyl-
ene geminal protons had split into sets of two distinct signals.
Intriguingly, only one of the two split signals of the methylene
proton Hmet2, next to the pyridylurea unit, gave a NOESY
crosspeak with Hnaph, thus excluding the possibility that the
split proton signals originated from a one to one mixture of
the proposed pseudo-diastereomers (Figure 4a).[25] The Hmet2
and Hmet2’ peaks were therefore strongly anisochronous
1
[D0.36 ppm in H NMR at (258C)], thus indicating a strong
handedness bias.[26] In an attempt to increase the likelihood of
observing the unfavored pseudo-diastereomer, the derivative
1d, possessing longer linkers to acquire a more flexible folded
structure, was synthesized (Figure 3b). A key difference
between cis-1c (L1 = L2 = C4) and cis-1d (L1 = L2 = C6)
was found: the NOESY crosspeak detected between Hpyr2 and
the protons at the Me group in cis-1c was no longer observed
for cis-1d, perhaps validating the assumption that derivatives
having longer linkers could be less restricted in their move-
ment, and thus more able to accommodate the steric bias of
Analogous to achiral helical compounds, photoisomeriza-
tion of linear trans-1 presumably gave rise to a mixture of the
two pseudoenantiomeric species, cis-1-minus and cis-1-plus,
upon folding (Figure 3a). Given that external chirality is able
to manipulate the screw-sense of a folded helical structure,[12]
we sought to investigate whether the direction of folding
1
the linker in empty space. While the H NMR spectra of cis-
1c and cis-1d appeared similar at room temperature, addi-
tional splitting of peaks was observed for cis-1d upon cooling
(À608C), most clearly in the NH region as well as for Hmet1
and Hmet2 (Figure 4b,c). NOE data could not distinguish
between these new peaks, however the ratio appeared to be
63:37. Chiral derivatives 1e (L1 = C6, L2 = C4) and 1 f (L1 =
C4, L2 = C6) displayed a similar tendency at reduced temper-
ature, with ratios of 67:33 and 71:29, respectively (Figure 4d).
Remarkably, the less flexible 1c (L1 = L2 = C4) gave only
1
one set of H NMR signals over the same range of temper-
ature (À608C to RT; Figure 4b), thus suggesting that folding
of 1c occurred in almost quantitative fashion to furnish (S)-
Me-M-cis-1c [and pseudoenantiomeric (R)-Me-P-cis-1c]
over (S)-Me-cis-P-1c [and (R)-Me-cis-M-1c; Figure 4a).[27]
Having established the stereoselective folding of trans-1c,
enantioenriched 1c was synthesized and its chiroptical
transition was evaluated by circular dichroism (CD) spec-
troscopy (Figure 5). Compared with the faint CD signals of
geometrically pure (R)-Me-trans-1c (91% ee), the corre-
sponding folded compound (R)-Me-P-cis-1c (91% ee, cis
content 82%) exhibited a large negative Cotton effect, likely
reflecting the P-helicity of the folded structure (Figure 5).[8c]
Accordingly, the S antipode of folded (S)-Me-M-cis-1c (95%
ee, cis content 82%) gave a complementary CD pattern. The
large enhancement of CD intensity by trans to cis transition is
ascribed to the chirality transfer from the Me group at the L2
Figure 3. Schematic representation of folding/unfolding of achiral and
chiral foldamers 1. a) Achiral trans-1a and trans-1b gave the mixtures
of pseudoenantiomeric M-cis-1 and P-cis-1 upon photoisomerization.
b) In the case of chiral derivatives trans-1c-f, bearing a pendant Me
group on the L2 linker, (S)-Me-M-cis-1, and (S)-Me-P-cis-1 [and (R)-Me-
M-cis-1 and (R)-Me-P-cis] are pseudo-diastereomeric and potentially
have a different stability profile.
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3
These are not the final page numbers!