Molecular torsion balances: Evidence for favorable orthogonal dipolar interactions between organic fluorine and amide groups

Article abstract of DOI:10.1002/anie.200702497

(Figure Presented) Finding the right balance: An indole-extended molecular torsion balance has the geometry for measuring a truly orthogonal noncovalent interaction between a C-F bond dipole and an amide carbonyl group (see picture, green F, red O, blue N). Employing a double-mutant cycle approach, negative interaction free enthalpies were determined. Thus orthogonal dipolar interactions can be a new tool for stabilizing protein-ligand complexes and assembling supramolecular architectures.

Full text of DOI:10.1002/anie.200702497

Communications
DOI: 10.1002/anie.200702497
Supramolecular Interactions
Molecular Torsion Balances: Evidence for Favorable Orthogonal
Dipolar Interactions Between Organic Fluorine and Amide Groups**
Felix R. Fischer, W. Bernd Schweizer, and François Diederich*
Dedicated to Professor David Reinhoudt on the occasion of his 65th birthday
ꢀ
Dipolar interactions are omnipresent in chemistry and
biology.[1] It is theorized that owing to steric constraints,
bond dipoles prefer an orthogonal alignment at closest
contact distance. The observation of an apparently attractive
H···F C hydrogen-bond-like interaction. Herein, we present
two new sets of molecular torsion balances in which the
Tröger base scaffold, bearing the rotor, has been extended to
an indole moiety. In two newly designed double-mutant
cycles, the concerns addressed above are overcome and we
give final proof for the existence of an attractive noncovalent
ꢀ
=
=
orthogonal C F···C O interaction between a backbone C O
ꢀ
of the serine protease thrombin and the Caryl F dipole of a
ꢀ
=
bound inhibitor initiated our program to determine exper-
imentally the energetics of such interactions.[2] The expected
weakness prevented any attempts at their energetic quantifi-
cation by studying a bimolecular protein–ligand or even well-
defined host–guest complexes. Therefore, we relied on a
monomolecular system, the molecular torsion balance, intro-
duced by Wilcox et al.,[3] which enables the quantification of
weak interactions with a greater accuracy than with a
comparable bimolecular system. Using a chemical double-
mutant cycle approach, popularized by Hunter et al.,[4] we
showed that the orthogonal interaction between an aryl-CF3
group and a secondary CH3CONH-aryl group in apolar
solvents was energetically favorable, with a free enthalpy
contribution of DDG = ꢀ1.05 ꢁ 0.25 kJmolꢀ1 in CDCl3 and
ꢀ0.85 ꢁ 0.25 kJmolꢀ1 in C6D6.[5] However, a number of
serious uncertainties remained after this first study: 1) The
validity of the chemical double-mutant cycle could be
questioned since the changes in substitution, made to extract
the energetics of the dipolar interactions, were substantially
perturbing the primary edge-to-face aromatic interactions in
the molecular torsion balance (see below).[6] 2) The dipolar
momentum of a freely rotating CF3 group is not oriented
dipolar Csp2 F···C O interaction in a broader range of
solvents.[7]
In the first set of new molecular torsion balances, the CF3
group on the edge component used in the earlier system was
maintained, while an acetylated indole moiety ensures the in-
plane orientation of the interacting acetamido group which no
ꢀ
longer features an N H fragment. The envisioned double-
mutant cycle resulting from this modification is shown in
Scheme 1. Equation(1) provides the incremental free inter-
action enthalpy between the CF3 and the acetamido carbonyl
group.
DDGCF
¼ DGðꢁÞ-1ꢀDGðꢁÞ-2ꢀDGðꢁÞ-3 þ DGðꢁÞ-4
ð1Þ
3ꢂꢂꢂC¼O
ꢀ
ꢀ
along a single C F bond but is aligned along the CF3 C bond,
thereby poorly reproducing the orthogonal interaction geom-
ꢀ
=
etry observed at short C F···C O distances. 3) It could not be
precluded that the interaction free enthalpy initially reported
ꢀ
arises from a weak, but geometrically possible Narylamide
[*] F. R. Fischer, Dr. W. B. Schweizer, Prof. Dr. F. Diederich
Laboratorium für Organische Chemie
ETH Zürich
Hönggerberg, HCI, 8093 Zürich (Switzerland)
Fax: (+41)44-632-1109
E-mail: diederich@org.chem.ethz.ch
[**] This work was supported by a grant from the ETH Research Council
and the Fonds der Chemischen Industrie. We thank Prof. W.
van Gunsteren (ETH Zurich), Dr. H. Rüegger (ETH Zurich), and
Prof. B. Jaun (ETH Zurich) for valuable discussions.
Scheme 1. Double-mutant cycle of indole-extended molecular torsion
balances for the determination of the interaction free enthalpy between
a CF3 and an acetamide group. The change from (ꢁ)-2 to (ꢁ)-4 takes
into account how the effect of substitution alters the edge-to-face
aromatic–aromatic interaction which is the primary force behind the
folding of the molecule.
Supporting information for this article (synthesis and character-
1
ization of compounds (ꢁ)-1 to (ꢁ)-4, H,19F NOESY experiments
of (ꢁ)-1 and (ꢁ)-7, X-ray crystal structure data of molecules (ꢁ)-3,
(ꢁ)-8, and (ꢁ)-10, error analysis of the physical data) is available
on the WWW under http://www.angewandte.org or from the author.
8270
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 8270 –8273

Angewandte
Chemie
The four molecules (ꢁ )-1 to (ꢁ )-4 were all synthesized
starting from the common precursor (ꢁ )-5 (Scheme 2, see the
Supporting Information). Compound (ꢁ )-5 was prepared in a
newly developed 15-step synthesis starting from commercially
Figure 1. ORTEP plot of (ꢁ)-3. Thermal ellipsoids at 203 K are set at
50% probability.
folding equilibrium of the two rotamers.[11] The resulting
folding free enthalpies DG are summarized in Table 1 for five
different solvents. Applying the double-mutant cycle
approach and Equation (1) provides the incremental free
Scheme 2. Synthesis of (ꢁ)-1 to (ꢁ)-4. a) Ac2O, CH2Cl2, 08C–248C,
ꢃ
98%. b) Me3SiC CH, CuI, [Pd(PPh3)4], Et3N, 758C, 99%. c) nBu4NF,
THF, 708C, 92%. d) LiOH, MeOH/H2O, 508C, 91%. e) Phenol or 4-
(trifluoromethyl)phenol, BOP, Et3N, CH2Cl2, 248C, 72–83%. f) Ac2O,
DMAP, Et3N, CH2Cl2, 248C, 57–72%. BOP=benzotriazol-1-yloxy-tris-
(dimethylamino)phosphonium hexafluorophosphate, DMAP=4-N,N-
dimethylaminopyridine.
enthalpies DDGCF ···C
for the interaction between the
=
O
3
Table 1: Folding free enthalpies for compounds (ꢁ)-1 to (ꢁ)-10.
Compound
DG [kJmolꢀ1 [a]
]
CDCl3
ꢀ3.54
ꢀ3.36
ꢀ0.90
ꢀ1.03
ꢀ1.41
ꢀ1.33
ꢀ0.55
ꢀ1.06
C6D6
CD2Cl2
ꢀ2.92
ꢀ2.78
ꢀ1.16
ꢀ1.20
ꢀ1.18
ꢀ0.95
ꢀ0.78
ꢀ1.05
CD3OD
ꢀ3.80
ꢀ3.09
ꢀ2.62
ꢀ2.33
n.s.[b]
C2D2Cl4
ꢀ3.00
ꢀ2.62
ꢀ0.25
ꢀ0.68
ꢀ1.08
ꢀ0.78
ꢀ0.18
ꢀ0.62
available materials.[8] Acetylation of the amino group and
(ꢁ)-1
(ꢁ)-2
(ꢁ)-3
(ꢁ)-4
(ꢁ)-7
(ꢁ)-8
(ꢁ)-9
(ꢁ)-10
ꢀ4.59
ꢀ4.36
ꢀ1.36
ꢀ1.91
ꢀ1.60
ꢀ1.31
ꢀ0.76
ꢀ1.67
ꢃ
Sonogashira cross-coupling of the bromide with Me3Si-C CH
led to an intermediate that underwent thermal cyclization to
indole (ꢁ )-6 under rather mild conditions in excellent yield.[9]
Deprotection of the methyl ester and subsequent esterifica-
tion with 4-(trifluoromethyl)phenol or phenol led to target
molecules (ꢁ )-2 and (ꢁ )-4. Acetylation of (ꢁ )-2 and (ꢁ )-4
provided (ꢁ )-1 and (ꢁ )-3, respectively, in satisfying yields.
The structure of the molecular torsion balances with the
extended indole scaffold was confirmed by X-ray crystallog-
raphy. Comparison of the crystal structure of (ꢁ )-3 (see
Figure 1 and the Supporting Information) with structures of
torsion balances previously reported by Wilcox and our group
only shows slight deviations from the expected geometry.[10]
The phenyl ester is not exactly centered on the indole moiety,
therefore the H atom on C34 (the CF3-substituted position in
(ꢁ )-1) resides at a distance of 3.84 Š and an angle of
n.s.[b]
n.s.[b]
n.s.[b]
[a] Determined by integration of the line-fitted (100% Lorentz functions)
1H NMR (500 MHz) spectra of 10 mm solutions at 298 K recorded on a
Bruker AMX-500 spectrometer. Uncertainty: ꢁ0.12 kJmolꢀ1. For an error
analysis see the Supporting Information. [b] n.s.=not soluble.
appended dipoles. In apolar solvents, such as C6D6 (ꢀ0.78 ꢁ
0.25 kJmolꢀ1) and C2D2Cl4 (ꢀ0.82 ꢁ 0.25 kJmolꢀ1), the inter-
action is strongest, consistent with the previously reported
values.[5] In smaller dipolar chlorinated solvents, such as
CDCl3 (ꢀ0.31 ꢁ 0.25 kJmolꢀ1) and CD2Cl2 (ꢀ0.18 ꢁ
0.25 kJmolꢀ1), the interaction is substantially weaker. We
attribute this to competing solvation of the extended Tröger
base structure, with the positively polarized hydrogen atoms
of solvent molecules interacting with the electron-rich indole
moiety, thereby shifting the equilibrium in favor of the
unfolded conformation. In the polar solvent CD3OD (ꢀ0.43 ꢁ
0.25 kJmolꢀ1), we observe a stronger interaction energy than
in the previously reported system. In polar protic solvents, the
indole-based torsion balances apparently are less solvated
than the previously reported, CH3CONH-bearing systems,
which feature a strong hydrogen-bond donor.[5]
aH···C O = 1038 over C37 of the acetamide group.
=
Since the interconversion between the rotational con-
formers (atropisomers) in (ꢁ )-1 to (ꢁ )-4 is slow on the
1H NMR spectroscopy timescale (DG° ꢄ 67 kJmolꢀ1 at
298 K), two separate signals for the aromatic methyl group
of the rotor can be observed (see the Supporting
Information).[3a] 1H,19F NOESY experiments demonstrate
the expected proximity of the CF3 group and the CH3CO
moiety in the folded conformation of (ꢁ )-1 (see the
Supporting Information). The 1H NMR spectra were line-
fitted to Lorentz functions and integrated to determine the
Angew. Chem. Int. Ed. 2007, 46, 8270 –8273
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8271

Communications
Analysis of the thermodynamic data in Table 1 reveals
that the acetylation of the indole nitrogen atom ((ꢁ )-4!
(ꢁ)-3) causes a small unfavorable change in the folding free
enthalpy of + 0.04 to + 0.55 kJmolꢀ1 in benzene and the
chlorinated solvents. In the case of CD3OD, an inverse effect
is observed (ꢀ0.29 kJmolꢀ1). The change in folding free
enthalpy for the substitution of the edge component with a
CF3 group ((ꢁ )-4!(ꢁ)-2) instead is much larger (ꢀ0.76 to
ꢀ2.45 kJmolꢀ1) and similar to the first system reported.[5] This
large change raises questions regarding the validity of the
double-mutant cycle.[6] The application of the double-mutant
cycle is restricted to systems that can be separated into
independent subsystems. Only then can the total free
enthalpy DG be expressed as a sum of its components, one
of them being the dipolar interaction. The substantial
energetic change upon moving from (ꢁ )-4 to (ꢁ )-2, however,
shows a strong perturbation of the primary edge-to-face
interaction. The acidification of the edge protons of the
phenyl ester by the CF3 group in (ꢁ )-2 causes a stronger
ꢀ
primary edge-to-face aromatic–aromatic C H···p interaction,
showing that both interactions are strongly cooperative and
cannot be regarded as independent subsystems. Therefore,
the validity of this double-mutant cycle is limited.
In an attempt to better meet the requirements for a valid
double-mutant cycle, we developed the model system shown
in Scheme 3. The four target molecules (ꢁ )-7 to (ꢁ )-10 were
synthesized along the route shown in Scheme 2.[8] Incorpo-
ration of a 5-fluoronaphth-2-yl ester in (ꢁ )-7 allows for an
orthogonal dipolar interaction since the local dipole moment
ꢀ
is oriented along the C F bond. Control compounds (ꢁ )-9
and (ꢁ )-10 bear an unsubstituted naphth-2-yl ester. X-ray
crystal-structure analyses of (ꢁ )-8 and (ꢁ )-10 (Figure 2)
reveal only negligible structural changes upon mutating
Figure 2. Top: ORTEP plot of ( ꢁ)-8. Thermal ellipsoids at 233 K are
set at 50% probability. Bottom: ORTEP plot of (ꢁ)-10, arbitrary
numbering. Thermal ellipsoids at 298 K are set at 50% probability.
naphth-2-yl ester (ꢁ )-10 to 5-fluoronaphth-2-yl ester (ꢁ )-8.
ꢀ
In particular, no additional primary C H···p interaction of the
proton on C35 of the naphthyl fragment in (ꢁ )-10 with the
1
indole system is possible. H,19F NOESY experiments dem-
onstrate the expected proximity of the fluorine atom and the
CH3 group of the acetamide in the folded conformation of
(ꢁ)-7 (see the Supporting Information).
The resulting folding free enthalpies are summarized in
Table 1 for four of the previously used solvents. Adapting
Equation (1) to the second double-mutant cycle provides the
free interaction enthalpy for a truly orthogonal interaction
ꢀ
=
between a Csp2 F and a C O dipole. In the apolar solvents
C6D6 (ꢀ1.21 ꢁ 0.25 kJmolꢀ1) and C2D2Cl4 (ꢀ0.74 ꢁ
0.25 kJmolꢀ1), the interaction is strongest while the interac-
tion free enthalpy in the dipolar solvents CDCl3 (ꢀ0.59 ꢁ
0.25 kJmolꢀ1) and CD2Cl2 (ꢀ0.49 ꢁ 0.25 kJmolꢀ1), which, as
discussed above, compete by favorable solvation of the indole
moiety in the unfolded conformer, only amounts to half of the
magnitude. The difference in free enthalpy between the
Scheme 3. Double-mutant cycle for the determination of the interac-
ꢀ
tion free enthalpy between a Csp2 F bond dipole and an acetamide
group.
8272
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 8270 –8273

Angewandte
Chemie
solvents CDCl3 and CD2Cl2 of 0.10 kJmolꢀ1 correlates with
the difference in dipole moment.
[1] R. Paulini, K. Müller, F. Diederich, Angew. Chem. 2005, 117,
1820 – 1839; Angew. Chem. Int. Ed. 2005, 44, 1788 – 1805.
[2] a) J. A. Olsen, D. W. Banner, P. Seiler, U. Obst-Sander, A.
DꢀArcy, M. Stihle, K. Müller, F. Diederich, Angew. Chem. 2003,
115, 2611 – 2615; Angew. Chem. Int. Ed. 2003, 42, 2507 – 2511;
b) J. A. Olsen, D. W. Banner, M. Kansy, K. Müller, F. Diederich,
ChemBioChem 2004, 5, 666 – 675.
[3] a) S. Paliwal, S. Geib, C. S. Wilcox, J. Am. Chem. Soc. 1994, 116,
4497 – 4498; b) E. Kim, S. Paliwal, C. S. Wilcox, J. Am. Chem.
Soc. 1998, 120, 11192 – 11193.
[4] a) H. Adams, F. J. Carver, C. A. Hunter, J. C. Morales, E. M.
Seward, Angew. Chem. 1996, 108, 1628 – 1631; Angew. Chem. Int.
Ed. Engl. 1996, 35, 1542 – 1544; b) S. L. Cockroft, C. A. Hunter,
Chem. Soc. Rev. 2007, 36, 172 – 188.
[5] For a discussion on the validity of a double-mutant cycle
approach applied to torsion balances see: F. Hof, D. M. Scofield,
W. B. Schweizer, F. Diederich, Angew. Chem. 2004, 116, 5166 –
5169; Angew. Chem. Int. Ed. 2004, 43, 5056 – 5059.
The data in Table 1 shows that acetylation of the indole
ꢀ
N H ((ꢁ )-10!(ꢁ)-9) is accompanied by a small (+ 0.27 to
+ 0.91 kJmolꢀ1) unfavorable change in folding free enthalpy,
consistent with the observations in the first double-mutant
cycle. The mutation of a naphth-2-yl to a 5-flouronaphth-2-yl
((ꢁ )-10!(ꢁ)-8) instead is almost neutral in energy (ꢀ0.27 to
+ 0.36 kJmolꢀ1)—as expected from the analysis of the X-ray
crystal structures—and certainly meets the criteria for the
validity of a double-mutant cycle.
In summary, we have presented the design, synthesis, and
evaluation of two novel, indole-extended molecular torsion
balances giving final proof for the existence of attractive
ꢀ
orthogonal dipolar interactions between a Csp2 F bond and an
amide carbonyl group. The measured interaction free enthal-
pies in nonpolar solvents lie in the range of ꢀ0.8 to
ꢀ1.2 kJmolꢀ1. Furthermore we have shown that the restric-
tions imposed on the validity of a double-mutant cycle
approach can be met by a careful design of the model system
combined with a thorough analysis of the influence of
substituent effects on the energetic contributions to the
folding equilibrium. Attractive orthogonal dipolar interac-
tions clearly represent a new promising tool to enhance the
stability of protein–ligand interactions in medicinal chemistry
and to assemble supramolecular architectures.
[6] A. E. Mark, W. F. van Gunsteren, J. Mol. Biol. 1994, 240, 167 –
176.
[7] S. L. Cockroft, C. A. Hunter, Chem. Commun. 2006, 3806 – 3808.
[8] All compounds were fully characterized by melting points, 1H
and 13C NMR spectroscopy, IR, MS, and high-resolution MS.
The preparation of (ꢁ )-1–(ꢁ )-4 starting from (ꢁ )-5 is described
in full detail in the Supporting Information; the synthesis of (ꢁ )-
5 and the new molecular torsion balances (ꢁ )-7–(ꢁ )-10 will be
part of a future full paper, that also addresses the solvent
influences discussed in Ref. [7].
[9] Y. Kondo, S. Kojima, T. Sakamoto, Heterocycles 1996, 43, 2741 –
2746. A. Yasuhara, Y. Kanamori, M. Kaneko, A. Numata, Y.
Kondo, T. Sakamoto, J. Chem. Soc. Perkin Trans. 1 1999, 529 –
534.
Received: June 8, 2007
Revised: August 22, 2007
[10] The X-ray crystal structures of three previously reported torsion
balances can be found in the Cambridge Structural Database:
PIWYAV,[3a] PIWYEZ,[3a] and WAFLOF.[5] To date we have not
been able to obtain crystals of (ꢁ )-1 and (ꢁ )-7 suitable for X-
ray analysis.
Published online: September 26, 2007
Keywords: amides · dipolar interactions · fluorine ·
[11] All compounds were measured and processed three times to
determine the experimental standard deviation.
.
mutant cycles · torsion balance
Angew. Chem. Int. Ed. 2007, 46, 8270 –8273
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8273