Selective complexation of N-alkylpyridinium salts: Binding of NAD + in water

Article abstract of DOI:10.1002/chem.200400603

A new class of receptor molecules is presented that is highly selective for N-alkylpyridinium ions and electron-poor aromatics. Its key feature is the combination of a well-preorganized molecular clip with an electron-rich inner cavity and strategically p

Full text of DOI:10.1002/chem.200400603

FULL PAPER
Selective Complexation of N-Alkylpyridinium Salts:
Binding of NAD+ in Water
Michael Fokkens,[a] Christian Jasper,[a] Thomas Schrader,*[a] Felix Koziol,[b]
Christian Ochsenfeld,*[b] Jolanta Polkowska,[c] Matthias Lobert,[c] Bjçrn Kahlert,[c] and
Frank-Gerrit Klꢀrner*[c]
Abstract: A new class of receptor mol-
ecules is presented that is highly selec-
tive for N-alkylpyridinium ions and
electron-poor aromatics. Its key feature
is the combination of a well-preorgan-
ized molecular clip with an electron-
rich inner cavity and strategically
interactions, and leads to their highly
efficient desolvation. NAD+ and
NADP, the important cofactors of
many redox enzymes, are recognized
by the new receptor molecule, which
embraces the catalytically active
nicotinamide site and the adenine unit.
Even nucleosides such as adenosine
are likewise drawn into the clipꢂs
cavity. Complex formation and struc-
tures were examined by one- and two-
dimensional NMR spectroscopy, Job
plot analyses, and isothermal titration
microcalorimetric (ITC) measurements,
as well as quantum chemical calcula-
1
placed,
flanking
bis-phosphonate
tions of H NMR shifts. The new recep-
monoester anions. This shape and ar-
rangement of binding sites attracts pre-
dominantly flat electron-poor aromat-
ics in water, binds them mainly by p–
cation, p–p, CH–p, and hydrophobic
tor molecule is a promising tool for
controlling enzymatic oxidation pro-
cesses and for DNA chemistry.
Keywords: cofactors
systems molecular clips
pi interactions
· host–guest
·
·
Introduction
extrusion from carboxylic acids, while many more related
processes in various biosynthetic pathways all rely on the
use of NAD+.[1–4] In the enzyme protein environment the
pyridinium unit of NAD+ is usually noncovalently buried in
a cleft called the Rossman fold.[5–7] Adenine and nicotin-
amide are both immerged in cavities with at least one hy-
drophobic side, facilitated by strong hydrophobic interac-
tions and dispersive forces.[8] Additional electrostatic inter-
actions and hydrogen bonds with basic amino acids are used
to keep the phosphate anions in place. We asked ourselves
whether such a microenvironment might be imitated by arti-
ficial receptor molecules, which could in turn act as carriers
for this important cofactor. To this end we joined our efforts
and combined two powerful binding motifs into a new class
of synthetic hosts for N-alkylpyridinium ions.
Many enzymes use NAD+ or NADP as cofactors for stereo-
selective oxidations or reductions. Thus, dehydrogenases cat-
alyze the oxidation of secondary alcohols to carbonyl com-
pounds or the reverse process, decarboxylases facilitate CO2
[a] M. Fokkens, C. Jasper, Prof. Dr. T. Schrader
Fachbereich Chemie der Universitꢀt Marburg
35032 Marburg (Germany)
Fax : (+49)6421-282-5544
E-mail: schradet@staff.uni-marburg.de
[b] Dr. F. Koziol, Prof. Dr. C. Ochsenfeld
Institut fꢁr Physikalische Chemie und
Theoretische Chemie der Universitꢀt Tꢁbingen
72076 Tꢁbingen (Germany)
Fax : (+49)7071-295-490
E-mail: christian.ochsenfeld@uni-tuebingen.de
State of the art: In recent years the p–cation interaction[9–12]
has been shown to be a major noncovalent force used in nu-
merous proteins.[13–19] Consequently, theoretical investiga-
tions have detailed the electrostatic, hydrophobic and dis-
persive contributions to this attraction, which works so well
in aqueous solution.[20–23] In the past decade, several groups
have designed macrocyclic receptor structures with embed-
ded electron-rich aromatic moieties, which use the p–cation
[c] Dr. J. Polkowska, M. Lobert, B. Kahlert, Prof. Dr. F.-G. Klꢀrner
Institut fꢁr Organische Chemie
der Universitꢀt Duisburg–Essen
45117 Essen (Germany)
Fax : (+49)201-183-4252
E-mail: frank.klaerner@uni-essen.de
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2005, 11, 477 – 494
DOI: 10.1002/chem.200400603
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
477

interaction to bind organic cations such as quaternary am-
monium ions,[24–30] guanidinium ions,[31,32] and N-alkylpyridi-
nium ions.[33–36] However, these macrocyclic structures have
a different selectivity towards their substrates; most of them
prefer spherical over flat guests.
An alternative approach consists of designing molecular
clips that offer a high degree of preorganization and are
usually quite selective for flat electron-poor guests. Porphyr-
in clips derived from diphenylglycoluril have been shown to
accommodate N-alkylpyridinium cations (e.g., viologen cat-
ions) between their tips.[37] Unfortunately, they also bind to
electron-rich aromatics like pyridine, hydroxypyridine, and
phenols.[38] In addition, to date they have been only studied
in organic solution.[37] Three earlier reports have appeared
on molecular recognition of NAD+ by artificial hosts:
Schmidtchen[39] used a macrocyclic anion receptor to com-
plex the bis-phosphate bridge of NAD+. Lehn et al.[40] pre-
sented mono- or bisacridinium-functionalized azacrown
ethers, which complex nucleotide anions; a little later, Bian-
chi et al.[41] studied azacrowns in their protonated form,
which also embrace the diphosphate core of NAD+. Cyclo-
voltammetric measurements indicated significant alterations
in the electrochemical behavior of the NAD+/NADH
couple due to its complexation by the azacrown. However,
in all three cases, NAD+ complexation mainly involves the
interaction of the positively charged host with the diphos-
phate anion, and not with the catalytically active nicotina-
mide moiety. This, however, is a major prerequisite for se-
lective interference with dehydrogenases aimed at complex-
ation of their cofactor.
Figure 1. Left: Structures of clips 1a–g. Right: Schematic design of the
new hybrid receptor system as electron-rich molecular clip and bis-phos-
phonate tweezer.
mary ammonium cations, including ammonium alcohols like
adrenaline or aminosugars, should fit into the clip, whereas
bulky substituents, as in tertiary and quaternary ammonium
salts such as alkaloids or acetylcholine should be rejected.
Results
Synthesis of bis-phosphonate clips 1a,b: The synthesis of
1a,b followed
a
straightforward modular strategy
(Scheme 1). The deprotonated form of hydroquinone clip 1c
Design: During the past few years Klꢀrner et al. have intro-
duced various molecular tweezers and clips designed for the
ꢀ
M
inclusion of electron-poor guests such as aromatics with
substituents, pyridinium cations, or even sulfonium cati-
ons.[42] These guests were mainly bound by p–p, CH–p, and
p–cation interactions inside their electron-rich concave
cavity.[42] However, due to the lipophilic nature of these
hosts, molecular recognition was restricted to organic sol-
vents, mainly chloroform. Introduction of hydrophilic
OCH2CO2R substituents in the central arene spacer units of
the tweezers and clips led to an unfavorable folding of these
side chains toward the receptor cavity and prevented effi-
cient binding in polar solution.[43,44] Schrader et al. have used
convergent bis-phosphonate anions extensively for recogni-
tion of ammonium alcohol and guanidinium.[31,32,45] These
electron-poor guests are bound by the bis-phosphonate
hosts in a chelate arrangement that maximizes electrostatic
attraction and establishes at the same time a network of
ionic hydrogen bonds. In a joint effort we have now adorned
the molecular clip 1 with short phosphonate monoester
anions and obtained water-soluble host molecules 1a,b
(Figure 1).[46] They not only have an open cavity, but they
are also flanked by two negatively charged functionalities
for additional ion pairing with cationic guests. Molecular
modeling studies (molecular mechanics calculations in
water, MacroModel 7.1, Amber*)[47,48] suggested that pri-
Scheme 1. Synthesis of bis-phosphonate clips 1a and 1b and model com-
pounds 2a and 2b.
was treated with methyl- or phenylphosphonic acid dichlor-
ide, followed by hydrolysis of the remaining chloro substitu-
ents. The resulting acids 1d,e were subsequently deprotonat-
ed with a suitable base in a precision titration providing the
desired counterion, for example, for maximum solubility.
With tetra-n-butylammonium counterions, the bis-phospho-
nate clips 1a,b are soluble in a wide range of solvents, rang-
ing from chloroform over DMSO and methanol to water.
Clips 1a,b were obtained in good to moderate yields and an-
alytically pure form. In addition to 1a,b with the whole clip
scaffold, two model compounds 2a,b were synthesized that
478
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
lacked the naphthalene side-
walls for reasons of comparison.
Compounds 2a,b turned out to
be valuable tools for identifica-
tion of proton-transfer process-
es and other unspecific binding
events.
Binding experiments: The pos-
sible self-association of pure
clip 1a in D2O was investigated
by means of the concentration
dependence of its 1H NMR
spectrum (Supporting Informa-
tion: Table S1). No significant
changes in the chemical shifts
of the clip protons (ca.
0.033 ppm) were observed after
dilution of the solution of 1a by
a
factor of 32 ([1a]=2.70–
0.084mm), and this indicates
negligible tendency of clip 1a
for self-association in aqueous
solution. However, the larger
1H NMR downfield shifts found
for the NCH2CH2 protons
(Dd=0.138 and 0.112 ppm) of
Scheme 2. Potential cationic guest molecules: ammonium salts 3–6, guanidinium and imidazolium derivatives
9–12 (which may serve as models for basic amino acid side chains), and pyridinium chloride 13, as well as the
neutral nicotinamide 14 are not included into the cavity of clip 1a in aqueous solution within the limits of
1H NMR detection.
the tetra-n-butylammonium cation in 1a on dilution is initial
evidence for complexation of the cation inside the clip
cavity (vide infra).
In preliminary experiments 1:1 mixtures of methylphosph-
onate clip 1a and representatives of various classes of or-
ganic cations 3–14 were examined for their ability to form
host–guest complexes in polar solvents (data for a selection
of cations are given in Table 1). The functional groups range
1a), which is positioned inside the clip cavity in water (vide
infra), by one of these cations is energetically not favored.
In methanol the phosphonate anions in 1a are certainly
more basic than in water and may deprotonate the N-H-am-
monium or -iminium groups instead of forming a complex.
Since the cations shown in Scheme 2 are the major compo-
nents in all side chains of basic amino acids, we assume that
the new bis-phosphonate clip 1a does not bind to solvent-
exposed amino acid residues on
protein surfaces. Control ex-
Table 1. Binding constants Ka [mꢀ1] for 1:1 complexes between primary ammonium salts and the methyl-
phosphonate clip 1a or phenylphosphonate clip 1b in various solvents, determined by 1H NMR titrations at
208C.
periments with histidine and ar-
ginine derivatives confirmed
this assumption: no binding
occurs even with the isolated
N/C-protected cationic amino
acids 8 and 12 in equimolar
Guest
Ka (1a, DMSO)
Ka (1b, DMSO)
Ka (1a, MeOD)
Ka (1a, D2O)
aniline·HCl 3
benzylamine·HCl 5
propylamine·HCl 6
acid–base reaction
–
acid–base reaction
no shifts
no shifts
–
–
2400 (ꢁ8%)[a]
2100 (ꢁ9%)[a]
–
6500 (ꢁ44%)[a]
no shifts
amounts
with
clip
1a
[a] The calculated errors represent standard deviations resulting from the curve-fitting procedure. –=not de-
termined.
(Scheme 2). Histidineꢂs pro-
nounced tendency for proton
transfer renders the evaluation
from aliphatic and aromatic primary ammonium cations
over imidazolium and guanidinium ions, to N-H, N-alkyl-,
and N-arylpyridinium or -pyrazinium compounds. In metha-
nol and water, no binding could be observed for any of the
ammonium, N-H-pyridinium, -imidazolium, or -guanidinium
cations shown in Scheme 2. The nonexistent affinity in water
or methanol for the pyridinium, imidazolium, and guanidini-
um cations is surprising in view of their flat nature. We
assume that in these cases the displacement of one of the
tetra-n-butylammonium cations (used as counterion in clip
of the respective binding experiments problematic. Al-
though in both histidine derivatives 10 and 12 considerable
shifts occurred in the imidazolium ring, these do not neces-
sarily indicate inclusion inside the cavity. Because of histi-
dineꢂs low pKa of about 6, dissociation is even observed
during dilution, and in unbuffered solution proton transfer
to the host releases the free base at the end of most titra-
tions. This even occurred in the binding experiment of 12
with 1a in a 65-fold excess of buffer (sodium hydrogenphos-
phate, pH 7). However, since no shifts were observed for
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
479

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
the host, experimental evidence
indicates that histidine deriva-
tives are not included in the
cavity. Since p-xylylene bis-
phosphonates have been shown
to complex primary ammonium
cations in less polar solvents,
we checked the ability of clip
1a,b to recognize these guests
in DMSO. This time, moderate
chemical shifts were produced
in the 1H NMR signals of the
host and guest protons, and
Figure 2. Left: model compound 2a lacking the side walls (side view). Middle: proposed binding mode for al-
kylammonium ions by 1 and 2 in DMSO. Right: Monte Carlo simulation of complex 5·2a in chloroform (Mac-
romodel 7.1, Amber*/H2O, 1000 steps).[47,48]
well-defined binding isotherms were obtained, which al-
lowed excellent fits for a 1:1 stoichiometry.[49–51] Job plots in
the same solvent confirmed this assumption.[52–54] The associ-
ation constants are on the order of 103 mꢀ1 and agree very
well with similar values found for adrenaline derivatives
with p-xylylene bis-phosphonates. However, in our complex-
ation experiments with clips 1a,b the maximum complexa-
tion-induced shifts Ddmax remain relatively small (ca.
0.1 ppm) even for the guest protons. Together with the mod-
erate binding constants this makes a strong argument for a
purely electrostatic interaction
tive substrate structures (Scheme 2) proves that pyridinium
cations are generally included into the electron-rich cavity
of clip 1a, irrespective of the substitution pattern or the
electronic character of functional groups attached to the
heterocycle. But also other types of heterocycles, such as
pyrazinium or imidazolium cations 16 or 21 (Scheme 3),
form stable complexes with clip 1a. The complexation of
the tetra-n-butylammonium cation inside the clip cavity
could be detected by the use of the lithium phosphonate
clip 1g.
in a chelate binding mode, simi-
lar to that already found for xy-
lylene
DMSO. Force-field calculations
(MacroModel 7.1, Amber*/
bis-phosphonates
in
H2O)[47,48] suggest that in these
cases the ammonium cation is
held by both phosphonate arms
above the central benzene ring
syn to the methylene bridges of
the host molecule (Figure 2).
However, as soon as an alkyl
or aryl substituent is introduced
at the ring nitrogen atom, the
resulting permanently charged
pyridinum cation is an excellent
guest for clip 1a. Very large
complexation-induced
shifts
1
Ddmax of the H NMR guest sig-
nals are observed, which reach
maximum values of up to
+3.5 ppm at saturation. This
time, no shifts are observed
with the model compounds
2a,b lacking the naphthalene
side walls, so the strong effect
must be explained by highly ef-
ficient inclusion of the guest
molecule in the electron-rich in-
terior of the molecular clips
leading to a shielding environ-
ment for the aromatic guest
Scheme 3. Structures of N-alkyl- and N-arylpyridinium, -pyrazinium, and -imidazolium substrates 15–21 and
the maximum complexation-induced 1H NMR upfield shifts Ddmax of the guest protons in their 1:1 complexes
with clip 1a in water. 15–17: simple pyridinium or pyrazinium derivatives; 18–20: dyes with pyridinium-based
1
chromophores. * In the cases of 19* and 20* the complexation-induced H NMR upfield shifts (Ddobs) are given
1
which were observed in the H NMR spectra of a 1:1 mixture of 1a and 19 or 20 (each 10ꢀ3 m) in aqueous solu-
1
protons. The list of representa- tions. ** In the case of nBu4N+Brꢀ the H NMR titration was performed with the lithium salt 1g as host.
480
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
1H NMR titrations: The maximum complexation-induced
shifts Ddmax, binding constants Ka, and hence the Gibbs en-
thalpies of association DG, were determined by means of
1H NMR dilution titration experiments, which offer the op-
portunity to observe the complexation-induced shifts in both
components (host and guest) at the same time. They also
allow the binding curves to be determined for very stable
complexes, which is mostly not feasible by means of “con-
ventional” NMR titration experiments in which either the
guest or host concentration is varied, while the concentra-
tion of the other component is kept constant. In the initial
experiments, already reported in a communication,[46] we
calculated these data by using an iterative nonlinear regres-
sion analysis from the dependence of the complexation-in-
1
duced shifts of the H NMR signals of 1a Ddobs on dilution
of the NMR sample containing both host and guest. To
obtain comparable results we used the same three receptor
1
signals for each calculation. A comparison of the H NMR
spectra of clip 1a with those of the model compound 2a and
of mixtures of 1a with various guest molecules indicates,
however, strong interactions of the alkyl side chains of the
tetra-n-butylammonium cation with the arene units of the
clip framework in aqueous solution; these interactions are
diminished by changing the solvent from D2O over D2O/
CD3OD (1:1) to pure CD3OD (Table 2). The weak solvent
Table 2. Comparison of the 1H NMR shifts d of the tetra-n-butylammoni-
um protons of clip 1a, model compound 2a, and mixtures of 1a or 2a
with N-methylnicotinamide iodide (17) or NAD+ (22).
Compound Solvent
d(a-CH2) d(b-CH2) d(g-CH2) d(d-CH3)
1a
D2O
2.37
0.96
1.22
1.51
1.37
1.42
1.64
1.32
0.96
1.14
1.32
1.21
1.26
1.40
1.18
0.73
0.85
0.96
0.87
0.91
1.01
0.85
D2O/CD3OD[a] 2.70
CD3OD
D2O
3.06
2.86
1a+17
1a+22
2a
D2O/CD3OD[a] 2.94
CD3OD
D2O
3.21
2.80
D2O/CD3OD[a]
D2O
3.17
1.62
1.64
1.32
1.41
0.91
0.98
D2O/CD3ODa) 3.20
CD3OD
D2O
D2O
3.20–3.29 1.60–1.74 1.35–1.49 1.03
2a+17
2a+22
3.21
3.22
1.67
1.67
1.38
1.38
0.96
0.97
Figure 3. 1H NMR spectra in the aromatic (top) and aliphatic regions
(bottom) of a) 17, b) mixture of 1a (c=9.57mm) and 17 (c=9.90 mm),
and c) clip 1a in D2O at 258C.
[a] 1:1 mixture.
1
dependence of the chemical shifts of the NCH2CH2CH2CH3
protons in the spectrum of 2a is a good indicator that these
interactions are less important in the model compound 2a
lacking the naphthalene sidewalls.
The pronounced upfield shifts of the NCH2CH2CH2CH3
protons in the spectrum of 1a in D2O suggest that the alkyl
side chains of the tetra-n-butylammonium cation are partial-
ly included into the clip cavity (Figures 3 and 4). This as-
sumption could be confirmed by the finding that the lithium
phosphonate clip 1g forms a stable complex with tetra-n-bu-
tylammonium bromide in aqueous solution. The association
constant (Ka =3800ꢁ150mꢀ1) and the Ddmax values of the
guest signals (Scheme 3) were determined for the formation
of this complex in water by H NMR dilution titration. Ad-
dition of an equimolar amount of N-methylnicotinamide
(17) to a solution of tetra-n-butylammonium phosphonate
clip 1a in D2O leads to a substantial downfield shift of the
signals assigned to the n-butyl protons and to an even more
substantial upfield shift of the signals assigned to the pro-
tons of 17 (relative to signals found in the spectra of pure
1a and pure 17). This finding allows us to conclude that 17
is indeed included in the clip cavity by replacing the tetra-n-
butylammonium cation, that is, it forms the more stable
complex. Similar results were obtained when NAD+ (22)
was added to a solution of 1a in D2O (Figures 3 and 5,
Table 2). The addition of tetra-n-butylammonium bromide
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
481

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
ions in 1a came from two-dimensional NOESY experiments.
In spectra of both 1a and a 1:1 mixture of 1a and 17 in
D2O, intense cross-peaks between the signals of the
NCH2CH2CH2CH3 protons and all of the clip protons indi-
cate close spatial contacts between these protons. In D2O/
CD3OD none of these cross-peaks can be detected any
more. Apparently, methanol solvates both species individu-
ally, and the solvent shells prevent their close spatial ap-
proach, which is necessary to observe the nuclear Overhaus-
er effect. (The NOESY spectra are shown in the Supporting
Information.)
The complexation-induced shifts Ddobs of the 1H NMR sig-
nals of the guest protons were used to determine the Ka and
Ddmax values from the binding curves, because the larger
shifts of these signals make them a more sensitive probe for
complex formation than the host signals. In addition to the
good fit of the binding isotherms, Job plots confirmed the
assumed 1:1 stoichiometry of the inclusion complexes.
Figure 6 shows the binding curves and the Job plot analysis
Figure 4. Structure of 1a calculated by a Monte Carlo conformer search
(5000 structures, AMBER*/H2O, MacroModel 6.5[47,48]).
Figure 5. 1H NMR spectra (aromatic region) of NAD+ (22) before and
after addition of clip 1a in water. Note the strong upfield shifts of the
pyridinium protons a and b, c and f as well as the diverging behavior of
both anomeric protons g and h upon complexation. (See above Lewis
structure for the assignment of the guest protons.)
Figure 6. Top: 1H NMR dilution titration curves for complex formation
between 1a and N-methylnicotinamide (17) showing the dependence of
~
*
the complexation-induced shifts Dd of protons 2-H ( ), 4-H ( ), and 5-H
&
( ) of 17 in D2O. Bottom: Job plot for complex formation between 1a
*
*
and 17 in CD3OD, shown for the guest protons 2-H ( ) and 5-H ( ); c=
mole fraction.
(0.5m) to an aqueous solution of the complex between clip
1a and N-methylnicotinamide iodide ([17·1a]=10ꢀ4 m) leads
to a complete dissociation of this complex within the limits
1
of H NMR detection. Evidently, the tetra-n-butylammoni-
for complex formation between clip 1a and N-methylnico-
tinamide (17). The complex 1a·17 is by far the most stable
one, since many of the experimental points lie in the satura-
tion region of the curve (Figure 6). In this case the experi-
mental error is certainly larger than the standard deviation,
and the Ka value given in Table 3 is only a lower limit. The
um cation in high concentration can again displace the N-
methylnicotinamide cation in the complex with 1a, although
it binds more weakly to 1a.
Further support for the close spatial contact between the
clip framework and the butyl groups of the ammonium cat-
482
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
Table 3. Association constants Ka and Gibbs enthalpies of association DG for complex formation between clip
1a and guest molecules 17–26, determined by H NMR dilution titration in methanol and water at 208C.
tween clip 1a and N-methylni-
cotinamide (17) or NAD+ (22).
From the sign of the heat
pulses produced during the ti-
tration of the aqueous guest so-
lution to an aqueous clip so-
lution it is immediately appa-
rent that host–guest binding in
both systems is exothermic.
Both titration curves, however,
exhibit unusual, but reproduc-
ible, shapes (Figures 8 and 9).
In the case of 17 and 1a, no
plateau was found at the begin-
ning of the titration, and the
curve (even after correction for
the heats of dilution) cannot be
fitted with the 1:1 complex stoi-
chiometry that was found by
1
[a,b]
[a,b]
Substrate
Ka [mꢀ1
]
ꢀDG
Ka [mꢀ1
in D2O
]
ꢀDG
in CD3OD
[kcalmolꢀ1
]
[kcalmolꢀ1
]
nicotinamide (17)
Kosower salt (15)
pyrazinium salt (16)
pyridinium dye (18)
imidazolium salt (21)
NAD+ (22)
NADP (23)
2’-desoxyadenosine (24)
AMP (25)
17250ꢁ1950
5.8
4.9
3.8
4.7
–
–
–
–
–
83000ꢁ9400
4800ꢁ1300[c]
10000ꢁ2500
broad signals[d]
7900ꢁ1700
9100ꢁ3450
3900ꢁ400
6.7
4.9
5.3
–
5.2
5.4
4.8
5.0
4.0
5.1
4300ꢁ450
600ꢁ30[c]
2950ꢁ300
–
precipitation[d]
–
–
–
–
5300ꢁ320
1100ꢁ200
NMN (26)
–
6800ꢁ1500
[a] Determined from the complexation-induced 1H NMR shifts of the guest proton signals. [b] The given
errors are standard deviations between the Ka values from different NMR signals; the standard deviations
from the nonlinear regressions were consistently lower. [c] Determined from host signals. [d] Precipitation of
1
the complex after mixing 1c and 22 and broad signals in the H NMR spectrum of a mixture of 1c and 18 pre-
vent the determination of Ka.
results obtained for the complexation of various N-alkyl- or
N-arylpyridinium, N-alkylpyrazinium, and N-alkylimidazoli-
um cations with clip 1a are summarized in Table 3 (Ka and
DG values) and in Schemes 3
Job plot analysis. Regression analysis of the experimental
data with stoichiometry factor n as an additional variable
parameter leads to a good fit for n=0.889 (Figure 8). We re-
and 4 (Ddmax values). Remarka-
bly, neutral guests like 2’-des-
oxyadenosine (24) also form
rather stable complexes with
1a.
Mass spectrometric analysis:
The complexes of clip 1a can
be also detected by mass spec-
trometry. Strong molecular ion
peaks for these complexes ap-
peared in the ESI-MS spectra
(negative ions). Figure 7 shows
the ESI-MS spectrum of the
complex between clip 1a and
2’-desoxyadenosine (24) as an
example. Although the complex
of 1a with NAD+ (22) was only
sparingly soluble in [D4]metha-
nol, the supernatant solution
produced a clean ESI spectrum,
which showed peaks for both
the free host 1a and its 1:1
complex with the fully deuterat-
ed guest 22.
Calorimetric titrations: Isother-
mal titration microcalorimetry
(ITC) is frequently applied for
the quantification of the enthal-
py and entropy changes DH
and DS resulting from complex
Scheme 4. Nucleosides and nucleotides (NAD+ derivatives) 22–26 bound by the clip 1a and the maximum
complexation-induced H NMR upfield shifts Ddmax of the guest protons in their 1:1 complexes with clip 1a in
formation.[55] We used ITC to
1
study complex formation be- water.
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
483

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
Figure 7. ESI-MS spectrum for the complex between clip 1a and 2’-des-
oxyadenosine (24). The spectrum was recorded in the negative ion mode.
The calculated masses are: [1a+24]2ꢀ
:
421; [1aꢀOPOMe]ꢀ: 515;
[1a+H+]ꢀ: 593; [1+Na+]ꢀ: 615; [1a+(nBu)4N+)]ꢀ: 834; [1a+24+Na+]ꢀ:
866; [1a+(nBu)4N+)+24]ꢀ: 1085. No ion peaks were found beyond
m/z=1100.
Figure 9. Top: Plot of the heat pulses measured during the titration of 1a
with 22 in H2O ([1a]=832 mm, [22]=7610 mm). Bottom: Heat of dilution
of 22; 40 10 mL portions of a solution of 22 (7610 mm) in H2O were added
to 1 mL of H2O. *artefact of the measurement.
tion, Table S3) and not for formation of the 1:1 complex.
The discrepancy between the NMR and ITC measurements
can be explained as follows: At the beginning of the ITC ex-
periment there is a large excess in the host concentration, so
that formation of a 2:1 host–guest complex is favored. As
the titration progresses, the guest concentration increases
and the more stable 1:1 complex is formed. The combina-
tion of these two processes taking place during the titration
explains why the stoichiometry factor is smaller than unity
(n<1). In the case of the NMR dilution experiment, we
start from a nearly 1:1 mixture of host and guest and there-
fore the formation of a 2:1 complex is highly unlikely. Frac-
tional stoichiometry factors have been also observed in
other ITC measurements on host–guest interactions.[56,57]
The solubility of clip 1a in water is, however, too low to
carry out a reverse ITC titration. This experiment, in which
a saturated solution of the clip ([1a]0 =0.001m) in water was
added to N-methylnicotinamide iodide ([17]0 =0.000012m),
produced only heat pulses that were not significantly larger
than those obtained by dilution of a solution of 1a in pure
water.
Figure 8. Top: Plot of the heat pulses measured during the titration of 1a
with 17 in H2O ([1a]=819 mm, [17]=5723 mm). Bottom: The fit from the
iterative regression analysis of this plot for the stoichiometry factor n=
0.889.
peated the ITC measurements several times using different
starting concentrations [1a] and [17] The stoichiometry
factor n varied in the range from 0.856 to 0.889, and the ap-
parent association constants Ka derived from the fits of
these curves are substantially smaller (1390–5340mꢀ1) than
1
that obtained from the H NMR titration experiment. The
Ka values obtained from the ITC measurements are only
valid for stoichiometry factors n<1 (Supporting Informa-
The heat pulses evolved at the beginning of the titration
of an aqueous solution of 1a with an aqueous solution of
484
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
NAD+ (22) are smaller than those after the addition of sev-
eral portions of 22, and they pass through a maximum
before decreasing. At first glance, this is surprising and not
expected for a simple host–guest binding (Figure 9, top). In-
dependent measurement of the heat of dilution of 22 in
water shows, however, that 22 forms an self-aggregate in
aqueous solution which must be dissociated before the com-
plex 22·1a can be formed. According to the heat of dilution
measured by ITC, the dissociation of the self-aggregate of
22 is an endothermic process (Figure 9, bottom), which must
be overcompensated by the exothermic complex formation
between dissociated 22 and 1a. We evaluated the two plots
shown in Figure 9 under the assumption that the self-aggre-
gate of 22 is a dimer and dissociates prior to complex forma-
tion according to the equation in Figure 9. Neglecting the
first five heat pulses measured in the titration of the aque-
ous solution of 22 to that of 1a leads to the thermodynamic
parameters for the formation of the 1:1 complex 22·1a given
in Figure 10. The binding constant Ka =8800ꢁ500mꢀ1 de-
rived from the ITC measurements is in good agreement
more stable than the corresponding ones in methanol (Ka =
600–17250mꢀ1) are good evidence for a substantial contribu-
tion of a hydrophobic effect, besides the cation–p interac-
tion and salt bridges in the receptor–substrate binding pro-
cesses observed here.[59] A computational study[9] predicts
that, especially in water, the cation–p interaction should be
even more important than the salt bridges and thus largely
responsible for the complex stabilities.
The binding of NAD+ (22) and NADP (23), which are
the most important redox coenzymes in nature, to clip 1a is
particularly remarkable. In methanol the complex 22·1a pre-
cipitated. In water (or methanol:water 1:1) no precipitation
occurred, and distinct upfield shifts were observed in the
1H NMR signals arising from the protons of the nicotina-
mide subunit as well as in the adenine nucleoside of NAD+
(22). These findings unambiguously indicate the complex
formation between both subunits of NAD+ (22) and clip 1a.
Similar results were obtained for the binding of NADP 23
to clip 1a (Table 3, Scheme 4). The binding constant (Ka =
9100ꢁ3450mꢀ1) determined for complex 22·1a is substan-
tially smaller than that of 17·1 (Ka =83000ꢁ9400mꢀ1), al-
though both guests are N-alkylnicotinamide derivatives.
There are two reasons which may explain this difference in
the complex stability:
1
with that determined by H NMR titration (Table 3).
1) The ribose unit of 22 is bound to the nicotinamide
moiety by a tertiary carbon atom in the N,O-acetal,
which may cause an additional steric hindrance in the
complex. This effect should also influence the stability of
the complex between NADP (23) and 1a.
2) According to ITC measurements NAD+ 22 forms a
dimer in aqueous solution, and in the process of com-
plexation the equilibrium between dimeric and mono-
meric 22 must be shifted toward the monomeric form,
which leads to a smaller Ka value.
The aggregation of NAD+ 22 in water observed by ITC
measurements could not be detected by H NMR spectros-
1
Figure 10. The fit from the iterative regression analysis evaluating the
plots shown in Figure 9 by means of the stoichiometry factor n=1.0 re-
sults in the enthalpy, entropy, and Gibbs enthalpy of association DH=
copy. Dilution of a solution of 22 in D2O by a factor of 32
(from 28 to 0.9mm, Supporting Information, Table S2) led to
ꢀ7.3ꢁ0.1 kcalmolꢀ1
,
DS=ꢀ6.7ꢁ0.5 calmolꢀ1 Kꢀ1
,
DG=ꢀ5.4ꢁ
1
0.1 kcalmolꢀ1, and the association constant Ka =8800ꢁ500mꢀ1
.
a shift of the H NMR signals by less than 0.09 ppm. Small
shifts of the 1H NMR signals have been observed in the dilu-
tion of aqueous NADP (23) by a factor of 20 (from 2.6 to
0.13mm), and this suggests an aggregation of 23 in water
comparable to that of 22. ITC measurements with NADP
are planned to confirm this suggestion.
Discussion
Complex stabilities: In contrast to complexes of dimethyl-
ene-bridged clips of type 1 in aprotic organic solvents,[58] clip
1a forms highly stable complexes with the N-alkylpyridini-
um and -imidazolinium salts 15–21, NAD+ (22), NADP
(23), 2’-desoxyadenosine (24), AMP (25), and NMN (26) in
protic solvents such as methanol and water (Table 3). For
example, Kosower salt 15 forms a substantially more stable
complex with clip 1a in D2O (Ka =4800ꢁ1300mꢀ1, DG=
ꢀ4.9 kcalmolꢀ1) than with the corresponding diacetoxy-sub-
stituted clip 1 f in CDCl3 (Ka =137ꢁ14mꢀ1, DG=
ꢀ2.9 kcalmolꢀ1).[44] This result and the finding that the com-
plexes of 1a formed in water (Ka =4800–83000mꢀ1) are
Extension of the guest profile: The upfield shifts observed
in the pyrimidine ring of adenine on binding of NAD+ (22)
and NADP (23) by clip 1a gave a first hint at the possible
inclusion of nucleosides. We checked this interesting hypoth-
esis with the naturally occurring nucleosides 2’-desoxyade-
nosine (24) and adenosine monophosphate AMP (25). A re-
markably high binding constant of Ka =5300ꢁ300mꢀ1 was
determined for complex formation between the neutral
1
guest molecule 24 and clip 1a from the H NMR titration
curve. The significantly smaller binding constant obtained
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
485

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
for the complex between negatively charged 25 and clip 1a
(Ka =1100ꢁ200mꢀ1) indicates that the interaction between
the negatively charged AMP phosphate and clip phospho-
nate groups in the complex 24·1a is repulsive and leads to
weakening of the complex. To compare the contributions of
the nicotinamide with the adenine subunit we cut the
NAD+ molecule into two halves, each carrying a ribose
monophosphate unit. Intriguingly, the binding constant for
nicotinamide mononucleotide (NMN, 26, Ka =6800ꢁ
1500mꢀ1) is significantly higher than that of adenosine
monophosphate (AMP, 25, Ka =1100ꢁ200mꢀ1). This obser-
vation is further good evidence that in the complexes 22·1a
and 23·1a both ring systems, that is, the nicotinamide and
adenine units, are included in the clip cavity. According to
the different binding constants of the complexes 25·1a and
26·1a, preferential inclusion of the nicotinamide ring of
NAD+ (22) or NADP (23) in the cavity of clip 1a is expect-
ed.
Electrostatic potential surface (EPS) calculations were
used to explain why the clips of type 1, analogous to the
structurally related tweezers,[42] selectively bind electron-de-
ficient guest molecules in organic solvents such as chloro-
form. The EPS of bis-phosphonate clip 1a is calculated by
AM1 to be more negative on the concave face than that of
the corresponding parent clip 1 or diacetoxy-substituted clip
1 f. This result can certainly provide an explanation of why
1a is found to be such a good binder to the strongly elec-
tron-deficient cationic guest molecules investigated here.
The comparative binding study of NAD+ and truncated ver-
sions thereof (22–26) shows, however, that in aqueous so-
lution the complementary host and guest EPS, which is cal-
culated for the isolated molecules in the gas phase
(Figure 11), is less important for complex formation in aque-
ous solution. If the electrostatic host–guest interaction is
predominant for complex formation, only the nicotinamide
ring and not the adenine moiety of 22–26 should be bound
to the clip cavity according to the large difference in the
EPS of the nicotinamide and adenine units shown in
Figure 11. Therefore, besides the electrostatic interactions
other effects such as the hydrophobic effect seem to be im-
portant for the strong binding of clip 1a to the various
guests in aqueous or methanolic solution reported here.[59–63]
Complex structures: The maximum complexation-induced
1H NMR shifts of the substrate signals Ddmax are a very sen-
sitive probe of the complex structures. The large Ddmax
values determined for the pyridinium cations 15–20
(Scheme 3) in their complexes with 1a are a good indicator
that the guest molecules are positioned inside the clip cavity
and experience the magnetic anisotropy of the host arene
units. The comparison of the Ddmax values of the Kosower
salt 15 in the complexes with clip 1a (Ddmax =1.96 (Ha), 3.45
(Hb)) and with diacetoxy-substituted clip 1 f (Ddmax =1.82
(Ha), 2.40 (Hb)) shows that the complex structures are simi-
lar and independent of the substitution pattern at the central
spacer unit of the clip. According to the single-crystal struc-
ture analysis of the 1:1 complex between N-ethyl-4-carb-
ethoxypyridinium triiodide (a derivative of 15) and 1 f
(Figure 12), the naphthalene sidewalls of the clip embrace
the guest molecule tightly by reducing the distance between
the naphthalene tips from 10 to 8 ꢄ.[58] Provided the com-
plex structures in the crystal and in solution are similar, the
1
observation of only one H NMR signal for the nonequiva-
lent guest protons Ha, Ha’, or Hb, Hb’, of 15 or 16 in the com-
plexes with clip 1a indicates that exchange of these protons
by rotation of the guest molecule inside the clip cavity and/
or by mutual host–guest dissociation/association is fast with
respect to the NMR timescale.
The complex between the dye 18 and clip 1a provides
some information about the selectivity of 1a toward guest
structures. In the complex
18·1a, the NMR signals arising
from the pyridinium ring pro-
tons exhibit large upfield shifts,
whereas the positions of the sig-
nals of the protons of the
amino-substituted
benzene
moiety remains almost un-
changed, that is, only the pyridi-
nium ring is included inside the
cavity of 1a. In all examined
pyridinium and amidinium dyes
(18–20), no change in color or
UV/Vis extinction maxima are
found, although NMR data
clearly reveal their inclusion
within the naphthalene side
walls of the cavity. Evidently,
the sandwich complex does
not lead to charge transfer be-
tween host and guest aromat-
ics.[64]
+
Figure 11. Structure (top) and electrostatic potential surface (EPS, bottom) of clip 1a with (NH4
) instead of
2
+
(NBu4
)
as counterions (left), NAD+ (22, middle), and 2’-desoxyadenosine (24, right) calculated by AM1.
2
The color code spans from ꢀ25 (red) to +25 kcalmolꢀ1 (blue). The clip geometry was optimized by AM1,
whereas in the case of NAD+ a single-point AM1 calculation was performed using the NAD+ geometry ob-
tained from the Monte Carlo conformer search of the complex 22·1a (Figure 16, bottom). Note that the EPS
of the nicotinamide subunit in 22 is highly positive compared to that of the adenine moiety.
486
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
Figure 13. The lowest-energy structure of the complex 17·1a including
the tetra-n-butylammonium cations calculated by force-field AMBER*/
H2O with a Monte Carlo conformer search (MacroModel 6.5, 5000 struc-
tures).[47,48]
of theory indicates that although its hydrophobic alkyl
chains are calculated to aggregate with the outside of the
naphthalene sidewalls of the clip, the position of the guest
within the cavity of the clip seems to be only weakly influ-
enced. Therefore, the complex structures were calculated in
the following without the tetra-n-butylammonium cations at
the force-field level. These are used as starting points for
the quantum-chemical studies.
Figure 12. Single-crystal structure of the complex between N-ethyl-4-
carbethoxypyridinium triiodide and diacetoxy-substituted clip 1 f.[58]
Starting from two lowest energy Amber*/H2O structures,
geometry optimizations were performed at the Hartree–
Fock level (HF/6-31G**), with constraints on the guest posi-
tion within the clip (Figure 14). Based on these structures
the chemical shifts were computed by using gauge-including
atomic orbitals (GIAO-HF) and SVP basis sets[65] relative to
the TMS reference computed at the same level. All quan-
tum chemical calculations were performed with the program
packages Q-Chem[66] and TURBOMOLE.[67] In the follow-
ing, we consider only guest protons. The 1H NMR complexa-
tion-induced chemical shieldings of NMNA-1 (see
Figure 14), computed with the different structural parame-
ters obtained with Amber*/H2O and HF/6-31G**, differ by
0.4 ppm at most. Basis-set influences, determined by com-
paring the SVP basis with a TZP basis[65] at the GIAO-HF
level, are 0.1 ppm at most. This is consistent with previous
studies on similar tweezer complexes, for which the accuracy
1
In the H NMR spectra of the complexes 22·1a and 23·1a,
the signals arising from the protons of both subunits (nico-
tinamide and adenine moieties) are shifted upfield. This ob-
servation already allows us to conclude that each complex
must exist in at least two conformations in which either the
nicotinamide or the adenine subunit is included inside the
cavity of 1a. Accordingly, the two conformations must equi-
librate rapidly on the NMR timescale, so that only averaged
Ddmax values are observed. The Ddmax values determined for
the guest protons in the complexes 24·1a, 25·1a, and 26·1a
are of similar magnitude to those of the NAD and NADP
complexes 22·1a and 23·1a. The upfield shifts of the signals
of the complexed nicotinamide protons in 22·1a, 23·1a, and
26·1a are significantly smaller than in 17·1a. This indicates
that the nicotinamide rings of 22, 23, and 26, each attached
to a ribose unit, are positioned inside the clip cavity differ-
ently to the nicotinamide moiety of 17, which is attached to
a methyl group. To gain further insight into this structural
problem we performed quantum chemical 1H NMR shift
calculations.
1
of H NMR chemical shifts computed at the GIAO-HF/SVP
level on HF/6-31G* structures (as compared to calculations
using electron correlation methods) was found to be in the
order of 0.2–0.5 ppm.[68–70] Complexation-induced shieldings
are obtained as the difference of the chemical shifts for the
bound and isolated guest (the structure of the isolated guest
was optimized at the same level of theory) and are listed in
Table 4. Computed and experimental relative NMR shield-
ings for the isolated nicotinamide 17 agree within 0.5 ppm.
For host–guest complex NMNA-1 the maximum difference
is 3.0 ppm, and for NMNA-2 2.1 ppm. For the complexa-
tion-induced chemical shieldings the difference between
theory and experiment is reduced to 2.7 and 1.5 ppm for
NMNA-1 and NMNA-2, respectively. Since the measure-
ments were performed in water as solvent, a strong devia-
Quantum chemical calculations for N-methylnicotinamide
complex 17·1a: The host–guest complex of N-methylnico-
tinamide 17·1a (NMNA for short) was investigated with
quantum chemical methods. In a first step, possible lowest
energy structures were determined by performing a Monte
Carlo conformer search at the force-field level (Amber*/
H2O; MacroModel 6.5; Figure 13).[47,48] Investigation of the
influence of the tetra-n-butylammonium cations at this level
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
487

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
Table 4. Relative (d) and complexation-induced (Dd) 1H NMR chemical
shifts [ppm] computed at the GIAO-HF/TZP level for the HF/6-31G**
optimized N-methylnicotinamide host–guest structures NMNA-1 and
NMNA-2 and the isolated guest structure 17 in comparison with the ex-
perimental results.
Isolated 17
Calcd Exptl
17·1a
NMNA-2
Dd
Proton
NMNA-1
Exptl
d
d
d
Dd
d
d
Dd
H2
H4
H5
H6
9.8
9.2
8.2
8.8
9.3
8.9
8.2
9.0
7.8
3.9
5.8
6.3
2.0
5.3
2.4
2.5
7.1
8.0
6.5
5.5
2.7
1.2
1.7
3.3
8.3
6.9
6.4
7.6
1.2
2.6
2.3
1.8
tion is not surprising. Besides structural flexibility, the influ-
ence of the water environment can be estimated to be on
the order of at least 0.7 ppm for the aromatic guest protons
of NMNA-1.[71] This data was estimated by a preliminary
simulation of the nicotinamide host–guest complex 17·1a in
a water environment by using a snapshot of a force-field
molecular dynamics simulation with the water molecules
closest to the center of the guest ring. We used the snapshot
structure to compute the influence of the water molecules
on the NMR shieldings explicitly at the GIAO-HF/6-31G**
level for a system with up to 1003 atoms using our newly de-
veloped linear-scaling NMR chemical shift method.[71] Clear-
ly, for an accurate study of solvent effects a multitude of
snapshots must be computed. Nevertheless, our calculations
provide a first estimate. Further simulations are in progress.
This indicates that the deviation between computed and
measured complexation-induced shieldings of up to 2.7 ppm
for the NMNA-1 structure seems to be larger than experi-
mental and theoretical error bars, whereas the agreement
for NMNA-2 is within the error bars.
Figure 14. The lowest-energy AMBER*/H2O structures of the complex
17·1a optimized at HF/6-31G** level (with constraints): NMNA-1 (top)
and NMNA-2 (bottom).
1
Clip influences on H NMR shifts of NMNA-2: In addition
to the relative and complexation-induced chemical shield-
ings of 17·1a, we studied the influence of different parts of
clip 1a on the 1H NMR chemical shifts in the NMNA-2
structure. Therefore the clip was partitioned into three
parts: 1) the upper unit including the phosphonate groups,
1a) the upper part without phosphonate groups, and 2) the
naphthalene side walls (Figure 15, Table 5). Open bonds
ꢀ
were saturated with protons (C H 110 pm). The maximum
influences of parts 1 and 2 are 0.7 and 4.3 ppm, respectively
Table 5. Influences of different parts of the clip on the relative chemical
shifts [ppm] (GIAO-HF/SVP) of guest 17 in the NMNA-2 structure (see
Figure 15) optimized at Amber*/H2O force-field level.
[a]
[a]
[a]
[b]
[c]
Proton
disolated
D1
D2
D1a
disolated +D1+2
dtotal
H2
H4
H5
H6
10.0
9.3
8.4
ꢀ0.7
ꢀ0.5
0.2
ꢀ0.3
ꢀ2.1
ꢀ0.8
ꢀ1.7
ꢀ4.3
ꢀ0.3
ꢀ0.3
ꢀ0.5
ꢀ1.9
7.2
8.0
6.9
4.6
7.5
8.2
6.6
4.8
9.2
[a] Change in the chemical shifts of the isolated guest molecule due to
the influence of parts 1, 2, and 1a of the clip. [b] Sum of D1 and D2 added
to the chemical shifts for the isolated guest molecule. [c] Chemical shifts
computed for the full host–guest complex.
Figure 15. Partitioning of the clip within the NMNA-2 structure
(AMBER*/H2O). For part 1a the phosphonate groups (OP(CH3)OOꢀ)
have been removed.
488
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
(GIAO-HF/SVP at Amber*/H2O structure). The maximum
influence of part 1a of 1.9 ppm is stronger than that of
part 1; this result indicates that the effects of the phospho-
nates and the ring currents are in opposite directions and
lead to partial compensation. Overall the sum of the influen-
1
ces of parts 1 and 2 yields H NMR shieldings for the guest
protons that are in good agreement (within 0.3 ppm) with
the full calculation. This indicates a weak influence of the
linking units, as found similarly in other tweezer-shaped
host–guest systems.[70]
1
Computed H NMR chemical shifts for 22·1a: Since for the
NMNA-2 complex 17·1a the influence on the 1H NMR
guest chemical shifts of the HF/6-31G** structure optimiza-
tion was found to be quite small (less than 0.4 ppm) relative
to those calculated with the Amber*/H2O structure, the
Amber*/H2O structures (again resulting from a Monte
Carlo conformer search) were used for the study of the
NAD+ system 22·1a. Although there is a multitude of possi-
ble host–guest structures, we consider here just two arrange-
ments of NAD+ within the clip cavity, as shown in
Figure 16: bound through the nicotinamide (NAD-1) and
adenine parts (NAD-2). The corresponding computed NMR
shieldings are listed in Table 6. The chemical shifts for the
Table 6. Relative (d) and complexation-induced (Dd) 1H NMR chemical
shifts [ppm] computed at the GIAO-HF/SVP level on the Amber*/H2O
structures of the NAD+ host–guest complex: structures NAD-1 and
NAD-2 and the isolated guest structure NAD+ (22) are compared to the
experimental results.
Isolated NAD+ (22)
22·1a
Proton
Calcd
Exptl
NAD-2
NAD-1
Exptl
d
d
d
Dd
d
Dd
d
Dd
H2
H4
H5
H6
H7
H8
H9
9.0
10.1
8.5
9.7
8.4
9.4
8.9
8.3
9.3
8.6
6.1
6.2
8.4
7.7
7.3
8.0
5.8
4.8
5.4
0.6
2.4
1.2
1.7
2.6
1.0
0.1
7.6
3.4
6.1
6.8
7.6
4.5
5.5
1.4
6.7
2.4
2.9
0.8
1.3
0.0
9.0
8.2
7.5
8.8
8.2
5.8
5.9
0.4
0.7
0.8
0.4
0.3
0.3
0.2
Figure 16. The lowest energy AMBER*/H2O structures of the complex
22·1a: NAD-1 (top: the nicotinamide ring inside the clip cavity) and
NAD-2 (bottom: the adenine moiety inside the clip cavity).
5.8
5.5
isolated NAD+ guest 22 agree within 1.2 ppm with experi-
ment, which is within the error bars of theory and experi-
ment. As discussed above, relatively large theoretical error
bars arise due to the missing solvent in the quantum chemi-
cal calculations. Comparison of the chemical shifts for
NAD+ within the clip indicates that the NAD-1 structure
leads to a difference of 4.8/6.0 ppm with respect to relative/
complexation-induced chemical shieldings between theory
and experiment. The second possible structure NAD-2
agrees better with the experiment: deviations are at most
2.4 and 2.3 ppm, respectively. Although this allows the
NAD-1 structure to be discarded, no final and safe conclu-
sion about how the adenine or nicotinamide moiety of
NAD+ is bound to the clip can be drawn at this stage of our
quantum chemical studies due to the multitude of possible
structural arrangements. Here, further investigations are
planned, in particular with the smaller ribose-substituted
systems such as 2’-desoxyadenosine (24), AMP (25), and
NMN (26).
Energetics of NAD+ binding within clip 1a: In addition to
the NMR considerations, we studied the binding energetics
of NAD+ within the clip. For reliable results electron-corre-
lation effects must be accounted for. We used the RI-MP2
approach (resolution of identity Møller–Plesset second-
order perturbation theory)[72] with an SVP basis set. All data
are listed in Table 7. In a first step, we used for the computa-
tion of binding energies only the parts of NAD+ that are
bound within the clip (free bonds were always saturated
ꢀ
with protons, N H 100 pm). Within the clip of charge ꢀ2
the nicotinamide part is bound by 152 kcalmolꢀ1 in NAD-1
as compared to the adenine part bound with 37 kcalmolꢀ1 in
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
489

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
Table 7. Influence of charge and guest structures on binding energies of
NAD+ guest 22 within the clip 1a computed at the RI-MP2/SVP level
[in kcalmolꢀ1].
porting Information, Figure S1). In the future, we will try to
create highly selective NAD+ sensors using modified clips
with built-in ribose recognition elements.[73] Another impor-
tant area is redox chemistry with NAD+. If the electrochem-
ical potential of NAD+ is changed upon complexation, the
substrate profile of dehydrogenases may be altered.[74] Effi-
cient trapping of NAD+ in enzymatic reductions may also
lead to shifted equilibria or even ultimately reverse the
course of the naturally occurring reaction. Finally, we intend
to create artificial enzyme models with noncovalently bound
zinc ions and complexed NAD+ without any protein pres-
ent. In the future, we will carry out a systematic investiga-
tion on the inclusion properties of the other naturally occur-
ring nucleosides. Binding of ATP or GTP might open new
areas of interference in numerous biological pathways. It is
also in principle conceivable that the well-preorganized clip
is capable of a double intercalation mode into electron-poor
base pairs of intact DNA.[75,76] Any of these new interactions
would render the bis-phosphonate clip a valuable new tool
in DNA chemistry. Already our first prototype of a NAD+
or NADP binder, clip 1a, can favorably compete with alco-
hol dehydrogenases for the oxidized form of their substrate:
NAD+ is bound by the natural enzyme with Ka values
around 103 mꢀ1, in contrast to NADH which is grasped much
more tightly (105 mꢀ1). In the future, we wish to exploit this
distinct Ka difference to interfere with enzymatic reductions.
Charge
NAD-1
NAD-2
Clip 1a
NAD+ (22)
ꢀ2
ꢀ2
0
0
0
+1[a]
0[a]
0[a]
+1
0
ꢀ151.7
ꢀ46.4
ꢀ34.1
ꢀ78.9
ꢀ78.3
ꢀ171.4
–
ꢀ36.9
ꢀ35.3
ꢀ71.1
ꢀ70.8
ꢀ144.3
ꢀ2
+1
[a] Only the part of the guest molecule bound inside the clip is consid-
ered (see text).
NAD-2. This is due to the positive charge of the nicotin-
amide unit. If this charge is neutralized, the nicotinamide
binding energy decreases to 46 kcalmolꢀ1. This trend is re-
versed if the neutral guests are considered within a charge-
ꢀ
neutralized clip (saturation with protons, O H 96 pm),
where adenine is slightly more strongly bound: 34 versus
35 kcalmolꢀ1. In the second step, we consider the full NAD+
guest (charge +1) within the clip with charge ꢀ2: binding
through the nicotinamide unit leads to a stabilization energy
of 171 kcalmolꢀ1, and binding through the adenine moiety
to 144 kcalmolꢀ1. If all the charges are neutralized (which
can be assumed to be more realistic for the system in water
as solvent), the stronger stabilization of the nicotinamide
versus adenine binding is much less pronounced: 78 versus
71 kcalmolꢀ1. Although it is clear that these energetics are
highly influenced by structural changes and possibly also by
methodological effects, they provide some insight into influ-
ences on the recognition process and suggest that both sub-
units are bound by the clip with a preference for the nicotin-
amide ring, in agreement with the experimental findings.
Experimental Section
1,4,5,8-Tetrahydro-1,4:5,8-dimethanoanthracene-9,10-bismethylphosphon-
ic acid ester (1b): The hydroquinone precursor[58] (100 mg, 420 mmol) and
methylphosphonic acid dichloride (150 mg 1.13 mmol, 2.7 equiv) were
Conclusion
In this paper we have presented the water-soluble clip 1a as
a host molecule which binds N-alkylpyridinium ions and
electron-poor aromatics highly selectively. The experimental
results obtained for the complexation of the naturally occur-
ring nucleosides or nucleotides 22–26 with 1a and the quan-
tum chemical calculations are good evidence for our as-
sumption that NAD+ (22) and NADP (23) are indeed
bound in a double complex geometry, in which either the
pyridinium or the pyrimidine ring is inserted into the cavity
of 1a. A low-temperature NMR experiment should provide
further insight into the dynamics of complexation. It would
be interesting to examine whether at low temperatures the
fast equilibrium between the complexed and free nicotin-
amide or adenine subunit is slowed down to below the
NMR timescale and two isolated sets of signals appear. This
experiment has been, however, prevented by the low solu-
bility of the complex 22·1a in CD3OD, but it might be possi-
ble to measure the temperature dependence in a mixture of
methanol and water, in which the complex is soluble (Sup-
dissolved in anhydrous THF (10 mL) and cooled to 08C. Then triethyl-
amine (175 mL, 127 mg, 1.26 mmol, 3 equiv) was added dropwise, and a
white solid precipitated after a few seconds. After 1 h, the precipitate
was filtered off under argon, washed with a small amount of anhydrous
THF, and the combined filtrates were treated with 2.5% aqueous HCl
(3 mL). After 15 min of n-hexane (5 mL) was added, and the resulting
two-layer system was stirred overnight, resulting in strong precipitation.
The colorless solid was filtered off, washed with a little 2.5% aqueous
HCL, and dried in vacuo to furnish 110 mg (279 mmol, 66%) of the bis-
phosphonate bismonoester as a colorless solid. M.p. decomp>3008C.
1H NMR (300 MHz, [D6]DMSO): d=1.48 (d, 2J(P,CH3)=17.1 Hz, 6H;
CH3), 2.10 (brs, 4H; 11-Hi, 11-Ha, 12-Hi, 12-Ha), 4.05 (brs, 4H; 1-H, 4-H,
5-H, 8-H), 6.75 ppm (brs, 4H; 2-H, 3-H, 6-H, 7-H); 13C NMR (50 MHz,
1
[D6]DMSO): d=12.42 (d, J(P,C)=139.2 Hz; CH3), 47.90 (d, C-1, C-4, C-
5, C-8), 69.50 (t, C-11, C-12), 136.53 (d, 3J(P,C)=7.4 Hz; C-4a, C-8a, C-
9a, C-10a), 142.32 (d, C-2, C-3, C-6, C-7), 142.92 ppm (d, C-9, C-10);
490
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
31P NMR (81 MHz, [D6]DMSO): d=29.86 ppm (s); MS (ESI, MeOH):
C-8a, C-9a, C-10a), 143.28 (m, C-2, C-3, C-6, C-7), 144.10 ppm (d, C-9, C-
m/z: 196 [Mꢀ2H+]2ꢀ
,
393 [MꢀH+]ꢀ, 425 [MꢀH++MeOH]ꢀ, 787
10); 31P NMR (81 MHz, [D4]MeOH): d=25.29 ppm (s); MS (ESI,
+
2ꢀ
MeOH): m/z: 197 [Mꢀ2NBu4
]
,
415 [Mꢀ2NBu4++Na+]ꢀ, 425
[2MꢀH+]ꢀ, 1181 [3MꢀH+]ꢀ; elemental analysis calcd (%) for
[Mꢀ2NBu4++H++MeOH]ꢀ, 634 [MꢀNBu4+]ꢀ; HRMS (ESI, MeOH):
C18H20P2O6: C 54.83, H 5.11; found: C 54.77, H 5.65.
m/z: calcd for C34H54NP2O6ꢀꢀNBu4+: 634.342; found: 634.341.
6,8,15,17-Tetrahydro-6,17:8,15-dimethanoheptacene-7,16-bis(methylphos-
phonic acid ester) (1a): The hydroquinone precursor[58] (100 mg,
Bis(tetra-n-butylammonium)
6,8,15,17-tetrahydro-6,17:8,15-dimethano-
heptacene-7,16-bismethylphosphonate (1c): the respective phosphonic
acid precursor 1a (51.8 mg, 87.1 mmol), which contained small amounts
228 mmol) and methylphosphonic acid dichloride (80 mg, 0.60 mmol,
2.7 equiv) were dissolved in anhydrous THF (10 mL) and cooled to 08C.
Then triethylamine (80 mL, 58.4 mg, 0.58 mmol, 3 equiv) was added drop-
wise, and a colorless solid precipitated after a few seconds. After 1 h, the
solution was warmed to room temperature and stirred for another 1 h.
The reaction mixture was quenched with 2.5% aqueous HCl (3 mL).
After 20 min, n-hexane (5 mL) was added, and the resulting two-layer
system was stirred overnight. Subsequently the aqueous layer was sepa-
rated, the organic phase washed with of 2.5% aqueous HCl (3 mL), and
the combined organic phases evaporated to dryness on the rotavapor.
Subsequent drying at 0.1 mbar and chromatographic purification over
silica (300ꢅ10 mm, gradient elution with CH2Cl2:MeOH 3:1!2:1) af-
forded 90 mg (0.15 mmol, 66%) of the phosphonate clip as a colorless
solid. TLC: Rf =0.02 (CH2Cl2/MeOH 2:1); m.p. decomp >2708C;
1H NMR (300 MHz, D2O): d=1.49 (d, 2J(PCH3)=16.4 Hz, 6H; PCH3),
(ca. 5%) of silica gel, was suspended in dichloromethane (10 mL) and
treated with aqueous tetra-n-butylammonium hydroxide (135 mL, 1m,
135 mmol, 0.75 equiv). The mixture was stirred at room temperature for
2 h, and then the solvent was evaporated to dryness. The residue was re-
dissolved in methanol and subsequently membrane-filtered and again
evaporated to dryness. If the NMR analysis of this crude product indicat-
ed a small excess of tetra-n-butylammonium hydroxide, the product was
again dissolved in methanol, and the phosphonic acid starting material
(10 mg, 16.8 mmol) was added with vigorous stirring. A second membrane
filtration, evaporation of the clear solution to dryness, and drying at
0.1 mbar furnished 59 mg (67.3 mmol, 99.7% with respect to the amount
of tetra-n-butylammonium hydroxide) of clip 1c as a light brown solid.
M.p. 1408C; 1H NMR (300 MHz, [D4]MeOH): d=0.96 (t, 3J(CH2CH3)=
7.3 Hz, 24H; CH2CH3), 1.24–1.36 (m, 16H; CH2CH3), 1.40 (d, 2J(P-
CH3)=16.6 Hz, 6H; P-CH3), 1.43–1.56 (m, 16H; NCH2CH2), 2.37 (dm,
2J(19-Hi, 19-Ha)=8.0 Hz, 2H; 19-Hi, 20-Hi), 2.65 (dm, 2J(19-Ha, 19-Hi)=
7.8 Hz, 2H; 19-Ha, 20-Ha), 2.99–3.08 (m, 16H; NCH2CH2), 4.79 (brs, 4H;
6-H, 8-H, 15-H, 17-H), 7.19 (ddm, 3J(H-1, H-2)=6.1 Hz, 4J(H-4, H-2)=
3.2 Hz, 4H; 2-H, 3-H, 11-H, 12-H), 7.54 (ddm, 3J(H-2, H-1)=6.1 Hz,
4J(H-4, H-2)=3.2 Hz, 4H; 1-H, 4-H, 10-H, 13-H), 7.59 ppm (brs, 4H; 5-
H, 9-H, 14-H, 18-H). 13C NMR (50 MHz, [D4]MeOH): d=13.43 (d,
1J(P,C)=137.2 Hz; C-21, C-22), 13.90 (q, C-26), 20.59 (t, C-25), 24.64 (t,
C-24), 59.33 (t, C-23), 65.77 (t, C-19, C-20), 120.82 (d, C-5, C-9, C-14, C-
18), 125.96 (d, C-2, C-3, C-11, C-12), 128.62 (d, C-1, C-4, C-10, C-13),
133.51 (s, C-4a, C-9a, C-13a, C-18a), 142.09 (s, C-5a, C-8a, C-14a, C-17a),
148.91 ppm (d, C-7, C-16); the signals for C-6, C-8, C-15, C17 are expect-
ed under the deuterium-coupled septet of [D4]MeOH; the signals for C-
6a, C-7a, C-15a, C-16a were too weak to be detected in [D4]MeOH;
31P NMR (81 MHz, [D4]MeOH): d=25.29 ppm (s); MS (ESI, MeOH):
2
2
2.42 (dm, J(19-Hi, 19-Ha)=8.5 Hz, 2H; 19-Hi, 20-Hi), 2.66 (dm, J(19-Ha,
19-Hi)=8.3 Hz, 2H; 19-Ha, 20-Ha), 4.69 (brs, 4H; 6-H, 8-H, 15-H, 17-H),
7.17 (ddm, 3J(H-1, H-2)=6.2 Hz, 4J(H-4, H-2)=3.3 Hz, 4H; 2-H, 3-H,
3
4
11-H, 12-H), 7.48 (ddm, J(H-2, H-1)=6.1 Hz, J(H-2, H-4)=3.3 Hz, 4H;
1-H, 4-H, 10-H, 13-H), 7.39 ppm (brs, 4H; 5-H, 9-H, 14-H, 18-H);
31P NMR (81 MHz, D2O): d=15.32 ppm (s); MS (ESI, MeOH): m/z: 296
[Mꢀ2H+]2ꢀ, 593 [MꢀH+], 615 [Mꢀ2H++Na+], 625 [MꢀH++MeOH]ꢀ.
Bis(tetra-n-butylammonium) 1,4,5,8-tetrahydro-1,4:5,8-dimethanoanthra-
cene-9,10-bismethylphosphonate (1d): The corresponding phosphonic
acid 1b (50 mg, 127 mmol) was suspended in dichloromethane (ca.
+
+
m/z: 296 [Mꢀ2NBu4
]
2ꢀ, 615 [Mꢀ2NBu4++Na+]ꢀ, 625 [Mꢀ2NBu4
+H++MeOH]ꢀ, 834 [MꢀNBu4+]ꢀ; HRMS (ESI, MeOH): m/z: calcd for
C50H62NP2O6ꢀꢀNBu4+: 834.405; found: 834.407.
1,4,5,8-Tetrahydro-1,4:5,8-dimethanoanthracene-9,10-bis(phenylphos-
phonic acid ester) (2b): The hydroquinone precursor (100 mg, 420 mmol)
and phenylphosphonic acid dichloride (117 mL, 164 mg, 841 mmol,
10 mL) and treated with aqueous tetra-n-butylammonium hydroxide
(253 mL, 1m, 253 mmol, 2.0 equiv). The mixture was stirred for 2 h at am-
bient temperature, and the solvent evaporated to dryness. Drying at
0.1 mbar afforded a quantitative yield of tetra-n-butylammonium salt 1d
as a colorless solid. M.p. 1788C; 1H NMR (300 MHz, [D4]MeOH): d=
1.03 (t, 3J(CH2CH3)=7.3 Hz, 24H; CH2CH3), 1.26 (d, 2J(PCH3)=
16.4 Hz, 6H; PCH3), 1.35–1.49 (m, 16H; CH2CH3), 1.60–1.74 (m, 16H;
NCH2CH2), 2.15 (brs, 4H; 11-H, 12-H), 3.20–3.29 (m, 16H; NCH2CH2),
4.16 (brs, 4H; 1-H, 4-H, 5-H, 8-H), 6.79 ppm (brs, 4H; 2-H, 3-H, 6-H, 7-
H); 13C NMR (50 MHz, [D4]MeOH): d=13.27 (d, 1J(P,C)=136.5 Hz; C-
13, C-14), 13.93 (q, C-18), 20.73 (t, C-17), 24.79 (t, C-16), 59.51 (t, C-15),
70.63 (t, C-11, C-12), 139.62 (dd, 3J(P,C)=8.0 Hz, 4J(P,C)=1.9 Hz; C-4a,
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
491

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
2.0 equiv) were dissolved in anhydrous THF (10 mL) and cooled to 08C.
Then triethylamine (175 mL, 127 mg, 1.26 mmol, 3 equiv) was added
dropwise, and a white solid precipitated after a few seconds. After 1 h,
the precipitate was filtered off under argon, washed with a little anhy-
drous THF, and the combined filtrates were treated with 2.5% aqueous
HCl (3 mL). After 15 min, n-hexane (5 mL) was added, and the resulting
two-layer system was stirred overnight, resulting in strong precipitation.
The white solid was filtered off, washed with a little 2.5% aqueous HCL
and dried in vacuo to furnish 110 mg (212 mmol, 50%) of the bis-phos-
1
phonate bis-monoester as a white solid. M.p. decomp at 2908C; H NMR
(300 MHz, [D6]DMSO): d=1.82 (dm, 2J(12-Hi, 12-Ha)=6.9 Hz, 2H; 11-
Hi, 12-Hi), 1.92 (dm, 2J(12-Ha, 12-Hi)=7.1 Hz, 2H; 11-Ha, 12-Ha), 3.79
(brs, 4H; 1-H, 4-H, 5-H, 8-H), 6.38 (brs, 4H; 2-H, 3-H, 6-H, 7-H), 7.44–
7.63 (m, 6H; HPh), 7.66–7.83 ppm (m, 4H; HPh); 13C NMR (50 MHz,
[D6]DMSO): d=47.93 (d, C-1, C-4, C-5, C-8), 69.34 (t, C-11, C-12),
128.59 (d, 2J(P,C)=14.8 Hz; C-14, C-14a, C-18, C-18a), 130.80 (d,
1J(P,C)=185.4 Hz; C-13, C-13a), 131.70 (d, 3J(P,C)=9.7 Hz; C-15, C-15a,
C-17, C-17a), 132.22 (m, C-16, C-16a), 136.50 (dd, 3J(P,C)=7.9 Hz,
4J(P,C)=2.3 Hz; C-4a, C-8a, C-9a, C-10a), 142.30 (m, C-2, C-3, C-6, C-7),
142.92 ppm (m, C-9, C-10); 31P NMR (81 MHz, [D6]DMSO): d=
17.45 ppm (s); MS (ESI, MeOH): m/z: 258 [Mꢀ2H+]2ꢀ, 539 [Mꢀ2H+
+Na+]ꢀ, 517 [MꢀH+]ꢀ.; elemental analysis calcd (%) for
C28H24P2O6·2H2O: C 60.43, H 5.43; found: C 60.37, H 5.10.
(ca. 10 mL), and was treated with aqueous tetra-n-butylammonium hy-
droxide (232 mL, 1m, 232 mmol, 2.0 equiv). The mixture was stirred for
2 h at ambient temperature, and the solvent evaporated to dryness.
Drying at 0.1 mbar afforded a quantitative yield of tetra-n-butylammoni-
um salt 2d as
a
brown solid. M.p. 1848C; 1H NMR (300 MHz,
[D4]MeOH): d=1.02 (t, 3J(CH2CH3)=7.3 Hz, 24H; CH2CH3), 1.34–1.49
(m, 16H; CH2CH3), 1.59–1.73 (m, 16H; NCH2CH2), 1.82 (brs, 4H; 11-H,
12-H), 3.18–3.29 (m, 16H; NCH2CH2), 3.74 (brs, 4H; 1-H, 4-H, 5-H, 8-
H), 6.28 (brs, 4H; 2-H, 3-H, 6-H, 7-H), 7.28–7.44 (m, 6H; HPh), 7.68–
7.79 ppm (m, 4H; HPh). 13C NMR (50 MHz, [D4]MeOH): d=13.94 (q, C-
22), 20.72 (t, C-21), 24.79 (t, C-20), 59.49 (t, C-19), 70.33 (t, C-11, C-12),
128.60 (d, 2J(P,C)=13.7 Hz; C-14, C-14a, C-18, C-18a), 131.07 (m, C-16,
C-16a), 133.26 (d, 3J(P,C)=9.1 Hz; C-15, C-15a, C-17, C-17a), 137.02 (d,
1J(P,C)=177.38 Hz; C-13, C-13a), 139.48 (dd, 3J(P,C)=8.0 Hz, 4J(P,C)=
1.9 Hz; C-4a, C-8a, C-9a, C-10a), 143.17 (m, C-2, C-3, C-6, C-7),
6,8,15,17-Tetrahydro-6,17:8,15-dimethanoheptacene-7,16-bis(phenylphos-
phonic acid ester) (2a): The hydroquinone precursor (100 mg, 228 mmol)
was dissolved in anhydrous THF (10 mL) and cooled to 08C. Then tri-
143.78 ppm (m, C-9, C-10); 31P NMR (81 MHz, [D4]MeOH): d=
+
14.25 ppm (s); MS (ESI, MeOH); m/z: 259 [Mꢀ2NBu4
]
2ꢀ, 517 [Mꢀ2
NBu4++H+]ꢀ, 549 [Mꢀ2NBu4++H++MeOH], 758 [MꢀNBu4+]ꢀ;
HRMS (ESI, MeOH): m/z: calcd for C44H58NP2O6ꢀꢀNBu4+: 758.37394;
found: 758.36129.
NMR titrations: Ten NMR tubes were filled each with 0.6 mL of a so-
lution of the host compound ([H]=0.5–4 mm) in a deuterated solvent
([D6]DMSO or CD3OD). The guest compound G (about 1.5 equiv rela-
tive to the host) was dissolved in 0.61 mL of the same solvent, and the re-
sulting solution was added with increasing volumes from 0 to 5 equiv to
the host solution in ten NMR tubes. Volume and concentration changes
were taken into account during analysis. The association constants were
calculated by nonlinear regression methods [Eq. (1)] in which K=[H]0/
[G]0.
ethylamine (95 mL, 69.4 mg, 686 mmol, 3.0 equiv) and phenylhydrazine
(55 mL) were added and stirred for 15 min, after which phenylphosphonic
acid dichloride (100 mL, 140 mg, 718 mmol, 3.1 equiv) was added drop-
wise. After stirring for 3 h at room temperature, the precipitate was fil-
tered off under argon, washed with a little anhydrous THF, and the com-
bined filtrates were treated with 2.5% aqueous HCl (3 mL). After
15 min n-hexane (10 mL) was added, and the resulting two-layer system
was stirred overnight. Subsequently, the aqueous layer was separated, the
organic phase again washed with 2.5% aqueous HCl (3 mL), and the
combined organic phases were dried with MgSO4 and evaporated to dry-
ness on a rotavapor. Subsequent drying at 0.1 mbar and chromatographic
purification over silica (300ꢅ10 mm, elution with CH2Cl2/MeOH 3:1) af-
forded 40 mg (55.7 mmol, 24%) of the clip as a brown solid. TLC: Rf =
0.21 (CH2Cl2:MeOH 2:1); m.p. decomp at 2958C; 1H NMR (300 MHz,
D2O): d=2.11 (dm, 2J(19-Hi, 19-Ha)=8.3 Hz, 2H; 19-Hi, 20-Hi), 2.24
(dm, 2J(19-Ha, 19-Hi)=8.3 Hz, 2H; 19-Ha, 20-Ha), 4.13 (brs, 4H; 6-H, 8-
H, 15-H, 17-H), 7.22 (ddm, 3J(H-1, H-2)=6.0 Hz, 4J(H-4, H-2)=3.3 Hz,
4H; 2H, 3-H, 11-H, 12-H), 7.39 (brs, 4H; 5-H, 9-H, 14-H, 18-H), 7.51
(ddm, 3J(H-2, H-1)=6.0 Hz, 4J(H-2, H-4)=3.3 Hz, 4H; 1-H, 4-H, 10-H,
13-H), 7.55–7.63 (m, 6H; HPh), 7.80–7.88 ppm (m, 4H; HPh); 13C NMR:
could not be obtained due to poor solubility; 31P NMR (81 MHz, D2O):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ꢀ
ꢁ
Ddmax
K
½Hꢂ0Ka
2K2
½Hꢂ0Ka
K2
2 K
½Hꢂ0Ka
Dd ¼
K þ 1 þ
ꢀ
K2 þ
ꢀ2K þ
þ
þ 1
ð1Þ
2
½Hꢂ0K2a
2
For dilution titrations equimolar amounts of host and guest compound
were dissolved in deuterated methanol (0.5 mL) or water (0.5 mL) ([G]=
[H]=0.5–4mm). From this reference sample, aliqouts of 250, 125, 75, 50,
und 25 mL were taken, and deuterated solvent was added to a total
volume of 0.5 mL. Only those signals were used for quantitative evalua-
tion which could be clearly detected during the whole titration. Binding
constants Ka were determined by nonlinear regression.
Job plots: Equimolar solutions (10 mmol/10 mL, ca. 10 mm) of host and
guest compound were prepared and mixed in various amounts. 1H NMR
spectra of the mixtures were recorded, and the chemical shifts were ana-
lysed by Jobꢂs method, modified for NMR data.
Salt-effect experiment: A dilution titration was carried out according to
the general protocol outlined above (host and guest concentration:
10ꢀ4 m), but in the presence of 0.5m aqueous tetra-n-butylammonium bro-
mide. The change in chemical shifts became much smaller and the resul-
tant binding constant from the nonlinear regression of the binding curves
between host and N-methylnicotinamide dropped to almost zero (ca.
50mꢀ1).
d=15.32 ppm (s); MS (ESI, MeOH); m/z: 358 [Mꢀ2H+]2ꢀ, 717 [MꢀH+]ꢀ,
ꢀ
749 [MꢀH++MeOH]ꢀ; HRMS (ESI, MeOH) calcd for C44H31P2O6
:
717.15959; found: 717.16184.
Bis(tetra-n-butylammonium) 1,4,5,8-tetrahydro-1,4:5,8-dimethanoanthra-
cene-9,10-bisphenylphosphonate (2d): The corresponding phosphonic
acid dihydrate 2a (64.2 mg, 116 mmol) was suspended in dichloromethane
Microcalorimetry experiments: All titration experiments were performed
on
a TAM 2277 microcalorimeter (Thermometric, Jꢀrfꢀlla, Sweden)
492
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494

Molecular Clips
FULL PAPER
using the ampoule unit 2277-201. The temperature during the experi-
ments was 298 K and we used water as solvent. 1 mL of the receptor so-
lution was filled into the cell of the microcalorimeter. The substrate so-
lution was added during the titration experiment by a syringe pump
6120-031 (Lund, Sweden).
[17] O. Livnah, E. A. Stura, D. L. Johnson, S. A. Middleton, L. S. Mulca-
hy, N. C. Wrighton, W. J. Dower, L. K. Jolliffe, I. A. Wilson, Science
1996, 273, 464–471.
[18] D. A. Dougherty, H. A. Lester, Nature 2001, 411, 252–254.
[19] K. Brejc, W. J. van Dijk, R. V. Klaassen, M. Schuurmans, J. van der
Oost, A. B. Smit, T. K. Sixma, Nature 2001, 411, 269–276.
[20] J. C. Ma, D. A. Dougherty, Chem. Rev. 1997, 97, 1303–1324.
[21] D. A. Dougherty, Science 1996, 271, 163–168.
Mass spectrometry: ESI mass spectra were recorded on a Finnigan MAT
95. Samples (20 mL) were introduced as 10ꢀ5 m solutions in methanol at
flow rates of 20 mLminꢀ1. Heated capillary temperature: 1508C. Ion
spray potential: 3.5 kV (positive ESI), 3.0 kV (negative ESI). About 20–
30 scans were averaged to improve the signal-to-noise ratio.
[22] H.-J. Schneider, T. Schiestel, P. Zimmerman, J. Am. Chem. Soc.
1992, 114, 7698–7703.
[23] H. J. Schneider, Chem. Soc. Rev. 1994, 23, 227–234.
[24] M. Dhaenens, L. Lacombe, J.-M. Lehn, J.-P. Vigneron, J. Chem. Soc.
Chem. Commun. 1984, 1097–1099.
[25] R. Meric, J. P. Vigneron, J. M. Lehn, J. Chem. Soc. Chem. Commun.
1993, 129–131.
[26] H. J. Schneider, R. Kramer, S. Simova, U. Schneider, J. Am. Chem.
Soc. 1988, 110, 6442–6448.
[27] K. Araki, H. Shimizu, S. Shinkai, Chem. Lett. 1993, 205–208.
[28] K. Murayama, K. Aoki, Chem. Commun. 1997, 119–120.
[29] L. Garel, B. Lozach, J. P. Dutasta, A. Collet, J. Am. Chem. Soc.
1993, 115, 11652–11653.
[30] S. Kubik, J. Am. Chem. Soc. 1999, 121, 5846–5855.
[31] S. M. Ngola, P. C. Kearney, S. Mecozzi, K. Russell, D. A. Dougherty,
J. Am. Chem. Soc. 1999, 121, 1192–1201.
[32] S. Rensing, M. Arendt, A. Springer, T. Grawe, T. Schrader, J. Org.
Chem. 2001, 66, 5814–5821.
Simulation methods: Molecular mechanics calculations, Monte-Carlo
simulations, and molecular dynamics: The program MacroModel 7.1 or
6.5[47,48] was used for model building procedures and as graphical inter-
face. Force-field parameters were taken from the built-in force fields,
which were in some cases modified versions of the classical published
versions. OPLS-AA and Amber* produced very similar results; the latter
was subsequently chosen for all minimizations and Monte Carlo simula-
tions. Minimizations were initially carried out in the gas phase, then in
aqueous solution. Most complex structures were virtually identical under
both conditions; this indicates strong enthalpic preference and hence sta-
bility of these arrangements. Energy minimizations were conducted over
1000 iterations on a Silicon Graphics O2 workstation or IBM workstation
RS/6000 34P model 260. The best structures were subjected to conforma-
tional searches with 5000-step Monte Carlo simulations.
[33] T. J. Shepodd, M. A. Petti, D. A. Dougherty, J. Am. Chem. Soc.
1988, 110, 1983–1985.
[34] A. Schenning, B. Debruin, A. E. Rowan, H. Kooijman, A. L. Spek,
R. J. M. Nolte, Angew. Chem. 1995, 107, 2288–2289; Angew. Chem.
Int. Ed. Engl. 1995, 34, 2132–2134.
Acknowledgement
[35] D. Philp, J. F. Stoddart, Angew. Chem. 1996, 108, 1242–1286;
Angew. Chem. Int. Ed. Engl. 1996, 35, 1154–1196.
[36] M. Lamsa, J. Huuskonen, K. Rissanen, J. Pursiainen, Chem. Eur. J.
1998, 4, 84–92.
[37] A. E. Rowan, J. Elemans, R. J. M. Nolte, Acc. Chem. Res. 1999, 32,
995–1006.
[38] J. Elemans, M. B. Claase, P. P. M. Aarts, A. E. Rowan, A. Schenning,
R. J. M. Nolte, J. Org. Chem. 1999, 64, 7009–7016.
This work was supported by the DFG (Deutsche Forschungsgemein-
schaft, Sonderforschungsbereich SFB 452) and the Fonds der Chemischen
Industrie. C.O. acknowledges financial support by an Emmy Noether re-
search grant of the DFG and the BMBF (Bundesministerium fꢁr Bildung
und Forschung) within the “Zentrum fꢁr Multifunktionelle Werkstoffe
und Miniaturisierte Funktionseinheiten” (BMBF 03N 6500). We thank
Professor Monika Mazik (Universitꢀt Braunschweig) for her assistance
with the interpretation of the calorimetric measurements.
[39] F. P. Schmidtchen, Chem. Ber. 1981, 114, 597–607.
[40] H. Fenniri, M. W. Hosseini, J. M. Lehn, Helv. Chim. Acta 1997, 80,
786–803.
[41] A. Domenech, E. Garcia-Espana, J. A. Ramirez, B. Celda, M. C.
Martinez, D. Monleon, R. Tejero, A. Bencini, A. Bianchi, J. Chem.
Soc. Perkin Trans. 2 1999, 23–32.
[42] F. G. Klꢀrner, B. Kahlert, Acc. Chem. Res. 2003, 36, 919–932.
[43] F.-G. Klꢀrner, J. Benkhoff, R. Boese, U. Burkert, M. Kamieth, U.
Naatz, Angew. Chem. 1996, 108, 1195–1198; Angew. Chem. Int. Ed.
Engl. 1996, 35, 1130–1133.
[1] D. Voet, J. G. Voet, Biochemistry, VCH, Weinheim, 1994.
[2] A. Fersht, Enzyme Structure and Mechanism, Freeman, New York,
1985.
[3] K. Dalziel, The Enzymes Vol. XI, Academic Press, New York, 1975.
[4] A. Wilkinson, J. Day, R. Bowater, Mol. Microbiol. 2001, 40, 1241–
1248.
[5] M. G. Rossmann, D. Moras, K. W. Olsen, Nature 1974, 250, 194–
199.
[44] F. G. Klꢀrner, J. Polkowska, J. Panitzky, U. P. Seelbach, U. Burkert,
M. Kamieth, M. Baumann, A. E. Wigger, R. Boese, D. Blꢀser, Eur.
J. Org. Chem. 2004, 1405–1423.
[6] A. M. Lesk, Curr. Opin. Struct. Biol. 1995, 5, 775–783.
[7] C. A. Bottoms, P. E. Smith, J. J. Tanner, Protein Sci. 2002, 11, 2125–
2137.
[45] M. Herm, O. Molt, T. Schrader, Chem. Eur. J. 2002, 8, 1485–1499.
[46] C. Jasper, T. Schrader, J. Panitzky, F.-G. Klꢀrner, Angew. Chem.
2002, 114, 1411–1415; Angew. Chem. Int. Ed. 2002, 41, 1355–1358.
[47] F. Mohamadi, N. G. J. Richards, W. C. Guida, R. Liskamp, M.
Lipton, C. Caufield, G. Chang, T. Hendrickson, W. C. Still, J.
Comput. Chem. 1990, 11, 440–467.
[48] Macromodel, v. 7.1, Schrçdinger, 1500 SW First Ave., Ste. 1180,
Portland, OR 97201.
[49] C. S. Wilcox, Frontiers in Supramolecular Chemistry, Photochemistry,
VCH, Weinheim, 1991.
[8] S. Ramaswamy, M. ElAhmad, O. Danielsson, H. Jornvall, H.
Eklund, Protein Sci. 1996, 5, 663–671.
[9] J. P. Gallivan, D. A. Dougherty, J. Am. Chem. Soc. 2000, 122, 870–
874.
[10] W. L. Zhu, X. J. Tan, C. M. Puah, J. D. Gu, H. L. Jiang, K. X. Chen,
C. E. Felder, I. Silman, J. L. Sussman, J. Phys. Chem. A 2000, 104,
9573–9580.
[11] S. Tsuzuki, M. Yoshida, T. Uchimaru, M. Mikami, J. Phys. Chem. A
2001, 105, 769–773.
[12] S. Mecozzi, A. P. West, D. A. Dougherty, J. Am. Chem. Soc. 1996,
118, 2307–2308.
[13] S. K. Burley, G. A. Petsko, FEBS Lett. 1986, 203, 139–143.
[14] J. B. O. Mitchell, C. L. Nandi, I. K. McDonald, J. M. Thornton, S. L.
Price, J. Mol. Biol. 1994, 239, 315–331.
[50] L. Fielding, Tetrahedron 2000, 56, 6151–6170.
[51] We would like to thank Prof. H. J. Schneider for a copy of his pro-
gram for 1:1 complexes. Binding constants were determined with
Sigma Plot 3.02 from Jandel Corporation.
[52] P. Job, C. R. Hebd. Seances Acad. Sci. 1925, 180, 925.
[53] K. A. Connors, Binding Constants, The Measurement of Molecular
Complex Stability, Wiley, New York, 1987.
[15] S. Karlin, M. Zuker, L. Brocchieri, J. Mol. Biol. 1994, 239, 227–248.
[16] A. M. de Vos, M. Ultshc, A. A. Kossiakoff, Science 1992, 255, 306–
312.
Chem. Eur. J. 2005, 11, 477 – 494
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
493

T. Schrader, C. Ochsenfeld, F. G. Klꢀrner et al.
[54] M. T. Blanda, J. H. Horner, M. Newcomb, J. Org. Chem. 1989, 54,
4626–4636.
[55] S. L. Tobey, E. V. Anslyn, J. Am. Chem. Soc. 2003, 125, 10963–
10970.
H. Daschel, W. M. Zhang, P. P. Korambath, J. Baker, E. F. C. Byrd,
T. Van Voorhis, M. Oumi, S. Hirata, C. P. Hsu, N. Ishikawa, J. Flo-
rian, A. Warshel, B. G. Johnson, P. M. W. Gill, M. Head-Gordon,
J. A. Pople, J. Comput. Chem. 2000, 21, 1532–1548.
[56] M. Berger, F. P. Schmidtchen, Angew. Chem. 1998, 110, 2840–2842;
Angew. Chem. Int. Ed. 1998, 37, 2694–2696.
[67] R. Ahlrichs, M. Bꢀr, M. Hꢀser, H. Horn, C. Kçlmel, Chem. Phys.
Lett. 1989, 162, 165.
[57] P. J. Garratt, A. J. Ibbett, J. E. Ladbury, R. OꢂBrien, M. B. Hurst-
house, K. M. A. Malik, Tetrahedron 1998, 54, 949–968.
[58] F.-G. Klꢀrner, J. Panitzky, D. Blaser, R. Boese, Tetrahedron 2001, 57,
3673–3687.
[68] C. Ochsenfeld, Phys. Chem. Chem. Phys. 2000, 2, 2153–2159.
[69] S. P. Brown, T. Schaller, U. P. Seelbach, F. Koziol, C. Ochsenfeld, F.-
G. Klꢀrner, H. W. Spiess, Angew. Chem. 2001, 113, 740–743; Angew.
Chem. Int. Ed. 2001, 40, 717–720.
[59] E. A. Meyer, R. K. Castellano, F. Diederich, Angew. Chem. 2003,
115, 1244–1287; Angew. Chem. Int. Ed. 2003, 42, 1210–1250.
[60] C. Tanford, The Hydrophobic Effect, Wiley, New York, 1980.
[61] W. Blokzijl, J. Engberts, Angew. Chem. 1993, 105, 1610–1648;
Angew. Chem. Int. Ed. Engl. 1993, 32, 1545–1579.
[70] C. Ochsenfeld, F. Koziol, S. P. Brown, T. Schaller, U. P. Seelbach, F.-
G. Klꢀrner, Solid State Nucl. Magn. Reson. 2002, 22, 128–153.
[71] C. Ochsenfeld, J. Kussmann, F. Koziol, Angew. Chem. 2004, 116,
4585–4589; Angew. Chem. Int. Ed. 2004, 43, 4485–4489.
[72] K. Eichkorn, O. Treutler, H. ꢆhm, M. Hꢀser, R. Ahlrichs, Chem.
Phys. Lett. 1995, 240, 283–289.
[62] N. Muller, Acc. Chem. Res. 1990, 23, 23–28.
[63] T. Lazaridis, Acc. Chem. Res. 2001, 34, 931–937.
[64] This may have two reasons: 1) The distance between the pyridinium
cation and the embracing walls is still too large for van der Waals
contact; 2) the HOMO–LUMO gap between the naphthalene donor
and the pyridinium donor is too large.
[73] T. D. James, K. Sandanayake, S. Shinkai, . Angew. Chem. 1996, 108,
2038–2050; Angew. Chem. Int. Ed. Engl 1996, 35, 1910–1922.
[74] The natural redox potential has the following value: E0 =
ꢀ0.315 V.[1] Inside the clip, it will most probably be lowered.
[75] L. S. Lerman, J. Mol. Biol. 1961, 3, 18.
[65] A. Schꢀfer, H. Horn, R. Ahlrichs, J. Chem. Phys. 1992, 97, 2571.
[66] J. Kong, C. A. White, A. I. Krylov, D. Sherrill, R. D. Adamson, T. R.
Furlani, M. S. Lee, A. M. Lee, S. R. Gwaltney, T. R. Adams, C. Och-
senfeld, A. T. B. Gilbert, G. S. Kedziora, V. A. Rassolov, D. R. Maur-
ice, N. Nair, Y. H. Shao, N. A. Besley, P. E. Maslen, J. P. Dombroski,
[76] K. E. Erkkila, D. T. Odom, J. K. Barton, Chem. Rev. 1999, 99, 2777–
2795.
Received: June 15, 2004
Published online: November 24, 2004
494
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 477 – 494